Geographic Macro and Regional (GMR) Model for EU Policy Impact Analysis of

Intangible Assets and Growth

Attila VargaPéter Járosi

Tamás Sebestyén

PTE KTK KRTI

Development policy instruments

• Knowledge-based development policy

• Policy instruments:– Promoting firms’ technological potential (start-up and

investment supports, tax credits, low interest rate loans or venture capital)

– Local technological environment support (R&D promotion: universities and private firms, human capital improvement, support of public-private interactions in innovation, financing physical infrastructure building)

• GMR: Geographic Macro and Regional Modelling

Why should geography be incorporated into development policy impact modeling?

• Geography and policy effectiveness:

1. Interventions happen at a certain point in space and the impacts appear there / spill over to proximate locations to a considerable extent.

2. The initial impacts could significantly be amplified/reduced by short run agglomeration effects.

3. Cumulative long run process resulting from migration of K and L:- further amplification/reduction of the initial impacts in the region- the spatial structure of the economy (K, L, Y, w) might eventually change in a significant manner.

4. Different spatial patterns of interventions might result in significantly different growth and convergence/divergence patterns.

Why „regional”

Why „macro”?

GMR-Eurozone

• The particular model developed for the Eurozone NUTS 2 regions includes:

– a KPF model (to model: 1 and 2)– an SCGE model (for 3)– a macro DSGE model (for 4)

Introduction

• Antecedents:

– Empirical modeling framework (Varga 2006)– EcoRet model (Schalk, Varga 2004, Varga,

Schalk 2004)– GMR-Hungary model (Varga, Schalk, Koike,

Járosi, Tavasszy 2008)– Dynamic KPF model for EU regions (Varga,

Pontikakis, Chorafakis, 2009)

Outline

• Model structure– The KPF model– The SCGE model– Dynamism and macro effects: macro DSGE

model (QUEST III)

• Policy simulations

The role of the KPF model

• To generate initial TFP changes as a result of technology policy interventions

• NOT for forecasting but for impact analysis

Equations in the TFP block 1. Log(PATENTS) = 1.325381*(-2.3006 + BETAPAT*Log(GRD(-2)) + 0.1804* Log(PSTCKN(-2)) + 0.4614* PAHTCORE) + U1 .....Knowledge Production [2. Log (PUBLICATIONS) = 2.6137 + BETAPUB*Log(NETRD(-2))]* Log(GRD(-2)) + 0.3293* PUBCORE+ U2] ............................ Publication Production 3. BETAPAT = [(0.7088 + 0.1439*Log(δ(-2))] .................................................................................................................................RD Productivity (patents) 4. BETAPUB = [0.4317 + 0.0003* WFP5_Log(RD(-2))] ..........................................................................................................RD Productivity (publications) 5. (GRD-GRD(-3)) = -391.369+ 352.437*BETAPAT(-3) + 325.33*BETAPUB(-3) + 266.917*RDHCORE-280.882*NL5REG+U3 .Endogeneous RD Growth 6. (HTEMP-HTEMP(-3)) = 11168.3 + [(0.0262 + 5.624E-06* GRD(-3))]* HTEMP(-3) + 21321.1*RDCORE+ U4 Endogeneous High-Tech Employment Growth

7. PSTCK = PSTCK(-1) + PATENTS ................................................................................................................................................................... Patentstock 8. PSTCKN = SUM(PSTCK) .................................................................................................................................................................... National Patentstock 9. HTEMP = HTEMP(-3) + (HTEMP – HTEMP(-3)).......................................................................................................................... High Tech Employment 10. HTEMPEU = SUM(HTEMP)............................................................................................................................................ National High Tech Employment 11. TOTEMP = FROM SCGE.............................................................................................................................................................................. Employment 12. TOTEMPEU = SUM(TOTEMP) ........................................................................................................................................................... Global Employment 13. ..................................................................................................................................................................................................Knowledge concentration 14. TFP = 57.42*(HUMCAP(-2)) 0.0004*SOCKAP(-2) (PATSTCK(-2)) 0.0056*ln(DENS(-2)) TFP equation

δi = [(EMPKIi / EMPKIEU) / (EMPi / EMPEU)] / [(1 - ∑ j (EMPKIi,j / EMPKIj,EU)][1 – (EMPi / EMPEU)]

The TFP equationTable 1. Regression Results for Log (TFP) for 135 Eurozone regions, 2004

Model (1) (2) (3) (4) (5) Estimation OLS OLS OLS OLS IV (2SLS)

Spatial Lag (INV1)

Constant Log(HUMCAP) Log(HUMCAP)*SOCKAP Log(PATSTCK) Log(PATSTCK)*Log(DENS) W_Log(TFP)

3.6425*** (0.2105)

0.0722*** (0.0175)

4.0850*** (0.0460)

0.0008*** (7.9577E-5)

3.9331*** (0.0425)

0.0003*** (8.7574E-05)

0.0623*** (0.0078)

3.9832*** (0.0385)

0.0004*** (7.5823E-5)

0.0073*** (0.0008)

3.9309*** (0.0414)

0.0004*** (7.4023E-5)

0.0054*** (0.0010)

0.0015*** (0.0005)

R2-adj Sq. Corr.

0.11 0.41 0.60

0.63 0.65

Multicollinearity condition number White test for heteroskedasticity LM-Err INV1 INV2 LM-Lag INV1 INV2

22

8.8335**

154.48*** 19.56***

52.47*** 29.31***

6

11.1798***

57.35*** 9.00***

38.11*** 22.03

9

10.5357*

1.57 0.61

14.98*** 11.33***

7

7.7393

3.27* 0.02

7.67*** 3.09*

1.38

Require the integration of TFP with the SCGE and MACRO models

• BUT:– How strong these processes are?– What are the economic impacts on the

regions?– What are the macro (EU level) economic

impacts?

The role of the SCGE model

• To generate dynamic TFP changes that incorporate the effects of agglomeration externalities on labor-capital migration

• Agglomeration effects depend on:- centripetal forces: local knowledge (TFP)- centrifugal forces: transport cost, congestion

• To calculate the spatial distribution of L, I, Y, w for the period of simulation

The SCGE model

• C-D production function, cost minimization, utility maximization, interregional trade, migration

• Equilibrium:

- short run (regional equilibrium)

- long run (interregional equilibrium)

Main characteristics of the SCGE model

• NOT for historical forecasting• The aim: to study the spatial effects of

shocks (technology policy intervention)• Without interventions: it represents full

spatial equilibrium - regional and interregional (no migration)

• Shock: interrupts the state of equilibrium, the model describes the gradual process towards full spatial equilibrium

The role of the MACRO model

• Regional technology policy impacts depend to a large extent on macro level variables (fiscal/monetary policy shocks, exchange rates, international trade etc.)

• Dynamising the (static) SCGE model

The MACRO model

• The QUEST III Dynamic stochastic general equilibrium (DSGE) model for the EURO area

• A-spatial model

• Macro effects of exogenous TFP shocks

• Baseline: TFP growth without interventions

• Policy simulations: describe the effects of TFP changes on macro variables

Regional and national level short run and long run effects of TFP changes induced by regional technology

policy interventions

1. Intervention in any region changes regional TFP level

2. „Short run” effect: - price of the good decreases

- decreasing demand for both L and K (substitution effect - SE) - increasing regional and interregional demand for the good increases demand for L and K (output effect - OE)- if OE>SE: increased regional demand increases wages and

utility levels of consumers in the region

3. „Long run” effects: increasing utility levels induces labor migration into the region (until congestion does not prevail) followed by capital migration

- resulting in a further increase in TFP- and finally a changed spatial economic structure

4. Macroeconomic variables reflect the long run equilibrium TFP level resulting from dynamic agglomeration effects

Policy Models, Procedures State of Equilibrium

MACRO modelDynamic

supply and demand side effects

Regional SCGE modelAgglomeration effects on regional and interregional

variables

Regional KPF model

Regional TFP effectsPolicy

intervention

A

CB

Dynamic impact on macroeconomic variables

Dynamic impact on regional economic

variables

Data, software environment

• The model is build for the NUTS 2 regions of the EURO zone

• Regional KPF model estimated in SpaceStat • The complex model is programmed and run in MATLAB• Easy to run/make simulation changes with an Excel

interface• The regional model is large considering that equilibriums

have to be found for 144 interconnected (interregional trade and migration) regions

• A simulation with 20 periods needs the computer time of about 20 minutes

Regional R&D policy impact assessment: The EU FP6 program

• EURO zone 144 NUTS 2 regions (QUEST constraint)

• Interventions: 2003-2007

Regional shares of FP6 funds0 - 0.0030.003 - 0.0090.009 - 0.0170.017 - 0.0370.037 - 0.153

Figure 2. Regional distribution of FP6 funds in the Euro-zone, 2003-2007

-0,40%

-0,20%

0,00%

0,20%

0,40%

0,60%

0,80%

1,00%

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022

TIER1 TIER2 TIER3 TIER4 EU Figure 3. Average FP6 impacts on GDP in regions belonging to different agglomeration

tiers: percentage differences between scenario and baseline values

Y2022-0.007 - -0.003-0.003 - 0.0010.001 - 0.0060.006 - 0.0180.018 - 0.029

Figure 4. Regional impacts of FP6 funds on GDP of Euro-zone regions, year 2022:

percentage differences between scenario and baseline values

0,00%

0,05%

0,10%

0,15%

0,20%

0,25%

0,30%

0,35%

0,40%

0,45%

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022

SCEN_Y/BASELINE_Y

Figure 5. Impacts of FP6 funds on EU GDP, Euro-zone, period 2003-2022: percentage differences between scenario and baseline values

0,000%

0,005%

0,010%

0,015%

0,020%

0,025%

0,030%

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022

SCEN_Y-BASELINE_y_growth Figure 6. The impact of FP6 funds on EU-level GDP growth rates, Euro-zone, 2003-2022:

percentage point differences between scenario and baseline values

-0,60%

-0,40%

-0,20%

0,00%

0,20%

0,40%

0,60%

0,80%

1,00%

1,20%

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022

TIER1 TIER2 TIER3 TIER4 EU Figure 7. The effect of EU FP6 research support augmented with an annual 1 percent

quality-oriented redistribution of national R&D expenditures, Euro-zone, 2003-2022: percentage differences between scenario and baseline values

-0,40%

-0,20%

0,00%

0,20%

0,40%

0,60%

0,80%

1,00%

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022

TIER1 TIER2 TIER3 TIER4 EU Figure 8. The effect of a 0.5 percent annual increase of human capital in Tier 2, 3 and 4

regions to compensate for the impact of the quality-oriented redistribution of national R&D expenditures, Euro-zone, 2003-2022: percentage differences between scenario and baseline values

-0,40%

-0,20%

0,00%

0,20%

0,40%

0,60%

0,80%

1,00%

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022

TIER1 TIER2 TIER3 TIER4 EU Figure 9. The effect of a 0.05 percent annual increase of social capital in Tier 2, 3 and 4

regions to compensate for the impact of the quality-oriented redistribution of national R&D expenditures, Euro-zone, 2003-2022: percentage differences between scenario and baseline values

Compared to the relatively small share of EU Framework Program research support in Member States’ R&D budgets regional and EU level economic impacts of FP6 expenditures are considerable. It suggests that this policy instrument is an effective tool not only for promoting scientific publication activities but also for supporting regional and macro level productivity and economic development.

Redistributing R&D funds to regions where research productivity is the highest is a

promising economic policy instrument in the hands of Member States. This instrument increases regional GDP in the most agglomerated regions as well as at the level of the European Union. However, as expected there is a small negative effect on regions with average development and a more adverse effect on lagging regions.

There are policy instruments to compensate for the negative effects of specialization in the

form of a spatial quality redistribution of R&D resources. Continuous regional human capital development can successfully overcompensate the adverse effects in regions where technological knowledge is about medium developed. There is also a considerable impact of regional human capital development on GDP at the macro level.

Policy implications

Policy implications (cont.)

Compensating for R&D specialization in the form of persistent social capital development is also a powerful tool for Member States to improve economic positions of regions with medium-level agglomeration of technological knowledge. This policy option results in a significant macro level GDP impact as well.

It is clear from the policy analyses that EU regions where agglomeration of technological

knowledge shows the lowest levels are not responsive to compensations in forms of either human capital or social capital development. These regions should be considered separately when local development policies are formed. They are not (yet) able to be the sites of future knowledge-based development. Instead, specific sectoral policies aiming at leisure or tourism would be more effective for those regions.