construction and evolution of a multidimensional well-being index for the spanish regions
TRANSCRIPT
Construction and Evolution of a MultidimensionalWell-Being Index for the Spanish Regions
Antonio Jurado • Jesus Perez-Mayo
Accepted: 20 September 2010 / Published online: 13 April 2011� Springer Science+Business Media B.V. 2011
Abstract The study presented here is an attempt to calculate a comparative multidimen-
sional index of economic well-being for the Spanish Autonomous Communities. Based on the
dimensions of adjusted consumption, real wealth, equity and economic security per inhabi-
tant, we obtain one relative index using a system of uniform weightings, a second resulting
from a factor analysis and a third provided by a DEA analysis. We elaborate the index for the
year 2000 and for 2006, thereby providing the relative results with greater robustness and
permitting some conclusions to be reached with regard to the evolution over time of the index.
Keywords Economic well-being � Factor analysis � Indicators � DEA
1 Introduction
The disparities in well-being evidenced by the Spanish Autonomous Communities1 (the 17
political-administrative regions of Spain, CC.AA. in their Spanish initials), while having
been for some time a cause for concern in economic and political debate, are currently
entering a phase in which their quantification is increasingly important.
On the one hand, European structural funds or the Spanish Interterritorial Compensation
Fund (FCI) use principally macromagnitudes2 as economic delimiters of shares, and could
A. JuradoDepartment of Economics, University of Extremadura, Caceres, Spain
J. Perez-Mayo (&)Department of Economics, University of Extremadura, Badajoz, Spaine-mail: [email protected]
1 Some authors, such as Ayala et al. (2011), Garcıa-Luque et al. (2009), Perez-Mayo (2008), or Poggi (2007)for example, included the territorial issue in the analysis of poverty, deprivation and social exclusion in Spain.2 The distribution of European structural funds is based, in the Objective 1 regions, solely on a percentageof average European income that indicates which regions will be recipients of such resources. However, theSpanish FCI (Interterritorial Compensation Fund) has to date employed a formula which, in addition to grossadded value, takes into consideration: the de jure population, the migratory balance, the number ofunemployed, geographical extent and population dispersion. Nevertheless, as we shall see below, our indexis much more ambitious with regard to the multidimensionality captured.
123
Soc Indic Res (2012) 107:259–279DOI 10.1007/s11205-011-9835-4
be more ambitious in the quantification of this matter. On the other, increasing interest in
regional fiscal balances3 in the framework of a statutory restructuring should be accom-
panied by a parallel and more exhaustive broadening of the indicator to be employed when
taking decisions of an inter-regional nature.
In the national and international panorama of this type of study, in the majority of cases
the analysis has been based on a macromagnitude such as GDP, income or similar,
combined with some indicator of economic equity (inequality indices and/or poverty
indices). These are the so-called abbreviated welfare functions.
Many authors,4 such as Nordhaus and Tobin (1972), Sen (2001), Khan (1991), Stewart
(2005), or Stiglitz et al. (2009), have raised criticisms of the use of a single macromag-
nitude. Consequently, many studies have moved beyond purely economic indicators and
entered the field of social well-being. The measurement of social well-being includes
dimensions such as material well-being, education, health, and participation in the social
activity of the environment, or may even consider questions such as crime or the climate,5
which, although clearly influential, sharply increase the problem of arbitrariness in the
choice of variables and weightings.
Although inter-country comparisons are more common, regional-level studies have also
been undertaken, such as the above-mentioned work by Stewart (2002 and 2005), applied
to the European regions.
In the field of the most recent comparative regional studies in Spain, Marchante et al.
(2006) elaborate a well-being index for the Autonomous Communities, based on a
broadening of the Human Development Index (HDI) of the United Nations Development
Program (UNDP) and by using the following variables: life expectancy, infant mortality
rate, illiteracy, school enrolment and gross added value per capita (GAV p.c.). Ayala et al.
(2006) focus on the same territorial units, calculating abbreviated welfare functions as a
multiplicatory trade-off between an income component and an inequality component.
Studies such as that by Osberg and Sharpe (2000, 2003, 2005 and 2009) indicate how a
wider notion of economic well-being must include dimensions of wealth, inequality and
insecurity, components which neither the work by Marchante et al. (2006) nor the HDI
itself claim to encompass and which, in the case of wealth and insecurity, the study by
Ayala et al. (2006) preferred to leave untouched. In that same year, Murias et al. (2006)
estimated an economic well-being index based on the structure supplied by Osberg and
Sharpe. Their work uses only one of the three systems of weightings applied in the present
study, does not study evolution over time and selects the Spanish provinces as the terri-
torial units of study. Bearing in mind the high and increasing degree of decentralisation of
the Spanish administration, and that almost all policies that affect citizens’ well-being are
already the responsibility of the regional administrations (Autonomous Communities), we
consider it essential to take the regions as the territorial unit of reference.6
3 The study ‘‘The fiscal balances of the Autonomous Communities with the Central Public Administration,1991–2005’’, Uriel and Barberan (2007), was published on 28 November 2007. It is the first scientific reporton the subject, since Spanish governments had never previously made public results of this nature.4 In addition to the references in the text, brief reviews of some attempts to measure well-being, taking intoaccount different aspects than macromagnitudes, can be found in: Zolotas (1981), Daly and Cobb (1989),Cobb and Cobb (1994), Cobb et al. (1995), Anielski and Rowe (1999), Jackson (2004), or Wolff et al.(2005).5 A wider analysis of wellbeing can be found in, for example, Diener et al. (1995), Lelkes (2005), or Rojas(2007).6 A brief description of the Spanish administrative structure can be found in Annex 1.
260 A. Jurado, J. Perez-Mayo
123
Returning to the structure established by Osberg and Sharpe, they consider the HDI to
be a social well-being index (possibly the best known worldwide) with three components:
health, education and ‘‘economic resources’’. The authors focus their attention on this last
point. The index is based on GDP per capita and is intended to capture the capacity of
persons to ‘‘lead a long and healthy life, acquire knowledge and have access to the
resources necessary to achieve a decent standard of living’’ (UNDP 2004, p. 127). They
believe the measurement of the economic aspect of the HDI can clearly be improved and
propose the construction of an economic well-being index to replace the GDP employed.
The design used in their studies is, in our view, both the most efficient and simplest of
the economic well-being indices constructed to date. Thus, with the aim of making a
deeper contribution to this subject, in the ambit of the Spanish regions (where we believe
controversy will increase in the near future), we develop an initial approximation of a
regional economic well-being index, founded on the basic structure used by Osberg and
Sharpe in a comparison of a group of developed countries.
We must insist upon the point that our index is ‘‘economic’’ and, consequently, aspects
such as health and education are not included because would convert it into an index of
social well-being. That would be a much more ambitious project, which should wait, at
least, until the present research has matured.
The paper is structured in four sections. The first deals in detail with the methodological
aspect; it defines the dimensions which the index is intended to capture, the variables we
have been able to use to represent each dimension, and the different techniques employed
to treat the weightings. The second section quantifies the index for each of the Autonomous
Communities in the year 2000, using three different weighting systems. The third analyses
the evolution over time in the period 2000–2006, while the final section ends the study with
a brief description of conclusions and future intentions.
2 Methodology of the Study
The well-being index designed has been structured in four dimensions: adjusted con-
sumption, real wealth, equity and economic security. We consider that these capture the
most influential effects of the purely economic variables operating in what we have
attempted to include within the concept of economic well-being.
The consumption flow is perhaps the factor having the clearest influence on family well-
being and that of its members. We consider it important to find data that reflect the
consumption of both market goods and services and of public goods and services provided
by the distinct administrations.
Within the consumption flow, there are some factors for which it is totally or partially
impossible to obtain data in the regional sphere, such as: the consumption of leisure, the
shadow economy, disability-free life expectancy, some consumption costs such as that of
the journey to work, the consumption of products which seriously damage health or the
consumption of illegal products which lead to crime and produce collateral costs.
Adjusted private consumption was calculated using monetary and non-monetary
expenditure7 (separately, to detect possible differences in the weightings), extracted from
the Continuous Household Budget Survey (CHBS) for the year 2000 and from the
7 The expenditures registered in CHBS and HBS refer to both the monetary flow that the household andeach of its members spend on the payment of certain goods and services, and the non-monetary flowcontaining self-consumption, self-supply, wages in kind, free or subsidised food and rent imputed to the
Construction and Evolution of a Multidimensional Well-Being Index 261
123
Household Budget Survey (HBS) for the year 2006. In the former case, we employed the
longitudinal file updated annually by the Spanish National Statistics Institute (INE) from
the quarterly surveys. Specifically, we employed in both cases the adjusted datum with a
parametric equivalence scale, with k equal to 0.5 and multiplied by life expectancy.
The life expectancy index has been used for several authors, i.e. Osberg and Sharpe
(2000), in measuring economic well-being. This expectancy has increased in the last years
in many countries and regions but not with the same strength. Despite the inter-regional
differences in life expectancy are not very notable we consider that long life is an
important factor of well-being in a society. The ‘‘extra years’’ of some regions in relation to
average life could have an economic value that should be included in the consumption
flows of individuals. Osberg and Sharpe (2000) said, ‘‘People care both about how much
they consume per year, and how many years they get to consume it’’.
The life expectancy index was obtained by the INE from the Analysis and demographic
studies (mortality data).
Adjusted public consumption was obtained from the data for total public consumption
supplied by the Regional Economic Balance Sheet (Autonomous Communities and
Provinces), years 2000–2005, and Cuadernos de Informacion Economica No. 203, April
2008, published by the Foundation of Confederate Savings Banks (FUNCAS).
Another variable taken into account in most of the studies is the working hours.
Especially, in a group of regions with similar leisure activities, the differences in hours
worked yearly per worker must influence on the well-being level. Moreover, the increasing
decentralization in Spain permits differences in labour legislation.
Any discussion of well-being must refer to intra- and inter-generational economic
equity. The traditional method of quantifying the first of these is to use inequality indices
or poverty indices. Ayala et al. (2006) offer a more exhaustive analysis of inequality in the
regional sphere. The indices of inequality and poverty intensity employed here were the
Gini index and the FGT 1, compiled by the authors from the HBS 2006 microdata and from
the longitudinal file 2000 obtained by the INE from the CHBS.
With regard to inter-generational equity, our intention is to provide a number of vari-
ables which attempt to measure the legacy which society will foreseeably leave to the
following generation, consisting of an accumulated capital which would serve as a pro-
ductive structure in the future society. However, this transfer should also include an
environmental equity, a solid foundation of research and a controlled public debt,8 which
would not excessively reduce the future capacities of the economy.
To quantify physical capital we use the nominal net capital stock obtained by the
Valencian Economic Research Institute (IVIE) in its study of the period 1964–2005.
Human capital is represented by the ratio of the working age population with higher
education to the total working age population. This data is obtained from the IVIE for the
year 2000, and for the year 2006 it has been calculated from the HBS 2006.
Footnote 7 continueddwelling in which the household resides (when it is the owner of the same, or the dwelling was granted freeof charge or at a very low price by other households or institutions).8 Regional public debt was initially included as an explanatory variable. However, when applying thefactorial analysis, an insignificant weighting, which recommended its elimination, was obtained. Further-more, from a theoretical point of view, regional debt would not be the same here as national debt because ofthree reasons. First, the Autonomous Communities’ debt is tightly controlled and regulated by the CentralTreasury. Second, it is principally assigned to investment and it is, therefore, debatable that it subtracts realwealth from the next generation. And, finally, figures for it may not be complete because the regionaladministrations usually spend in excess of their budgetary allocation (especially in recent years).
262 A. Jurado, J. Perez-Mayo
123
Expenditure on R&D is captured by internal expenditure on R&D by Autonomous
Community, provided by the INE. Expenditure on environmental protection is taken from
the same source, specifically its survey of industrial companies’ expenditure on environ-
mental protection.
Lastly, economic insecurity is, from our point of view, one of the principal causes of the
absence of well-being in families in developed countries. The economic problems that may
accompany illness, divorce,9 old age and unemployment (especially in the absence of
benefit coverage) enormously reduce both present economic well-being and certainty
surrounding the standard of living in the near future.
Using once more the CHBS 2000 and the HBS 2006, we calculated the percentage of
expenditure on health in relation to monetary income10 and the poverty rate for over-65s.
The INE’s database used for obtaining the unemployment rate is the Economically Active
Population Survey (Annual results, Autonomous Communities). In addition, the unem-
ployment benefit coverage, calculated as the number of recipients in relation to the total
number of unemployed, comes from the State Public Employment Service.
To take into account the disparities in prices that may exist among the different regions,
the monetary expenditure, non-monetary expenditure, Gini index, FGT1 index and over-
65s poverty rate variables have been adjusted by applying purchasing power parities
(Alcaide and Alcaide 2008).
2.1 Calculation of the Index
To obtain the well-being index from the indicators (central column, Table 1 above), each
has been rescaled at the interval [0,1] in accordance with the methodology employed by the
UNO in the elaboration of the Human Development Index (HDI). The maximum and
minimum reference values are taken for the set of the Autonomous Communities and for
each indicator. Subsequently, using the expression (1), the value of each indicator in each
of the regions is determined. These values will range between 0 for the region having the
lowest value of the indicator and 1 for the Autonomous Community showing the greatest
value.
indicator of region i ¼ value of the variable in i�minimum reference value
maximum reference value�minimum reference valueð1Þ
Having transformed the indicators [0,1], it is possible to determine the values of each
dimension of economic well-being of the Autonomous Communities (adjusted consump-
tion, real wealth, equity and economic security). To obtain the values of the dimensions
9 We initially included the rate of divorce as a variable representative of the risk of single-parent poverty;however, when applying the weightings based on the factorial analysis the system assigned a positiverelation between well-being and divorce. Although the relation we intended to quantify was theoreticallynegative, the results show that in the Autonomous Communities where well-being was greatest the divorcerate was higher. The reason may be that upper-income families are more likely than lower-income familiesto be able to ‘‘permit themselves’’ to undergo divorce and the consequent loss of family income. In view ofthe results, we decided to eliminate this variable from our index.10 This variable is used as proxy for health status and its main economic effect on well-being. The healthexpenditure of the households in the HBS includes medicines, other pharmaceutical products, therapeuticmaterials, extra hospital medical and paramedical services and hospital services. Although the SpanishHealth System has a universal coverage, medicines prices or some therapeutic materials and services arepartially covered. Although there are some differences in certain expenditures (not very relevant) among theregional Health Systems, it is assumed in the analysis that the public health sectors are covering the samebenefits (in cash or in kind) because the evaluation of Regional Health Systems is not the aim of this paper.
Construction and Evolution of a Multidimensional Well-Being Index 263
123
from the indicators, we applied uniform weighted averages,11 taking into account the
positive or negative sign corresponding to each indicator. Because the dimensions had
already been obtained, we did not proceed to the rescaling described above, and thus the
values are no longer between 0 and 1.
Subsequently, it was necessary to establish a system of weightings that would permit
their combination to construct the general index. One of the methodological options would
have been to initiate the calculation assuming that distribution is uniform (i.e. an impor-
tance of 25% for each dimension) and equitable for each indicator within each dimension.
However, as Osberg and Sharpe (2000) state, different weightings may be considered, such
as 40% for adjusted consumption, 10% for real wealth and 25% for each of the remaining
dimensions. Similarly, with regard to equity from a Rawlsian point of view, it is possible to
dedicate more attention to poverty than to inequality, with the resulting changes in the
weightings.
It must be borne in mind that a greater magnitude of ‘‘adjusted consumption’’ or of ‘‘real
wealth’’ represents greater well-being, while the opposite occurs with the other two
components. That is to say, the greater the value of ‘‘adjusted consumption’’ or of ‘‘real
wealth’’, and the lower the value of ‘‘equity’’ and ‘‘economic security’’, the greater will be
the well-being index obtained for the region.
Table 1 Structure of the well-being index
Structure of the economic well-being index for the CC.AA.
Dimensions Indicatorsa Variables
1.-Adjusted consumption 1.1.-Adjusted privateconsumption
1.1.1.-Equivalent adjusted monetaryexpenditureb
1.2.-Adjusted publicconsumption
1.1.2.-Equivalent adjusted non-monetaryexpenditureb
1.3.-(-) Hours worked yearlyper worker
1.1.3.-Life expectancy index
1.2.1.-Total public consumption per capita
1.3.1.-(-)Hours worked yearly per worker
2.-Real wealth—intergenerational equity
2.1.-Physical capital 2.1.1-Net capital stock p.c.
2.2.-Human capital 2.2.1-Rate of postgraduates/population
2.3.-R&D 2.3.1-Expenditure on R&D p.c.
2.4.- Environment 2.4.1-Company expenditure onenvironmental protection
3.-Equity 3.1.-(-) Gini index 3.1.1.-(-) Gini indexb
3.2.-(-) Poverty index 3.2.1.-(-) FGT 1 indexb
4.-Economic security 4.1.-(-) Risk of illness 4.1.1.-(-) % expenditure on health
4.2.-(-) Risk of over 65poverty
4.2.1.-(-) Poverty rate over 65sb
4.3.-(-) Risk ofunemployment
4.3.1.-(-) Unemployment rate
4.3.2.-Unemployment benefit coverage
a The minus sign in brackets before indicators means that a negative relation with the well-being index isassumedb Adjusted by the purchasing power parities provided by Alcaide and Alcaide (2008)
11 Our study also calculates the weightings using other methodologies (factorial analysis and efficiencyanalysis), to eliminate the arbitrariness which the exogenous application of the weightings may involve.
264 A. Jurado, J. Perez-Mayo
123
That is, for example we have used four variables for constructing the dimension ‘‘Eco-
nomic Security’’. We put a positive sign in the variables ‘‘health expenditure’’, ‘‘poverty rate
over 65s’’ and ‘‘unemployment rate’’, and we put a negative sign in the variable ‘‘unem-
ployment benefit coverage’’. Then we added the four adjusted values (rescaled at the interval
[0,1]) with the mentioned signs. Finally, we put a negative sign to the whole dimension to
calculate the well-being index. Therefore, we are assuming that ‘‘health expenditure’’,
‘‘poverty rate over 65s’’ and ‘‘unemployment rate’’ has an obvious negative effect on the
well-being index while ‘‘unemployment benefit coverage’’ has the opposite effect.
In the case of the dimension ‘‘Equity’’, the procedure is the equivalent of that. We put
positive signs in the variables ‘‘Gini index’’ and ‘‘FGT1 index’’ but the whole dimension
has a negative sign.
2.2 Estimation of the Well-Being Index by Factor Analysis
An alternative methodology to that proposed above consists of the construction of an
indicator of the level of well-being of the distinct Autonomous Communities using mul-
tivariate statistical techniques. In the specialised literature, such models are known as latent
variable models, and are differentiated by the measurement scale of the observed variables
and that assumed for the non-observed variable, as Table 2 shows.
When constructing a metric well-being index based on variables of the same type, factor
analysis is the technique selected for the estimation. It is very similar to the standard
regression model, with the sole difference that some of its variables are unknown. Since it
is impossible to establish their value, this relationship is analysed indirectly.
Let p denote the observed variables and q the latent factors or variables. Thus, the
general model is expressed as:
xi ¼ ai0 þ ai1y1 þ ai2y2 þ � � � þ aiqyq þ ei; i ¼ 1; . . .; p ð2Þ
where y1, y2, …,yq are the factors, ei the residuals and ai0, ai1, …,aiq the factor loads. From
among the assumptions of the model, it should be emphasised that the factors are estimated
as standardised variables,12 i.e. with zero mean and unitary variance. Generally, when
applying this analysis the number of underlying factors is unknown. To decide the structure
of the factors it is necessary to examine the proportion of the variance they each explain.13
However, while Eq. 2 expresses the observed variables as linear combinations of the
factors, the objective of this study involves the inversion of this relationship, that is to say
the expression of the factors as linear combinations of the observed variables.
It can be demonstrated that Eq. 2 can be transformed into another model, given in
expression (3).14
yi ¼ ci1x1 þ ci2x2 þ � � � þ cipxp; i ¼ 1; . . .; q ð3Þ
This system of equations permits the estimation of well-being based on the information
contained in the set of variables selected. Before presenting the results, some comments
must be made regarding the application performed. Firstly, as the different dimensions of
12 Similarly, the indicators have been standardised to eliminate the possible effects of the distinct units ofmeasurement units.13 A common criterion in the literature consists of revising the eigenvalues of the correlation matrix andselecting those components whose eigenvalue is greater than one.14 Although the punctuations cij must be estimated, if the principal components model is chosen to extractthe factors (as in our case), the punctuations c are related with the coefficients a of Eq. 2.
Construction and Evolution of a Multidimensional Well-Being Index 265
123
well-being have been established previously based on economic theory, it is not necessary
to explore further and decide how many factors exist. Additionally, in order to avoid
possible interference from some variables in other dimensions different to those to which
they are assigned, a dual process of estimation is undertaken:15 firstly, an index for each
dimension and, secondly, the joint well-being index from these indices.
Adjusted consumption = 0.3293*AdjPubCon + 0.3257*AdjMonExp
þ 0:4547 � AdjNonMonExp � 0.2719*HoursWork
Real Wealth = 0.3863*R&D Exp + 0.2757*Phys Cap + 0.3714*Human Cap
þ 0:2034 � Env Exp ð4ÞEquity = 0.5516*FGT1 + 0.5516*Gini
Security ¼� 0:0059 � Health + 0.1813*Pop over 65 + 0.5201*Unempl
� 0:5060 � Unemp Cov
Having estimated the partial indicators, the next step is to determine the model that permits
us to calculate the index of well-being for each of the regions, as shown by the following
expression.
Well-being = 0:3655 � Adjusted consumptionþ 0:4044 �Wealth � 0:2273
� Equity � 0:3129 � Security ð5Þ
Since a different method has been used, the coefficients of the model are not directly
comparable with the weightings proposed in the previous section.16 Nevertheless, the
analysis of their signs and values provides information concerning the relative importance
of each of the aspects considered. Thus, it is demonstrated that factors such as real wealth
or adjusted consumption are of great importance in the economic well-being of the Spanish
regions. This divergence between the relative importance of each indicator, depending on
whether use is made of the uniform weightings determined a priori or those estimated by
statistical analysis, may facilitate a sensitivity analysis of the uniform weightings. On this
point, the data appear to contradict the alternative proposal made by Osberg and Sharpe,
consisting of increasing the weight of adjusted consumption and reducing that of real
wealth.
These disparities with regard to the uniform distribution of weights are greater, and
more interesting, within each dimension. For example, R&D expenditure or the stock of
human capital are the most important indicators within the most significant dimension,
Table 2 Latent variable models
Observed variables
Latent variables Continuous Categorical
Continuous Factor analysis Latent feature models
Categorical Latent profile models Latent class models
Source Bartholomew et al. (2002)
15 It must be underlined that what is determined is not a structure of weightings, but instead the effect ofeach variable upon the factor.16 It can be shown that the coefficients of Eq. 5 do not sum to unity.
266 A. Jurado, J. Perez-Mayo
123
‘‘real wealth’’. They are assumed to be, therefore, two key factors in the explanation of
regional differences in economic well-being.
2.3 Estimation by Efficiency Analysis
Efficiency analysis has principally been employed in the analysis of the production
function, to estimate how productive units maximise their output from a set of inputs or,
alternatively, how they minimise the inputs used for a set of outputs. There exist few
applications of this technique to the study of well-being or the standard of living, although
among the studies present in the literature mention should be made of Lovell et al. (1994),
Ramos and Silber (2005) or Ramos (2008). In contrast to these articles, in which para-
metric distance functions are employed to determine the weights of each indicator and to
determine the joint index of well-being, the present study uses the method of Data
Envelopment Analysis (DEA) to summarise the information contained in the indicators, to
calculate the weightings for each variable and, finally, to estimate the well-being index.
DEA was proposed by Charnes et al. (1978)17 to obtain an envelopment formed by all
the efficient units and leaving the remaining units under it. The measurement of efficiency
provided by DEA is relative, because each unit is compared with the rest. Consequently,
the data available determine the form and situation of the production boundary. To obtain
this envelopment, it is necessary to resolve a problem of mathematical optimisation for
each unit (in this case, Autonomous Community). The solution of this problem permits an
efficiency index to be assigned to them.
Confronting this problem as intuitively as possible, it is possible to formulate a frac-
tional programme, in which a maximisation or minimisation is performed (oriented to
output or oriented to input) of a total productivity ratio for each unit evaluated:18
Max h0ðu; vÞ ¼Ps
r¼1 uryr0Pmi¼1 vixi0
under
Psr¼1 uryrjPmi¼1 vixij
� 1; vi; ur � 0
j ¼ 1; . . .; n r ¼ 1; . . .; s i ¼ 1; . . .;m
ð6Þ
where h0 is the measure of efficiency of unit 0, yr0 is the quantity of output produced by the
unit, xi0 is the quantity of input i used by the unit, yrj; xij are the outputs and inputs of unit j
and vi; ur � 0 are the weightings determined by the solution to the problem.
This problem may be transformed into a problem of linear programming in order to
resolve it better. To achieve this, it is sufficient to maximise the numerator of the objective
function while keeping the denominator constant:
Max h0ðuÞ ¼Xs
r¼1
uryr0
underXm
i¼1
vixi0 ¼1;Xs
r¼1
uryrj �Xm
i¼1
vixij� 0; vi; ur � 0
j ¼ 1; . . .; n r ¼ 1; . . .; s i ¼ 1; . . .;m
ð7Þ
The linear programme selects the weights which maximise the virtual output of the unit
17 This model is named CCR, after the names of its authors.18 The problem oriented to output only presents itself when using this formulation in the empiricalapplication.
Construction and Evolution of a Multidimensional Well-Being Index 267
123
ðuryr0Þ, given a unitary virtual input ðvixi0Þ. Furthermore, the virtual output can never
exceed the virtual input. Nevertheless, it is easier to calculate the efficiency indices if the
dual programme is used:
Max h0
underXn
j¼1
kjyrj� hyr0;Xn
j¼1
kjxij� xi0;Xn
j¼1
eij� xi0; kj� 0
j ¼ 1; . . .; n r ¼ 1; . . .; s i ¼ 1; . . .;m
ð8Þ
where eij is a vector comprised of ones.
3 Economic Well-Being Indices for the Year 2000
3.1 Results from the Application of Uniform Weightings
Observing the overall well-being index (Table 3), there are clearly two regions at the top
(Navarre and the Basque Country) and three at the bottom (Extremadura, Andalusia and
the Canary Islands). However, attending to the dispersion (mean ± standard deviation), it
is possible to create three groups of regions. In the group number 1 (well-being index over
mean ? s.d.) the Basque Country, Navarre, Madrid and Catalonia are, in the group number
2 (well-being index between mean - s.d. and mean ? s.d.) Castile-La Mancha, Balearic
Islands, Aragon, Asturias, La Rioja, Valencian Community, Castile and Leon, Murcia,
Galicia, Cantabria and Canary Islands are; and, finally, in the group number 3 (well-being
index under mean - s.d.) Andalusia and Extremadura are included.
Table 4 shows the values of the four components of the well-being index. It must be
remembered that a greater magnitude of ‘‘adjusted consumption’’ or of ‘‘real wealth’’
represents greater well-being, while the opposite occurs with the other two components. In
Table 3 Well-being index 2000PPP
Source Authors’ elaboration.Ranks appear in brackets
CC.AA Well-being
Andalusia (16) -0.2919
Aragon (7) 0.0044
Asturias (8) -0.0392
Balearic Islands (6) 0.0207
Basque Country (1) 0.2302
Canary Islands (15) -0.1878
Cantabria (14) -0.1701
Castile and Leon (11) -0.0875
Castile-La Mancha (5) 0.0306
Catalonia (4) 0.1219
Extremadura (17) -0.3055
Galicia (13) -0.1097
Madrid (3) 0.1504
Murcia (12) -0.0984
Navarre (2) 0.1744
Rioja (9) -0.0397
Valencian community (10) -0.0444
268 A. Jurado, J. Perez-Mayo
123
other words, the greater the value of ‘‘adjusted consumption’’ or ‘‘real wealth’’, and the
lower that of ‘‘equity’’ and ‘‘economic security’’, the greater will be the index of well-being
obtained by that region. Following the same logic, negative values of ‘‘economic security’’
(because of high relative values in unemployment coverage) increase it.
Table 4 Components ofwell-being index 2000 PPP(uniform weightings)
Source Authors’ elaboration
CC.AA Adjustedconsumption
Realwealth
Equity Economicsecurity
Andalusia 0.1833 0.1409 0.9410 0.5512
Aragon 0.3994 0.334 0.6908 0.0783
Asturias 0.1284 0.2887 0.4703 0.1268
Balearic Islands 0.1625 0.2307 0.3922 -0.0408
Basque Country 0.5913 0.2443 0.1532 0.1926
Canary Islands 0.3836 0.3602 0.9291 0.3198
Cantabria 0.2391 0.3548 0.8572 0.3055
Castile and Leon 0.2940 0.4123 0.7080 0.2685
Castile-La Mancha 0.0445 0.2993 0.0000 0.1334
Catalonia 0.3684 0.1151 0.3271 0.1089
Extremadura 0.0452 0.2032 0.6922 0.5878
Galicia 0.0297 0.3522 0.3745 0.2408
Madrid 0.5998 0.2234 0.5508 0.1918
Murcia 0.1251 0.1607 0.4126 0.3750
Navarre 0.4228 0.3514 0.2032 0.0619
Rioja 0.2802 0.2866 0.4065 0.3725
Valencian community 0.0578 0.1433 0.4222 0.1764
Table 5 Well-being indexcomponents 2000 PPP(factor analysis weightings)
Source Authors’ elaboration
CC.AA Adjustedconsumption
Realwealth
Equity Economicsecurity
Andalusia -0.4682 -0.8562 1.6905 1.9279
Aragon 0.7496 0.1545 0.7301 -0.6474
Asturias -0.6798 -0.2253 -0.1125 0.3020
Balearic Islands -0.4027 -0.4242 -0.4256 -1.9025
Basque Country 1.7455 1.6064 -1.3302 0.1546
Canary Islands 0.5566 -0.9665 1.6061 -0.4426
Cantabria -0.0508 -0.5436 1.3435 0.5681
Castile and Leon 0.0876 -0.0778 0.8087 0.5265
Castile-La Mancha -1.1614 -0.7176 -1.9469 -0.1579
Catalonia 0.5729 1.0601 -0.6253 -0.7860
Extremadura -1.4135 -1.4566 0.6841 2.1546
Galicia -1.0499 -0.8837 -0.4767 0.1060
Madrid 2.1674 2.5691 0.2655 0.0106
Murcia -0.8217 -0.3135 -0.4341 0.4470
Navarre 1.0399 0.9735 -1.1182 -1.5974
Rioja 0.1197 0.0105 -0.3746 -0.5135
Valencian community -0.9909 0.0910 -0.2845 -0.1499
Construction and Evolution of a Multidimensional Well-Being Index 269
123
In the case of the Basque Country, its extremely high adjusted consumption, equalled
only by Madrid, together with the best data regarding equity after Castile-La Mancha,
launches it to top position because of a regional per capita budget far above the values of
the remaining Communities. It must be underlined that Navarre and the Basque Country
have in force systems of Economic Contract and Economic Agreement, respectively,
which grant great financial and fiscal autonomy to the Foral Authorities. They manage the
greater part of taxes and contribute to general State revenue for the competencies not
assumed, via the payment of a quota or contribution established in the Contract/Agreement
with the State.
Third on the economic well-being index is Madrid. In this region, the highest values in
Spain for expenditure (monetary and non-monetary), expenditure on R&D, and human
capital are softened by lower than average values regarding environmental expenditure per
capita and, above all, regarding inequality, where it occupies the penultimate position.
Extremadura, Andalusia and the Canary Islands occupy the three lowest positions, using
practically whatever methodological option. The very high figures for inequity in Anda-
lusia and the Canaries and the low level of consumption and wealth in Extremadura are the
principal determinants of the lowest positions of these three Communities.
As explained in the introduction, the indicator most commonly employed to quantify
standards of living or well-being in distinct territorial units has been GDP per capita or an
equivalent macromagnitude. Figure 1 compares the distribution produced by our multi-
dimensional well-being index and the ranking that would result from using GDP per capita
(transformed to a 0–1 scale for an easier analysis). Sharp differences are apparent in some
Communities, which in our view confirm the need to construct an index such as that
proposed here. There are notable differences shown by Castile-La Mancha, for its
important improvements in equity in recent years, or by cases to the contrary such as La
Rioja, the Canary Islands, Cantabria or the Balearic Islands.
3.2 Results of the Factor Analysis
Despite the fact that when examining the extremes of the dimensions we find, in general,
the same regions, a common pattern among the Autonomous Communities is hard to find.
Fig. 1 Well-being index and GDP per capita 2000. Source Authors’ elaboration
270 A. Jurado, J. Perez-Mayo
123
Concretely, Madrid, the Basque Country and Navarre show the most favourable situations
in ‘‘Adjusted consumption’’; Madrid, the Basque Country and Catalonia in ‘‘Real Wealth’’;
Castile-La Mancha is notable in ‘‘Equity’’, followed by the Basque Country and Navarre;
and in ‘‘Economic security’’ the top positions are occupied by the Balearic Islands, Navarre
and Catalonia. The previous statement is confirmed: there is no clear and unequivocal
distribution of the indicators among the Autonomous Communities (Table 5).
At the other extreme of the partial indices, the Canary Islands, Extremadura and
Andalusia are noteworthy as the territories worst situated in the majority of the dimensions
considered.
Extremadura is dragged down due to its levels of consumption and wealth, Andalusia by
its poor figures for equity and security and the Canary Islands by a negative combination of
low figures for wealth and high levels of poverty and inequality (Table 5).
Note should be made of the broad consistency of the results obtained when calculating
the well-being indices with uniform weightings and through factor analysis: two much
differentiated groups can be observed at the extremes of the distributions. Factor analysis
results show relative improvements for Madrid (which moves to top position) and Navarre
and a decline for Castile-La Mancha (Table 6).
On this point, it is essential to remember the divergences, although slight, between the
uniform weightings and the factor analysis methods: the relative importance of adjusted
consumption and real wealth, and within the latter, of expenditure on R&D and of human
capital is greater than that considered in the system of uniform weightings. Thus, in
contrast to the descriptive analysis of the latter system, factor analysis, both in its direct
application and the weightings derived from it, permits the determination of the most
important factors in reducing regional disparities in well-being, with their corresponding
implications for economic policy. In other words, investment in research and development
and improved human capital appear to be key instruments for regional development.
Table 6 Well-being index2000 PPP (factor analysis)
Source Authors’ elaboration.Ranks appear in brackets
CC.AA Well-being
Andalusia (16) -1.5048
Aragon (6) 0.3731
Asturias (11) -0.4085
Balearic Islands (5) 0.3734
Basque Country(2) 1.5415
Canary Islands (12) -0.4139
Cantabria (15) -0.7215
Castile and Leon (10) -0.3480
Castile-La Mancha (9) -0.2227
Catalonia (4) 1.0262
Extremadura (17) -1.9354
Galicia (14) -0.6659
Madrid (1) 1.7674
Murcia (13) -0.4684
Navarre (3) 1.5277
Rioja (7) 0.2938
Valencian community (8) -0.2138
Construction and Evolution of a Multidimensional Well-Being Index 271
123
3.3 Application of the Data Envelopment Analysis
With the objective of reflecting the negative effect of certain variables (i.e. when certain
variables increase, well-being decreases), maintaining at the same time positive values in
them all, we undertake a change of variable, calculating its inverse following the practice
usually applied in the literature.
Although the weightings obtained using this technique vary for each variable and
region, thereby permitting the CC.AA. to emphasise their strong points, the results are very
similar to those previously obtained (Table 8).
With regard to the factor analysis, a decline in the Canary Islands and improvements in
Asturias and Castile-La Mancha are notable. Once more, real wealth proves to be the key
to the improvement of regional well-being (Table 7).
4 Evolution in the Period 2000–2006
The present section aims to fulfil two objectives, by including the calculation of indices for
the year 2006. Firstly, to attempt to confirm the robustness of the results obtained for the
year 2000 and, secondly, to appreciate certain tendencies, despite the short time period
involved.
Although we have constructed similar indicators for earlier years (1980), the method-
ological problems posed by the absence of certain data, the changes of sources and, above
all, the existence of a regional political map extremely diverse with regard to competen-
cies, discouraged us from working with observations prior to the year 2000.
Table 7 Well-being indexcomponents 2000 PPP (DEAanalysis)
Source Authors’ elaboration
CC.AA. Adjustedconsumption
Realwealth
Equity Economicsecurity
Andalusia 90.73 51.7 68.24 63.54
Aragon 94.7 77.12 74.51 99.71
Asturias 87.13 69.27 81.07 100
Balearic Islands 90.52 39.89 83.73 100
Basque Country 100 100 92.84 85.51
Canary Islands 96.01 39.98 68.61 73.82
Cantabria 92.49 60.71 70.29 76.55
Castile and Leon 95.94 70.04 74 81.57
Castile-La Mancha 84.89 58.36 100 90.06
Catalonia 94.6 93.57 85.84 89.88
Extremadura 82.86 29.96 74.63 64.46
Galicia 85.52 53.19 84.28 79.8
Madrid 100 93.98 78.3 84.59
Murcia 86.8 67.68 83.37 71.45
Navarre 96.15 96.46 90.67 100
Rioja 91.84 72.93 83.25 77.17
Valencian community 87.61 74.77 82.61 86.15
272 A. Jurado, J. Perez-Mayo
123
Tables 9 and 10 show the components and indices for the year 2006 according to the
three methodologies previously used. Without being exhaustive, two points with regard to
the time trend should be noted.
Table 8 Well-being index2000 PPP (DEA analysis)
Source Authors’ elaboration.Ranks appear in brackets
CC.AA. Well-being
Andalusia (16) 80.93
Aragon (6) 95.21
Asturias (7) 94.21
Balearic Islands (12) 86.48
Basque Country (1) 100
Canary Islands (13) 81.28
Cantabria (15) 85.81
Castile and Leon (8) 90.69
Castile-La Mancha (4) 96.71
Catalonia (5) 96.12
Extremadura (17) 69.66
Galicia (14) 84.63
Madrid (3) 97.46
Murcia (11) 86.05
Navarre (1) 100
Rioja (9) 89.16
Valencian community (10) 88.76
Table 9 Well-being index 2006 PPP
CC.AA. Uniform weightings Factor analysis DEA analysis
Andalusia -0.3487 (17) -1.7224 (17) 81.73 (15)
Aragon 0.0454 (6) 0.5347 (5) 93.39 (7)
Asturias 0.0607 (5) 0.4342 (7) 97.02 (3)
Balearic Islands 0.0742 (4) 0.4378 (6) 89.82 (10)
Basque Country 0.1965 (2) 1.4281 (3) 98.94 (2)
Canary Islands -0.2237 (14) -1.0495 (14) 69.66 (17)
Cantabria 0.0328 (8) 0.3433 (8) 96.25 (4)
Castile and Leon 0.0353 (7) 0.3260 (9) 93.90 (6)
Castile-La Mancha -0.1743 (13) -0.8217 (13) 84.49 (13)
Catalonia 0.0067 (9) 0.6078 (4) 92.74 (8)
Extremadura -0.2852 (15) -1.4069 (15) 80.24 (16)
Galicia -0.1144 (11) -0.2700 (10) 88.02 (11)
Madrid 0.1355 (3) 1.5263 (2) 94.20 (5)
Murcia -0.3050 (16) -1.4171 (16) 83.39 (14)
Navarre 0.2435 (1) 1.7627 (1) 100.00 (1)
Rioja -0.0847 (10) -0.3058 (11) 89.87 (9)
Valencian community -0.1412 (12) -0.4076 (12) 87.10 (12)
Source Authors’ elaboration. Ranks appear in brackets
Construction and Evolution of a Multidimensional Well-Being Index 273
123
Tab
le1
0C
om
po
nen
tso
fw
ell-
bei
ng
ind
ex2
00
6P
PP
CC
.AA
Un
ifo
rmw
eig
hti
ng
sF
acto
ran
aly
sis
DE
Aan
aly
sis
Ad
j.co
nsu
mp.
Rea
lw
ealt
hE
qu
ity
Eco
no
mic
secu
rity
Ad
j.co
nsu
mp
Rea
lw
ealt
hE
qu
ity
Eco
nom
icse
curi
tyA
dj.
con
sum
pR
eal
wea
lth
Eq
uit
yE
con
om
icse
curi
ty
An
dal
usi
a0
.022
10
.15
93
0.9
85
10
.590
9-
0.8
81
9-
0.9
73
51
.86
12
1.8
64
39
2.0
45
2.2
37
5.1
96
6.2
3
Ara
go
n0
.252
70
.38
07
0.3
21
20
.130
40
.445
9-
0.0
12
7-
0.6
54
8-
0.7
28
79
6.2
67
1.2
29
4.3
91
.74
Ast
uri
as0
.276
40
.51
00
0.3
15
60
.227
80
.368
10
.42
98
-0
.67
31
0.0
86
69
6.6
61
00
94
.57
81
.13
Bal
eari
cIs
lands
0.0
556
0.3
119
0.2
677
-0
.197
0-
0.5
50
4-
0.4
18
1-
0.8
50
7-
1.9
64
39
2.1
49
.34
96
.32
10
0
Bas
qu
eC
ou
ntr
y0
.487
60
.69
72
0.2
09
70
.189
21
.369
91
.56
23
-1
.07
36
-0
.16
50
10
01
00
98
.54
82
.82
Can
ary
Isla
nds
0.1
08
00
.15
94
0.7
44
50
.417
7-
0.4
07
9-
0.9
31
30
.96
21
0.9
74
99
3.5
92
2.9
81
.38
71
.27
Can
tab
ria
0.0
74
00
.46
08
0.2
94
40
.109
1-
0.5
80
60
.34
90
-0
.74
74
-0
.78
15
93
.11
77
.86
95
.32
10
0
Cas
tile
and
Leo
n0
.382
80
.36
43
0.3
38
30
.267
51
.039
3-
0.1
27
9-
0.5
82
80
.43
00
10
07
0.9
59
3.8
57
7.5
8
Cas
tile
-La
Man
cha
0.0
524
0.1
884
0.5
575
0.3
804
-0
.364
8-
0.9
58
10
.24
43
0.7
84
19
3.3
75
1.3
18
6.5
57
3.7
Cat
alonia
0.1
856
0.5
187
0.4
129
0.2
645
0.0
076
0.6
563
-0
.30
38
-0
.86
48
94
.77
82
.02
91
.18
83
.25
Ex
trem
adu
ra0
.170
40
.04
73
0.8
51
00
.507
4-
0.0
74
6-
1.4
63
11
.35
57
1.5
33
49
4.9
64
0.1
17
8.4
56
9.2
3
Gal
icia
0.1
019
0.2
325
0.5
046
0.2
876
-0
.288
0-
0.7
04
10
.03
97
-0
.41
21
93
.72
63
.69
88
.11
76
.38
Mad
rid
0.4
29
70
.73
22
0.5
52
80
.067
01
.953
41
.80
45
0.2
44
6-
0.4
41
71
00
67
.78
86
.91
90
.5
Mu
rcia
-0
.009
40
.20
50
0.9
66
00
.449
4-
0.9
71
7-
0.8
34
11
.78
99
1.0
15
99
1.5
85
7.3
77
5.6
57
0.9
6
Nav
arre
0.3
63
80
.75
26
0.1
77
9-
0.0
35
40
.871
91
.70
48
-1
.18
96
-1
.54
78
98
.36
10
01
00
10
0
Rio
ja0
.170
40
.04
73
0.8
51
00
.507
4-
1.5
59
30
.11
63
-0
.67
09
-0
.20
64
90
.43
74
.65
93
.94
80
.1
Val
enci
anC
om
m.
0.0
57
00
.34
09
0.5
58
60
.404
3-
0.3
76
7-
0.2
00
20
.24
93
0.4
22
89
2.6
66
7.5
86
.54
70
.74
So
urc
eA
uth
ors
’el
abora
tio
n
274 A. Jurado, J. Perez-Mayo
123
– Strongly heterogeneous evolutions in the period: Regions which improve in 2006 can
be found using any of the methodologies (Asturias, Cantabria, Castile and Leon,
Extremadura, Galicia and Navarre), others decline using any of them (Castile-La
Mancha, Catalonia, the Canary Islands, the Valencian Community, Madrid, Murcia and
the Basque Country), and the sign of the four remaining regions changes with the
method employed.
– In general, although the use of various techniques may generate a degree of initial
confusion, it also helps to provide robustness to certain conclusions. One example is
that the group of regions19 with greatest well-being, independently of the technique
employed, is comprised of the Basque Country, Navarre and Madrid. One-step below is
Catalonia, slightly above a large mean group (the Balearic Islands, Castile-La Mancha,
Asturias, Aragon, Castile and Leon, the Valencian Community). At the bottom of the
classification are the Canary Islands, Andalusia, Murcia and, lastly, Extremadura. In
this last case, it should be noted that Extremadura, following the adjustment for PPP,
leaves the final position. This fact is due to the lowest relative prices in Spain allow it to
exceed in well-being Andalusia and Murcia or Canary Islands in 2006 (depending on
methodology).
In general, greater parallelism can be observed among the results from uniform
weightings and the application of DEA than between either of them and the factor analysis.
5 Conclusions
The present study offers an attempt to construct a multidimensional index of economic
well-being for the Spanish Autonomous Communities. Given the high degree of decen-
tralisation of the nation and the need to have available instruments for the evaluation and
Table 11 Groups of regions according well-being index 2006 PPP
Well-beinglevel
Uniform weightings Factor analysis DEA analysis
Low Andalusia, Canary Islands,Extremadura, Murcia
Andalusia, Canary Islands,Extremadura, Murcia
Andalusia, Canary Islands,Extremadura
Medium Asturias, Aragon, BalearicIslands, Castile and Leon,Castile-La Mancha,Catalonia, ValencianCommunity
Asturias, Aragon, BalearicIslands, Castile and Leon,Castile-La Mancha,Catalonia, ValencianCommunity
Asturias, Aragon, BalearicIslands, Castile and Leon,Castile-La Mancha,Catalonia, Madrid, Murcia,Valencian Community
High Basque Country, Madrid andNavarre
Basque Country, Madrid andNavarre
Basque Country and Navarre
Source Authors’ elaboration
19 In Table 11, regions are classified according to their relative well-being levels according to the previouscriteria: low class is defined as well-being index under mean - s.d., high class as well-being index overmean ? s.d. and medium class contains the remaining cases (well-being index between mean – s.d. andmean ? s.d.).
Construction and Evolution of a Multidimensional Well-Being Index 275
123
tracking of national public policies, we believe it interesting to be able to evaluate and
compare the effects of such policies upon the well-being of citizens.
When it comes to measuring this well-being, we have attempted to overcome the
limitations of the most commonly used indicator, Gross Domestic Product per capita, by
adding factors such as the aggregate accumulation of productive stocks and the hetero-
geneity of individuals, both in the present (the distribution of potential consumption, that is
to say poverty and inequality) and the future (taking into consideration the insecurity of
future income).
In the construction of economic well-being indices for the comparison of territorial
units, there are two aspects in which arbitrariness is unavoidable: on one hand, in the
selection of the multidimensional indicators chosen to configure the index, and on the
other, in the way in which these indicators are weighted. In the first case, we have
attempted to reduce this arbitrariness by basing ourselves on a dimensional structure
created by Osberg and Sharpe and internationally recognised by many prestigious publi-
cations. In the second case, the use of three different, and we believe adequate, method-
ologies permits the results to be strengthened, although it limits the possibility of obtaining
a large number of quantifiable conclusions.
The empirical application of the methodologies proposed reveals great differences in the
national panorama. It can be observed that two groups appear at the extremes of the
distribution of the aggregate index, and this is repeated in the majority of the dimensions
considered.
Navarre, the Basque Country, Madrid and, at a certain distance, Catalonia, are notable
in the leading group, while Extremadura, Andalusia and the Canary Islands display the
worst results. When attempting to minimise arbitrariness in the weighting of variables by
using three different weighting systems, there appear changes of a certain importance for
some regions; however, the principal features of the regional ranking of well-being are
maintained, with few exceptions.
To conclude, we would like to make some comments regarding possible extensions of
this study. It would be useful to collect certain additional variables, such as hours of leisure
(or to adjust consumption by such hours), expenditure derived from travelling to work and
the level of durable goods, and better data for human and social capital, as well as adding
to the regional budget the expenditure of the regionalised Central Administration or some
information on the minimum incomes existing in each Autonomous Community. Presently,
the restricted availability of data is the principal obstacle to the majority of such advances.
We are aware that the construction of an index of this type will always be improvable
and permanently exposed to criticisms, which aim to contribute ideas. Nevertheless, we
consider that its arbitrary construction and considerable data restrictions in the regional
ambit must not discourage research aimed at filling, even partially, a lacuna of increasing
interest.
Acknonwledgments The authors would like to thank the Extremadura Government (Junta de Extrema-dura PRI08A137) and European Union Funds FEDER for the funding they have received. They are verygrateful for the useful referees’ comments.
Annex 1: A Brief Introduction to the Political Division in Spain
An autonomous community (A.C.) is the first-level political division of the Kingdom of
Spain, established in accordance with the Spanish Constitution. The Central Government
and 17 autonomous communities share political power in Spain. There are also two
276 A. Jurado, J. Perez-Mayo
123
autonomous cities, Ceuta and Melilla, situated in the African continent, though the lack of
significance of its samples and its particular social and economic situation make difficult to
include them in Spanish regional studies. We have decided, like most of the Spanish
regional researchers, not to include these two autonomous cities in this analysis.
Since 1978, a wide and fast decentralization process has been developed from the
Central Government to the Regional Authorities. Nowadays, almost all the policies have
been transferred to them, excepting some common areas as foreign policy, national
defense, pensions or the main fiscal taxes as VAT or Corporation Tax (the income tax is
divided into a national and regional shares). Moreover, Regional Governments and Par-
liaments are the main subjects in issues as education, health, culture, housing or social
exclusion.
Although, this decentralization has improved the efficiency of the regional spending in
the sense of a better knowledge of the people’s preferences, different problems related to
cohesion and solidarity between regions are becoming discussion topics.
The main characteristics of the Spanish autonomous communities are shown in
Table 12.
Table 12 Main characteristics of Spanish regions
GDP perhead2006 (%)
Unemploymentrate (%)2006
Activityrate2006
Population1-1-2007
%pop.(%)
Immigrants2007
% immigrant(%)
Andalucıa 77.5 12.68 55.32 8059.5 17.8 555.8 6.9
Aragon 107.2 5.54 56.59 1296.7 2.9 110.0 8.5
Asturias 90.5 9.31 49.85 1074.9 2.4 48.1 4.5
Balearic Islands 109.9 6.46 64.11 1030.7 2.3 180.4 17.5
Basque Country 128.5 6.97 58.07 2141.9 4.7 100.7 4.7
Canary Islands 89.2 11.68 61.02 2026.0 4.5 276.8 13.7
Cantabria 98.8 6.56 55.32 572.8 1.3 30.1 5.2
Castile and Leon 95.1 8.11 53.09 2528.4 5.6 121.8 4.8
Castile-LaMancha
77.7 8.81 55.03 1977.3 4.4 135.8 6.9
Catalonia 118.0 6.6 62.17 7210.5 16.0 923.2 12.8
Extremadura 67.9 13.43 51.58 1090.0 2.4 27.9 2.6
Galicia 82.9 8.48 53.61 2772.5 6.1 152.4 5.5
Madrid 130.7 6.37 63.57 6081.7 13.5 882.3 14.5
Murcia 83.6 7.85 58.98 1392.1 3.1 184.5 13.3
Navarre 125.5 5.3 60.67 605.9 1.3 51.1 8.4
Rioja 107.1 6.18 59.48 309.0 0.7 34.8 11.3
ValencianCommunity
91.7 8.37 59.62 4885.0 10.8 691.3 14.2
Ceuta(auton. city)
91.1 20.96 52.79 76.6 0.2 6.5 8.5
Melilla(auton. city)
90.4 13.38 52.5 69.4 0.2 13.2 19.1
Total Spain 100.0 8.51 58.32 45200.7 100.0 4526.5 10.0
Source INE (Spanish National Statistical Institute). Population and inmigrants are in thousands
Construction and Evolution of a Multidimensional Well-Being Index 277
123
References
Alcaide, J., & Alcaide, P. (2008). Balance economico regional 2000–2007. Madrid: FUNCAS.Anielski, M., & Rowe, J. (1999). The genuine progress indicator 1998 update: Long-term environmental
damage. San Francisco: Redefining Press.Ayala, L., Jurado, A., & Pedraja, F. (2006). Desigualdad y bienestar en la distribucion intraterritorial de la
renta, 1973–2000. Investigaciones Regionales, 8, 5–30.Ayala, L., Jurado, A. & Perez-Mayo, J. (2011). Income poverty and multidimensional deprivation: Lessons
from cross-regional analysis. Review of Income and Wealth, 57(1), 40–60.Bartholomew, D. J., Steele, F., Moustaki, I., & Galbraith, J. I. (2002). The analysis of interpretation of
multivariate data for social scientists. Boca Raton: Chapman & Hall/CRC.Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units.
European Journal of Operational Research, 2, 429–444.Cobb, C. W., & Cobb, J. B. (1994). The green national product: A proposed index of sustainable economic
welfare. Lanham, MD: University Press of America.Cobb, C. W., Halstead, T., & Rowe, J. (1995). The genuine progress indicator: Summary of data and
methodology. San Francisco: Redefining Progress.Daly, H. E., & Cobb, J. B., Jr. (1989). For the common good. Redirecting the economy toward community,
the environment and a sustainable future. Boston: Beacon Press.Diener, E., Diener, M., & Diener, C. (1995). Factors predicting the subjective well-being of nations. Journal
of Personality and Social Psychology, 69, 851–864.Garcıa-Luque, O., Lafuente, M., & Faura, U. (2009). Disparidad territorial de la pobreza dinamica en
Espana. Estudios de Economıa Aplicada, 27–2, 417–436.Jackson, T. (2004). Chasing progress: Beyond measuring economic growth. London: New Economic
Foundation.Khan, H. (1991). Measurement and determinants of socioeconomic development: a critical conspectus.
Social Indicators Research, 24, 153–175.Lelkes, O. (2005). Knowing what is good for you: Empirical analysis of personal preferences and the
‘‘objective good’’, CASE paper 94.Lovell, C. A. K., Richardson, S., Travers, P., & Wood, L. (1994). Resources and functionings: A new view
of inequality in Australia. In W. Eichhorn (Ed.), Models and measurement of welfare and inequality(pp. 787–807). Heidelberg: Springer-Verlag.
Marchante, A., Ortega, B. & Sanchez J. (2006). Las dimensiones del bienestar en las comunidades auto-nomas espanolas. Un analisis de sigma y gamma—convergencia. Documento de Trabajo: FEDEA.
Murias, P., Martınez, F., & De Miguel, C. (2006). An economic well-being index for the Spanish provinces:A data envelopment analysis approach. Social Indicator Research, 77, 395–417.
Nordhaus, W. D., & Tobin, J. (1972). Is growth obsolete? In Fiftieth Anniversary Colloquium V, NationalBureau of Economic Research, New york: Columbia University Press.
Osberg, L., & Sharpe, A. (2000). Estimates of an index of economic well-being for OECD countries. In 26thGeneral conference of the international association for research in income and wealth. Poland.
Osberg, L. & Sharpe, A. (2003) Human well-being and economic well-being: What values are implicit incurrent indices?. Centre for the Studying of Living Standards, Research Report.
Osberg, L., & Sharpe, A. (2005). How should we measure the ‘‘economic’’ aspects of well-being? Review ofIncome and Wealth, 51(2), 311–336.
Osberg, L. & Sharpe, A. (2009). New estimates of the index of economic well-being for selected OECDcountries, 1980–2007. Centre for the Study of Living Standards (CSLS), Research Report 2009–11.
Perez-Mayo, J. (2008). La dimension territorial de la pobreza y la privacion en Espana. FundacionAlternativas.
Poggi, A. (2007). Does persistence of social exclusion exist in Spain? Journal of Economic Inequality, 5(1),53–72.
Ramos, X. (2008). The econometric approach to efficiency analysis. In N. Kakwani & J. Silber (Eds.),Quantitative approaches to multidimensional poverty measurement (pp. 155–177). London: Palgrave-MacMillan.
Ramos, X., & Silber, J. (2005). On the application of efficiency analysis to the study of the dimensions ofhuman development. Review of Income and Wealth, 51(2), 285–309.
Rojas, M. (2007). The complexity of well-being: A life-satisfaction conception and a domains-of-lifeapproach. In I. Gough, & A. McGregor (Ed.), Wellbeing in developing studies. From Theory toResearch. Cambridge (UK): Cambridge University Press.
Sen, A. (2001). El nivel de vida. Madrid: Editorial Complutense.
278 A. Jurado, J. Perez-Mayo
123
Stewart, K. (2002). Measuring well-being and exclusion in Europe’s regions. CASE paper 53, LondonSchool of Economics.
Stewart, K. (2005). Dimensions of well-being in EU regions: Do GDP and unemployment tell us all we needto know? Social Indicators Research, 73, 221–246.
Stiglitz, J., Sen, A. K. & Fitoussi, J. P. (2009). Report of the Comission on the measurement of economicperformance and social progress. available in. http://www.stiglitz-sen-fitoussi.fr/documents/rapport_anglais.pdf.
UNDP. (2004). Human development report. New York, USA: United Nations Development Programme,available in. http://hdr.undp.org/en/media/hdr04_complete.pdf.
Uriel, E., & Barberan, R. (2007). Las balanzas fiscales de las comunidades autonomas con la administracionPublica central, 1991–2005. Bilbao: Fundacion BBVA.
Wolff, E. N., Zacharias, A., & Caner, A. (2005). Household wealth, public consumption and economic well-being in the United States. Cambridge Journal of Economics, 29(6), 1073–1090.
Zolotas, X. (1981). Economic growth and declining social welfare. Athens: Bank of Greece.
Construction and Evolution of a Multidimensional Well-Being Index 279
123