With over 500 millions of citizens, the European Union (EU) as a whole generates, in
2009, an estimated 28% share of the nominal gross world product and about 21.3% of the
gross world product. Compared to the EU average, the United States per capita gross
domestic product (GDP) is 35% higher and the Japanese per capita GDP is approximately
15% higher. However, there are substantial economical disparities within the EU. On the
high end, Inner London has a per capita GDP equal to EUR 83,200 (334% of the EU27
average), while the poorest region Severozapaden, in Bulgaria, has a per capita GDP of
only EUR 6,400 (26% of the EU27 average).
Regional policy has been at the heart of EU policies since the Treaty of Rome in 1957.
The Treaty’s preamble refers to the need “to strengthen the unity of their economies and
to ensure their harmonious development by reducing the differences existing among the
various regions and the backwardness of the less-favoured regions”. To this goal there are a
number of Structural Funds and Cohesion Funds supporting development of poor regions.
Nowadays with the current financial crisis, EU is also facing a new range of challenges
owing to the enlargement of the Union which is actually strengthening European economic
stability and welfare.
In this thesis we will try to answer to the following questions: Which are the determinants
of productivity dynamics of EU regions? Which policy was adopted by the EU in
the last thirty years to achieve its goals of convergence and competitiveness of European
regions? Has this policy been effective? In particular, the thesis is organized in three
Chapter 1 traces the history of the EU regional policy explaining how it has evolved
from 1957 to 2006. We discuss how the use as unit of observation of the so called NUTS
2 regions (see http://ec.europa.eu) can be not appropriate for the type of issues under
scrutiny. Then we show how we built our database on Structural and Cohesion Funds
for the first three programming periods (i.e. 1975-1988, 1989-1993 and 1994-1999), for
which no databases are directly available by the European Commission. In this we paid
particular attention to the reallocation across regions when information was available only
at multiregional or country level, in the light of the constant changes that the EU regional
policy underwent from 1975 to 1999.
We find that Structural and Cohesion Funds mainly flowed to regions with lower per
capita income and, within the Objective 2 regions, with higher unemployment rate and
employment share in the industrial sector. However, recipients were not always regions
with the least favourable economic conditions and there exists a significantly share of
funds allocated independently of the eligibility criteria. Moreover, we find that over time
there is an increasing discrepancy between the funds committed and those actually spent
by the regions. Finally, taking the ratio of payments on commitments as a measure of the administrative capacity for using EU funds, we find that countries greatly differ in their
efficiency in managing the funds.
Chapter 2 analyzes the impact of the European Union regional policy on the productivity
growth of European regions over the period 1980-2002. In particular, we separately
consider three programming periods (1975-1988, 1989-1993 and 1994-1999), and study
the effects of various types of funds on labour productivity (i. e. Structural, Cohesion,
Objective 1, etc.). In order to capture the main features of the funds, i.e. their size and
composition, we also propose a simple growth model, which is subsequently utilized as a
guide to the empirical analysis.
In all the three programming periods we find that, on average, funding had a positive
and concave effect on productivity growth. In particular, a share of funds on GVA of
10% is estimated to raise the regional growth rate of about 0.9% per year. However, by
separately considering the three programming periods and the composition of the funds
according to different “objectives”, we find that: i) only the funds allocated in the second
and third programming periods, when their amount remarkably increased, had a significant
impact; and ii) only Objective 1 and Cohesion funds had a positive and significant
impact, while the impact of funds devoted to Objectives 2, 3, 4 and 5 appears negative
or non significant. The results are robust to funds’ endogeneity and spatial dependence.
Chapter 3 analyzes the determinants of the distribution dynamics of labour productivity
of European regions over a shorter period, i.e. 1992-2002. We propose a novel methodology
which combines the growth regression approach `a la Barro and Sala-i-Martin, in a
semiparametric framework, with the stochastic kernel approach `a la Quah. In particular,
the distributional impact of a given variable is evaluated by the comparison of actual
and counterfactual distributions and the related actual and counterfactual stochastic kernels
and ergodic distributions. Counterfactual distribution is calculated by the estimated
growth regression, taking the variable to sample average for all regions. The methodology
also allows for measuring the marginal effect of the variable on distribution and for testing
for possible presence of distributional effects in the residuals of growth regression.
We find that initial productivity accounts for a large decrease in the dispersion of productivity.
Instead, country unexplained component (country dummies) has an ambiguous
effect (benefiting regions around but below the average but hurting regions far below),
while employment growth has not any distributional effect. Objective 1 and Cohesion
Funds have a reducing-dispersion effect, but their very limited size produces a negligible
effect on the overall distribution. This also holds for structural change, as measured by
the change in the share of Agriculture sector on total GVA, and Wholesales and Retail;
on the opposite Hotel and Other Market Services result enhancing-dispersion sectors. Finally,
financial sector has an ambiguous effect, mostly benefiting regions with productivity
around but below the average. No variable considered in the analysis appears to affect
the polarization of productivity, but initial productivity.