University of Groningen Dynamic modelling of energy stocks and flows in the economy … ·...

University of Groningen

Dynamic modelling of energy stocks and flows in the economyBattjes, Jacobus Johannes

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:1999

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Battjes, J. J. (1999). Dynamic modelling of energy stocks and flows in the economy: an energy accountingapproach. s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 21-04-2021

https://research.rug.nl/en/publications/dynamic-modelling-of-energy-stocks-and-flows-in-the-economy(64366553-8de3-4b3c-a14e-3db467da9db9).html

Dynamic Modelling of Energy Stocks and Flows in the Economy

An Energy Accounting Approach

Beoordelingscommissie:

Prof. dr. W.A. HafkampProf. dr. J. KommandeurProf. dr. J. Oosterhaven

Copyright (C) 1999 by J.J. Battjes

ISBN: 90-367-1063-4

RIJKSUNIVERSITEIT GRONINGEN

Dynamic Modelling of Energy Stocks and Flows in the Economy

An Energy Accounting Approach

Proefschrift

ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen

op gezag van deRector Magnificus, dr. D.F.J. Bosscher

in het openbaar te verdedigen opvrijdag 7 mei 1999

om 16.00

door

Jacobus Johannes Battjes

geboren op 22 oktober 1969

te Roden

Promotor: Prof. dr. A.J.M. Schoot Uiterkamp

Referenten: dr. K.J. Noormandr. H.C. Moll

Aan mijn vader en aan Wouter die beiden op hun wijze aan de wieg stonden van dit proefschrift

Contents

Voorwoord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Chapter 1Dynamic Modelling of Energy Stocks and Flows in the Economy . 131.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.2 Economic Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.2.1 Neoclassical Economics . . . . . . . . . . . . . . . . . . . . . . . . 171.2.2 Resource Economics . . . . . . . . . . . . . . . . . . . . . . . . . . 191.2.3 Ecological Economics . . . . . . . . . . . . . . . . . . . . . . . . . 211.2.4 Thermodynamics and the Economic System . . . . . . . . . 24

1.3 Resource Accounting Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 261.3.1 General Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261.3.2 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

1.4 ECCO-Modelling Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . 291.5 Scope of this Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301.6 Research Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311.7 Overview of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Chapter 2General Concepts of the ECCO-Modelling Approach . . . . . . . . . . 332.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.2 Embodied Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.3 ERE-Values and Energy Savings . . . . . . . . . . . . . . . . . . . . . . . . . 362.4 Utility versus Real Energy Use . . . . . . . . . . . . . . . . . . . . . . . . . . 392.5 System-Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.6 General Concepts of ECCO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

2.6.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412.6.2 Production from an Energy Perspective . . . . . . . . . . . . 422.6.3 Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452.6.4 Energy Supply System . . . . . . . . . . . . . . . . . . . . . . . . . 462.6.5 Imports and Exports . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Chapter 3Regional Input-Output Analysis for OECD-Europe . . . . . . . . . . . . 493.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

3.1.1 General Methodology of Determining Energy Intensitieswith IO-analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.1.2 Comparison of National Energy Intensities . . . . . . . . . . 533.2 A Second Single Region Approach for Assessing Energy Intensities of

Imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.2.1 Methodology of the single region approach of OECD-Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.2.2 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573.2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

3.3 Multi-Regional Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.3.1 Methodology of the Multi-Regional Approach . . . . . . . 603.3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.4 Energy Flows in OECD-Europe . . . . . . . . . . . . . . . . . . . . . . . . . . 623.5 Conclusions and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

Chapter 4Structure of ECCO-Models for OECD-Europe . . . . . . . . . . . . . . . 654.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654.2 Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.2.1 Sector’s Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674.2.2 Investments and Capital Stock . . . . . . . . . . . . . . . . . . . 67

4.3 Energy System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684.3.1 Energy Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684.3.2 Energy Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

4.4 Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754.4.1 Population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754.4.2 Material Standard of Living . . . . . . . . . . . . . . . . . . . . . 754.4.3 Consumption Goods and Services . . . . . . . . . . . . . . . . 764.4.4 Dwellings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774.4.5 Private Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

4.5 Balance of Trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 784.6 Multi-regional Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

4.6.1 Trade in Goods and Services . . . . . . . . . . . . . . . . . . . . 804.6.2 Trade in Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

Chapter 5Scenario Results of the Regional ECCO-Models . . . . . . . . . . . . . . 835.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 835.2 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

5.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 835.2.2 Imports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855.2.3 Energy and Electricity Demand . . . . . . . . . . . . . . . . . . 865.2.4 Intermediate Deliveries . . . . . . . . . . . . . . . . . . . . . . . . . 875.2.5 ERE-values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 905.2.6 Electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925.2.7 Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 935.2.8 General Conclusions of Sensitivity Analyses . . . . . . . . . 94

5.3 The Impact of Regional Differences . . . . . . . . . . . . . . . . . . . . . . . 955.3.1 Regional Differences . . . . . . . . . . . . . . . . . . . . . . . . . . 97

5.3.2 Comparison between Results of Single Region Model andMulti-regional Model . . . . . . . . . . . . . . . . . . . . . . . 109

5.3.3 General Conclusions of Studying Regional Aspects . . . 114

Chapter 6Bridging Part I and Part II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

Chapter 7Dynamic Resource and Economy Accounting Model . . . . . . . . . . 1197.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1197.2 Main Dynamics of ECCO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1197.3 A Demand-driven Modelling Approach . . . . . . . . . . . . . . . . . . . 120

7.3.1 Arguments for a Demand-driven Modelling Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

7.3.2 Major Concepts of the DREAM-Modelling Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

7.4 Detailed Overview of the DREAM-Modelling Approach . . . . . . 1267.4.1 Investments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1267.4.2 Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1277.4.3 Balance of Trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

7.5 Case Study of the Dutch DREAM-model . . . . . . . . . . . . . . . . . . 1307.5.1 From NLECCO to NLDREAM . . . . . . . . . . . . . . . . . 1307.5.2 Starting Points of Scenarios . . . . . . . . . . . . . . . . . . . . 1317.5.3 Scenario Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

7.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

Chapter 8Scenario Results of OECD-DREAM . . . . . . . . . . . . . . . . . . . . . . . 1418.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1418.2 Comparison of DREAM and the ECCO-Scenarios in OECD-Europe

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1428.3 A Number of Growth Scenarios Developed with DREAM . . . . . 146

8.3.1 Examples of Growth Rates . . . . . . . . . . . . . . . . . . . . 1468.3.2 Growth Rates Observed for the 1985-1995 Period . . . 150

8.4 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1548.5 Regionalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

8.5.1 Comparison between the Multi-Regional Approach with theSingle Region Approach . . . . . . . . . . . . . . . . . . . . . 157

8.5.2 Regional Differences in Electricity Supply . . . . . . . . . 1618.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

Chapter 9Conclusions and Reflections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1659.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

9.2 Theoretical and Methodological Backgrounds . . . . . . . . . . . . . . 1659.2.1 Perspectives of this Thesis . . . . . . . . . . . . . . . . . . . . . 1659.2.2 Regional Input-Output Energy Analyses . . . . . . . . . . . 167

9.3 Regionalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1689.3.1 Developing a Multi-Regional ECCO-Model of OECD-

Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1689.3.2 Outcomes of the Mulit-Regional Modelling Approach . 1699.3.3 General Remarks on Multi-Regional ECCO-Approach

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1719.4 The Dynamic Resource and Economy Accounting Model . . . . . . 172

9.4.1 Developing the DREAM-Modelling Approach . . . . . . 1729.4.2 A DREAM-Model for The Netherlands . . . . . . . . . . . 1739.4.3 DREAM-Models for OECD-Europe . . . . . . . . . . . . . . 1749.4.4 General Remarks about the DREAM-Modelling Approach

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1759.5 General Conclusions and Reflections . . . . . . . . . . . . . . . . . . . . . 1769.6 Suggestions for Future Research . . . . . . . . . . . . . . . . . . . . . . . . 177

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

Appendix Model Listing and Assumptions of the ECCO-Models and theDREAM-Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Samenvatting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

11

Voorwoord

Ruim vier jaar geleden begon ik aan mijn promotietraject dat uiteindelijk resulteerdein dit proefschrift. Het schrijven van een proefschrift is echter geen eenmansactie. Ikben daarom verschillende mensen uiterst dankbaar voor hun betrokkenheid bij detotstandkoming van dit werk.

Allereerst ben ik Wouter Biesiot enorm veel dank verschuldigd. Niet alleen washij zeer bertrokken bij de voortgang van mijn onderzoek, maar met name zijnbevlogenheid heeft inspirerend op mij gewerkt. Zonder hem had ik dit proefschrift, omverschillende redenen, waarschijnlijk nooit geschreven. Helaas heeft hij de voltooiingervan niet meer mogen meemaken.

Naast de begeleiding van Wouter heb ik ook zeer veel steun gehad van Klaas JanNoorman. Hij heeft de ‘taak’ als directe begeleider zeer goed opgepakt na het groteverlies van Wouter. Sterker nog, hij is in de loop der jaren meer geworden dan eenbegeleider. Daarnaast ben ik ook Ton Schoot Uiterkamp en Henk Moll erg dankbaarvoor het commentaar dat zij de afgelopen tijd gegeven hebben. Verder heb ik ook veelbaat gehad bij het commentaar van Phil Smith. Hij raakte pas aan het eind van hettraject betrokken bij het project maar zijn commentaar was daarom zeker niet mindernuttig. Al met al kan ik met veel genoegdoening terugkijken op mijn begeleidingsteam,mede door hun inspanning is het geheel vlot verlopen. Mijn dank daarvoor.

Dank ook aan de leden van de leescommissie, prof. dr. W.A. Hafkamp, prof. dr.J. Kommandeur en prof. dr. J. Oosterhaven, voor het lezen van het manuscript.

Tevens wil ik ook de Rijksuniversiteit Groningen bedanken voor het financierenvan mijn AIO-plaats.

Natuurlijk zijn niet alleen de begeleiders belangrijk tijdens een promotietraject.Mijn collega’s bij de IVEM hebben gezorgd voor een prettige werksfeer, waardoorik altijd graag naar de IVEM ging. In het bijzonder wil ik mijn kamergenoot Renébedanken aangezien hij het laatste jaar met grote rust mijn ‘gestress’ onderging.

Voor het doen van promotie-onderzoek is een leven naast het werk enormbelangrijk aangezien men tijdens het doen van redelijk solo-onderzoek zo nu en dantoe is aan een ‘uitlaatklep’. Voor mij bestond deze de afgelopen jaren voornamelijkuit hockey en met name ‘het gedoe’ er omheen. Ik mag mezelf gelukkig prijzen dat ikin zo’n mooi team ben terechtgekomen als Men Ten. Hierdoor stond niet alleen dezondag in het teken van ‘hockey’ maar ook de dinsdag en de vrijdag. Met dezogenaamde harde kern van Men Ten, en Jan Jaap en Bas in het bijzonder, heb ikmenig uur doorgebracht in ‘t Vaatje. Het klinkt tegenstrijdig maar deze (late) uren

12

hebben me veel energie gegeven om de volgende dag weer fris aan mijn onderzoek tewerken.

Een gedeelte van die energie heb ik ook gegenereerd en ben ik meteen ook weerkwijtgeraakt tijdens het zaalvoetballen op de donderdagavond. Heren van de v.v.Folkingedwarsstraat Vooruit mijn dank voor de mooie partijen en de gezellige borrels.

Tot slot ben ik ook mijn familie zeer erkentelijk voor hun ‘support’ door de jarenheen.

CoosGroningen, maart 1999.

13

Chapter 1Dynamic Modelling of Energy Stocks and Flows in the

Economy

1.1 Introduction

In general economic activities are associated with a demand for natural capital.Generally, two types of natural capital are distinguished; renewable and non-renewable resources, referring to the difference between the ability and the rate atwhich resources regenerate. Renewable resources, such as fisheries and forestry, havethe capacity to regenerate themselves at a timescale relevant for mankind. In contrast,the timescale at which non-renewable resources regenerate is of no direct relevanceto humans. However, renewable resources can be depleted too when the regenerationrate lags significantly behind the rate at which the resource is exploited, theirsurroundings become unfavourable or the stock becomes too small. Since both typesof resources have their own characteristics, their supply and use may be restricted bydifferent boundaries or environmental limits (e.g. natural laws). As economicactivities are associated with the use of both types of resources, meeting present andfuture human demands has to take place within these natural laws or environmentallimits. This thesis mainly addresses the use of non-renewable resources and inparticular that of (fossil) energy. Peet [1992], among others, argues that energy is acritical factor for economic activities since all consumption and production activitiesrequire energy inputs.

The demand for energy resources has grown enormously in the 20th century,mainly as a result of population growth and increasing welfare levels notably in theindustrialized world. For instance, world population more than tripled and the averageglobal GDP per capita grew more than fivefold between 1890 - 1990 [KleinGoldewijk and Battjes, 1997]. As a result, the global energy use grew abouttwelvefold between 1900 and 1990 [ibid]. The availability of energy is restricted toaccessible stocks of non-renewable resources (such as fossil fuels) and of thequantities furnished exogenously on a flow basis as self-renewing fluxes (such assolar radiation) [O’Connor, 1998b]. One may question whether the availability ofthese resources constrains the (growing) global energy demand and herewith futureeconomic activity Moreover, the current energy supply system, which is dominatedglobally by exploiting fossil fuels, gives rise to a range of environmental problems,among others those associated with greenhouse gas emissions and emissions ofacidifying gasses (cf. [Meadows, 1972]).

As a result of the environmental problems related to economic activity and theassociated energy use, the concept of sustainable development has become prominentduring the last decade. The World Commission on Environment and Development(WCED, also known as the Brundtland Commission) [1987] defines it as follows.

14

Environmental Metabolism

renewal and decomposition by the environment

renewableresources

nonrenewableresources

SocietalMetabolism

decomposablewaste flows

nondecomposable

waste flows

Figure 1.1: societal and environmentalmetabolism [Moll, 1993]

‘In essence, sustainable development is a process of change in which theexploitation of resources, the direction of investments, the orientation oftechnological development, and institutional change are all in harmony andenhance both current and future potential to meet human needs and aspirations’

According to this definition sustainable development should be considered as aprocess between three subsystems: the ecological system (exploitation of resources),the economic system (investments and technological development) and the socio-cultural system (institutional changes) [RMNO, 1992]. Clearly, the WCED advocatesan integrated approach to study the economic system and the associated demand fornatural resources. This concept of sustainable development is commonly accepted asa general guideline for (economic) development planning. However, translating thisconcept into operational policies turned out to be very complicated due to differencesin status and interpretations.

Although Daly [1993] agrees on the relevance of the Brundlandt report, hecriticizes the commission’s thought that economic growth is necessary for sustainabledevelopment. Daly argues that the term growth reflects a natural increase in size bythe addition of material through assimilation or accretion while the term developmentrefers to expanding or realising the potentials, to bring to a fuller, better or greaterstate. Hence, sustainable development should be interpreted as reducing physicalthroughput. In this perspective, Daly [1993] presents three criteria for managingrenewable and non-renewable resources. First the harvest rates should not exceed theregeneration rates. Second, waste emission rates should not go beyond the naturalassimilative capacities of the eco-systems into which the waste is emitted. Third, non-renewable resources, which by definition can not be maintained intact, should be usedin such a way that the depletion rate is limited by the rate of introduction of renewablesubstitutes. In this perspective, Moll [1993] considers the sustainable use of resourcesby considering the throughput of these resources through society. In this respect, Molluses the term societal metabolism to point out the interdependency between the useof energy and materials in society and the associated environmental loads. Figure 1.1presents a simplified scheme ofthe current societal metabolism.Waste flows from society into theenvironment can be distinguishedin decomposable and non-decomposable waste flows. Flowsin the society indicateconsumption and recycling. Therenewal and decomposition ofnatural capital are also presentedin figure 1.1. The large differencebetween the rate of consumption

15

Socio-EconomicSystem

EnvironmentalSystem

Energy- Material Flows

Figure 1.2: The economy-environment interface(based on [Noorman et al., 1998])

(cross-hatched arrow within society) and the recycling rate (white arrow withinsociety), resulting in relatively large inputs of non-renewable resources flows intosociety and non-decomposable waste flows into sinks, indicates that the currentsocietal metabolism is not sustainable. Non-sustainability is addressed from aphysical point of view in this thesis.

Clearly, there are more aspects related to the sustainability concept besidesnatural resource use alone. Agenda 21, which is an international consensus-basedworking program, was produced in Rio to direct global sustainable development.Agenda 21 also acknowledges the importance of growing income disparity betweenrich and poor, population growth and environmental problems such as atmosphericpollution, deforestation, desertification and loss of bio diversity (in: [Mulder andBiesiot,1998]). Besides focussing on the capacity of the natural system as a sourcefor our wealth and as a sink for our waste, aspects such as equity, quality of life andenvironmental quality should also be included in concepts of sustainable development[ibid].

Figure 1.2 presents thelinkage between the socio-economic system and theenvironmental system in anotherway than figure 1.1 as itillustrates the relations betweentwo subsystems (based on[Noorman et al., 1998]). Withinthe socio-economic system,consumption and productionactivities take place to maximizeeconomic welfare. Moreover, thesoc io -economic sys temconsiders the notion thatmaximizing welfare depends to time-related sets of norms and values. As presentedin figure 1.1, the environment provides a biophysical basis for economic activity orbetter, economic activity can not be sustained without using natural resources. Asboth systems are linked, both environmental and socio-economic aspects should betaken into account in studying the economic system within a sustainable framework.From the above, it may be concluded that analysing sustainable economicdevelopment and the associated energy use requires an integrated approach. Such anapproach is outlined in this thesis.

Different schools of economics study the relationships between the environmentand the economy from different perspectives. Table 1.1 lists a number of theseperspectives.

16

Table 1.1: Perspectives of different schools of economics in the way the relationshipbetween the economic system and environmental system is studied.

School Perspective Focus

neoclassicaleconomics

Socio-economicSystem

economic system

resourceeconomics

Socio-economicsystem

inputs from the environmental systemto the economic system

environmentaleconomics

Socio-economicsystem

outputs from economic system to theenvironmental system

ecologicaleconomics

EnvironmentalSystem

relationship between economic systemand environmental system

Essentially, the field of neoclassical economics (or main-stream economics)studies the economic system in an isolated way (i.e. only partly taking into accountthe environmental counter parts of economic activity). Resource economics addresses,based neoclassical economic principles, the inputs from the environmental system tothe economic system [Faber et al., 1996]. Similarly, the school of environmentaleconomics studies the outputs from the economic system to the environmental system.Both the school of environmental economics and resource economics belong to theschool of neoclassical economic [ibid]. The school of ecological economics studiesthe interactions between the economic system and the environmental system from theperspective of the environmental system. One of the methodologies used in ecologicaleconomics is Resource Accounting which incorporates physical information intoeconomics models. The concepts of the Resource Accounting methodology areapplied to two energy-accounting approaches that form the main topic of this thesis.The first energy-accounting approach involves the ECCO-methodology. The mainfeatures of the ECCO-methodology are described in chapter 2. The second energy-accounting approach is the DREAM-modelling approach. This approach, which isbased on ECCO, is outlined in the second part of this thesis.

Before the concepts of the Resource Accounting approach are outlined, thischapter first describes concisely the main characteristics of resource economics, beingpart of neoclassical economics, and ecological economics. These sections are includedto address the differences in the perspectives associated with the Resource Accountingapproach, being part of ecological economics, and neoclassical economics.

1.2 Economic Perspectives

Since it is beyond the scope of this thesis to describe the concepts of economicsin detail, only a concise overview is given here of the concepts of economics whichare relevant for this thesis. Herewith, these subsections focus on the way the field of

17

neoclassical economics addresses environmental aspects and in particular naturalresources.

1.2.1 Neoclassical Economics

Samuelson [1967] defined economics as a science that studies how men andsociety choose, with or without the use of money, to employ scarce productiveresources, which could have alternative uses, to produce various commodities overtime and distribute them for consumption, now and in the future, among variouspeople and groups in society. Following this definition, the field of economics studiesthe allocation mechanisms of choosing among relative scarce or limited resources(means capable of alternative uses) in order to achieve best goals (or ends). In thisperspective, economics can be regarded as a discipline that studies the wayindividuals (i.e. producers as well as consumers) cope with relative scarce means torealise a maximum level of satisfaction. Consumers maximise utility by purchasinggoods and services on the basis of individual preferences. These goods and servicesare supplied by production sectors. These production sectors are responsible for thetransformation of raw resources into goods and services. As the amount of resourcesare limited for consumers (income type, capacities, abilities), choices among differentgoods and services are needed to achieve certain ends. Moreover, choices based onconsumer preferences are constrained by physical and technological endowments ofthe economy. The coherence between goods and services utilised by consumers whichare realised by producers coincide with allocation mechanisms which involve choiceof product and production process and spatial and distribution issues [Dietz et al.,1994].

The market mechanism is expected to balance demand for and supply of goodsand services as under perfect conditions both the consumer’s utility and theproducer’s profits are maximised. Perfect market conditions involve pure competition,complete security, complete information and uniform prices for homogenous products[Blaug, 1985]. Under these perfect conditions, an optimal consumption/productionlevel is derived which is referred to as the Pareto optimum. This Pareto optimumplays a crucial role in welfare economics as it is assumed to be a condition formaximizing welfare. The process of obtaining a Pareto optimum is referred to as theTheorem of the Invisible Hand and this concept was introduced by Adam Smith (in:[Blaug, 1985]). The market is assumed to coordinate all economic actions where costsand prices are used as indicators. Moreover, the discount rate is assumed to the reflecttime preferences or otherwise the trade-off between current and future consumptionthat consumers are willing to accept.

In the above, markets are assumed to be perfect. However, markets may beimperfect due to monopolies (no full competition), ill-defined property rights (lack ofinformation of which individual owns certain resources), or side effects which are notreflected by the market (lack of information about the costs of some factors). The

18

concept of imperfect markets is fully addressed in economics as these markets mayno longer result in Pareto optima (see among others [Samuelson, 1967; Blaug, 1985;Tietenberg, 1988, 1993; Perman et al., 1996]). Negative side effects of economicactivity that are not included in the market, and which, therefore, may not result in aoptimal allocation, are referred to as negative externalities. Ideally, these negativeexternalities should be internalized in the market. In this perspective, Pigou [1920]argued that negative effects have to be defined by authorities which allows them tomonetarise these effects. An economic tax should be used to shift these burden ofsocial costs to the polluter, restoring the optimal allocation.

Above, it is argued that the market mechanism plays a key role in describing theeconomic system as markets are assumed to balance production and consumption.With regard to production, the field of neoclassical economics has a reductionist viewof the economic system. In this perspective, Solow [1956] proposed a model ofeconomic growth which characterises the economy by a single aggregate productionfunction with two inputs, labour and capital (Y=f(L,K)). The Cobb Douglas functionis a commonly used production function.

Various economic schools have different views on the driving forces of economicgrowth. That is, do consumers determine what goods and services should be producedor do producers set the consumption patterns? In the view of the Walresians (in:[Blaug, 1985]), consumer preferences have a central place and production isdependent on the demand for consumer goods hence producers play a rather passiverole. In other words, the economy is demand-driven in Walras’ view. In addition,Keynes also refers to a demand-driven economy and emphasises the importance of theeffective demand. Keynes’ view is mainly developed from a macroeconomic point ofview whereas the theories of Walras are based on partial microeconomics [Nentjes,1983]. On the opposite of Keynes’ and Walras’, economists who revert to the workof Say start from a supply-driven economy implying that the producers are the drivingforce of the economy [van Ierland et al., 1994]. It is their willingness to invest thatdecides the development potentials of the economy.

The field of microeconomics is the study of how individual consumers andproducers behave, and how the market system allocates relatively scarce resources[Hall and Taylor, 1988]. So, the field microeconomics deals with most of theallocation concepts described above. On the other hand, the field of macroeconomicsmainly studies fluctuations in economic activity. Microeconomics andmacroeconomics should not be considered as two isolated studies as finding anexplanation for fluctuations at a macroeconomic level relies on microeconomics[ibid].

1.2.2 Resource Economics

19

Based on the principles of neoclassical economics, the field of resourceeconomics focuses on aspects relating to the effect of resource flows from nature tothe economy [Faber et al., 1996]. In Solow’s model, the capacity of nature to deliverenergy and materials and to absorb effluents generated because of economic activityis not incorporated and it is thus assumed unimpaired by economic growth. Resourceeconomists recognise that natural resources involve different optimality conditions forthe production than ordinary goods. Therefore, it is argued to expand Solow’s modelof economic growth by introducing resources as a specific factor (Y=f(L,K,R)).Physical interdependencies of the economic system and its environment only receiveattention if they are associated with prices and costs. In this perspective, the pricesystem has to include in its computation all present and future supply of resources aswell as labour and capital, and the demand of present and future generations[Dasgupta and Heal, 1969; Faber et al., 1996]

Below, some basic concepts of resource economics are listed which concern howto deal with resource use in decision processes. In resource economics, resource useand scarcity are integrated from a standard economic point of view. In this field,resource scarcity holds that resources are relatively scarce as the availability is fixedand finite at any point in time. However, where a market exists for a resource, theexistence of any positive price is viewed as evidence of scarcity. Where markets donot exists, the existence of a positive shadow price (i.e. the implicit price that wouldbe necessary if the resource were to be used economically efficiently) is an indicatorof absolute scarcity where the absolute scarcity holds that the availability of resourcesare fixed and finite at any point of time [Perman et al., 1996]. From this perspective,a perfect free market is a key condition for an optimal use of natural resources (e.g.declining resource stocks). Under perfect market conditions, the optimal harvest rateis in principle equal to the sustainable yield in the case of renewable resources, andtherefore the stock of the renewable resources would not be depleted. Heijman [1991]mentions the additional condition that the discount rate may not exceed the renewalrate. However in the case of exhaustible resources, one can not speak aboutsustainable yield. Moreover, the resource will be depleted as the use rate is positive(that is not zero). Therefore in the case of exhaustible resources, the optimal rate ofdepleting the resource is the main item. Alternatives of the Hotelling rule (see textbox) are used to determine the optimal depletion rate (see among others [Pearce andTurner, 1990]). It is assumed that even in case of exhaustible resources the outputwould not automatically decline in time as sustainable production is possible whenreproducible inputs or renewable resources inputs are substitutable for the exhaustibleinputs [ibid]. In this view, substitution and technological progress are the answer toresource supply problems.

20

The above assumes the existence of perfect market conditions. However, undermost circumstances, economic systems will not sustain efficient allocations. Marketimperfections may occur as a result of externalities such as common property rightsand imperfect market structure (e.g. monopolies) and these market imperfections mayresult in less optimal outcomes. The way producers and consumers use resourcesdepends on the nature of the property rights governing their resource use. In thisperspective, Hardin [1968] argues that common property resources are always overexploited since individuals as rational beings seek to maximise their individual gaineven though society as a whole suffers. This concept is more widely known as ‘thetragedy of the commons’. One way of avoiding ‘this tragedy’ is to replace thecommon property right system by a system of private property rights. Besides theconcept of ill-defined property rights, problems may arise as a significant number ofproduction factors are not included in the market. As these factors may influenceeconomic activities or welfare negatively, main stream economics argue that thesefactors should be internalised into the market, for instance, by levying Pigouviantaxes (see above and [Nentjes, 1983]).

Hotelling’s Rule

Hotelling’s rule is one of the oldest theocratical concepts involving the extraction ofdepletable resources [Dietz et al., 1994]. Let C(Vt) and Pt denote the extraction costs and theresource price at time step t, respectively. Moreover, define respectively Vt and Kt as theannual use and the total stock of the resource at time step t. The optimal resource extractionis then equal to the following maximisation problem when the discount rate is equal to r[Hotelling, 1931]:

MAX ( 3t [ Pt Vt - C(Vt) ]e-rt)

subject to

Kt = Kt-1 - Vt

KT $ 0 (where T denotes the end of the period)Vt $ 0

As the most simple version of the Hotelling’s rule implies that the marginal extraction isconstant (which means ( dC(Vt)/dVt = c ) one can show that the following holds forthe resource price at time step t [ibid]:

Pt = c + $ e-rt

In this case, the price of the resource increases at a similar rate as the discount rate.In addition, the Hotelling’s rule states that the present discounted value of aresource should be the same at all dates [Perman et al., 1996; Pearce and Turner,1990].

1 A clear distinction should be made between isolated systems, closed systemsand open systems. Isolated systems exchange neither energy nor matter with theirsurrounding environment while closed systems exchange energy but not matter andopen system can exchange both energy and matter with their surrounding environment[Faber et al., 1996].

21

1.2.3 Ecological Economics

While resource economics is mainly based on concepts stemming from the fieldof neoclassical economics, the field of ecological economics addresses the overallrelationship between economic systems and ecosystems [Costanza et al., 1991]. It isbeyond the scope of this overview to describe how all the environmental relatedaspects are dealt with in ecological economics. Hence, only some relevant aspectsrelated to resource use are described here.

In ecological economics, the economy is considered as an open system in the sensethat the economy withdraws resources from the environment. Because of solar inflow,the global system is a thermodynamically closed system1. This physical feature setslimits to the economic system. In this perspective, Boulding [1966] introduced twoimages of the world to underline the difference between an open and a closedeconomy. Boulding refers to the open economy as the 'cowboy economy', thecowboy being symbolic of the illimitable plains and also associated with reckless,exploitive, and violent behaviour, which is characteristic for open societies.Opposed to this view Boulding introduced the metaphor ‘Spaceship’ for the closedeconomy, in emphasizing the earth’s dependency on the limited reservoirs of naturalresources and nature’s ability to absorb waste residues stemming from humanactivity. In this perspective, mankind must find his place in the cyclical ecologicalsystem which is capable of continuous reproduction of material. However, itcannot escape from having inputs of energy. The difference between these twoperspectives on the economic system becomes most apparent in the attitudetowards consumption. In the cowboy economy, high consumption is regarded aspreferable and production likewise. Success of the economy is measured by thethroughput of the ‘factors of production’. On the other hand in the spaceshipeconomy, throughput is regarded as something to be minimized rather thanmaximized. The essential success of this economy is the nature, extent, quality andcomplexity of the total capital stock. In the cowboy economy, resources are regardedas flows whereas in the spaceship economy resources are considered as stocks.

So, from the broader ecological-economic perspective, the expansion of theeconomic subsystem is limited by its dependence on fundamental laws of nature. Thislimitation raises the question of the scale of the economy activity. Does an optimalscale exist for economic activity? Various authors (such as Daly [1993] ) havestudied this question extensively and argue that the stationary state economy is anecessity, since exponential growth in a finite world is not consistent with physical

22

laws. Going even further, Georgescu-Roegen [1993] states that the most desirablestate of economic activities is a declining one (instead of a stationary).

The concept of optimal scale can be illustrated by considering the carryingcapacity of a ship. The allocation of the cargo is naturally of great importance for thestability of a ship. However, when a ship is loaded too heavily it will become unstableeven when the cargo is allocated optimally. In maritime institutions, the absoluteoptimal scale of a cargo is recognised by the Plimsoll line [Daly and Townsend,1993]. When the waterline hits the Plimsoll line, the boat is fully loaded, it hasreached its safe carrying capacity. Of course, the Plimsoll line is touched sooner whenthe cargo is badly allocated. The Plimsoll line, however, is not one single line whichholds that there only exists one optimal scale. The Plimsoll line is a band whichindicates a relative optimal scale depending on external factors such as the weatherconditions. The concept of the Plimsoll line can be regarded as an analogy of theoptimal scale of the economy. The optimal allocation of resource flow in the economy(a microeconomic concept) does not imply an optimal scale of the whole economy (amacroeconomic problem). In addition, the relative optimal scale of economic activityis also influenced by external factors (i.e. environmental conditions). The strictmeaning of carrying capacity is used in a broader sense to evaluate the carryingcapacity of the environment related to economic activity. Carrying capacity has itsroots in population ecology and is defined as the number of people sharing a givenarea or territory who can, for the foreseeable future, sustain the existing standard ofliving (through the utilization of) energy, land, water, skill and organization [Unescoand FAO,1985]. In this sense, the major task of ecological economics is to design anindicator analogous to the Plimsoll line. However, finding an indicator for carryingcapacity of the environment is difficult as the given environment is not constant[Gilbert, 1991].

In the perspective of the optimal scale of the economy, Daly [1993] states thatsustainable growth is impossible as growth by definition implies an increase in sizeby the addition of material through assimilation or accretion. As mentioned before,Daly prefers to focus on development which holds an expansion or realization of thepotentials or bringing something to a fuller or, greater or better state. With respect tothe Plimsoll line, development refers to improve the allocation of the cargo on a shipwhile growth means to increase the cargo on a ship. Development instead of growthis the idea behind the ‘zero-growth’ concept.

Munda [1997] disagrees with the ‘zero growth’ idea and believes that the greatappeal of sustainable development is to establish a simultaneous realisation ofeconomic growth and environmental objectives. Moreover, sustainable developmentinvolves a multidimensional concept and multi-criteria analysis teaches us that it isimpossible to maximise different objectives at the same time.

These studies bring forward the need for different concepts of sustainability:‘strong sustainability’ and ‘weak sustainability’. Strong sustainability gives priority

23

to ecosystem resilience, and does not accept human-made-capital accumulation as anadequate substitute for natural capital depletion. The socio-economic systems andenvironmental systems are so interrelated that the constraints imposed by thesecomponents should be considered jointly [Harris and Goodwin, 1995]. The criticalcharacter of some natural capital justifies the introduction of non-monetary indicatorsof ecological sustainability based on direct physical measurements [Faucheux andO’Connor, 1997]. Opposed to strong sustainability, ‘weak sustainability’ requires themaintenance of natural capital and human-made capital with the implicit assumptionof infinite substitution possibilities over time [Turner,1998] and herewith involvesless stringent constraints.

Resource Accounting is the field of ecological economics which attempts toexpand conventional economic information with physical principles imposed by theenvironment. This might, at first sight, seem somewhat repetitive. After all, thePhysiocrats already recognized the important role of nature for the economy whereassome classical economists as Ricardo and Malthus (in: [Blaug,1985]) took naturalresource scarcity (notably agricultural land) into account in their analyses. TheResource Accounting approach also deals with ‘physical’ facts. From thisperspective, one can argue that what is not physically possible cannot beeconomically possible. This evolving trend of integrating physical aspects intoeconomics has let to an increasing (renewed) interest in the concept of sustainabledevelopment. In this sense Slesser [1991] states that the role of Resource Accountingis to identify the physically possible within which the economically feasible mustoperate by the laws of nature.

Within the field of Resource Accounting, energy analysis focuses on studyingrelationships between the economy and energy supply system as energy is a criticalproduction factor. Georgescu-Roegen [1971, 1986] was among the first to stress theconnection between physical principles and economic processes using the second lawof thermodynamics. This law holds that in an isolated system, the available energycontinuously and irrecoverably degrades into unavailable states. Herewith Georgescu-Roegen related economic activity with ‘entropy production’. As energy is essential foreconomic activity it may become a limiting factor. Within the economic theory,thermodynamics involves resource scarcity and physical limits on technologicalimprovements. Integrating the role of thermodynamics in economics is elaborated insection 1.2.4.

Although energy-based valuation may appear a newcomer in the economy, it hasalready been applied for quite some time. Martinez-Alier [1987] lists scientificresearch in which the use of energy is included in economics in the period 1865-1950.By doing so, he demonstrates that the evaluation of energy use in the economy is nota prerogative of the last quarter of this century.

From the above, it can be seen that ecological economists reject the reductionist

24

view of mainstream economists by accepting the notion that the economy is part ofa complex larger system. The acknowledgement of the complexity of the largersystem implies that the system can no longer be depicted by simple linearrelationships (cf. [Prigogine, 1976]). Just as most biologists would admit thatecosystems, species and individual organisms are all significant levels of naturalreality, economists should face the fact that social classes, corporate organizations,and the state are levels of economic reality that cannot be reduced to atomistic,optimizing individuals [Simon, 1981]. Next to incorporating the environment in thestudies, the role of socio-psychological aspects should be regarded. These items aretaken into account by some institutional economists who focus on actors, their worldviews, habits and the institutional arrangement. The latter term refers toorganisations, rules, power relationships, entitlements and other types of control overresources. Ecological economics and institutional economics share the recognition ofthe impossibility of a value-free science, emphasis the distribution of property rights,and the strong criticism of monetary reductionism (cf. [Aguilera-Klink, 1994;Opschoor and van der Straaten, 1993]). Institutional economics involved inenvironmental concepts mainly focus on the question of how essential ecologicalknowledge can be incorporated into economic theories [Dietz and van der Straaten,1992]. In this sense, institutional economics can be exemplified by itsinterdisciplinary orientation and its tendency to regard values, technologies, andinstitutions as endogenous [Söderbaum,1990; Opschoor, 1991].

1.2.4 Thermodynamics and the Economic System

Thermodynamics addresses the conservation of quantity of energy and the changein quality of energy in a system. The first law of thermodynamics states that theamount of energy is conserved in a closed system. The second law of thermodynamicstates that the entropy, which is a measure of disorder or unavailable energy will notdecrease in a isolated system. This law implies that all physical processes, naturaland technological, proceed in such a way that the availability of useful energyinvolved decreases. So, energy consumption means that the availability of usefulenergy is consumed and not energy itself. Hence, non-renewable energy carriers,such as coal, oil and natural gas, can only be used once. The increase of entropy ina isolated system makes time an important factor in studying the system as it relatesto the state of the system. The second law of thermodynamics limits the efficiency atwhich energy can be transformed and, thus, imposes restrictions on growth of theeconomic system.

According to Ruth [1993], physics has been employed in economic theory forquite some time, as well in the form of analogies as in the form of principles. Forinstance, the use of thermodynamic concepts for explaining economic conceptsencouraged the recognition that the laws of thermodynamics set upper bounds to theefficiencies of material and energy transformations. Among others, Kneese et al.

25

[1970], Boulding [1976], Faber and Proops [1985], and Perrings [1987] alsoacknowledged the importance of thermodynamics and incorporated its concepts intothe theoretical scheme of economics. On the other hand, some economists ignore thesignificance of thermodynamics and for instance Samuelson [1972] even stated thatthere is nothing more pathetic than to have an economist try to force the analogiesbetween the concepts of physics and economics.

Although some economists acknowledge the importance of thermodynamics withrespect to economic activity, the concepts of thermodynamics are still not widelyincorporated into the concepts of economics. England [1997] poses three reasons forthe failure of evaluating the economy from the perspective of thermodynamics:1. Recognition of thermodynamics would force economics to take history more

seriously as the entropy law induces a distinct orientation in time. Herewith,economics have to focus more on historic rationality.

2. Long-term economic models may become obsolete as thermodynamic principlesreduce the freedom for future development.

3. Thermodynamics appears to contradict the optimistic belief in economic progressheld by modern economists.Especially the third reason seems appealing as it parallels the doomsday thinking

that environmentalists are sometimes accused of. The first two arguments reflect thenotion that past use of resources may limit future use and therefore may constraineconomic growth.

Thermodynamics generates an additional view on economic activity as it relatesthe economic system with the natural system. In economics, processes are valued inmonetary terms. However, money is an abstraction; that is money has no physicalvalue but its value depends on the people’s set of beliefs and confidence. Monetaryvalue is the opinion of the seller and of the buyer, which reflects their preferences.Opinions may change and hence the price of goods and services may change.Furthermore, money can only be used to manage and organise the process ofproducing goods and services. It is not money itself that provides the physicalmaterials. Or in other words, money is cycled within the economic system (fromconsumers to producers and vice versa) and as a consequence using money as the solevariable in economics brings with it the danger of ignoring the physical reality ofgoods and services. From a physical perspective, it is possible to quantify the amountof energy or materials required to change the physical state of the input materialsfrom their initial to a final state. This quantification is the keystone of ResourceAccounting and is outlined in section 1.3.

26

EnergyResources

Raw resources

Consumption Production

E.T.S.

Pollution

$

$

Goods and services

Neoclassical Economics Resource Accounting

Figure 1.3: Gilliland’s [1978] model of thelinkage between the economy and theenvironment. Solid lines represent physicalflows and dashed lines represent monetaryflows. The dotted lines indicate the scope ofthe two disciplines neoclassical economics andResource Accounting. E.T.S. stands for theenergy transformation system.

1.3 Resource Accounting Approach

1.3.1 General Concepts

Resource Accounting is the field which aims to expand conventional economicinformation by including information on the use of flow and stock resources from theenvironment [Wright, 1989]. The long-term aim of this approach is to identifychanges in the sustainability of national economies or economic sectors and, ifnecessary, to take corrective actions [Peet, 1992]. Resource Accounting consists ofdescribing, in terms of physical throughput, the use of natural resources and the stresson the environment as a result of consumption and production processes. The physicalthroughput can be expressed in energy as well as other resources. This thesis focuseson valuing production and consumption in terms of energy since current economicactivities depend heavily on the use of non-renewable energy resources. Thisapproach is referred to as energy-accounting and it is based on physical conservationlaws, and thus generates an additional perspective (next to economic analyses) onlong-term strategies and assessments of various options for sustainable development.

Resource Accounting ingeneral enables the study ofoptions to reduce the use of naturalresources and the correspondingenvironmental stress from anintegrated perspective. Naturalresources do not determine whathumans can and can not do, butthey do set important constraints,as humans cannot overridephysical constraints. Thus,Resource Accounting linksexplicitly processes in the economywith processes in the environment.In Resource Accountingmethodology, the economic systemis embedded in the broaderbiophysical system. Thebroadening of the system can beillustrated by the model of Gilliland [1978] (see figure 1.3).

Gilliland focuses on resources in general and on energy in particular as energy isregarded as a key factor in the economy for three reasons. The first reason hasalready been mentioned and relates economic activities to the constraints set by thelaws of thermodynamics. The second reason is that energy is potentially a goodindicator of physical flows in the economy as available energy is expended in the

2 It should be noted that Peet only refers to energy-accounting in the case of theschool of Slesser. He uses the term systems ecology for the school of Odum. However inthis thesis, energy-accounting is used in a broader sense than Peet does.

27

Structure

Facts

Design

OE

MB DF

SD

OFBF

Legend:MB: material balance

descriptionSD: system-dynamical

modelOE: optimal

equilibrium model

DF: dynamic flow modelBF: blue print for the

futureOC: Optimal control

model

Figure 1.4: Taxonomy of models for long-term environmental studies [Moll, 1993]

production of goods and services. Third, energy use and especially fossil fuel use isclosely related to pollution. Hence, determining energy flows provides insight into theeconomic structure as well as in the environmental stress associated with the energyproduction.

Peet [1992] discusses two schools within the energy-accounting methodology. Theschool of Odum follows the maximum power principle which holds that in a systemof strong competition among alternatives, the surviving alternative is the one thatmaximise their energy throughput [Odum, 1994; Peet,1992; Slesser et al., 1997].Within the approach, all forms of energy are linked by relating them to itsfundamental form (i.e. the sun). Despite some criticism, Odum’s approach has shownsuccess when it is applied to ecological systems but it seems less appropriate forapplying it to the economic system [Peet, 1992]2. The second school Peet refers tois led by Slesser and involves the ECCO-methodology. The ECCO-methodology iselaborated further section 1.4 and in chapter 2.

1.3.2 Models

In the Resource Accountingapproach, models play animportant role in investigating thecomplex relationship between theeconomic and the environmentalsystem. The linkage betweenthese two systems can be studiedfrom different perspectives andtherefore various types of modelsare designed and applied inResource Accounting.

Moll [1993] presents ataxonomy for models aimed atinvestigating the long-termenvironmental problems related tomaterial use and energyconsumption which, in turn, arethe results of economic activity(see figure 1.4). In thisperspective, the variation in

28

Resource Accounting models is represented by this cube. Different models can beplaced in a three dimensional space defined by three axes: facts, structure and design.Models placed near the facts-axis are mainly developed to describe and explainaccurately the use of energy or flows of materials in time (e.g. material balance orenergy balance descriptions). Models near the design-axis are mainly intended todevise a structure for future society derived from normative principles (e.g. blueprintsfor tomorrow). Models that are situated near the structure-axis are developed toanalyse the structure and theories which determine and predict changes in energy useor material flows (e.g. system-dynamic models). It is beyond the scope of this thesisto describe all possible combinations in the space set by these axes.

In essence, system-dynamic flow models that integrate structural aspects withfactual aspects are of special interest as dynamic energy-accounting models which arethe main subject of this thesis involve system-dynamic or dynamic flow models) andare thus on the line DF-SD). These models focus on behavioural aspects of the systemin order to present a holistic rather than a reductionist view on the system [Slesser etal., 1997]. So by using these types of models, one tries to cope with the complexityof the system by focussing on the significant causal influences. Feedback loops anddynamics such as delays and accumulation of stocks are important features. Ryan[1995] discusses the important aspects of the balance between complexity andsimplicity. Simple models are easier to handle. However, one must be careful foroversimplification. Ruth [1993] addresses another important aspect about system-dynamic models by stressing the essential role of the system boundaries. Since inthermodynamics, system boundaries are necessary because thermodynamics isconcerned with the change of systems properties between alternative end states of welldefined closed or isolated systems. Moreover, Costanza [1980] stresses that thechoice of system boundaries is important as it distinguishes net inputs from internaltransactions. Net inputs are independent and exogenous whereas interaction areinterrelated and endogenous. For instance in an open economy, where the energy issolely regarded as an input, energy plays a minor role in the output (gross nationalproduct). However energy can also be regarded as endogenous since the economy ispart of the closed environmental system. In this way the interdependencies of energyand economy are considered.

System-dynamic models aim at gaining understanding the functioning of theeconomic system and its linkage to the environment. This type of models contrastswith econometric models that aim at predicting the major economic indicators overa relative short time period on the basis of carefully analysing historical data[Werbos, 1990]. Nijkamp [1987] addresses various limitations of these system-dynamics models. Nijkamp, among others, stresses that the integration of policymodels and institutional configurations in such models is usually poor, thebehavioural characters of many of these models are fairly limited and the majority ofthese models generate conditional pictures of the evolution of a sector, but fail toprovide reliable predictions on solid statistical/econometric techniques. In this

29

perspective, O’Connor [1998a] refers to the system-dynamic approach as being partof structural economics which is introduced to study the process of economic andecological resource management aiming at the jointly delivery of economic andecological benefits.

1.4 ECCO-Modelling Approaches

Within the framework of energy-accounting models, a number of ECCO-models(Enhancement of Capital Creation Options) have been developed that use energy asa numeraire to study economic activity and its impact on the environment [Noorman,1995; Ryan, 1995; Slesser et al., 1997]. The ECCO-modelling approach is used tostudy the interdependency between economic activities and the available naturalresources. Moreover, it takes into account the notion that economic activity may belimited by the availability of human-made capital stock as human-made capital stockis also required to produce goods and services in order to sustain or enhance thecurrent standard of living of the population. Human-made capital stock comprises allfixed capital and is the result of past labour, energy and material input. The ECCO-modelling approach can be characterised as a dynamic energy-accounting approachthat links the production of human-made capital to the natural capital that physicallyenables a given production level. The economy is described by a physical model thataddresses the conversion of raw material by means of energy into goods and services,taking into account the fundamental physical laws [Noorman, 1995]. In principle, theECCO-modelling approach is situated on the line determined by points DF and SDin the ‘Moll cube’ (see figure 1.4). However when more normative features are addedto the ECCO-approach, it can also move up the Design-axis. A more detailedoverview of the concepts of the ECCO-modelling approach is presented in chapter2.

In the ECCO-modelling approach, resources are generally quantified through theenergy required to release them and to produce any good or service in the economy.Inputs and outputs associated with all activities are thus expressed in energy terms,based on the concepts of energy analyses [IFIAS, 1974; King, 1991]. It describes theeconomy in terms of energy stocks and flows. Moreover, ECCO is a dynamicsimulation modelling approach which quantifies feedback among the different sectorsin physical terms. In addition, the original modelling approach, as introduced bySlesser, is supply-driven which holds that the level of economic activity is determinedby the production sectors. Feedback loops between investments and industrial output,therefore, form the main influences of the modelling approach. This supply-drivenapproach resulted from the aim to study the potentials of an economy to grow undercertain (energetic) constraints. The ECCO-methodology has been developed over theyears and one might now distinguish three generations. 1 The first generation of ECCO-models is introduced by Slesser. These models may

be characterised of having overshoot and collapse patterns. The models mainly

30

concerned developing countries. In a later stage also ECCO-models were built forindustrial countries. By using these models, the potentials were studied by whichdeveloping countries are able to develop their economy to a higher level of wealthunder certain environmental constraints. By incorporating aspects of ResourceAccounting and the carrying capacity assessments, it is argued that ECCO is arealistic method of measuring the attainable, testing development programs andthus defining for each region its capacity to develop in an appropriate way[Loening, 1991].

2 Noorman [1995], Crane [1995] and Ryan [1995] introduced the secondgeneration of ECCO-models as they introduced the idea of the double sets ofaccounts which distinguishes the utility level of an output from the energy valuein terms of real energy consumption. The system of the double sets of accountscan be compared by distinguishing output in terms of real (or constant) dollarsfrom nominal (or current) dollars (see also chapter 2). The introduction of thedouble sets of accounts resulted in more realistic outcomes. Moreover, the ECCO-models have a less forced character of having an overshoot and collapse pattern.The models of the second generation all involve industrial countries and aremainly used to study the environmental consequences of economic growthassociated with energy efficiency improvements and the substitution of a non-renewable energy supply by a renewable energy supply.

3 The third generation of the ECCO-modelling approaches involves modellingchanges introduced in this thesis. Two major changes are advocated here. First,it describes the energy flows at a multi-regional level in order to address theconsequences of open economies. Second, a demand-driven modelling approachis introduced in order to study scenarios which are closer to changingconsumption patterns in relation with changes in the production processes. Thisapproach is more consistent with the notion that consumers play a key role indriving the economy. The latter change brings about a considerable alternation ofthe key feedback loops of the ECCO-model. Hence, the latter model is referred toby a different name (i.e. DREAM, Dynamic Resource and Economy AccountingModel).

1.5 Scope of this Research

The ‘third generation’ of model changes referred to in section 1.4 are motivated bythe following arguments. Imports and exports are becoming increasingly importantfor national economies in the world and certainly since the free-trade regulationsbrought about by the WTO. Moreover, the introduction of the euro will mostprobably increase trade in Europe in the coming decades. In addition, liberalisationof energy markets in Europe will most probably increase international energy flowsin OECD-Europe. A proper treatment of imports almost unavoidably necessitates theintroduction of regional models in which regional differences are, by definition, dealt

31

with. The necessity of developing a regional model also applies to the ECCO-modelling approach as the industrial structures and the energy supply and demanddiffer from country to country. Most ECCO-models only involve studies at a countrylevel. Although some theoretical research has been done on regional ECCO-models,no regional model has yet been published that is based on an existing (‘large’) regionMaybe with the exception of a pilot study on the EC [Slesser and de Vries, 1990].Therefore, the regional ECCO-model of OECD-Europe, which is described in thisthesis, forms a next step in this methodology. Braat et al. [1987] identify groups ofconcepts which span many regions and therefore require multi-regional approaches.Two of these groups comprise the concepts of shared-markets and shared-resources.The concepts corresponding of imports can be assigned to the former.

Current ECCO-models are all supply-driven which means that one assumes thatthe consumption level is determined by producers. This assumption is consistent withthe macroeconomic ideas of the theory of Keynes which holds that production is thedriving force of the economy but it is in contrast to the microeconomic point of theWalresians who state that the consumers are the driving force of the economy. It isincreasingly recognised that consumers or households play a key role in driving theeconomy. Hence, it is argued that one should assign energy costs associated withproduction activities to consumption as most goods and services eventually end up asconsumer goods. In this way, the energy costs and the related environmental stresscan be determined for different scenario assumptions about future consumer activity. 1.6 Research Goals

The above indicates that the energy flows associated with economy should bestudied at a multi-regional level as the economies involved are open implying thatimports contribute considerably to economic activity. Moreover, it is argued thatconsumers play a key role driving in the economy and that this notion should also becovered in the ECCO-modelling approach. This thesis addresses these two topics: inthe first part, the role of imports is addressed by developing a regional the ECCO-modelling approach for OECD-Europe. The DREAM approach (a demand-drivenversion of the ECCO-modelling approach) is presented in the second part to study thedifferences between a supply-driven model and a demand-driven model. Hence, theresearch topics are aimed at answering the following questions:

1 Do the results from the multi-regional modelling approach differ from those ofthe so-called ‘one region’ modelling approaches? In order to study this topic,the results are compared of two ECCO-modelling approaches. The firstinvolves a multi-regional ECCO-model in which OECD-Europe is divided into6 subregions. The second consists of an ECCO-model in which OECD-Europeis regarded as one large region.

32

2 What are the consequences for the results of scenarios developed with theDREAM modelling approach compared to that of the ECCO-modellingapproach? Or in other words, do the outcomes of a supply-driven model differsubstantially from those of a demand-driven modelling approach? Thesedifferences may be the results of some drastic changes in the key feedbackloops of the ECCO-modelling approach in order to realise the shift towards aDREAM-model.

1.7 Overview of this Thesis

A general overview of the ECCO-methodology is presented in chapter 2. Input-output analyses form the base of an ECCO-model. The role of imports in (regional)input-output analyses are discussed in chapter 3. The single and multi-regionalmodels of OECD-Europe are described in chapter 4. Scenario results to study theimpact of regional aspects are presented in chapter 5. Chapter 6 bridges the first andthe second parts of thesis. Chapter 7 describes the main differences andmethodological aspects of the demand-driven type of model called DREAM. Inaddition, a demand-driven version of the Dutch ECCO-model [Noorman, 1995] ispresented to illustrate the potentials of the DREAM-modelling approach. Chapter 8describes the behaviour of the DREAM-model of OECD-Europe and compares thescenario results with the corresponding ECCO-model. In addition, chapter 8addresses regionalisation of the DREAM-model of OECD-Europe. Finally, generalconclusions and reflections are listed in chapter 9. Table 1.2 presents a schematicoverview of this thesis.

Table 1.2: Schematic overview of this thesis.

Part I Chapters 2-5 Comparison between the multi-regional ECCO-model of OECD-Europe and the single region ECCO-model for OECD-Europe

Part II Chapters 7and 8

Comparison between the single region DREAM-model of OECD-Europe and the single region ECCO-model for OECD-EuropeCase study of the single DREAM-model for The NetherlandsComparison between the multi-regional DREAM-model of OECD-Europe and the single region DREAM-model for OECD Europe

Chapter 9 Conclusions and reflections

33

Chapter 2General Concepts of the ECCO-Modelling Approach

2.1 Introduction

This chapter describes the general concepts of the ECCO-modelling approach asdeveloped in the past. Before describing these main concepts, it is necessary tointroduce a number concepts which form the basis of the ECCO-approach and areassociated to energy analysis and system-dynamics.

In chapter 1, the ECCO modelling was introduced as a dynamic energy-accounting approach which considers the interdependencies between economicactivity and the available physical resource base. The approach focuses on therequirement of energy to produce human-made capital that in turn is required tosustain and expand economic activity. Therefore, all economic activity is expressedin terms of energy by using the concept of embodied energy. The basis lies in thedetermination of the energy content of economic activity, or more specifically, theenergy content of a sector’s output. Section 2.2 outlines the basic principles ofdetermining the embodied energy content of a sector’s output or a product. Theembodied energy content of a product or a certain amount of output can be reducedthrough energy savings. In chapter 1, it was stressed that all economic activity isassociated with the use of energy. So, the processes of making energy carriersavailable (i.e. mining and refining) also require energy. This notion, which is referredto as the Energy Requirement for Energy (ERE), is addressed in section 2.3. Inaddition, section 2.3, describes the most relevant aspects related to energy savings.When the energy content of a product changes for instance as result of energysavings, it is necessary to distinguish the energy costs of a product from the utilityvalue. Both are defined in chapter 2.4. The above shows that analysing economicactivity by means of the energy costs involves time-related issues such as energyefficiency improvements. Therefore, the ECCO-modelling approach determines theenergy required to produce goods and service in a system-dynamic way. The majorconcepts of system-dynamics are presented in section 2.5. The ECCO-modellingapproach itself is outlined in section 2.6.

2.2 Embodied Energy

In the ECCO-modelling approach, all goods and services are expressed in termsof the energy requirement to produce that good or service. This total amount ofenergy is referred to as gross energy requirement or the embodied energy of aproduct. The embodied energy of products are determined according to theconventions set by the IFIAS [1974]. The embodied energy value of a product isdefined here as the amount of primary energy resources sequestered from the earthin the process of making a good or service. When calculating the embodied energy

34

of a product, the system boundary must embrace all resources at the point when theyare still in the ground, oceans or air [TNO, 1976]. In other words, the energyrequirements must be assessed that make all inputs available, untill each input istraced back to the energy resource in its original state.

The quality of energy is a very important aspect in evaluating economic activityin terms of energy as, for instance, 1 MJ of electricity has more potential to do workthan 1 MJ of coal does. But in principle, most fuels can be separated into two qualitygroups: electricity and hydrocarbons (e.g. coal, oil and natural gas). In thisperspective, indicators such as ‘free energy’, ‘exergy’ and ‘negentropy’ areintroduced to depict the ability of energy to do work instead of depicting the totalamount of available energy (mostly expressed in terms of the heating value of theenergy carrier (e.g. MJ)). Moreover, Ryan [1995] discusses the pros and cons ofthree types of energy analysis all including different energy forms depending on thepurpose of the analysis: commercial energy analysis, solar energy analysis and fossilenergy analysis. Commercial energy analysis involves commonly traded fuels, suchas coal, oil, gas, biomass and electricity, that can be used directly or indirectly in theeconomy. Commercial energy analysis only focuses on human use and mainly in thesense of technological and economic development. Solar energy analysis mainlycomprises Odum’s eMergy concept. This concept is already described briefly inchapter 1. The approach seeks to value both the transactions associated with moneyand the contributions from nature involved with those transactions by concentratingon the solar energy [Odum, 1994]. The (solar) eMergy evaluates the units of workdone to make a product. Work is defined here as the available energy qualitativelydegraded in an energy transformation. In this sense, eMergy is a measure of (solar)energy use in the past [ibid]. Solar energy flows are indirectly limited by theavailability of land. In fossil energy analysis, the use of depletable energy sourcesis stressed. Within this approach, economic output is expressed in terms of primaryfossil embodied energy.

A characteristic difference between solar energy analysis, fossil energy analysisand commercial energy analysis involves the determination of the system boundaries.For instance, the system boundaries of the fuel use of a car is different among thethree approaches. Commercial energy analysis only studies the fuel use in terms ofgasoline. The fossil energy analysis would also include the fossil energy use requiredto make gasoline available (i.e. energy required for mining and refinery). In the solareMergy concept a link is made between economy and ecology by assigning fossil fuelresources to past solar radiation. Table 2.1 summarises the main characteristics ofthe three types of energy analysis.

35

Figure 2.1: Schematic overview of the sector’sinput and output in terms of energy. The blackarrows represent flows of goods and services interms of energy and the white arrow represents thedirect input of energy.

Table 2.1: Main characteristics of the three types of energy analysis (based on [Ryan,1995])

numeraire system

commercial energyanalysis

secondary energycontent

economy

fossil energy analysis primary (fossil) energycontent

economy + environment

solar energy analysis (past) solar radiation economy + environment + solarradiation

The ECCO-modelling approach focusses on the fossil energy requirementassociated with economic activity. It addresses to what extent the availability of fossilenergy requirement may limit economic activity. Thus, the ECCO-modellingapproach belongs to the category of fossil energy analysis. The energy content ofgoods or services equals the fossil energy embodied in that product.

As indicated above, the fossil energy required to produce goods and services notonly includes the energy that is used directly to produce the goods and services (e.g.electricity, oil and natural gas). Italso incorporates the energyrequired to produce the goods andservices which are used as inputfor producing that good or service.Obviously, the same principlesalso apply to the sector level thatis, the embodied energy (energycontent) of the output of a sector isequal to the total energy content ofall the inputs of that sector. On aaggregate level, besides direct fueluse, the sector’s input consists ofgoods and services which arep r o d u c e d d o m e s t i c a l l y(intermediate deliveries) and ofimports (cf. figure 2.1). Theconcept of embodied energyapplies to the key principles of input-output analysis since the output of a sector isequal to the total inputs (see [Miller and Blair, 1985; Bullard and Herendeen, 1975;Costanza, 1986; Nieuwlaar, 1988; Wilting, 1996]). Consequently embodied energyis a measure of the total energy required to produce goods and services in a sector.For the base year, the embodied energy values are computed with the aid of enegy

36

input-output analysis. This procedure is described in detail in chapter 3.In chapter 1, using energy as an indicator for economic activity was justified as

it links the economic system with the environmental resource base. Slesser et al.[1997] state that another advantage of using energy as a numeraire is that itfacilitates the prediction of the value of future productivity. Their statement evolvesfrom the notion that economic valuation (in monetary terms) fails to fully address thephysical conservation of goods and services produced. Only part of the conventional‘price’ is related to physical aspects of producing these goods and services (i.e. themonetary value of energy and the material inputs). The monetary value of goods andservices also includes non-physical aspects such as human preferences andperceptions. Changing perceptions and preferences might (drastically) change thevalue in monetary terms without corresponding changes in physical features. Inparticular, the physical features of producing goods or services are constrained bysome physical laws. Naturally, products can be made more efficiently but there arelimits to improving the energy efficiency (see text box Carnot principle). On theother hand, depleting fossil fuel reserves may increase the amount of energy requiredto make fossil fuels available.

2.3 ERE-Values and Energy Savings

From an energy perspective, there is no such thing as ‘a free lunch’ which impliesthat the ‘production’ of energy also requires energy. This notion is clearly illustrated

Carnot Principle

Carnot showed that the efficiency with which work (energy available for ‘use’) can be extracted from a heat source is restricted by the following equation:

(T - T0 )/ T In this equation T represents the temperature of the heat source and T0 thetemperature of the environment. Both temperatures are expressed in degreesKelvin.

For example, work can be generated at a maximum efficiency of 20% froma heat source of 100 oC (373 K) in an environment of 25 oC. (298 K)

Generating electricity from fossil fuels by combustion is a good example ofan energy conversion process constrained by the Carnot principle. Hence, thereare upper limits for the efficiency by which electricity can be generated fromcoal, oil or natural gas.

37

by Gilliland’s model (see figure 1.3) and it is also addressed in the first part of thissection. In addition, this section deals with the possibility that production processescan become more energy-efficient.

The amount of primary fossil energy that is needed to make one unit of energycarrier available is referred to as the Energy Requirement for Energy or the ERE-value and it is defined as the ratio between the sequestered energy used to deliver anamount of energy divided by the amount of energy delivered [IFIAS, 1974]. In otherwords, the ERE-value of a fuel (EREfuel) is equal to the gross energy requirement(denoted by GERfuel) of that fuel divided by amount of energy delivered by that fuelor the combustion value (denoted by CVfuel) (see equation (2.1)):

As the gross energy requirement (GERfuel) also includes the energy delivered by thatfuel (CVfuel), the ERE-value is always larger than one in case of the IFIAS definition.

As mentioned above, only the fossil energy requirements are taken into accountin the ECCO-modelling approach involved. The ERE-value as defined by IFIASincludes all energy carriers. The definition of the ERE-value has to be adjustedslightly in order to only consider fossil energy requirement (GERfossil). The fossilenergy requirements for energy (denoted by EREfossil) is defined as:

In equation (2.2), CVfuel refers to the energy delivered by the energy carrier. Aspresented in equation (2.3), the gross fossil energy requirement (GERfossil) refers tothe fossil energy sequestered in that fuel, which states that the energy delivered onlyconsiders the combustion value of fossil fuels (CVfossil). Hence, CVfossil is equal toCVfuel in the case of fossil energy carriers and CVfossil is equal to 0 in the case of non-fossil energy carriers (e.g. nuclear and solar energy). Moreover, ECEfossil representsthe fossil energy costs for energy that is the amount of fossil energy required toproduce CVfuel. The fossil energy costs for energy includes all the direct energy inputsas well as the indirect inputs, that is the energy needed for mining, conversion,distribution and for making the capital stock required are all taken into account. Forexample, the major energy inputs needed to make coal or natural gas availablecomprises the (direct) energy use for mining and transport, distribution losses and theconsumption of capital stock required for mining and transportation. The directenergy use required for mining may decrease as result of new technology which isoften referred to as the learning-by-doing effect. On the other hand, the energy

EREfuel = GERfuel / CVfuel (2.1)

EREfossil = GERfossil / CVfuel (2.2)

GERfossil= ECEfossil + CVfossil (2.3)

38

required for mining may increase when resources are becoming depleted, as the moreeasily accessible resources are obviously depleted first. It is more complex todetermine the ERE-value of oil as oil is converted from crude oil to various refinedoil products. Besides fossil fuels, also renewable energy sources and nuclear energyare regarded in ECCO, albeit in a different manner. Renewable energy, among others,involves wind energy, solar energy, energy from biomass, etc. Only, the fossil energyinputs are regarded in the case of renewable energy sources as each activity isexpressed in fossil primary energy use in ECCO (CVfossil = 0). This means that theERE-value of renewable energy is less than one whereas the ERE-values of fossilfuels are by definition larger than one (as CVfossil = CVfuel).

Mulder and Biesiot [1998] acknowledge the dynamic character of an ERE-value,as the value not only changes as result of learning-by-doing and depleting resourcesbut it also depends on, among other things, the fuel mix of electricity generation,especially when the fuel mix changes from a fossil based supply to a renewablesupply. In most recent ECCO-models, the energy costs due to mining are setexogenously. A more dynamic or endogenous way of computing the energy requiredfor mining is introduced in chapter 4.

Energy savings (at a sector level) are often the result of energy efficiencyimprovements which means that a certain output can be produced with less energy,implying that the total energy inputs decrease. These efficiency improvementscomprise changes in the physical system. Although energy efficiencies have improvedconsiderably over the last years, physical upper bounds exist for these improvements(cf. Carnot principle). Energy savings induced by efficiency improvements arereferred to as efficiency effects. In addition, energy savings can also be accomplishedby structural changes at a sectoral level. These changes or substitutions are, forinstance, the result of shifts in the social economic system that influence the energyuse within a (aggregate) sector (e.g. changing production due to shifting consumptionpatterns). Structural changes may have huge impacts in ECCO models as theindustrial sector is considered at a rather high aggregation level. There might be someoverlap between efficiency effects and structural changes effects as the boundariesbetween social economic and physical economic systems are not always clear [Hiltenet al., 1996]. Hence, energy savings refers to a broad range of options to reduce theenergy use;

• substitution of raw materials• more efficient use of raw materials• increase in productivity of labour or capital • improvements in energy savings at the end-use• shifts in production or consumption patterns

Energy efficiency improvements at the end-use implies reducing the direct energy

39

Figure 2.2: Primary energy use per unit of GDP inkilogram of oil equivalent (kgoe) per 1990 US$.Derived from [Nakiƒenoviƒ, 1996]

use per unit output. These efficiency improvements can, for instance, be realised byusing more energy efficient appliances. Substitution of material inputs (both raw andintermediate) can reduce the energy contents of product when energy intensivematerials (such as aluminum) are substituted by more energy extensive materials(such as steel) [Moll,1993]. In addition, the energy content of products decreaseswhen materials are used more efficiently. For instance, the energy requirements torecycle aluminium are much lower than to produce it from raw resources [ibid].Increasing the productivity of labour and capital holds that the output increases perunit labour and capital input. Hence, the input output ratio decreases which impliesthat more can be done with a certain amount of (energy) input. Changing productionor consumption patterns means that other goods and services are produced andconsumed which may involve a lower energy content of the overall production-consumption chain.

The development of energy intensities is often used as an indicator for thedevelopment of energy savings (and the ‘state of development’ of an economicsystem). The energy intensity of a sector is defined as the total (direct) energy use ofa sector divided by total output of that sector expressed in monetary terms (cf.[Bullard and Herendeen, 1975]). At a national level, annual energy savings can beindicated by dividing the total energy use of the production sectors by the nationalincome (GDP). Nakiƒenoviƒ [1996] gives an overview of the developments in energyintensities at a national level for a number of countries for the period 1855 - 1990(see figure 2.2). Figure 2.2 shows that the energy intensities decreased strongly fora number of countries. On average, energy intensities decreased at a rate of 1% peryear at a global level [ibid], which is about the same as in the US between 1890-1990[WEC and IIASA, 1995]. In scenarios developed for integrated assessment models,assumptions on global futureenergy intensity improvementsvary from 0.4% to 1.9% per year[EMF, 1997; Morita and Lee,1997].

2.4 Utility versus Real EnergyUse

When a sector’s output isexpressed in terms of energy andthe corresponding productionprocesses become more energy-efficient, a distinction should bemade between the utility value ofproducts and the (real) energy

40

costs of that product. The former involves the amount of useful products whereas thelatter refers to the energy costs of producing that product. For instance, when due toenergy savings the embodied energy content of a car is reduced from 100 to 50energy units, 100 energy units no longer represents the energy value of one car butof two cars. As a result, the energy required to produce that car no longer representsthe utility value (i.e. the car).

The necessity of distinguishing the utility output from the real energy output canbe dealt with by taking into account two values for production. The first one consistsof the real energy content of output or the actual energy costs of the production andrepresents the energy requirements of all inputs. The second value consists of theutility level of the output (e.g. number of cars produced or amount of cropsharvested). The utility value can be determined by expressing the output in so-calledconstant energy terms that is the energy content of a certain product output as if itwas produced in the initial or base year. Or in other words, the energy content of theoutput is calculated without incorporating changing energy requirements. In this way,a double set of accounts is introduced for determining the energy content of a certainoutput.

The concept of considering two sets of accounts is not only applied to energyanalysis. In the field of economics, it is also common practise to consider two sets ofaccounts to deal with inflation. The two values described above can best be comparedwith the concept of ‘current’ and ‘constant’ dollars which is common in the field ofeconomics.

Noorman [1995], Crane [1995] and Ryan [1995] introduced the concept of doublesets of accounts in the ECCO-methodology as they were the first to make,independently, a clear distinction between the utility value and the real energy contentof the output in ECCO-models.

2.5 System-Dynamics

As mentioned before, the ECCO-modelling approach can be considered as asystem approach in the field of energy analysis. The ECCO-methodology belongs tothe system-dynamic modelling approach which describes the trajectory of variablescharacterizing the structure and evolution of the economy in relation to itsenvironmental aspects [Nijkamp, 1987].

System-dynamics was developed to address problems encountered by managersin corporate systems [Forrester, 1961]. Its use has been extended to includeeconomic, social, biological and physical systems [Radzicki and Sterman, 1994].From a mathematical point of view, system-dynamic models consist of sets ofdifferential or difference equations providing more or less complex mathematicaldescription of the main processes of a system [Jeffers, 1987]. System-dynamicmodels can be characterised as structural, disequilibrium, behavioural models.Herewith, they differ from familiar econometric models or general equilibrium models

41

[Radzicki and Sterman, 1994]. System-dynamic models are formulated in terms ofthe relationships between stocks, flows, queues, decision rules and influences.

Stocks characterise the states of a system, while flows represent the rates ofchange of stocks. In system-dynamic models, the decision processes of agents areaddressed by including the way people response to certain circumstances whichemerge from the assumptions about the system’s structure and interaction offeedback loops [ibid]. Feedback loops exist when the outcome of a variable is directlyor indirectly corrected by or related to the variable itself [Sontag, 1990]. In order to get more feeling with the features relating to system-dynamics, asimple example is presented of how the energy content of output is described by themeans of system-dynamics. Assume that oil is extracted from an oil well (stock) andused to produce a car (stock), the oil resources decrease (flow) with the increasingcar stock (flow). The car-producing process is represented in terms of flows. Theflows as well as the stocks are all expressed in terms of energy. The flows are relatedby feedback loops. For instance, the ERE value of oil increases when wells aredepleting, implying that more oil is required to extract a certain amount of oil andthus more oil needs to be extracted. Moreover, the energy supply sector may need tobe expanded as more capital may be needed to extract oil. A decision rule to producemore energy-efficient cars will have consequences on the fuel use of cars and thus onthe oil demand.

2.6 General Concepts of ECCO

Above, some of the general concepts regarding energy analysis are discussed tostress the importance of energy use in the economic system. Moreover, the aboveshows that economic activity can be expressed in terms of embodied energy and thatthe primary energy requirement of a certain amount of output has a dynamiccharacter and therefore the ECCO-methodology involves a system-dynamicapproach. Below some of the main features of the ECCO-model are presented. Thesefeatures apply to the ECCO-modelling approach described by Noorman [1995] andRyan [1995]. The modifications that play the central role in this thesis are discussedin chapters 4 and 7.

2.6.1 History

Originally, the ECCO-modelling approach was developed by Slesser at therequest of UNESCO to study issues of population growth, development of materialstandard of living (which may be seen as an indicator for material wealth) and thecarrying capacity of the environment in developing countries. Pilot studies were madefor countries such as Kenya, Mauritius, and Thailand (cf. [Owino,1991; Baguant andSlesser, 1991; Sintunawa, 1991]). In these first studies, the ECCO acronym stoodfor Enhancement of Carrying Capacity Options as the main goal of these models was

42

quantifying the concepts of carrying capacity in the context of sustainabledevelopment. At a later stage ECCO-models were also designed for developed (orindustrial) countries such as the UK [Slesser et al., 1994; Crane, 1995], NewZealand [Ryan, 1995], and the Netherlands [Noorman, 1995]. These models aimedat studying physical limitations on long-term economy-environment questions froman energy perspective. The same natural capital accounting method is used in modelsfor both developing countries and developed countries, yet the emphasis shifted frominvestigating the carrying capacity of a developing country’s environment to enablea growth of wealth for a growing population, to studying the ability to transformnatural capital into human-made capital in order to sustain a given production andconsumption level in a developed country. This shift was induced by differences inthe available resource base, environmental conditions, economic structure andpopulation issues between developing and developed countries [Noorman, 1995].Changing the emphasis to the transformation from natural capital to human-madecapital can be justified by the fact that human-made capital (capital stock such asfactories, infrastructure, means of transport, etc.) is essential to produce output.Natural capital provides the materials and energy required to produce human-madecapital.

2.6.2 Production from an Energy Perspective

Production can be viewed from a micro, meso, or macro perspective. From amicro level, the field of neoclassical economics focuses on the role of capital andlabour in describing production activities. In this context, Heyman [1991], amongothers, noticed that one might say that neoclassical economists have a reductionisticview on the production function:

In equation (2.4), y indicates the production value whereas K and L refer tocapital and labour, respectively. The Cobb-Douglas function ( y = K"L1-" where 0 <" <1) is the production function most commonly known. These production functions,however do not reflect that economic activities also depend on the physical inputssuch as materials and energy. In chapter 1, it was argued that the physicaldependency of the economy should also be taken into account. In order to internalizethe inputs of materials (M) and energy (E), the production function of equation 2.4is extended to equation (2.6) which is also known as the Generalised Cobb-Douglasfunction [Perman et al., 1996; Tietenberg, 1988]. The generalised Cobb-Douglasfunction views production from a micro to meso perspective.

y = f(K,L) (2.4)

y = f(K,L,E,M) (2.5)

43

In energy analysis, the energy requirements of output are determined by summingup all energy inputs. Slesser and King [1988] introduced a production function inwhich all inputs are traced to labour and energy. In their production function, theydistinguish past and present inputs (equation (2.6)). In this perpective, Dürr [1994]stresses that capital is not a production factor but an organisational factor and it thuscan be left out of the production function.

In equation (2.6), production is represented as a function of operational labour(L), past labour (L’), operational energy (E) and, past energy (E’). Operationalenergy use involves the direct energy use of a production process whereas the latterholds the energy used in that past to produce the goods and services required asinputs for the production process. The same idea also applies to the input of labour.In this representation, the production function is shifted from a micro/mesoperspective to meso/macro perspective.

Energy is the most fundamental factor of production from a physical point ofview. Even though labour itself plays an important role in production, the physicalcontribution of labour is negligible. Hence, equation (2.6) can be reduced to equation(2.7).

According to equation (2.7), production can be expressed in terms of direct (E)and indirect (E’) energy use. Indirect energy use can be considered as the cumulationof past (direct) energy use required to enable a certain output. In the ECCO-modelling approach, the economy is described form this perspective. The basicprinciple of this methodology is that a sector requires capital to produce any output,where capital stock represents means of production such as factories, machinery andother equipment. In addition, the utility output of a sector is assumed to beproportional to the capital stock of that sector which holds that the load factor isassumed to be constant. For instance, the number of cars produced is doubled if thecapital stock of the car industry is doubled. Capital stock itself is the result of pastmaterial and energy use (E’) also known as investments. Investments are essential forgrowth as the output does not increase without an increasing capital stock. So, theavailability of energy not only influences directly the level of production as energyis required in the production process but also indirectly as it may limit the room forinvestments. The room for investments is calculated differently for each sector. Anumber of sectors can be distinguished in the production sector; agriculture, industry,energy supply sector, transport, market services, and non-market services. Theenergy sector can be separated into number of subsectors (see 2.6.4). Moreover,

y = f(E,E’,L,L’) (2.6)

y = f(E,E’) (2.7)

44

ExportsFinal Consumption

IntermediateDeliveries

Investments in Non- Industrial SectorsIndustrial

Investments

IndustrialOutput

Figure 2.3: Example of allocation of industrialoutput

public transport and freight transport can be dealt with individually. The room forinvestments in the industrial sector forms a crucial factor in the ECCO-modellingapproach as it is the balancing term of the model and therefore it influencesultimately the growth rate of the total economy. The main influences are describedbelow.

In the ECCO-modellingapproach, the total output of asector is equal to the total input ofthat sector. The approach isconsistent with the concepts ofIO-analysis. The inputs comprisethe total (direct) primary fuel use,the embodied energy content ofintermediate deliveries, capitaldepreciation and imports. Asmentioned above, a sector’soutput in terms of utility isproportional to the total amount ofcapital stock in that sector andthus the inputs (i.e. primary fuel use, intermediate deliveries and imports), in termsof utility, are, by definition, also proportional to the sector’s capital stock. In bothcases, the capital stock is also expressed in terms of utility. The extent of the inputflows in terms of utility is then converted into terms of real energy value by takinginto account the effects of energy savings and changing ERE-values. In the ECCO-modelling approach, the ERE-value and energy savings are mostly set exogenouslyand are varied in different scenarios.

The output of a sector is subsequently distributed over the items: domesticdeliveries, exports, final consumption and investments. Domestic deliveries comprisethe flows of goods and services to other domestic production sectors. Finalconsumption refers to the consumption by households and governments. Naturally,the distribution differs per sector. In case in of the industrial sector, it is assumed thatthe output that is not used for deliveries to any sector (i.e. total domestic intermediatedeliveries originating in the industrial sector), exports or for investments in non-industrial sectors is available for (final) consumption or investment in the domesticindustrial sector. Intermediate deliveries also include deliveries to the industrial sectoritself. These deliveries from the industrial sector to the industrial sector illustrate thatthe embodied energy concept involves double accounting, as a part of the outputs isalso regarded as inputs. Figure 2.3 illustrates an example of the allocation ofindustrial output.

The distribution of the balancing term over the entries consumption andinvestments depends on the level of wealth which is calculated by determining the

45

CS-industryInvestments Depreciation

Output

Consumption

Fraction invested

Wealth

Investments in other sectors

Exports

Deliveries to other sectors

+

Figure 2.4: The key feedback loops of the ECCO-model. Solid lines indicate physical (i.e. energy)flows and dotted lines indicate influences. CSstands for capital stock

material standard of living. In thisway, the room for investments isinfluenced by the total capital stockof the industrial sector (positivefeedback loop) and by the level ofwealth (negative feedback loop). Inother words, more capital stockresults in more room forinvestments whereas a higher(demand for) wealth results in lessroom for investments. Both keyfeedback loops of the ECCO-modelling approach determine theallocation mechanism of theindustrial output and they areillustrated by figure 2.4.

2.6.3 Consumption

Besides production, consumption is obviously also an essential factor in economicsystems. In principle, the specific dynamics of both consumption and productionshould be determined simultaneously. As already mentioned in the preceding chapter,the ECCO-modelling approach is supply driven which means that consumption levelsand patterns are mainly determined by the production sector. This point of view isconsistent with the macro economic perspectives of the school of Keynes implyingthat the industry is the driving force of the economy as it is their willingness to investthat decides economic development paths (in: [van Ierland et al., 1994]). Also in theECCO-modelling approach the consumption patterns are set by the productionsectors. It is assumed that consumers purchase all goods and services which aresupplied by the production sector. The aggregation level involving consumption issimilar to that of production.

The level of consumption which is supplied by the industrial sector depends on aspecific demand for a material standard of living in which the material standard ofliving can be regarded as an indicator for (material) wealth. The material standardof living is expressed in terms of utility and comprises consumption of goods andservices, direct energy use (such car fuel, electricity and natural gas for heating), andthe depreciation of capital such as dwellings and private cars. The demand formaterial standard of living per capita depends on the rate of investments in theindustrial sector. Herewith, consumption growth is coupled to industrial growth orin a broader sense to economic growth.

46

2.6.4 Energy Supply System

In the ECCO-modelling approach, the energy supply sector is described in moredetail compared to other sectors. In general, mining, conversion and distributionaspects are dealt with separately at a level of a fuel type. Similar to other sectors,capital stock is essential to produce output (energy in this case) in the energy sector.Without sufficient capital stock the energy supply may decrease. Such an event hasan enormous impact on other economic activities as energy is considered to be anessential input. The formation of capital stock by investments plays, therefore, animportant role in transforming the energy supply from a fossil fuel based supply intoa renewable one. Mulder and Biesiot [1998] illustrate clearly these events bysketching two possible transition scenarios (see figure 2.5). The first transitionscenario involves a scenario in which there is sufficient energy to enable the shiftfrom a fossil energy supply into a renewable energy supply. In this scenario, thetransition has a smooth character. In the second option, the fossil fuel reserves are alimitation. In this scenario, not enough was invested in renewable energy capital toenable a smooth transition. Hence, the energy supply drops drastically. The samewould of course apply to economic activity.

Electricity generation forms an important part of the energy supply sector.Electricity is generated from fossil fuel as well as from nuclear or renewable energy.In OECD-Europe, the ERE-value in fossil energy terms of electricity ranges from 0for Iceland to 3.5 Greece depending on the fuel mix and to a lesser extent on theefficiency with which electricity is generated [OECD, 1991a-b]. The amount ofcapital required to meet the electricity demand depends on the total capacity of powerplants. This is influenced by the desirable electricity output and the load factor ofwhich the latter is defined as the relative time that a power plant is operational.

Figure 2.5: Two possible transition paths towards a renewable energy supply. The left figurerepresents a scenario in which fossil energy reserves are sufficiently available and thereforethis scenario involves a smooth transition. The right graph depicts a scenario where asmooth transition is hindered by limited reserves and therefore the energy supply drops asthe reserves run out and the capacity of renewables falls short. All units of time and energyare chosen arbitrarily. Based on Mulder and Biesiot [1998].

47

Energy cannot only be made available domestically but it can also be imported.The energy transport costs are determined in the energy supply sector and depend onthe country of origin. The extent to which imports should be allowed to contribute tothe total energy supply of a country may be limited if a country does not totally wantto rely on imports.

2.6.5 Imports and Exports

Imports and exports are important factors in an economy and especially in anopen economy. Hence, imports and exports should be dealt with sufficiently in theECCO-modelling approach.

One of the major problems concerning imports is determining the energy value ofthe imports. In most existing ECCO-models, the energy content of imports arecomputed under the assumption that the energy content of foreign products is equalto that of a similar domestic product which is consistent with the methodology ofinput-output analysis. This assumption may introduce errors in the calculations asall calculations have taken place at a sectoral level. This issue is outlined more in thenext chapters as it is one of the major subjects of this thesis.

Similar to main stream economics, an import-export balance is introduced in theECCO-modelling approach as it is believed that it is not a sustainable situation whena country relies substantially on net imports. Hence, exports are coupled withimports to balance trade somewhat. Industrial exports may limit the room forinvestment when too much of the industrial outputs have to be exported as result ofbalancing trade (see also figure 2.3).

2.7 Summary

In this chapter, the general concepts of the ECCO-modelling approach aredescribed briefly. In the ECCO-modelling approach, all economic activity is expres-sed in terms of energy use (embodied energy). The ECCO-modelling approach canbe regarded as a (system) dynamic input-output analysis in which sectoral outputdepends on the sector's capital stock. Two sets of accounts are used to distinguish thereal energy costs of output and the utility value of that output. Growth of the(economic) system is mainly determined by the room for investment in the industrialsector which is assumed to be the balancing term of the model. Consumption levelsare set by industrial growth. A detailed overview of the (regional) ECCO-modellingapproach is presented in chapter 4.

3 This chapter is based on [Battjes et al., 1998; Battjes and Noorman,1998]

4 Note that figure 3.1 is similar to figure 2.1

49

Sector

Imports

Intermediate Deliveries (domestic origin)

Energy(direct)

Production

Figure 3.1: Schematic overview of the sector’sinput and output in terms of energy. The whitearrow represents the direct input of energy and theblack arrows represent flows of goods and servicesin terms of energy.

Chapter 33

Regional Input-Output Analysis for OECD-Europe

3.1 Introduction

As mentioned in chapter 2, input-output energy analysis is used tocompute the initial energy flowsin OECD-Europe. Not only dothese energy flows comprisedirect energy use (e.g. the use ofcoal, oil products, gas andelectricity) but also indirectenergy use: energy required forthe production of the goods andservices purchased to be used inthat sector. Figure 3.14 indicatesthat the energy content for thetotal production in a sector isequal to the combined energycontent of goods and servicespurchased and of the energydirectly used. This total is referred to as the embodied energy content for the totalproduction of that sector (cf. chapter 2). Embodied energy intensities (expressed inMJ/US$) are often used to calculate the total energy content of deliveries from onesector to another as the embodied energy intensity of a sector equals the total amountof energy (direct as well as indirect) required to produce one unit of economic outputin that sector. Input-output analysis (IO) appears to be an appropriate method forcomputing embodied energy intensities (denoted by EEI) because IO provides asystematic and all-inclusive framework in which indirect energy use is taken intoaccount [Costanza, 1986]. This methodology has been outlined in several studies[Bullard and Herendeen, 1975; Costanza, 1986; Proops, 1988; Nieuwlaar, 1988;Wilting, 1996].

Ideally, imports should be included in the embodied energy intensities. They are,however, frequently neglected (e.g. [Nieuwlaar, 1988; Gowdy and Miller, 1991;Common and Salma, 1992; Hetherington, 1994]) because the data required to

50

calculate the energy value of these imports are generally not available for mostexporting countries. Naturally, this problem does not concern a sector’s exports asthese exports are part of the total production in that sector and thus, by definition,included in the calculations of that production.

In addressing the energy content of imports, the energy intensities of the importedproducts are often assumed to be similar that of domestically produced goods andservices [Office of Technology Assessment, 1990; Wilting, 1996; Wyckoff andRoop, 1994; Noorman, 1995]. This assumption may introduce errors in thecalculations as energy intensities vary from country to country. Different countrieshave different economic structures and levels of technology. This chapter addressesthe impact of these differences in assessing the EEI of a sector’s output or moreparticular on the EEI of imports. Differences in the electricity generating systemsform a striking example. In countries such as France and Belgium, the generation ofelectricity depends heavily on nuclear energy, whereas in Norway most electricity isgenerated in hydro electric installations. As mentioned in section 2.6.4, in OECD-Europe, the fossil ERE-value of electricity ranges from about 0 for Iceland to 3.5 forGreece [OECD, 1991a-b]. In Iceland, all electricity is generated in hydro orgeothermal plants, while in Greece electricity is mostly generated from thecombustion of lignite. To illustrate that differences in national energy intensitiesexist, the energy intensities are compared at a sectoral level for a number ofcountries. But first a general overview is presented in what way the energy intensitiesare computed.

3.1.1 General Methodology of Determining Energy Intensities with IO-analysis

As mentioned before, input-output analysis is used to compute embodied energyintensities. The general concepts of the input-output methodology are described inthis section.

Let Xi denote the total production of a sector and let Yi denote the final demand ofproducts originating from sector i. Final demand includes final consumption, changesin stocks, gross fixed capital formation as well as exports. Let zij denote theintermediate deliveries from sector i to sector j. If for each sector, total productionequals the sum of intermediate output and final demand, one obtains the followingequation:

œ i; Xi = Yi + 3j zij (3.1)

Note that column totals (3i zij) of the Z-matrix represent the total domestic inputs(or costs) of goods and services in that sector. From the Z-matrix, a matrix of inputcoefficients can be constructed which represents the domestic costs incurred per unitoutput. The matrix of input coefficients, denoted by A, is computed by dividing each

5 The matrix of input coefficients (aij) is calculated by the following equation:œ i,j; aij = zij /Xj

51

industrial input by the total production5. Let I be the n x n identity matrix. From equation (3.1), it follows that:

X = (I - A)-1 * Y (3.2)

In this equation, (I - A)-1 is referred to as the Leontief inverse and determines thedirect and indirect inputs of a final demand.

The Leontief inverse is frequently used to compute the embodied energy intensities(denoted by ,) [Bullard and Herendeen, 1975; Miller and Blair, 1985; Peet, 1991;Wilting,1996]. If one assumes that energy is conserved in the production of goodsand services (i.e. the energy content by the total production of a sector is equal toadding up the energy contents of its domestic intermediate inputs and imports, andits direct energy use), the following equation can be obtained for each sector:

œ j; 3i (,doj * aij * Xj ) + 3c 3i (,c

j * bcij * Xj ) + ddo

j * Xj = ,doj *Xj (3.3)

bcij represents the input originating in sector i and country c

required to produce one unit output for sector j.ddo represents the domestic direct fossil energy intensity.,do

j represents the domestic embodied energy intensity ofsector j and ,c

j the embodied energy intensity of sector j incountry c.

In equation (3.3), it is assumed that for each sector imports are proportional toproduction (bc

ij * Xj). Ideally, the domestic embodied energy (,do) should be computed by means of

equation (3.3), that is by regarding the embodied energy intensities of imports at acountry level (,c). However, the embodied energy intensities of the imports specifiedby country are generally unknown. Therefore, one often assumes that the EEI ofimports are equal to that of domestic products. This assumption results in thefollowing equation:

,do = ddo * (I - A - B•)-1 , where B• = 3c Bc (3.4)

The embodied energy intensities that can be computed by means of equations (3.3)and (3.4) do not include the capital required to produce these goods and services.According to Casler [1983], ignoring the flow of embodied energy in capital inputsresults in significant underestimations of energy intensities because approximately20% of all energy used in the production processes is consumed in the production of

6 The sector classification, with the conversion from the NACE-CLIO groupindicated in parentheses, is as follows [van der Linden and Oosterhaven, 1995; Greenand le Grontec, 1976]:1 agriculture, hunting, forestry and fishery [011/030]; 2 Mining and quarrying [120,131, 132, 133, 140, 151/152]; 3 Food, beverages and tobacco [411, 412, 413 414, 423,424, 428, 429]; 4 Textile, wearing apparel and leather industries [431/439, 441, 442,451, 453/456,]; 5 Wood and wood products [461/467]; 6 Paper and paper products [471,472/473]; 7 Chemical and chemical petroleum, coal, rubber and plastic products[252/259, 260, 481/483]]; 8 Non-metallic products except those from sector 7 [231/239,241, 242, 243/246, 247, 248]; 9 Basic metal industries [211/221, 222/223, 224]; 10Fabricated metal products, machinery and equipment [311/316, 321/328, 330, 314/347,

52

capital goods. Similarly, Wilting [1996] and Noorman [1995] found that neglectingthe embodied energy in depreciated capital results in an underestimation of Dutchenergy intensities of about 15% for both 1985 and 1987.

There are no detailed data available for the (energy) input of sector i in the totalconsumption of fixed capital in sector j (CFCj). It is assumed that the relative inputof sector i to the capital consumption of sector j (denoted by cij) is equal to therelative input of the sector i to the gross fixed capital formation of the reference year(i.e. GFCFi / 3i GFCF, where GFCF denotes the gross fixed capital formation). Thenew ‘technology’ matrix is now obtained by acij = (zij + cij )/Xj. In equations (3.3) and(3.4), matrix A should thus be replaced by matrix AC where capital is included.

The embodied energy intensities can be computed with the resulting technologymatrix and the energy intensities involved in the direct consumption of fossil fuelfuels within any particular sector. In making this calculation, one assumes thatmonetary transactions in the IO-table are proportional to the corresponding physicaltransactions. Since there are large differences in the purchasing prices of differentsectors, this assumption is certainly not valid for deliveries from the energy sectors.To resolve this problem, the use of fossil fuels in terms of primary energy iscomputed by multiplying the final energy demand per sector per energy carrier by theERE-value of that energy carrier [Wilting, 1996]. The total ERE-values arecomputed by multiplying the ERE-values of extraction by those of conversion andtransportation. The ERE-values of extraction are assumed to be constant for eachcountry and are derived from Noorman [1995]. The direct fossil energy intensities inthe energy sectors are set at zero to avoid double counting [Van Engelenburg et al.,1991; Wilting, 1996; Noorman, 1995].

Eurostat [1992a-e] only presents direct energy use in monetary terms and thus notin physical terms as required in the calculations described above. The energystatistics presented by the OECD [1991a] are used to compute the direct energy useof the 15 sectors for which the direct energy intensities are computed. The sectorclassification which is presented in Eurostat [1992a-e] is converted to that of theOECD6 [1991a-b]. The use of two different data sets based on different

351/353, 361/365, 371/374]; 11 Other manufacturing industries [491/495]; 12Electricity, gas and water [151/152, 161, 162, 163]; 13 Construction [205/509, 620, 671,672]; 14 Transportation, storage and communication [721/7, 761/764, 771/773, 790]; 15Services [93a/c, 94a/c 95a/c, 96a/c, 97a/c 660, 710, 830, 840, 850, 99 981/984].

53

Figure 3.3: The direct and embodied energyintensities for the commercial sector arepresented in terms of primary fossil energyuse per US$. Data involves the year 1985except for The Netherlands (1987).

Figure 3.2: The direct and embodied energyintensities for the industrial sector arepresented in terms of primary fossil energyuse per US$. Data involves the year 1985except for The Netherlands (1987).

classifications may give rise to errors in the calculations as the sector classificationsmay involve some inconsistencies. Besides the conversion of the sector classification,the high level of aggregation may also introduce errors in the calculations. The latterproblem has been fully addressed in several studies (e.g. [Blin and Cohen, 1977;Fisher, 1986; Miller and Blair, 1985]).

3.1.2 Comparison of National Energy Intensities

The embodied energy intensities of France, Germany, Italy, Ireland, TheNetherlands, Spain and UK are calculated using equation (3.4). The IO-tablesrequired are taken from Eurostat [1992a-e] and from van der Linden [1999] andOosterhaven and van der Linden [1995], and the direct energy intensities derivedfrom the statistics of the OECD [1991a]. All calculations are based on theperformance of 15 different sectors in 1985, except for The Netherlands, where thecalculations are based on 1987 statistics, as Eurostat does not present consistent datafor 1985. However, Wilting [1996] shows that the embodied fossil energy intensitiesdid not change much on average at an aggregate level between 1985 and 1987.

Figures 3.2-3.3 show the direct and the embodied fossil fuel derived energyintensities (or in short the fossil energy intensities) of the sectors services and theindustry sector encompassing all industrial sectors. Figures 3.2-3.3 show that boththe direct and embodied energy intensities vary from country to country.

54

direct and embodied energy intensitiesagriculture

direct embodied

countryfra ger ire ita net(87) spa uk OECD

0

10

20

30

40

Figure 3.4: The direct and embodied energyintensities for the agricultural sector arepresented in terms of the primary fossilenergy use per US$. Data involves the year1985 except for The Netherlands (1987).

Figure 3.5: The direct and embodied energyintensities for the transportation sector arepresented in terms of primary fossil energyuse per US$. Data involves the year 1985except for The Netherlands (1987).

Both figures show the differences in energy intensity in the industrial and theservices sector for the various countries within OECD-Europe. The relatively highenergy intensity of the Dutch industrial sector is remarkable. The Dutch chemicalsector (oil refining included) shows a very high embodied energy intensity of 43MJ/US$ compared to the average energy intensity of 13 MJ/US$ for OECD-Europe.The energy intensity of the chemical sector raises the direct and embodied averageenergy intensities of the Dutch industry sector [OECD, 1991a, 1993a]). About 63%of the energy directly used by Dutch industrial sectors is used in the chemicalindustries. This is more than twice the average share of 30% for OECD-Europe. Asmight be expected, the direct as well the embodied energy intensity of the commercialsectors is rather low.

Figures 3.4 and 3.5 show the energy intensities of the agricultural and thetransport sector. The relatively high energy intensities of the Dutch industrial andagricultural sector are remarkable. The direct energy intensity of Dutch agricultureis about equal to that of the Dutch industry. This energy intensity reflects the highlevel of mechanization in Dutch agriculture: The energy use of greenhouses inhorticulture and the intensive use of fodder and fertilizers in cattle-breeding (cf.Battjes [1993]). For obvious reasons, direct energy intensities constitute a large partof the embodied energy intensities in the transportation sector of each country.Differences between countries are not necessarily the result from structuraldifferences, as the energy use related to private car use is included in the energystatistics of the road transportation sector presented in OECD [20]. To overcome thisproblem, it is assumed that the use of gasoline and LPG is associated with private caruse, while the use of diesel is assigned to public road and freight transportation. Inmaking this distinction, the total amount of energy used for road transportation thatis, in fact, used for private transportation can be estimated by combining the statistics

55

Figure 3.6: Schematic overview of trade amongcountries within one region. The grey and blackarrows represent foreign trade with countriesinside and outside the region, respectively. Thewhite arrows indicate domestic trade. The outerellipse represents the boundary of the region.

of OECD [1991a] and of OECD [1991b].

The energy intensities presented above also include imports. The latter iscomputed under the assumption commonly used that the EEI of imports equal thatof domestic products. However, the figures shown above indicate clearly that thisassumption is generally not valid. For instance in case of the Netherlands thisassumption will most likely result in an overestimation of the EEI of imports. Hence,computing the EEI of imports by using average values of domestically producedgoods and services may introduce errors.

Foreign trade is illustrated in figure 3.6. In this figure, arrows indicate traderelations among three countries, together forming one region. Imports include tradefrom countries within OECD-Europe (grey arrows of figure 3.6) as well as tradefrom countries outside OECD-Europe (black arrows of figure 3.6). Ideally, theembodied energy intensities of all imports are computed by taking into account theenergy intensities for each country of origin. This exercise requires an interregionalinput-output table in which data on the intermediate deliveries from each sector of acountry to the sector of any other country (i.e. outside as well as within the region)are available. Such an interregional input-output table is not yet available for OECD-Europe.

Sections 3.2 and 3.3 presentdifferent estimation procedures fora number of countries withinOECD-Europe. These proceduresshould be regarded as a first step incomputing the EEI of imports moreaccurately. Two additionalapproaches (denoted by ‘OECD’and ‘Region’) to assess the EEI ofimports are introduced below andare compared with the approachcommonly used (denoted by‘domestic’). In the first (singleregion) approach (‘OECD’), theEEI of imports are assumed to beequal to the average value ofOECD-Europe. In the second(mul t i - r eg iona l ) approach(‘Region’), the EEI of importsoriginating within OECD-Europe (grey arrows of figure 3.6) are determined byconsidering the energy intensities of the subregion. In addition, the EEI of the otherimports (black arrows of figure 3.6) are assumed equal to average value of OECD-Europe. In the three approaches presented, all calculations are performed at a level

56

of 15 sectors.

3.2 A Second Single Region Approach for Assessing Energy Intensities of Imports

Taking into account the prominent role of imported goods and services, thequestion arises whether the use of the EEI of the aggregated OECD-Europe region(see figure 3.6) leads to a more accurate estimate of the EEI of imports of an OECD-Europe member state than the use of domestic energy intensities alone would provide.The latter assumes that the EEI of imports are equal to the EEI of the domesticcountry and it is introduced in section 3.1 (see equation (3.4)). The former assumesthat the EEI of imports are equal to the average values of OECD-Europe and thisapproach is outlined in this section.

3.2.1 Methodology of the single region approach of OECD-Europe

As mentioned above, the first single region approach involves the assumption thatthe energy intensities of imports are equal to that of domestically produced goods andservices (which holds that ,c

j = ,doj for each country see equations (3.3) and (3.4)).

In the second option, the EEI of the imports for each country (,cj in equation (3.3))

are assumed to be equal to an average value that is calculated for the aggregatedregion (in this case study the relevant region is OECD-Europe and this option is,therefore, denoted by ‘OECD’). Equation (3.5) results from equation (3.3) (whichholds that ,c

j = ,OECDj for each country c).

,do = ddo * (I - AC)-1 + ,OECD * B• * (I - AC)-1 (3.5)

As mentioned before, imports include trade from countries within OECD-Europe(grey arrows of figure 3.6) as well as trade from countries outside OECD-Europe(black arrows of figure 3.6). Lack of data concerning imports from certain countrieswithin OECD-Europe as well as from countries outside OECD-Europe influences theconstruction of equation (3.5). Similar to the adjustments related to equation (3.4),the technology matrix AC of equation (3.5) represents the technology matrix adjustedfor capital.

An IO-table of OECD-Europe is required in order to compute the Leontief inverseof equation (3.5). Such a table is not yet available. Therefore, the technology matrixis estimated by using the consolidated IO-table of a number of countries in theEuropean Union as a starting point [Van der Linden and Oosterhaven , 1995]. Theso-called RAS-method is used as the estimating procedure [Bacharach, 1970; Millerand Blair, 1985]. The technology matrix of OECD-Europe is determined by thefollowing equation:

(AC +B•)OECD= R * (AC0 + B•0)* S (3.6)

57

Matrix (AC0 + B•0) represents the technology matrix used as starting point (seeabove), where ACo and B•0 represent the cost incurred per unit output by bothdomestically produced and imported goods and services (capital included). Both(diagonalised) matrices R and S are computed iteratively and are the result ofsuccessively adjusting the row and columns of matrix (AC + B•) until the row andcolumn totals of the matrix Z = (AC + B•) * XOECD finally approximate the observedrow and column totals of OECD-Europe. For a detailed outline of the RAS-procedure, see Bacharach [1970] and Miller and Blair [1985]. By following thismethod, it is assumed that the technology matrix of OECD-Europe deviatesminimally from the technology matrix derived from the consolidated IO-table of theEC (presented by Van der Linden and Oosterhaven [1995]).

The use of the RAS-method for estimating the input-output table of OECD-Europe introduces errors in the energy intensities. Although the row and columntotals of the estimated Z-matrix approximate the original row and column totals, theelements of the estimated Z-matrix can deviate considerably from the target matrixof intermediate deliveries. This problem is also addressed in Miller and Blair [1985].They present a case study in which the elements of an estimated Z-matrix deviatestrongly from the target matrix of intermediate deliveries. However, these large errorsappear to have much less impact on the Leontief inverse associated with this matrix,a consequence which is of more concern in our case.

3.2.2 Assumptions

The row and column totals which are required to carry out the RAS-method arecomputed by using the data of total production, imports, final demand and valueadded. The required data are not available for all countries of OECD-Europe.Therefore, a number of assumptions are made in order to estimate the input-outputtable of OECD-Europe. These assumptions are listed below.! The production figures are derived from the statistics provided by the UN [1991],OECD [1995] and Eurostat [1992a-e]. Production data for Belgium and Luxembourgwere estimated by using the average ratio between the value added and the productionin the same sector of other OECD-countries. The same method is applied to a numberof sectors of Greece (sector 1, 13-15), Portugal (sector 2) and Turkey (sector 1, 2,13-15). ! For each country, the totals of final consumption expenditure of both householdsand governments (PC and GC), gross fixed capital formation (GFCF) and changesin stocks (ChSt) are derived from OECD statistics [OECD, 1993a]. The finaldemand of OECD-Europe is computed by converting all data into (1985)US$ (WorldTables [1992]). For most countries, the sectors of origin are not known for the finaldemand. The column of change in stocks (ChSt) is added to the column of finalconsumption expenditure (PC and GC). The RAS-method is now applied to the Z-matrix ((AC + B•)OECD * XOECD) and to the two final demand columns (final

58

consumption expenditure & changes in stocks and gross fixed capital formation). Aseparate treatment of exports to countries outside OECD-Europe and which are alsopart of the final demand is described below:! The total export of each sector is determined by using the foreign trade statisticsof commodities provided by the OECD [1992]. For the purpose of this study, eachcommodity is allocated to its sector of origin. Trade in commodities does not involvetrade in services, transportation and construction. It is, therefore, assumed that thetransportation, services and construction sectors of OECD-Europe are not exportingto the rest of the world. Note that only exports to countries outside OECD-Europeare taken into account (the black arrows of figure 3.6) as the intra-regional trade(grey as well as white arrows in figure 3.6) are considered to be ‘domestic’intermediate deliveries. Another problem concerning the trade statistics concerns theinternal consistency of import data: for instance, imports into country A from countryB that are given by country A not always match the export figures (from country Bto country A) presented by country B. Boomsma et al. [1991] state that theseinconsistencies are often caused by the differences in the valuation of the imports andexports. These inconsistences may take place for a variety of reasons.! Data on value added per sector are derived from statistics provided by the OECD[1993a], the UN [1991] and Eurostat [1991a-e]. The value added figures containedin these sources deviate slightly for a number of sectors. ! Although total imports are included in the matrix of intermediate deliveries,imports specified by the sector of origin are required to compute the total resources(i.e. domestic production and imports). These imports should only involve the importsfrom outside OECD-Europe, since the intra OECD-European trade should beregarded as domestic trade (In figure 3.6 this trade is represented by the greyarrows). The import statistics are derived from the statistics on the foreign trade ofcommodities (OECD [1992]). They do not cover construction, transportation andservices. Therefore, it is assumed that there are no imports from the transportation,services and construction sectors from outside OECD-Europe. Note that exportsfrom these sectors are also not covered.

With the data obtained above, it is possible to estimate the technology matrix ofOECD-Europe at a level of 15 sectors. The technology matrix and the relative shareof the deliveries from the sectors of origin to the two final demand columns in theconsolidated IO-table of the EC [Van der Linden and Oosterhaven 1995; van derLinden, 1999] are used as a starting point. The RAS-method has been applied toestimate the Z-matrix and the two final demand columns. The first final demandcolumn includes private consumption (PC), governmental consumption (GC) andchanges in stocks (ChSt) The second column comprises gross fixed capital formation( GFCF). In order to execute the RAS-method, the total inputs should by definitionbe equal to the total outputs. This need not be the case here because the totals arederived from various sources and not from one IO-table. These two totals deviate

59

sector1 2 3 4 5aver 1 2 3 4 5aver 1 2 3 4 5aver

0

5

10

15

20

25

30

35

domestic OECD

Embodied energy intensities

Figure 3.7: The embodied energy intensities forfive aggregated sectors and the average value forGermany, The Netherlands and Ireland inMJ/US$. ‘Domestic’ and ‘OECD’ represent theassumptions that the EEI of imports are equal tothe domestic EEI and the average EEI of OECD-Europe, respectively.

about 2% in the case of OECD-Europe, and therefore the output totals are adjustedin order to satisfy equation (3.4) or (3.6).

Capital depreciation is internalised after executing the RAS-procedure becausethe column of gross fixed capital formation should first be computed first. Data oncapital depreciation per sector are scarce and are computed by using the totalconsumption of fixed capital as presented by OECD [1993a] and the distributionpatterns derived from statistics of the OECD [1995]. These distribution patterns arebased on capital depreciation data computed by using the data of total fixed capitaland of lifetime (although the computed totals of capital depreciation may notcorrespond to the totals presented by OECD [1993a]).

3.2.3 Results

The impact of differentassumptions regarding the EEI-ofimports is studied by comparingthe German, the Dutch, and theIrish EEI. The energy intensitiesare calculated in the usual way(assuming that the energyintensities of imports are equal tothe domestic EEI), with the EEIcalculated by using the averageintensities of OECD-Europe toestimate the energy intensities ofimports. The German embodiedenergy intensities are about equalto the OECD-Europe averagewhile the Dutch and Irishembodied energy intensities arerespectively higher and lower thanthe average values of OECD-Europe. Figure 3.7 shows the results. From figure 3.7,it can be concluded that estimating the EEI of imports using the average EEI ofOECD-Europe (option ‘OECD’) instead of the domestic EEI (option ‘domestic’) hasa large impact (up to about 20%) on the calculation of embodied energy intensities.As might be expected, the assumption that the EEI of imports is equal to that ofdomestically produced goods and services results in an overestimation of theembodied energy intensities in The Netherlands and an underestimation for Ireland.This certainly applies to the industrial and agricultural sectors of both countries. Notonly is this the result of differences in EEI but also of the large contribution ofimports to the total resource of both countries: 20% in The Netherlands and 23% in

60

Ireland (OECD [1991c]). For Germany, the method of calculation (option ‘domestic’versus ‘OECD’) appears to have only a small affect on the outcome. This is notsurprising, since the domestic EEI of Germany are about equal to the EEI of OECD-Europe except in a number of industrial sectors. Furthermore, imports only compriseabout 13% of the total resources of Germany.

Using the average values of the OECD seems a first step in avoiding errors inassessing the EEI of imports. A next step to avoid these errors is to take into accountregional differences within the OECD-Europe region by specifying the trade amongEuropean OECD-countries (or subregion) of origin (grey arrows of figure 3.6). Thismulti-regional approach and is outlined in section 3.3

3.3 Multi-Regional Approach

3.3.1 Methodology of the Multi-Regional Approach

For each region, data on the average embodied energy intensities and trade flowsare required to specify imports to the country (or subregion) of origin within OECD-Europe. The EEI of imports can then be computed by using the so-called multi-regional approach that estimates an inter-regional IO-table by combining technologydata at a country level with the corresponding trade statistics. Ideally, the calculationsof this regional approach are performed at a country level. Such an exercise requiresthat input-output tables are available for each country of OECD-Europe. However,consistent IO-tables are only available for several countries [Eurostat, 1992a-e].Moreover, OECD-Europe consists of 19 countries and considering all these countriesseparately results in too large and complex matrices. Hence, the countries of OECD-Europe are grouped into subregions in which at least the most countries of OECD-Europe with the most dominant economies (i.e. France, Germany, Italy and UK) areconsidered separately. The region classification is presented in table 3.1. So at least,the regional differences in the most dominant economies are taken into account.Grouping the other countries into a region is not trivial since there are no obviouscriteria for making clear region classification. Hence, the region classification thatis introduced here may seem somewhat arbitrary. However, it is assumed thatconsidering the most dominant countries separately is the most important step incomputing the embodied energy intensities at a regional level. Countries are groupedtogether into one (sub)region based on similarities in their fuel mix and on theadditional requirement that countries within a region should be adjacent or near by.Similarity in fuel mix is mainly based on the contribution of fossil fuels to the totaldirect fuel use and the fossil fuel mix and nuclear fuels for the electricity generation.Similarities in economic structure are assumed to be reflected in the GDP per capitaand the contribution of the different sectors to the total GDP. However the similarityin economic structure only played a minor role in the regional classification. In orderto investigate the impact of region determining classifications on assessing the

61

embodied energy intensities, a case study was carried in which Switzerland wasshifted from region 3 into region 4. This shift hardly affected the embodied energyintensities of boh regions 3 and 4.

Table 3.1: Region classification

Region 1 UK and Ireland

Region 2 Denmark, Finland, Iceland, Norway, and Sweden

Region 3 Austria, Germany and the Netherlands

Region 4 Belgium, France, Luxembourg, and Switzerland

Region 5 Spain and Portugal

Region 6 Greece, Italy and Turkey

Let z rsi denote the deliveries of goods and services from sector i of region r to

region s, irrespective of its destination. These deliveries will find their destination inthe producing sectors of region s as well as in the final demand of region s. Let zCsi

(= 3r z rs

i ) represent the total of deliveries of i to region s. When each element z rsi is

divided by zCsi one obtains the proportion of the products used in region s that isproduced in region r. This coefficient is denoted by trrs

i (thus tr rsi = z rs

i / zCs

i). The embodied energy intensities can now be computed as follows (equation (3.7)):

œ j,s: , sj = d s

j+ 3r,i (, ri * tr rs

i * ACrij ) (3.7)

In equation (3.7), ACr represents the technology matrix of region r adjusted to inputswhich includes the consumption of capital. The matrices ACr

ij are not available at aregional level and therefore these matrices are estimated. The estimation methodologyused is similar to that of the whole region of OECD Europe. That is, the RAS-method is used to assess the technology matrix for each region (see section 3.2.1).Data sources and the assumptions involved in this procedure are also similar to thesources and assumptions listed in section 3.2.2, as otherwise both the single regionand the multi-regional approach would not be comparable. Hence, trade statistics todevelop matrix trrs are also derived from [OECD, 1992].

3.3.2 Results

Following the considerations above, OECD-Europe has been divided into 6regions and the non-European OECD-countries are lumped together in one largeregion (i.e. the rest of the world). The multi-regional input-output table derived thisway facilitates a more accurate assessment of the embodied energy intensities of

62

sector1 2 3 4 5aver 1 2 3 4 5aver 1 2 3 4 5aver

0

5

10

15

20

25

30

35 domestic OECD

region

Embodied energy intensities

Figure 3.8: The embodied energy intensities for 5aggregated sectors and the average value forGermany, The Netherlands and Ireland inMJ/US$. ‘Region denotes the results of theregional approach to compute the EEI of imports.‘Domestic’ and ‘OECD’ represent theassumptions that the EEI of imports are equal tothe domestic EEI and the average EEI of OECD-Europe, respectively

imports. Figure 3.8 clearly showsthat the regional approach (thisoption is denoted by ‘region’)influences (although less clearly)the total embodied energyintensities observed for theaggregate sectors of Ireland, TheNetherlands and Germany (notethat figure 3.8 is similar to figure3.7 only it includes the option‘region’). Apparently, assessingthe imports with the aid of theaverage energy intensities ofOECD-Europa appears to be auseful first step to calculate theEEI of imports. However, itshould be noted that the EEI ofimports of non-European OECD-countries are still assessed byusing the average values ofOECD-Europe. The resultspresented in figure 3.8 (and 3.7) do not show striking differences. The figurespresented involve the embodied energy intensities and, therefore, direct energy isincluded. The direct energy input is equal in all three options. Only EEI of the inputsof goods and services (domestically produced goods as well as imported) differamong the options (i.e. intermediate domestic deliveries and imports). 3.4 Energy Flows in OECD-Europe

Not only can the energy intensities be computed more accurately by means of thismulti-regional input-output table but the energy flows among regions can bedetermined too. Studying these energy flows in a dynamic way is one of the majorobjectives of the ECCO-modelling approach. The (embodied) energy flows amongthe 6 defined European regions are presented in figure 3.9. In this figure, the widthof the arrows indicates the magnitude of the embodied energy flows. Circle arrowsrepresent the intermediate deliveries of a region (note that these intermediatedeliveries include trade among the countries within the subregion).

Figure 3.9 shows that the regions 3, 4, 5, and 6 have a negative import-exportbalance which means that these regions depend heavily on energy resources outsidethe region. Notably for region 4 and 6, the net (embodied) energy imports comprisea relatively large part of the total energy flows within these regions. Region 1 and 2appear to be more self sufficient and both regions are even net (embodied) energy

63

region 1: U.K. and Irelandregion 2: Denmark, Finland, Iceland,

Norway and Swedenregion 3: Germany, Austria and the

Netherlandsregion 4: France, Belgium, Luxembourg

and Switzerlandregion 5: Spain and Portugalregion 6: Italy, Greece and TurkeyR.O.W. : Rest of the World

13.0

0.4

1.6

1.1

0.2 0.3

0.4

1.7

3.4

0.3

0.3

0.3

1.324.7

0.60.2

0.68.4

0.2

0.2

0.2

8.9

4.18.6

7.5

7.2

2.9

0.5

1.9 1.0

1.8

1.3

2.7

R.O.W.

R.O.W.

2.3

1.1

Figure 3.9: Overview of embodied energy flows (in EJ) in 1985 in OECD-Europe. Flowsless than 0.1 EJ are not shown.

exporting regions. However, the aggregate OECD-Europe region is in terms ofembodied energy a net importing region: net imports are about 21 EJ compared to'domestic' intermediate deliveries of about 75 EJ in 1985. The relatively highmagnitude of imports originating from non-European OECD-regions is due to theenormous contribution (between 85-95% for all regions) of fossil fuels to theseimports, that is the energy carriers themselves (e.g. crude oil).

Another remarkable difference between the regions is the magnitude of theembodied energy flows. The energy content of intermediate deliveries of region 3(24.7 EJ) is by far the largest of OECD-Europe and is about three times as high asthat of region 4, although the total production of region 3 is less than twice theproduction value of region 4 (about 1900 billion US$ for region 3 compared to 1200billion US$ for region 4 and 800 billion US$ for region 1). The relatively low valueof region 4 can be explained by the fact that the electricity supply in region 4 dependsheavily on nuclear energy (about 60% [OECD, 1991a]) and nuclear energy is notincluded in these calculations. Electricity is generated mostly from coal in both region1 and 3 (58% and 55%, respectively) in 1985 (note that former East Germany is alsoincluded in region 3 for the year 1985). Region 2 being the smallest region of OECD-Europe, in terms of population and its total production being relatively low too(expressed in monetary terms, total production equals 455 billion US$), supports the

64

low value of the intermediate deliveries of this region. Moreover, the electricitygeneration depends for about 85% [OECD, 1991a] on non-fossil fuels. The relativelylow energy content observed for region 5 and 6 can partly be justified by therelatively low production level (about 365 and 945 billion US$, respectively) and forregion 5 also by the electricity generation as in region 5 non-fossil fuels contributefor about 50% to electricity production.

Naturally the flows presented in figure 3.9 have static character. The dynamicsof these flows can be studied by means of the regional ECCO-model or regionalDREAM-model of OECD-Europe which is described in the next chapter. In thischapter, the energy flows of figure 3.9 and the input-output tables presented aboveare used as staring point.

3.5 Conclusions and Discussion

There are large national differences in embodied energy intensities. For example,the direct and the embodied energy intensities of the Netherlands and Ireland differsubstantially from the average values of OECD-Europe. This may be due to ratherspecific economic structures and the contribution of imports to the total resources.For example, the oil refining industry has a large impact on the Dutch (chemical)industry sector and imports comprises 20% of the total resources in the Netherlands.

The usual assumption that the embodied energy intensities (EEI) of imports canbe calculated in terms of the domestic EEI introduces errors in the calculation of theEEI for a country. These errors may be substantial for countries with a relativelyhigh contribution of imports and a rather specific economic structure. Hence, twoother approaches are introduced to avoid these errors. It has been demonstrated thatthe EEI of imports can be based on the average EEI of OECD-Europe. Moreover,the EEI of imports can also be determined by specifying the imports in terms of theirorigin (i.e. by dividing OECD-Europe in 6 subregions). It is believed here thatavoiding these errors in the calculations by using the two approaches introducedabove results in more accurate assessments of the energy intensities. However, itshould be noted that all calculations were performed at a level of 15 sectors whichmight exaggerate somewhat the results compared to involving more sectors.

A next step in avoiding errors in assessing the EEI of imports consists ofspecifying the imports in terms of their origin: the OECD-Europe versus the non-European OECD-countries and the rest of the world. This alternative requires theavailability of average embodied energy intensities data for each region.

The energy flows among the regions are very region specific and given the totaloutput in terms of money in these flows illustrate the energy intensiveness of theproduction of the various regions. Especially, regions 2 and 3 appear to be ratherenergy intensive which can be ascribed largely to the fuel mix in the electricityproduction sector.

65

Production

Energy Demand and Supply

Consumption

Balance of Trade

Figure 4.1: Overview of subsystems of the ECCO-modelling approach and their relationships.

Chapter 4Structure of ECCO-Models for OECD-Europe

4.1 Introduction

Two ECCO-models of OECD-Europe are described in detail in this chapter; a singleregion model and a multi-regional model. In the latter, OECD-Europe is divided into6 regions. The region classification is already discussed in chapter 3. The two modelsare introduced to compare the results of a single ECCO-modelling approach with thatof a multi-regional approach. Intuitively, a multi-regional modelling approachemphasises on regional differences which may result in more accurate outcomes.Although the multi-regional model forms the main topic of the first part of this thesis,the single region model of OECD-Europe is described here first as it less complexthan the multi-regional model and therefore lends itself more easily for presentationand explanation. Obviously, the multi-regional model and the single regional modelhave similar structures. Therefore, only the real regional aspects will be discussedfor the multi-regional model. Inboth models, four mutuallyinterrelated subsystems can bedistinguished: the productionsystem, the consumption system,the energy demand and supplysystem, and the balance of tradesystem (see figure 4.1). These foursubsystems are described below bypresenting the main characteristics.A model listing is presented inAppendix I. In addition, all datasources and the assumptions madein developing this model are also

listed in Appendix I. Throughout thischapter, diagrams are presented thatillustrate the main stocks (representedby rectangles), flows (denoted bysolid lines) and influences (denoted bydashed lines) of the subsystems.Positive and negative signs represent aco-productive influences andcounteracting influences, respectively.

Both the single and the multi-

Subscripts used below

r, s : agr, ind, ene, tra, serf, g : coal, oil, gas, bio, nucl, hydro,

solara : ac1, ac2b,c : reg1, reg2, reg3, reg4, reg5,

reg6

Diagram 4.1: Subscripts

66

regional models are developed by using of a system dynamic software packet calledVensim [1997a-c]. Vensim allows the use of subscripts (cf. diagram 4.1) whichreduces the number of equations drastically as similar equations, for instance in thecase of the various production sectors, can all be presented by one equation. Notonly is the number of equations reduced by the use of subscript but it also makes themodel more accessible as flows and influences are simulated in a more orderly andsimilar manner. The factor time is mostly an implicit variable in the model althoughit is used explicitly in determining efficiency improvements and stocks (i.e.accumulations over time).

In the models, subscripts r and s are introduced to denote the different productionsectors considered: agriculture, industry, energy, transport and services. Twosubscripts are introduced for the sectors to facilitate describing trade between sectors.Subscripts f and g are defined for the various fuel types: coal, crude oil, natural gas,biomass, nuclear energy, hydro, and solar. The latter also includes wind, geothermalenergy, and tidal and wave energy. Two subscripts are introduced for the energycarriers to facilitate describing energy conversion processes. Subscript a isintroduced for dealing with the double sets of accounts (cf. section 2.4). Throughoutthe model, ac1 denotes the first account which uses the actual energy value of a flow

or stock meaning that energy savingsand changing ERE-values areconsidered. The index ac2 refers to thesecond account in which the utilityvalue of the stocks and flows (e.g.number of cars or factories) are used.The concept of the double set ofaccounts can be compared to the ideaof current and constant dollars.Finally, both b and c indicate theregions 1-6 in case of the multi-regional model (cf. section2.4).

4.2 Production

The sectoral outputs are determinedin the production subsystem.Moreover, the (net) investments in thesector’s capital stock are computedhere.

out[s]Cs[s]

pedprod[s]

imp[s]

res[s]

+

+

+

Cs[s] : Capital stock of a sectorpedprod[s] : sector’s primary energy

demandres[s] : total amount of

intermediate deliveries toa sector [s] (includingcapital depreciation)

imp[s] : sector’s importsout[s] : sector’s output which is

equal to all inputs

Diagram 4.2: sector’s output.

67

4.2.1 Sector’s Output

As mentioned before, the ECCO-modelling approach can be regardedas a (system) dynamic input-outputmodel in the sense that the outputs ofthe sectors equal the total inputs.Hence, the total output of a sector(out[s]) is equal to total directprimary energy use of that sector(pedprod[s] and the energy value ofthe total intermediate inputs, theimports (res[s] + imp[s]), and theconsumption of capital referred to ascapital deprecation (included inres[s]). All sector’s inputs areassumed to be proportional to thecapital stock of that sector (cf.diagram 4.2).

4.2.2 Investments and Capital Stock

It is assumed here that capital stock isessential for creating output (cf.diagram 4.3). For a sector, thegrowth of the capital stock (Cs) isequal to the investments (rate ofcapital formation, rcf) diminishedwith the depreciation of capital (rdc).The capital depreciation rate equalsthe total capital stock divided by theaverage lifetime of that capital stock.The rate of capital formation iscomputed differently for each sector.The rate of capital formation of the

industrial sector is essential as it is one of the key parameters of the ECCO-modellingapproach in general. In this approach, it is assumed that the output of the industrialsector (out[ind]) that is not used for intermediate deliveries to sectors, export, or forcapital investments in other sectors is available for consumption of the domestichouseholds (incons[ind]) or for investments in the industrial sector (rcf[ind]). Thesum of rcf[ind] and incons[ind] is referred to as left. The allocation of this ‘available’industrial output among the various demands (i.e. domestic investments in the

out[ind]

rcf[ind] rdc[ind]Cs[ind]

export[ind]tintout[ind]rcfos

left

+

incons[ind]

F(msolf)

fnc

++

+

-

Cs[ind] : capital stock of industryrdc[ind] : deprecation of industrial

capital stockrcf[ind] : rate of capital formation

of industryout[ind] : industrial outputleft : output available for

consumption andinvestments in industry

incons[ind] : consumption of industrialgoods produceddomestically

F(msol) : function of the materialstandard of living

fnc : fraction not consumedexport[ind] : exports of industrial

productstintout[ind] : total intermediate

deliveries of industrialproducts

rcfos : investments in non-industrial sectors

Diagram 4.3: investments in the industrialsector.

68

industrial sector and the domestic consumption by households) depends on aparameter which is referred to as the fraction not consumed (fnc). Fnc can becompared to the savings rate implying the less is consumed the more is savedresulting in more capital or, in the case of the ECCO-methodology, the more outputis allocated to investments in capital stock. How much output is allocated toconsumption depends on the level of (material) wealth referred to as the materialstandard of living. The fraction not consumed is a function of the material standardof living (msol), see subsection 4.4.2. High levels of the material standard of livingresult in a relatively high demand for consumer goods implying that more outputmust be allocated to consumption leaving less output for investments. Herewith, thecomputation of the investment rate in the industrial capital sector involves twofeedback loops. The first comprises a positive feedback loop as more output isassociated with more room for investments. The second consists of a negativefeedback loop as an increasing industrial output results in more consumptionimplying a higher demand for consumption. In the latter case, the room forinvestments is influenced negatively by a higher demand for consumption. These twofeedback loops form the basis of the ECCO-modelling approach and they are alreadypresented in figure 2.4.

4.3 Energy System

The energy supply system is demand-driven implying that the aggregated energydemands in the various sectors determines the capacity required to meet this demand.In the energy demand module a distinction is made between the demand for thermalenergy and electricity in the production sectors as well as the domestic sector (i.e.households).

4.3.1 Energy Demand

The primary energy demand of the production sectors is essential in the ECCO-modelling approach as it determines the energy content of goods and servicesproduced domestically and which are ultimately consumed or exported. Energydemand may either increase due to growing output or decrease as result of energysavings (cf. section 2.3). Energy savings imply that a ceratin amount of output canbe produced with less energy. Producing more output with less energy can result inthe so-called rebound effect [Noorman,1995; Ryan, 1995]. This rebound effectinvolves a situation in which more output becomes available due to energy savings.When this output is allocated in a similar manner, more output will be allocated toindustrial investment implying industrial growth. The resulting industrial growth mayexceed the energy savings effect and in this way it may undermine the initial goal ofenergy saving and result in an increasing overall energy demand.

The concept of the ERE-value introduced in chapter 1 and 2 refers to the energy

69

required to make energy available. The energy savings options in the conversionprocesses of energy (from crude resource to fuel or electricity) are included in theERE-value (see chapter 2). The ERE-value may decrease as a result of efficiencyimprovements but it may also increase as it costs more and more energy to extractenergy when resources are depleting. So, the actual energy use per output can

fluctuate over time. These patternscan be considered in ECCO.

Production sectors

For each sector, the primaryenergy demand is computed at a fueltype level (cf. diagram 4.4). Thedemand for a fuel is determined byconsidering the contribution of thatfuel type in generating the electricitydemand (eedprod) and thermalenergy demand (tedprod) of thatsector. Both thermal energy andelectricity demands depend on thelevel of sector’s capital stock and onenergy savings. Both demands areexpressed in terms of secondaryenergy. These energy savings are setexogenously in this ECCO-model.Both types of energy demand shouldbe converted into a primary energydemand as all activities are expressedin terms of primary energy use in thisECCO-model. In case of electricity,the total demand for electricity(eedprod) of a sector is convertedinto primary energy terms by thevariable perel. This variablecomprises variations in the fuel mix,efficiency, and distribution losses ofgenerating electricity (see subsection4.3.2).

Households

The amount of primary energy used directly by consumers (pedhh) comprises the

pedprod[s,f,ac1]Cs[s]

eedprod[s,ac1]

tedprod[s,f,ac1]

Efficiency

-

-+

ERE[f,ac1]

perel[ac1]

+

fuelmix +/-

+/-

+/-

Cs[s] : capital stock of sectors

eedprod[s,ac1] : sector’s electricitydemand in terms

tedprod[s,f,ac1] : sector’s thermalenergy demand in termsof fuel type f

pedprod[s,f,ac1] :sector’s primaryenergy demand for fuelf

efficiency : refers to energyefficiencyimprovements in theend-use sector

fuel mix fuel mix of thermalenergy demand

ERE ERE-Value of fuel fperel primary energy

requirement forgenerating electricity

Diagram 4.4: Energy demand of productionsectors.

70

energy use for the purpose of heating,private transportation and the use ofelectricity (cf. diagram 4.5). Theenergy use for private transportationis described in subsection 4.4.5. Bothenergy use for heating (tedhh) andelectricity use (eedhh) are influencedby the capital stock of dwellings(Csdwell); not only may the numberof houses increase but also the size ofhouses may increase. Efficiencyimprovements are set exogenously.

4.3.2 Energy Supply

The total energy demand describedabove determines the amount ofenergy that must be made available.As mentioned above, producingenergy or generating electricity alsorequires energy (concept of ERE-value). The determination of the ERE-values of non-electric energy carriersas well as electricity are important todescribe the energy supply subsystem.In addition the required electricity

capacities are also determined in this subsection.

Non-Electric Energy Carriers

The energy supply of non-electric energy carriers involves energy sources suchas coal, (crude) oil, natural gas, nuclear energy, biomass and other renewables.Energy carriers are either produced indigenously or imported. Formalising thecontribution of imports and indigenous production in the domestic energy supply israther straightforward and is therefore not outlined here.

In case of non-electric energy carriers, the ERE-values are computed bymultiplying the total energy required to make the primary energy carrier available(erepes) with the energy required to (if necessary) convert the energy carrier and todistribute it to the end-user (ereconvdist). Erepes comprises the energy requirementsto make crude energy resources available, such as the energy requirements formining and transportation of the crude energy resources. Ereconvdist consists of theenergy requirements for conversion and for distribution to the end-users.

Csdwell

eedhh tedhh[f]private transport

pedhh[f]

+

+ +

+/-ere[f]perel[f]

+/-

+/-

+

pedhh[f] : primary energy demandof households per fueltype f

eedhh : electricity demandtedhh[f] : thermal energy demand

of households in terms offuel type f

ere[f] : ere-value of energycarrier f

perel[f] : primary energyrequirement of fuel type ffor generating electricity

Csdwell :capital stock of dwellingsprivate transportation refers to the

Diagram 4.5: Energy demand of households.

71

The total energy required to make the crude energy resources available (erepes)can be separated into two parts (cf. diagram 4.6). The first part (eremining)addresses ERE-value of energy produced domestically in which case only the energyrequirements for mining are computed. The second part (ereimp) involves the ERE-values of crude energy resources that are imported. In this case, the energy requiredfor transportation as well as for mining are taken into account.

The energy required for mining(eremining) includes both direct andindirect energy use (capital). Energyrequired for mining can eitherdecrease due to efficiencyimprovements or increase as result ofdepleting resources. Energy efficiencyof mining is assumed to improvefollowing a learning-by-doing processimplying that the more experienceexists in mining a certain fuel thelower the energy costs are to make itavailable. Experience in mining isdetermined by taking into account acumulative indigenous productionfactor. Depletion of the reservescoincides with increasing energyrequirements to mine a unit of energyas it is assumed that the more easilyaccessible resources are exploitedfirst. A relative depletion factorindicates the rate of depletion of thereserves of the various fuel types.The depletion factors are defined aselasticities and are specifiedseparately for each fossil fuel type [deVries and Janssen, 1996]. In this way,the ERE-value of domestic mining is

computed endogenously. On the other hand, the ERE-value of imports (ereimp) isset exogenously as the efficiency improvements are defined as scenario variables.

Conversion of primary energy sources into secondary energy and distribution ofthe secondary energy carriers are associated with energy losses (ereconvdist) so thatless energy becomes available to do work (cf. diagram 4.7). Next to energy lossesalso capital is required to enable the distribution and conversion processes.Distribution and conversion losses consist of thermal energy and electricity

eremining[f]

efficiency

learning-by-doingdepleting reserves

- +

ereimp[f]

erepes[f]

efficiency

+/-

+/-

+/-

erepes :ERE-value for makingenergy carriers availablein its crude or originalform

ereimp : energy required forimporting primary energycarriers

eremining : energy required forproducing primary energycarriers domestically

efficiency : energy efficiencyimprovements in mining ortransport

Diagram 4.6: ERE-values of energy supply.

72

(eretherm and ereelec, respectively)and may decrease over time as resultsof efficiency improvements. In bothcases, the efficiency improvements areset exogenously. The indirect energyrequirement which can be seen as theresult of the consumption of capital inthe energy distribution and conversionsectors (csconvdist) is computed bycalculating the so-called relativeconsumption of capital. This relativec o n s u m p t i o n o f c a p i t a l(erercdconvdist) holds that the totalrate of capital depreciation(rdcconvdist) is divided by the totalenergy output in terms of secondaryenergy (enerout). The total energyoutput (enerout) is equal to the totalprimary domestic energy demand(ped) and the total exports(pfexport). The rate of capitalformation in the energy conversionand distribution sector (rcfconvdist)is computed by considering thedesired capital stock (dcsconvdist)which depends on the energy output(enerout). So, the indirect energycomponent of the energy requirementfor distribution and conversion isdetermined endogenously.

Electricity Supply

Electricity is a high quality energycarrier in the economy and thereforeof great importance. Electricity can begenerated from fossil fuels as well asnuclear energy, biomass and therenewables hydro and solar.Electricity generated from fossil fuelsinvolves a different computation ofthe corresponding ERE-values than

eretherm[f]

efficiency

-

ereconvdist[f]

ereelec[f]-

efficiency

-

+ererdconvdist[f]

rdconvdist[f]Csconvdist[f]rcfconvdist[f]

enerout[f]

-

desired capital stock

+

+/--

enerout+

ereconvdist[f] : energy requirementsor losses of fuel type fduring conversion fromprimary energy carriersinto secondary carriersas well as distributionlosses

ereelec[f] : electricity required forconversion processes interms of primary energyuse of fuel type f

eretherm[f] : primary energyrequired of fuel type ffor conversionprocesses

Csconvdist :Capital stock requiredin the conversion anddistribution processes

rdcconvdist : capital depreciation ofthe energy conversionand distribution sector

rcfconvdist :investments in theenergy conversion anddistribution sector

enerout : total amount of energyoutput delivered by theenergy conversion anddistribution sector

ererdcconvdist : Consumption ofcapital required in theconversion anddistribution processesassigned to one unit of

Diagram 4.7: ERE-values of conversion anddistribution

73

that of electricity is generated fromnon-fossil fuels. This distinction is alsomentioned in section 2.2. In the casefossil energy, determining the ERE-value involves the fuel use as well asconsumption of capital. In the case ofnon-fossil fuels, determining the ERE-value only involves consumption ofcapital.

The primary energy requirementsfor producing one GJ of electricity(perel) depends on the fuel mix of theelectricity generation (fuelmixelec), theERE-value (ere), the electricity systemlosses (syselec), average efficiency ofthe power plant (efficiency), and theconsumption of capital (ererdccspp)(cf. diagram 4.8). Shifting the fuel mixfor generating electricity from fossilfuels to nuclear energy and renewablesresults in a decrease of the fossilenergy requirements. Since, the use ofenergy carriers should be expressed interms of primary energy, the ERE-value of the energy sources areincluded. Besides efficiency losses,electricity and thus energy is also lostdue to internal use and distributionlosses. Electricity system losses(syselec) cover both the energy lossesdue to internal use and the distributionlosses. Efficiency improvements ingenerating electricity and in systemlosses are set exogenously. Due to lackof data, the parameter consumption ofcapital (ererdccspp) only involves thecapital stock of powerplants. It isassumed here that this omission willhave minor impact on the result as itwill only increase the energy

requirement for energy. Sensitivity analysis presented in chapter 5 seems to confirm

fuelmixelec[f]

perel[f]

efficiency

-

+ererdccspp[f]

rdccspp[f]Cspp[f]rcfcspp[f]

ppout[f]

-

desired capacity

+

+/--

+/-

syselec

ere[f]

perel[f] : primary energyrequirement of fuel typef for generatingelectricity

fuelmixelec : contribution of fueltype f in electricitygeneration

ere[f] : ere-value of energycarrier f

efficiency : power plant’s ofgenerating electricity

syselec : distribution losses ofelectricity

Cscspp :Capital stock of powerplants in terms ofprimary energy

rdccspp : capital depreciationof power plants

rcfxspp : investments in powerplants

ppout : total amount ofelectricity output ofpower plants

ererdccspp : Consumption ofcapital of power plantsassigned to one unit ofelectricity output

Diagram 4.8: Primary energy requirement forelectricity (perel).

74

this assumption. The capital stock required for distribution and transformation is notincluded in the calculations. The consumption of capital is computed in similar wayas the consumption of capital of the energy conversion and distribution sector for thenon-electric energy carriers.

The power plant’s capacityindicates its ability to generate anamount of electricity in one hour.Total capacities of power plants(Cpp) are separated by fuel type. Notonly does the output of a plant dependon its capacity but it is also influencedby the load factor (LF). The loadfactor indicates the relativeoperational time of a power plant (cf.diagram 4.9).

The total capacity requireddepends on the total output(reqoutpp) needed to meet the totalelectricity demand. The total domesticelectricity demand is equal to theelectricity demand of the productionsectors (eedprod) and the electricitydemand of the households (eedhh).Additional capacity is needed whenthe required output exceeds thecurrent output (reqoutpp - outpp).The deprecation of capacity shouldalso be taken into account. The rate ofcapacity formation (rcfpp), therefore,equals the sum of additionallyrequired capacity formation(depending on reqoutpp - outpp) andthe total rate of capacity depreciation( rdcpp). The total rate of capacitydepreciation (rdcpp) equals the totalcapacity divided by the averagelifetime of the power plants.

Resource Base

Fossil fuels are associated with finite resources. BP [1994], WRI [1994], and

rdcpp[f]Cpp[f]rcfpp[f]

desired capacity

+

ppout[f]+

reqppout[f]

eedprod eedhh

+ +

+

-

load factor (LF)

+/-

Cpp[f] : total capacity of powerplant which generateelectricity from fuel type f

rcfpp : rate of formation ofcapacity

rdcpp : reduction of capacity ppout[f] :amount of electricity

generated from fuel type fLF : load factor that is

relative time that a powerplant is operational

reqppout[f] :required electricityoutput generated fromfuel type f to meet thedemand

eedprod : electricity demand ofproduction sectors

eedhh : electricity demand byhouseholds

Diagram 4.9: Capacities power plants

75

Nakiƒenoviƒ et al. [1997] give estimates of reserves of coal, natural gas and crudeoil which are proven to be recoverable under present economic conditions. Driven byeconomics, technological advances and policy decisions, the resource base has beenexpanded over time and may expand in the future. Energy resource quantities thatcould become economically attractive are indicated as additional resources[Nakiƒenoviƒ et al.,1997]. Hence in this ECCO-model, the energy resource base isseparated into two classes approach; proven recoverable resources and additionalrecoverable resources.

Carbon Dioxide Emissions

The use of fossil energy is associated with the emissions of greenhouse gasses.The ECCO-modelling approach described here only considers the greenhouse gas,carbon dioxide. The total carbon emissions are determined by multiplying the fossilfuel supply with the corresponding carbon dioxide emission factor (i.e. in g C / MJ[Edmonds and Reilly, 1985]. 4.4 Consumption

As mentioned above, the ECCO-model is supply driven, implying thatconsumption is mainly determined by the production sector. Besides the dependenceon the production sector, the level of consumption is also influenced by populationgrowth and the level of (material) wealth. The latter is indicated by the materialstandard of living which comprise the consumption of goods and services as well asenergy carriers.

4.4.1 Population

In this model, the population is determined in a rather simple way. Totalpopulation is assumed to grow at a rate set exogenously. This simple way ofsimulating population does not mean that the significant role of the population is notacknowledged. However, it is beyond the scope of this thesis to simulate populationin a more a complex way among others by including age distributions and categorizedbirth and death rates.

4.4.2 Material Standard of Living

The material standard of living indicates the consumption level in terms of goods,services, capital and energy (cf. diagram 4.10). The material standard of living canalso be regarded as an indicator for wealth. The material standard of living isexpressed in terms of utility as the volume of consumption is of interest here and notthe actual energy value. The relative growth in the material standard of living (msolf)

76

is equal to the material standard ofliving in a time step (msol) divided bythat of 1985. The material standardof living (msol) is equal to the grossmaterial standard of living (gmsol)divided by population (pop). Thegross material standard of living(gmsol) is equal to the totalconsumption of goods and services(tcons), direct energy (pedhh).

It should be noted that theconsumption of total capital stock ofdwellings is included in theconsumption of products stemmingfrom the services sector. Renting ofhouses should be considered as adelivery of services since a large partof the dwellings are rented fromhouse owners who belong to theservices sector. The same may beapplied to private house owners witha mortgage at the bank. Hence, theconsumption of dwellings is notincluded separately in the materialstandard of living. A same principleapplies to the use of cars. The capitaldepreciation should be included in the

material standard of living. However, car sales are included in the consumptionfigures and are therefore not included separately in the gross material standard ofliving, herewith, assuming that the consumption of cars (rate of capital formation)is about equal to the number of new cars.

4.4.3 Consumption Goods and Services

Consumption of goods and services refers to the purchase of products stemmingfrom the domestic production sectors and from imports. The consumption ofindustrial goods produced domestically (incons) is proportional to the rate of capitalformation in the industry (rcf[ind,ac2]) (cf. diagram 4.11). Herewith, consumptionis coupled with industrial growth. Moreover, the consumption of industrial goodsdepends indirectly on changes in the total material standard of living (rgcf). Thisassumption results from the key feedback loops in the ECCO-modelling approach(see chapter 4.2.2 and 2.5.2). The variable rgcf is equal to the initial ratio between

msolf

gmsol

msol msol85

pop

+

+

tcons pedhh

+ +

+

msolf : relative growth rate ofmaterial standard of living

msol : material standard of living percaput

pop : population sizegmsol :gross material standard of

livingtcons : total consumption of goods and

services (including consumptionof dwellings)

pedhh : primary energy demand byhouseholds

Diagram 4.10: Material standard of living

77

consumption of domestic industryproducts and the rate of capitalformation of the sector industry. Inaddition, this ratio is multiplied withrelative growth rate of the population(popf) and the relative growth rate ofthe energy content of consumer goods(rgech). The relative growth of theenergy content of consumption byhouseholds (rgech) is a function ofthe growth in the material standard ofliving (msolf) and the energy toexpenditure elasticity. This elasticityrefers to the assumption that energyexpenditure and the level of welfareare positively correlated. This notionis corroborated by a study of Vringerand Blok [1993] and alreadyintroduced in the ECCO-modellingapproach by Noorman [1995]. Theconsumption of products stemmingfrom the non-industrial sectors, whichare produced domestically, is equal tothe contribution of that sector to thetotal consumption. The energy contentper unit consumption good maychange differently over the sectors dueto various energy saving optionswithin the sectors. The change inenergy content per unit consumptiongood or service may change theconsumption pattern as the energyvalue and the utility value may showdifferent distribution patterns. Besides

goods and services which are produced domestically, consumption goods are alsoimported

4.4.4 Dwellings

The capital stock of dwellings (Csdwell) is computed in a straightforward way;that is the increase of capital is equal to the rate of capital formation (rcfdwell)diminished with the rate of capital depreciation (rdcdwell) (cf. diagram 4.12). The

incons[ind,ac2]

rcf[ind,ac2] rgcf

popfrgech

msolf

+

+

+

incons[ind,ac2] : consumption ofindustrial goods interms of the utility value

rcf[ind,ac2] : investments inindustrial sector interms of the utility value

rgcf : ratio betweenconsumption ofindustrial goods andinvestments inindustrial sector

popf : relative populationgrowth rate

rgech : relative growth rate ofthe energy content ofconsumer goods

msolf : relative growth rate ofmaterial standard ofliving

Diagram 4.11: Consumption of industrialgoods produced domesticallly

78

rate of capital formation (rcfdwell) is determined by considering a desired level ofcapital stock of dwellings (dcsdwell). The desired level of capital stock is assumedto depend on the relative growth of population (popf) and of the relative growth onthe material standard of living (msolf).

4.4.5 Private Transport

Private transport involvestransport of individuals by privatecars (cf. diagram 4.13). Publictransport is included in the transportsector. Energy use of private transportcomprises the direct use of car fueland the indirect use of capital inwhich capital only includes the energyused to produce cars. Due to lack ofdata, energy requirements to developand maintain infrastructure are notincluded in these calculations. It isassumed here that this omission willhave minor impact on the overallresult as it will only increasesomewhat the level of consumption.Sensitivity analysis presented inchapter 5 seems to confirm thisassumption.

The total number of cars (Cars) ina region is equal to the number of new cars (rbars) minus the number of carsscrapped (rdcars). The number of new cars is a function of the total number of carsthat is required (carsreq). Which in turn is equal to the average number of cars perperson (carspp) times the total population (pop). The number of cars per persondepends on a growth rate set exogenously.

The total fuel use by private car transportation (fuelcars) is a function of thepersonal travel kilometres per person (pkmpp), the population (pop), and the fuelrequirement per km (fuelreqkm). The personal travel kilometres per person (pkmpp)are assumed to be proportional to the relative growth of the material standard ofliving (msolf). Efficiency improvements in the the fuel requirement per km per fueltype are set exogenously.

4.5 Balance of Trade

The balance of trade (impexpbalance) is an important parameter in ECCO as is

rdcdwellCsdwellrcfdwell

popf

+

msolf

+

dcsdwell

+

+

Csdwell : capital stock of dwellingsin terms of energy

rcfdwell : rate of capital formationof dwellings

rdcdwell : capital depreciation ofdwellings

dcsdwell : desired level of capitalstock of dwellings

popf : relative population growthrate

msolf : relative growth rate ofmaterial standard of living

Diagram 4.12: Dwellings

79

it indicates the dependence on foreign resources. An import-export balance largerthan zero means that the region is a net importer of energy. A relevant aspect of beinga net importing region lies within the framework of sustainable development and isrelated to the distribution of natural resources between nations or world regions. Inthis perspective, the question arises to what extent a region can import energy.

In the import-export balance, bothimports (totimports) and exports(totexports) involve goods andservices as well as primary energyresources (cf. diagram 4.14). Importsand exports of goods and services bythe production sectors (imp andexport, respectively) are assumed tobe proportional to the capital stock ofthe sectors involved. Imports ofconsumer goods and services (mcons)are determined by the consumptionsector (tcons). The imports andexports of primary energy (pfimportand pfexport) involve the fossilprimary energy resources (i.e. coal,crude oil, and natural gas) as well asrefined oil products which areexpressed in terms of primary energy.Imports of primary energy depend onthe domestic primary energy demand(ped). The exports of primary fuelsdepend on the primary energy supply(tpes) which involves the total amountof energy that is made available withinthe region.

4.6 Multi-regional Model

In sections 4.2-4.5, a detailed description is presented of the single region modelof OECD-Europe. Besides a number of new features (e.g. the endogenous way ofcalculating the energy requirement for mining), this model is still rather similar toother ECCO-models [Noorman, 1995; Ryan,1995; Crane, 1995]. Developing amulti-regional model is, however, one of the main goals of this thesis. It involves anumber of specific methodological aspects which are described below.

The multi-regional model involves a distinction of 6 regions within OECD-Europe. The region classification was already introduced in chapter 3 and is therefore

rdcarsCars

Carsreq

Material standard of living (msolf)

popfuelreqkm

car fuels- + +

+pkmpp

-

efficiency

rbcars

-

+pop carspp

+

+

fuelcars : total fuel use byprivate transport

fuelreqkm : fuel requirement perkm

pkmpp : personal travelkilometres per person

pop : populationCars : number of carsrdcars : number of cars

scrappedrbcars : number of new carsCarsreq : number of cars

requiredcarspp : number of cars per

person

Diagram 4.13: Private transport.

80

only listed here:

Region 1: Ireland and UK;Region 2: Denmark, Finland, Iceland, Norway, and Sweden; Region 3: Austria, Germany, and The Netherlands;Region 4: Belgium, France, Luxembourg, and Switzerland ;Region 5: Portugal and Spain;Region 6: Greece, Italy ,and Turkey.

In general, the multi-regional modelling approach has the same structure as thesingle region modelling approach, but in the multi-regional model, all calculations aremade for 6 regions instead of one. In this model, subscripts b and c both denote theregions 1-6. Two subscripts are introduced for the regions to facilitate describingtrade.

In most cases, the shift to multi-regional modelling approach is ratherstraightforward as most parametersare not dependent on other regions.Most equations can be ‘regionalised’just by adding the subscript b or c.Obviously, this method does notapply to trade of goods and servicesand of energy as these flows are muchmore complex. Regions can alsocontribute to investments in otherregions and are expected to do somore frequently in the liberalisedenergy market in future years. Hence,equations associated with these tradeflows are adjusted to incorporatethese regional aspects. Below, theequations concerned are discussedand together with the unchangedequations which have already beendescribed above, they form thestructure of the regional model ofOECD-Europe.

4.6.1 Trade in Goods and Services

Trade in goods and servicesinvolves imports and exports to

impexpbalance

totexports

mcons+

-

+ imp

totimports

exports pfexports

CS production sectors

+

+ ++

tpes

+

+

++

pfimports

ped tcons

+

impexpbalance : import-export balancetotimports : total amount of

importstotexports : total amount of

exportsimp :imports of goods and

servicespfimports : imports of energy

carriers in terms ofprimary energy

exports :exports of goods andservices

pfexports : exports of energycarriers in terms ofprimary energy

ped : total domestic primaryenergy demand

CS : capital stock

Diagram 4.14: Balance of trade.

7 Note that the regions outside OECD-Europe are aggregated to one artificiallylarge region referred to as Rest Of the World (R.O.W.).

81

production sectors and consumers. Imports of a region originate in either anotherregion of OECD-Europe or outside OECD-Europe. Likewise, exports can have theirdestination in other regions of OECD-Europe or outside OECD-Europe. A cleardistinction is made between trade within OECD-Europe and trade with regionsoutside Europe7. This distinction is required to allow a comparison between the multi-regional and the single region model in which trade only involves imports andexports with regions outside OECD-Europe. Imports of the production sector consistof goods and services required in the production process or in investments. Theequations which are adjusted mostly involve the output (out) and the room forinvestment in the domestic industrial sector and the consumption of industrial goodsproduced domestically (denoted by left, see also above). The total intermediateinputs (res) of a sector not only includes goods and services domestically producedbut also the goods and services originating in other regions of OECD Europe (i.e.all inputs originating in OECD-Europe are included). That is, the total intermediatedeliveries to sector s in region c is equal to the sum of the deliveries from each sectorin any region to sector s of region c. All these intermediate deliveries are assumedproportional to the capital stock in sector s of region c. As mentioned in chapter4.3.1, it is assumed that the output of the sector industry (out[ind]) that is not usedfor any intermediate delivery to another sector, export, or for the investment in non-industrial sectors (this total is referred to as left) is available for consumption of thedomestic households or for investments in the domestic industrial sector (rcf[ind]).The room for investment in the domestic industrial sector and the consumption ofindustrial goods domestically produced is much more complex in case of the multi-regional model because deliveries to other sectors and the contribution in investmentsin other sectors may also involve deliveries to and investments in other regions.Moreover, the domestic industrial sector may also contribute to the final consumptionin other regions. These terms should obviously also be excluded from the room fordomestic industrial consumption and investments in the domestic industrial sector.The variables export and imp still only involve trade with a region outside OECD-Europe.

4.6.2 Trade in Energy

Trade in energy involves imports and exports in primary energy sources (such ascoal and crude oil) and in secondary energy sources such as electricity and refinedoil products. Similar to trade in goods and services, a clear distinction is madebetween imports from and exports to regions within OECD-Europe and with regions

82

outside OECD- Europe. The multi-regional approach influences the way in which the primary energy

supply is modelled. The total primary energy supply (tpes) is equal to the primaryenergy that is produced indigenously (indig) and the primary energy that is importedfrom regions within as well as outside OECD-Europe (pfimport). As mentionedbefore, crude oil and refined oil products are dealt with separately although both areexpressed in terms of primary energy (i.e. crude oil). Therefore, it is convenient todeal with the fuel type oil separately from the non-oil fuel types. For each non-oil fueltype, the total indigenous production of the fuel in region c is equal to contributionin the energy demand of any region within OECD-Europe. Thus, the production offuels is influenced by the energy demand of other regions. In addition, the energysupply also depends on exports to regions outside OECD-Europe. The distributionof fuel supply to the region of origin is set exogenously for each region of destination.

The indigenous production of crude oil (indig[oil]) is computed in a similarmanner as the non-oil fuel types. However in this case, the demand for crude oil isnot dependent on the primary energy demand of the region of destination. It dependsindirectly on the production of refined oil in the region of destination as crude oil isnot used directly by end-users but is refined into other products first. The distributionpattern of crude oil to the region of origin is also set exogenously for each region ofdestination. Trade of refined oil (refoilprod) is computed similarly since the non-oilfuel types as the indigenous production of refined oil products is also dependent onthe primary energy demand in the region of destination. For all the energy sources theprimary fuel trade described above only involve trade within OECD-Europe. Exportsto and imports from regions outside OECD-Europe are, of course, also included inthe regional ECCO-modelling approach.

The total domestic demand for generating electricity in region c (totdemgen[c])depends on the total domestic electricity demand of that region (domeedprod[b] anddomeedhh[b]). It is also possible to trade electricity with a region outside OECD-Europe (melec and eelec).

83

Chapter 5

Scenario Results of the Regional ECCO-Models

5.1 Introduction

This chapter discusses scenario results of the regional ECCO-models of OECD-Europe as described in chapter 4. Section 5.2 describes the outcomes of sensitivityanalysis. Section 5.3 focuses on the impacts of regional differences. The outcomesof the multi-regional model are compared with results derived from analysis with theaggregated model of OECD-Europe with the aim of studying the impact of regionaldifferences. Finally, conclusions about the multi-regional ECCO-approach arepresented in section 5.4.

5.2 Sensitivity Analysis

5.2.1 Introduction

Sensitivity analysis is of major importance for developing a model as it givesinsight in the significance of the outcomes. Moreover, it indicates the most relevantparameters in the sense that these parameters have the most impact on the outcomesof the model. Below, a number of constants are varied in order to study the effectsof corresponding variations. The constants varied are used in the key equations of themodel. These equations involve determining the magnitude of imports, intermediatedeliveries, thermal energy use, electricity use, the primary energy demand ofelectricity generation, and the ERE-values.

Not only does varying the constants influence the parameters directly associatedwith the constants involved but it has also impact on other outcomes. As mentionedin chapter 4, the rate of capital formation is the balancing term in ECCO-modelswhich means that the room for investments is sensitive to changes in other inputs andoutputs. As a result, it can be expected that sampling the constants in the sensitivityanalysis has the most effect when the equations involved influence directly the rateof capital formation. For instance, the room for investments decreases in absoluteterms if more embodied energy is allocated to consumption or intermediate deliveries.

8 In the reference scenario, developments of the relative growth rates of non-industrial sectors and the energy savings are based on the average annual growth ratesand the average annual improvements observed for the 1985-1995 period. In the

84

year1985 1995 2005 2015 2025 2035 2045

0

0.05

0.1

0.15

0.2

Region 1 Region 2 Region 3

Region 4 Region 5 Region 6

Net investments

2050

Figure 5.1: Regional net investments (in(1985)EJ) for the reference scenario

Sensitivity analysis can alsobe associated with regionaldifferences as for instance netinvestments (i.e. net growth ofcapital, i.e. rate of capitalformation minus depreciation)can be more sensitive to smallchanges for a number of regions.Figure 5.1 illustrates regional netinvestments in the industrialsectors for the referencescenario8. This figure shows that

Monte Carlo sensitivity analysis

Monte Carlo simulations, also known as multi variate sensitivity analysis,areused to carry out the sensitivity testing. Within Monte Carlo simulations theparameter considered is sampled randomly according to a given probabilitydistribution [Winston,1987].

Here, fifty simulations are performed in which the constant involved issampled over a range of values according to a triangular distribution. Thetriangular distribution is a member of the family of continuous probabilitydistributions which is easy to interpret. In essence, it is a good approximationof the log normal distribution [Schenk,1998].

The constants are varied between a span of ±10% of its value (i.e. themaximum value = 1.1 * CST and the minimum value = 0.9 * CST). Clearlythe peak of the triangular distribution coincides the reference value of theconstant (i.e. peak = CST).

Below, the results are only presented for the major parameters. In thissubsection each graph shows the deviations resulting from changing theconstants. The dotted area covers 65% of the results implying that the resultsof about 33 simulations lie within the dotted area. An additional 30% of theresults is indicated by one of the slashed areas. The remaining 5% of theresults lie within the other slashed area.

reference scenario, these growth rates and improvements are moderated compared to thegrowth rates and the improvements observed to slow down these changes rates (see alsoscenario ‘sec2’ in section 5.3).

85

sensitivity_imports65% 95% 100%

out[ind,reg1,ac2]10.82

8.678

6.526out[ind,reg2,ac2]

4.466

3.839


35.55

27.63


14.99

12.10


7.228

5.568


14.91

11.89

8.8641985 2018 2050

Time (a)

Figure 5.2: Variation in regional industrialoutput (in (1985)EJ) as result of sampling theenergy intensities of imports originatingoutside OECD-Europe.

the net investments decrease for eachscenario which is the result of adecreasing room for investments. Inturn, the latter is due to theprerequisite that industrial output isallocated first to the demands in non-industrial sectors. In addition, itshows that the net investments ofregions 2, 4, 5 and 6 become aboutzero in the reference scenario. As aresult, capital growth may expectedto be rather sensitive to smallchanges in these regions as thedifference between growth or decayappears to be small. Subsections5.2.2-5.2.7 describe separately theresults corresponding to samplingone constant. The methodology usedin the sensitivity analysis is outlinedin the following text box.

5.2.2 Imports

Chapter 3 discussed extensivelythe problems associated withassessing the energy content ofimports. In chapter 3, it was arguedthat the assessment of energyintensities may still introduce errorsin the calculations although anadvanced approach is presented todetermine the energy intensities at acountry or a regional level. Lack ofdata aggravates the difficultly ofdetermining the exact value ofimports, especially in the case ofimports originating outside OECD-Europe. Therefore, the embodied

86

energy content of imports involves a high degree of uncertainty. Hence, sensitivityanalysis is performed on the energy content of imports in order to study theirsignificance on the total industrial output. In this case, energy intensities are solelyvaried for imports by the industrial sectors and originating outside OECD-Europe.

Figure 5.2 shows the results for the industrial output. Apparently, varying theenergy intensities of imports hardly affects the industrial output. Only for regions 2and 4, the variation in the energy of imports has a moderate impact on the industrialoutput. The minor impact of these changes is not surprising as the imports involvedonly contribute little (i.e. less than 5% for most regions) to the total sector’s input.The minor impact of varying the energy intensities of imports means that accidentalerrors in determining these energy intensities will have little effect on the outcomes.

5.2.3 Energy and Electricity Demand

Subsection 5.2.2 showed that the regional ECCO-model of OECD-Europe is notsensitive to changes in energy intensities of the imports originating outside OECD-Europe, maybe with the exception of regions 2 and 4. The thermal energy andelectricity demands contribute substantially to the total input. In most regions, thetotal input consists for about 30% of (direct) primary energy use. So, one couldexpect that varying the thermal energy and electricity demand would result in a highervariation in the industrial output. In this subsection, the results are presented ofrespectively changing the ratio between the thermal energy use and the capital stockand the ratio between the electricity demand and the capital stock (see the parametersTEI and EEI as described in section 4.3)

Figure 5.3 shows the results of varying the ratio between the thermal energy useand the capital stock. Especially for regions 2, 3 and 4, altering the thermal energyuse ratio results in a substantial variation of the industrial output (10-15%). The highsensitivity of the rate of capital formation to small changes results in widespreadvariation of output. For regions 1, 5, and 6, the deviations in the output are about 5-10% which is rather similar to the deviations in the total inputs. The distribution ofthe variation in the industrial output is somewhat skewed for regions 2 and 4, that isthe higher output levels show more variation than the lower ones. Figure 5.4illustrates this effect as it shows the net room for investments for each region. Theroom for net investments shows a somewhat skewed distribution for these regions.Moreover, the net investments even become negative for regions 2, 4, and 6 implyingthat the capital stock and output is decreasing. For regions 1, 3, 5, and 6, thevariations in the net investments diminishes in time as the variations are mainly theresult of the initial fluctuations. For regions 2 and 4, the variation stabilises orincreases as the net investments approach zero or even becomes negative whichappears to have more impact on the outcomes.

87

sensitivity_TEI65% 95% 100%


8.887


4.700

3.942


37.36

28.38


16.15

12.62


7.449

5.641


15.71

12.21

8.7121985 2018 2050

Time (a)

Figure 5.3: Variation in regional industrialoutput (in (1985)EJ) as result of sampling thethermal energy demand to capital ratio.

Figure 5.5 shows that varying theratio between the sectoral electricitydemand and the capital stock of thatsector results in outcomes rathersimilar to varying the ratio between thesectoral thermal energy demand andthe capital stock. Regions 2 and 6 aremost sensitive to the variations on thesectors inputs (i.e. the electricitydemand) only the outcomes show asomewhat smaller variation than in thecase of varying the thermal energydemand. Especially for region 4, theoutcomes show less deviationcompared to figure 5.3. This differenceis due to the fuel mix of the electricitygeneration. In region 4, electricitygeneration mainly depends on nuclearenergy which is hardly taken intoaccount in the energy intensities.

Again, the variation is somewhatskewed. Only in this case, the loweroutput levels show more variation thanthe higher ones. A decreasing room forinvestment may have more impact thanan increasing one or alternativelydecreasing growth rates may havemore impact than increasing ones.

5.2.4 Intermediate Deliveries

Sections 5.2.2 and 5.2.3 dealt withthe sensitivity of development ofindustrial output for varying theinputs. This section describes thesensitivity of the industrial output forchanges in that output. In this section,the energy intensities of theintermediate deliveries from theindustrial sector to non-industrialsectors are sampled. These samples may be expected to influence substantially therate of capital formation in a number of cases. An increase in industrial deliveries to

88

sensitivity_EEI65% 95% 100%


8.771


4.594

3.888


36.30

27.94


14.83

12.03


7.353

5.613


15.37

12.07

8.7691985 2018 2050

Time (a)

Figure 5.5: Variation in regional industrialoutput (in (1985)EJ) as result of samplingthe electricty demand to capital ratio.

sensitivity_TEI65% 95% 100%

netinv[ind,reg1,ac2]0.1

0.05

0netinv[ind,reg2,ac2]

0.04

0.016

-0.008netinv[ind,reg3,ac2]

0.2

0.1


0.2

0.08

-0.04netinv[ind,reg5,ac2]

0.06

0.03


0.2

0.09

-0.021985 2018 2050

Time (a)

Figure 5.4: Variation in regional netinvestments in the industrial sector in(1985)EJ as result of sampling the thermalenergy demand to capital ratio.

other sectors implies that relatively more output should be allocated to intermediatedeliveries which decreases the room for investments.

89

sensitivity_intermediate deliveries65% 95% 100%


9.108


5.345

3.791


39.50

29.64


17.92

13.23


7.563

5.739


16.87

12.87

8.8811985 2018 2050

Time (a)Figure 5.6: Variation in regional industrialoutput (in (1985)EJ) as result of samplingthe energy intensities of intermediatedeliveries originating in the industrialsectors.

Varying the energy intensities of industrial intermediate deliveries to non-industrial sectors results in higher deviations than in the case of varying the ratiobetween thermal energy demand orelectricity demand and the amount ofcapital stock in a sector. Regions 2 and 4still show the highest deviations (between-50% and +20% in 2050) in industrialoutput (see figure 5.6). For both regions,industrial output collapses in some cases.Also for region 6, the industrial outputdecreases as result of changes in energyintensities of the intermediate deliveriesfrom the industrial output to non-industrial sectors.

Similar to figure 5.5, the variation inthe industrial output is skewed. However,there is a substantial difference betweenvarying the energy intensities of outputcompared to altering the energyintensities of the inputs. Altering the ratiobetween the thermal energy or theelectricity demand results in deviations inthe inputs of the industrial sector. Sincetotal output is set equal to the totalamount of inputs, the output of a sectordecreases when one of the inputsdecreases. If the total output decreasesand other items such as intermediatedeliveries to other sectors and exportsremain constant, the room forinvestments decreases as less output canbe allocated to this balancing term. So, adecrease in either the electricity demandto capital ratio or in the thermal energydemand to capital ratio results in adecreasing room for investments. Thedecreasing room for investments inducesthe collapsing type of behaviour inregions 2, 4, and 6. The opposite holdsfor varying the energy intensity of theintermediate deliveries stemming from theindustrial sector. Less output is allocated

90

sensitivity_ERE65% 95% 100%

totped[reg1,ac1]13.09

10.41

7.730totped[reg2,ac1]

4.435

3.594


26.09

20.47


12.99

10.76


5.856

4.424


19.66

14.02

8.3861985 2018 2050

Time (a)Figure 5.7: Variation in regional totalprimary energy demand (in EJ) as result ofsampling the ERE-values.

to industrial deliveries to sectors when the energy intensities of intermediate deliveriesdecrease. The level of output remains, however, almost the same as the decreasingenergy intensities of deliveries mostly affect the allocation of the industrial output.Hence, the room for investments increases as less output is allocated to deliveries. So,the room for investments increases if theenergy intensities of intermediatedeliveries decrease. Although the resultsappear to be rather similar, altering theinputs and allocation of output influencethe potential industrial outputdevelopment differently.

5.2.5 ERE-values

The ERE-value (energy requirementfor energy) is one of the key parametersof the ECCO-modelling approach as itinfluences, directly or indirectly, almostall other parameters of the model.Section 4.3 describes the way the ERE-values are computed and it shows thatmost of the parameters involved indetermining the ERE-values are setexogenously. On the one hand, settingthese parameters exogenously means thatthe parameters involved can be adjustedeasily and therefore be part of scenarioassumptions, but on the other hand smallchanges may have much impact on theoutcomes of the whole system andespecially on the energy supply anddemand. The impact of small changes isstudied by sampling the overall ERE-value.

Unlike most parameters used fordetermining the ERE-values, the ERE-value for mining is computedendogenously by considering a reservesdepletion factor and a so-called learning-by-doing factor. The reserve depletion

91

sensitivity_depletion factor65% 95% 100%

ere[gas,reg1,ac1]1.270

1.186

1.101ere[gas,reg2,ac1]

1.505

1.292


1.148

1.104


1.080

1.074


1.215

1.120


1.121

1.089

1.0571985 2018 2050

Time (a)

Figure 5.8: Variation in regional ERE-values (in EJ) as result of sampling thereserve depletion factor.

factor takes into account that easily accessible reserves are depleted first. Hence,mining requires more energy if reserves are depleting. In contrast, advancedtechniques make mining more energy efficient. This aspect is embodied by thelearning-by-doing factor implying that better techniques become available withincreasing experience in mining. The reserve depletion factor uses an elasticity tocouple the reserve depletion rate and theenergy required for mining. Thiselasticity is derived from [De Vries andJanssen, 1996] and it involves a highlevel of uncertainty as this elasticity issolely based on money instead ofenergy. The same applies to the dataused to compute the elasticityassociated the learning-by-doing factor.The elasticity of the reserve depletionfactor is also varied in order to study itsimpact on the whole system.

Figure 5.7 shows the impact ofvarying the overall ERE-values of allfuel types on the total primary energydemand. The outcomes vary the mostfor region 3 (more than 25%) as resultof the relatively high dependence onfossil fuels. Varying the ERE-value hassimilar consequences for the growthpotential of the industry compared tochanges made in subsection 5.2.3 andthe industrial output shows variationssimilar to those in figure 5.6. Not onlydo different growth rates of theindustrial sector impose variations onthe total primary energy demand of aregion but also the energy demand byhouseholds influences the total primaryenergy demand. For regions 1, 2, 4, 5and 6 the total primary energy demanddeviates between 10-20% of the meanvalue (i.e. value of the reference run)which is somewhat higher than thevariation in the ERE-values.

Figure 5.8 shows the results ofvarying the elasticity associated to the

92

sensitivity_perel65% 95% 100%

totped[reg1,ac1]11.52

9.939


3.841

3.566


22.25

19.44


11.27

10.36


5.412

4.307


17.35

13.12

8.9021985 2018 2050

Time (a)

Figure 5.9: Variation in regional totalprimary energy demand (in EJ) as result ofsampling the primary energy requirementfor generating electricity (perel).

reserves depletion factor in the case of natural gas. Although the overall effect on thetotal ERE-value of natural gas is rather moderate (less than 5%, note the relativelysmall scale on the y-axis), four types of behaviour patterns in the variation can berecognised in figure 5.8. Thedevelopment of the ERE-values ofregions 1 and 3 shows a very interestingpattern. In the beginning, due todepletion of reserves the ERE-valuesincrease. The deviations also increase asa consequence of sampling theelasticity. Subsequently, the ERE-valuereaches its highest level and then startsto decrease. At that point, the reservesare depleted and all natural gas isimported. Therefore, the ERE-value isno longer influenced by a varyingelasticity and this explains the lack ofvariation in this part of the curve. Inregions 2 and 6, the ERE-value simplyincreases due to depletion of thereserves as the reserves are sufficient tomeet the total cumulative energydemand in this period. In region 4,efficiency improvements first exceed theeffects of depleting reserves but theERE-values also starts to increase whenthe reserves become more depleted inthis region. For region 5, the ERE-valueis indifferent for varying the elasticityas almost all natural gas is imported inthis region. The relatively smallvariation in the ERE-values has littleeffect on the total primary energydemand in all regions.

5.2.6 Electricity

Nowadays, electricity is one of themost important energy carriers withinthe economy. The primary energyrequirement for generating electricity(denoted by the parameter perel, see

93

sensitivity_EEE65% 95% 100%


8.669


4.454

3.838


35.66

27.72


14.65

11.97


7.486

5.700


14.85

11.86

8.8811985 2018 2050

Time (a)

Figure 5.10: Variation in regional industrialoutput (in (1985)EJ) as result of samplingthe energy to expenditure elasticity.

also section 4.3) depends on the fuel mix, efficiencies in the power plants, distributionlosses, the use of capital, and the ERE-values of the corresponding energy carries.Most of these parameters are set exogenously and most of them are scenariovariables.

Figure 5.9 shows the impact ofsampling the perel parameter on thetotal primary energy demand. The totalprimary energy demand clearly showsless deviation than by changing theoverall ERE-values. For most regions,the deviation is about half the deviationof changing the ERE-values. Thisresult is not very surprising aschanging the ERE-values alsoinfluences the perel-value. So, changingERE-values affects all direct energyuse while changing the perel-value onlyhas impact on the primary energyrequirement for generating electricity.

5.2.7 Consumption

The energy to expenditure elasticitydetermines the energy requirement of acertain consumption level. Thiselasticity can considered to be aconversion factor between growth ofwealth (indicated by the materialstandard of living) and the energyrequirements for the consumption ofmaterial goods. For all regions, thiselasticity is based on data of a smallcountry such as the Netherlands[Vringer and Blok, 1993]. Applyingthis elasticity to OECD-Europeinvolves a high level of uncertainty. Forthis reason, this elasticity is sampled inorder to study the impact of thisconstant.

Figure 5.10 shows that varying theenergy to expenditure elasticity hardly

94

(1985)EJindustrial output

Figure 5.11: Comparison of variation in industrial output for region 3 in 2050 (in(1985)EJ) as result of sampling several constants.

affect on the industrial output. This minor effect can be explained by the fact thatconsumption only forms a small part of industrial output. Small changes in a minorterm of the output does not influence the room for investments too much.

5.2.8 General Conclusions of Sensitivity Analyses

In subsections 5.2.2-5.2.7, a number of constants are sampled in a ±10% rangeof its mean value to study the sensitivity of the model on the constant involved. Theabove has shown that changes in inputs and outputs of the industrial sector may haveconsiderable consequences for the outcomes. The potential of the industry to growis especially sensitive in a number of regions (i.e. regions 2 and 4) where the netinvestments (total investments minus depreciation) are about zero. In the otherregions, deviations in the output more or less coincide with the variation in theconstants involved.

Fortunately, a number of constants associated with a high level of uncertainty (i.e.energy intensities of energy requirement for mining to reserve depletion elasticity andenergy to expenditure elasticity) show relatively low deviations in the industrialoutput indicating that possible errors in determining these constants have little or noeffect on the overall behaviour of the model.

95

Figure 5.11 presents a comparison between the variations in the industrial outputfor region 3 in the year 2050. This figure also shows that the variation in theconstants with a high level of uncertainty is negligible compared to the variationsassociated to sampling the other constants. Sampling the constants related to theintermediate deliveries originating in the industrial sector shows the highest deviation.In addition, the distribution is clearly skewed. The same applies to sampling theelectricity intensity. Next to sampling the constants associated to the intermediatedeliveries originating in the industrial sector, sampling the constants directly relatedto the energy demand (ERE, TEI and EEI) result in substantial variations.

5.3 The Impact of Regional Differences

Section 5.2 showed that the potential development paths differ per region. Forinstance, regions 2 and 4 are much more sensitive to small changes than the otherregions. What are the consequences of these differences on the total industrial outputof OECD-Europe? Or in other words, does it matter whether OECD-Europe isconsidered as one large region or whether regional aspects are included in thecalculations? This section addresses these questions. First, results are shown at aregional level for a number of parameters and for a number of scenarios. In addition,the outcomes of scenarios developed with the aid of the regional ECCO-model arecompared with that of the aggregated model of OECD-Europe in order to study theimpact of regional differences. The primary intention of designing these scenarios isto study regional differences and not to predict future developments, meaning that thescenarios are designed from a methodological point of view. Therefore, the scenariosintroduced encompass a wide span of development potentials. In a number of cases,this wide span may result in options that are not very likely to happen.

96

Scenario variables are varied, here, according to three alternatives. In the first twoalternatives, the parameters involved are determined by using the average annualgrowth rates or improvements observed for the 1985-1995 period (the growth ratesand improvements observed are denoted by o.r.). The alternatives are outlined in thetext box presented below.

Energy intensities and the desired relative growth rates of the non-industrialsectors are the sole parameters that vary over the different scenarios. The desiredrelative growth rate of a non-industrial sector consists of the desired output growthrate of that sector related to the growth rate of industrial output. This desired growthrate is defined in order to couple the growth of non-industrial sector to industrialgrowth and in addition by considering the notion that non-industrial sectors may growat a different rate than the industrial sector. These rates are determined by using thegrowth rates of the ratio between the value added of the non-industrial sectors and theindustrial sector observed for the 1985-1995 period [OECD, 1997b]. Energy savingsinclude efficiency improvements as well as structural changes and are determined bycoupling the trends in final energy demand observed for the 1985-1995 period to the

Scenario alternatives

In studying the regional differences, parameters can vary according to threealternatives. The parameters are changed by defining different growth rates andimprovements among the alternatives. The growth rates or improvements are definedas the annual change of the parameter implying that the value of parameter (V) in theyear (y) is equal to Vy = V0 * (a.c.y)

y In this equation V0 refers to the initial value. Inaddition, the factor a.c.y is referred to as the annual change in year y (which maychange in the course of time). Note that between 1985 and 1995 the annual changeis equal to (1+o.r.), where o.r. refers to growth rates and improvements observed forthat period.

The first alternative (denoted by as ‘sec1’) refers to the assumption that the annualchanges of 1985-1995 are maintained in the 1996-2050 period (for scenario‘sec1’,a.c.y = (1+o.r.) for y between 1996-2050).

In the second alternative (denoted by ‘sec2’), the annual changes observed arereduced to slow down the changes. For the 1996-2025 period, these annual changesare equal to the square root of the annual changes observed for the 1985-1996 period(i.e. for scenario ‘sec2’, a.c.y = (1+o.r.)1/2 for y between 1996-2025). For the 2026-2050 period, these annual changes are equal to the square root of the annual changesof the 1996-2025 period (i.e. for scenario ‘sec2’, a.c.y = (1+o.r.)

1/4 for y between 2026-2050). The second scenario is also used as reference scenario in the sensitivityanalysis of section 5.2.

In the third alternative (referred to as ‘seccst’), the parameters involved areassumed constant after 1995 (o.r.=0 implying that a.c.y = 1 in scenario ‘seccst’).

97

growth rates in the value added observed for that period. Data needed to calculate allthe scenario variables are derived from the OECD [1991a-c, 1993a-c, 1994, 1995a-b, 1997a-d], Fritsche et al. [1994] and the UN [1987, 1988a-b, 1990, 1996] andinvolve the years 1985 (starting point) and 1995 (reference point).

However, energy efficiency improvements are set in a different manner in the caseof region 6 (including Italy, Greece and Turkey). Unlike the other regions, thethermal energy and the electricity intensities increase considerably for most sectorsin region 6. These increasing intensities mean that more thermal energy and electricityis required to produce one unit of output and they might be the result of a kind of‘catching up’ with the other regions in OECD-Europe. These considerations arediscussed more in chapter 9. Extrapolating these observed trends for the thermalenergy and the electricity intensities would result in an extremely high primary energydemand. For instance, extrapolating the observed trend in the electricity intensity ofthe agricultural sector would imply that this sector will require more than hundredtimes more electricity to produce one unit of output by the year 2050 compared to1985. This is obviously very unlikely to happen. Although, the scenarios arepresented here to study methodological aspects and not to present predictions aboutthe future, such high energy requirements are considered undesirable as they totallydominate the overall outcomes of OECD-Europe. Therefore, observed changes in thethermal energy intensity and the electricity intensities are reduced for the 1996-2050period (by applying (an additional) square root to the annual changes). The energyuse per unit output still increases for most sectors of region 6 after 1995 withoutresulting in absurdly high energy requirements.

5.3.1 Regional Differences

The growth rates of the parameters changed are varied according to the threealternatives given above. In addition, the parameters involved are varied accordingto the same alternative for each region that is only three scenarios are developed. Inthe first scenario (denoted by ‘sec1’), all parameters are assumed to developaccording to alternative ‘sec1’ for each region. The same applies to the second andthird scenario which are respectively denoted by ‘sec2’ and ‘seccst’. For each region,parameters are varied according to the same alternative to avoid studying andconcluding the obvious. For instance, it can be expected that the outcomes of theaggregate ECCO-model differ from that of the multi-region approach when, for anumber of regions, the desired relative growth rates of non-industrial sectors areassumed to be constant after 1995 whereas that of other regions are assumed toevolve according to the trends observed in the 1985-1995 period.

This subsection presents the results for the main parameters for each of thesubmodels of the ECCO-model for OECD-Europe; production module, consumptionmodule, energy demand and supply module, and the import-export module (see alsofigure 4.1). As the results are presented to study regional differences the shapes of

98

the curves corresponding to a scenario should be compared for the various regions.Substantial regional differences in the shapes of the curves indicate that regionsevolve differently under the same scenario assumptions (i.e. the same alternative),showing the essential advantage of the multi-regional approach. For instance, aparameter may increase in a specific scenario for a number of regions whereas thesame parameter can decrease in the corresponding scenario for other regions. Thesekind of regional differences can be studied in the figures presented in this subsection.Table 5.1 presents an overview of these main regional scenario variables for thetrends observed and for each scenario for the sectors industry and services. From thistable, it can be seen that the scenario variables presented have evolved differently indifferent regions in the 1985-1995 period.

99

Table 5.1: Overview of the energy savings, electricity savings and desired relative growthrates or improvements for the trends observed and for each scenario in case of the sectorsindustry and services.parameter region 1985-1995

(o.r.)scenario

sec1scenario

sec2scenario

seccst

thermal energy savings region 1 2.8 2.8 1.1 0.0

industry region 2 0.7 0.7 0.3 0.0

region 3 3.9 3.9 1.5 0.0

region 4 1.6 1.6 0.6 0.0

region 5 3.4 3.4 1.3 0.0

region 6 -0.2 -0.1 0.0 0.0

OECD 1.7 1.7 0.7 0.0

thermal energy savings region 1 2.1 2.1 0.8 0.0

services region 2 -2.9 -2.9 -1.1 0.0

region 3 3.0 3.0 1.2 0.0

region 4 3.0 3.0 1.1 0.0

region 5 -1.9 -1.9 -0.7 0.0

region 6 6.7 6.7 2.7 0.0

OECD 2.3 2.3 0.9 0.0

electricity savings industry region 1 0.4 0.4 0.2 0.0

region 2 0.0 0.0 0.0 0.0

region 3 2.3 2.3 0.9 0.0

region 4 -1.3 -1.3 -0.5 0.0

region 5 4.0 4.0 1.6 0.0

region 6 -1.8 -1.8 -0.7 0.0

OECD 0.0 0.0 0.0 0.0

electricity savings services region 1 -0.1 -0.1 0 0.0

region 2 -0.8 -0.8 -0.3 0.0

region 3 -0.2 -0.2 -0.1 0.0

region 4 -2.2 -2.2 -0.8 0.0

region 5 -4.7 -4.7 -1.8 0.0

region 6 -4.5 -1.7 -0.7 0.0

OECD -1.6 -1.6 -0.6 0.0

relative desired growth region 1 0.1 0.1 0.0 0.0

rate services region 2 0.2 0.2 0.1 0.0

region 3 0.7 0.7 0.0 0.0

region 4 1.3 1.3 0.5 0.0

region 5 -2.5 -2.5 -1.0 0.0

region 6 0.8 0.8 0.3 0.0

OECD 0.9 0.9 0.3 0.0

100

Scenario ‘sec1’Scenario ‘sec2’Scenario ‘seccst’

Industrial output

year1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

12

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

5

10

1520

25

30

3540

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

12

3

4

5

67

8

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

region 6

2050

in (1985)EJ

Figure 5.12: Industrial output in (1985)EJ for each region and for scenarios ‘sec1’,’sec2’,and ‘seccst’.

Figure 5.12 shows the industrial output for the three scenarios. Similar to theprevious section, industrial output deviates the most in regions 2, 4 and 6. This resultis not very surprising as figures 5.4 and 5.6 already showed that the industrial outputin these regions is most sensitive to changes in the energy demand. The industrialoutput appears to collapse in regions 2, 4, and 6 for scenario ‘sec1’. In addition,industrial output also diminishes slightly for region 5 in scenario ‘sec1’. For allregions, the lowest industrial output occurs in scenario ‘sec1’ and the highest outputin scenario ‘seccst’ although industrial output is almost similar for all scenarios inregions 1, 3 and 5.

The conversion factor of the industrial sector is influenced by energy savings inthe industrial sector as well as the energy savings in non-industrial sectors sincepurchased goods and services are also included in the conversion factor. For eachscenario, the developments of the conversion factor of the industrial sector isillustrated in figure 5.13. For regions 1, 3, 4, and 5, energy savings are tempered asa result of increasing ERE-values. The latter is due to depleting reserves and a higherdependence on imports. In the case of region 2, the industrial conversion factorincreases after 2030 and it finally exceeds its initial value. This increase is mainlydue to the trade-off with the increasing conversion factor of the services sector (seefigure 5.14). In turn, the increasing conversion factor of the services sector is the

101

Conversion factor industry

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

00.20.40.60.8

11.21.41.6

region 6

2050

index


Figure 5.13: Conversion factor for the industrial sector in EJ/(1985)EJ for eachregion and for scenarios ‘sec1’,’sec2’, and ‘seccst’.

result of increasing energy intensities (see table 5.1) and ERE-values. Albeit to alesser extent, the same also applies to region 5 although the conversion factor of theservices sector increases even more strongly in this region (see figure 5.14). In region6, growing thermal energy and electricity intensities were observed for the industrialsector between 1985-1995. Extrapolating these growing trends to the year 2050(alternative ‘sec1’ and ‘sec2’) result in an increasing conversion factor especiallysince the growing energy intensities are associated with growing ERE-values. Forregions 1, 3, and 4, the development in the industrial conversion factor changesaround 2030 in the scenario ‘sec1’ which is mainly due to increasing ERE-values.

Figure 5.14 shows the development of the conversion factor of the services sectorfor each region. For regions 2, 5, and 6, this conversion factor grows substantiallyimplying that much more energy is required to produce a certain output in terms ofutility. As mentioned before the growing conversion factors are mainly due toincreasing thermal energy and electricity intensities (see table 5.1). Note, thedevelopment of electricity saving rates is moderated for region 6 to avoid that theconversion factor increases to very unlikely values. For regions 1, 3, and 4, theconversion factor of the services sector develops similarly to that of the industrialsector.

Figure 5.15 shows the overall primary energy demand of the production sectors(i.e. industry, agriculture, transport and services). The primary energy demandresults, among others, from the development of the output in terms of utility and the

102

Conversion factor services

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5

3

3.5

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

region 6

2050

index


Figure 5.14: Conversion factor for the services sector in EJ/(1985)EJ for eachregion and for scenarios ‘sec1’,’sec2’, and ‘seccst’.

energy savings (note that although the conversion factor of a sector includes dependsconsiderably on the energy savings in that sector, it also includes other aspects). Formost regions, the primary energy demand of the production sectors is the lowest forscenario ‘sec1’. For most regions, these results are consistent with the fact thatenergy savings improve the most in this scenario.

Note that energy intensities of the services sector increase for regions 2, 4, and 5which might be expected to coincide with a high energy demand. However, increasingenergy saving options (table 5.1) are associated with a decreasing industrial outputin terms of utility (see figure 5.12). For these regions, the decreasing energy demandof the industrial sector exceeds the increasing energy demand of the services sector.For region 1, the sharp increase in the primary energy demand of the productionsectors after 2030 is mainly due to the increasing ERE-value of imports based onextrapolation of the observed trends

For most regions, the primary energy demand of the production sectors is thehighest for in case of scenario ‘seccst’ which is consistent with the assumption thatthe energy saving potentials remain constant after 1995. Unlike most regions, theprimary energy demand of production sectors is the highest in region 6 for scenario‘sec1’ which coincides with the growing energy intensities of that region.

Besides the overall primary energy demand of the production sectors, the totalenergy demand is also influenced by the energy demand of households. The primary

103

Primary energy demand of production sectors

year1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

7

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5

3

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

12

14

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

11.5

2

2.53

3.5

4

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

2

46

8

10

1214

16

region 6

2050

in EJ


Figure 5.15: Primary energy demand of productions sectors in EJ for each region and forscenarios ‘sec1’,’sec2’, and ‘seccst’

energy demand of households is mainly determined by energy savings and thedevelopment of the capital stock of dwellings. In turn, the latter is determined by thematerial standard of living (see figure 5.20) and the population growth. Figure 5.16shows that the primary energy demand of households has a somewhat differentpattern than that of the production sectors for regions 5 and 6. For region 5, theprimary energy demand of households is the highest in scenario ‘sec1’ whereas theprimary energy demand of the production sectors is the highest in scenario ‘seccst’(c.f. figure 5.15). The opposite holds for region 6.

The primary energy demand by households varies remarkably among thescenarios. This deviation is the result of differences in the assumption regarding theenergy saving options and the growth levels of the capital stock of dwellings. Forregions 2, 4, and 6, these two effects even magnify each other resulting in strongvariations. For region 5, the energy demand of households increases although thecapital stock of dwellings decreases for scenario ‘sec1’. The higher energy demandis due to the extrapolating the relatively higher energy use by households observedfor the period 1985-1990.

For the other regions, the patterns of the primary energy demand for householdsare rather consistent with that of the capital stock of dwellings. For these regions, theelectricity demand of households increased between 1985-1995 in most regions while

104

Primary energy demand of households

year1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

7

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

24

6

8

10

12

14

16

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5

3

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

7

region 6

2050

in EJ


Figure 5.16: Primary energy demand of households in EJ for each region and for scenarios‘sec1’,’sec2’, and ‘seccst’.

the thermal energy demand decreased in the same period. Both developments have alevelling effect on the total primary energy demand. Therefore, the relatively lowprimary energy demand in scenario ‘sec1’ is not the result of high energy saving ratesbut it is caused by the level of capital stock of dwellings.

Figure 5.17 shows the total primary energy demand resulting from the demand inthe production sectors and in households. The development of the total primaryenergy demand varies substantially among the regions. For instance in region 5 and6, the differences in the primary energy demand of the production sectors and that ofhouseholds compensate each other in the total primary energy demand. For the otherregions, developments in the primary energy demand by household and by theproduction sector show similar patterns. The same obviously applies to the totalprimary energy demand in these cases. With the exception of the regions 5 and 6, thetotal primary energy demand is the highest in scenario ‘seccst’ and the lowest inscenario ‘sec1’.

105

Total primary energy demand

year1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

12

14

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

30

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

02

46

810

1214

16

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

region 6

2050

in EJ


Figure 5.17: total primary energy demand in EJ for each region and for scenarios‘sec1’,’sec2’, and ‘seccst’.

Total resources comprise the proven recoverable reserves as well as additionalresources (see section 4.3). Figures 5.18 and 5.19 show the results of the differencesin the total primary energy demand in the total available resources in the case of coaland natural gas, respectively. The total resources decrease for each fuel type andregion as it is assumed that no additional resources will be found. Clearly, the highestresource depletion rates coincide with the highest primary energy demand as the fuelmixes and the distribution over the regions of origin are assumed constant after 1995implying that for most regions scenario ‘seccst’ is associated with the highestdepletion rates.

The results shown above mostly involve the production sectors. Figure 5.20 showsthe results of the development in the material standard of living. The gross materialstandard of living is an indicator of the level of material wealth of a region and itincludes consumption of goods and services and the primary energy demand.According to figure 5.20, the development of the material standard of living variessubstantially in the different scenarios among the regions. Scenario ‘sec1’ isassociated with the largest variations.

106

Total gas resources

year1985 1995 2005 2015 2025 2035 2045

02040

6080

100120

140160

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

300

350

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

300

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

7

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

300

350

region 6

2050

in EJ


Figure 5.19: Total natural gas resources in EJ for each region and for scenarios‘sec1’,’sec2’, and ‘seccst’.

Total coal resources

year1985 1995 2005 2015 2025 2035 2045

4400

4450

4500

4550

4600

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

6200

6300

6400

6500

6600

6700

6800

6900

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

4045

50556065707580

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

80

90

100

110

120

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

60

80

100

120

140

160

region 6

2050

in EJ


Figure 5.18: Total coal resources in EJ for each region and for scenarios ‘sec1’,’sec2’,and ‘seccst’.

107

Material standard of living

year1985 1995 2005 2015 2025 2035 2045

0

0.20.4

0.6

0.8

11.2

1.4

1.6

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.60.8

1

1.2

1.41.6

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

0.20.4

0.60.8

11.2

1.41.6

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

region 6

2050

index


Figure 5.20: Relative changes in the material standard of living (index) for each regionand for scenarios ‘sec1’,’sec2’, and ‘seccst’.

Region 3 is the only region in which the material standard of living keeps ongrowing in scenario ‘sec1’. Moreover, the material standard of living hardly differsamong the various scenarios for this region which is the result of the similarities inthe output of the industrial and services sectors among the scenarios. These resultscould indicate that the industrial sector of region 3 is rather stable or robust as itappears to be rather insensitive to different growth rates of the non-industrial sector.

In contrast, the material standard of living decreases to about half the level of1985 for regions 4 and 6. For region 2, the material standard of living decreases tothe initial value. The material standard of living decreases to the initial value after atemporary increase for region 5. A decreasing material standard of living isultimately the outcome of the growth rates of industrial output. For regions 2, 4, and6, the decreasing material standard of living corresponds with the decreasingindustrial output (see figure 5.12) In addition, the development of the output ismagnified by a so-called positive feedback loop. This feedback loop involves theprimary energy demand by households, which is part of the material standard ofliving. In turn, the primary energy demand of household is influenced indirectly bythe material standard of living as the primary energy demand is set by the capitalstock of dwellings which itself depends on the material standard of living.

108

Import-export ratio

year1985 1995 2005 2015 2025 2035 2045

0

0.51

1.5

22.5

3

3.54

region 1

2050year

1985 1995 2005 2015 2025 2035 2045

0

0.5

1

1.5

2

2.5

3

region 2

2050year

1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

region 3

2050

year1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

region 4

2050year

1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

region 5

2050year

1985 1995 2005 2015 2025 2035 2045

0

1

23

4

5

67

8

region 6

2050

(1985)EJ/(1985)EJ


Figure 5.21: import-export ratio (total imports/total exports) for each region and forscenarios ‘sec1’,’sec2’, and ‘seccst’.

For regions 1 and 5, the material standard of living starts to decrease after 2025although the industrial output does not decrease substantially (see figure 5.12).However for both regions, the increase in industrial output levels off after 2025indicating that the investments in the industrial sector are decreasing. Since thematerial standard of living is coupled to the investments in the industrial sector, thematerial standard of living also starts to decrease after 2025 in these two regions. Thedeceasing investments in the industrial sector are the result of the increasing energyintensities (or conversion factors) in both regions implying that relatively more outputshould be allocated to intermediate deliveries.

In scenarios ‘seccst’ and ‘sec2’, the material standard of living develops similarlyfor regions 1, 3, and 5. Regions 2, 4, and 6 show more deviations in the developmentof the material standard of living since the industrial output in these three regions ismore sensitive to small changes in some of the main parameters (see section 5.2).

Although the three scenarios were set up for methodological reasons and not topredict future economic activity and the corresponding energy use, the patterns of thematerial standard of living, which is coupled to the rate of investments in theindustrial sector and to consumption, indicates that from an energy perspectiveeconomic activity is more sensitive to small changes in some regions than others. Thedevelopment of the economy in region 3 seems rather insensitive whereas region 2appears to be highly sensitive to small changes. For regions 1 and 3, these outcomesappear to be are rather consistent with the findings resulting from national ECCO-models [Noorman and Crane, 1995].

109

The import-export ratio (total amount of imports divided by total exports)indicates a region’s dependence on foreign resources, in this case resources fromoutside OECD-Europe. Figure 5.21 shows that the development of the import-exportratio deviates substantially for the different scenarios and the various regions. Forregions 1, 2, 4, and 6, the highest import-export ratios coincide with scenario‘seccst’. For regions 3 and 5, the import-export ratio is the highest in scenario ‘sec1’and ‘sec2’. For regions 1, 3, and 5, the development in import-export ratio changesabruptly around 2025 which is mainly the result of increasing ERE-values ofimported energy.

Differences in the import-export ratio are mainly due to increasing imports ofenergy. Not only do increasing imports of energy result form higher primary energydemand but they are also due to depleting reserves especially in the case of oil.Moreover, additional imports required to meet the desired output influences theimport-export ratio in a number of regions. In addition, imports and exports ofproducts also affects the import-export ratio. Both imports and exports of goods andservices are proportional to the industrial capital stock. So at first sight, one mayassume that the import-export ratios of these products should be constant. However,it is the utility value of these imports and exports that is proportional to the capitalstock and not the real energy costs of producing those imports and exports. Clearly,the development of the conversion factor between the real energy costs and the utilitylevel can be totally different for imports and exports (compare, for instance, theregional differences in the conversion factors within OECD-Europe). Hence, theimport-export ratio of goods and services can change over time.

5.3.2 Comparison between Results of Single Region Model and Multi-regionalModel

Subsection 5.3.1 showed the results of regional differences for a number ofregions for three scenarios. Moreover, various parameters are influenced differentlyby the changes in the three scenarios. The next step in studying regional differencesis to compare the aggregated results of the multi-regional with that of the singleregion or aggregate model of OECD-Europe. The aggregated results of the multi-regional approach hold that, for instance, the industrial output of the 6 regions isadded up in order to determine the industrial output of OECD-Europe. The singleregion model involves an ECCO-model in which OECD-Europe is considered as onelarge region (see chapter 4). In this way, the consequences of the regional approachcan be investigated.

In this subsection, the results of the multi-region model are denoted by ‘reg’ andthat of the single region model by ‘agr’. The results are presented for the samescenarios as developed in the previous section which holds that in the first scenario(denoted by ‘sec1’), all parameters are assumed to develop according to alternative

110

year1985 1995 2005 2015 2025 2035 2045

0

20

40

60

80

100

regsec1 regsec2 regseccst

agrsec1 agrsec2 agrseccst

industrial output

2050

Figure 5.22: Comparison between results ofindustrial output (in (1985)EJ) for the threescenarios developed with both models.

‘sec1’ for each region. The same applies to the second and third scenario which arerespectively denoted by ‘sec1’ and ‘seccst’. These three scenarios are developed withboth models.

Figure 5.22 shows that thereare differences between the resultsof the multi-regional model andthe single region model. Theoutcomes of scenarios ‘sec2’ and‘seccst' differ about 10% betweenthe two models by the year 2050.In the case of scenario ‘sec1’, theoutcomes generated by bothmodels differ about 15% by theyear 2050. Although thesedeviations are smaller compared tovariations in the results of thethree different scenarios, adifference of 10% is stillconsidered to be significant here implying that the developments in the industrialoutput differ significantly in the two models. The outcomes of the scenariosdeveloped with the aid of the single region model are somewhat higher than thecorresponding scenarios developed with the multi-regional model.

These differences are most probably mainly the result of differences in the desiredrelative growth rates of the non-industrial sectors as energy saving options appear toinvolve two counteracting influences. On the one hand, energy savings hold that themore can be produced with the same amount of energy implying that energy savingsshould coincide with a higher output in utility terms. It might be expected that moreindustrial output can be allocated to investments when more output can be generatedwith the same amount of energy as presumably more output becomes available.However, energy savings also imply that the total inputs of the industrial sectordecrease and the same, by definition, holds for the amount of output that is at leastin terms of real energy costs. As the room for investments is determined at the levelof real energy costs, a lower output results in a lower room for investments and thusa lower growth of the output. Therefore, industrial output growth is expected to beinfluenced more by desired relative growth rates of the non-industrial sectors. Thistopic is readdressed at the end of this subsection.

Another interesting result is that the outcomes of the industrial outputs of bothmodels already start to deviate slightly right from the beginning although the initialvalues are obviously equal in both models. These variations are most likely the resultof statistical differences. However, it may also be that the whole system evolves insomewhat different directions. In that case, it is hard to point out precisely which

111

year1985 1995 2005 2015 2025 2035 2045

25

30

35

40

45

50



primary energy demand of production sector

2050

Figure 5.24: Comparison between results ofprimary energy demand of the production sectors(in EJ) for the three scenarios developed with bothmodels.

year1985 1995 2005 2015 2025 2035 2045

0.6

0.7

0.8

0.9

1



conversion factor industry

2050

Figure 5.23: Comparison between results ofconversion factor of the industrial sector (inEJ/(1985)EJ) for the three scenarios developedwith both models.

parameters are giving rise to these differences. For instance, not only does the roomfor investments of the industrial sector set the growth rates of industrial output butit also has impact on the consumption level of industrial products. In turn,consumption of industrial products influences the room for investments as a higherconsumption level of industrial products holds that more industrial output is allocatedto consumption reducing the room for investments.

Figure 5.23 shows the resultsof the conversion factor of theindustrial output for wholeOECD-Europe. The conversionfactors differ between the twocorresponding scenarios developedwith the two models. For scenarios‘sec2’ and ‘seccst’, the conversionfactors of the industrial sectordevelop according to a similarpattern although they deviateabout 5%. For scenario ‘sec1’,these conversion factors areremarkably similar for bothmodels until 2030. Then, theystart to deviate as result ofincreasing energy intensities in regions 1, 2, 5, and 6.

Figure 5.24 shows thedevelopment of the total primaryenergy demand of the productionsectors in OECD-Europe for thesix scenarios. As mentionedbefore, note that although theconversion factor of a sectordepends considerably on theenergy savings in that sector, italso includes other aspects. Forscenario ‘seccst’ and ‘sec2’, theresults deviate about 5% by theyear 2050. In the case scenario‘sec1’, the results differ about15% for the two models in 2050.The former does not mean that areno significantly regionaldifferences between the two models as again there are appear to be two counteracting

112

year1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

30

35

40



primary energy demand of households

2050

Figure 5.25: Comparison between results ofprimary energy demand of households (in EJ) forthe three scenarios developed with both models.

relationships. The industrial output is higher in the scenarios corresponding to themulti-regional model compared to that of the single region model (figure 5.22)whereas the opposite holds for the conversion factor of the industrial sector (figure5.23) indicating that the overall energy savings improve more in the scenarioscorresponding to the multi-regional model.

The impact of regionaldifferences is also illustrated byfigure 5.25 which shows theprimary energy demand ofhouseholds. The outcomes showmore deviations for the primaryenergy demand of households thanfor the primary energy demand ofthe production sectors. Theoutcomes are substantially higherfor the multi-regional approachcompared to the single regionapproach. For scenarios ‘sec2’and ‘seccst’, the total primaryenergy demand is respectivelyabout 10% and 20% higher in themulti-regional model than in the single region model in the year 2050. For scenario‘sec1’, this difference is even about 30-40% in 2050.

This effect can be explained by the regional differences in the energy savings ofhouseholds. The energy savings observed for the 1985-1995 period appear to varyconsiderably among the regions. Extrapolating these trends result in somewhat higherenergy demands by households, since the contributions of the less energy efficientregions in the primary energy demand increase relatively. The deviations are also theresult of another influence. The capital stock of dwellings is higher for the multi-regional model than in the aggregated model which is due to differences in thedevelopment in the material standard of living (see also figure 5.26). The primaryenergy demand of households is proportional to the capital stock when energy savingsare excluded. Except for scenario ‘sec1’, the outcomes of the primary energy ofhouseholds are higher for the multi-regional model than for the aggregated model.Hence, energy savings result in higher absolute deviations when the reference energydemand is higher. Although the total capital stock of dwellings in the aggregatedmodel exceeds that of the multi-regional model for scenario ‘sec1’, the primaryenergy demand is still much higher for the multi-regional compared to the aggregatedmodel. This result is due to the extrapolating the regional differences in the energysavings.

Figure 5.26 shows the results of the material standard of living. The regionalrelative growth rates of the material standard of living already start to deviate in

113

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

1.4



material standard of living

2050

Figure 5.26: Comparison between results ofmaterial standard of living (index) for the threescenarios developed with both models.

year1985 1995 2005 2015 2025 2035 2045

0

20

40

60

80

100



total primary energy demand

2050

Figure 5.27: Comparison between results of totalprimary energy demand (in EJ) for the threescenarios developed with both models.

1985. The major differencesbetween the outcomes of the twomodels result from differences inthe growth potentials of the wholesystem. For scenarios ‘sec2’ and‘seccst’, the relative growth ratesin the material standard of livingare about 10% and 25% higher forscenarios corresponding to themulti-regional model than that ofthe single region model in 2050,respectively. In contrast, theseresults are about 15% lower forscenario ‘sec1’ corresponding tothe multi-regional model than thatof the single region model in 2050.For the scenario ‘sec1’, the differences are the result from differences in growthpotentials of the whole economic system. Figure 5.26 also explains the deviations inthe results of the primary energy demand of households for the scenarios generatedby the multi-regional model and by the single region model.

Figure 5.27 shows the totalprimary energy demand of OECD-Europe which is the result ofadding up the primary energydemand of the production sector(figure 5.25) and the primaryenergy demand of households(figure 5.26). The total primarye n e r g y d e m a n d s d i f f e rsubstantially (i.e. ranging from5% for scenario ‘seccst’ to 20%for scenario ‘sec1’ in 2050)between the scenarios developedwith the two models.

Above, it is argued that changing the energy saving options involvescounteracting influences. Hence, figure 5.28 shows the results of keeping the energysaving options constant after 1995 (i.e. alternative ‘seccst’) while the relative growthrates of the non-industrial sectors change according to alternative ‘sec1’ (i.e. thechanges in the growth rates of the non-industrial sectors observed between 1985 and1995 are extrapolated to the period 1996-2050). The scenario obtained is referred toas scenario ‘sec3’.

114

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60

70

80

regsec3 agrsec3

industrial output

2050

Figure 5.28: Comparison between results ofindustrial output (in (1985)EJ) for scenario ‘sec3’developed with both models.

From figure 5.28, it can beseen that the industrial output asdeveloped with the multi-regionalmodel deviates considerably fromthat of the single region model(about 20% in 2050). Thisdeviation is somewhat highercompared to the results shown infigure 5.21 in which thecorresponding deviation (i.e. thatof scenario ‘sec1’) is about 15% in2050. So, as argued before thedesired relative growth rates of thenon-industrial sectors and theenergy savings appear to becounteracting. In addition, these results indicate that these deviations are mainly theresult of differences in the desired relative growth rate of non-industrial sectors.

5.3.3 General Conclusions of Studying Regional Aspects

The room for investments in the industrial sector forms a key parameter in theECCO-models presented above. Not only does it set the growth of the industrialoutput but it is also the balancing term of the model and it may, therefore, besensitive to changes of other parameters. The sensitivity analysis presented in section5.2 showed that intra-regional differences in the structure of a region’s economy mayresult in various growth potentials of the industrial output. Moreover, a number ofregions is more sensitive to deviations than other regions. This especially applies toregions in which the net investments of the industrial sector becomes about zero.

The energy savings and the desired relative growth rates of the output of non-industrial sectors set exogenously are altered according to three alternatives to studythe impact of regional differences. The three alternatives encompass a wide span ofdevelopment potentials to illustrate the spread of the possible deviations. In the firstalternative, the trends observed for the 1985-1995 period are also applied to 2050.In the second alternative, the these observed trends are reduced for the 1996-2050period. In the third alternative, all scenario variables are assumed constant after1995.

The regions that are most sensitive to small changes in a number of parametersalso show the largest variations in the energy savings and the relative growth of theoutput of the non-industrial sectors set exogenously. The impact of the differentscenario assumptions on the outcomes varies per parameter.

The results of a number of main parameters (e.g. industrial output and total

115

primary energy demand) differ substantially among the various regions although thescenario variables are varied, here, according to the same alternative for each region.In a number of cases, these parameters evolve in totally different directions. Thesedeviations are most probably magnified when the scenario variables are assumed todevelop according to different alternatives.

The regional differences in the outcomes of the multi-regional model have asignificant effect on most aggregate outcomes of OECD-Europe in the sense that theresults of corresponding scenarios developed with the two models deviate more than5% for the main parameters. In a number of cases, albeit rather extreme ones, theoutcomes of one region may substantially influence the outcomes of the aggregateregion OECD-Europe. A single region version of the model hardly comes up withthese influences when the same alternatives are selected. Moreover, the outcomes ofthe multi-regional approach may result in more as well as less deviations among thevarious scenarios compared to that of the aggregated model. In addition, regionaldifferences appear to influence the potential growth rates of the aggregated regionsas the impact of rather energy inefficient regions increases over time especially whenthe observed trend for 1985-1995 period is extrapolated to 2050.

Although some of the outcomes shown above may not be very likely to happen,they support the idea that multi-regional models result in more accuraterepresentation of the aggregate region than single region models. Of course, it isdifficult to conclude which method results in the most accurate outcomes when youcompare two methods that both use rather unlikely scenario variables as a startingpoint. However, the preponderance of evidence suggests that taking into account thedifferences among subregions of an aggregate region result in a better description ofthe dynamics than by only regarding the average values of the aggregate region. Inthis perspective, the deviating results of the various scenarios developed by the multi-regional model support this view point.

117

Chapter 6Bridging Part I and Part II

Taking the field of ecological economics as a starting point, energy analysis studiesthe energy flows and stocks associated with economic activity by expressingeconomic activity in embodied energy terms. In this perspective, the ECCO-modellingapproach is introduced to investigate the interaction between economic activity andenergy demand and supply in a system dynamic way. Development potentials ofproduction sectors play a central role within this approach. These potentials mainlydepend on the room for investments in the industrial sector which in turn is mainlyinfluenced by the allocation of industrial output, among others, various demands. Asmaterial consumption is regarded to be a function of among other the industrialsector, production levels as well as consumption levels are mainly determined by theindustrial sector. By emphasising on the production structure of the economy, theECCO-modelling approach is characterised as supply-driven.

Within the increasing economic integration of OECD-Europe (i.e. the introductionof the euro and the liberalisation of energy markets) the interdependency betweennational economies and economic sectors of different sectors is increasingcontinuously. A proper treatment of these international interdependencies almostunavoidably necessitates introducing regionalisation in the modelling approaches.The necessity to develop a regional model also applies to the ECCO-modellingapproach as the industrial structures and the energy supply and demand differ fromcountry to country. Therefore, the first part of this thesis (i.e. chapters 2 through 5)focused on regionalisation aspects within the ECCO-modelling approach. A multi-regional energy-based model of OECD-Europe was developed to study the impactof regional differences in OECD-Europe. The consequences of a full treatment ofregionalisation is studied by comparing the results for two ECCO-modellingapproaches. The first involves an ECCO-model in which OECD-Europe is regardedas one large region. The second consists of a modelling approach involving a multi-regional ECCO-model in which OECD-Europe is divided into 6 subregions. Regionaldifferences are illustrated by varying the energy efficiency improvements and therelative growth rates of the sectors agriculture, transport and services within thedistinguished region according to three alternatives. Scenario results show thatregional differences substantially influence the outcomes of the aggregate OECD-Europe region in particular in a number of extreme cases in which even one subregion dominates the aggregate outcomes. The results presented in chapter 5 supportthe idea that the multi-regional approach results in more detailed outcomes in theECCO-methodology.

118

As mentioned above the ECCO-modelling approach involves a supply-driveneconomy. This assumption is consistent with the theory of Say [van Ierland et al.,1994] perspectives which holds that production is the driving force of the economy.However there are other perspectives on driving forces underlying economicdevelopment in which it is assumed that the consumers are the driving force of theeconomy. Changes in consumption patterns affect the investment rates in theproduction sectors. This prominent role of consumers in directing economic activitycan be stressed by relating economic activity to consumer demands based on thenotion that these production activities take place to meet the demand for consumergoods and service. Note that in case of consumer goods and services, the concept ofembodied energy involves a similar notion, as the embodied energy content ofconsumer goods and services comprises the energy sequestered in the process ofmaking these goods and services, implying that energy costs of production processesare assigned to that good or service. A demand-driven version of the ECCO-methodology can combine the notion of the central role of consumers in steeringeconomic activity with the determination of the embodied energy content of consumergoods and services patterns in dynamic way. Therefore, it is argued that such aversion of ECCO forms an adequate approach to investigate the overall energy costsand the related environmental stress associated with consumption patterns.

A demand-driven version of the ECCO-modelling approach, which is referred toas Dynamic Resource and Economy Accounting Model (DREAM), forms the maintopic of the second part of this thesis (i.e. chapter 7 and 8). The shift from a supply-driven modelling approach to a demand-driven one implies structural changes in themodel. These structural changes can affect the outcomes of the model. The impactof these changes is outlined in the second part of this thesis by comparing the resultsof corresponding scenarios developed with both modelling approaches. Thiscomparison is based on models developed for OECD-Europe (chapter 7 and section8.2). In addition, the impact of different consumption and export growth rates ispresented (section 8.3) and the results of the sensitivity analysis performed are given(section 8.4). Next to these matters, the impact of regionalisation is also studied inthe second part (section 8.5). Besides the models for OECD-Europe, a case study ofthe Netherlands is presented in section 7.5 to illustrate the way the DREAM-modelling approach determines the energy costs and the related environmental stressfor different scenario assumptions about future consumer activity.

119

C.S.

output

left

investments depreciation

Consumption

wealth

fnc+

-+

+

+

-

+

Otherdeliveries

ECCO

Key flows and influences of the Industry

Figure 7.1: Key flows and influences in theECCO-modelling approach. Solid lines indicatephysical flows and dotted lines indicate theinfluences.

Chapter 7Dynamic Resource and Economy Accounting Model

7.1 Introduction

A demand-driven version of the ECCO-model referred to as Dynamic Resourceand Economic Accounting Model (DREAM) is introduced in this chapter to studythe energy demand associated with (changing) consumption patterns. In theDREAM-modelling approach a number of the key feedback loops are changedcompared to the ECCO-approach to facilitate a switch from a supply-driven modelto a demand-driven model.

The main changes as well as the consequences of these changes form the maintopics of part two of this thesis. First, section 7.2 lists some of the basic principlesof ECCO as a reminder. Moreover, the reasons for developing a demand version arediscussed in this section. Basic concepts and model changes are listed in the sections7.3 and 7.4. Results of a Dutch DREAM model are presented in chapter 7.5. Anumber of general remarks concerning the DREAM-modelling approach and theDutch case study are made in chapter 7.6.

7.2 Main Dynamics of ECCO

Compared to the generalECCO-modelling approach, anumber of the main dynamicfeedback loops are changed in theDREAM modelling approach. Inorder to highlight the differencesbetween both approaches, themain dynamics of the ECCO-methodology are briefly repeated.The dynamics presented heremainly involve investments in theindustrial sector and the totalconsumption as these twoparameters are changedpredominantly in shifting fromECCO to DREAM. In the ECCOmethodology, all sector’s input in terms of utility (e.g. number of cars) are assumedto be proportional to the capital stock of that sector. As the total output of a sectoris equal to the total input of that sector, output is, by definition, also proportional tothe capital stock. So, the growth of the output of a sector depends on the growth of

120

the capital stock. Capital stock growth rates are determined by the room forinvestments and play a key role in the growth of the output. Notably, the room forinvestments in the industry (denoted by fnc, i.e. fraction not consumed) is of greatimportance since it is the balancing term in the ECCO modelling approach. It islimited by consumption activities, and other demands such as contributions to non-industrial capital investments, intermediate deliveries and exports (see figure 7.1). Itis, thus, a key indicator as it measures the growth potential of the industrial sector.An example of the allocation mechanism is presented by figure 2.3. To summarise,economic growth, which is indicated by growth of utility output in ECCO, mainlydepends on the room for investments in the industrial sector. Thus, the physicalgrowth potentials of the system depend on the allocation of the industrial output overa number of terms.

One of these terms consists of consumption. The growth of the consumption levelis directly influenced by the investments in the industrial sector. Albeit indirectly,the allocation of the industrial sector’s output towards consumption is directed fromthe industrial sector. Hence, the consumption level is mostly set by the productionsector. The ECCO-modelling approach is, therefore, based on the assumption thatthe economy is supply-driven which means that each output creates its demand(theory of Say; in [van Ierland et al., 1994]).

7.3 A Demand-driven Modelling Approach

This section discusses the reasoning for developing a demand-driven modellingapproach. Moreover, the general concepts are outlined in this section.

7.3.1 Arguments for a Demand-driven Modelling Approach

In chapter 1, the ECCO-modelling approach is introduced to study the societalmetabolism and the associated energy flows through society in particular (see figure1.1). It is stressed above that the dynamics of the ECCO-modelling approach aremainly supply-driven. This supply-driven perspective was generally chosen byenvironmental scientists over the last decades as they tended to emphasize industrialand agricultural production and the associated environmental impacts. In doing so,they followed the traditional approaches of economic historians in whose view lowerprices for industrial goods and rising incomes serve as a stimulus for a growingdemand for products. [Schuurman, 1997]. More and more, consumers are assumedto play an important role in driving the economy. For instance, the theories ofKeynes involve an economy that is demand-driven instead of supply-driven. Hisgeneral theory, following a macro economic perspective, holds that income and thusconsumption decreases as a result of declining investments. In turn, the demand forproducts declines which subsequently influences negatively total production andemployment. Herewith, Keynes disagrees with the theory of Say which holds that

121

every supply creates its own demand (in: [van Ierland et al., 1994]). In addition,McKendrick [1982] pointed out that the industrial revolution was the result of anincreasing demand for consumer goods in eighteenth-century England. Moreover, deVries [1993] concluded that people began to work harder to accumulate moreconsumer goods which resulted in higher production levels. Furthermore, Brewer andPorter [1993] argue that the time is ripe for studies of what societies consume insteadof what they produce. Pietilä [1997] also advocates the central role of householdsin the economy and states that households, as a basic economic unit, form theprimary base of the economy and all other economic functions should serve asauxiliaries. He puts forward that the whole picture will change if we start looking atproduction, trade and economic activities from the household point of view.Christensen [1997] stresses that traditionally policies involve the impact of theproduction process during the whole cycle and therefore the impacts from existinglifestyles are ignored. Moreover, consumption pattern related lifestyles must bechanged towards consumption of goods and services which are produced in moreenvironmentally friendly ways. Noorman and Schoot Uiterkamp [1998] argue thathouseholds influence environmental conditions directly through the consumption ofenergy, the use of materials and the generation of waste. In addition, householdsinfluence environmental conditions indirectly implying that the goods and servicesconsumed are also associated by the use of energy and materials and the generationof waste. The notion of using consumer activity as starting point to studyenvironmental impacts economic activity is also outlined in Biesiot and Moll [1995],Duchin [1995], RIVM [1997], Biesiot and Noorman [1998], and Noorman andKamminga [1998].

Ehrlich and Holdren [1971] also acknowledge the relationship between population(consumers), consumption and environmental stress and introduced an equivalent ofthe following simple relationship which is referred to as the IPAT-equation:

In equation (7.1), the increment of stress on the environment is the result ofpopulation change, per capita consumption change, and technology change.

Biesiot and Noorman [1998] link consumer activities, centred around households,to, among others, energy inputs and outputs in the economy and to the associatedenvironmental impacts. Hence, all inputs and outputs in the production sector, withthe exception of those involving exports, can be assigned to consumption byhouseholds as most output finally ends up in either domestic or foreign consumergoods and services. Taking the prominent role of households into regard, the societalmetabolism concept described in chapter 1 and illustrated by figure 1.1 can bechanged into figure 7.2. Figure 7.2 illustrates the way the impact of economic activityare attributed to consumption by households.

Impact = Population * Affluence * Technology (7.1)

122

Waste and Metabolic Leaks

Natural Resources

Production ofGoods & Services

EnergySupply

WasteManagement

Figure 7.2: Concept of Societal metabolism in whichHouseholds play a central role as industrial activity andthe energy use associated are all assigned to consumptionby households in this figure. Derived from [Noorman etal., 1998]

The concepts presented above are conceptualised in a research program calledHOMES (HOuseholds Metabolism Effectively Sustainable) that was initiated tostudy the physical throughput of energy and materials through households [Noormanet al., 1998]. The aim of this program is to develop and apply the concepts,operational approaches, methodologies and instruments relevant for the diagnoses,evaluation, and change options of household metabolic rates in the Netherlands [ibid].Or more specifically, it addresses the relationships between trends in consumptionand the consequences for spatial and environmental quality, for socio-cultural aspectsand for natural resources. In this way, the program takes the consumption side ofeconomic activities as a starting point to study the relationship between economicactivity and the environment. The program relies on the notion that consumeractivities (centred around households) can be linked through integral assessments andlife cycle analysis to the complex patterns of inputs and outputs of the economy andto the associated environmental loadings [ibid]. Households are considered as basicconsumption units.

Within the HOMES-research program, measuring the consumption patterns interms of energy is utilised as a mean towards understanding how to direct themtowards sustainable objectives. The demand for natural resources (energy inparticular) is not only determined by the number of households and the consumptionper household but is also a function of biophysical, economic, technological, spatialand behaviour aspects [Biesiot and Noorman, 1998]. Among others, the dynamics oflifestyles play a key role in these studies.

123

The energy throughput through households is determined by assessing the energycosts of household expenditures. Assessing the energy content of householdsexpenditures involves a hybrid method of energy analyses that is a mixture ofprocess analyses and input-output analyses [Wilting, 1996]. Process analysis isfrequently used to determine the energy intensity of materials and input-outputanalysis is used to compute the energy intensities at the sectoral level. The input-output methodology was already outlined in chapter 3. The methodologies involvedhave a quasi-static or semi-dynamic character implying for instance that some energyefficiency improvements which will probably be implemented in the productionsectors are taken into account to predict changes in the indirect energy use ofhouseholds [Noorman and Moll, 1998]. This quasi-static character can be regardedas a shortcoming in studying the long-term consequences of changing consumptionpatterns as this approach assumes that the system considered is almost in a steadystate. In this perspective, the ECCO-modelling approach seems to be appropriate tostudy consequences of changing consumption patterns at a higher level of aggregationand in a (full) dynamic way. Not only can the consequences of changing consumptionpatterns be assessed in this way but also the changes in the production can be takeninto account.

Therefore, a demand-driven version of ECCO (called DREAM) is developed withwhich these studies can be performed. In this perspective, Ryan et al. [1998] andSchembri [1998] also developed a more demand-driven modelling approach based ona ECCO-type model as they also note the important role of households and theirconsumption activity in the economy. Applying the concepts of HOMES to theEnergy Accounting approach holds that all the primary fossil energy is extractedfrom the physical environment in order to produce goods and services to meet theneeds, wants and aspirations of the population (or households). The goal of theDREAM-modelling approach is to calculate dynamically the total energy costs ofnational consumption. As a consequence , the purpose of the DREAM-modellingapproach has somewhat shifted away from that of ECCO-models which aredeveloped primarily for studies of production potentials. Besides final domesticconsumption, also exports are included in the DREAM-modelling approach. Mostof these exports also ultimately end up as consumer goods but are consideredseparately as it is too complex to assign these exports to foreign or domesticconsumption. So in the DREAM-modelling approach, the overall energy costsassociated with certain consumption and exports levels are computed by linking themto the primary energy costs of the production of the goods and services involved. Thisapproach also involves an energy accounting approach and in addition it also uses theconcepts of input-output analysis. From this perspective, it is not really differentfrom ECCO although several key feedback loops differs in the two approaches. InECCO, all sectoral outputs were eventually allocated over consumption exports andinvestments whereas in the DREAM-modelling approach the sectoral output requiredis based on desired consumption and export levels. The computation of the total

124

energy costs can best be compared to the derivation of the Leontief inverse (seechapter 3) as the Leontief inverse can, by definition, be used to determine thechanging direct and indirect inputs of a changing final demand. The Leontief inversecan only be used in static analyses while the DREAM-modelling approach assessesthe energy costs related to consumption and export level in a system dynamic way.

One of the topics of this part of this thesis is to compare from a methodologicalpoint of view the results developed with the DREAM-modelling approach with thatof the ECCO-modelling approach. The ECCO-modelling approach describes theeconomy at a macro/meso level. Hence, the DREAM-methodology approachpresented here describes also the energy content of household expenditures atmacro/meso level in order to be consistent with the ECCO-modelling approach.Changes in household expenditures, therefore, mainly focus here on growingconsumption levels. The models and associated results presented in this part are lesssuitable for assessing the energy costs of (qualitative) shifts in householdsconsumption patterns. Such studies require data availability at a much more detailedlevel. However, such studies are very interesting for future research.

As stressed above, the demand-driven approach discussed in this chapter involvesdifferent balancing terms than the ECCO-model. Compared to the ECCO-approach,some of the key feedback loops are drastically changed in the DREAM-modellingapproach (the main changes are outlined in section 7.4). The major feedback loopsor influences of the DREAM-modelling approach are described below.

7.3.2 Major Concepts of the DREAM-Modelling Approach

The goal of the DREAM-modelling approach is to determine the energy costs ofchanging consumption levels in a dynamic way and therefore consumption growthrates are considered to be scenario variables (i.e. these growth rates are setexogenously). Growing consumption levels will result in growing productions levelseither domestically or in foreign economies. Capital stock of the production sectorswill also have to increase in order to meet these growing consumption levels. In thisway, consumption levels determine a desired level of capital stock which determinesthe amount of capital needed to produce the output required to satisfy the desireddemand for consumer goods and services and exports. Differences between thedesired capital stock and the existing capital stock set the rate of investments. Thus,investments in the production sector are set by consumption growth rates and act nolonger as the balancing term. The amount of industrial output produced domesticallythat can be allocated to consumer goods now forms the balancing term. Hence, thefraction not consumed (see figure7.3), which represents the room for investment inthe ECCO-methodology, is replaced by a fraction available for domesticconsumption. So, the allocation of industrial output produced domestically is nolonger balanced by the room for investments but it is by balanced by the room for

125

consumption. In case domestic production does not satisfy the domestic demand,imports balance domestic supply and demand. The key flows and influences of theindustrial sectors are presented for the DREAM-modelling approach as well as forthe ECCO-approach in figure 7.3.

From the two figures, it can be concluded that one of the most striking differencesbetween both models is the absence of a negative feedback loop in the DREAM-modelling approach. Therefore, the DREAM-model can be considered to be a growthmodel. Herewith, the DREAM-approach are more consistent with common economicmodelling approaches in the sense that the DREAM-approach is also based onextrapolating historical trends. The substantial change in the structure of theDREAM-approach compared to that of the ECCO-approach may have a strongeffect on the outcomes of the model as the growth of the production sector no longerappears to be decelerated endogenously. Hence, it is less likely that the productionsector will face a severely declining output indicating that the environment may setlimits to the industrial growth.

In the DREAM-modelling approach, imports increase if the level of consumptioncan no longer be met by domestic industrial production. These imports may beregarded as a kind of loans or foreign investments as these imports allow theeconomy to grow in the sense that these imports do not restrict the investments in thedomestic sectors. A number of ECCO models also have the option to include loans(see among others [Crane, 1995; Foran and Crane, 1998]). These loans can amongothers be used to expand industrial growth. This borrowing feeds into the total debts.Repayments of these debts also include interest rates accrued on the debt [Crane,1995]. Herewith, these models incorporate the common practice of getting a loan in

CS

output

left


Consumption

+-

+

+

+

-

+

Otherdeliveries

ECCO


fraction not invested

Wealth

C.S.

output

left


consumption

+

+

+

+

+

otherdeliveries

DREAM

desired investments

+

desiredconsumption andexport levels

-imports+


Figure 7.3: Comparison between the key flows and influences of the DREAM-modellingapproach and of the ECCO-approach. Solid lines indicate physical flows and dashed linesindicate the influences.

9 The return on investments (ROI) of a simple project is determined as follows:

ROIZ

Inv rt

ttt

n=

+=∑0

0 1*( )

In this equation, Zt represents the annual average yield of the project, Invt0 indicates thetotal investment costs. In addition, n denotes the life time of the project and r thediscount rate (based on [Bouma, 1988]).

126

order to invest under the assumption that the return on investments exceeds one9.However, a number of models (e.g. the Dutch ECCO-model [Noorman, 1995] andthe ECCO-model for OECD-Europe as presented in chapter 4) does not have theoption to issue loans to increase investments. The reasons for not including loans canbe found in the fundamental laws of thermodynamics. In issuing loans capital iscreated and therefore one might get the impression that energy is ‘created’ too. As aconsequence of not issuing loans, scenarios developed with the aid of these modelsmight show less economic growth than observed in reality. The topic of loanscomplicates the study of determining the all-inclusive energy costs of a number of(consumption) growth paths with the aid of these models. In DREAM, the matter ofissuing loans has been dealt with by adjusting the balancing term implying that theadditional imports which balance supply and demand can be regarded as a kind ofloan as it involves foreign or strange capital.

7.4 Detailed Overview of the DREAM-Modelling Approach

In section 7.3, the main differences between the ECCO and the DREAM-modelling approach are described in more or less general terms. This section presentsa more detailed overview of the parameters which are changed in theDREAM-modelling approach compared to the ECCO-modelling approach of OECD-Europe as presented in chapter 4. A complete overview of the DREAM-methodologycan be obtained by combining the listing below with the detailed description of theaggregate ECCO-model for OECD-Europe in chapter 4. This section describes asingle region model for OECD-Europe. As might be expected the changes mainlyinvolve investments, consumption and the import-export balance. Below, influencesare represented by dashed lines and flows by solid lines. The equations changedsubstantially are listed in Appendix I.

7.4.1 Investments

In the DREAM-modelling approach, the rate of investments of a sector dependson a desired capital stock (Dcsut[s]) implying that investments are demand-driven.Investments are, therefore, determined differently in the DREAM-modelling approachcompared to the ECCO-modelling approach. The demand for capital is influenced by

127

the desired output of a specific sector(dout[s]), that is the demand forcapital is assumed to be proportionalto the amount of output (cf. diagram7.1). The desired output is allocatedover the items: exports, total(intermediate) deliveries to sectors(tintout[s]), deliveries to the finaldemand (del2fc[s]), and to the rate ofcapital formation (del2rcf[s]).Deliveries to the final demand involvethe contribution of that sector to thefinal consumption by households orgovernment whereas deliveries to therate of capital formation involve thecontribution of a sector’s output to thetotal investment rate. Intermediatedeliveries to sectors are determined bythe demand for input in the sector ofdestination as in each sector the inputsare proportional to the capital stock(cf. subsection 4.2.1). The sameapplies to deliveries to finalconsumption and to exports asconsumers set consumption levels. Inaddition, the deliveries are alsodetermined by the sectors ofdestination as growth of the sectorsinvolved regulates the total rate ofcapital formation and thus the extentof the total deliveries to capitalformation.

7.4.2 Consumption

The shifts in the balancing term have made the consumption sector more complexthan in ECCO. The consumption level (consut), which is expressed in terms of utilitydepends on the the relative population growth (popf) and the growth rate ofconsumption per capita (grwthconsut) of which the latter is set exogenously (cf.diagram 7.2). The total consumption level is assigned to the sectors of origin byintroducing a factor. In other words, the demand for consumption (dcons) is coupledto a demand for goods and services in the production sectors. These production

dout[s]

rcf[s] rdc[s]CS[s]

export[s]tintout[s]del2fc[s]del2rcf[s]

++Dcsut[s]

+ -

+

Cs[s] Capital stock of sectors

rcf[s] Rate of capitalformation (investments)of sector s

rdc[s] Rate of capitaldepreciation of sector s

Dcsut[s]Desired capital stock of sector sdout[s] Desired output of

sector sexports[s] Exports from sector stintout[s] Total intermediate

deliveries of sector s toall sectors

del2rcf[s,a] Deliveries to rate ofcapital formation ofsector s

del2fc[s] Deliveries to finalconsumption of sector s

Diagram 7.1: Investments

128

sectors can either be domestic orforeign implying that the consumptionlevel can be met by domestic deliveriesor by imports. The ratio of domesticdeliveries (ctrbincons, incons) toimports (ctrbmcons, mcons) needed tomeet the consumption level is setexogenously as a starting point.However, it remains to be seenwhether this demand for goods andservices domestically produced can bemet by the output of the associatedsector as consumption is the balancingterm in the DREAM-modellingapproach. In other words, the startingpoint results in a desired demand forgoods and services domesticallyproduced (dcons). The room forconsumption determines whether or notthis desired demand can be completelymet by the domestic sectors. The roomfor consumption is computed in asimilar manner as the room forinvestments in the ECCO-modellingapproach which holds that the room forconsumption (tacons) depends on theoutput of a sector (out) and to theallocation of this output to exports(exports), total intermediate deliveries(tintout) and investments. The latter isdealt with by introducing an allocationfactor called ‘fraction available forconsumption (fac)’ which is similar tothe ‘fraction not consumed’ describedin section 4.2. The fraction availablefor consumption depends on theavailable output (out) and on theinvestments rates required (rcf). Theroom for consumption (tacons) issufficient when it is not smaller thanthe desired demand for goods andservices domestically produced

incons[s,ac1] addimp[s,ac1]

consut[s]

dcons[ind,ac1]

grwthconsut[s]

tacons[s,ac1]

fac[s]

rcf[s,ac1]

export[s,ac1]tintout[s,ac1]

out[s,ac1]

++

+

+

-+

-

+-

+

+

mcons[s,ac1]

+

popf

+

incons[s] consumption ofproducts stemming fromdomestic sector s

addimp additional importsrequired to meet thedesired consumptionlevel

dcons[s]desired consumption of productsstemming from sector s

consut[s] consumption ofproducts stemming fromsector s in terms oftheir utility value

grwthconsut growth of consumption of products stemmingfrom sector s in termsof the utility value

mcons[s] imported consumergoods stemming from aforeign sector s

tacons[s] total domestic outputavailable forconsumption

fac[s] fraction available forconsumption

out[s] domestic output ofsector s

exports[s] exports from sector stintout[s] total intermediate

deliveries of sector s toall sectors

rcf[s] Rate of capital

Diagram 7.2: Consumption

129

(dcons). In this case, the initial allocation of consumption over imports (mcons) anddomestic production (incons) is maintained. When this desired level of consumptionexceeds the room for consumption, the additional demand for consumption is met byadditional imports (addimp) which holds that consumption is balanced by anincrease in imports in the DREAM-modelling approach (see below).

7.4.3 Balance of Trade

The import export balance (impexp-balance) is equal to the differencesbetween the total imports(totimports) and the total exports(totexports) (cf. diagram 7.3). Theimportance of the import-exportbalance was already emphasised inchapter 4.5. The import-exportbalance is even more essential to theDREAM-modelling approach asimports are a balancing term.Consumption and exports whichcannot be met by domestic output aremet by additional imports (addimp).As mentioned in subsection 7.4.2,these imports are additional toimports of goods and services by theproduction sectors (imp), importsassociated with consumption(mcons), and the imports of primaryfuels (pfimports). So, additionalimports balance the desired output ofsector (dout, see 7.3.1) and theactual output of a sector (out).Exports and imports are linkeddirectly. The level of exports ofgoods and services is based on agrowth rate (grwthexp) which is setexogenously. Besides goods andservices (exports), exports also

comprise energy sources (pfexports) which depends on the domestic energy supplyin that country (see section 4.3).

impexpbalance

totexports mcons

-

+ imp

totimports

exports pfexports+

+ ++

+

++

pfimports

grwthexp

+ +

total primaryenergy supply

CS productionsectors

addimp

+

+dout

+

out

-

impexpbal import-export balancetotimports total importsimp imports of goods pfimports imports of primary

energyaddimp additional imports

required to meetdesired consumptionlevel

out domestic outputmcons imports of consumption

goodsdout desired domestic outputCS Capital Stocktotexports total exportsexports exports of goodsGRWTHEXP export growth rate (set

exogenously)pfexports exports of primary

energy

Diagram 7.3: Balance of trade.

130

7.5 Case Study of the Dutch DREAM-model

Results of scenarios developed with the DREAM-model for OECD-Europe arepresented throughout chapter 8. In addition, these scenario results are compared tothat of corresponding scenarios developed with the ECCO-model of OECD-Europe.Before presenting these scenario results, this subsection presents the case study ofdeveloping and applying a DREAM-model for the Netherlands (denoted byNLDREAM). This case study serves to illustrate how the energy costs associatedwith future consumption growth rates can be studied by using the DREAM-modellingapproach. The ECCO-model, which Noorman [1995] developed for the Netherlands(NLECCO), forms the basis of NLDREAM. A number of the basic principles ofNLECCO are outlined in chapter 2 and 4. Noorman [1995] presents a detailedoverview of NLECCO.

7.5.1 From NLECCO to NLDREAM

Sections 7.3 and 7.4 described the consequences of shifting the model from asupply-driven to a demand-driven perspective. Besides shifting the model’sperspective from supply-driven to demand-driven, a number of other features arechanged in NLDREAM compared to NLECCO. This subsection describes thesechanges briefly. Appendix I presents a detailed overview of these model changes.

Households are included in NLDREAM and direct consumption of goods andservices and private transportation. Five types of households are distinguished inNLDREAM: one person households, two persons households, three personshouseholds, four persons households and households with five or more persons.Within the household groups the same age distribution is used as in NLECCO (i.e.young people (19), young adults (20-44), major adults (44-64), and elderly people(>65)) [CBS, 1986,1996].

In NLDREAM, private transportation is included differently in comparison to theprocedure in NLECCO. In NLECCO, private transport is influenced by the materialstandard of living whereas in NLDREAM it is determined by the development in thecomposition of households. Herewith, NLDREAM uses the aggregate results of thestudy of Schenk [1998] who describes the private transportation sector at a verydetailed level. Schenk calculates the emissions associated with private transport bydistinguishing five related subsystems in which the driving forces specific for thatmodule are determined at a detailed level; population module, households, car stock(ownership), kilometres driven per household and car emission module. From theseresults, average growth rates are derived for the kilometres driven per householdstype and the car ownership per household type.

Energy conservation is also altered in NLDREAM. In NLECCO, autonomous andadditional energy savings are distinguished. Autonomous energy savings are assumedto take place without the requirement of additional investments whereas additional

131

energy saving require investments implying that additional capital stock must bedeveloped. In NLECCO, the extent of the additional energy savings depends on thelevel of capital stock developed in the energy conservation sector. Moreover, theintroduction of both types of energy savings is set exogenously and is based onMelman et al. [1990]. In NLDREAM, energy savings are divided in efficiencyimprovements and structural changes. Both types of energy savings are setexogenously. A part of both energy saving options can take place autonomouslywhereas the other part of the energy saving options require investments. So unlikeNLECCO in NLDREAM, the extent of additional energy savings set the demand forthe level of capital stock required to enable these energy savings. In many cases, theenergy saving options introduced may not agree with the energy saving optionspresented by Melman et al. [1990] as in some other studies (cf. [de Beer et al., 1994;Hilten, et al., 1996]) other energy savings options are presented. These differentenergy saving options will most likely not correspond with the associated capitalrequirements presented by Melman et al. [1990]. Despite the possible inconsistences,the capital requirements of Melman are still used in order to take into account thenotion that a part of energy savings requires additional investments.

7.5.2 Starting Points of Scenarios

NLDREAM is introduced as a model developed in order to investigate the overallenergy costs associated with household consumption patterns in a system dynamicway. Four scenarios are developed here to illustrate the potentials of NLDREAM aswell as its limitations. Two of these scenarios are consistent with the EuropeanCoordination scenario (EC) and Global Competition scenario (GC) as developed byCPB [1996, 1997]. Both scenarios can be considered as two variations on a‘business-as-usual’ scenario. The former assumes a strong European Communityinvolvement. The European economy grows considerably together with the Asianeconomy while the economy of North America lags behind. The level of technologydevelops strongly. Consumption patterns become more immaterial andenvironmentally friendly. The GC-scenario involves an economy which growsglobally at a high rate. This scenario also assumes a considerable development intechnology but the developments are mainly market orientated. Consumption patternsare associated with a high level of product differentiation.

It would be interesting to compare these two ‘business-as-usual’ scenarios witha sustainable development scenario. Simplifying its concepts, sustainabledevelopment is regarded only from an energy perspective and involves an economyin which the energy demand is totally met by renewable energy and the energy.

10 The target energy service use of 1-1.5 kW per persons equals a maximalannual energy demand of 30-50 GJ per person.

132

demand does not exceed the 1-1.5 kW per person10. The latter is conceptualised byDürr [1994] and it is also applied by Biesiot [1998] and Mulder and Biesiot [1998].The target of 1.5 kW per person to be met in 2050 is based on the maximal amountof renewable energy which can be produced per person if one assumes full globalequity in energy service terms (i.e. every person in the world uses the same amountof energy). This target energy demand involves both direct energy use and indirectenergy use. In case of the Netherlands, Biesiot [1998] shows that such a long-termsustainability objective requires structural changes in the current consumptionpatterns and thus in the whole economy. Potential energy savings options alone areinsufficient to reduce the current Dutch energy consumption level of about 5-6 kWper person to the target level of 1-1.5 kW per person. Such a ‘sustainable’ scenariocannot be developed with the version of NLDREAM described above since changesin consumption patterns can hardly be studied with the model in its present statebecause the production sectors are described at a too higly aggregated level (i.e ata macro level instead of a micro/meso level). Therefore, two ‘semi-green’ scenariosare developed to illustrate the differences between ‘business as usual’ and a transitiondirected towards a more sustainable future. Both ‘semi-green’ scenarios assume arenewable public electricity supply by the year 2020. In addition, the energy savingoptions follow the maximum saving potentials as presented by ICARUS-3 [ de Beeret al., 1994]. In the first semi green scenario (SG1), consumption and export growthrates are assumed to be similar to the EC-scenario which is consistent with theassumptions of these scenario improvements. In the second semi-green scenario(SG2) are coupled to the growth rates presented in the Divided Europe (DE) scenario.The DE-scenario is also introduced by CPB [1996, 1997] and involves a morepessimistic scenario from a macro-economic point of view implying lower economicgrowth rates. The basic scenario variables of the four scenarios are listed in table 7.1.The differences in the energy saving potentials between scenarios SG1 and SG2 arethe results of differences in structural changes (derived from EC and DE,respectively).

133

Table 7.1: Average annual growth rates and improvements (in %) of driving exogenousfactors of the presented scenarios for the period 1995-2020. Thermal energy savings andelectricity savings include efficiency improvements as well as structural changes. Theobserved annual growth rates are also presented for the period 1985-94.

1985-94 EC GC SG1 SG2

consumption volumea 2.5 2.7 3.2 2.7 1.7

export volume: agriculturea 6.7 3.0 3.4 3.0 -0.5

industrya 5.9 4.3 5.2 4.3 2.3

transporta 4.7 3.5 4.4 3.5 2.7

market servicesa 4.7 5.0 6.2 5.0 3.1

non-market servicesa 4.7 4.1 4.9 4.1 2.8

thermal energy savings: agricultureb 0.5 1.7 1.2 1.0 1.8

industryb 2.9 1.5 2.7 5.4 4.5

transportb 2.1 1.5 1.7 3.8 3.6

market servicesb 2.3 1.3 1.9 1.7 1.5

non-market servicesb 0.2 0.7 0.8 0.6 0.9

electricity savings: agricultureb 0.5 1.0 0.7 1.4 1.6

industryb 1.5 1.0 1.5 5.2 4.9

transportb - - - - -

market servicesb 1.6 0.9 1.6 2.6 1.9

non-market servicesb -0.7 0.3 0.5 1.5 1.3

thermal energy savings: householdsb 0.6 0.0 0.6 0.7 0.1

electricity savings: householdsb -3.0 -2.0 -2.0 0.4 0.4

Fuel mix public electricity supply: fossilc,d 0.93f 0.88g 0.88g 0.00g 0.00g

renewablec,d 0.07f 0.12g 0.12g 1.00g 1.00g

total capacity of cogeneration in GWc,e 1.3f 17.4g 14.8g 16.5g 16.5g

a Data for 1985-1995 are derived from [CBS, 1987, 1997]. Except, data for export growth ratesof transport, market services, and non market services are estimated by combining OECDstatistics [1997] with Eurostat statistics [Oosterhaven and van der Linden, 1995; van der Linden,1999]. Data for the four scenarios are derived from [CPB, 1996, 1997]; b Data for 1985-1995 are determined by using [OECD, 1991a-b, 1996,1997]. Data for EC andGC-scenarios are derived from [CPB, 1997; ECN, 1997, 1998] and data for SG1 and SG2-scenarios are derived from [de Beer et al.; 1994].c Data for 1985 derived from [Noorman, 1995]; Data for the GC and EC-scenarios are derivedfrom [ECN, 1998]

d Data involve shares instead of growth rates; e Data involve absolute values instead of growthrates f Data for the year 1985; g Data for the year 2020.

134

year1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

7

8

GC EC

SG1 SG2

total consumption

2050

Figure 7.4: Total consumption level in (1985)EJfor the scenarios GC, EC, SG1 and SG2.

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60

GC EC

SG1 SG2

total exports of goods and services

2050

Figure 7.5: Total exports from the industrial sectorin (1985)EJ for the scenarios GC, EC, SG1 andSG2.

7.5.3 Scenario Results

The growth rates presented intable 7.1 are applied to the year2050 to develop scenarios whichcover the 1985-2050 period. Figure7.4 shows that the totalconsumption grows considerably inall scenarios as result of thisassumption. For the GC-scenario,the total consumption level is morethan eight times as high in 2050compared to 1985. Even in the caseof SG2-scenario, the totalconsumption level more than doubles in the same time period. The EC-scenario andthe SG1-scenario have similar consumption levels and end up with having a level sixtimes as high compared to 1985.

Compared to the consumption level, the total export levels even increase at amuch higher rate as shown infigure 7.5 (see scenarioassumptions presented in table7.1). Expanding these rates to theyear 2050 results in extremeexport levels in the case ofscenario GC where the exportlevels are about 25 times higherin 2050 than in 1985. Also in thecase of scenarios EC and SG1,the total export levels areextremely high as they are morethan 15 times as high in 2050 asin 1985. Even in 2020, the exportlevels are about 4-5 times higherfor the GC, EC and SG1scenarios compared to 1985. In addition for scenario GC, total exports are 7 timeshigher than the consumption level in 2050 whereas they are twice as high in 1985.

In section 7.3, it is argued to assign the energy use in the production sectors to theconsumption by households. One can argue whether or not exports should also beassigned to the consumption of households. One can state that most exports end upas consumer goods for foreign households and thus the associated energy costsshould be assigned to those households. One can also argue that most of these exportscompensate the imports of, among others, energy carriers. Therefore, these exports

135

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

GC EC

SG1 SG2

total energy consumption per housholds

2050

Figure 7.7: Total energy (direct and indirect)consumption per household in kW for thescenarios GC, EC, SG1 and SG2.

year1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

GC EC

SG1 SG2

total energy consumption per person

2050

Figure 7.6: Total energy (direct and indirect)consumption per person in kW for the scenariosGC, EC, SG1 and SG2.

can be assigned to Dutch households. However, the latter may involve a kind ofsnowball effect as increasing exports to balance imports is associated with anincreasing output of the production sectors implying growing imports which alsohave to be balanced.

Figure 7.6 shows the totalenergy consumption per personwhereas figure 7.7 presents thetotal energy consumption perhousehold. In both cases, the totalenergy consumption includes thedirect energy demand and theenergy content of consumer goodsand services. The energy contentof exports are not taken intoaccount. The energy content ofconsumption of common goodssuch as infrastructure, schoolingand medical care should beincluded in the consumption ofgoods and services. These collective goods and services comprise about 30% of thetotal consumption of goods and services [van Engelenburg et al.,1991; Kramer et al.,1998].

In figures 7.6 and 7.7, theenergy consumption per capitaabout stabilises for SG2-scenariowhile it increases considerably inthe case of the GC-scenario. Thetotal energy consumption perhouseholds more than triples forthe GC-scenario between 1985 and2050 and about doubles for theSG1 and EC-scenarios. Besides thestructural changes, the SG1-scenario is associated with a highlevel of efficiency improvementand a renewable public electricitysupply after 2020. These energysavings and renewable public energy supply are exceeded by the consumption growthrates resulting in the increase in the energy use in SG1-scenario. The energyconsumption by households only remains about the same for SG2-scenario whichinvolves a lower consumption growth rate (1.7. for SG2 compared to 2.7 for EC and

11 In 1985, export levels and consumption level were respectively about 2.0(1985)EJ and 1.0 (1985)EJ compared to the total production of 4.5 (1985)EJ.

136

year1985 1995 2005 2015 2025 2035 2045

0

2

4

6

8

10

12

GC EC

SG1 SG2


2050

Figure 7.9: Primary energy demand of theNetherlands in EJ for the scenarios GC, EC, SG1and SG2.

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60

70

80

GC EC

SG1 SG2

total output production sectors

2050

Figure 7.8: Total output of all the productionsectors in (1985)EJ for the scenarios GC, EC, SG1and SG2.

SG1-scenarios). The results shown in figure 7.7 indicate that the loweringconsumption growth rates are necessary to decrease the energy consumption perhousehold.

Figure 7.8 shows that the totalproduction (in terms of the utilityvalue) increases substantially as aresult of the growing consumptionlevels and especially of thegrowing export levels. For theGC-scenario and the SG2-scenario, total industrial outputincreases at an average annualgrowth rate of about 4.5% and1.8%, respectively. The scenariosEC and SG1, which have similarexport growth rates andconsumption growth rates,increase at an average annualgrowth rate of about 3.5%. In 1985, total export and the total consumption togethercomprised about 66% of the total output11 in the year 1985 implying that about 34%of the total output is allocated to intermediate deliveries or investments. By 2050,about 80% of all output is allocated to the exports and consumption. This change isthe result of the relatively larger contribution of exports.

Figure 7.9 shows the total primary energy demand of the Netherlands. Not onlydoes the primary energy demandresult from these developmentsof the output of the productionsectors but it also includes theprimary energy demand byhouseholds. The productionsectors contribute the most to theprimary energy demand. For theGC and the SG1-scenario, thiscontribution increases fromabout 70% in 1985 to 88% in2050. For the EC-scenario, thecontribution of he productionsectors to the primary energy

137

year1985 1995 2005 2015 2025 2035 2045

0

0.002

0.004

0.006

0.008

0.01

GC EC

SG1 SG2

perel

2050

Figure 7.10: Primary fossil energy requirement forgenerating electricity (perel) in GJ/kWh for thescenarios GC, EC, SG1 and SG2.

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

GC EC

SG1 SG2

total carbon dioxide emissions

2050

Figure 7.11: Total carbon dioxide emissions forthe scenario in Tg C for the scenarios GC, EC,SG1 and SG2.

demand increases from 70% to 85% while this contribution remains about the samefor scenario SG2. For each scenario, the primary fossil energy demand of theproduction sectors increases at a lower rate than the total output of the productionsectors. These lower growth rates are clearly the result of energy savings. Theprimary energy demand of production sectors involves average growth rates of 2.2%and 1.4% for the GC and EC-scenarios, respectively. For scenarios the SG1 andSG2, the corresponding growth rates are about 0.6% and -0.5% respectively.Compared to 1985, the primary energy demand is about 45% higher for scenarioSG1 in 2050 whereas it is about 30% lower for scenario SG2 in that year.

So in scenario SG1, the fossil primary energy demand is still increasing despitethe high level of energy savings and a public electricity supply which is fully basedon renewables. Only for scenario SG2, the total primary energy demand decreasesslightly as this scenario involves alower consumption growth rate(1.7%) besides the high level ofenergy savings and a publicelectricity supply which is fullybased on renewables. Forscenarios GC and EC, the fossilprimary energy demand is about4.5 to 2.5 times higher in 2050than in 1985, respectively.

Figure 7.10 shows thatprimary fossil energy requirement(in GJ) for producing one kWh ofelectricity decreases somewhatdue to efficiency improvementsfor scenarios GC and EC. As aresult of the public renewableelectricity supply, the primaryfossil energy becomes zero after2015 for scenarios SG1 and SG2.

Clearly, the fossil primaryenergy demand for scenario GC isassociated with the highest carbondioxide emissions while the SG2-scenario is associated with lowestcarbon dioxide emissions (seefigure 7.11). For each scenario,the carbon dioxide emissionchange rates are slightly higher

12 Backstop technology can regarded to be similar to end-of-pipe technology asit also involves dealing with emissions at the sink and not at the source implying that theCO2-emissions are reduced by ‘putting’ it in a sink other than the air. Beeldman et al.[1998] consider a very broad interpretation of backstop technology by including importsof biomass and imports of electricity derived from nuclear energy and renewables nextto CO2-removal and storage.

138

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60

70

GC EC

SG1 SG2

natural gas reserves

2050

Figure 7.12: Dutch natural gas reserves in EJ forthe scenarios GC, EC, SG1 and SG2.

than that of the fossil primary energy demand which is the result of small changes inthe fuel mix. According to the Kyoto (i.e. the Climate Change Conference in 1997)targets, The Netherlands is committed to reduce the CO2-emission with 6-10% in2010 compared to the emission level of 1990 [VROM, 1998]. Only in the case of theSG1 and SG2-scenarios, these emission targets are met as these scenario areassociated with high energy efficiency improvements and a renewable public electricsupply. Although the emission targets of 2010 are met, emissions only decreaseslowly after 2010 in case of scenario SG2 as this scenario involves a relatively lowconsumption growth rate.

Beeldman et al. [1998] investigated in what way the CO2-emission can be reducedby 32% in 2020 compared to 1990 for the GC-scenario when more energy savingoptions are used and by so-called ‘backstop’ technologies12 are introduced. Inprinciple, backstop technologies should be considered as end-of-pipe technologiesand can therefore be regarded as unfavourable. Depending on the way the technologyimprovements are stimulated, about 55-77% of reducing the CO2-emission reductionare realised by backstop technologies in order to meet this reduction target. Theseresults illustrate that high consumption levels are almost unavoidably associated withhigh CO2-emissions. Hence, substantially reducing the CO2-emissions necessitateschanging consumption patterns. This conclusion corresponds with the findings ofBiesiot [1998]. Note that the results in this study, changing consumption patternsmainly concerns lowering consumption growth rates. Changing the consumptionpackage requires a more detaileddescription of the productionsectors in NLDREAM and shouldbe introduced in future versionsof the model.

Figure 7.12 shows that theDutch natural gas reserves aredepleted firstly in the GC-scenario. The gas reservesdepletion is delayed with aboutten years (from 2012 to 2021) forscenario SG2. The natural gasreserves may appear to depletewithin a rather short time period.

139

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60

70

GC EC

SG1 SG2

total imports

2050

Figure 7.14: Total imports in (1985)EJ for thescenarios GC, EC, SG1 and SG2.

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

GC EC

SG1 SG2

import-export ratio

2050

Figure 7.13: Import-export ratio (defined as totalimports divided by total exports) for the scenariosGC, EC, SG1 and SG2.

However it is assumed here that the contribution of imports to the total domesticsupply remains constant after1990 whereas it may increase inpractice.

The depleting gas reserves areassociated with an increase in theimports of energy carriers. Figure7.13 shows the import-export ratio(defined as total imports dividedby total exports in which bothconcern fuels as well as goods andservices) increases as result. Theyear in which the gas reserves aredepleted can also derived fromfigure 7.10 as this year isindicated by a strong increase inthe import-export ratio. Therefore,the import-export ratio may be regarded as an indicator for (un)sustainability in thesense that it indicates the dependence on foreign resources. The Netherlands evenshift from a net energy exporting country (i.e. import-export ratio is lower than onefrom 1985 to 1990) to a heavily net energy importing country.

The high dependence onimports is also illustrated byfigure 7.14. For the GC-scenario,total imports are about 20 timeshigher in 2050 than in 1985implying an annual growth rate of4.6%. For the SG1 and EC-scenarios, total imports are aboutten times higher in 2050compared to1985.

7.6 Conclusions

In this chapter, the supply-driven ECCO-modelling approach is changed into thedemand-driven DREAM-approach. A dynamic energy accounting approach isdeveloped based on the notion that most production finally ends up in consumergoods and services. The consequences of changing consumption patterns on theoverall energy requirements of an economy can be determined by assigning mostenergy use resulting from economic activity to the consumption of households.Compared to the ECCO-approach, a number of key feedback loops are altered to

140

facilitate the shift from a supply-driven model to demand. The impact of thesechanges are studied in chapter 8 in which the results of DREAM are compared withthat of ECCO.

The DREAM-model of The Netherlands illustrates that a number ofenvironmental consequences of consumption (i.e. reserve depletion, CO2-emissionsand an increasing dependence on foreign resources) can be determined by using theDREAM-approach. Unlike NLECCO, the current version of NLDREAM can beused to study dynamically the energy costs of different consumption growth rates andexport growth rates combined with various energy saving potentials. Besidesdetermining the overall energy costs related to consumer activities, NLDREAMoffers the possibility to focus on some specific aspects related to consumer activitiesas all activities are expressed in terms of energy (e.g. fuel use due to privatetransportation, investments in the services sector, investments in electricity supply,etc). As the current version of NLDREAM involves a rather aggregated level, theenergy costs of different aspects related to consumer activity cannot be consideredat a detailed level so that the energy costs of changes in consumer packages cannotbe studied yet. However, these studies can be carried out when more productionsectors are considered.

The results of the GC and EC-scenarios developed with NLDREAM follow theresults of the ECN/CPB-studies in the sense that the deviations in the total primaryenergy demand of the corresponding scenarios are less than 10% in the year 2020[ECN,1997-1998; CPB, 1998].

Both scenarios involve a growing primary fossil energy demand as result ofconsumption and export growth. The results may seem rather straightforward as theenergy demand associated with growing consumption and export levels is notrestricted by, for instance, adopting the condition that imports and exports should bebalanced more in terms of energy or by setting the CO2-emission limits (e.g.consistent with the Kyoto target). In principle, these kind of items can be studied withthe aid of NLDREAM but they require some modelling adjustments.

Scenario SG1 showed that despite substantial energy savings potentials and apublic renewable electricity supply, relatively high consumption growth rates are stillassociated by an increasing total energy demand. The results of scenario SG1 andSG2 indicate that a decreasing energy demand can only be realised by lowering theconsumption growth rates. However, it should be noted that the consumption patternsare not changed in these scenarios. Changing consumption patterns towards moreenergy-extensive products will lower the total energy demand. It is not clear whetherthese changes will result in much more decreasing energy demands. Again, this topiccan be studied more properly with DREAM when more sectors are included in theDREAM-modelling approach.

13 The target for the annual energy use of 30-50 GJ per capita is equivalent tothe 1-1.5 kW as introduced by Dürr [1994] (see also section 7.5).

141

Chapter 8Scenario Results of OECD-DREAM

8.1 Introduction

This chapter discusses the results of the main parameters for a number scenariosdeveloped with the DREAM-model of OECD-Europe introduced in chapter 7. Oneof the most relevant differences in the DREAM and the ECCO-modelling approachis that the DREAM-modelling approach does not have a negative feedback loopassociated to industrial growth (cf. figure 4.3). Hence, industrial output growth is notdecelerated by consumption as it is the case in ECCO (see figure 4.1). In contrast,consumption is expected to boost industrial output as production is assumed to beinfluenced by consumption in the DREAM-modelling approach. Industrial output isallocated in such a way that the investments required for the desired growth arealways totally met.

The scenarios described in this chapter are developed in order to study four topics:

• Comparative assessments of the performance of DREAM-model for OECD-Europe compared to the ECCO-model of that region. Section 8.2 describes theresults of a number of scenarios which are developed to investigate this topic.

• The impacts of various consumption and export growth rates on the energy use,CO2-emissions and the import-export ratio. These topics are described in section8.3. In DREAM, imports balance the available output and the desired output.Therefore, the import-export ratio is a key factor as it indicates the dependenceon foreign energy sources (i.e. energy carriers purchased directly as well as theenergy required to produce imported goods and services). A growing dependenceon imports not only makes the region involved vulnerable to events which cannotbe influenced by the region itself, but it may also give rise to issues of equity thatis what are a region’s rights to exploit resources of other regions. In thisperspective, Mulder and Biesiot [1998] stress a long-term global target forannual energy use of 30-50 GJ13 per capita taking into account full equity and theglobal capacity of renewable energy production. Besides a growing dependenceon imports and the associated ‘equity challenge’, section 8.3 describes the impactof economic growth on the CO2 emissions

• Sensitivity analysis; section 8.4 outlines the results of sensitivity analysis

142

performed with the single DREAM-model for OECD-Europe

• Study of regional aspects in the DREAM-modelling approach. In section 8.5, theresults of scenarios of the multi-regional DREAM model are compared with thatof the single region DREAM-model.

Some general conclusions and discussion of the scenario results of the sections 8.2-8.4 are presented in section 8.6.

8.2 Comparison of DREAM and the ECCO-Scenarios in OECD-Europe

The aggregated DREAM-modelling approach differs from the ECCO-modellingapproach of OECD-Europe described in chapter 4 in the sense that a number of keyinfluences are different in the two approaches. These distinctions in the key feedbackloops may result in the outcomes of the two modelling approaches evolvingdifferently. Besides the alterations in the key feedback loops, the two modellingapproaches are rather similar in many other aspects. For example, similar data setsare used to compute the initial values as well as the most of the structure of themodel.

Below, the results of two scenarios developed with the ECCO-modelling approachare compared with those of the DREAM-modelling approach to assess in what waythe two models generate different outcomes. Average growth rates of consumptionper capita and of exports are derived from two ECCO scenarios and are applied tothe DREAM-modelling approach to obtain consistent starting points. As a result, theconsumption and export levels involve similar starting points as well as end pointsin a scenario corresponding to both models.

The two ECCO scenarios differ in the assumptions regarding annual changes(denoted by a.c.y) of a number of scenario variables for the period 1995 and 2050 (cf.section 5.3). These future annual changes are based on the growth rates andimprovements observed between 1985-1995 (denoted by o.r.). The first alternative(denoted by ‘sec1’) refers to the assumption that the annual changes of 1985-1995are maintained in the 1996-2050 period (i.e. for scenario ‘sec1’, a.c.y = (1+o.r.) fory between 1996-2050). In the second alternative (denoted by ‘sec2’), the observedgrowth rates and improvements are reduced to slow down the changes. For the 1996-2025 period, these annual changes are equal to the square root of the annual changesobserved for the 1985-1996 period (i.e. for scenario ‘sec2’, a.c.y = (1+o.r.)1/2 fory between 1996-2025). For the 2026-2050 period, the annual changes are equal tothe square root of the 1996-2025 period (i.e.for scenario ‘sec1’, a.c.y = (1+o.r.)1/4 fory between 2026-2050). The scenario variables reflect energy savings and consumption per capita growthrates and export growth rates. In the two scenarios, the structure of the technologymatrix (i.e. the structure of the intermediate deliveries) is assumed to be constant for

143

the 1985-2050 period as no data are available that observe changes between 1985-1995. In addition, fuel mixes and energy conversion efficiencies are assumed to beconstant after 1995. Table 8.1 lists the growth rates for the key scenario variables.

The results of the two scenarios, which are simulated with the aid of the twomodels, are presented below for a number of key parameters. Figures 8.1 and 8.2show that the shapes of the curves representing the consumption and the export levelsare different for the two models. In the scenarios developed with theDREAM-modelling approach, consumption and exports grow exponentiallyreflecting the constant average growth rates which are set exogenously. Theexponential growth patterns are not so clearly shown in the figures due to therelatively low growth rates in both scenarios.The growth rates of consumption andof exports decrease in both ECCO-scenarios. In ECCO, the growth of consumptionand export are computed endogenously and therefore consumption and export neednot grow exponentially. In addition, the scenario assumptions result in differentoutcomes. In particular in the case of the industrial export levels, the scenario resultsevolve differently. Exports increase gradually for both models in case of scenario‘sec2’. In case of scenario ‘sec1’, the export level starts to decrease after 2020 forthe ECCO-model. This decrease is due to a declining industrial output which in turnis the result of a declining room for investment. These differences between the curvesof figure 8.1 and figure 8.2 have a strong influence on the other parameters such asthe output (see figure 8.3).

Table 8.1: Annual growth rates and improvements (in %) of a number of scenario for theperiod 1985-2050

scenario ‘sec1’ scenario ‘sec2’

consumption 0.9 0.8

exports agriculture 1.8 1.1

industry 0.9 0.6

transport - -

services - -

thermal energy savings agriculture -2.7 -1.3

industry 1.7 0.8

transport 0.6 0.3

services 2.3 1.1

electricity savings agriculture -2.8 -1.3

industry 0.0 0.0

transport - -

services -1.7 -0.8

thermal energy savings dwellings 1.8 0.9

electricity savings dwellings 0.1 0.1

144

year1985 1995 2005 2015 2025 2035 2045

4

5

6

7

8

ECCO sec1 ECCO sec2

DREAM sec1 DREAM sec2

total exports

2050

Figure 8.2: Total exports in (1985)EJ for scenariosec1 and sec2 developed with DREAM andECCO.

year1985 1995 2005 2015 2025 2035 2045

25

30

35

40

45

ECCO sec1 ECCO sec2


total consumption level

2050

Figure 8.1: Total consumption level in (1985)EJfor scenario sec1 and sec2 developed withDREAM and ECCO.

As mentioned before, ECCOdetermines the developmentpotential of an economy. From thisenergetic perspective, economicgrowth depends on how muchoutput is allocated to industrialinvestments as well as consumption,exports, and other deliveries. Theroom for investments diminisheswhen too much output is allocatedto these latter terms implying thatthe capital stock grows at lowerrate. Subsequently, the output alsogoes down as the total output isproportional to the capital stock. Inthe DREAM-modelling approach, the balancing terms involve consumption andimports that is the demand forconsumption which can not bemet domestically is imported. InDREAM, investments are afunction of the demand for capitalstock which is based on desiredoutput levels set by consumptionand exports. Hence rather thandetermining the output level by theroom for investments, the demandfor output determines theinvestments rate.

Figure 8.3 shows that theoutput fluctuates somewhatinitially in the case of thescenarios developed with theDREAM-model. This decrease is the result of a somewhat lower desired outputcompared to that in the scenarios developed with ECCO. The lower desired outputis mainly due to lower consumption and export levels (see figure 8.1 and 8.2). Inaddition, lower desired output levels result in lower investments levels as less capitalis required to produce the output. Lower investments also reduce the demand forindustrial output since output also comprises investments.

Figure 8.4 shows the net investments which are equal to the rate of capitalformation minus the rate of capital depreciation. The oscillations in the curvesassociated to the two modelling approaches eventually disappear and are most likelythe result of small inconsistencies in the initial data set. Investments in the ECCO-

145

year1985 1995 2005 2015 2025 2035 2045

40

50

60

70

80

ECCO sec1 ECCO sec2


industrial output

2050

Figure 8.3: Industrial output in (1985)EJ forscenario sec1 and sec2 developed with DREAMand ECCO.

year1985 1995 2005 2015 2025 2035 2045

-0.4

-0.2

0

0.2

0.4

0.6

ECCO sec1 ECCO sec2


net investments in the industrial sector

2050

Figure 8.4: Net investments in the industrialsector in (1985)EJ for scenario sec1 and sec2developed with DREAM and ECCO.

scenarios show a moderately decreasing trend. In the DREAM-modelling approachon the other hand, net investments first decrease substantially before they stabilize.The temporary set back in netinvestment is the result of adecreasing desired output andthus a decreasing desired capitalstock (see also above). In case ofscenario ‘sec2’, net industrialinvestments of the DREAM-scenario exceed that of theECCO-scenario after 2012. In thecase of scenario ‘sec1’, netinvestments of the DREAM-scenario exceed that of theECCO-scenario after 2005.

This section showed that thescenarios developed with the aidof the DREAM-model result indifferent development paths andend points than correspondingscenarios developed with theECCO-model. Clearly, thestarting points are similar in thecorresponding scenarios andtherefore the differences betweenthe outcomes can be assigned todifferences in the structures of thetwo modelling approaches as thetwo models are developed fromdifferent perspectives. As aconsequence, industrial outputgenerally involves a diminishinggrowth rate in scenarios developed with ECCO-models. The diminishing growth rateis mainly the results of the prerequisite that more output is allocated to growingconsumption and investments in non-industrial sectors thus limiting the investmentsin the industrial sector. In DREAM, growing consumption and export levels resultin higher demand for output and thus in higher investment rates. Hence, the industrialoutput generally grows exponentially.

146

8.3 A Number of Growth Scenarios Developed with DREAM

8.3.1 Examples of Growth Rates

Three scenarios are presented in this section to study the impact of variousconsumption and export growth rates on the industrial output, the import-exportbalance, and environmental stress. In three scenarios, annual growth rates of bothconsumption per capita and exports are set at 1%, 3%, and 5%. The scenario aredenoted by ‘gr1’, ‘gr3’ and ‘gr5’, respectively. Note that these scenarios are notconsistent with the scenarios in section 8.2. Energy efficiency improvements areassumed equal in all 3 scenario and are similar to those of scenario ‘sec1’ of theprevious section which means that the growth rates in energy improvements observedin 1985-1995 are maintained in the period 1996-2050. The first scenario, whichassumes consumption annual and export growth rates of 1%, is, therefore, aboutequal to scenario ‘sec1’ of the previous section which also resulted in consumptionand export growth rates of about 1%.

Table 8.2 lists the annual growth rates and improvements for the key scenariovariables.

Table 8.2: Annual growth rates and improvements (in %) for a number of scenario for theperiod 1985-2050

scenario ‘gr1’ scenario ‘gr3’ scenario ‘gr5’

consumption per capita 1.0 3.0 5.0

exports agriculture 1.0 3.0 5.0

industry 1.0 3.0 5.0

transport - - -

services - - -

thermal energy savings agriculture -2.7 -2.7 -2.7


transport 0.6 0.6 0.6

services 2.3 2.3 2.3

electricity savings agriculture -2.8 -2.8 -2.8


transport - - -

services -1.7 -1.7 -1.7

thermal energy savings dwellings 1.8 1.8 1.8

electricity savings dwellings 0.1 0.1 0.1

147

year1985 1995 2005 2015 2025 2035 2045

0

500

1000

1500

2000

gr1 gr3 gr5

industrial output

2050

Figure 8.6: Industrial output in (1985)EJ for thescenarios gr1, gr3, and gr5.

year1985 1995 2005 2015 2025 2035 2045

0

200

400

600

800

1000

gr1 gr3 gr5

total consumption levels

2050

Figure 8.5: Total consumption level in (1985)EJfor the scenarios gr1, gr3, and gr5.

Figure 8.5 shows theconsumption level in case of thethree scenarios. The consumptiongrowth rates are somewhat higherthan 1%, 3%, and 5% since thegrowth rates of total consumptionalso includes population growthbesides growth of consumptionper capita. The total consumptiongrowth rates are 1.6%, 3.6%, and5.6% in case of scenario ‘gr1’,‘gr3’ and ‘gr5’, respectively.

Clearly as result of the annualgrowth rates, the consumptionlevels increase considerably in thecorresponding scenarios. Especially in the case of scenario ‘gr5’, it increasesenormously as it is more than thirty-five times larger in 2050 than in 1985. Thisconsumption boost is the result of the high growth rate of 5% (factor 22) and to amuch lesser extent to population growth (factor 1.5). But also in the case ofconsumption growth rate of 3%, the consumption level grows about tenfold in 65years. Consumption level in scenario ‘gr1’ seems almost constant compared toscenario ‘gr3’ and ‘gr5’ but it still doubles over that time period.

As a result of the high growthrates of consumption and export,industrial output also shows highgrowth rates in the correspondingscenarios (see figure 8.6). Thecorresponding industrial outputgrowth rates are about equal tothe consumption and exportgrowth rates (i.e. that is thegrowth rate of industrial output is1.4%, 3.5%, and 5.6%).

The higher dependence onimports is illustrated by theimport-export ratio (i.e.imports/exports). This ratio is presented in figure 8.7. For each scenario, the import-export ratio increases substantially between 1985 and 1995. These increases are theresult of additional import required to meet the domestic consumption level (see alsosection 7.4.1). The import-export ratio continues to increase after a small decreasearound 2000 for each scenario.

148

year1985 1995 2005 2015 2025 2035 2045

0

1

2

3

4

5

6

7

gr1 gr3 gr5

import -export ratio

2050

Figure 8.7: Import-export ratio (imports/exports)in (1985)EJ/(1985)EJ for the scenarios gr1, gr3,and gr5.

year1985 1995 2005 2015 2025 2035 2045

0

200

400

600

800

1000

1200

1400

gr1 gr3 gr5


2050

Figure 8.8: Total primary energy demand in EJ forthe scenarios gr1, gr3, and gr5.

In case of scenario ‘gr3’ and ‘gr5’, the import-export ratio about doubles in 65years. For both scenarios, imports are about six to seven times higher than the totalexports which indicates a relatively high dependence on ‘foreign’ resources that isresources which have their origin outside OECD-Europe. Moreover in the case ofscenario ‘gr5’, imports are almost equal to the total domestic output by the year2050, whereas imports are only one third of the total domestic output in 1985.

Figure 8.7 shows that theimport-export ratio in terms ofenergy is not in balance forOECD-Europe. However, importsand exports are much more inbalance when the ratio isexpressed in monetary terms ofproduct. A production structurewhich is not in balance from anenergetic perspective may be inbalance from a monetary point ofview as in practice the productionsectors generate sufficient valueadded to balance import andexports.

The import-export ratio alsoincreases due to a relativelyhigher dependence on the importsof fuels. Not only does theprimary energy demand increaseconsiderably but also thecontribution of indigenousproduction of fuels decreases dueto depleting reserves. The growthrates of total primary energydemand (figure 8.8) are less thanthat of the output due to energysavings. In case of the industrialsector, output can be producedabout 40% more efficiently in 2050 than in 1985.

For crude oil and natural gas, the contribution of indigenous production to thetotal domestic energy supply decreases gradually as result of depleting resources (seefigure 8.9). In the DREAM-modelling approach, the contribution of the indigenousenergy production to the energy supply is assumed to decrease gradually after theresource base has decreased to 60% of its initial level. This explains why thecontribution of indigenous production to the domestic energy supply starts to

149

year1985 1995 2005 2015 2025 2035 2045

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

gr1 (oil) gr3 (oil)

gr5 (oil) gr1 (nat. gas)

gr3 (nat. gas) gr5 (nat. gas)

contribution indigenous production to domestic supply

2050

Figure 8.9: Contribution of indigenous oil andnatural gas production to the total domestic oiland natural gas supply for scenarios gr1, gr3, andgr5.

year1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

gr1 gr3 gr5

carbon dioxide emissions

2050

Figure 8.10: Carbon dioxide emissions in Pg C forthe scenarios gr1, gr3, and gr5.

decrease in scenario ‘gr5’ first asthis scenario is associated with thehighest energy use and thuscumulative indigenous production.In scenario ‘gr3’ and ‘gr3’,indigenous production of crude oilbecomes almost negligibleimplying that almost all crude oilis imported after 2030. In addition,the contribution of indigenousproduction to the domestic energysupply of natural gas alsodecreases considerably especiallyin the case of scenario ‘gr5’. Forcoal, it stays almost constant in allthree scenarios due to ampleresources. Note that coal is notpresented in figure 8.9 but the contribution of indigenous production to the domesticenergy supply remains about 70% of the total domestic supply. So especially in thecase of scenario ‘gr5’ and ‘gr3’, OECD-Europe becomes considerably dependent onoil and natural gas resources from outside OECD-Europe.

In each scenario, fuel mixesare assumed to be constant after1995 and therefore an increasingprimary energy demand results inhigher CO2-emission levels (seefigure 8.10). Obviously, scenario‘gr5’ shows the highest emissionlevels. Besides scenario ‘gr5’,scenario ‘gr3’ also involves CO2-emission levels that exceed theKyoto protocol targets forOECD-Europe as most countriesof OECD-Europe committedthemselves to reduce theemissions of 1990 with about 8% for the period 2008-2012 [UN, 1997]. On thecontrary, carbon dioxide emission are respectively about 5 and 15 times higher inscenario ‘gr3’ and ‘gr5’ for 2010 compared to 1990. Even, scenario ‘gr2’ does notmeet the Kyoto protocol targets as the emissions still increase in this scenario.

150

8.3.2 Growth Rates Observed for the 1985-1995 Period

In order to compare the consumption and export growth rates presented insubsection 8.3.1 with observed trends, two additional scenarios are obtained byconsidering the observed trends in annual export growth rates and annualconsumption growth rates between 1985-1995. Within the first scenario (denoted by‘sdr1’), the observed annual growth rates are also applied to the period 1996-2050.In the second scenario (denoted by ‘sdr2’), the growth rates of exports andconsumption between 1996-2050 are assumed to be equal to the square root of theobserved growth rates between 1985-1995. The second scenario is introduced to slowdown the growth rates somewhat. Both scenarios are obtained in a similar way as theECCO-based scenarios ‘sec1’ and ‘sec2’ which are introduced in chapter 5.

Besides the growth rates of consumption and export, also the energy savings inthe end-use sectors (i.e. energy efficiency improvements realized by productionsectors and households) differ between the two scenarios. Relative changes in theseenergy savings are determined according to alternatives similarly to the growth ratesof consumption and exports. In the case of scenario ‘sdr1’, this holds that the annualchanges (1+o.r.) in energy savings observed for the 1985-1995 period are alsoapplied to the period 1996-2050. For scenario ‘sec2’, the square root of the averageannual changes observed for the 1985-1995 period are applied to the period 1996-2025' (i.e.between 2026-2050, the annual changes of scenario ‘sec2’ are equal to(1+o.r.)1/2 ). For the 2026-2050 period, the annual changes are equal to the squareroot of that of the 1996-2025 period (i.e.between 2026-2050, the annual changes ofscenario ‘sec2’ are equal to(1+o.r.)1/4 ). The other scenario variables (i.e. parametersset exogenously) are assumed to be constant after 1995).

The results of scenarios ‘sdr1’ and ‘sdr2’ are presented below for a number ofparameters. In addition, the results of the growth scenarios ‘gr1’ and ‘gr3’ are alsopresented in order to relate the latter two to the observed trends. Scenario ‘gr5’ is notincluded in this comparison as it results in too extreme outcomes. Table 8.3 lists thegrowth rates of consumption and exports among the different scenarios. In case ofscenario ‘sdr2’, the growth rates change gradually.

Table 8.3: Annual growth rates (in %) of consumption and exportconsumption growth per capita total exports growth

1995 2025 2050 1995 2025 2050

gr1 1.0 1.0 1.0 1.0 1.0 1.0

gr3 3.0 3.0 3.0 3.0 3.0 3.0

sdr1 1.9 1.9 1.9 5.1 5.1 5.1

sdr2 1.9 0.9 0.5 5.1 2.6 1.3

151

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

300

gr1 gr3 sdr1 sdr2


2050

Figure 8.11: Total consumption level in (1985)EJfor the scenarios gr1, gr3, sdr1, and sdr2.

year1985 1995 2005 2015 2025 2035 2045

0

100

200

300

400

500

600

700

800

gr1 gr3 sdr1 sdr2

total output production sectors

2050

Figure 8.13: Total output of the production sectorsin (1985)EJ for the scenarios gr1, gr3, sdr1, andsdr2.

year1985 1995 2005 2015 2025 2035 2045

0

20

40

60

80

100

120

140

160

gr1 gr3 sdr1 sdr2

total exports

2050

Figure 8.12: Total exports in (1985)EJ for thescenarios gr1, gr3, sdr1, and sdr2.

As result, consumption andexports develop according to thecurves presented in figures 8.11and 8.12, respectively. Clearly,scenario ‘gr3’ is associated withthe highest consumption levelswhereas scenario ‘sdr1’ shows thehighest export levels. Scenario‘sdr2’ show the lowestconsumption levels. By the year2050, the total final demand forconsumption and exports is thehighest in case of scenario ‘gr3’(i.e. final demand of about 300(1985)EJ) compared to scenario‘sdr1’ (i.e. final demand of about270 (1985)EJ). This final demandof the other two scenarios is muchlower (i.e. final demand of about65 and 85 (1985)EJ for scenario‘sdr2’ and ‘gr1’, respectively).

Figure 8.13 presents the totalindustrial output (in terms ofutility that is in 1985(EJ))corresponding to the final demandof consumption and exports incase of the four differentscenarios. In 2050, total domesticoutput of the production sectorsvaries from about 600-700(1985)EJ for scenario ‘sdr1’ and‘gr3’ to 150-200 (1985)EJ forscenario ‘gr1’ and ‘sdr2’. Foreach scenario, the total output isabout twice as high as the totalfinal demand. So, about half ofthe domestic output is allocateddirectly to the final demand forconsumption and export whereasthe other half is allocated tointermediate deliveries andinvestments. The latter two are

152

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

gr1 gr3 sdr1 sdr2

primary energy demand of production sectors

2050

Figure 8.14: Primary energy demand of theproduction sectors in EJ for the scenarios gr1, gr3,sdr1, and sdr2.

required to produce the final demand. The total domestic output being higher than the total final demand may at first

sight appear to be in contrast with the assumption that all the energy use in theproduction process is assigned to the final demand which is the starting point of theDREAM-methodology. However, the total domestic output figures presented here arecalculated by summing the output in terms of embodied energy of the differentsectors. As outlined in chapter 3, the embodied energy of the output of a sector alsoincludes the energy costs of the intermediate deliveries which obviously originate inproduction sectors. This holds that a part of a sectors’s output is also taken intoaccount as input of another sector and thus it is also included in the output of thatsector. Therefore , the output of the production sectors, by definition, involves doublecounting. The industrial output figure presented above should be regarded as anindicator of the volume of the output.

For both scenario ‘gr3’ and ‘sdr1’, the total output of the production sectors isabout 7-10 times higher in 2050 compared to 1985 which implies that the domesticproduction has an average annual growth rate of 3-3.5%.

Figure 8.14 shows the primaryenergy use of the productionsector corresponding to the totaloutput presented in figure 8.13. Incase of scenario ‘gr3’ and ‘sdr1’,the primary energy demand isabout 5-8 times higher in 2050compared to 1985 which impliesthat the domestic output isproduced 20-30% more efficientlyin 2050 than in 1985. Theseef f ic iency improvementsincorporate improvements at theend-use sector as well as changesin the ERE-values. In case ofscenario ‘gr1’ and ‘sdr2’, the energy demand about doubles between 1985 and 2050.This relatively low increase coincides with the increase in the domestic output of theproduction sectors and is mainly due to the low growth rates of both consumption andexports.

The development of the primary energy use by households is presented in figure8.15 for the different scenarios. In this case, the results of scenario ‘gr3’ and ‘sdr1’deviate considerably. The differences between the results of the two scenarios can beascribed to the differences in consumption growth rates. Figure 8.11 showed thatscenario ‘gr3’ involves a much higher consumption growth rate than in scenario‘sdr1’. Therefore, the material standard of living grows also much faster in scenario‘gr3’. In the DREAM-approach, the capital stock of dwelling is assumed to be

153

year1985 1995 2005 2015 2025 2035 2045

0

20

40

60

80

100

120

140

160

gr1 gr3 sdr1 sdr2

primary energy demand of households

2050

Figure 8.15: Primary energy demand ofhouseholds in EJ for the scenarios gr1, gr3, sdr1,and sdr2.

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

300

350

400

gr1 gr3 sdr1 sdr2


2050

Figure 8.16: Total primary energy demand in EJfor the scenarios gr1, gr3, sdr1, and sdr2.

proportional to the relative growthrate in the material standard ofliving. In turn, the electricitydemand and the thermal energydemand are determined by the levelof capital stock of dwellings. Inshort, a growing consumption levelis associated with an increasingprimary energy demand byhouseholds in case energyefficiency improvements are notincluded. Besides the higherconsumption level of ‘gr3’, theenergy efficiency improvements aresomewhat lower in ‘gr3’ than in‘sdr1’ which also results in arelatively higher energy primary energy demand.

The primary energy demand ofhouseholds and of the productionsectors together form the totalprimary energy demand. In case ofscenario ‘gr3’, the total primaryenergy demand is about 7 timeshigher in 2050 than in 1985implying that it involves anaverage annual growth rate of 3%.This growth rate is remarkablysimilar to the 3% growth rate ofthe total consumption and exportsbut this similarity is presumablycoincidental. Therefore, it wouldbe inappropriate to conclude from this result that the primary energy demand growsat a similar rate as consumption and export.

This section showed in what way consumption and export growth rates affect theoutput of the production sectors and the associated primary energy requirements. Inaddition, the development of parameters such as CO2-emissions, imports-export ratioand reserves depletion factors are presented to illustrate the impact of consumptionand export growth patterns on these topics.

154

year1985 1995 2005 2015 2025 2035 2045

0

20

40

60

80

100

120

140

160

consumption exports

reference scenario

2050

Figure 8.17: Development in consumption andexport according to the reference scenario ‘sdr1’.

8.4 Sensitivity Analysis

This section describes the results of sensitivity analysis. The sensitivity analysisis carried out to a lesser extent for the aggregated DREAM model than that of theECCO model which is presented in chapter 5. The latter was carried out for theregional ECCO-model involved studying the influence of a number of parametersassociated with a relatively high level of uncertainty. From the results presented inchapter 5 it was concluded that these parameters moderately or barely influence theoverall behaviour. For instance the industrial output was almost indifferent tochanges in these parameters.

Section 8.3 already showed that differences in the consumption per capita growthrates and the export growth rates strongly effect the outcomes of the model. Bothgrowth rates, which are set exogenously, are varied in order to study the sensitivityof the model to changes in these two variables. In addition, also the initial values inthe thermal energy intensities and the electricity intensities of the production sectorare altered in order to compare these sensitivity results with that of the consumptionper capita growth rate and the export growth rates. Section 5.2 showed that theregional ECCO model was rather sensitive for changes in the initial values of thethermal energy intensity and the electricity intensities of the production sectors(which are respectively denoted by TEI and EEI in chapter 4) .

Similar to the sensitivityanalyses in chapter 5, MonteCarlo simulations are used tocarry out the sensitivity analyses.In this way 1000 simulations wereperformed in which the constantinvolved is sampled over a rangeof values according to a triangulardistribution. The constantsaddressed above are also variedbetween a span of ±10% of itsvalue (i.e. the maximum value =1.1 * CST and the minimum value= 0.9 * CST). Clearly, the peaklies at the reference value (i.e.CST). The latter is derived from scenario ‘sdr1’. In other words, scenario ‘sdr1’ isused as reference scenario. Figure 8.17 shows the development of total exports andtotal consumption in case of this reference scenario ‘sdr1’. One important resultshown by this figure is that the total exports exceed the total consumption after theyear 2045 which means that the growth rates of exports are higher than that of totalconsumption. This result has strong implications for the sensitivity analyses.

155

sensitivity_consumption65% 95% 100%

out[ind,ac2]565.49

436.98

308.48

179.97

51.461985 2001 2018 2034 2050

Time (a)

Figure 8.18: Variation in industrial output due tosampling the consumption per capita growth rate.

sensitivity_exports65% 95% 100%

out[ind,ac2]623.24

480.30

337.35

194.41

51.461985 2001 2018 2034 2050

Time (a)

Figure 8.19: Variation in industrial output due tosampling the export growth rate.

sensitivity_energy intensities65% 95% 100%

out[ind,ac2]544.43

420.79

297.15

173.50

49.861985 2001 2018 2034 2050

Time (a)

Figure 8.20: Variation in industrial output due tosampling the thermal energy intensities and theelectricity intensities of the production sector.

Figures 8.18 and 8.19 showthe variation in the industrialoutput as result of sampling theconsumption per capita growthrate and the export growth rate,respectively. The variation inindustrial output is much higher incase of sampling the exportgrowth rates compared tosampling the consumption percapita growth rate. This outcomeis the result of the export growthrates being much higher than theper capita consumption growthrate. Corresponding relativechanges in the growth rates leadto higher absolute differences inthe case of export growth ratescompared to the per capitaconsumption growth rates.

Figure 8.20 shows thevariation in industrial outputresulting from sampling thethermal energy intensities and theelectricity intensities of theproduction sector. Compared tothe export growth rates and percapita consumption growthrates, the industrial output seemsto be indifferent to the changesin the thermal energy intensitiesand the electricity intensities ofthe production sector. Clearly,changing growth rates of exportsand consumption have moreimpact than changing the ratioof energy use per unit outputwhich explains the relativeindifference in case of varyingthe thermal energy intensitiesand the electricity intensities.

The differences between the

156

(1985)EJ

Figure 8.21: Comparison of variation in industrial output in 2050 (in(1985)EJ) as result of consecutively sampling the consumption percapita growth rate, the export growth rates, and the thermal energyintensities and the electricity intensities of the production sectors.

variation in industrial output as result of sampling the various parameters is clearlyillustrated by figure 8.21 which presents the variations in industrial output in the year2050 as result of consecutively sampling the consumption per capita growth rate, theexport growth rates, and the thermal energy intensities and the electricity intensitiesof the production sectors. In case of sampling the thermal energy intensities and theelectricity intensities of the production sectors, almost 100% of the outcomes of theindustrial output lie within a range of ±5% of the reference industrial output of 525(1985)EJ in 2050 (i.e. between 500-540 (1985)EJ). In the case of sampling theexport growth rate, only about 50% of the outcomes of the industrial output lie withinthis range in 2050. The outcomes of sampling the export growth rates, totally coverthe range of 440-620 (1985)EJ which is about ±20% of the reference industrialoutput. In case of sampling the consumption per capita growth rate, the outcomescover the range 480-580 (1985)EJ which is about ±10% of the reference industrialoutput.

This subsection showed that the outcomes of the model are sensitive to changesin the export growth rates and the consumption growth rates. These findings parallelwith those of subsection 8.3 where the results of difference consumption per capitaand export growth rates are described.

157

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

region 1 region 2 region 3



2050

Figure 8.22: Regional total consumption level in(1985)EJ for scenario sdr1.

8.5 Regionalisation

The first part of thesis dealt with the methodological issues concerning a multi-regional approach of the ECCO-methodology. It is asserted that the multi-regionalapproach resulted in a more accurate representation of the system. The results of theDREAM-modelling approach shown in sections 8.2-8.4 are derived from a singleregion model only. A multi-regional DREAM-model of OECD-Europe was comparedwith a single region model in order to study the consequences of regional differenceswithin OECD-Europe. This section describes the results of this comparison. Themulti-regional DREAM-model of OECD-Europe is developed in a similar way as themulti-regional ECCO-model of OECD-Europe. The basic concepts of the multi-regional model are, therefore, not outlined in this section. The first part of this sectioncompares the result of the multi-regional DREAM-model with that of the singleregion DREAM-model. The second part of this section deals with a number ofregional differences in the electricity supply.

8.5.1 Comparison between the Multi-Regional Approach with the Single RegionApproach

This section compares the results of two scenarios developed with the multi-regional DREAM-model with that of two corresponding scenarios developed with thesingle region approach. The two scenarios used in this comparison were alreadyintroduced in section 8.3 as the scenarios ‘sdr1’ and ‘sdr2’. These two scenarioswere selected to study regional differences as the main exogenous parameters arebased on the growth rates observed between 1985-1995. In this way, regionaldifferences observed between1985 and 1995 are extrapolated tothe 1996-2050 period. Bothscenario refer to the scenario asdefined in subsection 8.3.2. Inaddition, the parameters involvedare varied according the samealternative for each region growthrates. Besides the export and percapita consumption growth rates,the energy saving potentials arealso different in the two scenarios.The energy savings are to asimilar alternative as the growthrates of consumption per capitaand exports. Similar to the scenarios presented in chapter 5, the changes of the energyintensities are moderated for region 6 to avoid extreme outcomes.

158

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60



industrial exports

2050

Figure 8.23: Regional industrial exports in(1985)EJ for scenario sdr1.

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250



industrial output

2050

Figure 8.24: Regional industrial output in(1985)EJ for scenario sdr1.

Before comparing the outcomes of the multi-regional DREAM model with thoseof the single region model, the results of scenario ‘sdr1’ are presented at a regionallevel in order to illustrate thedifferences. Figures 8.22 and 8.23present the outcomes of the totalconsumption and exports. From thetwo figures, it can be seen that thegrowth rates differ per region. Inthe case of total consumption, thedifferences in the growth rates isdue to regional differences inconsumption per capita growthrates as well as in populationgrowth rates.

The consumption growth ratesare the highest in region 1 and 3. Inthese regions the average annualgrowth rates are equal to about 2.7%. Region 2 shows the lowest annual consumptiongrowth rates (1.3%). The total exports of region 6 increase considerably and end upthe highest within OECD-Europe albeit close to that of region 3. Region 6 togetherwith region 5 have the highest export growth rates (6.5-7%) whereas the growth ratesof the other regions are about 4.5-5%.

Figure 8.24 shows the results ofthe industrial output correspondingto the exports and consumptiongrowth rates of figures 8.22 and8.23. The industrial output appearsto grow the most in regions 5 and6. In both regions, the averageannual growth rate of industrialoutput is about 4%. For region 1and 3 the average annual growthrate of industrial output is about3.5%. The industrial output growthrate is the lowest in region 2 and 4where the industrial output onaverage grows with an annual rate of 3%. Although the industrial output of regions6 and 5 catches up with that of other regions, region 3 has the highest industrialoutput in OECD-Europe. Besides the growth rates of consumption and export, theimprovements in energy efficiencies differ also per region. The regional differences

159

year1985 1995 2005 2015 2025 2035 2045

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6



conversion factor of the industrial sector

2050

Figure 8.25: Regional conversion factor of theindustrial sector in EJ/(1985)EJ for scenario sdr1.

year1985 1995 2005 2015 2025 2035 2045

0

10

20

30

40

50

60

70



primary energy demand of the industrial sector

2050

Figure 8.26: Regional primary energy demand ofthe industrial sector in EJ for scenario sdr1.

in energy efficiency improvements are clearly illustrated by figure 8.25 which showsthe conversion factor of the industrial sector.

As mentioned before, theconversion factor is an indicatorof overall energy savings as it isdefined as the ratio between theindustrial output expressed interms of real energy use and theindustrial output expressed interms of its utility value. Fromfigure 8.25, it can be seen that theenergy requirements to produceone unit output increases in thecase of region 6. Thisphenomenon is already addressedin the beginning of thissubsection. Regions 1, 3, and 5 show rather similar energy saving options. The samecan be applied to regions 2 and 4. This grouping of regions is due to the electricitysupply in the regions involved. The electricity supply of regions 1, 3, and 5 is mainlybased on fossil fuels whereas the electricity supply of regions 2 and 4 heavily dependson nuclear and hydro electric energy, respectively.

Within the DREAM-modelling approach, determining the energy content ofoutput only takes into account fossil energy carriers. This methodology implies that(fossil) energy requirements of generating electricity are higher in the case of regions1, 3, and 5 compared to regions 2 and 4. Efficiency improvements in the use ofelectricity have, therefore, more impact in the overall energy efficiency improvementsin case of regions 1, 3, and 5.

Figure 8.26 shows the primaryenergy demand of the industrialsector. The most notable result isthe strong increase in the primaryenergy demand of region 6. Forregion 6, the primary energydemand has an average annualgrowth rate of about 5%. Theprimary energy demand of theindustrial sector also increasesconsiderably in regions 2 and 4where the primary energy demandannually grows on average about2.5%.

160

year1985 1995 2005 2015 2025 2035 2045

0.5

0.6

0.7

0.8

0.9

1

regsdr1 regsdr2

agrsdr1 agrsdr2

converion factor of the production sectors

2050

Figure 8.28: Comparison of results conversionfactor of the industrial sector (in EJ/(1985)EJ)developed with the single region model and themulti-regional model for scenario sdr1 and sdr2.

year1985 1995 2005 2015 2025 2035 2045

0

100

200

300

400

500

600

700

800

regsdr1 regsdr2

agrsdr1 agrsdr2

total output of the production sectors

2050

Figure 8.27: Comparison of results of industrialoutput (in (1985)EJ) developed with the singleregion model and the multi-regional model forscenario sdr1 and sdr2.

The first part of this subsection clearly shows that there are regional differenceswithin OECD-Europe if the observed trends for 1985-1995 are extrapolated to the1996-2050 period. The last part of this subsection shows the effects of these regionaldifferences on the aggregated results. Besides the results of scenario ‘sdr1’, theresults are also shown for scenario ‘sdr2’ where the regional differences arediminished. The aggregated results of the multi-regional model are compared withthat of the single region model.

Figure 8.27 shows the resultsfor the total output of theproduction sectors. Although theoutcomes may appear rathersimilar for the correspondingscenarios, the total output of theproduction sectors differconsiderably by the year 2050.For instance in 2050, the totaloutput of the production sector isabout 20% higher for the multi-regional model compared to thesingle region model in case ofscenario ‘sdr1’. In the case ofscenario ‘sdr2’, total output ofthe production sector is about10%higher for the multi-regional model compared to the single region model.

Figure 8.28 shows thedifferences between the conversionfactor of the two modellingapproaches. In comparing theresults of the correspondingscenarios, the conversion factorsare mostly lower for the multi-regional approach than for thesingle region approach. Howeverby the year 2050, the differencebetween the conversion factors ofthe corresponding scenariosdiminishes. In the case of scenario‘sdr1’, the conversion factorincreases in case of the multi-regional approach and finallyexceeds that of the single region approach. The increasing conversion factor is mainlydetermined by region 6 as the increasing energy intensities of this region more and

161

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

regsdr1 regsdr2

agrsdr1 agrsdr2

primary energy demand of the production sectors

2050

Figure 8.29: Comparison of results of primaryenergy demand of the production sectors (in EJ)developed with the single region model and themulti-regional model for scenario sdr1 and sdr2.

more dominate the overallconversion factor of the productionsectors of OECD-Europe.

As a result of the differences inthe output growth rates ofproduction sectors and thedifferences in the energy efficiencyimprovements, the primary energydemand of the production sectoralso shows substantial differencesbetween the two approaches andfor scenario ‘sdr1’ (see figure8.29). For this scenario, theprimary energy demand is about30% higher for the multi-regionalmodel compared to the singleregion model. In case of scenario‘sdr2’, the primary energy demand is about 15% higher for the multi-regional modelcompared to the single region model.

This subsection showed that regional differences in consumption growth rates andexport growth rates and the energy efficiency improvements considerably affect theoutput of the production sectors and the associated primary energy demand. Not onlydoes this effect hold when the observed differences are extrapolated for the 1996-2050 period but it also appears to be valid, albeit to a lesser extent, when thesedifferences are moderated somewhat. As a consequence, the multi-regional approachresults in rather different outcomes than the single region approach.

8.5.2 Regional Differences in Electricity Supply

Subsection 8.5.1 showed that regional differences in the observed consumptiongrowth rates and the energy efficiency improvements have impact on the developmentof the aggregate primary energy demand or total output of the production sectorswhen these growth rates or improvements are extrapolated to the future. Basically,this subsection showed the results of the impact of regional differences under moreor less business-as-usual circumstances. This subsection also shows some results ofthe effects of rather drastic changes within one region. A nuclear-energy-freeelectricity supply is chosen as a case study. In this case study, it is assumed that thenuclear power plants are replaced by fossil electricity plants in region after 2010. Thefossil primary energy requirements of generating electricity forms an important factorin the methodology. A transition towards a nuclear-energy-free electricity supply asproposed above will increase the fossil energy requirements of generating electricity

162

year1985 1995 2005 2015 2025 2035 2045

0

5

10

15

20

25

30

35

sdr1 (region 3) nfreg3 (region 3)

sdr1 (region 4) nfreg4 (region 4)

fossil primary energy requirement for electricity generation

2050

Figure 8.30: Fossil primary energy requirement ofelectricity generation for regions 3 and 4 in case ofscenario sdr1 and the corresponding nuclear-energy-free scenario (i.e. nfrreg3 and nfreg4).

and therefore influence the total primary energy demand. Similar to Sweden, atransition towards a nuclear-energy-free has become official governmental policy inGermany. The electricity supply of region 3, which includes Germany, depends forabout 20% on nuclear energy. In the case of region 3, the initial contribution ofnuclear energy in the electricity supply might be too small to have an effect in theaggregate fossil primary energy demand of OECD-Europe when nuclear energy issubstituted by fossil energy carriers. The electricity supply of region 4, on the otherhand, depends heavily on nuclear energy (about 70%). A transition towards anuclear-energy-free electricity supply in region 4 may, therefore, affect substantiallythe aggregate fossil primary energy demand of OECD-Europe.

Four scenarios have beendeveloped to study the issuesaddressed above. The first isscenario ‘sdr1’, already defined insubsection 8.5.1 and serves as thereference scenario. In the secondand third scenarios (denoted by‘nfreg3’ and ‘nfreg4’), theelectricity supply of regions 3and 4 are assumed to be nuclear-energy-free af ter 2010,respectively. In the fourthscenario (denoted by ‘nfeur'), thetotal electricity supply of OECD-Europe is assumed to be nuclear-energy-free after 2010. Except forthe fuel mix of the electricity supply, the latter three scenario are consistent with thereference scenario ‘sdr1’.

Figure 8.30 shows the total fossil energy requirements for generating electricityin regions 3 and 4. In case of region 4, the fossil primary energy requirements abouttriple when the electricity demand is met by a nuclear-energy-free electricity supply.Due to the transition and an increasing electricity demand, the total primary fossilenergy requirements are about 15 times higher in 2050 compared to 1985 whichcorresponds to an average annual growth rate of 4%. The fossil primary energyrequirements also increase considerably when this regions no longer generateselectricity from nuclear energy.

Figure 8.31 shows the aggregate fossil primary energy requirements forgenerating electricity in OECD-Europe. Clearly, scenario ‘nfeur’ deviates the mostfrom the reference scenario ‘sdr1’ as a nuclear-energy-free European electricitysupply has the largest effect. A nuclear-energy-free European electricity supplywould mean that about 40% more fossil primary energy is required to meet the

163

year1985 1995 2005 2015 2025 2035 2045

0

20

40

60

80

100

120

sdr1 nfreg3

nfreg4 nfeur

fossil primary energy requirement for electricity generation

2050

Figure 8.31: Fossil primary energy requirement ofelectricity generation for OECD-Europe for thescenarios sdr1, nfrreg3, nfreg4 and nfeur.

year1985 1995 2005 2015 2025 2035 2045

0

50

100

150

200

250

300

sdr1 nfreg3

nfreg4 nfeur


2050

Figure 8.32: Total fossil primary energy demandfor OECD-Europe for the scenarios sdr1, nfrreg3,nfreg4 and nfeur.

electricity demand in 2050 underthe assumptions of scenario ‘sdr1’.The aggregate fossil primaryenergy requirements even increasessubstantially when only region 3shifts towards a nuclear-energy-free electricity supply. In this case,about 6% more fossil primaryenergy is required to meet theelectricity demand in 2050.Making the electricity supply ofregion 4 nuclear-energy-freerequires about 25% more primaryfossil energy to meet the electricitydemand in 2050.

The above shows that anuclear-energy-free electricitysupply considerably affects theaggregate primary fossil energydemand even if this transition onlytakes place at a regional level.These differences are less clearwhen the total primary energydemand is taken into accountinstead of only focusing on theprimary energy requirements ofthe electricity generation. Figure8.32 shows the aggregate primaryenergy demand for the fourscenarios. In case of scenario‘nfreg3’, the total primary energy demand is only about 1% higher compared to thereference scenario in 2050. So, the nuclear-energy-free electricity supply in region3 hardly influences the aggregate primary energy requirement of OECD-Europe.However, the opposite holds for region 4 as the aggregate primary energy demand isabout 8% higher in scenario ‘nfreg4’ than in scenario ‘sdr1’ in 2050. Under thescenario assumptions of scenario ‘sdr1’, a nuclear-energy-free OECD-Europe wouldresult in an additional primary energy demand of 15% in 2050.

The scenarios presented in this subsection showed that changes in the electricitysupply which only takes place in one region can affect the aggregate results such asthe aggregate primary energy demand. These results support the use of multi-regionalapproaches as they indicates that changes in a subregion can influence the aggregate

164

outcomes.

8.6 Concluding Remarks

In chapter 7, DREAM-methodology was introduced as a tool to study the long-term consequences of various consumption and export growth rates on thedependence on net imports and on environmental stress. The DREAM-methodologycomputes dynamically the overall energy costs to meet the demand for consumptionpatterns and exports. The DREAM methodology is based upon the ECCO-methodology but some of the key influences in ECCO are changed. As result of thesechanges, ECCO-models and DREAM-models behave differently. Scenariosdeveloped with the aid of the DREAM-modelling approach involve exponentialgrowth whereas in ECCO the growth rates of the economy tend to decrease over time.Depending on the assumptions made, scenarios developed with theDREAM-modelling approach show that high growth rates of consumption coincidewith high CO2-emissions not compatible with the Kyoto protocol targets. In addition,the primary energy demand may increase considerably and it easily exceeds the long-term target of an annual energy use of 30-50 GJ per capita. This long-term target isstressed by Mulder and Biesiot [1998] and takes into account full global equity interms of energy service and the global capacity of renewable energy production.Moreover, a high energy demand is associated with a relatively high dependence onnet imports. Sensitivity analyses showed that the outcomes of the DREAM-modellingare sensitive to changes in the export growth rates and the consumption per capitagrowth rates.

The multi-regional DREAM-model again showed that regional differences canresult in other aggregated results than a single region approach. This result does notonly hold for extrapolating regional differences in the consumption and export growthrates but it can also hold for certain changes in one region (e.g. in developing anuclear-energy-free electricity supply). These outcomes support the idea that multi-regional approaches result in a better determination of the energy use associated withchanging consumption patterns in OECD-Europe than a single region approach.

165

Chapter 9Conclusions and Reflections

9.1 Introduction

In essence, this thesis addressed two methodological aspects of dynamic energyaccounting modelling approaches. The first involved regionalisation of the ECCO-modelling approach as developed by Slesser et al. [1997], Ryan [1995], andNoorman[1995]. The second involves the development and application of a demand-driven dynamic energy accounting approach. In this approach, energy costsassociated with changing consumption patterns are assumed to be driving forcesunderlying changes in the production system and thus in the system’s energyrequirement. This demand-driven model is mainly based on the principles of theECCO-approach in which the latter is used as a tool to assess economic developmentpotentials.

Section 9.2 addresses the perspectives and methodological aspects which formthe basis of this study. Section 9.3 concerns the regionalisation procedures appliedto the ECCO-modelling approach whereas section 9.4 addresses the DREAM-modelling approach. Some general conclusions and reflections on both approachesare presented in section 9.5. Finally, suggestions for future research are given insection 9.6.

This thesis started from describing how the interrelationship between the economicand the environmental systems are considered in the energy accounting approach. Anumber of methodological aspects were subsequently addressed throughout thisthesis. To begin with, a regional energy input-output picture was introduced to studythe energy contents of (international) trade in a static way. However, the main partof this thesis dealt with dynamic energy modelling approaches and in particular itaddressed methodological topics considering regionalisation and supply versusdemand-driven approaches.

9.2 Theoretical and Methodological Backgrounds

This section describes the background of this thesis. Not only are variousperspectives on the economy-environment interface discussed but also regional input-output analysis is considered as the background, since the latter forms amethodological starting point of both modelling approaches.

9.2.1 Perspectives of this Thesis

Different schools of economics study the relationship between the environmental

166

system and the socio-economic system from different perspectives and focus ondifferent aspects of the economy-environment interface. In essence, the school ofneoclassical economics (including resource economics and environmental economics)considers this interrelationship from a socio-economic point of view by monetarisingthe inputs from and the outputs to the environment. The school of ecologicaleconomics studies the relationship between the environmental system and the socio-economic system from a broader perspective by using methodologies and approachesfrom various scientific disciplines. One of these approaches involves the energyaccounting approach in which all economic activities are expressed in energy terms.The approach is based on the notion that economic activities are subject tofundamental physical laws as all economic processes occur on earth (i.e.thermodynamics) and it, therefore, applies physical concepts to (macro) economics.

Within the framework of the energy accounting approach, the ECCO-modellingapproach was developed to study the potentials of economic development taking intoaccount the physical laws rather than taking the standard economic decisionframework as a starting point. The ECCO-modelling approach can be characterisedas a dynamic energy accounting approach in which resources are quantified throughthe primary energy required to release them and to produce any good or service in theeconomy. Based on the principles of system dynamics, the ECCO-modellingapproach describes the trajectory of variables characterizing the structure andevolution of the economy in relation to physical conditions. In essence, the approachis supply-driven implying that the level of economic activity is determined by theproduction sectors. Feedback loops between investments and industrial output,therefore, form the main elements of the modelling approach. This supply-drivenapproach resulted from the objective to study the potentials of an economy to growunder certain physical conditions.

Nijkamp [1987] addresses various limitations of these system dynamics models.He stresses, among others, that the integration of policy models and institutionalconfigurations in such models is usually poor, the behavioural characters of many ofthese models are fairly limited and the majority generate conditional pictures of theevolution of a sector, but fail to provide reliable predictions on solidstatistical/econometric techniques. Most neoclassical economists will probably agreewith Nijkamp’s arguments and add the limitation that market mechanisms are poorlydealt with in ECCO. While it is recognised that the physical aspects of the economicsystem described in the ECCO-modelling approach cannot fully account for standardeconomic decision rules, the approach yields a number of insights into the functioningof economic systems in relationship with the environmental system. Unlikeneoclassical economic models, the ECCO-approach is not designed to focus onunderstanding and predicting human behaviour or short-term optimal allocations butit is developed to analyse the long-term physical limits on economic activity. TheECCO-approach forms a framework within which economic policies must be(although this is mostly not -fully- recognised by neoclassical economics) formulated.

14 The outcomes of this study are also outlined in [Battjes et al., 1998].

167

Therefore, the ECCO-approach should be regarded as an additional approach nextto economic/econometric models instead of a substitute. Both type of models shouldbe part of integrated analysis to study the relationships between the economic systemand the environmental system in order to direct economic activity towards sustainableroutes of development.

9.2.2 Regional Input-Output Energy Analyses

Regional input-output analyses are used to compute the initial values of theECCO-models of OECD-Europe, and notably for determining the initial energy valueof intra and inter regional trade. The multi-regional approach which was used tocompute the energy intensity of inter and intra regional trade resulted in someremarkable outcomes especially in the case of assessing the energy intensity ofimports14.

There appear to be large national differences in embodied energy intensities withinthe OECD-Europe region. For instance, the direct and the embodied energy intensitiesof The Netherlands and Ireland differ substantially from the average values ofOECD-Europe. This may be the result of rather specific economic structures and ofthe contribution of imports to the total resources.

The usual assumption in input-output energy analysis that the embodied energyintensities (EEI) of imports can be set equal to the EEI of corresponding domesticproducts introduces errors in the calculations of the EEI for a country. This certainlyapplies to countries with a relatively high contribution of imports and a ratherspecific economic structure. Hence, two other approaches are introduced to assessthe EEI. It was demonstrated that, in these cases, the EEI of imports can be based onthe average EEI of OECD-Europe, although the EEI of OECD-Europe are computedby using a similar method. Moreover, it has been shown that estimating the EEI ofimports by specifying the imports in terms of their origin (i.e. by dividing OECD-Europe in 6 subregions) appears to be a next step in avoiding errors in assessing ofthe energy intensities. However, it should be noted that all calculations wereperformed at an aggregated level which might exaggerate somewhat the resultscompared to involving more sectors. Intuitively, it is felt that avoiding these errorsresults in more accurate assessments of the EEI.

The energy flows among the regions are very region specific and with the totaloutput in terms of money in mind these flows illustrate the energy intensiveness of theproduction of the various regions. A number of regions appears to be rather energyintensive which can largely be ascribed to the fuel mix in the electricity productionsector. The energy flows associated with a region also indicated its (relative)

168

dependence on foreign resources. This relative dependency on foreign resources alsovaries per region.

In the analyses described in chapter 3, the energy flows and energy intensities aredetermined by taking into account regional differences within OECD-Europe. Theenergy flows originating from non-European OECD-regions are still poorly dealtwith as they are assumed to stem from one large region (called the ‘rest of theworld’). However for most regions, these import only contribute moderately (about5-10%) to the total domestic supply of goods and services, perhaps with theexception of region 4. Dividing this large region into a number of homogenousregions could be the next step in making more accurate assessment of the energycontent of international trade flows and of the energy intensities of import stemmingfrom these regions. However, this next step makes the analysis more complex as itrequires much more consistent data. 9.3 Regionalisation

Imports and exports are becoming increasingly important for national economiesin the world and in particular in the case of Europe as the result of the introductionof the euro and the liberalisation of the energy markets in the European Union. Aproper treatment of imports almost unavoidably necessitates the introduction ofmulti-regional models dealing with regional differences. The necessity of developinga regional model also applies to the ECCO-modelling approach as the industrialstructures and the energy supply and demand differ from country to country.Therefore, the first part of this thesis addressed the following research question:

Do the results from regional modelling approach differ from those of the so-called‘one region’ modelling approaches?

The results of two ECCO-modelling approaches were compared in order to studythis question. The first modelling approach involved a multi-regional ECCO-modelling approach in which OECD-Europe was divided into 6 subregions. Thesecond consisted of an ECCO-modelling approach in which the OECD-Europe wasregarded as one large region.

9.3.1 Developing a Multi-Regional ECCO-Model of OECD-Europe

The regional ECCO-modelling approach as presented in chapter 4 was mainlybased on the common concepts as presented by Slesser et al. [1994], Noorman[1995], and Ryan [1995]. The changes with the most drastic consequences involvedthe introduction of a regional approach. Most ECCO-models consider one countryor region, whereas the ECCO-model presented in the first part of this thesis coveredthe whole region of OECD-Europe and distinguished 6 subregions. It was postulated

169

that in this way the outcomes of a region are less vulnerable for foreign differencesor changes as the trade within OECD-Europe is dealt with in a more accuratemanner.

Besides these regional aspects, the ECCO-models presented also have a numberof other new features. One major change is, for example, the more endogenouscalculations of the ERE-values (energy required for energy). In former modelling theERE-values were mostly set exogenously and were treated as scenario variables. Inparticular the energy required for mining is adjusted in the model of OECD-Europe.The energy requirements of mining are of importance as one might expect that miningrequires more energy as reserves become depleted. On the other hand, energyefficiency of mining may increase when new and more efficient techniques becomeavailable. In the ECCO-model of OECD-Europe, the evolving ERE-values wereassessed endogenously by introducing an elasticity between the relative energyrequirements and the depletion ratio. In addition, the elasticity of the efficiencyimprovements was determined by using a learning-by-doing factor, that is the energyefficiency of mining is considered to be a function of the cumulative production.

Lack of data is a disadvantage of considering a large region as OECD-Europe.As a consequence, the development of the multi-regional model of OECD-Europerequired a number of assumptions regarding data used for determining the initialvalues of the model. Moreover, it also restricts the level of detail of the model as inmost cases detailed and consistent statistics were lacking. The availability of dataalso influenced the region classifications as consistent input-output tables have onlybeen presented in the literature for several European countries. Fortunately, thesecountries involve the more dominant economies of OECD-Europe and therefore it isbelieved that the most important economic structures are properly covered in themulti-regional approach. However, it is hard to point out in what way theassumptions introduce errors in the outcomes of the model although sensitivityanalysis showed that these possible errors hardly affect the overall outcomes of themodel in a number of assumptions. 9.3.2 Outcomes of the Mulit-Regional Modelling Approach

Below the main conclusions of the outcomes of the model are described. Theoutcomes presented in the first part of this thesis can be separated in two types ofanalysis. The first one comprises the outcomes of sensitivity analysis and the secondconsists of studying the impact of regional differences.

Sensitivity Analysis of the Model

Sensitivity analysis showed that outcomes of the model, or in particular thegrowth potentials of industry, are rather sensitive for a number of major inputs.Varying the variables mostly resulted in rather moderate deviations which were

170

parallelled with the variations in the parameters changed. This implies that thesechanges mainly influence the parameters involved and have a moderate impact on thewhole system. However, in a number of cases varying these inputs may make thedifference between growth or decay. The room for net investments appeared to be acrucial parameter in the sensitivity of the growth potential of region as regions inwhich the net investments were around zero show the highest deviations in theoutcomes of industrial output. In addition, lower output levels showed more variationthan the higher ones. This effect may mean that a decreasing room for investmenthas more impact on the growth potential than an increasing one, or that declininggrowth rates have more impact than inclining ones.

In addition the multi-regional model appeared to be rather insensitive to varyingthe constants which are associated with assumptions due to lack of data. Theseassumptions involve parameters such as the energy to expenditure elasticity ofconsumers, the energy requirement to reserve depletion elasticities, and the energyintensities of imports originating outside OECD-Europe.

The Impact of Regional Differences

A number of key parameters were varied according to three alternatives in orderto study regional differences. These parameters were the energy intensties and thedesired relative growth rate of the output of non-industrial sectors. The threealternatives encompassed a wide span of development potentials to illustrate thespread of the possible deviations. However for each scenario, the scenario variableswere varied according to the same alternative for each region to avoid studying theobvious since it might be expected that regions are developing differently when theyinvolve totally different starting points.

Although the scenario variables were varied according to the same scenarioalternative, the regional economies evolved according to different patterns among theregions, especially for scenario ‘1’ in which the observed growth rates wereextrapolated to 2050. The results of the primary energy demand of householdsdeviated more among the scenarios than that of the corresponding primary energydemand of the production sectors. This difference is due to the model being supply-driven. Small changes in the investments rates strongly influence the consumptionand thus the material standard of living. In turn, the material standard of livingdetermines the primary energy demand by households.

The existence of regional differences was clearly shown by the total primaryenergy demand of regions 5 and 6. For both regions the results appeared rathersimilar for the various scenarios. However for both regions, these similarities werethe result of counteracting developments in the primary energy demand of householdsand of production sectors. It is postulated here that these kinds of effects stress theadvance and necessity of multi-regional models. Clearly, the regions that were mostsensitive to small changes in a number of parameters also showed the highest

171

deviations among the different scenarios.

As a result of the differences in regional development paths, considering the multi-regional differences generally resulted in significant deviations compared to that ofthe single region approach. In a number of cases, the outcomes of the aggregatedregion OECD-Europe were substantially influenced by extreme developments in oneregion. The single region version of the model did not reveal these results as it failedto represents the underlying dynamics. For example, the average conversion factorof the industrial sector is increasingly influenced by the developments in the ratherenergy inefficient regions 5 and 6 as the relative contributions of these regionsincreased over time (especially when the observed trend for 1985-1995 wereextrapolated to 2050). These dynamics were not represented by the single regionmodel.

9.3.3 General Remarks on Multi-Regional ECCO-Approach

In the first part of the thesis, it was demonstrated that the outcomes of a multi-regional ECCO-model differ substantially from that of a single region ECCO-model.Of course, it is difficult to conclude which method results in more accurate outcomeswhen two methods are compared that both use rather unrealistic scenario variablesas starting point. However, the preponderance of evidence indicates that taking intoaccount the differences among subregions of an aggregate region results in a betterdescription of the dynamics than by only regarding the average values of theaggregate region. In this perspective, the deviating results for a number of parametersamong the various scenarios developed by the multi-regional model support this viewpoint. The scenarios presented were developed by extrapolating the trends observedfor the 1985-1995 period. Some of the regional differences resulting from thisextrapolation may be the consequence of the process that a number of regions (e.g.regions 5 and 6) are catching up to other regions (e.g. region 3) implying that in thecourse of time the regional differences will disappear. Clearly, a homogenouseconomy throughout OECD-Europe makes a multi-regional approach less necessaryto study regional differences although it still facilitates the study of changing energystocks and flows within OECD-Europe. Another consideration of developing a multi-regional model is that the economy will most probably not become completelyhomogenous as for instance there are still substantial differences in the electricitysupply. In addition, many policy makers will be mostly interested in the nationalconsequences of their decisions and less in that of OECD-Europe a whole.

The scenarios presented here were developed from a methodological point of view,that is the scenarios do not indicate expectations about the potential growth rates ofthe economies and the related energy efficiency improvements of the various regionsbut the scenarios are developed to show the impact of possible regional differences.

172

Summarising the research question studied in the first part of this thesis, theoutcomes of the multi-regional ECCO-model differ substantially from thecorresponding results of the single-region model. Going even further, it is assertedhere that the multi-regional ECCO-modelling approach results in more accurateoutcomes than the single regions ECCO-model.

9.4 The Dynamic Resource and Economy Accounting Model

In principle, ECCO-models are all supply-driven implying that one assumes thatthe consumption level is determined by producers. It is increasingly recognised thatconsumers or households play a key role in driving the economy. Hence, it is arguedto assess the energy costs associated with changing consumption activities underlyingthe changes in production systems and the system’s energy costs. In this way, theenergy costs and the related environmental stress can be determined for differentscenario assumptions about future consumer activity. The second part of this thesisaddressed the following research questions:

What are the consequences for the results of scenarios developed with theDREAM-modelling approach compared to that of the ECCO-modellingapproach? Or in other words, do the outcomes of a supply-driven model differsubstantially from those of a demand-driven modelling approach?

9.4.1 Developing the DREAM-Modelling Approach

In the second part of this thesis, the supply-driven ECCO-modelling approach waschanged into the demand-driven DREAM-approach. That is, a dynamic energyaccounting approach was developed based on the notion that most production finallyends up in consumer goods and services. As a result of shifting from a supply-drivenmodelling approach to a demand-driven approach, a number of key feedback loopsand influences were altered in DREAM. DREAM does not include the negativefeedback loop in which a higher demand for consumption limits the room forinvestments in the industrial sector. The latter forms the basis of ECCO together withthe positive feedback loop that a higher output implies a larger room for investmentin the industrial sector.

In DREAM, the investment rates are no longer determined by the allocation ofindustrial output but depend on the demand for industrial output. The demand forindustrial output ultimately depends on the consumption and export growth rates.Unlike ECCO, the demand for industrial output for the items consumption, exports,intermediate deliveries and investments is always met in the DREAM-modellingapproach. Output that cannot be produced domestically due to insufficient capital ismet by imports. So, imports form the balancing term in the DREAM-modellingapproach. Compared to the ECCO-approach, the balancing term shifted from the

173

room for investments to the imports.

9.4.2 A DREAM-Model for The Netherlands

The DREAM-model of The Netherlands illustrates that a number ofenvironmental consequences of consumption (i.e. reserve depletion, CO2-emissionsand an increasing dependence on foreign energy resources) can be related to changesin consumption and export levels. Besides determining the overall energy costs ofconsumer activity, specific aspects related to consumer activities and productionactivities can be studied separately. As the current version of NLDREAM considersthe production sectors at an aggregated level, the energy costs of different aspectsrelated to consumer activity cannot be considered at a detailed level. This implies thatdetermining the energy costs of changes in consumer packages cannot be studied yet.However, these studies can be carried out when more production sectors areconsidered in more detail.

The results of the scenarios carried out with NLDREAM showed that relativelyhigh consumption growth rates (cf. the results of scenario SG1) are associated withan increasing total energy demand despite the assumptions in a number of scenariosthat energy saving potentials increase substantially and the public electricity supplyis totally based on renewables. These results indicate that a decreasing energydemand can only be realised by redirecting consumer activities which can be realisedby decreasing consumption growth rates (cf. the results of scenario SG2) or bychanging the package of consumer goods and services. In the current version ofNLDREAM, the energy costs related to changes in consumption packages cannot beassessed at a detailed level as the production sector involves a too aggregated level.These kind of studies should be made possible in future versions of NLDREAM asthe effect of changing consumption patterns on the overall energy use is a major topicin meeting the targets of sustainable development (cf. [Biesiot, 1998; Biesiot andNoorman, 1999; Noorman and Schoot Uiterkamp, 1998, Kramer et al., 1998]).These studies should address the question whether changes in the packages ofconsumer goods and services can result in sustainable levels. Assessments ofconsumer activities in terms of suitability for achieving sustainable developmenttargets and paths requires (normative) criteria concerning the desired relationshipbetween mankind and nature. In this perspective, a long-term target was introducedbased on the global capacity of renewable energy production and global equity inenergy terms. Dürr [1994] conceptualised this issue by arguing a long-term energytarget of 1-1.5 kW per person to be met in 2050 which is based on the maximalamount of renewable energy which can be produced per person if one assumes fullglobal equity in energy service terms. In case of The Netherlands, Biesiot [1998]showed that this energy target cannot be met solely by technology improvements andchanging consumer packages implying that the consumption growth rates shoulddecrease. Biesiot’s study mainly depends on a semi-dynamic or quasi static

174

approaches which holds that the corresponding dynamics are poorly dealt with. It isadvocated here that future versions of NLDREAM which involve a more detailedrepresentation of the production sectors should be used to study dynamically thesetopics.

9.4.3 DREAM-Models for OECD-Europe

As mentioned before, DREAM-methodology was introduced as a tool to study thelong-term consequences on net imports and on environmental stress when changingconsumption and export growth rates. As result of differences in the key influences,ECCO-models and DREAM-models behave in different ways. Economic activityshowed exponential growth patterns in the scenarios developed with the aid of theDREAM-modelling approach whereas in ECCO-models the growth rates of theeconomy tended to decrease over time. That is, as a consequence of the absence ofa negative feedback loop the DREAM-approach leads to increasing growth patterns.In principle, economic growth is not decelerated as long as consumption and exportlevels remain growing. This does not imply that sustainability aspects related to theseeconomic development paths cannot be indicated by the DREAM-modellingapproach. For instance, CO2-emissions due to fossil energy use and the import-exportbalance indicating the dependence on foreign resources are used to study the impactof economic growth levels. In the DREAM-approach, options can be included thatrestrict endogenously economic growth, for instance by adopting conditions such asconstraining the import-export ratio or by setting CO2-emission limits (e.g. consistentwith the Kyoto target). These kinds of additional constraints require someadjustments in the models and they are complementary to the main objectives of theDREAM-approach which involves assessing dynamically the energy costs ofchanging energy patterns. Herewith, the DREAM-modelling approach seems to bea useful tool to address in a system dynamic way the relationship between changesin consumer activity underlying the corresponding structural changes in productionssectors and environmental related issues.

Similarly to NLDREAM, scenarios developed with the DREAM-modellingapproach for OECD-Europe showed that high growth rates of consumption coincidewith high energy demands. These high energy demands were associated by increasingCO2-emissions and by an increasing dependence on foreign energy resources. Thelatter was also the result of depleting energy reserves in OECD-Europe. The‘growth’ scenarios presented illustrated the impact of consumption growth rates inOECD-Europe. However, the starting points of these scenarios involved many moreassumptions on future energy savings, fuel mixes, consumption and export growthrates than the scenarios developed with NLDREAM. Extrapolating the trends inconsumption and export growth rates observed for the 1985-1995 period indicatedthat economic development in OECD-Europe is still associated with an increasingenergy use even when the corresponding growth rates are decelerated.

175

Sensitivity analyses showed that the outcomes of the DREAM-modellingapproach are most sensitive to changes in the export growth rates and theconsumption per capita growth rates. Compared to these two parameters, the modelseemed insensitive to changes in the energy intensities in the production sectorsimplying that the outcomes of the model are hardly affected by errors in the initialdata set.

The multi-regional modelling approach showed that regional differences can resultin other aggregated results than a single region approach. Not only does this resulthold for extrapolating regional differences in the consumption and export growthrates but it can also hold for certain changes in one region (e.g. in obtaining a nuclearfree electricity supply). These outcomes are consistent with that of the ECCO-modelling approach and support the idea that multi-regional approaches result in amore accurate outcomes than a single region approach.

9.4.4 General Remarks about the DREAM-Modelling Approach

The DREAM-modelling approach was developed in order to be used in theresearch program called HOMES (HOuseholds Metabolism Effectively Sustainable)that was initiated to study the physical throughput of energy and materials throughhouseholds [Noorman et al., 1998]. In this context, the DREAM-modelling approachshould be used to investigate the long-term consequences of the energy costsassociated with changing consumption patterns by households for The Netherlands.The second part demonstrated that in principle the DREAM-modelling approach isan appropriate tool to study the energy costs of changing consumption patterns in asystem dynamic way. However, the results presented are developed at a macro levelimplying that principally various consumption growth rates can be considered, butthat changes in consumer packages cannot be dealt with properly. Therefore, theproduction sectors should preferably be dealt with at micro/meso level to facilitatethe assessment of changes in energy cost due to changes in consumption packages.The level of aggregation is mainly limited by the availability of data concerning itemssuch as input-output tables and the direct energy use by the production sectors whichare used to determine the initial values of the models. In case of The Netherlands,these data are availably at a level of 59 sectors. This kind of aggregation level shouldbe sufficient for making a first step in linking changes in consumers packages tochanging demand for output in the production sectors. In this perspective, Kramer etal. [1998] and Biesiot [1998] determine in a quasi-static way changes in energy usedue to changes in consumer activity at a very detailed level, for instance byaddressing the energy costs of purchasing a waterbed. The energy costs of these kindof changes in consumption patterns are based on a hybrid method of energy analysisthat is a mixture of process analysis and input-output analysis involving anaggregation level of 59 sectors [Wilting, 1996]. Besides the data requirement fordetermining the initialising the model, data should also be available for consumption

176

patterns per household type to include demographic changes. One disadvantage of considering a detailed level of aggregation involves the

assumption that should be made about future consumption patterns per householdtype, demographic changes, energy saving potentials, and the technological changes(represented by the technology matrix). Assumptions about these topics may havestrong implications for the uncertainty of outcomes. Despite these shortcomings, itis advocated that these detailed aspects should be included in the DREAM-modellingapproach as it facilitates assessing the energy costs associated with consumptionpatterns in a system dynamic way. Hence, it can be used to investigate how to directconsumption patterns towards sustainability.

Summarising the main research question of the second part of this thesis, theoutcomes of the ECCO-modelling approach differ considerably from that of theDREAM-modelling approach. In addition, the DREAM-modelling approach appearsto be a useful tool to study the long-term primary energy demand consequences ofchanging consumption and export growth rates.

9.5 General Conclusions and Reflections

Returning to the perspectives of the energy accounting approach, both the ECCO-modelling approach and the DREAM-approach study the relationship between theeconomic system and the environmental system from a physical perspective. Bothapproaches address the energy use associated with economic activity.

The ECCO-approach emphasises the notion that energy is a key factor ineconomic production processes by constraining economic development by theavailability of energy resources or energy embedded in capital. In this way, itdetermines the potential of the economy to grow under physical constraints.

In essence, the DREAM-modelling approach was developed to dynamicallyaddress the long-term consequences of the energy use as result of technologyimprovements, changing consumption patterns, energy savings and demographicchanges. The approach, thus, determines the environmental impacts correspondingto specific economic development paths. In addition, the approach can be used toinvestigate which changes or measures are required to direct consumption patternstowards sustainability.

Obviously, the two modelling approaches do not result in ‘scientifically solving’the problems concerning economic activities and the stress on the environment.However by using both approaches, the insights gained from studying the (physical)interactions between the economic system and the environmental system makes theincorporation of physical approaches into economic theory worthwhile. Bothapproaches should be used together with conventional economic approaches in orderto investigate the relationships between the economic system and the environmentalsystem in an integrated way.

177

9.6 Suggestions for Future Research

Throughout sections 9.2-9.5, shortcomings or limitations of the ECCO-approachas well as the DREAM-approach are addressed and in most cases options arepresented to deal with them. These findings are summarised in this section.

The DREAM-models presented in the second chapter involve a rather highaggregation level which hampered assessing the energy costs of changingconsumption patterns at a micro/meso level. Hence, the production sectors should beconsidered in much more detail. In addition, consumer activities should bedistinguished at the level of households in order to take into account demographicchanges. Obviously these findings apply to the DREAM-model for The Netherlandsas well as for OECD-Europe. In the particular in case of the DREAM-model forOECD-Europe, improvements are necessary in the way demographic topics are dealtwith. In both models, a number of calculation methods calls for improvements suchas the determination of the energy associated with private transportation anddwellings and the investments required to make the energy savings possible.

Both the ECCO-models and the DREAM-models for OECD-Europe involve anumber assumptions as a result of lack of data. A number of aspects can be dealtwith more properly when more accurate and consistent data become available. Thisespecially applies to data and the corresponding level of aggregation concerning theregional input-output data and the direct energy use by the production sectors whichare used to calculate the initial values of the models for OECD-Europe.

Differences between OECD-Europe and non-European regions are poorlyaddressed in the multi-regional models for OECD-Europe. This also applies todetermining the energy intensities of the production sectors in OECD-Europe. Theseregional aspect can be dealt with in more detail by specifying the imports in terms oftheir origin: OECD-Europe versus non-European regions. This alternative requiresthe availability of average embodied energy intensities data for each region outsideOECD-Europe.

It is argued here that despite some shortcomings (mainly as result of lack of data)both the DREAM and the ECCO-modelling approaches presented in this thesis areuseful tools to study the economy-environment interface from a physical point ofview. Dealing with these shortcomings as suggested above will most probably resultin more accurate descriptions of this interface. In particular, developing DREAM-models at a micro/meso level would facilitate studying dynamically the consequencesfor the total primary energy demand as result of changing consumer activitiesunderlying structural changes in the production sectors.

179

References

Aguilera-Klink, F. (1994); Some Notes on the Misuse of Classic Writings on the Subjectof Common Property, Ecological Economics, 9(4), pp.221-228.

Atomwirtschaft (1986); Jahrbuch der Atomwirstschaft, 17 Jahrgang, VerlagsgruppeHandelsblatt, Düsseldorf (in German).

Atomwirtschaft (1987); Jahrbuch der Atomwirstschaft, 18 Jahrgang, VerlagsgruppeHandelsblatt, Düsseldorf (in German).

Bacharach, M. (1970); Biproportional Matrices & Input-Output Change, University Press,Cambridge.Baguant, J. and Slesser, M. (1991); Exploring the Potential for Enhanced National Welfare

in Mauritius Using ECCO ‘Economic Co-ordination Model’, in: Gilbert, A.J. andBraat, L.C. (Editors); Modelling for population and Sustainable Development,Routledge London.

Battjes, J.J. (1993); Resource Accounting in de Nederlandse Landbouw, naar een meerduurzame landbouw, IVEM students report no. 68, Centre for Energy andEnvironmental Studies (IVEM), University of Groningen, The Netherlands (in Dutch).

Battjes, J.J. and Noorman, K.J. (1998); Multi-regional ECCO-Modelling Approach forOECD-Europe, paper prepared for: Resource Accounting Modelling Workshop,Groningen, 16-17 September 1998, IVEM-research report no. 96, University ofGroningen, The Netherlands.

Battjes, J.J., Noorman, K.J., and Biesiot, W. (1998); Assessing the Energy Intensities ofImports, Energy Economics, 20 ,pp.67-83, Elsevier.

Beeldman, M., van Dril, A.W.N., Jeeninga, H., and Koustaal, P.R. (1998); Mogelijkhedenvoor CO2-Reductie in 2000, ECN-report no. C98-041, Energy Research Centre, Petten,The Netherlands (in Dutch).

Benders R.J.M. (1996); Interactive Simulation of Electricity Demand and Production,Ph.D.-thesis, University of Groningen, Groningen.

Biesiot, W. (1998); Fragmenten voor een Droom, Duurzame Onwikkeling Dicht bij Huis,Van Arkel, Utrecht, The Netherlands (in Dutch).

Biesiot, W. and Moll, H.C. (Eds.) (1995); Reduction of CO2-emissions by LifestyleChanges, IVEM-research report 80, Center for Energy and Environmental Studies,University of Groningen.

Biesiot, W. and Noorman, K.J. (1999); Energy requirements of Household Consumption(a case study of The Netherlands), Ecological Economics (in press).

Blaug, M. (1985); Economic Theory in retrospect, 4th edition, Cambridge University Press.Blin, J.M. and Cohen, C. (1977); Technological Similarity and Aggregation in Input-

Output Systems: a cluster-analytic approach, The Review of Economics and Statistics,Vol LIX, 1977, p 82-91.

Boomsma, P., van der Linden, J. and Oosterhaven J. (1991); Note on Comparability ofExport and Import Data from E.C. Input-Output Tables and Trade Statistics, MimeoUniversity of Groningen, Department of Economics.

Bouma, J.L. (1988); Leerboek der Bedrijfseconomie, Deel II, 2nd edition, Delwel,Wassenaar (in Dutch).BP (1994); Statistical Review of World Energy June 1994, British Petroleum Company,

180

Dorset, England.Boulding, K.E. (1966); The Economics of Spaceship Earth, in: Jarret, H. (Ed.);

Environmental Quality in a Growing Economy, pp. 3-14, John HopkinsPress,Baltimore.Boulding, K.E. (1976); The Great Laws of Change, in: Tang, A., Westfield, F., and

Worley, J. (ed.); Evolution, Welfare, and Time in Economics, Lexington Book,Massachusetts.

Braat, L.C., Brouwer, F.M., Gilbert, A.J., Hulzebos, E., and Hafkamp, W.A. (1987);Integrated Modelling of Renewable Natural Resources: the Economic-EcologicInterface, Fast Occasional paper No.172, Commission of the European Community.

Brewer, J. and Porter, R. (Eds.) (1993); Consumption and the World of Goods, Routledge,London.

Bullard, C.W. and Herendeen, R. (1975); The Energy Costs of Goods and Services, EnergyPolicy, December 1975, p. 268-278.

Casler, S. (1983); Correcting Input-Output Coefficients for Capital Depreciation, EnergySystems and Policy, Volume 7, no. 3, p. 171-193.

CBS (1986); Uitkomsten Bevolkings- en Huishoudensprognose 1985, Mndstat 86/1,Central Office for Statistics, The Hague, The Netherlands, (in Dutch).

CBS (1987); Nationale Rekeningen, Central Office for Statistics, The Hague, TheNetherlands, (in Dutch).

CBS (1996); Uitkomsten Bevolkings- en Huishoudensprognose 1995, Mndstat 96/1,Central Office for Statistics, The Hague, The Netherlands, (in Dutch).

CBS (1997); Nationale Rekeningen, Central Office for Statistics, The Hague, TheNetherlands (in Dutch).

Central Electricity Generating Board (1992); Annual Report and Accounts 1985/1986,Department of Public Affairs Central Electricity Generating Board, London.

Chapman, P.F., Leach, G., and Slesser, M. (1974); The Energy of Costs Fuels, EnergyPolicy, Vol 2, p 231-243.

Christensen, P. (1997); Different Lifestyles and Their Impact on the Environment,Sustainable Development, Vol.5 No 1, pp. 30-36, John Wiley & Sons, Chisester.

Common, M.S. and Salma, U. (1992); Accounting for Changes in Australian CarbonDioxide Emissions, Energy Economics, July 1992, pp. 217-225, Buuterworth-Heinemann Ltd..

Costanza, R. (1980); Embodied Energy and Economic Valuation, Science, Vol. 210, pp.1219-1224.

Costanza, R. (1986); Embodied Energy and Economic Valuation, in: [Sohn, 1986], p. 432-445.

Costanza, R., Daly, H.E., and Barthlomew, J.A. (1991); Goals, Agenda and PolicyRecommendations for Ecological Economics, in: Costanza R. (ed.); EcologicalEconomics: the science and management of sustainability, pp. 1-20, ColumbiaUniversity Press, New York.

CPB (1996); Omgevingsscenario’s Lange Termijn Verkenning 1995-2020, Workingdocuments No 89, Dutch Central Planning Buro, Hague, The Netherlands (in Dutch).

CPB (1997); Economie en Fysieke Omgeving, beleidsopgaven en oplossingsrichtingen1995-2020, Dutch Central Planning Buro , the Hague, The Netherlands (in Dutch).

Crane, D. (1995); Regional UK ECCO, Users Manual, Centre for Human Ecology,University of Edinburgh.

181

Daly, H.E. (1993); Sustainable Growth: An Impossible Theorem, in: [Daly and Townsend,1993], pp. 267-274.

Daly, H.E. and Townsened, K.N. (1993); Valuing the Earth, Economics Ecology Ethics,387 pg., the MITT Press Massachusetts.

Dasgupta, P.S. and Heal, G.M. (1969); Economic Theory and Exhaustible Resources,Cambridge University Press.

Davis, S.C. (1996); Transportation Energy Data Book, edition 15 on Internet, http://www-cta.ornl.gov.cta/data/tedb15.

De Beer, J.G., van Wees, M.T., Worrell, E., and Blok, K. (1994); ICARUS-3, the Potentialof Energy Efficiency Improvements in The Netherlands up to 2000 and 2015,Department of Science, Technology and Society, Utrecht University, The Netherlands.

De Vries, H.J.M. (1993); Between Purchasing Power and the World of Goods:Understanding the household economy in early modern Europe, in: Brewer, J. andPorter, R. (Eds.) (1993); Consumption and the World of Goods, Routledge, London.

De Vries, H.J.M. and Janssen, M.A. (1996); Global Energy Future: An IntegralPerspective with the TIME -model, GLOBO Report Series No. 18., Institute for PublicHealth and the Environment (RIVM, Bilthoven, The Netherlands.

Dietz, F., Hafkamp, W., and van der Straaten, J. (1994); Basisboek Milieu-Economie,Boom, Amsterdam (in Dutch).

Dietz, F., and van der Straaten, J. (1992); Rethinking Environmental Economics: MissingLinks between Economic Theory and Environmental Policy, Journal of EconomicIssues, Vol. XXVI, No. 1, pp. 27-51.

Duchin, F. (1995); Global Scenarios about Lifestyle and Technology, paper presented atthe conference ‘The sustainable Future of the Global System’,United NationsUniversity, October 1995, Tokyo.

Dürr, H.P. (1994); Sustainable, Equitable Economics, the Personal Energy Budget, in:Smith, P.B., Okoye, S.E., de Wilde, J., and Deshnkar, P.; The World at the Crossroads,Towards a Sustainable, Equitable and Liveable Word, Earthscan, London.

ECN (1997) Bijlage met energietabellen basisscenario’s Nationale Energie Verkenningen1995-2020, Dutch Energy Research Centre, Petten, The Netherlands (in Dutch).

ECN (1998); Nationale Energie Verkenningen 1995-2020, Dutch Energy Research Centre,Petten, The Netherlands (in Dutch).

EDF (1992); Resultats Techniques d’Exploitation 1991, Electricité de France (in French).Edmonds, J. and Reilly, J.M. (1985); Global Energy, Assessing the Future, Oxford

University Press, New York. Ehrlich, P. and Holdren, J. (1971); Impact of Population Growth, Science 171, pp.1212-

1217.EMF (1997); EMF14 database on diskette, Energy Modelling Forum. England, R.W. (1997); Evolutionary Concepts in Contemporary Economics, 4th edition,

pp. 193-212, University of Michigan Press.Ente azionale per l’energia Elettrica (1987); Production and Consumption of electricity in

Italy 1986 (in Italian).Eurostat (1992a); National Accounts ESA, Input-Output tables for France 1985,

Luxembourg.Eurostat (1992b); National Accounts ESA, Input-Output tables for Germany 1985,

Luxembourg.

182

Eurostat (1992c); National Accounts ESA, Input-Output tables for Ireland 1985,Luxembourg.

Eurostat (1992d); National Accounts ESA, Input-Output tables for Italy 1985,Luxembourg.

Eurostat (1992e); National Accounts ESA, Input-Output tables for Spain 1985,Luxembourg.

Faber, M. and Proops, J. (1985); Interdisciplinary Research between Economists andPhysical Scientists: Retrospect and Prospect, Kyklos, 38, pp.299-616.

Faucheux, S. and O’Connor, M. (1997); Valuation for Sustainable Development: methodsand policy, Edward Elgar, Cheltenham, UK.

Fisher, W.D. (1986); Criteria for Aggregation in Input-Output Analysis, in: [Sohn, 1986],p 210-225.

Foran, B. and Crane, D. (1998) the OzECCO Embodied Energy Model of Australia’sPhysical Economy, paper presented at International Workshop on Advances in EnergyStudies: Energy Flows in Ecology and Economy, May 27-31 1998, Porto Venere, Italy.

Forrester, J. (1961); Industrial Dynamics, Cambridge MIT Press. Fritsche, U.R., Leuchtner, J., Matthes, F.C., Rausch, L., and Simon, K.H. (1994); Gesamt-

Emissions-Modell Integrierte Systeme (GEMIS) version 2.1, Oeko-Institut, Kassel (inGerman).

Georgescu-Roegen, N. (1971); The Entropy Law and the Economic Process, HarvardUniversity Press, Cambridge, Massachusetts.

Georgescu-Roegen, N. (1986); The Entropy Law and the Economic Process in Retrospect,Eastern Economic Journal, 12, pp.3-25.

Georgescu-Roegen, N. (1993); The Entropy and the Economic Problem, in: [Daly andTownsend, 1993], pp. 75-88.

Gilbert A.J. (1991); Introduction: Issues and Perspectives, in: Braat, L.C. and Gilbert,A.J.; Modelling for Population and Sustainable Development, Routledge, London.

Gilliland, M. (1978); Energy Analyses: a new public tool, AAAS Symposium series,Washington DC.

Gowdy, J.M. and Miller, J.L.; ‘An Input-Output Approach to Energy Efficiency in the USAand Japan (1960-1980)’, Energy, Vol 16, No. 6, pp. 897-902 (1991).

Green, M.J. and le Grontec, P. (1976); Methodology of the communal input-output tables,1970-1975 (in Dutch), Eurostat, Luxemburg.

Hall, R.E. and Taylor, J.B. (1988); Macro-Economics, 2nd edition, W.W. Norton &Company Inc, New York

Hardin, G. (1968); Tragedy of the Commons, Science, Vol. 162, pp. 1243-1248. Harris, J.M. and Goodwin, N.R. (1995); A Survey of Ecological Economics, Island Press,

Washington.Heijmann, W.J.M. (1991); Depletable Resources and the Economy, Ph.D.-thesis,

University of Wageningen.Hetherington, R. (1994); An Input-Output Assessment of Carbon Dioxide Release within

the UK, Working paper no. 3, Resources Research Unit, School of Urban and RegionalStudies, Sheffield Hallam University.

Hilten, O. Beeldman, M., Boonekamp, P.G.M., van Dril, A.W.N., Gielen, D.C., Kok, I.C.,and Kroon, P. (1996); De ECN-Bijdrage aan de Derde Energie Nota, uitgebreidebeschrijving energieschtesen 2020, report no. ECN-C-96-014, Energie Centrum

183

Nederland, Petten, the Netherlands (in Dutch).Hotelling, H.(1931); The Economics of Exhaustible Resources, Journal of Political

Economy 39.IFIAS (1974); Energy Analysis, Workshop report no. 6, Energy Analysis Workshop on

Methodology and Conventions 25-30 August, Guldsmedshyttan, Sweden.Jeffers, J.N.R. (1987); Ecological Modelling: Shortcomings and Perspectives, in: Braat,

L.C. and van Lierop, W.F.J. (Eds); Economic Ecological Modelling, Elsevier SciencePublisher, New York.

King, J. (1991): The ECCO-Approach to Planning Sustainable Development, in: Braat,L.C. and Gilbert, A.J.; Modelling for Population and Sustainable Development,Routledge, London.

Klein Goldewijk, C.G.M. and Battjes, J.J. (1997); A Hundred Year (1890-1990) Data Basefor Integrated Environment Assessment (HYDE version 1.1), RIVM report no.422514002, National Intsitute of Public Health and Environmental Protection (RIVM),Bilthoven, The Netherlands.

Kneese, A.V., Ayres R.U., and d’Arge R.C. (1970); Economics and the Environment, aMaterial Balance Approach, John Hopkins University Press, Baltimore.

Kramer, K.J., Wiersema, G. Gartesleben, B.C.M., Noorman, K.J., and Biesiot, W. (1998);Huishoudelijk Energiegebruik Gemeten, Aard Omvang en Veranderingspotentieel,IVEM-research report no. 93, Centre for Energy and Environmental Studies (IVEM),University of Groningen, The Netherlands (in Dutch).

Loening, E.U. (1991); Introductory Comments: the Challenge for the Future, in: Braat,L.C. and Gilbert, A.J.; Modelling for Population and Sustainable Development,Routledge, London.

Martinez-Alier, J. (1987); Ecological Economics, 186 pg., Basil Blackwell, Oxford, UK.McKendrick, N. (1982) Commercialization and the Economy, in: McKendrick, N., Brewer,

J., Plumb, J.H.P. (1982); The Birth of Consumer Societ, Commercialization ofEighteenth-Century England, Europa, London.

Meadows D., Randers, J., and Behrens III, W . (1972); The Limits to the Growth, UniverseBooks. New York.

Melman, A.G., Boot, H., and Gerritse, G (1990); Energiebesparingspotentiëlen 2015,eindrapport, TNO-Milieu en Energie Apeldoorn, The Netherlands (in Dutch).

Miller, R.E. and Blair, P.D. (1985); Input-Output Analysis: foundations and extensions,Prentice-Hall Inc., New Jersey.

Moll, H.C. (1993); Energy Counts and Matter Matters in Models for SustainableDevelopment, Dynamic Lifecycle Modelling as a Tool for Desig and Evaluation ofLong-term Environmental Strategies, Ph.D.-dissertation University of Groningen,STYX publications, Groningen, the Netherlands.

Morita, T. And Lee, H-C. (1997); Development of an Emission Scenario Database forSimulations Conducted since 1994 and a Preliminary Analysis Using the Data Base,draft version for contribution to special IPCC report on Greenhouse Gas Emissions.

Mulder, H.A.J. and Biesiot, W. (1998); Transition to a Sustainable Society; A BackcastingApproach to Modelling Energy and Ecology, Edward Elgar, Cheltenham, UK.

Munda, G. (1997); Environment Economics, Ecological Economics, and the Concept ofSustainable Development, Environmental Values, 6, pp. 213-233, White Horse Press,Cambridge UK.(1997)

184

Nakiƒenoviƒ, N. (1996). Freeing Energy from Carbon, Daedadus Volume 125, no.33, pp.95-112. Nakiƒenoviƒ, N., Grübler, A., Ishitani, H., Johansson, T., Marland, G., Moreira, J.R., and

Rogner H-H (1997); Energy Primer, IIASA-report RR-97-1, Laxenburg, AustriaNentjes, A. (1983); De Ontwikkeling van de Economische Theorie, 5th edition, pp. 175-208,

Wolters Noordhof, Groningen, the Netherlands.Nieuwlaar, E. (1988); Developments in Energy Analysis, thesis, University of Utrecht,

Utrecht.Noorman, K.J. (1995); Exploring Futures From an Energy Perspective, A Natural Capital

Accounting model to study into the long-term economic development of theNetherlands, Ph.D.-thesis, University of Groningen, the Netherlands.

Noorman, K.J. and Crane, D. (1995) Personal Communication during Second InternationalSymposium on Energy-Based Models, 28-30 June, Edinburgh, Scotland.

Noorman, K.J., Biesiot, W., and Schoot Uiterkamp, A.J.M., (1998); HouseholdMetabolism, in: Noorman, K.J. and Schoot Uiterkamp, A.J.M. (Eds.) (1998); GreenHousholds? Domestic Consumers, Environment and Sustainability, EarthscanPublications Ltd., London.

Noorman, K.J. and Kamminga, K.J. (1998); Reducing Residential Energy Use for aSustainable Future, Sustainable Development, Vol. 6 no.3 , John Wiley & Sons,Chisester.

Noorman, K.J. and Moll, H.C. (1998); Energy Flows in Ecology and Economy: Frommethodological approaches to real life applications, paper presented at theinternational workshop advances in energy studies, energy flows in ecology andeconomy, May 26th - 30th 1998, Porto Venere, Italy.

Noorman, K.J. and Schoot Uiterkamp, A.J.M. (Eds.) (1998); Green Housholds? DomesticConsumers, Environment and Sustainability, Earthscan Publications Ltd., London.

Nijkamp, P. (1987); Economic Modelling: Shortcomings and Perspectives, in: Braat, L.Cand van Lierop, W.F.J. (Eds); Economic Ecological Modelling, Elsevier SciencePublisher, New York.

O’Connor, M. (1998a); Ecological-Economics Sustainability, in: Faucheux, S. AndO’Connor, M.; Valuation for Sustainable Development, methods and policy indicators,Edward Elgar, Cheltenham, UK.

O’Connor, M. (1998b); Outer and Inner Energy Input Constraints on Economic Activity,paper presented at the Workshop ‘Advances in Energy’, Porto Venere, 27-31 May,1998.

Odum, H.T. (1994); The Emergy of Natural Capital, in: Jansson, A., Hammer, M., Folke,C., and Costanza, R; Investing in Natural Capital: the Ecological Economics approachto Sustainability, Island Press, Washington DC.

OECD (1987); Annual Oil and Gas Statistics, International Energy Agency, Organisationfor Economic Development and Co-operation, Paris.

OECD (1989); World Energy Statistics and Balances 1971-1987, International EnergyAgency, Organisation for Economic Development and Co-operation, Paris.

OECD (1991a); Energy Balances of OECD-countries 1980-1989, International EnergyAgency, Organisation for Economic Development and Co-operation, Paris.

OECD (1991b); Energy Statistics of OECD-countries 1980-1989, International EnergyAgency, Organisation for Economic Development and Co-operation, Paris.

185

OECD (1991c); Foreign Trade by Commodities, 1990, volume 1-5, International EnergyAgency, Organisation for Economic Development and Co-operation, Paris.

OECD (1992); Electricity Information 1992, International Energy Agency, Organisationfor Economic Development and Co-operation, Paris.

OECD (1993a); National Accounts, Detailed tables volume II, 1979-1991, Organisationfor Economic Development and Co-operation, Paris.

OECD (1993b); Industrial Structure Statistics, Organisation for Economic Developmentand Co-operation, Paris.

OECD(1993c); Energy Balances of OECD-countries 1990-1991, International EnergyAgency, Organisation for Economic Development and Co-operation, Paris.

OECD (1995a); International Sectoral Data Base, 1960-1994, statistics on diskette,Statistics Directorate, Organisation for Economic Development and Co-operation, Paris.

OECD (1995b); Statistical Trends in Transport 1965-1990, European Conference ofMinisters of Transport, Organisation for Economic Development and Co-operation,Paris.

Office of Technology Assessment (1990); Energy Use and the U.S. Economy, Backgroundpaper, congress of the United States.

Oosterhaven, J. and van der Linden, J.A. (1995); the Input-Output tables of the UK for1985 and of The Netherlands for 1987 are derived from the department of Economics,University of Groningen, The Netherlands (personal communciation).

Opschoor, J.B. (1991); Sustainable Development: An Institutional Approach, paperpresented at the Annual Meeting of the European Association of the Environmental andResource Economists, Stockholm School of Economics. June 10-14, 1991

Opschoor, J.B. and Van der Straaten, J. (1993); Sustainable Development: an InstitutionalApproach, Ecological Economics, 7(3), pp.203-222.

Owino, J.D. (1991); Sectoral Study - Exploring the Potential of Carrying CapacityAssessment, population growth and sel-sufficiency in Kenya Using ECCO, in: Gilbert,A.J. and Braat, L.C. (Editors); Modelling for population and Sustainable Development,Routledge London.

Pearce, D.W. and Turner, R.K. (1990); Economics of Natural Resources and theEnvironment, pp. 271-287, Harvester Wheatsheaf, New York.

Peet, J. (1992); Energy and the Ecological Economics of Sustainability, IslandPress,Washington D.C..

Perman, R., Ma, Y., and McGilvray, J., (1996); Natural Resource & EnvironmentalEconomics, pp. 116-142, Addison Longman, London.

Perrings, C. (1987); Economy and the Environment: A Theoretical Essay on theInterdependency of Economics and Environmental Systems, Cambridge UniversityPress.

Pietilä, H. (1997); The Triangle of the Human Economy, Household - Cultivation -Industrial Production, An attempt at making visible the human economy in toto,Ecological Economics 20, pp. 113-127, Elsevier.

Pigou, A.C. (1920); The Economics of Welfare, Macmillan, London.Prigogine, I. (1976); Order through Fluctuation: Self-Organisation and Social Systems, in:

Jantsch, E. and Waddington, C; Evolution and Consciousness: Human Systems intransition, pp. 93-133, Addison-Wesely, Massachusetts.

Proops, J.L.R (1988); Energy intensities, input-output analysis and economic development,

186

in: Ciaschini, M. (editor, 1988); Input-Output Analysis, current developments,Chapman and Hall, New York, pp. 201-216.

Radzicki, M.J. and Sterman, J.D. (1994); Evolutionary Economics and System-dynamics,in: England, R.W. (ed.); Evolutionary Conspets in Contemporary Economics, theUniversity of Michigan Press.

RIVM (1997); Nationale Milieuverkenning 1997-2020, National Institute of Public Healthand Environmental Protection (RIVM), Samson H.D., Tjeenk Willink BV Alphen aanden Rijn.

RMNO (1992) Duurzame Economie, een invertarisatie van onderzoeksbehoeften, Raadvoor Milieu- en Natuuronderzoek, publicatie no.75, Alphen aan de Rijn, theNetherlands (in Dutch).

Ruth, M. (1993); Integrating Economics Ecology and Thermodynamics, Kluwer AcademicPublishers, Dordrecht, the Netherlands.

Ryan, G.J. (1995); Dynamic Physical Analysis of Long-Term Economy-EnvironmentOptions, Ph.D.-Thesis, University of Canerbury.

Ryan, G., Méral, P., Schembri, P., and Zyla, E. (1998); Structural Economy-EnvironmentSimulation Modelling, in: Faucheux, S. And O’Connor, M.; Valuation for SustainableDevelopment, methods and policy indicators, Edward Elgar, Cheltenham, UK.

Samuelson, P.A. (1967); Economics,7th edition, McGraw-Hill Book Company, New York.

Samuelson, P.A. (1972); Maximum Principles in Analytical Economics, AmericanEconomic Review, Vol. 62 (June), pp. 249-162.

Schembri, P. (1998); Adaption Costs to Sustainable Development and EcologicalTransitions, A presentation of the structural model M3ED, paper presented at the ThirdInternational Symposium on Energy-Based Models and Beyond, Groningen, TheNetherlands, 16-17 September, 1998.

Schenk, E.A.M. (1998); Households, Automobility and Emissions, the Dutch case, Ph.D.-Thesis, University of Maastricht, The Netherlands.

Schuurman, A. (1997); Aards Geluk: consumptie en de moderne samenleving, in:Schuurman, A., de Vries, J. and van der Woude, A. (Eds), Aards Geluk, pp. 11-27,Balans, Amsterdam., The Netherlands (in Dutch).

Simon, H. (1981); The Science of the Artificial, MIT Press Cambridge.Sintunawa, C. (1991); The Integrated Approach to Long-Term Resource Planning and

National Carrying Capacity Assessment for Thailand, in: Gilbert, A.J. and Braat, L.C.(Editors); Modelling for Population and Sustainable Development, Routledge London.

Slesser, M and King, J. (1988); Resource Accounting: An Application to DevelopmentPlanning, World Development, Vol. 16, No. 2, pp. 293-303, Pergamon Press.

Slesser, M. (1991); Planning for Sustainable Development, in: Braat, L.C. and Gilbert,A.J.; Modelling for Population and Sustainable Development, Routledge, London.

Slesser, M., King, J., Revie, C., and Crane, D. (1994); ECCO, Evaluation of CapitalCreation Options, Non-Monetary Indicators of Sustainability ,A Report to EC DG XII.

Slesser, M., King, J., and Crane, D.C. (1997); The Management of Greed, a Bio-PhysicalAppraisal of Environment and Economic Paradigm, Resource Use Institute, Ltd.,Dunblane, Scotland.

Söderbaum, P. (1990); Neoclassical and Institutional Approaches to EnvironmentalEconomics, Journal of Economic Issues, Vol. XXIV, No. 2, pp. 481-492.

187

Sohn, I. (Editor, 1988); Readings in Input-Output Analysis, theory and applications,Oxford university press, New York.

Solow, R. (1956); A contribution to the Theory of Economic Growth, Quarterly Journalof Economics, 70, pp. 65-94.

Sontag, E.D. (1990); Mathematical Control Theory, Deterministic Finite DimensionalSystems, Springer-Verlag, New York..

Statistische Bundesamt (1994); Volkswirtschftliche Gesamtrechnungen, Input-Output-Tabellen 1986, 1988, 1990. Fachserie 18, Reihe 2, Wiesbaden (in German).

Szymer, J. (1989); The Trade-Off between and Information in the RAS-Procedure, in:Miller, R.E., Polenske, K.R. and Rose, A.Z. (editors), Frontiers of Input-OutputAnalysis, pp. 258-278, Oxford University Press, New York.

Tietenberg, T. (1988); Environmental Resource Economics, second edition, pp. 38-63, 113-136, Scott Foresman Company, Boston.

Tietenberg, T. (1993); Using Economic Incentives to Maintain our Environment, in: [Dalyand Townsend, 1993], pp. 315-324.

TNO (1976); The Energy Accounting of Materials, Products, Process and Services, 9th

International TNO Conference, 26-27 February, Rotterdam, the Netherlands.Turner, R.K. (1998) Household Metabolism and Sustainability, in: Noorman K.J. and

Schoot Uiterkamp A.S.U.; Green Households? Domestic Consumers, Environment andSustainability, Earthscan , London.

UN (1987); Gas Statistics for Europe, Economic Commission for Europe, United Nations,Geneva.

UN (1988a); Electric Energy Statistics for Europe, Economic Commission for Europe,United Nations, Geneva.

UN (1988b); General Energy Statistics for Europe, Economic Commission for Europe,Geneva.

UN (1990); National Accounts Statistics: main aggregates and detailed tables, 1988, partI & II, United Nations, New York.

UN (1996); National accounts Statistics: main aggregates and detailed tables, 1993, partI & II, United Nations, New York.

UN (1997); Kyoto Protocol to the United Nations Framework Convention on ClimateChange, Conference of the Parties, 1-10 December 1997, third session, agenda item 5,United Nations.

Unesco and FAO (1985); Carrying Capacity Assessment with a Pilot Study of Kenya: aresource accounting methodology for exploring national options for sustainabledevelopment, Paris and Rome.

Van den Broecke S.R. (1988); Groei en groeimogelijkheden personenenautpark inNederland, Project Integrale Verkeers en Vervoerstudies, Den Haag (in Dutch).

Van der Linden, J. (1999); Interdependence and Specialisation in the European Union,Intercountry Input-Ouput Analysis and Economic Integration, Ph.D.-Thesis, Universityof Groningen, The Netherlands

Van der Linden, J. and Oosterhaven, J. (1995); European Community Intercountry Input-Output analysis: Construction Method and Main results for 1965-85, Economic SystemsResearch, Volume, No. 3, p. 249-269.

Van Engelenburg, B.C.W., van Rossum, T.F.M., Blok, K., Biesiot, W., and Wilting, H.C.,(1991); Energiegebruik en Huishoudelijke Consumptie: handleiding en toepassing,

188

NW&S research report no. 91032, Utrecht, The Netherlands.Van Hilten, O., Beeldman, M., Boonekamp, P.G.M., van Dril, A.W.N., Gielen, D.J., Kok,

I.C., and Kroon, P. (1996); De ECN-Bijdrage aan de Derde Energienota, uitgebreidebeschrijving energieschetsen 2020, ECN-research report no. ECN-C--96-014, Petten,The Netherlands (in Dutch).

Van Ierland, E.C., Heijman, W.J.M., Kroese, E.P., and Oskam, E.A. (1994); Grondslagenvan de Macro-Economie, Stenfert Kroese, Houten, The Netherlands.

Vensim (1997a); Modelling Guide, Vensim 3.0, Ventana systems, USA.Vensim (1997b); Tutorial, Vensim 3.0, Ventana systems, USA.Vensim (1997c); User’s Guide, Vensim 3.0, Ventana systems, USA.Vringer, K. and Blok, K. (1993); The Direct and Indirect Energy Requirement of

Households in The Netherlands, Department of Science, Technology and Society,Utrecht University

VROM (1998); Nationaal Milieubeleidsplan 3, Ministry of Housing, Spatial Planning andthe Environment, The Hague, The Netherlands (in Dutch).

VWEW-Verlag (1992); Die Elektrizitätswirtschaft in der Bundesrepublik Deutschland1990, Verlags- Und Wirtschaftsgesellschaft der Elektrizitätswerke, Frunkfurt am Main(in German).

WCED (1987); Our Common Future, World Commission on Environment andDevelopment, Oxford University Press, Oxford/New York.

WEC and IIASA (1995); Global Energy Perspectives to 2050 and Beyond, Report 1995,World Energy Council and International Institute for Applies Systems Analyses,Laxenburg, Austria.

Werbos, R. (1990); Econometric Techniques; Theory versus practice, Energy, vol. 15. No.3/4, pp. 213-236.

Wilting, H.C. (1996); An Energy Perspective on Economic Activities, Ph.D.-thesis,University of Groningen, Groningen.

Winston, W.L. (1987); Operations Research, Applications and Algorithms, PWS-KentPublishing Company, Boston, USA.

WRI (1994); World Resources on diskette, World Resouce Institute, New York.Wright, J.C. (1989); Natural Resource Accounting-A technique for improving planning in

New Zealand?, In: [Peet, 1992].Wyckoff, W.A. and Roop, J.M. (1994); The Embodiment of Carbon in Imports of

Manufactured Products, implications for international agreements on greenhouse gasemissions, Energy Policy, March 1994, p. 187-194.

189

Appendix Model Listing and Assumptions of the ECCO-Models and the

DREAM-Models

Introduction

Below, a detailed listing is given for the single region ECCO-modelling approach ofOECD-Europe as discussed in chapter 4 and 7. The main equations are presented inthis section implying that this listing is not complete. Moreover, the changes made inthe multi regional model and the DREAM-modelling approach are presented here.Both the single and multi-regional models are developed with the aid of a systemdynamic software packet called Vensim [1997a-c]. Vensim allows the use ofsubscript which reduces the number of equations drastically as similar equations, forinstance in the case of the various production sectors, can all be presented by oneequation. Not only are the number of equations reduced by the use of subscript butit also makes the model more accessible as flows and influences are simulated in amore orderly and similar manner.

In the models, subscripts rand s are introduced todenote the differentp r o d u c t i o n s e c t o r sconsidered: agriculture,industry, energy, transportand services. Moreover,subscripts f and g aredefined for the various fuel

types: coal, crude oil, natural gas, biomass, nuclear, hydro, and solar energy. Thelatter also includes wind geothermal energy, and tidal and wave energy. In addition,subscript a is introduced for dealing with the double sets of accounts (seechapter2.4). Throughout the model, ac1 denotes the first account which holds the realenergy value of a flow or stock. This means that energy savings and changing ERE-values are taken into account here. In addition, ac2 refers to the second accountwhich involves the utility value of the stocks and flows (e.g. number of cars orfactories). The introduction of a double sets of accounts can be compared to theconcepts of current constant dollars. Subscript b and c indicate the 6 regions in caseof the multi-regional modelling approaches. Finally, subscript res indicates theresource base class; proven recoverable reserves or addition resources.

In the listings of equations presented for each subsystem, subscripts are putbetween brackets at the end of a variable name. Moreover, subscripts that aresummed are marked by an exclamation mark (e.g. SUM(export[s!])).

Subscripts used below

r, s : agr, ind, ene, tra, serf, g : coal, oil, gas, bio, nucl, hydro, solara : ac1, ac2b,c : reg1, reg2, reg3, reg4, reg5, reg6res : prr, add

190

Chapter 4

Production System

output

out[s] = SUM( pedprod[s,f!]) + res[s] + imp[s]res[s] = SUM(int[r!,s])int[r,s] = ICR[r,s] * Cs[s]imp[s] = MCR[s] * Cs[s]

investments and capital stock

Cs[s,a] = INTEG( rcf[s,a] - rdc[s,a], CS85[s])rdc[s,a] = Cs[s,a] / LT[s,a]rcf[ind,a] = left[a] * fnc[c]/ CTRBRCF[ind]fnc[a] = CTRBCF[ind]/ (CTRBRCF[ind] + rgcf[a])out[s,a] = SUM( pedprod[s,f!,a]) + res[s,a] + imp[s,a]left[a] = out[ind,a] + rdcdel[ind,a] - rcfos[a] - tintout[ind,a] -

export[ind,a] + CSTtintout[r,a] = SUM(int[r,s!,a])export[s,a] = ECR[s,c] * Cs[s]

Energy System

energy demand of production sectors

pedprod[s,f,a] = eedprod[s,a] * perel[f,a] + fuelmixtherm[s,f] * tedprod[s,a]* ere[f,a]

eedprod[s,ac1] = EEIPROD[s] * EFFelecprod[s] * Cs[s,ac2] eedprod[s,ac2] = EEIPROD[s] * Cs[s,ac2]tedprod[s,ac1] = TEIPROD[s] * EFFthermprod[s] * Cs[s,ac2]tedprod[s,ac2] = TEIPROD[s] * Cs[s,ac2]

energy demand of households

pedhh[f,a] = eedhh[a] * perel[f,a] + tedhh[f,a] * ere[f,a] + fuelcars[f,a]* ere[f,a]

tedhh[f,ac1] = TEIHH[f] * EFFthermhh * Csdwell[ac2]eedhh[ac1] = EEIHH * EFFelechh * Csdwell[ac2]

191

thermal energy supply

eremining[f,a] = EREMINING85[f] *EFFlbd[f]* EFFresdepl[f] EFFlbd = relcumindig[f] ^ (log(1-LBDCOEF[f],2))relcumindig[f] = IF THEN ELSE (INDIG85[f]=0, 1, cumindig[f]/INDIG85[f])cumindig[f] = INTEG(indig[f,ac1],INDIG85[f])effresdepl[f] = EFFRESDEPL(tresdeplf[f]) tresdeplf[f] = totresources[f] / TOTRESOURCES85[f]ereimp[f,ac1] = EREIMP85[f] * EFFereimp[f]

ereconvdist[f,a] = SUM(eretherm[g!,f,a]) + ereelec[f,a]* perel[f,a] +ererdcconvdist[f,a] + EREFF[f]

eretherm[g,f,a] = ERETHERM85[g,f] * EFFeretherm[f]ereelec[f,ac1] = EREELEC85[f] * EFFereelec[f] ererdcconvdist[f,a] = IF THEN ELSE(ped[f,a] pfexport[f,a]=0, 0,rdcconvdist[f,a] /

enerout[f,a])Csconvdist[f,a] = INTEG(rcfconvdist[f,a] + rdcconvdist[f,a],

CSCONVDIST85[f])rdcconvdist[f,a] = Csconvdist[f,a] / LTCONVDIST[f]rcfconvdist[f,a] = MAX(0, dcsconvdist[f,a] - Csconvdist[f,a]) +

rdcconvdist[f,a]dcsconvdist[f,a] = enerout[f,a] * RCSCONVDISTOUT[f]enerout[f,a] = (ped[f,a] + pfexport[f,a]) / (ere[f,a] + ERENFF[f])

electricity supply

perel[f,a] = (fuelmixelec[f] * ere[f,a] * syselec[a]) /effpp[f,a] +ererdcpp[f,a]

syselec[ac1] = SYSELEC85 * EFFsyseleceffpp[f,ac1] = EFFPP85[f] * EFFimprpp[f]ererdcpp[f,a] = IF THEN ELSE(outpp[f]=0, 0, rdccspp[f,a] / outpp[f])

Cpp[f] = INTEG(rcfpp[f]-rdcpp[f],CPP85[f])rdcpp[f] = Cpp[f]/ LTPP[f]rcfpp[f] = MAX(0,(reqoutpp[f,ac1] - outpp[f]) / (8760 * 3.6 * LF[f]) *

1e+006) + rdcpp[f]outpp[f] = Cpp[f] * 8760 * 3.6 * LF[f] / 1e+006reqoutpp[f,a] =totgen[a] * fuelmixelec[f]totgen[a] = totdomeed[a] * syselec[a]totdomeed[a] = SUM(eedprod[s!,a]) + eedhh[a]

Resource Base

Resources[res,ff,c] = INTEG(-useres[res,ff,c],RESOURCES85[res,ff,c])

192

Resources[res,nf,c] = INTEG(0,RESOURCES85[res,nf,c])useres[prr,ff,c] = MAX(0,MIN(indig[ff,c,ac1],Resources[prr,ff,c]))useres[add,ff,c] = MAX(0,MIN(indig[ff,c,ac1]-useres[prr,ff,c]

,Resources[add,ff,c]))useres[prr,nf,c] = indig[nf,c,ac1]resdeplf[res,ff,c] = DELAY FIXED ( ffdeplf[res,ff,c],0,1)resdeplf[res,nf,c] = 1ffdeplf[res,ff,c] = IF THEN ELSE ( RESOURCES85[res,ff,c] < 0.001, 0,

Resources[res,ff,c] / RESOURCES85[res,ff,c])tresdeplf[ff,c] = totresources[ff,c] / TOTRESOURCES85[ff,c]tresdeplf[nf,c] = 1totresources[f,c] = SUM( resources[res!,f,c]) Carbon Dioxide Emissions

CO2[c,a] = SUM(ped[f!,c,a] * CO2EMF[f!,c])

Consumption System

population

pop[c] = pop85[c] * GRPOP[c]^(Time-1985)popf[c] = pop[c]/pop85[c]

material standard of living

msolf = msol/msol85msol = gmsol / popmsol85 = GMSOL85/pop85gmsol = tcons[ac2] - govcons[ac2] + totpedhh[ac2] + rdccars[ac2]

consumption of goods and services

non-industrial sectors:incons[ind,ac2] = SMOOTHI(rcf[ind,ac2] * rgcf[ac2],3,INCONS85[s])rgcf[ac2] = rconsrcfind[ac2] * rgech * popfrconsrcfind[ac2] = RCONSRCFIND85 * cofarcf[ind] / cofaout[ind] rgech = ((msolf - 1 ) * EEE) + 1tcons[a] = incons[ind,a] / ctrbfc[ind,a]

non-industrial sectors:incons[s,a] = ctrbfc[s,a] * tcons[a] ctrbfc[s,ac1] = CTRBFCCST[s] * cofacons[s]

193

dwellings

Csdwell[a] = INTEG(rcfdwell[a] - rdcdwell[a],CSDWELL85)rdcdwell[a] = Csdwell[a] / LTDWELLrcfdwell[ac2] = DELAY1I( Dcsdwell[ac2] - Csdwell[ac2] + rdcdwell[ac2], 1,

RCFDWELL85)Dcsdwell = CSDWELL85 *popf * msolf

private transportation

Cars = INTEG(rfcars - rdcars, CARS85)rdcars = Cars /LTCARrfcars = MAX(0, Carsreq - Cars + rdcars)Carsreq = pop * CARSPP85 * GRWCARPPCscars[a] = Cars * gercar[a]/1000rdccars[a] = Cscars[a] / LTCARgercar[ac1] = GERCAR85 * EFFgercarfuelcars[f,a] = pkmpp * pop * fuelreqkm[f,a] /1000fuelreqkm[f,ac1] = FUELREQKM85[f] * EFFfuelcar[f]pkmpp = PKMPP85 * msolf

Balance of Trade System

impexpbal[a] = totimports[a]-totexports[a]totimports[a] = SUM(imp[s!,a]) +totpfimport[a]totexports[a] = SUM(export[s!,a]) + totpfexport[a]totpfimport[a] = SUM(pfimport[f!,a]) + refoilimport[a] totpfexport[a] = SUM(pfexport[f!,a]) + refoilexport[a] pfimport[nof,a] = mprimfuel[nof] * ped[nof,a] pfimport[oil,a] = mprimfuel[oil] * refoilprod[a]refoilimport[a] = MREFOIL * ped[oil,a]pfexport[nof,a] = EPRIMFUEL[nof] * tpes[nof,a]pfexport[oil,a] = EPRIMFUEL[oil] * tpes[oil,a]refoilexport[a] = EREFOIL * tros[a]

Multi-regional model

out[s,c] = SUM( pedprod[s,f!,c]) + res[s,c] + imp[s,c]res[s,c] = SUM(int[r!,s,b!,c])left[c] = out[ind,c] - rcfos[c] -fcoc[c] -tintout[ind,c] - export[ind,c]int[r,s,b,c] = ICR[r,s,b,c] * Cs[s,c]rcfos[c] = SUM (CTRBRCF[ind,c,b!] * trcf[b!]) - CTRBCF[ind,c,c] *

rcf[ind,c]fcoc[c] = SUM (CTRBFC[ind,c,b!] * tcons[b!]) - CTRBFC[ind,c,c] *

tcons[c]

194

trcf[c] = SUM(rcf[r!,c])tintout[r,b] = SUM(int[r,s!,b,c!]

non-oil primary energy sources (nof):indig[nof,c] = SUM(primfueltrade[nof,c,b!] * ped[nof,b!])/

(1-eprimfuel[nof,c])pfimport[nof,c] = (1 - primfueltrade[nof,c,c]) * ped[nof,c]pfexport[nof,c] = sum(primfueltrade[nof,c,b!] * ped[nof,b!]) -

primfueltrade[nof,c,c] * ped[nof,c]

oil:indig[oil,c] = sum(primfueltrade[oil,c,b!] * refoilprod[b!]) /

(1-eprimfuel[oil,c])refoilprod[c] = sum(refoiltrade[c,b!] * ped[oil,b!])/(1-erefoil[c])pfimport[oil,c] = (1 - primfueltrade[oil,c,c]) * refoilprod[c]refoilimport[c] = (1 - refoiltrade[c,c]) * ped[oil,c] + eprimfuel[nof,c] *

tpes[nof,c]pfexport[oil,c] = sum(primfueltrade[oil,c,b!] * refoilprod[b!]) - refoilprod[c]

* primfueltrade[oil,c,c] + eprimfuel[oil,c] * tpes[oil,c]refoilexport[c] = SUM(refoiltrade[c,b!] * ped[oil,b!]) - refoiltrade[c,c] *

ped[oil,c] + erefoil[c]tros[c] = refoilprod[c] + refoilimport[c]

electricity:totdemgen[c] = SUM (totdomeed[b!] * syselec[b!] * electrade[c,b!]) +

eelec[c]totdomeed[c] = SUM (eedprod[s!,c]) + eedhh[c]elecexport[c] = SUM (totdomeed[b!]* syselec[b!] * electrade[c,b!]) -

totdomeed[c] * syselec[c] * electrade[c,c]elecimport[c] = (SUM (electrade[b!,c]) + melec[c] - electrade[c,c]) *

(sum(eedprod[s!,c]) + eedhh[c])

Data and Assumptions

Above, the model listing is presented for the single region and the multi-regionalECCO-model for OECD-Europe. Sources of data required to develop both modelsare presented below. Moreover, the assumptions, which were made to deal with lackof data, are listed here. Both data sources and assumption are described at the levelof the sub-systems that was also used to describe the structure of the single regionmodel.

195

Production System

The capital stock of the production sector is computed by multiplying the capitalstock in 1985 in terms of money by the average energy intensity of capitaldepreciation The average intensity of capital deprecation is determined in chapter 3.The capital stock in terms of money is computed by multiplying the consumption offixed capital of a sector (in US$) [OECD, 1993a] with the lifetime of that sector. Therate of capital depreciation of the non-energy sectors is equal to the total capital stockdivided by the lifetime of that capital stock (LT). For each sector of a region, thelifetime of the capital stock is assumed equal to that of the economically dominantcountry within the region country or equal to the average value of a number ofdominant countries within the region except for region 2 where the lifetime of capitalis assumed equal to the average value for a number of European countries [OECD,1995a].

Data for determining the ratio between intermediate inputs and imports andbetween the capital stock (denoted ICR and MCR) are derived from the (regional)input-output table of OECD-Europe (in terms of energy) that is described in chapter3.

Energy System

Data required for determining the ratio between the thermal energy and the electricitydemand and between the capital stock (TEI and EEI) are also derived from chapter3. Additional data on the electricity and thermal energy demand are derived from[OECD, 1991a-b]

Data for computing ratios between the electricity and thermal energy demand andbetween the capital stock of dwellings are mainly derived from [OECD, 1991a]

For all non-oil fuel types, the import and export totals are derived from [OECD,1991a]. For natural gas this distribution is mainly derived from [OECD, 1987] onlyfor East Germany data are derived from [OECD, 1989; UN, 1986 and UN, 1988b].The total primary energy demand for bio fuels and nuclear, hydro and solar energyis assumed to be produced domestically. For coal, the distribution is derived from[UN, 1988b] where various coal types are distinguished. As the model only considersone aggregated coal type, the various coal types presented by the UN in tonnesproduct are converted into GJ by using the conversion data of [OECD 1991b]specified by the country of origin. In a number of cases, the import statistics appearnot to be consistent with the export statistics. These inconsistencies apply to dataderived from the OECD [1987] as well as from the UN [1988a and 1988b]. This ismost probably the result of transit trade. The production of a fuel of a certain regionmay be estimated too high when the import statistics are used as a reference as thatcountry may only transit (a part of) the fuel. Therefore, it is preferred here to estimate

196

the trade of fuel within OECD-Europe by using the export data. The statisticaldifferences introduced are assumed to be imports from outside OECD-Europe. Themost striking example of statistical differences is the imports of natural gas by theUK. Statistical differences, which involve imports of natural gas by the UK fromNorway, are the result of whether or not the imports/exports from/to the islandTeeside are included in the statistics [OECD, 1991c]. These imports (by the UK) areassumed to have their origin outside OECD-Europe as the country of origin is hardto determine. For Austria, the distribution figures of 1985 of the trade in coal areassumed equal to that of 1984 as no data are available for 1985. For Germany, nospecification is presented of the distribution figures of exports for natural gas to thecountry of origin and therefore these distribution figures are estimated by using theimport data of the other countries within OECD-Europe. The import and exporttotals of crude oil are derived from [OECD, 1991a]. Distribution data are derivedfrom [OECD, 1987] except for East Germany of which all data are derived from[OECD, 1989; UN, 1988b]. For refined oil products, the import totals and exporttotals are derived from [OECD, 1991a] and the distribution data from [OECD, 1987]except for East Germany of which all data are derived from [UN, 1988].

Energy required for mining consists of two components that is the direct andindirect energy use. Data on both components are not available at a regional level andare, therefore, assumed equal for all regions. For 1985, the direct energy requirementfor extracting one GJ (eremining) is derived from [Fritsche et al., 1994]. The initialcapital stock of the mining sector (CSMINING85) is also derived from [ibid]. Dataon the efficiency improvements as result of learning- by-doing are derived from [deVries and Janssen, 1996]. Curves used to compute the increasing energy requirementdue to depleting reserves are also derived from [ibid].

For each region, the energy requirement for transportation (ereimp) is computedby taking the weighed average of the energy requirement for transporting energy fromthe region of origin. The data, which is used in computing the average value, arederived from [Fritsche et al., 1994] where the energy requirement is presented fromimporting fossil energy (the energy costs of mining included) from a number ofregions to Germany. So, it is assumed here that the energy costs of transportingenergy from a number of regions to Germany is a good indication of the energy costsfor transporting energy from those regions to any region within in OECD-Europe.

The direct energy requirements for converting primary fossil energy intosecondary energy sources and of distributing the secondary energy sources to the end-user are determined by using the data of [OECD, 1991a]. For oil, the modeldistinguishes the imports of crude oil and of refined oil products. The energy costsof importing refined oil products is determined by using data of [Fritsche et al.,1994] where the energy requirement is presented from importing fossil energy (theenergy costs of mining included) from a number of regions to Germany. For theregions outside OECD-Europe, the energy requirement for the conversion of crudeoil into refined oil products is assumed equal to the average value of OECD-Europe

197

[1991a]. The capital stock requirement per GJ secondary output (RCSCONVDIST)is assumed equal for each region as no region-specific data have come available. Thedata are derived from [Fritsche et al., 1994] and are based on Germany.

The total rate of capacity depreciation of power plants (rdcpp) is equal to thetotal capacity divided by the average value lifetime (LT) of the power plants. Foreach region, the average life time of a power plant is assumed to be 25 years exceptfor hydro power plants where the lifetime is assumed to be 50 years [Benders,1996].The contribution of a region in the electricity supply of any region in OECD-Europe (ELECTRADE) is set exogenously and the data are for computing the initialvalues are derived from [OECD, 1991a and 1992]. Due to inconsistencies in thetrade statistics the distribution data are based on export statistics. Data for EastGermany are not included in these statistics and therefore the volume of trade inelectricity between East Germany and the rest of OECD-Europe is assumed to bezero.

The load factor depends on the capacity and the total output. The initial valuesof the total capacities (CPP85) are derived from [OECD, 1992; Atomwirtschaft,1986 & 1987]. Data includes multi-fired power plants, that is plants that use varioustypes of fossil fuels. The capacities of these multi-fired plants must be specified forone fuel type as the model does not consider multi-fired plants. Data of the totalcapacity specified per fuel type are also required in order to compute the load factor.For each region, the load factor and the initial capacity of the power plants fossil fuelfired are estimated by using the model PowerPlan, which is an interactive model tosimulate (future) electricity planning of a country/region [Benders, 1996]. PowerPlancomputes the electricity supply based on the annual simulated peak demand and aload duration curve and it uses a certain merit order of putting power plants inoperation (among others, based on operational costs) [Benders, 1996]. For eachregion, the load factor (which is an output of the model) and the capacities (which isan input of the model) are estimated by adjusting the distribution of the fuel mix overthe various capacities of the multi-fired plants (i.e. with trial and error). Data on thepeak demand and LDC curves are determined by regarding the dominant countriesof the various regions except for region 2 and 5 these values were estimated by usingthe average values of the OECD [VWEW-Verlag, 1992; Ente Nazionale per l’energiaElettrica, 1987; EDF, 1992; Central Electricity Generating Board, 1992; Benders,1996].

Initial data on the fuel mix of generating electricity (FUELMIXELEC), theelectricity system losses (SYSELEC) and efficiency of power planst (EFFPP85)are derived from [OECD, 1991a].

Two classes of resources are distinguished in the model (i.e. proven recoverablereserves and additional reserves). For each region, the model computes the (depletionof) proven recoverable reserves and the additional resources of the fossil fuels. Theinitial values of the proven recoverable reserves (PRORECOVRES85) arecomputed by adding the cumulative production between 1985 and 1990 of the fossil

198

fuels [WRI, 1994] to the proven recoverable reserves in 1990 [BP, 1994]. Data onadditional resources are derived from [BP, 1994; Nakiƒenoviƒ et al.,1997]

Consumption System

The population data of 1985 are derived from [OECD, 1991a and 1993c]. Thegrowth rates are computed by taking the average value over 1980 - 1989 [OECD,1991a and 1993c].

The energy to expenditure elasticity (EEE) was introduced to compute the relativegrowth of the energy value of consumption by households as result of the growth inthe material standard of living (msolf). This energy to expenditure elasticity is set at0.63 [Vringer and Blok, 1993] for all regions. For each region the share thegovernment and private non-profit institutions in the final consumption is assumedequal that of the economically dominant country within the region except for region2 where this share is computed by taking the average value of a number of regionswithin OECD-Europe [Eurostat, 1992a-f].

The rate of capital depreciation (rdcdwell) is equal to the total capital stockdivided by the lifetime of the capital (LTDWELL). The average lifetime of dwellingis assumed equal to the economically dominant countries within the region except forregion 2 where the lifetime of capital is assumed equal to the average value for anumber of European countries [OECD, 1995a].

The number of cars scrapped is equal to the total number of cars divided by thelifetime of cars (LTCAR). For each region, the lifetime of cars is assumed equal tothe value of the economically dominant country within the region except for region2 where the lifetime of cars is assumed equal to the value of Norway which is aboutthe average value of Sweden, Norway and Denmark [Van den Broecke, 1988].

The number of cars required is equal to the average number of cars per person(CARSPP85) multiplied with total population (pop). For region 1, 3 and 4, thenumber of cars per person is assumed equal to the value of the economicallydominant country within the region [Davis, 1996]. For region 2, the number of carsper person is assumed equal to the value of Sweden [Davis, 1996] and for 5 regionit is assumed equal to the average value of Spain [OECD, 1995b]. For region 6, thenumber of cars per person is assumed equal to the average value of Italy and Turkey[Davis, 1996]. The capital stock of the private owned cars in terms of energy iscomputed by multiplying the number of cars (Cars) with the average energy valueof a car (the gross energy requirement of a car is denoted by GERCAR). The grossenergy requirement per car is assumed equal to 220 GJ/car (this figure includesconstruction and average repairs) [Henham in [Moll, 1993; pg. 210]].

For region 1, 3 and 4, the personal travel kilometres and the fuel requirement in

199

MJ per pkm (person kilometres) are assumed equal to the average values of theeconomically dominant country within the region [Davis, 1996]. For 2 region thepersonal travel kilometres and the fuel requirement in MJ per pkm are assumed equalto the average values of Sweden [Davis, 1996] and for region 5 to those of Spain[OECD, 1995b]. For region 6, the personal travel kilometres and the fuel requirementin MJ per pkm are assumed equal to the average values of Italy and Turkey [Davis,1996; OECD, 1995b]

Chapter 7

Model changes in DREAM-OECD compared to ECCO-OECD.

Below a listing is presented of the equation that are changed in the single regionDREAM-model for OECD-Europe compared to the single ECCO-model of OECD-Europe.

Investments:

rcf[s,ac2] = DELAY1I(Max(0,Dcsut[s] - Cs[s,ac2] + rdc[s,ac2]),5,RCF85[s])

Dcsut[s] = SMOOTH(dout[s,ac2] /OCR85[s],3)dout[s,a] = del2rcf[s,a] + del2fc[s,a] + tintout[s,a] + export[s,a]OCR85[s] = OUT85[s] / CS85[s]del2rcf[s,a] = ctrbrcf[s,a] * trcf[a]del2fc[s,a] = ctrbfc[s,a] * tcons[a]del2rcf[s,a] = ctrbrcf[s,a] * trcf[a]del2fc[s,a] = ctrbfc[s,a] * tcons[a]tintout[r,a] = SUM(int[r,s!,a])

Consumption:

consut[s] = CTRBCONSUT[s]* totconsuttotconsut = CONSUT85 * popf * (1 + (GRWTHCONSUT)) ^ (Time -

1985)incons[s,ac1] = MAX(0,MIN(dcons[s,ac1], tacons[ac1] )dcons[s,ac2] = ctrbincons[s,ac2] * consut[s]tacons[a] = out[s,a] * (fac[s,a]) - export[s,a] - tintout[s,a] fac[s,a] = (out[s,a] - ctrbrcf[s,a]* trcf[a])/out[s,a],mcons[ac2,s] = ctrbmcons[s,ac2] * consut[s]

200

Imports Exports:

impexpbal[a] = totimports[a]-totexports[a]totimports[a] = SUM(imp[s!,a] + mcons[s!,a] + addimp[s!,a])

+totpfimport[a]addimp[s,a] = MAX(0, dout[s,a] - out[s,a])export[s,ac2] = EXPORTS85[s] * (1 + GRWTHEXP[s])^(Time-1985)totexports[a] = SUM(export[s!,a]) +totpfexport[a]

Changes in NLDREAM compared to NLECCO

Below a listing is presented of the main equation of NLDREAM that differ fromNLECCO.

Investments structure in the Industry

rcf[s,ac1] = DELAY1(MAX(0,dutcs[ind] - Cs[ind,ac2] +rdc[ind,ac2])*cofaout,2)

rcf[s,ac2] = DELAY1(MAX(0,dutcs[ind] - Cs[ind,ac2] + rdc[ind,ac2]),2)utcs[s] = dout[s,ac2]/OCR85[s]dgrutout[s] = dout[s,ac2]/out[s,ac2]

Consumption

cons[ind,ac1] = MAX(0,MIN(dconsind[ac1], tacons[ac1] ))cons[ind,ac2] = cons[ind,ac1] / cofaout[ind] cons[nis,ac1] = cons[nis,ac2] *

cofaout[nis]cons[nis,ac2] = CTRBCONS[nis] * totconsutcons[ind,ac2] = cons[ind,ac1] / cofaout[ind]consaux[a] = cons[ind,a] + mcons[a]dconsind[ac1] = CTRBCONS[ind]* totconsut * cofaout[ind]dconsind[ac2] = CTRBCONS[ind]* totconsutconsphh[hh,ac1] = consut[hh] * cofaconsconsphh[hh,ac2] = consut[hh]totconsut = SUM(CONSUTPHH[hh!]) * consgr(time)^(Time-1985)consut[hh] = CONSUTPHH[hh] * numbhh[hh] /SUM( CONSUTPHH[hh!] *

numbhh[hh!]) * totconsutbalconsut = totconsut - SUM(consut[hh!])tcons[a] = SUM(cons[s!,a]) + mcons[a]mcons[ac1] = (1 - SUM(CTRBCONS[s!])) * totconsut * cofaout[ind] +

admcons[ac1]mcons[ac2] = mcons[ac1] / cofaprod

201

Households

numbphh[age,hh] = (pop[age]+ Netimm* agedistr[age]) * DISTRPOPHH[age,hh]numbhh[hh] = SUM(numbphh[age!,hh]) /POPPERHH[hh]avagehh[hh] = SUM(numbphh[age!,hh] * AVAGEPOP[age!])/

SUM(numbphh[age!,hh])tothh = SUM(numbhh[hh!])perhou = tpop/tothhDISTRPOPHH[age,lhh] = SMOOTH3I(DISTRPOPHHEND[age,lhh],

10,DISTRPOPHH1985[age,lhh]) ~~|DISTRPOPHH[age,mfrph] = 1 - SUM(DISTRPOPHH[age,lhh!])totpphh[age] = SUM(DISTRPOPHH[age,hh!])

Private Transportation

Cars = INTEG(rfcars-rdcars, CARS85)rdcars = Cars /LTCARrfcars = DELAY1(MAX(0, Carsreq - Cars + rdcars),0.8)Carsreq = SUM(Carsreqhh[hh!]) + nonhhcars(time)Carsreqhh[hh] = carpen[hh] * numbhh[hh]Carshh[hh] = INTEG(rfcarshh[hh]-rdcarshh[hh], CARS85hh[hh])rdcarshh[hh] = Carshh[hh] /LTCARrfcarshh[hh] = DELAY1(MAX(0, Carsreqhh[hh] - Carshh[hh] +

rdcarshh[hh]),0.8)KMCAR[hh] = KMCARHH85[hh] * 1000 * GRKMCARHH[hh]^(Time-1985)carkm = SUM(carkmhh[hh!])carkmhh[hh] = KMCAR[hh] * Carshh[hh]fuelcar[a] = SUM(fuelcarhh[hh!,a])~~|fuelcarhh[hh,a] = 2.86 * carkmhh[hh] * efffuelcar[a]/1000efffuelcar[ac1] = (1-Effcar)^(Time-1985) ~~|efffuelcar[ac2] = 1Effcar = structcar + savingscar

Energy Savings

eedprod[s,ac1] = EEIPROD[ind] *eedeff[s]* Cs[s,ac2] - eleccogenprod[s,ac1]tedprod[s,ac1] = TEIPROD[ind] * tedeff[s] * Cs[s,ac2]tedeff[s] = SMOOTH((1-EFFRATETC[s]-STRUCTEFFTC[s])^(Time -

1985),5)eedeff[s] = SMOOTH((1-EFFRATEEC[s]-STRUCTEFFEC[s])^(Time -

1985),5)autotc[ind] = RESAUTOTCind(Time)autotc[agr] = RESAUTOTCagr(Time)autotc[ene] = 1autotc[tra] = 1

202

autotc[ser] = RESAUTOTCser(Time)autoec[ind] = RESAUTOECind(Time)autoec[agr] = RESAUTOECagr(Time)autoec[ene] = 1autoec[tra] = 1autoec[ser] = RESAUTOECser(Time)

xtratc[s] = IF THEN ELSE(tedeff[s]<autotc[s], tedeff[s]/autotc[s], 1)xtraec[s] = IF THEN ELSE(eedeff[s]<autoec[s], eedeff[s]/autoec[s], 1)dcstc[s] = RESXTRATCind(xtratc[s])dcsec[s] = RESXTRAECind(xtraec[s])

pedhh[a] = eedhh[a]* perel[a] + SUM(tedhh[f!,a] * ere[f!,a]) + fuelcar[a] *ere[oil,a] + adgascogenpublic[a] * ere[gas,a]

tedhh[f,a] = (SUM(tedphhs[hh!,a])- hpcogenpublic[a]) * FUELMIXHH[f] +oildom(Time) * FUELMIXOILDOM[f]

eedhhgj[a] = eedhh[a] *3.6/1000~~|tedphhs[hh,ac1] = TEIHH[hh] * tedeffdwell * numbhh[hh] ~~|eedhh[a] = SUM(eedphhs[hh!,a])eedphhs[hh,ac1] = EEIHH[hh] * eedeffdwell * numbhh[hh] ~~|eedphhs[hh,ac2] = EEIHH[hh] * numbhh[hh]tottedhh[a] = SUM(tedhh[f!,a])eedeffdwell = SMOOTH((1-EFFRATEECDWELL(time)

-STRUCTEFFDWELLEC(time))^( Time - 1985),5)tedeffdwell = SMOOTH( (1-EFFRATETCDWELL(time)-

STRUCTEFFDWELLTC(time))^( Time - 1985),5)autotcdwell = 1autoecdwell = RESAUTOECdwell(Time)xtratcdwell = IF THEN ELSE(tedeffdwell< autotcdwell, tedeffdwell

/autotcdwell, 1)xtraecdwell = IF THEN ELSE(eedeffdwell< autoecdwell,

eedeffdwell/autoecdwell, 1)dcstcdwell = RESXTRATCdwell(xtratcdwell)dcsecdwell = RESXTRAECdwell(xtraecdwell)

Assumptions NLDREAM

NLDREAM is mainly based on NLECCO implying that most data assumptionare listed in Noorman [1995]. Three model changes were introduced in subsection7.5.2 regarding the introduction of households, changes in the private transportationand capital requirement for energy savings. The latter is not associated with changesin the data whereas the data required for changes in the private transportation arederived from [Schenk, 1998]. The assumption considering the introduction ofhouseholds are listed below.

203

Households

In the original model, consumption, direct energy use, and private transport arecomputed on the basis of the demand per caput. These demands may be computedmore accurately if one takes into account the number of households divided byhousehold type. Moreover, adopting this approach offers the possibility to use theNLECCO model within the framework of HOMES.

At first, calculating the demand is only applied to determining the consumptionof goods and services and the direct energy use. Five type of households aredistinguished that is one person households, two persons households, three personshouseholds, four persons households, and multi persons households. The number ofhouseholds per household type should, of course, coincide with the population dataof the original NLECCO model. Hence, the population divided by the 4 age groupsshould be distributed over the 5 households types. Determining this distribution is notan easy task as the data required are rather scarce especially historic data such as for1985. The distribution is therefore determined by combining different closely relatedtables, such as number of children per age group of the parents, the number ofpersons in a household divided by the positions in the household, and the age of thehead of the household [CBS, 1986, 1996]. However, a number of assumptions werestill required to calculate the distribution as from these table one cannot determine thedistribution exactly. The distribution of the number of children over the householdtype is assumed not to be depend on the age of the children who are living with oneparent as well as with two parents. The distribution of the total number of childrenis not dependent on the age of the person. Non-married couples and single parents areassumed to have maximally 1 child. The average number of children in house holdswith more than 2 children is set at 3.3. A three persons household is estimated toconsist of 1.3 children on average. While, a four persons household and a multi-persons household are assumed to comprise 2 and 3.3 children on average,respectively. The derived distribution is presented in table 1. At first sight one mightbe surprised that so many children are relatively living in larger households but oneshould note that when 3 children are, for example, living together with their parentsthen they are, of course, taken into account 3 times as the persons of an age groupare distributed over the households types.

205

Summary

The school of ecological economics studies the relationship between the economicsystem and the environment from a broader perspective than neoclassical economicsas it also considers methodologies and approaches from other disciplines. In thissense, energy accounting is introduced to study the relationship between economicactivity and the environment from a physical perspective. Energy accounting startsfrom the notion that the economic system is subject to the fundamental laws ofthermodynamics. In energy accounting, economic activity is expressed in terms of(past)energy use.

Within the framework of energy accounting, the ECCO-modelling approach(Enhancement of Capital Creations Options) is developed to study the potentials ofeconomic development taking into account physical laws rather than taking thestandard economic decision framework as a starting point. The ECCO-modellingapproach can be characterised as a dynamic energy accounting approach in whichresources are quantified through the primary energy required to release them and toproduce any good or service in the economy. Based on the principles of systemdynamics, the ECCO-modelling approach describes the structure and evolution of theeconomy in relation to the environment.

In essence, the approach is supply-driven implying that the level of economicactivity is determined by the production sectors. Feedback loops between investmentsand industrial output, therefore, form the main elements of the modelling approach.This supply-driven approach resulted from the objective to study the potentials of aneconomy to grow under certain physical conditions. Although the physical aspectsof the economic system described in the ECCO-modelling approach cannot fullyaccount for standard economic decision rules, the approach yields relevant insightinto the functioning of economic systems in an environmental content. Unlikeneoclassical economic models, the ECCO-approach is not designed to focus onunderstanding and predicting human behaviour or short-term optimal allocations butit is developed to analyse the long-term physical limits on economic activity.Therefore, the ECCO-approach should be regarded as an additional approach nextto economic/econometric models instead of a substitute.

The basic idea behind ECCO is that a sector requires capital stock to produce anyoutput and, therefore, the output of a sector is assumed to be proportional to thecapital stock of that sector. Moreover, the approach is consistent with the concept ofIO-analyses implying that the total output of a sector equals the total input of thatsector comprising: the total (direct) primary fuel use, the energy content ofintermediate deliveries and imports. All these inputs are assumed to be proportionalto the capital stock, that is if efficiency improvements are not included. RecentECCO-models distinguish two values for the output. The first one measures the realenergy content of output or the real energy costs of the production. The second valuecalculates the utility level of the output (e.g. number of cars produced or amount of

206

crops harvested). These two values can best be compared with the concept of‘current’ and ‘constant’ dollars which is commonly used in the field of economics.The output of a sector is distributed over the items: domestic deliveries, exports, finalconsumption and investments. Naturally, the distribution differs per sector. The(aggregated) industrial sector is of special interest as the room for investments in thissector is the balancing term of the model that is the allocation of industrial outputdetermines the room for investment. The determination of the room for investmentsforms one of the major equations of the model as it indirectly influences the growthof the entire production system.

This thesis focuses on two methodological aspects of the ECCO-modellingapproaches. The first comprises regionalisation and the second involves thedevelopment and application of a demand-driven version of the ECCO-modellingapproach. In the latter approach, energy costs associated with changing consumptionpatterns are assumed to be driving forces underlying changes in the productionsystem and thus in the system’s energy requirement. This demand-driven model ismainly based on the principles of the ECCO-approach in which the latter is used asa tool to assess economic development potentials. Given the focus of this thesis, thescenarios presented are also developed from a methodological point of view, that isthese scenarios do not indicate expectations about the potential growth rates of theeconomies involved and the related energy use but the scenarios are developed toshow the impact of methodological changes.

Regionalisation

Imports and exports are becoming increasingly important for national economiesin the world and in particular in the case of Europe as a result of the introduction ofthe euro and the liberalisation of the energy markets in the European Union. A propertreatment of imports almost unavoidably necessitates the introduction of multi-regional models dealing with regional differences.

A multi-regional energy-based model of OECD-Europe was developed to studythe impact of regional differences in OECD-Europe. The consequences of a fulltreatment of regionalisation are studied by comparing the results for two ECCO-modelling approaches. The first involves an ECCO-model in which OECD-Europeis regarded as one large region. The second consists of a modelling approachinvolving a multi-regional ECCO-model in which OECD-Europe is divided into 6subregions. Regional differences are illustrated by varying the development energyintensities and the relative growth rates of the sectors agriculture, transport andservices within the distinguished region according to three alternatives. Thesealternatives are based on the regional development paths observed for the 1985-1995period. Scenario results show that regional differences substantially influence the

207

outcomes of the aggregate OECD-Europe region. These results support the idea thatthe multi-regional approach results in more detailed outcomes in the ECCO-methodology.

In addition to the regional differences in the development path, there also appearto be large national differences in the embodied energy intensities (EEI) of sectorswithin OECD-Europe. The usual assumption within input-output (IO) energyanalysis that the embodied energy intensities of imports can be set equal to the EEIof corresponding domestic products may, therefore, introduce errors in thecalculations of the EEI for a country. Two approaches are introduced to avoid theseerrors. Intuitively, it is felt that avoiding these errors results in more accurateassessments of the EEI.

The first part of this thesis demonstrates that the outcomes of a multi-regionalECCO-model differ substantially from that of a single region ECCO-model. Ofcourse, it is difficult to conclude which method results in more accurate outcomeswhile two methods are compared that both use somewhat unrealistic scenariovariables as starting point. However, the preponderance of evidence indicates thattaking into account the differences among subregions of an aggregate region resultsin a better description of the dynamics than by only regarding the average values ofthe aggregate region.

Some of the regional differences shown may be the consequence of the processthat a number of regions are catching up to other regions implying that in due timethe regional differences will disappear. Clearly, a homogenous economy throughoutOECD-Europe makes a multi-regional approach less necessary to study regionaldifferences although it still facilitates the study of changing energy stocks and flowswithin OECD-Europe. Another consideration of developing a multi-regional modelis that the economy will most probably not become completely homogenous as forinstance there are still likely to be substantial differences in the electricity supply.

Demand-Driven Approach.

In principle, all ECCO-models are supply-driven implying that one assumes thatthe consumption level is determined by producers. It is increasingly recognised thatconsumers or households play a key role in driving the economy. Hence, it is arguedto assess the energy costs associated with changing consumption activities underlyingthe changes in production systems and the system’s energy costs. In the second partof this thesis, the supply-driven ECCO-modelling approach was changed into thedemand-driven DREAM-approach (Dynamic Resource and Economy AccountingModel). That is, a dynamic energy accounting approach is developed based on thenotion that most production finally ends up in consumer goods and services. In thisway, the energy costs and the related environmental stress can be determined fordifferent scenario assumptions about future consumer activities.

208

The second part demonstrates that the DREAM-modelling approach is anappropriate tool to study the energy costs of changing consumption patterns in asystem dynamic way. The demonstration is based on two case studies:

A case study of OECD Europe showed that the ECCO-model and the DREAM-model behave differently as a result of differences in the key influences. Economicactivity showed exponential growth patterns in the scenarios developed with the aidof the DREAM-modelling approach whereas in ECCO-models the growth rates ofthe economy tended to decrease over time. That is, as a consequence of the absenceof a negative feedback loop the DREAM-approach leads to increasing growthpatterns. In principle, economic growth is not decelerated as long as consumption andexport levels remain growing. However, sustainability aspects related to theseeconomic development paths can also be shown by the DREAM-modelling approach.For instance, CO2-emissions due to fossil energy use and the import-export balanceindicating the dependence on foreign resources are used to study the impact ofeconomic growth levels.

A case study of The Netherlands shows that relatively high consumption growthrates are still associated with an increasing total fossil energy demand despite theassumptions in a number of scenarios that energy saving potentials increasesubstantially and the public electricity supply is totally based on renewables. Theseresults indicate that a decreasing energy demand can only be realised by redirectingconsumer activities which in turn can be realised by decreasing consumption growthrates or by changing the package of consumer goods and services. In the currentversion of NLDREAM, the energy costs related to changes in consumption packagescannot be assessed at a detailed level as the production sector involves too aggregateda level. These kinds of studies should be made possible in future versions ofNLDREAM as the effect of changing consumption patterns on the overall energy useis a major topic in meeting the targets of sustainable development.

Final Remarks

Returning to the perspectives of the energy accounting approach, both the ECCO-modelling approach and the DREAM-approach study the relationship between theeconomic system and the environmental system from a physical perspective. Bothapproaches address the energy use associated with economic activity.

The ECCO-approach emphasises that energy is a key factor in economicproduction processes since it constrains economic development by the availabilityof energy resources or energy embedded in capital.

In essence, the DREAM-modelling approach was developed to addressdynamically the long-term consequences of the energy use as result of technologyimprovements, changing consumption patterns, energy savings and demographicchanges. The approach, thus, determines the environmental impacts correspondingto specific economic development paths. In addition, the approach can be used to

209

investigate which changes or measures are required to direct consumption patternstowards sustainability.

Obviously, the two modelling approaches do not result in ‘scientifically solving’the problems concerning economic activities and the stress on the environment.However by using both approaches, the insights gained from studying the (physical)interactions between the economic system and the environmental system makes theincorporation of physical approaches into economic theory worthwhile. Bothapproaches should be used together with conventional economic approaches in orderto investigate the relationships between the economic system and the environmentalsystem in an integrated way.

211

Samenvatting

Binnen ‘Ecological economics’ bestudeert men de relatie tussen de economie en hetmilieu vanuit een breder perspectief dan de gangbare neoklassieke economieaangezien ‘ecological economics’ ook gebruik maakt van benaderingen en methodesuit andere disciplines. Binnen dit relatief nieuwe vakgebied is de zogenaamde ‘energyaccounting’ geïntroduceerd die de relatie tussen economie en milieu beschouwt vanuiteen fysiek perspectief. Hiermee wordt benadrukt dat de economie gebonden is aan dewetten van de thermodynamica. Dat wil zeggen dat in ‘energy accounting’economische activiteiten worden uitgedrukt in termen van energie.

Binnen de ‘energy accounting’ is een methodologie ontwikkeld die ECCO wordtgenoemd (waar ECCO staat voor (Enhancement of Capital Creations Options).ECCO gebruikt men om de potentiële economische ontwikkelingen te bestuderenvanuit een fysisch kader in plaats van de standaard economische beslissingsregels.De ECCO-methodologie kan worden gekarakteriseerd als een dynamische 'energy-accountingsmethode' waarin de het gebruik van fossiele brandstoffen wordtgekwantificeerd door een koppeling te maken tussen de vraag naar deze brandstoffenen de productie van goederen en diensten in de economie. In tegenstelling totneoklassieke modellen is ECCO niet ontwikkeld om het gedrag te voorspellen hoeproducenten en consumenten omgaan met korte termijnschaarste, maar om de fysiekelange termijnlimieten van economische activiteiten te bestuderen. Daarom moetenECCO-modellen worden beschouwd als een aanvullende benadering op de gangbareeconomisch/econometrische modellen en niet als substituut.

De methodologie gaat uit van een aanbodsgestuurde economie. Dit houdt in datmen ervan uitgaat dat de economische groei voornamelijk wordt bepaald door deproducenten. Feedback loops tussen investeringen en industriële productie vormen deessentiële elementen van het model. De basisgedachte achter de ECCO-methodologieis dat een sector kapitaalgoederen nodig heeft om te kunnen produceren, vandaar datde output van een sector proportioneel is verondersteld aan dekapitaalgoederenvooraad. Verder is de ECCO-methodologie gebaseerd op de input-output analyse (IO) in de zin dat de totale output van een sector per definitie gelijkis aan de totale inputs van een sector. Onder de inputs van een sector wordenverstaan: binnenlandse intermediaire leveringen (leveringen van sector naar sector),primaire energiegebruik afschrijving van de kapitaalgoederenvoorraad en importen.Al deze inputs zijn dus proportioneel verondersteld aan de kapitaal goederenvoorraaddat is wanneer efficiëntieverbeteringen niet meegenomen zijn. Recente ECCO-modellen onderscheiden namelijk twee waarden van output. De eerste behelst dewerkelijke energiewaarde c.q. de energiekosten van de output: dat wil zeggen energieefficiëntieverbeteringen zijn hierin wel meegenomen. De tweede drukt de output uitin termen van het utiliteitsniveau (aantal geproduceerde auto's enz.). Deze tweewaarden kunnen het beste vergeleken worden met het concept nominale en reële

212

guldens. De output van een sector wordt verdeeld over de posten: investeringen, finalevraag (consumptie) van de overheid en huishoudens en exporten. Natuurlijk verschiltde verdeling per sector. De geaggregeerde sector industrie speelt een essentiële rol inECCO aangezien de investeringen in deze sector de groei van de alleproductiesectoren stuurt en daarmee ook de economie in haar geheel. Verder zijninvesteringen in de industriële sector de sluitpost van het model, dat wil zeggen dehoeveelheid output die niet gealloceerd is naar de overige posten is beschikbaar voorinvesteringen in de industriële sector.

Deze dissertatie richt zich op twee methodologische aspecten van debovengenoemde ECCO-benadering. De eerste behelst regionalisering en de tweedede ontwikkeling van een vraaggestuurde variant van de ECCO-methodologie waarinveranderende consumptiepatronen beschouwd worden als de drijvende kracht in deeconomie. In deze nieuwe benadering worden veranderingen in het totaleenergiegebruik gestuurd door veranderingen in de consumptiepatronen en de daaraangekoppelde veranderingen in de productiestructuur. Gegeven het methodologischekarakter van deze dissertatie geven de gepresenteerde scenario's geen voorspellingenvoor de toekomst maar verschaffen ze inzicht in uitkomsten van de methodologischeveranderingen.

Regionalisering

Importen en exporten vormen een steeds belangrijker wordend onderdeel vannationale economieën en in het bijzonder voor Europa als gevolg van de invoeringvan de euro en de liberalisering van de energiemarkten. Een goede beschrijving vandeze importen en exporten vraagt bijna onontkoombaar om multiregionale modellen.Dit geldt ook voor ECCO-modellen. Vandaar dat er een multiregionaal model isontwikkeld voor OECD-Europa. Het eerste deel van deze dissertatie bestudeert deconsequentie van het ontwikkelen van een multiregionaal ECCO-model voor OECD-Europa. In deze studie worden de uitkomsten van twee modellen met elkaarvergeleken. In het eerste model wordt OECD-Europa opgedeeld in 6 sub-regio’s enin het tweede model wordt OECD-Europa beschouwd als een grote regio. Het bestaanvan regionale verschillen wordt aangetoond door het variëren van de ontwikkelingenin energie-intensiteiten (energiegebruik per eenheid output in guldens) en de relatievegroei van de niet-industriële sectoren volgens drie alternatieven. Deze alternatievenzijn gebaseerd op de waargenomen ontwikkelingen tussen 1985 en 1995. Deuitkomsten tonen aan dat er substantiële verschillen bestaan tussen regio's. Verderblijkt dat de uitkomsten van het multiregionale model verschillen metovereenkomstige resultaten van het geaggregeerde model.

Verder blijkt ook dat er ook grote verschillen bestaan in energie-intensiteitentussen overeenkomstige sectoren van landen. De veelal gebruikte aanname dat deenergie-intensiteiten van importen gelijk zijn aan die van overeenkomstige

213

binnenlandse sectoren kan dus leiden tot foutieve berekeningen van de energie-intensiteiten van sectoren (waarbinnen ook leveringen van importen meegenomenzijn). Vandaar dat er twee berekeningsmethodes worden geïntroduceerd in dezedissertatie die deze fouten voorkomen. Intuïtief resulteren deze nieuwe methodes inbetere berekeningen van de energie intensiteiten.

In het eerste gedeelte van deze dissertatie wordt aangetoond dat de uitkomsten vaneen multiregionaal ECCO-model voor OECD-Europa substantieel afwijken van dievan een geaggregeerd model. Hoewel het moeilijk is om te concluderen welk modelde accuraatste uitkomsten genereert wanneer de resultaten van twee modellen metelkaar vergeleken worden, wijst de aard van de uitkomsten er op dat de inachtnemingvan de regionale verschillen tot een betere beschrijving leidt van de dynamiek binneneen regio dan door alleen rekening te houden met de gemiddelde waarden binnen deregio.

Enkele van de regionale verschillen kunnen het gevolg zijn van het feit datbepaalde regio’s bezig zijn met een ‘economische inhaalslag’ hetgeen er toe kanleiden dat de regionale verschillen in de toekomst zullen verdwijnen. Het mogeduidelijk zijn dat door een homogene economie binnen Europa het maken vanmultiregionale modellen minder noodzakelijk wordt. Het is echter nog steeds mogelijkom met een dergelijk model de energiestromen binnen Europa in kaart te brengen.Verder zal OECD-Europa niet snel geheel homogeen worden aangezien erbijvoorbeeld nog steeds substantiële verschillen zullen blijven in deelektriciteitsvoorziening.

Vraaggestuurde Benadering

ECCO-modellen zijn aanbodsgestuurd hetgeen wil zeggen dat hetconsumptieniveau bepaald wordt door de producenten. De rol die consumenten cq.huishoudens hebben als drijvende kracht in de economie wordt echter steeds meeronderkend. Vandaar dat hier beargumenteerd wordt om het totale nationaleenergiegebruik te bepalen door rekening te houden met veranderingen in deconsumptiepatronen en de onderliggende veranderingen in de productiesector.Daarom wordt in het tweede gedeelte van deze dissertatie een vraaggestuurde variant,genaamd DREAM (Dynamic Resource and Economy Accounting Model), van hetECCO-model geïntroduceerd. Deze benadering stoelt op het feit dat bijna alleproductie uiteindelijk eindigt als consumptiegoederen en –diensten. Op deze wijze kanhet energiegebruik en de daarmee gepaard gaande milieuproblemen bepaald wordenvoor verschillende scenario’s omtrent toekomstige consumptie-activiteiten.

Aan de hand van twee casestudies wordt aangetoond dat de DREAM benaderingeen goede methode is voor het bestuderen van veranderingen in het energiegebruik alsgevolg van veranderende consumptiepatronen.

214

De eerste casestudie behelst een DREAM-model voor OECD Europa. In dezecasestudie wordt aangetoond dat het DREAM-model voor OECD-Europa zich andersgedraagt in vergelijking tot het ECCO-model voor deze regio. Deze verschillen zijnhet gevolg van verschillen in de essentiële parameters. In DREAM is er meestalsprake van een exponentiele groei in de industriële output waar in ECCO de groei inde industriële output veelal afneemt. Dit verschil is de oorzaak van het ontbreken vaneen negatieve feedbackloop in DREAM waardoor economische groei niet geremdwordt. Dit betekent niet dat aspecten van duurzaamheid omtrent economische groeiniet kunnen worden aangetoond met DREAM. De import-exportbalans als indicatorvoor afhankelijkheid van buitenlandse hulpbronnen en CO2-emissies wordenbestudeerd om de gevolgen van economische groei te bepalen.

In de tweede casestudie worden resultaten van een DREAM-model voorNederland gepresenteerd. Deze tonen aan dat op lange termijn een relatieve hogeconsumptiegroei zal leiden tot een toenemend totale fossiele energiegebruik hoeweler enorme inspanningen zijn gedaan in energie-efficiënte verbeteringen en deelektriciteitsvoorziening volledig gebaseerd is op zonne-energie. Deze resultatenimpliceren dat een daling in het (fossiele) energiegebruik en de CO2-emissie alleengerealiseerd kunnen door een lagere consumptiegroei of door drastischeveranderingen in het consumptiepakket. De gevolgen voor het energiegebruik doorveranderingen in het consumptiepakket kunnen echter nog niet bestudeerd worden metde huidige versie van DREAM vanwege het aggregatieniveau van het model. Ditsoort studies zullen mogelijk gemaakt kunnen worden in toekomstige versies wanneerde productiesectoren in meer detail beschreven wordt.

Enkele Slotopmerkingen

Teruggrijpend op de uitgangspunten van de ‘energyaccounting’-benadering, iszowel ECCO als DREAM ontwikkeld voor het bestuderen van de fysieke relatiestussen het economische systeem en het milieusysteem. In beide benaderingen wordtde relatie tussen economische activiteiten en het energiegebruik benadrukt.

De ECCO-methodologie heeft als uitgangspunt dat energie een essentiëleproductiefactor is en limiteert economisch groei aan de beschikbaarheid van‘bruikbare’ energie (zowel als hulpbron als in de vorm van kapitaal).

De DREAM-benadering is ontwikkeld voor het bepalen van de langetermijnconsequenties voor het energiegebruik als gevolg van veranderendeconsumptiepatronen, energiebesparingen, technologische verbeteringen endemografische ontwikkelingen. Hiermee worden in deze benadering de milieugevolgenvan diverse economische ontwikkelingen bepaald. Verder kan deze benadering ookgebruikt worden voor het bestuderen van die maatregelen die nodig zijn voor hetsturen van de consumptiepatronen in een meer duurzame richting.

Natuurlijk resulteren beide benaderingen niet in een geheel toereikendewetenschappelijk oplossing voor de duurzame ontwikkelingsvraagstukken omtrent

215

economische activiteiten. Echter door het gebruik van beide modellen kunneninzichten verkregen worden in (fysieke) interacties tussen het economische systeemen het milieu. En deze inzichten maken het inpassen van de fysieke benadering ineconomische theorieën de moeite waard. Beide methodes zullen daarom gebruiktmoeten worden in combinatie met conventionele economische methodes voor hetintegraal bestuderen van de relatie tussen milieu en economie.

University of Groningen Dynamic modelling of energy stocks and flows in the economy … ·...

Documents

Transcript of University of Groningen Dynamic modelling of energy stocks and flows in the economy … ·...