The study of the territorial/regional development in Spain has nowadays a relatively long
Tuesday, April 28, 2009
1
Urban growth and territorial dynamics in Spain (1985-2001): A spatial
econometrics analysis
*
José M. Mella-Márquez
jose.mella@uam.es
Coro Chasco-Yrigoyen
coro.chasco@uam.es
Universidad Autónoma de Madrid
ABSTRACT
The study of the territorial/regional development in Spain has nowadays a relatively long
tradition, but from the point of view of cities development the number of studies and
documents decreases drastically. This paper tries to improve the knowledge of the Spanish
urban system. The aim of this paper is twofold: firstly, to determine the factors that explain
the urban growth of Spanish cities; secondly, to observe the cities situation in terms of
“winners” and “losers” after the long period of integration of Spain in the EU. A spatial
conditional β-convergence equation is specified and the Durbin-Wu-Haussman exogeneity
test is used to check on the existence of simultaneity between urban growth and the control
variables. The classic problems of spread and backwash are studied by including a spatial
autoregressive term and spatial regimes –convergence clubs- in the growth model.
Key words: urban growth, Spanish cities, conditional β-convergence, endogeneity, Durbin-
Wu-Haussman, spatial autocorrelation, spatial heterogeneity, convergence clubs.
JEL classification: C21, C31, C51, D14, O18, R11
*
Research partially supported by the “Instituto de Estudios Fiscales” and “Universidad Autónoma de Madrid”
(Spain). We would like to thank Julie Le Gallo for helpful discussions relating to certain issues considered in
this paper. We also appreciate the collaboration of Esteban Sanromá and Raúl Ramos, who have provided us
with some valuable data. Earlier versions of this paper were presented at the Regional Science Association
International World Congress at the PE Technikon, Port Elisabeth (South Africa) in April, 2004, Regional
Economic Applications Laboratory seminars at the University of Illinois at Urbana-Champaign in August 2004
and the First Seminar in Spatial Econometrics at the University of Zaragoza (Spain) in October 2004. 2
I. INTRODUCTION
The study of the territorial/regional development in Spain has nowadays a relatively
long tradition, especially since the birth in the early eighties of the Autonomous
Communities (“Comunidades Autónomas”) or regions, considered as NUTS II in Eurostat
nomenclature. There are plenty of articles and books written about the Spanish regional
development, and in general they can be considered as a rich economic literature. But when
one looks at the regional development topic from the point of view of cities, there are only a
few documents and the studies are very scarce (Trullén, 2002; Trullén et al., 2002;
Viladecans, 2002; Mayor and López, 2003). This paper tries to improve the knowledge of
the Spanish urban system and boost the urban studies in a country that has experienced a
fast urbanization process within the last four decades. It is evident that at the international
level this topic has received much more attention (in the EU, Cheshire, 2002; in the USA,
Henderson, 1986, 1995; Glaeser et al., 1992, Glaeser, 1998, among many other good
references).
In fact, our intention is to test the same hypothesis formulated by Cheshire (2002):
“The integration of Europe favours the core regions at the expense of the peripheral ones
(…) removing protection as a result of economic integration works to the relative
disadvantage of backward, peripheral regions and favours advanced core regions” (pp. 213).
Several analysis have been presented in the literature (from Clark et al., 1969 to Venables,
1996) that employing very different methodologies, converge on the 19th-century economic
geographer’s conclusion that “the best protection for a backward region is a bad road”,
interpreting “road” broadly as a shorthand for “costs of doing trade”, including all transport
costs and tariff and non-tariff barriers. On the other point, there are two exceptions in
Steinle (1992) and Cheshire and Carbonaro (1996) that reach to the conclusion that the
strongest gains from integration would continue, over time, to spread outwards from the
core to the near periphery: for example, to Catalonia, Valencia, the Ebro valley and to the
south and east of England.
In this paper, we analyse the performance of the 122 main Spanish cities (province
capitals and those with more than 50,000 inhabitants) in terms of per capita GDP. The
analysis period starts just before the Spanish adhesion to the EU in 1985 and ends in the
year 2001, for which we have available data. As a result of this analysis we can also
conclude that, in the group of main Spanish cities, there has been a significant spread from
the core to the near periphery in terms of economic development during the EU integration 3
period. In fact, if in 1985 the core was located in the northern and eastern cities, as well as
in Madrid and some nearby metropolitan towns, by 2001 there has been a shift towards
some near periphery cities, especially in the Castilian ones surrounding Madrid (Avila,
Ciudad Real, Cuenca, Salamanca, Segovia, Valladolid, etc.).
Therefore the aim of this paper is twofold: firstly, to determine the factors that
explain the urban growth of Spanish cities; secondly, to observe the situation of cities in
terms of “winners” and “losers” after a long period of integration of Spain in the EU or at
least how is the pattern of Spanish cities like after a period of Economic and Monetary
Union -Single European Market and Euro- impact. For the former, we estimate a β-
convergence model and spatial effects are tested with spatial econometric techniques. And
for the later, we also use exploratory spatial data analysis (ESDA).
We use the formal tools of spatial econometrics to identify and include the relevant
spatial effects in the estimation of the appropriate income growth model. In effect, in spatial
cross-sectional contexts, it is almost inevitable to test for the presence of spatial spillovers.
Spatial econometrics techniques, including the ESDA, are the appropriate tools to manage
with spatial dependence in the error terms.
The paper proceeds as follows: section II introduces the theoretical framework to
study the factors that explain the urban growth of Spanish cities and provides some insights
into the β-convergence model and spatial effects upon which the empirical estimations
described in the following sections rely. Section III presents the estimation of urban GDP
and exploratory spatial data analysis (ESDA) is used to detect spatial autocorrelation and
spatial heterogeneity among Spanish urban GDP data. In section IV, we present the model
and set of control variables. The estimation, testing and re-specification process of the
appropriate model is detailed in section V, in which these two spatial effects –spatial
autocorrelation and spatial heterogeneity- are included. In section VI, some economic
interpretations of the results are shown and the last section provides some concluding
remarks.
II. THEORETICAL APPROACH
This section introduces the theoretical framework to study the factors that explain the
urban growth of Spanish cities. The function that explains urban evolution in Spain could be
a cross-sectional specification of the neo-classical growth model (Solow, 1956), which is
considered as a natural starting point of the analysis of regional disparities, especially in 4
Europe (Fingleton, 2003). Although debatable, the neoclassical convergence specification
have a sound empirical track record and that is why we considered it could also be useful to
explain urban growth processes. The neo-classical model predicts that the growth rate of a
city is positively related to the distance that separates it from its steady state. That is to say,
if all urban economies are structurally identical and have access to the same technology,
they are characterized by the same steady state, and differ only by their initial conditions:
() ()
, 2
,0
,0
1
ln ln ; . . . 0,
it
ii i
i
y
yiid
ty
ε
α βεε σ
⎛⎞
=+ + ⎜⎟
⎜⎟
⎝⎠
Urban growth and territorial dynamics in Spain (1985-2001): A spatial
econometrics analysis
*
José M. Mella-Márquez
jose.mella@uam.es
Coro Chasco-Yrigoyen
coro.chasco@uam.es
Universidad Autónoma de Madrid
ABSTRACT
The study of the territorial/regional development in Spain has nowadays a relatively long
tradition, but from the point of view of cities development the number of studies and
documents decreases drastically. This paper tries to improve the knowledge of the Spanish
urban system. The aim of this paper is twofold: firstly, to determine the factors that explain
the urban growth of Spanish cities; secondly, to observe the cities situation in terms of
“winners” and “losers” after the long period of integration of Spain in the EU. A spatial
conditional β-convergence equation is specified and the Durbin-Wu-Haussman exogeneity
test is used to check on the existence of simultaneity between urban growth and the control
variables. The classic problems of spread and backwash are studied by including a spatial
autoregressive term and spatial regimes –convergence clubs- in the growth model.
Key words: urban growth, Spanish cities, conditional β-convergence, endogeneity, Durbin-
Wu-Haussman, spatial autocorrelation, spatial heterogeneity, convergence clubs.
JEL classification: C21, C31, C51, D14, O18, R11
*
Research partially supported by the “Instituto de Estudios Fiscales” and “Universidad Autónoma de Madrid”
(Spain). We would like to thank Julie Le Gallo for helpful discussions relating to certain issues considered in
this paper. We also appreciate the collaboration of Esteban Sanromá and Raúl Ramos, who have provided us
with some valuable data. Earlier versions of this paper were presented at the Regional Science Association
International World Congress at the PE Technikon, Port Elisabeth (South Africa) in April, 2004, Regional
Economic Applications Laboratory seminars at the University of Illinois at Urbana-Champaign in August 2004
and the First Seminar in Spatial Econometrics at the University of Zaragoza (Spain) in October 2004. 2
I. INTRODUCTION
The study of the territorial/regional development in Spain has nowadays a relatively
long tradition, especially since the birth in the early eighties of the Autonomous
Communities (“Comunidades Autónomas”) or regions, considered as NUTS II in Eurostat
nomenclature. There are plenty of articles and books written about the Spanish regional
development, and in general they can be considered as a rich economic literature. But when
one looks at the regional development topic from the point of view of cities, there are only a
few documents and the studies are very scarce (Trullén, 2002; Trullén et al., 2002;
Viladecans, 2002; Mayor and López, 2003). This paper tries to improve the knowledge of
the Spanish urban system and boost the urban studies in a country that has experienced a
fast urbanization process within the last four decades. It is evident that at the international
level this topic has received much more attention (in the EU, Cheshire, 2002; in the USA,
Henderson, 1986, 1995; Glaeser et al., 1992, Glaeser, 1998, among many other good
references).
In fact, our intention is to test the same hypothesis formulated by Cheshire (2002):
“The integration of Europe favours the core regions at the expense of the peripheral ones
(…) removing protection as a result of economic integration works to the relative
disadvantage of backward, peripheral regions and favours advanced core regions” (pp. 213).
Several analysis have been presented in the literature (from Clark et al., 1969 to Venables,
1996) that employing very different methodologies, converge on the 19th-century economic
geographer’s conclusion that “the best protection for a backward region is a bad road”,
interpreting “road” broadly as a shorthand for “costs of doing trade”, including all transport
costs and tariff and non-tariff barriers. On the other point, there are two exceptions in
Steinle (1992) and Cheshire and Carbonaro (1996) that reach to the conclusion that the
strongest gains from integration would continue, over time, to spread outwards from the
core to the near periphery: for example, to Catalonia, Valencia, the Ebro valley and to the
south and east of England.
In this paper, we analyse the performance of the 122 main Spanish cities (province
capitals and those with more than 50,000 inhabitants) in terms of per capita GDP. The
analysis period starts just before the Spanish adhesion to the EU in 1985 and ends in the
year 2001, for which we have available data. As a result of this analysis we can also
conclude that, in the group of main Spanish cities, there has been a significant spread from
the core to the near periphery in terms of economic development during the EU integration 3
period. In fact, if in 1985 the core was located in the northern and eastern cities, as well as
in Madrid and some nearby metropolitan towns, by 2001 there has been a shift towards
some near periphery cities, especially in the Castilian ones surrounding Madrid (Avila,
Ciudad Real, Cuenca, Salamanca, Segovia, Valladolid, etc.).
Therefore the aim of this paper is twofold: firstly, to determine the factors that
explain the urban growth of Spanish cities; secondly, to observe the situation of cities in
terms of “winners” and “losers” after a long period of integration of Spain in the EU or at
least how is the pattern of Spanish cities like after a period of Economic and Monetary
Union -Single European Market and Euro- impact. For the former, we estimate a β-
convergence model and spatial effects are tested with spatial econometric techniques. And
for the later, we also use exploratory spatial data analysis (ESDA).
We use the formal tools of spatial econometrics to identify and include the relevant
spatial effects in the estimation of the appropriate income growth model. In effect, in spatial
cross-sectional contexts, it is almost inevitable to test for the presence of spatial spillovers.
Spatial econometrics techniques, including the ESDA, are the appropriate tools to manage
with spatial dependence in the error terms.
The paper proceeds as follows: section II introduces the theoretical framework to
study the factors that explain the urban growth of Spanish cities and provides some insights
into the β-convergence model and spatial effects upon which the empirical estimations
described in the following sections rely. Section III presents the estimation of urban GDP
and exploratory spatial data analysis (ESDA) is used to detect spatial autocorrelation and
spatial heterogeneity among Spanish urban GDP data. In section IV, we present the model
and set of control variables. The estimation, testing and re-specification process of the
appropriate model is detailed in section V, in which these two spatial effects –spatial
autocorrelation and spatial heterogeneity- are included. In section VI, some economic
interpretations of the results are shown and the last section provides some concluding
remarks.
II. THEORETICAL APPROACH
This section introduces the theoretical framework to study the factors that explain the
urban growth of Spanish cities. The function that explains urban evolution in Spain could be
a cross-sectional specification of the neo-classical growth model (Solow, 1956), which is
considered as a natural starting point of the analysis of regional disparities, especially in 4
Europe (Fingleton, 2003). Although debatable, the neoclassical convergence specification
have a sound empirical track record and that is why we considered it could also be useful to
explain urban growth processes. The neo-classical model predicts that the growth rate of a
city is positively related to the distance that separates it from its steady state. That is to say,
if all urban economies are structurally identical and have access to the same technology,
they are characterized by the same steady state, and differ only by their initial conditions:
() ()
, 2
,0
,0
1
ln ln ; . . . 0,
it
ii i
i
y
yiid
ty
ε
α βεε σ
⎛⎞
=+ + ⎜⎟
⎜⎟
⎝⎠
