Integration, Regional Specialization and Growth Differentials in EU Acceding Countries
13
specific region-invariant effects, suggesting a weaker explanatory power in the two last
cases.
5.2 Testing for regional convergence
We first estimate the following standard convergence cross-section model:
(3) T1 ln(
GDPCAPj, t+τ
GDPCAP, t
j,t
) = a +blnGDPCAP t +ε tt T
j,t j,t,t+T
t = initial year
T = the number of years in the time interval
j = refers to region j
εj,t,t+T = the error term
The model described by (3) is a standard model testing for absolute convergence8. The
dependent variable is the average annual growth rate of real regional GDP per capita.
The independent variable is the initial level of regional real GDP per capita. The
estimation results are shown in Column (1) in Table A2.4. The positive and significant
coefficient for the initial level of the real regional GDP per capita indicates an absolute
regional divergence in the case of Hungary in the analyzed period. The higher the initial
regional real GDP per capita the higher the regional growth rate. In other words, rich
regions in Hungary grew faster in the period 1994-2000 than poor regions. On average,
other things being equal, a 100 percent increase in initial level of regional GDP per
capita is associated with 5.5 percent increase in the average annual growth rate of the
real regional GDP per capita.
We test further for conditional convergence. We investigate the effect of
regional manufacturing specialization change on regional growth and control for
regional economic structure, openness and the effect of being a region bordering the
EU. We cannot investigate in the cross -section model the knowledge spillover effects
because our measure for regional FDI intensity is highly correlated with the initial level
of the regional GDP per capita (see Table A2.3). We estimate the following model:
8 Barro / Sala i Martin (1995) suggest that regions are more likely to exhibit unconditional convergence
because of similarity of preferences and structural characteristics.