Post on 01-Aug-2020
Dealing with Complexity in Society:
From Plurality of Data to Synthetic Indicators
September 17th and 18th, 2015 1
Jan W. Owsiński Systems Research Institute, Polish Academy of Sciences
Newelska 6, 01-447 Warszawa
owsinski@ibspan.waw.pl
IS THERE ANY ‘LAW OF REQUISITE
VARIETY’ IN CONSTRUCTION OF INDICES
FOR COMPLEX SYSTEMS?
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 2
OUTLINE
1. The modelling project in the capital region of
Poland
2. The nature of variables & indicators
3. The diverse ways to synthesising
4. The purpose & the policy instruments
5. The requisite variety at work…
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 3
The modelling project in Masovia (1)
# In 2011-2013 a modelling project was carried out, commissioned by the
Masovian Bureau of Regional Planning (Masovia: capital province of
Poland)
# The project was financed from a program, supported by structural
funds of the EU, called Development Trends of Masovia
# The specifications for the modelling project included 19 domains of
modelling and the time horizon of 2025. It was assumed that
municipalities shall be the basic objects in the project
# The work was done by a team from the Systems Research Institute
and the Institute of Geography and Spatial Organization, both of the
Polish Academy of Sciences (some 15 persons in total)
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 4
The modelling project in Masovia (2)
# Within the 19 domains specified in the commission, altogether some
70 variables / indicators were formulated, along with more than 20
auxiliary variables / indicators
# For all of these, mostly relatively simple forecasting models, based on
various prerequisites, were developed
# In a vast majority of cases, the models were specified (forms,
parameters) for municipality types, but actually run for each of the
municipalities (315 in total) in annual steps until 2025
# A computer application was developed, along with a database, to run
the models for various assumptions („scenarios”), elaborated within
the project
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 5
The modelling project in Masovia (3)
1. Population Population numbers, various
demographic indicators
Birthrate, mortality and
migration scenarios
2. Social
capital
NGOs, sports clubs members,
art circle members + synthetic
indicator
For municipalities, but
also for counties,
separately + synthetic
3. Wealth of
municipalities
Own and total revenues, capital
expenditures, etc. + auxiliary
Model types for
municipality types
4. Municipality
investments
Revenues, capital expenditures,
debt, propensity to invest
An interactive model
5. Intellectual
capital
University graduates & students,
foreign capital firms + synthetic
indicator
Indicators at different
resolution levels +
synthetic one
6. Labour
market
Demand for labour (GDP);
auxiliary: productivity
Based on variables from
other domains
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 6
The modelling project in Masovia (4)
7. Social
exclusion
Synthetic indicators, based
on domains 1, 5, 14, 15
Gini-like measure is applied for
wages
8. Quality of
life
Single synthetic indicator
based on domains 3, 5, 7,
9, 15, 16, 17
Relative indicator, based on 7
variables (some of them
synthetic)
9. Technical
infrastructure
Shares of inhabitants using
water supply, sewage and
water treatment
Models for municipality types
and levels attained
10. External
investments
Value of external
investments per capita
Based on variables from other
domains
11. Inno-
vativeness
Composite indicators,
based on domains 3, 5, 13
Two indicators, for two levels
of resolution
12. Information
society
Based on 11 and internet in
schools
Two indicators (school levels
in municipalities and counties)
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 7
The modelling project in Masovia (5)
13. Fixed
assets
Assets of public bodies,
companies, total, per capita
Auxiliary: investments
14. Enterprise
& employment
Several derived indicators
and auxiliary variables
Models for municipality types
15. Transport
accessibility
Travel time based indicator of
population number
Based on road and
settlement networks
16. Quality of
environment
Synthetic indicator based on
several (+auxiliary) variables
Includes variables from
domains 17 and 19
17. Sustainable
development
Synthetic indicator based on
several auxiliary variables
Models for auxiliary variables
(e.g. car number)
18. GDP value Global and local GDP
dynamics
Simple macroeconomic
models with six scenarios
19. Scale&rate
of urbanisation
Several indicators based on a
couple of variables
Some variables from other
domains
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 8
The modelling project in Masovia (6)
-- a single run of the population model for a selected scenario
(births, mortality, migrations) produces 315 municipalities x 15
years x 15 age groups x two sexes x indicator categories (5) =
more than 700 000 numbers, constituting a forecast
-- altogether some 100 models (variables / indicators + auxiliary
variables) were developed, varying as to their complexity, many
of them featuring variants (parameter-wise) for the types of
municipalities, so that we deal with several hundred different
effective models
-- there are altogether more than 100 of feasible scenarios,
which can be selected by the user
-- all this costed the commissioning agency some 160 000 €…
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 9
The variables and indicators (1)
The variables (quantities) appearing in the models, can be very
roughly classified into:
-- original real-life variables (population number, number of
businesses, number of computers in schools,…)
-- first-order indicators (population density, forest share of
municipality area, university graduates per 1 000 inhabitants,…)
-- second-order indicators (GDP per capita, own revenues of
municipality per capita, employment per business,…)
-- composite variables (variables, defined effectively as models,
based on other variables)
-- synthetic indicators (these are no longer „models”…)
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 10
The variables and indicators (2)
The issues:
-- selection of magnitudes and their interpretations
-- availability of data, their formal definitions and
interpretations
-- precision of measurement and degree of „truth”
(certainty)
-- validity and possibility of verification of models
-- in case of synthetic indicators: can we check the
meaning?
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 11
The variables and indicators (3)
The objectives & purposes:
-- to know what is going on – a selection of „most important”
variables in „cognitive” terms
-- to assess the situation from a broader perspective, but still
within a well-defined domain (e.g. the joint meaning of a
couple of variables, pertaining to the same subject matter)
-- to assess the situation from a really broad perspective
(„quality of life”, „sustainable development”,…)
-- to determine the ways to act, the policy levers and the
undertakings meant to achieve definite goals…
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 12
The variables and indicators (4)
The summary:
-- sometimes a commonsense assessment is more valuable
than a million data items and a study worth a million…
-- sometimes there is an intuition as to „what it all boils down
to…”
-- sometimes there are variables that could be picked up and
(relatively easily) measured, which shed light on a much
wider landscape of phenomena
-- but beware of both these hunch-guesses and the academic
immovable truths…: keep to your ground knowledge
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 13
The ways to synthesise (1)
The summary:
-- synthesising can be easy: when variables pertain to the
same subject, are similarly measured, and correlated (A) -- synthesising can be challenging: when variables are [only]
interconnected, like through a model (often the question
arises: if there is a model, why synthesise?) (B) -- synthesising can be difficult: when variables are of widely
differing character, and the meaning of the resulting indicator
is doubtful… (C)
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 14
The ways to synthesise (2)
Case (A): synthesising can be [relatively] easy: when
variables pertain to the same subject, are similarly measured,
and correlated
An example:
9. Technical infrastructure: shares of municipality inhabitants with
access to (i) water supply; (ii) sewage; (iii) water treatment (in %):
# generally very high correlations (exceeding 0.85, even close to 0,98)
# classification into municipality types (kinds of needs…)
# upper limit of 100% (saturation) – many entities already there…
## choices: (a) average score; (b) minimum of the three – accounting for
municipality type; (c) weighted score, e.g.: w1=0.45, w2=0.45, w3=0.1
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 15
The ways to synthesise (3)
Case (B): synthesising can be challenging: when variables
are [only] interconnected, like through a model (often the
question arises: if there is a model, why synthesise?)
Examples:
6. Labour absorption capacity = GDP / productvity, both of which are
separately modelled
15. Transport accessibility = number of inhabitants who can reach a given
destination (distance+ population) withina given time limit
3. Wealth of the municipalities:
DWMit=DWMit-1+aSATDWMi + (1-a)a2typ(WWPAt – WWPAt-1);
a simple econometric model superposing trend, with an increment in the
trend, and the influence of an important independent variable
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 16
The ways to synthesise (4)
Case (C): synthesising can be difficult: when variables are
of widely differing character, and the meaning of the resulting
indicator is doubtful
Examples:
6. Quality of life: JZM = f(DWM,SWM,ZWS, [WOD,KAN,OCZ], DTR, [JSP, UWP,
UWS1]) = (DWM# + SWM# +ZWS#+(WOD+KAN+OCZ)#/3 + DTR#+
(JSP+UWP+UWS1)#/3))/6; a weighted sum of the unitarised variables, assumed to
contribute to quality of life of the inhabitants
11. Innovativeness: WIGit = f(SWM, PKZ, WWPA, WMI) = (SWMit* + PKZit* +
WWPAit* + WMIit*)/4; similarly as above…
16. Quality of natural environment: JSPit = f(ZIEL, ROLG, GZL, AUTO, GZB) =
ZIEL’i + ROLG’i - dGZL(typ)i – (GZL#it + AUTO#it + GZB(scen)#it)/3; ibidem
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 17
The ways to synthesise (5)
Some trivial conclusions:
-- the situations (may) differ very widely, both in statistical
and in substantive terms
-- each case must be considered separately, with application
of statistical and „theoretical” reasoning
-- it is highly probable that a given synthetic indicator turns
out to be poorly justified (apples and oranges summed up)
-- to see landscape, one has to dispose of three dimensions
(cognitive needs)
-- an important aspect is constituted by the „purpose” (the
required number & character of dimensions)
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 18
The purpose and the policy instruments (1)
What do we need an indicator for? Whether
and why should we go for a single, all-
embracing indicator?
(The answer „to simplify” is incorrect.)
If the purpose is policy making, then very
rarely one dimension suffices (e.g.
determining tax levels or fees).
In most cases, for practical reasons, more
than one indicator is needed.
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 19
The purpose and the policy instruments (2)
Even if we have only one policy instrument (e.g. subsidy), it
is not true that we need just one indicator.
This is because the instrument can almost always be used in
a variety of manners (e.g. subsidy to different areas of socio-
economic life or for different kinds of undertakings).
Thus, instead of trying to „aggregate everything”, a matching
should be sought between the diversity of encountered
situations and the possibility of diversifying policy execution.
The matching shall, definitely, very rarely be „perfect”,
anyway…
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 20
The law of requisite variety at work
The postulate of „matching” the indicators to the policy
instruments and their use gives rise to a different kind of
problem, involving the multidimensional scaling and other
data analysis techniques (clustering) and the MCDM
techniques.
Data / variables
Policy
instruments &
their use
MDS & other data
analysis
techniques The dimensionality and
granularity required,
and transformations
What is
measured, how
and why?
MCDM methods
Can we take a
rational decision?
The decision
Opening Session
The modelling project
The variables & indicators
The ways to synthesise
The purpose & the policy instruments
Jan W. Owsiński Dealing with Complexity in society 21
The law of requisite variety at work?
But that is an entirely different
story…
So,
for now, thank you very much for
your attention,
and/or, indeed, for your patience…
Opening Session
Title Section 1
Title Section 2
Title Section 3
Title Section 4
First Author et al. Dealing with Complexity in society
22
OUTLINE
2. Title Section 2
Title Subsection 1
Title Subsection 2
…….