Content uploaded by Sebastien Bourdin
Author content
All content in this area was uploaded by Sebastien Bourdin on Sep 14, 2018
Content may be subject to copyright.
Accepted version – Review : Economic Geography
Does the Cohesion Policy Have the Same Influence on Growth
Everywhere? A GWR Approach in Central and Eastern Europe
Dr. Sebastien BOURDIN
Normandy Business School (France) Metis Lab
Department of Regional Economics and Sustainable Development
9, rue Claude Bloch, 14 052 Caen (France)
sbourdin@em-normandie.fr
Keywords: cohesion policy, regional growth, convergence, GWR, Central and Eastern Europe
JEL Classification Codes: C21, O40, H50, R11
Acknowledgments
This research did not receive any specific grant from funding agencies in the public,
commercial, or not-for-profit sectors. I thank the four reviewers that has help me to improve
the quality of the paper.
Abstract
The latest successive enlargements to Central and Eastern European countries and their
differentiated convergence raise the question of the effectiveness of the European structural
funds that have been greatly enjoyed by such countries. The literature on this question is
nevertheless not unanimous. I therefore offer an analysis of the cohesion policy and its role in
regional growth specifically in Central and Eastern Europe, using a method developed in spatial
analysis, namely, GWR (geographically weighted regression) at the NUTS 3 level. My findings
reveal the existence of a multipolar convergence. The differentiated spatial variations of the
influence of European funds on regional economic growth call for reorientation of the cohesion
policy, especially in favor of a more territorialized policy.
Accepted version – Review : Economic Geography
In 2009, the EU’s cohesion policy had been in existence for twenty years. The
implementation of this policy acknowledges that the market forces are not necessarily sufficient
to significantly reduce regional disparities. The EU therefore created this tool of financial
solidarity between member states with the aim of improving the competitiveness of slow-
growth regions and correcting regional unbalance. Since its creation, this policy has always
aimed at reducing regional disparities, restructuring regional economies, creating jobs and
stimulating private investment. In their article, Becker, Egger, and Von Ehrlich (2012)
wondered whether this policy really helped to favor growth in the regions concerned. They
concluded that reorienting the European aid that was more concentrated in targeted regions
would improve the effectiveness of the spent funds. They were therefore able to demonstrate
the spatial heterogeneity of the effectiveness of the cohesion policy in Europe.
Despite the fact that the EU has been extended to eleven new Central and Eastern
European Countries (CEECs), the question of the future of European policies, particularly the
cohesion policy, remains problematic. The European regional policy has been developed in
stages since 1975 and funds development programs for EU regions that are backward or facing
structural difficulties, to use the official terms (OECD 2012; European Commission 2014).
Since this cohesion policy was created, backward regions have greatly enjoyed European
structural funds known as convergence funds with outcomes that exceed the expectations of the
European Commission (Farole, Rodríguez-Pose, and Storper 2011). When these Central and
Eastern European regions joined the European Community, they received European funds;
however, it can be noted that, today, only six of them have a per capita gross domestic product
(GDP) higher than the EU average while they have been part of the EU for over ten years. With
a budget of 351.8 billion euros for the period 2014–20—compared to 347 billion euros for the
Accepted version – Review : Economic Geography
previous period 2007–13—the EU has reiterated the importance of the cohesion policy while
the economic crisis could have driven the European members of parliament to vote for a scaled-
down budget. In such a context, this policy is seen as an income and growth accelerator,
especially in peripheral regions.
This article contributes to the literature on the question of evaluating the impact of the
European structural funds, which have grown extensively over the past few years. I would like
to make two contributions to the academic work on the subject. First, there are currently more
than seventy articles on this subject (Dall’Erba and Fang 2017), but none of them specifically
address the local effects of structural funds in lagging regions, especially in Central and Eastern
Europe. Given the amount of European structural funds allocated to regional growth of less
developed areas, it seems important to me to deal with this issue in particular. Specifically
studying these countries rather than considering them as a homogenous block (when performing
a study on the entire EU while considering the CEEC as a spatial regime in itself) enables me
to have a more thorough approach to the territorial reality of regional growth in these countries.
Second, as already pointed out by Le Gallo, Dall'Erba, and Guillain (2011), for a long
while, very few articles used a spatial econometric approach while the omission of the spatial
dimension of regional growth biases the findings. This spatial dimension has two aspects:
spatial autocorrelation, which refers to the lack of independence between geographic
observations; and spatial heterogeneity, which is related to the differentiation of variables and
behaviors in space. The omission of the spatial spillover effects can produce biased estimates
of the cohesion policy impact, due to the fact that it may also affect the growth rate of
neighboring regions, since it affects the growth of a particular region. Furthermore, the omission
of the spatial nonstationarity of the effects of the cohesion policy may also bias the findings.
Indeed, the value of the overall model does not represent the values of specific local study areas.
The effects of the cohesion policy detected in a global model would be weaker or stronger than
Accepted version – Review : Economic Geography
in some local study areas. To deal with the spatial dimension of regional growth, researchers
have especially tried to control the spatial spillover effects and dealt with spatial dependence
by using mainly global models (Spatial Error Model (SEM), Spatial Autoregressive Model
(SAR), Spatial Lag Model (SLX), Spatial Durbin Model (SDM), in particular), that is, those
that estimate a regression coefficient for the entire sample for the variable relating to European
funds. On the other hand, while the majority of the articles detect spatial heterogeneity, very
few of them deal with it explicitly (Postiglione, Andreano, and Benedetti 2013). As pointed out
by Billé, Benedetti, and Postiglione (2017), once heterogeneity has been detected, no test can
suggest how to correctly model the spatial data set and in what direction we must proceed for
further analyses. One solution, provided by the spatial econometricians, is to perform these
spatial dependence tests on defined spatial regimes (in the form of clubs, that is, the core versus
the periphery—Central Europe, Mediterranean Europe). These regimes, however, do not enable
us to take into account the structural instability as expressed by changing functional forms or
varying parameters across the space. My second contribution to the literature has the ability to
better deal with spatial heterogeneity. In order to do this, I use a method— geographically
weighted regression (GWR)—which has not yet been used to evaluate the spatially
differentiated influence of the cohesion policy on regional growth. Instead of modeling a global
equation, this local spatial analysis method generates separate equations for each observation.
By using it as a spatial microscope (Fotheringham, Brunsdon, and Charlton 2002), we can take
into account the location-specific nature of regional growth and its determinants, allowing the
global specification to be refined. This is all the more justified as, as recommended by Dall’Erba
and Fang (2017), it is important to be able to consider the presence of spatial heterogeneous
treatment effects in order to reconsider the one size fits all theory, and therefore demonstrate
that the European funds can have a significant positive impact in some regions and no or even
Accepted version – Review : Economic Geography
opposite effects in others. My thought process is therefore in line with recent literature on the
possible nonlinear effects of funds (Becker, Egger, and Von Ehrlich 2012).
The consideration of the role of spatial heterogeneity and effects of overflow—as has
already been suggested by the endogenous growth theory and the New Economic Geography
(NEG)—is an alternative to the neoclassical theory and is used to conduct territorialized
analysis of the convergence process involving several factors (including the cohesion policy).
In such a context, I use the GWR to assume that there is a significant spatial variation in the
influence of the factors instrumental in regional growth. One of my main objectives is to show
the spatial heterogeneity of the effectiveness of European funds (over the last two programming
periods) in the explanation of the growth of Central and Eastern European regions, or in other
words, to show that the cohesion policy does not have the same influence on growth
everywhere.
The analysis is based on a sample of 147 Central and Eastern European regions at the
NUTS 3 level over the period 2000-–16 using GWR. Although the favored level of allocation
of European funds is NUTS 2, I have made the decision to work on NUTS 3 regions in order
to provide more detailed understanding of the effects generated by the EU cohesion policy
transfer on growth (like Becker, Egger, and Von Ehrlich [2012] or Gagliardi and Percoco
[2017]). As mentioned above, the choice of (1) focusing on Central and Eastern European
regions, (2) this level of analysis, NUTS 3 and (3) this method is unprecedented in the literature.
This article offers a review of the literature on the measurement of the effectiveness of
the cohesion policy and, more precisely, on the consideration of the spatial dimension when
evaluating structural funds, before presenting the database used and the methodology followed.
The findings of my study are then presented and analyzed, before the conclusion, which offers
some points of discussion.
Accepted version – Review : Economic Geography
1. Taking the Spatial Heterogeneity of Regional Growth into account in
the new cohesion policy
Several studies have shown the polarization of regional growth in Central and Eastern
Europe (Petrakos 2000; Ezcurra, Pascual, and Rapún 2007; Aristei and Perugini, 2012;
Gorzelak, Maier, and Petrakos 2013; Gorzelak 2017). They particularly highlight quick growth
recorded in metropolitan regions due to a concentration of service activities, direct foreign
investment, and a significant number of start-up launches. The regions of Central and Eastern
Europe situated close to the former Iron Curtain tend to experience higher growth rates than the
regions situated in the East. This can be partially explained by the fact that trade is more intense
due to the existence of differentials (in cost, offering, structure by age, etc.) and associated
regional growth distribution phenomena. There is therefore a shift from the Iron Curtain to the
Golden Curtain (Bourdin 2015). On the contrary, Eastern regions find it difficult to reduce the
gap with developed regions. Their industry is declining, their agriculture is fragmented and not
very competitive, there is a lack of infrastructures, and hardly any money is spent on research
and innovation.
This spatial nonlinearity of regional development is complemented by the heterogeneity
of the impact of the European funds on economic growth requiring the reworking of the
cohesion policy. Research into the ex-post evaluation of the impact of the cohesion policy has
increased over the past few years (see in particular Rodríguez-Pose and Fratesi 2004; Dall’Erba
and Le Gallo 2008; Ertur and Le Gallo 2009; Arbia, Battisti, and Di Vaio 2010; Mohl and
Hagen 2010; Becker, Egger, and Von Ehrlich 2013; Bouayad-Agha, Turpin, and Védrine 2013;
Fratesi and Perucca 2014; Bachtler et al. 2017; Crescenzi, Rodríguez-Pose, and Storper 2007;
Percoco 2017; Gagliardi and Percoco 2017). It highlights the circumstances at the origin of the
spatially differentiated impact of the structural funds (Crescenzi and Giua 2018). The findings
of the work on the subject show that the effectiveness of this policy is dependent on (1) the
Accepted version – Review : Economic Geography
concentration of the funds on certain themes, (2) the regional environment (spatial spillover
effects), and (3) the local regional characteristics/the territorial capital of the region.
In a regional integration scheme such as that of the EU, the unequal distribution of wealth
is a highly important political issue. The general objective is to stimulate the least developed
countries/regions toward those that are more developed. The question of knowing whether the
cohesion policy helps to reduce or accentuate the disparities is not easy to deal with, and there
are arguments in the literature that are in favor of one or the other. That said, the spatial
heterogeneity of regional growth is a topical subject in the EU now more than ever. It questions
the design of this public territorial development policy, which has been entirely reworked for
the programming period 2014–20. This is why the modern approaches to territorial
development have taken into account the key role of geography in the policies targeting
economic growth (Varga 2017).
Two trends have oriented the reworking of the European regional policy (Barca, McCann,
and Rodríguez-Pose 2012; Bachtler et al. 2017). First, globalization and the arguments
developed by Porter (1990, 1) stating that “the competitive advantage is created and sustained
through a highly localized process” forced us to question local specificities. The result was a
thought process about the specific factors at the origin of competitive advantages such as the
quality of human capital, the presence of infrastructures related to knowledge, or the existence
of networks and clusters (Capello and Nijkamp 2009; Crescenzi and Rodríguez-Pose 2012). All
these factors reinforced the idea that regional policies should be thought out at the regional level
(Lagendijk 2011). Second, we are witness to a general trend to decentralize the different types
of development policies in terms of research and development in EU countries in particular
(OECD 2011). The approach supported by the OECD (2009) consists in developing the growth
potentials that exist in each region and implementing territorial development policies that aim
at helping lagging regions to realize these potentials. The challenge for the Central and Eastern
Accepted version – Review : Economic Geography
European regions is therefore to rely more on their territorial capital (Camagni 2002, 2008) by
making endogenous investment rather than relying on direct foreign investment, which is
volatile by nature.
In a context that is as heterogeneous as the EU’s regional system, the top-down approach
that was adopted from the one-size-fits-all for a long time is limited (McCann and Rodríguez-
Pose 2011; Barca, McCann, and Rodríguez-Pose 2012). Finding the best way to better adapt
the actions and political interventions to local heterogeneous contexts has become an essential
question today (McCann and Ortega-Argilés 2013). We can clearly notice a change in the
regional policy paradigm. It now includes a territorialized dimension that necessitates (1)
focusing on the local endogenous advantages and knowledge, (2) designing and adapting the
interventions to the specific contexts and their spatial relationships, and (3) stimulating the
choices and positioning of the local players (Barca 2009). It is also necessary to give more
weight to the subnational levels (Leonardi 2005). The answer was to introduce smart
specialization (European Commission 2010; OECD 2011) as we shift from a place-neutral (or
space-blind) logic to a place-based logic. Smart specialization is considered to be a key factor
in treating economic disparities in the European regions (European Commission 2010). It offers
a way to establish priorities when it comes to funding policies, aiming at improving territorial
development, by relying on the local opportunities that underlie innovation and are focused on
entrepreneurship.
This smart specialization policy (Foray 2011) has revived a significant academic
discussion about specialization and diversification (Asheim, Moodysson, and Tödtling 2011;
Asheim 2013; Boschma 2014). In spite of this discussion and the challenges to be faced from
an operational point of view (Lagendijk 2011; Camagni and Capello 2013), this concept has
become important in the EU (McCann and Ortega-Argilés 2013, 2015). By focusing on the
aspects that are at the origin of the best competitive potentials in a region, smart
Accepted version – Review : Economic Geography
specialization/differentiation helps to position the regions in global niches/markets and specific
international value chains (Foray and Van Ark 2007; Foray, David, and Hall 2009; Foray 2011).
It does not only depend on the provision of physical infrastructures; it also depends on the
training of its human capital and on the ability of its institutional environment to implement
such policies based on the knowledge economy. Some current thought processes (Benner 2014)
go even further and imply that this policy could be even more refined with the idea of
considering individual agents, their actions, and their relations in order to identify the
trajectories of economic development (concept of smart experimentation). In all cases, the new
approach of this place-based policy is to take the spatial heterogeneity of economic
development into account.
This new design of the cohesion policy reflects the shift from a nonspatial design to an
explicitly spatial and regional endogenous design. It aims at improving its effectiveness and
efficiency by taking into account the territorial variability of the effect of structural funds. In
this way, evaluating the spatial heterogeneity of the effectiveness of the cohesion policy
between 2000 and 2016 will enable us to justify the interest in a more territorialized approach
to this mechanism, measured by the findings.
2. Method and Data
As mentioned in the introduction, my contribution is to explicitly take into account
spatial heterogeneity using GWR. Far from opposing global and local models, I showed the
complementarity of these two methods by explaining that one of them (SDM—global model
and regime model taking the relative location of a region into account) is used to better capture
the spatial spillover effects, whereas the other one (GWR—local model taking the absolute
location of a region into account) is used to capture the spatial heterogeneity effects.
Accepted version – Review : Economic Geography
2.1.The GWR Method
Tobler’s law (1970) is a central hypothesis of the analysis of spatially referenced data and
shows that observations made close to other observations over space tend to be more similar
(positively correlated) than others that are more distant. This implies that not only the model’s
variables of interest are spatially dependent (nearby regions tend to have similar values) but
also that the residuals of the model may be spatially dependent. That said, the presence of spatial
autocorrelation of the model’s residuals is at the origin of biased parametric estimations.
Anselin (1988) suggested two types of model enabling the spatial structuring of data to be taken
into account. There is either a spatial structure of residuals, in which case I will use an SEM, or
the spatial structuring can be found in the model’s variables, in which case I will prefer an SLX.
Spatial heterogeneity is another phenomenon to be taken into account when building a model.
I make the hypothesis that modeled relations vary over space. This spatial nonstationarity
implies that some variables can have a positive effect on some territorial units while negative
effects can be observed in others (Meloche and Shearmur 2010).
The recent approach that has gained in popularity when it comes to considering and
representing geographic heterogeneity is GWR. It is a mainly exploratory method used to
identify the nature and patterns of spatial heterogeneity for the entire zone studied. Unlike linear
regression and autoregressive spatial models that generate an equation for the entire data table,
GWR generates an equation for each spatial entity and thus, local values of R², , , T of
Student. So, GWR can evaluate separate coefficients, possibly for each observation, in contrast
to the approach using multiple (global) linear regression.
When evaluating the regression of each individual region, the features of the regions taken
individually are weighted using their spatial proximity. This means that the observations that
are closer to a region have more weight in the estimation than observations that are more
distant. The result of GWR is both an overall figure for each estimated parameter and a set of
Accepted version – Review : Economic Geography
localized estimations of the parameters as well as localized versions of regression quality
measurements. As highlighted by Waller et al. (2007), these measurements can be used as an
exploratory tool to indicate the regions in which there are relations that can be interesting to
analyze. As the local estimations are all associated with specific locations, each set of
parameters can be mapped to illustrate the spatial variations of the measured relation. This
mapping can then be used to improve the understanding of the modeled processes as well as to
identify the local spatial anomalies for each exploratory variable.
Finally, from a methodological point of view, Leung, Mei, and Zhang (2000) and Huang
and Leung (2002) tested the presence of spatial autocorrelation in the GWR model’s residuals
using the statistic Moran’s I in order to evaluate the degree of dependence between close
observations. I measure whether the chosen model is biased and the degree of efficiency of its
estimators.
Up until now, only five research articles had studied the spatial nonstationarity of the
convergence process using this method. Bivand and Brunstad (2005) and Sassi (2010)
attempted to understand the spatial variations of the impact of the European agricultural policy
on convergence in Western Europe. Eckey, Kosfeld, and Türck (2007) and Yildlrim, Öcal, and
Özyildirim (2009) studied regional convergence for Germany and Turkey, respectively. Lastly,
Artelaris (2014) studied the convergence of European regions between 1995 and 2005 using
the same method and concluded that convergence was low. As far as I know, no academic
article has yet integrated the role of the cohesion policy into the modeling of regional growth
using GWR. I therefore suggest using GWR to confirm the existence of a varied convergence
rate for the European regions and to study the spatial variations of the impact of the cohesion
policy on regional growth. Let us now consider the mathematical expression of the GWR
model.
In a traditional linear regression model, the latter is as follows:
Accepted version – Review : Economic Geography
For GWR, the model is written as follows:
where is the location in a geographic space of the observation. When calibrating the
GWR model, I suppose that the data observed close to a point have more influence over the
estimation of the values of de than the data located far from. The equation therefore
measures the relations inherent to the model around each point. In GWR, an observation is
weighted with respect to its proximity to point . The choice of weighting scheme is an
important step in model specification, since it implies that the observations closest to the
location have more influence on the estimated parameters of this location than the
observations that are the furthest away. So, weight can be considered as a continuous,
ever-decreasing function of distance . For the purpose of my analysis, I chose the
Gaussian distance decay function with
where 0 and is defined as being the bandwidth of the function or, in other words, the radius
of the sphere of influence for point .
Fotheringham (2009) explained that the findings of GWR are hardly ever sensitive to
the definition of the spatial weighting function, but they are when it comes to choosing the
bandwidth of the function,. Since it is a priori difficult to specifically define the bandwidth,
the first approach may consist in performing cross-validation in order to minimize the root-
Accepted version – Review : Economic Geography
mean-square-error of prediction. Another approach is possible, and recommended by
Fotheringham, Charlton, and Brunsdon (2001). It is the corrected Akaike information criterion
(AICc
1
) method which is used to define the spatial reach as accurately as possible. As part of
my study, I chose the AICc method to calculate the radius of the sphere of influence, which
minimizes the value of the Akaike information criterion; in other words, I was looking to reduce
the difference between the observed value and the estimated value for. In my model, the
bandwidth varies between 52 and 57 regions.
2.2.The Data
The debate between the supporters of the schools of divergence or convergence has been
given a new dimension with the new growth theory. Several factors that can explain a spatial
distribution of economic growth that is far from homogeneous can be named such as access to
markets, human capital, technological change, international competitiveness, economies of
scale, institutional effectiveness, and geographic location.
Regional economic growth is the model’s dependent variable and is estimated in real
terms for the period 2000–16. This variable enables me to evaluate the effectiveness of the
cohesion policy, as I expect there to be a positive relation between these two variables.
Effectiveness is defined here as being the relation between the obtained results (economic
growth
2
) and the objectives set by the European decision-makers. Appropriate identification
strategies may be difficult to develop to measure the causal relation of a public policy on what
it is supposed to deal with. From this point of view, it is obvious that counterfactual methods
(using a regression-discontinuity design–RDD), such as the ones developed recently in some
articles (see, e.g., Becker, Egger, and Von Ehrlich 2013, 2017; Pellegrini et al. 2013; Percoco
1
The AICc is a corrected version of the AIC used for small samples. This is the method I chose for my analysis.
2
Crescenzi and Giua (2018) have recently reminded us that the methods for evaluating the cohesion policy should
not just be focused on the impact on regional economic growth but also on employment growth. My study
concentrates on economic growth, but it would be interesting if future studies could complete it by evaluating the
effects on employment.
Accepted version – Review : Economic Geography
2017; Gagliardi and Percoco 2017; Giua 2017; Crescenzi and Giua 2018), are still the most
adequate (Baslé 2006; Bradley and Untiedt 2007). By focusing on the net effects (impacts),
they enable me to check whether the aim pursued by the policy has been reached thanks to this
or because of exogenous factors (deadweight effect), by comparing the differences in growth
rates between treated and control regions.
3
They are therefore very well suited to a policy-
performance evaluation of the cohesion policy. In fact, given the method that I use, I will speak
more of the exploratory evaluation of effectiveness, rather than measurement of the net impact
(quantification of the effects).
The independent variables are measured at the beginning of the studied period (except
the institutional quality index) in order to capture the effects of the initial conditions on the
scope of growth. The collected data come from Eurostat and the ESPON (European Spatial
Planning Observation Network) database. The sample includes 147 NUTS 3 regions belonging
to 8 member states of Central and Eastern Europe (Baltic States, Poland, Czech Republic,
Slovakia, Hungary, Slovenia). Croatia was excluded as it did not enjoy European funds during
the studied period, as were Romania and Bulgaria, which did not receive European funds (apart
from preaccession funds) over the period 2000-2006. I used the geographic coordinate system
GCS_ETRS_1989 (European Terrestrial Reference System—Earth-Centered, Earth-Fixed) for
my analysis.
The unequal spatial distribution of economic activities highlighted in the literature
indicates the important role of geography in regional growth. Agglomeration (Agglo)
economies are measured using the regional population density proxy (inhabitants/km²). The
impact of agglomeration economies on regional economic growth is supposed to be positive
and significant. The regions with a denser population should record higher levels of productivity
3
Note that, for my analysis, all the regions received European funds for the period in question. It was therefore
impossible to conduct such an analysis, as I did not have any regions that had not been dealt with in my sample.
Accepted version – Review : Economic Geography
and therefore better economic performance (Geppert and Stephan 2008; Chapman, Cosci, and
Mirra 2012). As pointed out by Gagliardi and Percoco (2017), conducting empirical work at
the NUTS 3 level rather than at NUTS 2 level enables me to compare regions subject to the
same treatment status (funds are allocated at the NUTS 2 level) but characterized by different
degrees of urbanization (measured by the regional population density in my analysis) and a
more or less great proximity to large cities (measured by the distance to capital cities). The
analysis thereby underlines spatial heterogeneity of economic response to the structural funds.
We must nevertheless bear in mind that the use of the NUTS 3 level should be taken with
caution because the expenses are allocated at the NUTS 2 level, and therefore the use of
estimated and imputed values is very extensive.
The role of human capital in economic growth has been greatly demonstrated in the
literature since the seminal works of Becker (1964) and then of Lucas (1988). Education is an
important part of a regional development policy, and recent studies on the convergence of
European regions have confirmed that the level of human capital has a positive influence on
growth (Rodríguez-Pose and Fratesi 2004; Crescenzi 2005; Faggian and McCann 2010). This
is why I have included the level of education of the population (Educ) in my model, measured
using the percentage of the population aged between twenty-five and sixty-four with a
university degree or equivalent.
Changes in the orientation of regional policies call on empirical studies analyzing the
role of innovation and research and development (R&D) in regional development. Camagni
and Capello (2013) explained that today’s factors of development are above all innovation and
competitiveness. The works resulting from regional science have more than shown the virtuous
circle that exists between innovation and growth dynamics (Martin and Ottaviano 2001;
Ederveen, Groot, and Nahuis 2006; Molle 2007; Bachtler and Gorzelak 2007; Crescenzi,
Rodríguez-Pose, and Storper 2007; Crescenzi and Rodríguez -Pose 2011; Farole, Rodríguez-
Accepted version – Review : Economic Geography
Pose, and Storper 2011). This is why I have also chosen a variable acting as a proxy in my
model, to measure the level of investment in R&D in a particular region, namely, the overall
R&D expenditure per region expressed as PPP parity purchasing power per capita (R&D).
The Sixth Cohesion Report (European Commission 2014) and the Barca report (2009)
highlighted the importance of having sufficient institutional capacity at the regional level to
accelerate economic development and the effectiveness of European funds. Among the studies
that are interested in the factors explaining growth in Europe, recent literature has mentioned
the role of institutions in regional performance (Glaeser et al. 2004; Rodrik, Subramanian, and
Trebbi 2004; Ederveen, Groot, and Nahuis 2006; Bosker and Garretsen 2009; Arbia, Battisti,
and Di Vaio 2010; Farole, Rodríguez-Pose, and Storper 2011; Rodríguez-Pose 2013;
Rodríguez-Pose and Di Cataldo 2014; Rodríguez-Pose and Garcilazo 2015; Rodríguez-Pose
and Ketterer 2018). These works advocate the need to improve the quality of governance, on
the one hand, and reduce institutional heterogeneity (Charron, Lapuente, and Dykstra 2012), on
the other hand, in order to reduce regional inequalities in Europe and to make the cohesion
policy more effective (Charron 2016). I have therefore included this dimension (Insti) by
integrating the institutional quality index (and its components: control of corruption, rule of
law, government effectiveness, and government accountability) into the model (Charron,
Lapuente, and Dykstra 2012; Charron, Dijkstra, and Lapuente 2014). The European Quality of
Government Index (EQI) was designed from a survey conducted with 34,000 inhabitants of the
EU in different countries for 2010 and 2013.
4
The sixteen questions asked in this questionnaire
referred to the perception and experiences of the citizens interviewed regarding corruption in
the public sector and trust toward their regional institutions (in terms of financial management,
law enforcement, and governance). The fact that data at level NUTS 3 were unavailable led me
to allocate the values of level NUTS 2 to the corresponding lower level.
4
As I did not have data for this variable for the first year (2000), I chose to use the data generated in 2010.
Accepted version – Review : Economic Geography
For the programming period 2000-2006, 213 billion euros were allocated to this policy,
nearly 35 percent of the EU’s budget, representing 0.45 percent of its gross national product.
The percentage of budget allocated to the regional policy did not really change (35.7 percent)
during the subsequent programming period 2007–13, but as the EU’s budget had increased, the
overall amount for this period was 347 billion euros. I introduced a variable relating to the sum
of the European funds spent per region (EUFds) into my model. It includes the aggregate
amount of European funds spent by each region (the different objectives + the cohesion funds)
and cumulated over the period 2000-2006 and the period 2007–13. The data have been taken
from the databases available on the Directorate-General for Regional and Urban Policy’s
website.
5
I have chosen to integrate all the objectives and cohesion funds, since the allocation
mechanisms and objectives vary depending on the programming periods. In fact, I wanted to
measure the overall influence of the European funds received by the NUTS3 regions on their
economic growth. For my sample, the payments of Objective 1 make up the largest share of the
overall structural funds spent. It should be noted that the findings of the empirical studies
evaluating the effectiveness of the European funds in the regional catch-up process are
nevertheless contradictory and often depend on the methods used, the temporal nature of the
study, and the choice of territorial grid (Dall’Erba and Fang 2015). As the conclusions in the
literature are diverging (positive unconditional impact, positive conditional impact,
insignificant impact, and negative impact), the objective was to test the extent to which the
cohesion policy influences regional growth (see the recent meta-analyses regarding this
question by Le Gallo, Dall'Erba, and Guillain 2011; Dall’Erba and Fang 2015; Neumark and
Simpson 2015).
Two main interconnected issues relating to structural funds are traditionally mentioned
in the literature: endogeneity and reverse causality. In the academic literature, it is admitted that
5
http://ec.europa.eu/regional_policy/en/policy/evaluations/data-for-research/
Accepted version – Review : Economic Geography
the amount of structural fund payments is strongly correlated to the initial GDP, which implies
that there is an obvious problem of endogeneity for this variable. The numbers show that the
distribution of the funds for the whole of the EU (especially the funds relating to cohesion and
Objective 1—representing 72 percent and 81 percent, respectively, of the total cohesion policy
budget for 2000-2006 and 2007–13) is highly concentrated in the Objective 1 regions (the
poorer ones), while regions that have a higher initial wealth level are receiving less European
Structural and Investment Funds (ESIF)ESIF. To deal with this problem, previous work
(Dall’Erba and Le Gallo 2008; Dall’Erba, Guillain, and Le Gallo 2009; Hagen and Mohl 2011;
Bouayad-Agha, Turpin, and Védrine 2013) has divided their sample with the Europe of the
Fifteen, on the one hand, and, on the other, the Central and Eastern regions and the South
Mediterranean regions. But, as regards the Central and Eastern European regions of my sample,
the correlation between the initial level of economic development (ln2000) and the European
funds spent (lnEUFds) is relatively low (0.22). The least developed regions in Central and
Eastern Europe are not necessarily the ones that receive the most funds (Gorzelak 2017; Medve-
Balint 2017). Map 1 does not enable me to detect a distribution of aid concentrated in the
poorest regions, and justifying the treatment of the endogeneity of this variable in my model.
Furthermore, conducting a more disaggregate study (at the NUTS 3 level instead of the NUTS
2 level) enables me to take advantage of the mild exogeneity of the treatment status (Gagliardi
and Percoco 2017). The second issue related to the first one is the reverse causality. A lagging
region can receive a large amount of European funds (the allocation method is based on the
initial GDP per capita levels with respect to the EU average) and experience strong regional
growth, but there is nothing to say that it is related to the significant amount of European funds
received. Indeed, according to the neoclassical theory, a region’s growth will be all the stronger
the further it is from its steady state or, in other words, if its level of economic development is
Accepted version – Review : Economic Geography
low. It is difficult to differentiate between what depends on the actual effect of European funds
on regional growth and what the Solow model predicts (1956).
Map 1. Distribution of European funds spent per NUTS 3 region.
It is difficult to identify the causal effect of the European funds due to the fact that other
public regional development policies may be implemented by authorities other than the EU
(such as the state and the region). There are, however, no such data at such a level, and if there
were, they would be difficult to compile and compare. It therefore becomes tricky to
differentiate what depends on the actual effect of the cohesion policy (Gripaios et al. 2008). In
light of the restrictions mentioned above, I must interpret the findings of my analysis with
moderation and caution. This is what Molle (2007) explained when he said that the
effectiveness of the European funds must be considered in terms of plausibility rather than in
terms of evidence.
Accepted version – Review : Economic Geography
Finally, I added control variables to the convergence equation estimations. As the
cohesion policy consists of a subsidy for investment, I controlled the initial level of investment
using the variable relating to the gross fixed capital formation (Investot). I also added the
population growth rate (XPop). I have also introduced the distance (by road in km) of each
region from the border with the EU15 (ProxiEU15) and added another control variable
regarding absolute location. It is the distance (by road in km) of each region from a capital
region (ProxiCapi). I have made the hypothesis that proximity to a capital city has a positive
influence on the economic development of a neighboring region through the effect of
distribution of growth (Dijkstra and Poelman 2008; Dijkstra, Garcilazo, and McCann 2013). To
conclude, it seemed relevant to add a control variable relating to the crisis effect, as my
empirical analysis covers the 2000-2016 period. This economic and financial crisis that started
in 2008 had an asymmetric impact on regions, and its effects have been dramatically strong in
many (lagging) regions (Gorzelak 2017). The impact of the cohesion policy in the boom years
is different from the impact in the crisis periods (Pieńkowski and Berkowitz 2017). The latest
studies reached the conclusion of spatial heterogeneity of the costs of the crisis and of a more
or less high regional resilience. Camagni and Capello (2015) concluded that the hosting capital
cities of the CEECs have experienced higher growth rates. Dijkstra, Garcilazo, and McCann
(2015) calculated the average growth rates between 2008 and 2011 to take into account the
crisis effect on the decrease/increase in regional disparities. I have added the variable (Crisis)
in accordance with this calculation.
2.3.The GWR Model
Following the growth model developed by Durlauf, Jonhson, and Temple (2005), the
GWR equation can be written as follows:
Accepted version – Review : Economic Geography
+
where is the model’s dependent variable to be explained defined for a region over the
period (2000–2016) as follows:
is the location of the observation and , , …, are the parameters of the models
resulting from the conditional convergence models by Solow (1956) and Mankiv et al. (1992),
and from the endogenous growth model by Barro (1991).
Using the conditional convergence hypothesis, I can say that there is -convergence if
the estimated parameter () is significantly negative (Barro and Sala-i-Martin 1995).
Note that, as per Mohl and Hagen (2010), I also include country-fixed effects (. I only report
results for regressions with country-fixed effects. The correlation matrix comprising all
variables is given in Appendix 1.
In order to approach the spatial nonstationarity of the studied phenomenon, several
methods can be used. I first chose to use GWR without differentiating between the spatial
regimes within my sample. I then moved on to the stratification of the regions in my sample in
order to define two spatial regimes for the SDM and the GWR models. The ex ante delimitation
of the subzones may follow a predefined spatial hierarchy (Brunsdon, Fotheringham, and
Charlton 1996). In the case of my study, I defined two spatial regimes: the West and the East.
As far as I know, all previous work considered the CEECs as a block, even though the West-
Accepted version – Review : Economic Geography
East dissymmetry has been widely demonstrated in the literature (Dunford and Smith 2000;
Petrakos 2000; Ezcurra, Pascual, and Rapún 2007; Artelaris, Kallioras, and Petrakos 2010;
Gorzelak, Maier, and Petrakos 2013; Epstein and Jacoby 2014; Bourdin 2015). I therefore
wanted to take this limit into account by dividing my sample into two subregimes. This second
model with two regimes was conducted in order to test the differentiated effectiveness of the
cohesion policy between the (1) Western and (2) Eastern regions of my sample. I therefore
make the hypothesis that the effectiveness of European funds is differentiated depending on
whether the region is located in the West spatial regime or in the East. Generally speaking,
European funds tend to have a greater effect on growth in the most developed regions (Pinho
et al.2015). In order to define the two regimes, I used the data on the standardized multimodal
accessibility potential of the regions (by road, air and rail—source ESPON Database), since it
is understood that the Western regions are those with a higher accessibility potential compared
to the average of the regions in my sample. These two spatial regimes therefore enable the
relative location of each region to be taken into account.
Map 2. The two spatial regimes.
Accepted version – Review : Economic Geography
I conducted a geographic variability test for each of the model’s coefficients. I compared
the original GWR model and a switched model in which the tested coefficient is the only one
to be fixed, whereas the other coefficients are kept in their potential variability as per a GWR
model. By comparing the AICc, if the original GWR is better than the switched model testing
the coefficient, it is then possible to say that this coefficient varies over space. I conducted the
traditional hypothetical test using F statistics for a Gaussian model (Nakaya 2014).
An example of convergence model is taken hereunder, where I test the impact of
structural funds on growth (model 1—see Table 1):
Let us take a traditional GWR model:
To test the geographic variability of , I compare a fixed constant model and a fixed
slope model, respectively defined as follows:
(i) fixed constant model:
(ii) fixed slope model:
If the switched model outperforms the fixed slope model, this means that
significantly varies over space. For this, the difference between the original model and the
Accepted version – Review : Economic Geography
switched model is calculated. If the value is positive, I conclude that there is a lack of spatial
variability in the highlighted local term. If the switched GWR model has a statistically better
fit than the fixed slope model, the value of the model is negative, and the variable significantly
varies over space.
Table 1 Geographic Variability Test of the General Territorial Convergence Model
The performed test (see Table 1) shows that all the variables considered in the model
vary over space.
2.4.The SDM
As noticed by Billé, Benedetti, and Postiglione (2017), the spatial econometricians have
mainly tried to control the spatial spillover effects and less to explicitly take spatial
heterogeneity into consideration. One solution provided by them is to perform spatial
dependence tests on defined spatial regimes (clubs). From this point of view, GWR enables us
to better deal with spatial heterogeneity, since this method generates separate equations for each
observation instead of modeling a global equation. On the other hand, the spatial spillover
effects are less captured. So, in order to have an insight into both the impact of location (GWR)
and the impact of spatial organization (spatial parametric models) on regional growth, I have
also included a global model in my study that controls the effects of spatial dependence.
By considering the existence of a spatial autocorrelation phenomenon, the ordinary least
squares estimators are biased and ineffective, questioning the statistical inference approach. In
order to correct this bias, the estimated model integrates spatially lagged variables and,
Variable F DOF for F test DIFF of Criterion
lnEUFds 3.910 0.742 -5.259
ln2000 5.960 0.798 -92.326
lneduc 4.892 0.661 -6.489
lninstit 5.692 0.756 -5.385
lnRetD 3.750 0.511 -6.489
lnAgglo 5.699 0.859 -7.038
Accepted version – Review : Economic Geography
depending on the sources of spatial dependence, the autoregressive process does not apply to
the same variables. The SAR accounts for the spatial dependence of the explained variable and,
in my case, of the effect of regional growth spreading between neighboring observations. When
neighboring regions share common intrinsic or extrinsic characteristics, the SLX can be used
to take the spatial dependence of the explanatory variables into account. The SEM shows the
spatial dependence of error terms, which is the sign of a bad model specification (omission of
variables, unsuited functional form). As the spatial dependence of growth-related data can come
from different sources, I have decided to use an SDM specification, which is a combination of
the SAR and SLX. It is used to take both the spatial dependence related to the explained variable
and spatial dependence related to exogenous variables into account. It enables me to consider
the effects of growth spreading to neighboring regions as well as the impact of similar
environmental characteristics and neighborhood effects.
I consider the SDM where spatial dependence is included in both the endogenous
variables ( and the exogenous variables (:
where represents the development of growth between 2000 and 2016 for a region at a
time t;
is the growth of neighboring regions (spatially lagged endogenous variable),
represents the explanatory variables of the region i,
are the explanatory
variables of the regions neighboring i (spatially lagged exogenous variables); the country-
fixed effects; the constant; and are the vectors of the estimated coefficients and that
of the residuals.
Accepted version – Review : Economic Geography
3. The Contribution of European Structural Funds to Territorial
Convergence
3.1. A Sequential Model of Territorial Convergence
The growth rate has been modeled in the implementation of the GWR as a function of
several parameters described above. I made a sequential model (Table 2) in order to study the
spatial heterogeneity of the disparities in terms of regional growth. Several territorial
convergence models (Models 1 to 4 with structural funds and Models 5 and 6 without structural
funds) were tested. A second territorial convergence model was built with respect to the two
Core and Periphery spatial regimes. The only value to be carried over to the sequential model
table was the average of the coefficients. Mapping the estimated coefficients enabled me to
evaluate the scope of the spatial variability of the parameters of the different models.
In Models 5 and 6 without structural funds, the estimated coefficient associated with the
per capita GDP in 2000 is significant and negative, indicating a convergence process in
European regions. Moran’s index (I = 0,35***) shows that the growth rate of a region is
influenced by that of neighboring regions. In 2016, this index significantly decreases and is
estimated at 0,11 (***), indicating a decrease in the concentration of regional growth.
When I expound on the process of convergence between the Central and Eastern
European regions, I realize that there was a significant reduction between 2000 and 2016. In
this case (Models 3 and 4), the level of expenditure of European funds seems to have a positive
association with the rate of regional growth, confirming the work by Crescenzi, Fratesi, and
Monastiriotis (2017). Their direct influences can be isolated by using Models 1 and 2, which
estimate the regional growth rates only by including structural funds as parameters of the model.
The European aid given to the regions has a positive influence on growth, confirming the work
by Becker, Egger, and Von Ehrlich (2018) or by Ramajo et al. (2008). In the general Models 3
Accepted version – Review : Economic Geography
and 4, I show, like Rodríguez -Pose and Crescenzi (2008), that the local socioeconomic
conditions, such as the level of education, drive economic growth. On the other hand, it would
seem that the level of investment in R&D has a low negative influence on the growth of Central
and Eastern European regions. Finally, it seems that the influence of institutional quality on
regional growth is very high, confirming recent literature (Rodríguez-Pose and Garcilazo 2015;
Rodríguez-Pose and Ketterer, 2018) on regional government quality as a fundamental factor
affecting the economic performance of European regions.
According to the most recent studies indicating that the effectiveness of the funds
depends on the heterogeneity of the regional environment (Bachtler et al. 2017; Crescenzi,
Fratesi, and Monastiriotis 2017; Gagliardi and Percoco 2017), it is interesting to look into the
impact of proximity on regional growth. If I compare Models 1, 3, and 5, I can see that the
control variable regarding the proximity to a capital region has a low negative influence on
regional growth. In other words, the regions located close to metropolitan regions do not seem
to benefit from their spatial growth externalities. This makes me think of the agglomeration
shadow developed by Paul Krugman in 1996, according to which metropolitan regions stop
other nearby regions from developing due to the centripetal forces that characterize them. It
could also be interpreted as a statistical effect since NUTS3 regions bordering metropolitan
regions are often residential regions with residents commuting to the adjacent metropolitan
areas and producing economic growth there or trickledown effect of the business.
The results of models also show that the regions on the border of the Europe of Fifteen
and close to the metropolitan regions recorded higher growth rates, thus confirming the
hypothesis already made by Petrakos (2000), Ezcurra, Pascual, and Rapún (2007) or Gorzelak,
Maier, and Petrakos (2013). These regions take real advantage of their location. Overall, they
have more infrastructures than in the East and a large workforce with relatively low cost, and
Accepted version – Review : Economic Geography
have been able to attract European investors and stimulate their growth due to the fact that their
accessibility is higher.
Table 2 Sequential Model of Territorial Convergence
LQ MUQ LQ MUQ LQ MUQ LQ MUQ LQ MUQ LQ MUQ
Intercept -0,587** -0,535*** -0,49*** -0,578*** -0,521*** -0,484*** -0,561*** -0,547*** -0,522*** -0,561** -0,548** -0,522** -0,583*** -0,556*** -0,518*** -0,581*** -0,555*** -0,517***
ln2000 -0,104*** -0,084** - 0,067*** -0,147*** -0,13*** -0,112*** -0,105** - 0,082** -0,063** -0,146*** -0,089*** -0,093*** -0,104** -0,088** - 0,072** -0,148** - 0,118** -0,106**
lnEUFds 0,065*** 0,094*** 0,119*** 0,092*** 0,105*** 0,127*** 0,085*** 0,094*** 0,106*** 0,081*** 0,089*** 0,102***
lnEduc 0,075*** 0,089*** 0,107*** 0,052*** 0,093*** 0,095*** 0,012** - 0,058** 0,009** -0,002*** 0,079*** 0,079***
lnAgglo 0,011** 0,023** 0,032** 0,028** 0,064** 0,064** 0,057** 0,068** 0,076** 0,078*** 0,097*** 0,097***
lnInstit -0,154* -0,144* -0,129* -0,145* -0,12* -0,114* -0,147** -0,133** -0,122** -0,131** - 0,112** -0,111**
lnR&D -0,027** -0,017** - 0,008** -0,029** -0,015** -0,002** -0,016*** -0,009*** -0,004*** -0,023*** 0,001*** 0,001***
lnInvestot -0,06*** -0,029*** - 0,024*** -0,051*** -0,033*** -0,023*** -0,016** -0,011** -0,005** - 0,022*** - 0,016*** - 0,007*** -0,011** -0,006** -0,003** -0,019** - 0,004** -0,004**
lnXpop -0,083*** -0,015*** 0,017*** -0,065*** -0,038*** -0,003*** -0,015** -0,01** -0,006** -0,021** -0,015** -0,012** 0,003* 0,009* 0,018* -0,002* 0,008* 0,008*
lnProxiCap -0,075* -0,071* -0,064* -0,047*** -0,034*** -0,029*** -0,062** -0,029** -0,029**
lnProxiUE15 0,038** 0,07** 0,083** 0,018** 0,033** 0,436** 0,001* 0,032* 0,054*
AICc -19,03478 -13,33623 -56,70415 -58,8059 -31,96674 -27,18343
R² 0,585961 0,572797 0,707693 0,700976 0,645228 0,635916
Ajusted R²
0,541232 0,524386 0,659652 0,650336 0,592742 0,580428
I Moran of the stand. res. 0,082 0,098 0,087 0,054 0,069 0,102
Joint tests :
0.000
0.000
0.002
0.000
0.005
0.000
0.000
Lower Quartile (LQ), Median (M), Upper Quartile (UQ)
***, **, *: statistically significant at the 0.1%, 1%, 5%, respectively
F-stat. P-value (lnR&D)
F-stat. P-value (country dummies)
F-stat. P-value (ln 2000)
F-stat. P-value (lnEUFds)
F-stat. P-value (lnEduc)
F-stat. P-value (lnAgglo)
F-stat. P-value (lnInstit)
without SF
model 6
model 5
model 4
model 3
model 2
model 1
with SF
Accepted version – Review : Economic Geography
3.2. Spatial Heterogeneity and Spatial Dependence in the Regional Growth Process
Convergence and Influence of European Structural Funds
Building an SDM with regimes enables me to directly capture the spatial spillover
effects, taking spatial heterogeneity into consideration. This approach is therefore
complementary to GWR-based analysis that captures the spatial nonlinearity of regional
growth. I would like to test the hypothesis of a differentiated effect of regional European aid
with respect to the belonging of the regions to the core or peripheral group defined by their
accessibility potential.
Regardless of the spatial regime (Table 3), the influence of European funds on regional
growth is higher for core regions, thus confirming the work by Charron (2016). As European
funds are allocated on the principle of additionality, the effective payments of the European
Commission depend on the abilities of the regions to propose and cofinance projects. This
principle therefore introduces a bias that results from the fact that some regions are able to
double community support, and others are not. The initial wealth of a region and its ability to
set up coherent projects therefore play an important role. In light of this, the regions in the core
spatial regime seem to take more advantage of European funds.
Besides, results show a positive and significant influence of the level of education for
the two regimes. This confirms the work by Petrakos, Kallioras, and Anagnostou (2007), which
explains that an abundance of qualified human capital in a region generates innovation and
growth in the long run. There is also a positive association between the population density
(proxy of urban regions) and regional growth, in both the core and the periphery, confirming
the work by Dijkstra and Poelman (2008) and Dijkstra, Garcilazo, and McCann (2013). My
findings are also linked to what has recently been demonstrated in the literature about the
prevalence of centripetal forces in countries in transition, which also leads to a quicker tendency
to convergence (Monastiriotis 2014).
Accepted version – Review : Economic Geography
The findings of my sequential spatial regime model show that the core regions have
experienced more catch-up than peripheral ones (Table 3). At the same time, the local
estimations of regional growth (Map 3) enable me to give other explanations or slightly modify
the findings. Indeed, there are several regions located in the eastern parts of the CEECs that
have grown faster than national averages, especially in the last period. This could also be the
result of the fact that the major cities of these countries are located more in the West than the
East of these countries, and these cities are growing faster. In any case, the increase in regional
differences (Gorzelak 2017; Medve-Bálint 2017) has consequently been higher due to the fact
that the structural funds have more influence on the core regions than on the peripheral ones,
creating a virtuous circle for the core regions and a vicious circle for the peripheral regions,
since it is harder for the latter to catch up.
Table 3 Sequential Model of Territorial Convergence: Comparison of Two Spatial Regimes
Using GWR and SDM
LQ MUQ LQ MUQ total effects total effects
ln2000 -0,151** - 0109** -0,085** -0,074*** - 0,039*** -0,007*** -0,116*** -0,037***
lnEUFds 0,093*** 0,117*** 0,164*** 0,033*** 0,039*** 0,004*** 0,125** 0,039***
lnEduc 0,073*** 0,084*** 0,094*** 0,077** 0,100** 0,113** 0,085*** 0,091**
lnAgglo 0,017** 0,062** 0,099** 0,012*** 0,029*** 0,047*** 0,062*** 0,032***
lnInstit -0,136** -0,122** -0,109** -0,112* - 0,060* -0,006* -0,121*** -0,056**
lnR&D -0,080** -0,049** -0,021** -0,010** 0,012** 0,041** -0,051** 0,017***
lnInvestot -0,030*** -0,016*** -0,002*** -0,020*** -0,010*** 0,005*** -0,018*** -0,007***
lnXpop 0,001*** 0,011*** 0,022*** -0,012*** 0,005*** 0,014*** 0,011*** 0,001***
0,307*** 0,189***
W x ln2000 -0,212*** -0,122**
W x lnEUFds 0,049** 0,023***
W x lnEduc 0,057*** 0,014**
W x lnAgglo -0,003** 0,0027***
W x lnInstit -0,176** - 0,121***
W x lnR&D -0,112* -0,031**
W x lnInvestot -0,062*** -0,023***
W x lnXpop -0,011** -0,002**
AICc -45,589221 -84,240522
R² 0,73 0,80 0,58 0,60
Ajusted R² 0,62 0,71 0,53 0,55
LRCOM test 16,88*** 20,19***
Log likehood 177,42 153,43
Spatial Breusch–Pagan 11,72** 25,13***
I Moran of the stand. res.
0,082 0,098
Lower Quartile (LQ), Median (M), U pper Quartile (UQ)
***, **, *: statistically significant at the 0.1%, 1%, 5%, respectively
GWR
SDM
model 1 (core)
model 2 (periphery)
model 3 (core)
model 4 (periphery)
Accepted version – Review : Economic Geography
The LR test on the common factor hypothesis is significantly rejected, indicating that there are
externalities and, in this context, the unconstrained SDM is the appropriate specification.
Moreover, the Spatial Breusch–Pagan test against heteroskedasticity is significant and indicates
the presence of spatial heterogeneity. In order to interpret the results of this model, I focus on
the total effects as they take into account both the direct effects and the effect of spatial
dependence between all the observations using the spatial multiplier. The estimation of the
SDM confirms an effect of spatial spreading of regional growth both in the core spatial regime
and the peripheral one, but it is higher in core regions (ρ = 0,307 vs. ρ = 0,231). The level of
growth of a region depends both on its characteristics and on the regional growth of its
neighboring regions. As far as the analysis of the direct and total effect of European funds is
concerned, I note that the elasticity of the explanatory variable has a positive and significant
sign. So, any increase in European aid in the regions neighboring i lead to an increase in the
growth of i. More specifically, these first results indicate a nonlinear relation between growth
and the amount of European funds spent with an elasticity of 0,125 in the core and 0,039 in the
periphery, implying that when the amount of European aid spent increases by 10 percent,
growth increases by 1,25 percent. This echoes previous work such as that done by Becker,
Egger, and Von Ehrlich (2010), which showed that the funds of Objective 1 increased the GDP
per capita by nearly 2 percent.
3.3. A Territorialised Approach to the Convergence and Influence of European Structural
Funds
Just like Le Gallo, Dall'Erba, and Guillain (2011), I think that the local approach to
regional inequalities is the most adequate methodology to be used to get as close as possible to
what Solow (1956) wanted to show with his conditional beta-convergence model. From this
point of view, the GWR is a spatial microscope (Fotheringham, Brunsdon, and Charlton 2002)
enabling us to have a local approach to the great variation of economic growth between the
Accepted version – Review : Economic Geography
regions and the scope of structural instability. The visualization of the GWR model’s
coefficients made possible by this method highlights the spatial variations of the parameters.
Map 3: Spatial variations in the influence of European funds on regional growth (Model 1
global).
Mapping local t-values is an advantage since it allows us to measure the systematic
effect of the parameters on the dependent variable. Local t-values capture both the direction
(sign) and the standardized strength (amount) of local relations between regional growth and
the independent variables. GWR enables us to evaluate where and how the relation between
European funds and regional growth varies over space, in size and significance within the EU.
The spatial distribution of the estimated coefficients of In2000 for Model 1 has been mapped
(Map 3.a). It brings to light territorial convergence, that is, differentiated catch-up depending
on the region. The geographic distribution of the estimated coefficients is not random. Even if,
generally speaking, we observe a West/East differentiation logic, this is to be measured in the
sense that there are several regions located in the eastern parts of the CEECs that grow faster
than national averages, especially in the last period. This growth advantage concentrated more
Accepted version – Review : Economic Geography
in the West can also be explained by the fact that the major cities of these countries are located
more to the West than the East, and these cities are growing with the fastest rates. The spatial
spillover effects are more significant in the more accessible regions than in those that are less
accessible (Table 3).
The findings (Map 3.b) suggest that the amount of European structural funds received has a
significant influence on regional growth, heterogeneously over space. This is especially true
along the border with Western Europe. More specifically, I can identify what could be called a
golden crescent that includes all of the Czech regions, the regions that are the furthest to the
West of Slovakia (regions of Bratislava, Trnava, Trenčín, and Nitra) and Hungary
(Transdanubia) as well as Slovenia. The geographic proximity of these regions to the rich
regions of the Europe of Fifteen (especially Austria) is favorable to their growth. The increase
in the opening of borders linked to the integration of the CEECs in the EU has fostered foreign
direct investments, in particular in the capital regions and the regions that are the furthest to the
West (Gorzelak, Maier, and Petrakos, 2013). In terms of foreign direct investment, Western
Hungary is the reference thanks to its attractiveness, which is partially related to its geographic
position and also to its choice to bet on privatization open to international capital (a much
stronger choice than the rest of the countries in the golden crescent). This links in with the
endogenous growth literature on the initial conditions of the regions. The latter highlights the
role of forces of agglomeration, which tend to strengthen the regions with a favorable economic
environment (the golden crescent in this case). In this context, according to the principle of
additionality of the European funds, which stipulates that the regions must cofinance the
projects, the richest regions find it easier to complement European aid than the regions located
to the East (Del Bo and Sirtori 2016). The levels of institutional quality of these winning regions
are also higher than peripheral regions. We can find multiplier effects where sufficient human
capital and good governance enable the transfer of European funds to be transformed into
Accepted version – Review : Economic Geography
regional growth. This echoes the works written by scholars such as Becker, Egger, and Von
Ehrlich (2013) or Rodríguez-Pose and Ketterer (2018).
GWR consequently enables the spatial heterogeneity of the effectiveness of European
structural funds to be demonstrated, as already mentioned by Le Gallo, Dall'Erba, and Guillain
(2011) and Becker, Egger, and Von Ehrlich (2013). My work confirms those of ,ad-Agha,
Turpin and Védrine (2013), which stipulate that the initial conditions count. They also echo
those of Percoco (2017) according to which the effect of structural funds on growth is
heterogeneous and depends on the economic structure of the regions. In view of my findings,
the European funds do not seem to have the expected influences for the regions with a low
growth potential. The emphasis on multipolar economic growth, as already mentioned by
Bourdin (2015), indicates the differentiated effectiveness of European aid. On the one hand,
these findings support the idea that the regions must strengthen their capabilities (McCann and
Ortega-Argilés, 2013, 2015, 2016; Bachtler et al. 2017). On the other hand, they confirm the
importance of a new cohesion policy based on smart specialization, where we go from a logic
of convergence to a logic of competitiveness, from a sector-specific logic to a territorialized
logic, from a redistributive logic to a logic of local development (Thissen et al. 2013; McCann,
2016; Bachtler et al., 2017).
Discussions and Conclusions
GWR as a spatial microscope enables us to highlight the positive and negative relations
and their strength between regional growth and response variables at the local scale. In adopting
this approach, spatial spillover effects were not considered with the advantage of identifying
the causal effect of structural funds on growth. It was nevertheless important to take spatial
dependence into consideration, since it has been greatly demonstrated in the literature; GWR
does not enable this phenomenon to be taken into account. From this point of view, the two
Accepted version – Review : Economic Geography
approaches adopted must not be seen as competitive but as complementary, and enable us to
understand the spatial phenomena at work in Central and Eastern Europe (Table 4).
GWR
SDM
Spatial analysis
Absolute effect
Impact of location
Local models
Relative effect
Impact of spatial organisation
Spatially lagged variables
Spatial effects
Spatial heterogeneity
Spatial nonlinearity effects
Spatial autocorrelation
Spatial spillover effects
[insert table 4 here]
Among the limits of GWR, multicollinearity may arise, which makes it more suitable
for exploratory rather than confirmatory analyses. As an exploratory method (Fotheringham
and Brunsdon 1999), it measures the influence of the cohesion policy on growth more than the
actual, net impact, as can be done using a counterfactual method such as regression
discontinuity design, for example. One of the limits of my work lies in the fact that nowadays,
the possibility that other public policies (local, regional, national) may also influence regional
growth, along with the cohesion policy, is not taken into account. This is due to the fact that it
is currently difficult to have harmonized, compiled data that can be compared. This is what
Molle (2007) explained when he says that the effectiveness of the European funds must be
considered in terms of plausibility rather than in terms of evidence. New research avenues into
this should also be explored.
As far as I know, all previous work considered the CEECs as a block, even though the
West-East/core-periphery dissymmetry has been widely demonstrated in the literature. The
subject that has been discussed here is important for three reasons. It is of political importance,
since it is desirable to shed light on the question of the contribution of cohesion policy funds to
regional growth in a context where solidarity between member states is called into question. It
Accepted version – Review : Economic Geography
is of theoretical importance because the territorialized analysis of growth and the role played
by European aid implies understanding of the spatial heterogeneity of the factors instrumental
in regional economic development (endogenous growth theory and new economic geography).
It is of methodological importance, since this theoretical debate has been empirically tested
using the GWR method as a tool to evaluate this European regional public policy.
The findings show significant spatial heterogeneity of the local coefficients of the
explanatory parameters of regional growth. The differentiated influence of European funds can
be particularly linked to the capacity of the regions to use these funds efficiently (Charron
2016). This is confirmed by the separation into two spatial regimes (core and periphery) where
the findings show a significant positive influence of the cohesion policy on the regional growth,
which is higher for the core regions than for the peripheral regions of my sample. My findings
nevertheless question the capacity of the funds to be able to reverse the tendencies to
agglomeration and the selective regional growth dynamics. As pointed out by Sotiriou and
Tsiapa (2015), the spatial variation of the differentiated impact of structural funds, combined
with the heterogeneity of current regional economic development, questions the capacity of the
European funds to reduce this unbalance. Some regions have been able to take advantage of
their geographic location close to the agglomerations and the border of the Europe of Fifteen
(in particular, the golden crescent—spatial spillover effects). These findings have consequences
on the fact that it is necessary to take this parameter into account in the current thinking on the
post-2020 cohesion policy.
GWR enables us to ask new questions in order to understand the sources of this observed
heterogeneity. The map representation of the estimated local parameters brings to light the need
for a more territorialized cohesion policy (Petrakos, Kallioras, and Anagnostou 2011; Petrakos
2012). Also in keeping with the works by Becker, Egger, and Von Ehrlich (2012), I think that
a more targeted concentration of the aid would increase the effectiveness of the spent funds.
Accepted version – Review : Economic Geography
The expertise regarding the regional policy has developed in this direction and, at the end of
the first decade of the new millennium, a certain number of very convincing reports about the
intervention of the policy on regional development were published by important international
entities such as the OECD, the World Bank, and the European Commission. These reports
revealed two opposing points of view about an animated debate over European regional policy.
On the one hand, there is the idea of a space-neutral regional development policy, emphasizing
the advantages of agglomeration and the spillover from geographic concentration. On the other
hand, a territorial approach (called place-based), which supposes that it is essential to take the
geographic context into account—especially the role of institutions, the importance of local
knowledge and socioeconomic characteristics. My findings seem to suggest a territorialized
approach to the cohesion policy by taking account of its unequal effectiveness in Europe. On
the other hand, it is not said that an approach using intelligent specialization, based on
performance and competitiveness between the regions, makes the unbalance worse (Avdikos
and Chardas 2015). Some regions (the most developed ones) already have the adequate skills
(soft skills) to enable them to use the European structural funds in the best possible way, and
others (more fragile) are not in a position to make the most of what such funding allows. After
2020, when researchers have new data about the European funds received, the analyses will
enable us to compare the space-neutral and place-based approaches, and to draw conclusions
from them.
Capello (2009) suggested that a new growth theory in regional science is emerging. This
theory invests in the local specificities and conditions that allow an economic system to reach
high levels of competitiveness and innovation, and to more crucially maintain these levels over
time. My analysis of this highlights the spatial variations of these local specificities and
conditions. The political decision-makers must consequently take local characteristics into
Accepted version – Review : Economic Geography
account and encourage more targeted policies. From this point of view, the alternative spatial
regression methods may be particularly useful to shed light on the policies.
Among the future avenues of research, Crescenzi and Giua (2016) invited us to use
mixed methods (spatial econometrics—contextualisation approaches, and counterfactual
methods—identification approaches) as developed by Crescenzi, Fratesi, and Monastiriotis
(2017). As far as the data on European funds are concerned, I deplore the lack of more detailed
and long-term data at the NUTS 3 level about the funds’ expenditure categories (agriculture,
entrepreneurship, transport infrastructures, environment and urban infrastructures, R&D,
human capital). This would have enabled us to show the extent to which the different ways of
investing (hard investments [infrastructures] vs. soft investments [R&D and human capital])
heterogeneously influence growth as shown in other studies (e.g., Sotiriou and Tsiapa 2015).
By spatially comparing the sizes of the estimated parameters, GWR can be used to
conceptualize the relative influences of the independent variables as a continuous function over
space. In their article entitled “Can GWR Improve Regional Analysis and Policy Making?”,
Ali, Partridge, and Olfert (2007) questioned the interest and limitations of using such a method.
As they mentioned, the findings of GWR are more designed to be exploratory and to generate
hypotheses than to test hypotheses. As my findings demonstrate the spatial heterogeneity of
growth, they must be completed by more in-depth investigations into the reasons for this
heterogeneity.
References
Ali, K., Partridge, M. D., and Olfert, M. R. 2007. Can geographically weighted regressions
improve regional analysis and policy making? International Regional Science Review
30 (3): 300–29.
Accepted version – Review : Economic Geography
Anselin, L. 1988. Lagrange multiplier test diagnostics for spatial dependence and spatial
heterogeneity. Geographical Analysis 20 (1): 1–17.
Arbia, G., Battisti, M., and Di Vaio, G. 2010. Institutions and geography: Empirical test of
spatial growth models for European regions. Economic Modelling 27 (1): 12–21.
Aristei, D., and Perugini, C. 2012. Inequality and reforms in transition countries. Economic
Systems 36 (1): 2–10.
Artelaris, P. 2014. Local versus regime convergence regression models: A comparison of two
approaches, GeoJournal 80 (2): 263–77.
Artelaris, P., Kallioras, D., and Petrakos, G. 2010. Regional inequalities and convergence clubs
in the European Union new member-states. Eastern Journal of European Studies 1 (1):
113–32.
Asheim B. T., 2013. Smart specialisation—Old wine in new bottles or new wine in old bottles?
ERSA Conference Paper ersa13p476. Louvain-la-Neuve, Belgium: European Regional
Science Association.
Asheim B. T., Moodysson J. and Tödtling F. 2011. Constructing regional advantage: Towards
state-of-the-art regional innovation system policies in Europe? European Planning
Studies 19 (7): 1133–39.
Avdikos, V., and Chardas, A. 2015 European Union cohesion policy post 2014: More (place-
based and conditional) growth—Less redistribution and cohesion. Territory, Politics,
Governance 4 (1): 97–117.
Bachtler, J., and Gorzelak, G. 2007. Reforming EU cohesion policy: A reappraisal of the
performance of the structural funds. Policy Studies 28 (4): 309–26.
Accepted version – Review : Economic Geography
Bachtler, J., Berkowitz, P., Hardy, S. and Muravska, T., eds. 2017. EU cohesion policy:
Reassessing performance and directions. New York: Routledge.
Barca, F. 2009. An agenda for a reformed cohesion policy. A place-based approach to meeting
European Union challenges and expectations. Brussels, Belgium: European
Commission.
Barca, F., McCann, P., and Rodríguez-Pose, A. 2012. The case for regional development
intervention: Place-based versus place-neutral approaches. Journal of Regional Science
52 (1): 134–52.
Barro, R. J. 1991. Economic growth in a cross section of countries. Quarterly Journal of
Economics 106 (2): 407–43.
Barro, R. J., and Sala-I-Martin, X. 1995. Economic growth theory. Boston: McGraw-Hill.
Baslé, M. 2006. Strengths and weaknesses of European Union policy evaluation methods: Ex-
post evaluation of Objective 2, 1994–99. Regional Studies 40 (02): 225–35.
Becker, G. S. 1964. Human capital theory. New York: Columbia University Press.
Becker, S. O., Egger, P. H., and Von Ehrlich, M. 2012. Too much of a good thing? On the
growth effects of the EU's regional policy. European Economic Review 56 (4): 648–68.
______. 2013. Absorptive capacity and the growth and investment effects of regional transfers:
A regression discontinuity design with heterogeneous treatment effects. American
Economic Journal: Economic Policy 5 (4): 29–77.
______. 2018. Effects of EU regional policy: 1989–2013. Regional Science and Urban
Economics 69 (March) 143–52.
Benner, M. 2014. From smart specialisation to smart experimentation. German Journal of
Economic Geography 58 (1): 33–49.
Accepted version – Review : Economic Geography
Billé, A. G., Benedetti, R., and Postiglione, P. 2017. A two-step approach to account for
unobserved spatial heterogeneity. Spatial Economic Analysis 12 (4): 452–71.
Bivand, R., and Brunstad, R. 2005. Further explorations of interactions between agricultural
policy and regional growth in Western Europe—Approaches to nonstationarity in
spatial econometrics. Amsterdam: European Regional Science Association Congress.
Boschma, R. 2014. Constructing regional advantage and smart specialisation: Comparison of
two European policy concepts. Scienze Regionali 1:51–68.
Bosker, M., and Garretsen, H. 2008. Economic development and the geography of institutions.
Journal of Economic Geography 9 (3): 295–328.
Bouayad-Agha, S., Turpin, N., and Védrine, L. 2013. Fostering the development of European
regions: A spatial dynamic panel data analysis of the impact of cohesion policy.
Regional Studies 47 (9): 1573–93.
Bourdin, S. 2015. National and regional trajectories of convergence and economic integration
in Central and Eastern Europe. Canadian Journal of Regional Science 38 (1/3): 55–63.
Bradley, J., and Untiedt, G. 2007. Do economic models tell us anything useful about cohesion
policy impacts? GEFRA Working Paper 3. Münster, Germany: GEFRA.
http://www.gefra-muenster.de/downloads/doc/GEFRA-WP-3-2007.pdf.
Brunsdon, C., Fotheringham, A. S., and Charlton, M. E. 1996. Geographically weighted
regression: A method for exploring spatial nonstationarity. Geographical Analysis 28
(4): 281–98.
Camagni, R. 2002. On the concept of territorial competitiveness: Sound or misleading? Urban
Studies 39 (13): 2395–2411.
Accepted version – Review : Economic Geography
______. 2008. Regional competitiveness: Towards a concept of territorial capital. In Modelling
regional scenarios for the enlarged Europe: European competitiveness and global
strategies, ed. R. Capello, R. P. Camagni, B. Chizzolini, and U. Fratesi, 33–47. Berlin:
Springer Science and Business Media.
Camagni R., and R. Capello, 2013. Regional innovation patterns and the EU regional policy
reform: Toward smart innovation policies. Growth and Change 44 (2): 355–89.
______. 2015. Rationale and design of EU cohesion policies in a period of crisis. Regional
Science Policy and Practice 7 (1): 25–47.
Capello, R. 2009. Space, growth and development. In. Handbook of regional growth and
development theories, ed. R. Capello, and P. Nijkamp, 33–52. Cheltenham, UK: Edward
Elgar.
Capello, R., and Nijkamp, P. eds. 2009. Handbook of regional growth and development
theories. Cheltenham, UK: Edward Elgar.
Chapman, S. A., Cosci, S., and Mirra, L. 2012. Income dynamics in an enlarged Europe: The
role of capital regions. Annals of Regional Science 48 (3): 663–93.
Charron, N. 2016. Explaining the allocation of regional structural funds: The conditional effect
of governance and self-rule. European Union Politics 17 (4): 638–59.
Charron, N., Dijkstra, L., and Lapuente, V. 2014. Regional governance matters: Quality of
government within European Union member states. Regional Studies 48 (1): 68–90.
Charron, N., Lapuente, V., and Dijkstra, L. 2012. Regional governance matters: A study on
regional variation in quality of government within the EU. Regional Studies 48 (1): 68–
90.
Accepted version – Review : Economic Geography
Crescenzi, R. 2005. Innovation and regional growth in the enlarged Europe: The role of local
innovative capabilities, peripherality, and education. Growth and Change 36 (4): 471–
507.
Crescenzi, R., and Giua, M. 2016. Leveraging complementarities for evidence-based policy
learning. In EU cohesion policy: Reassessing performance and directions, ed. J.
Bachtler, P. Berkowitz, S. Hardy, and T. Muravska, 21-32. New York: Routledge.
______. 2018. One or many cohesion policies of the European Union? On the diverging impacts
of cohesion policy across member states. Paper 230. London: Spatial Economics
Research Centre, London School of Economics.
http://ec.europa.eu/regional_policy/sources/policy/analysis/sercdp0230_rdd_eu.pdf.
Crescenzi, R., and Rodríguez-Pose, A. 2011. Innovation and regional growth in the European
Union. Berlin: Springer Science and Business Media.
______. 2012. Infrastructure and regional growth in the European Union. Papers in Regional
Science 91 (3): 487–513.
Crescenzi, R., Rodríguez-Pose, A. and Storper, M. 2007. The territorial dynamics of
innovation: A Europe-United States comparative analysis. Journal of Economic
Geography 7 (6): 673–709.
Crescenzi, R., Fratesi, U., and Monastiriotis, V. 2017. The EU cohesion policy and the factors
conditioning success and failure: Evidence from 15 regions. Regions Magazine 305 (1):
4–7.
Dall'Erba, S., and Fang, F. 2017. Meta-analysis of the impact of European Union structural
funds on regional growth. Regional Studies 51 (6): 822–32.
Accepted version – Review : Economic Geography
Dall'Erba, S., and Le Gallo, J. 2008. Regional convergence and the impact of European
structural funds over 1989–1999: A spatial econometric analysis. Papers in Regional
Science 87 (2): 219–44.
Dall’Erba, S., Guillain, R., and Le Gallo, J. 2009. Impact of structural funds on regional growth:
How to reconsider a nine year-old black box. Région et Développement. 30 ( ): 77–99.
Del Bo, C. F., and Sirtori, E. 2016. Additionality and regional public finance–evidence from
Italy. Environment and Planning C: Government and Policy 34 (5): 855–78.
Dijkstra, L., Garcilazo, E., and McCann, P. 2013. The economic performance of European cities
and city regions: Myths and realities. European Planning Studies 21 (3): 334–54.
______. 2015. The effects of the global financial crisis on European regions and cities. Journal
of Economic Geography 15 (5): 935–49.
Dijkstra, L., and Poelman, H. 2008. Remote rural regions: How proximity to a city influences
the performance of rural regions. European Union Regional Focus 1.
http://ec.europa.eu/regional_policy/sources/docgener/focus/2008_01_rural.pdf.
Dunford, M., and Smith, A. 2000. Catching up or falling behind ? Economic performance and
regional trajectories in the “New Europe.” Economic Geography 76 (2): 169–95.
Durlauf, S. N., Jonhson, P. A., Temple, J. R. W. 2005. Growth Econometrics. In Handbook of
macroeconomics, ed. P. Aghion, and S. N. Durlauf.North Holland : 555-577.
Eckey, H. F., Kosfeld, R., and Türck, M. 2007. Regional convergence in Germany: A
geographically weighted regression approach. Spatial Economic Analysis 2 (1): 45–64.
Ederveen, S., Groot, H. L., and Nahuis, R. 2006. Fertile soil for structural funds? A panel data
analysis of the conditional effectiveness of European cohesion policy. Kyklos 59 (1):
17–42.
Accepted version – Review : Economic Geography
Epstein, R. A., and Jacoby, W. 2014. Eastern enlargement ten years on: Transcending the East–
West divide? Journal of Common Market Studies 52 (1): 1–16.
Ertur, C., and Le Gallo, J. 2009. Regional growth and convergence: Heterogeneous reaction
versus interaction in spatial econometric approaches. In Handbook of regional growth
and development theories, ed. R. Capello and P. Nijkamp, 374–88.Cheltenham, UK:
Edward Elgar.
European Commission. 2010. Regional policy contributing to smart growth in Europe 2020.
COM 2010–533. Brussels, Belgium: European Commission.
http://ec.europa.eu/regional_policy/sources/docoffic/official/communic/smart_growth/
comm2010_553_en.pdf.
______. 2014. Investment for jobs and growth: Promoting development and good governance
in EU regions and cities. Sixth report on economic, social and territorial cohesion.
Luxembourg: Publications Office of the European Union.
http://ec.europa.eu/regional_policy/en/information/publications/reports/2014/6th-
report-on-economic-social-and-territorial-cohesion.
Ezcurra, R., Pascual, P., and Rapún, M. 2007. The dynamics of regional disparities in Central
and Eastern Europe during transition. European Planning Studies 15 (10): 1397–1421.
Faggian, A., and McCann, P. 2010. Human capital and regional development. In Handbook of
regional growth and development theories, ed. R.Capello, and P. Nijkamp, 133–51.
Cheltenham, UK: Edward Elgar.
Farole, T., Rodríguez-Pose, A., and Storper, M. 2011. Cohesion policy in the European Union:
Growth, geography, institutions. Journal of Common Market Studies 49 (5): 1089–
1111.
Accepted version – Review : Economic Geography
Fratesi, U., and Perucca, G. 2014. Territorial capital and the effectiveness of cohesion policies:
An assessment for CEE regions. Annals of Regional Science 60 (2): 411–41
Investigaciones Regionales, 29.
Foray D., 2011. Smart specialisation: From academic idea to political instrument, the surprising
destiny of a concept and the difficulties involved in its implementation. MTEI Working
Paper, Integration Process in the New Regional and Global Settings, 269.
Foray D., and Van Ark, B. 2007. Smart specialisation in a truly integrated research area is the
key to attracting more R&D to Europe. Knowledge Economists Policy Brief 1:1–4.
Foray D., David, P., and Hall, B. 2009. Smart specialisation: The concept. Knowledge
Economists. Policy Brief, 9, 1–5.
Fotheringham A. S. 2009. Geographically weighted regression. In The Sage handbook of
spatial analysis ed. A. S. Fotheringham and P. Rogerson. London: SAGE : 243–253.
Fotheringham, A. S., and Brunsdon, C. 1999. Local forms of spatial analysis. Geographical
Analysis 31 (4): 340–58.
Fotheringham, A. S., Brunsdon, C. and Charlton, M. E. 2002. Geographically weighted
regression: The analysis of spatially varying relationships. Chichester, UK: Wiley.
Fotheringham, A. S., Charlton, M. E., and Brunsdon, C. 2001. Spatial variations in school
performance: A local analysis using geographically weighted regression. Geographical
and Environmental Modelling 5 (1): 43–66.
Gagliardi, L., and Percoco, M. 2017. The impact of European cohesion policy in urban and
rural regions. Regional Studies 51 (6): 857–68.
Geppert, K., and Stephan, A. 2008. Regional disparities in the European Union: Convergence
and agglomeration. Papers in Regional Science 87 (2): 193–217.
Accepted version – Review : Economic Geography
Glaeser, E. L., La Porta, R., Lopez-de-Silanes, F., and Shleifer, A. 2004. Do institutions cause
growth? Journal of Economic Growth 9 (3): 271–303.
Gorzelak, G. 2017. Cohesion policy and regional development. In EU cohesion policy:
Reassessing performance and directions, ed. J. Bachtler, P. Berkowitz, S. Hardy, and
T. Muravska. New York: Routledge, 33–54. .
Gorzelak, G., Maier, G., and Petrakos, G., eds. 2013. Integration and transition in Europe: The
economic geography of interaction. London: Routledge.
Gripaios, P., Bishop, P., Hart, T., and McVittie, E. 2008. Analysing the impact of Objective 1
funding in Europe: A review. Environment and Planning C: Government and Policy 26
(3): 499–524.
Giua, M. 2017. Spatial discontinuity for the impact assessment of the EU regional policy: The
case of Italian Objective 1 regions. Journal of Regional Science 57 (1): 109–31.
Hagen, T., and Mohl, P. 2011. Econometric evaluation of EU cohesion policy: A survey. In
International handbook on the economics of integration, ed. M. N. Jovanovic, 324–43..
Cheltenham, UK: Edward Elgar.
Huang, Y., and Leung, Y. 2002. Analysing regional industrialisation in Jiangsu province using
geographically weighted regression. Journal of Geographical Systems 4 (2): 233–49.
Krugman, P. 1996. Urban concentration: The role of increasing returns and transport costs.
International Regional Science Review 19 (1–2): 5–30.
Lagendijk A., 2011. Regional innovation policy between theory and practice. In Handbook of
regional innovation and growth, ed. P. Cooke, B. Asheim, R. Boschma , R. Martin, D.
Schwartz, and F. Todtling, 597–608.Cheltenham, UK: Edward Elgar.
Accepted version – Review : Economic Geography
Le Gallo, J., Dall'Erba S., and Guillain R. 2011. The local versus global dilemma of the effects
of structural funds. Growth and Change 42 (4): 466–90.
Leonardi, R. 2005. Cohesion policy in the European Union: the building of Europe.
Basingstoke, UK: Palgrave Macmillan.
Leung, Y., Mei, C. L., and Zhang, W. X. 2000. Testing for spatial autocorrelation among the
residuals of the geographically weighted regression. Environment and Planning A 32
(5): 871–90.
Lucas Jr., R. E. 1988. On the mechanics of economic development. Journal of Monetary
Economics 22 (1): 3–42.
Mankiv N. G., Romer, D., and Weil, D. N. 1992. A contribution to the empirics of economic
growth. Quarterly Journal of Economics 107 (2): 407–37.
Martin, P., and Ottaviano, G. I. 2001. Growth and agglomeration. International Economic
Review 42 (4): 947–68.
McCann, P. 2016. The Regional and urban policy of the European Union: Cohesion, results-
orientation and smart specialisation. Cheltenham, UK: Edward Elgar.
McCann, P., and Ortega-Argilés, R. 2013. Redesigning and reforming European regional
policy: The reasons, the logic, and the outcomes. International Regional Science Review
36 (3): 424–45.
______. 2015. Smart specialization, regional growth and applications to European Union
cohesion policy. Regional Studies 49 (8): 1291–1302.
______. 2016. Smart specialisation: Insights from the EU experience and implications for other
economies. Investigaciones Regionals 36:279–93.
Accepted version – Review : Economic Geography
McCann, P., and Rodríguez-Pose, A. 2011. Why and when development policy should be place-
based. OECD Regional Outlook 2011. https://read.oecd-ilibrary.org/urban-rural-and-
regional-development/oecd-regional-outlook-2011_9789264120983-en#page5.
Medve-Bálint, G. 2017. Funds for the wealthy and the politically loyal? How EU funds may
contribute to increasing regional disparities in East Central Europe. In EU cohesion
policy: Reassessing performance and directions, ed. J. Bachtler, P. Berkowitz, S. Hardy,
and T. Muravska New York: Routledge, 220–240.
Meloche, J. P., and Shearmur, R. 2010. Emploi dans le secteur public et croissance économique
régionale au Canada: Quelques limites inhérentes aux régressions spatiales [Public
sector employment and regional economic growth in Canada: Some limitations of
spatial regressions]. Canadian Journal of Regional Science, 33 (1), 101–114.
Mohl, P., and Hagen, T. 2010. Do EU structural funds promote regional growth? New evidence
from various panel data approaches. Regional Science and Urban Economics 40 (5):
353–65.
Molle, W. 2007. European cohesion policy. London: Routledge.
Monastiriotis, V. 2014. Regional growth and national development: Transition in Central and
Eastern Europe and the regional Kuznets curve in the East and the West. Spatial
Economic Analysis 9 (2): 142–61.
Nakaya, T. 2014. Semiparametric geographically weighted generalised linear modelling: The
concept and implementation using GWR4. In Geocomputation: A practical primer, ed.
C.Brunsdon, and A. Singleton. London: Sage, 1–5.
Neumark, D., and Simpson, H. 2015. Place-based policies. In Handbook of regional and urban
economics, vol. 4, ed. G. Duranton, V. Henderson, and W. Strange., 1197–1287.
Elsevier,
Accepted version – Review : Economic Geography
OECD. 2009. Regions matter: Economic recovery, innovation and sustainable growth. Paris:
OECD.
______. 2011. Building Resilient Regions for Stronger Economies. Paris: OECD.
______. 2012. Promoting growth in all regions: Lessons from across the OECD. Paris: OECD.
Pellegrini, G., Terribile, F., Tarola, O., Muccigrosso, T., and Busillo, F. 2013. Measuring the
effects of European regional policy on economic growth: A regression discontinuity
approach. Papers in Regional Science 92 (1): 217–33.
Percoco, M. 2017. Impact of European cohesion policy on regional growth: Does local
economic structure matter? Regional Studies 51 (6): 833–43.
Petrakos, G. 2000. The spatial impact of East-West integration in Europe. In Integration and
transition in Europe: The economic geography of interaction. ed. G. Gorzelak, G.
Maier, and G. Petrakos, 38–68. London: Routledge.
______. 2012. Integration, spatial dynamics and regional policy dilemmas in the European
Union. University of Thessaly, Discussion Paper Series 18 (2): 27–40.
Petrakos G., Kallioras D., Anagnostou A. 2007. A generalized model of regional economic
growth in the European Union. Working Paper DYNREG 12. Dublin, Ireland:
Economic and Social Research Institute.
https://www.researchgate.net/publication/23528906_A_Generalized_Model_of_Regio
nal_Economic_Growth_in_the_European_Union.
______. 2011. Regional growth and convergence in Europe: Understanding patterns and
determinants. European Urban and Regional Studies 18 (4): 375–91.
Pieńkowski, J., and Berkowitz, P. 2017. Econometric assessments of cohesion policy growth
effects. How to make them more relevant for policymakers? In EU cohesion policy:
Accepted version – Review : Economic Geography
Reassessing performance and directions, ed. J. Bachtler, P. Berkowitz, S. Hardy, and
T. Muravska New York: Routledge, 55–68.
Pinho, C., Varum, C., and Antunes, M. 2015. Structural funds and European regional growth:
Comparison of effects among different programming periods. European Planning
Studies 23 (7): 1302–26.
Porter, M. E. 1990. The competitive advantage of nations. Competitive Intelligence Review 1
(1): 14.
Postiglione, P., Andreano, M. S., and Benedetti, R. 2013. Using constrained optimization for
the identification of convergence clubs. Computational Economics 42 (2): 151–74.
Ramajo, J., Marquez, M. A., Hewings, G. J., and Salinas, M. M. 2008. Spatial heterogeneity
and interregional spillovers in the European Union: Do cohesion policies encourage
convergence across regions? European Economic Review 52 (3): 551–67.
Rodríguez-Pose, A. 2013. Do institutions matter for regional development ? Regional Studies
47 (7): 1034–47.
Rodríguez -Pose, A., and Fratesi, U. 2004. Between development and social policies: The
impact of European structural funds in Objective 1 regions. Regional Studies 38 (1):
97–113.
Rodríguez-Pose, A. and Di Cataldo, M. 2015. Quality of government and innovative
performance in the regions of Europe. Journal of Economic Geography 15 (4): 673–
706.
Rodríguez-Pose, A., and Garcilazo, E. 2015. Quality of government and the returns of
investment: Examining the impact of cohesion expenditure in European regions.
Regional Studies, 49 (8): 1274–90.
Accepted version – Review : Economic Geography
Rodríguez-Pose, A., and Ketterer, T. 2018. Institutional change and the development of lagging
regions in Europe regions. GSU Paper 1806. Atlanta: International Center for Public
Policy, Andrew Young School of Policy Studies, Georgia State University.
Rodrik, D., Subramanian, A., and Trebbi, F. 2004. Institutions rule: The primacy of institutions
over geography and integration in economic development. Journal of Eeconomic
Growth 9 (2): 131–65.
Sassi, M. 2010. OLS and GWR approaches to agricultural convergence in the EU-15.
International Advances in Economic Research 16 (1): 96–108.
Solow, R. M. 1956. A contribution to the theory of economic growth. Quarterly Journal of
Economics 70 (1): 65–94.
Sotiriou, A., and Tsiapa, M. 2015. The asymmetric influence of structural funds on regional
growth in Greece. Environment and Planning C: Government and Policy 33 (4): 863–
81.
Thissen, M., Van Oort, F., Diodato, D., and Ruijs, A. 2013. Regional competitiveness and smart
specialization in Europe: Place-based development in international economic
networks. Cheltenham, UK: Edward Elgar.
Tobler, W. R. 1970. A computer movie simulating urban growth in the Detroit region.
Economic Geography 46 (1): 234–40.
Varga, A. 2017. Place-based, spatially blind, or both? Challenges in estimating the impacts of
modern development policies: The case of the GMR policy impact modeling approach.
International Regional Science Review 40 (1): 12–37.
Waller, L. A., Zhu, L., Gotway, C. A., Gorman, D. M., and Gruenewald, P. J. 2007. Quantifying
geographic variations in associations between alcohol distribution and violence: A
Accepted version – Review : Economic Geography
comparison of geographically weighted regression and spatially varying coefficient
models. Stochastic Environmental Research and Risk Assessment 21 (5): 573–88.
Yildirim, J., Öcal, N., and Özyildirim, S. 2009. Income inequality and economic convergence
in Turkey: A spatial effect analysis. International Regional Science Review 32 (2): 221–
54.
[insert apendix 1 here]
lnEUFds ln2000 ln2014 Yit lneduc lninstit lnRetD lnAgglo ininvestot lnXpop lnproxicap lnproxiUE1 lnYit
lnEUFds 10,215 0,334 0,185 -0,331 -0,153 -0,008 0,411 0,096 0,253 -0,062 0,074 0,262
ln2000 0,215 10,878 -0,418 -0,010 0,253 0,246 0,607 0,073 0,411 - 0,274 - 0,401 - 0,426
ln2014 0,334 0,878 10,068 0,130 0,023 0,244 0,658 0,018 0,281 -0,379 - 0,227 0,044
Yit 0,185 -0,418 0,068 10,269 -0,484 -0,050 -0,017 -0,119 -0,323 -0,148 0,404 0,970
lneduc -0,331 -0,010 0,130 0,269 10,220 0,467 -0,250 0,025 -0,415 -0,369 -0,042 0,151
lninstit -0,153 0,253 0,023 -0,484 0,220 10,362 -0,300 0,224 -0,012 0,058 -0,341 -0,536
lnRetD -0,008 0,246 0,244 -0,050 0,467 0,362 10,036 0,282 -0,028 -0,329 0,008 -0,114
lnAgglo 0,411 0,607 0,658 -0,017 -0,250 -0,300 0,036 10,092 0,257 -0,139 0,073 0,019
ininvestot 0,096 0,073 0,018 -0,119 0,025 0,224 0,282 0,092 10,014 -0,049 -0,094 -0,126
lnXpop 0,253 0,411 0,281 -0,323 -0,415 - 0,012 - 0,028 0,257 0,014 1-0,069 -0,187 -0,272
lnproxicap -0,062 -0,274 -0,379 -0,148 -0,369 0,058 -0,329 -0,139 -0,049 -0,069 10,010 -0,089
lnproxiUE1 0,074 -0,401 -0,227 0,404 -0,042 -0,341 0,008 0,073 -0,094 -0,187 0,010 10,423
lnYit 0,262 -0,426 0,044 0,970 0,151 -0,536 -0,114 0,019 -0,126 -0,272 -0,089 0,423 1
