12 Anna Iara / Iulia Traistaru
effects. In order to correct for potential heteroscedasticity we estimate White-corrected
standard errors.
The estimation results are presented in Table A2.2. Column (1) shows the results
of the estimation including as control variables the shares of regional employment in
agriculture (AGRIC) and services (SERV), respectively. We find that FDI intensity and
regional manufacturing specialization are positively and significantly associated with
regional growth. On average and other things equal, a 100 percent increase (a doubling)
of the regional FDI intensity is associated with an increase by 1.8 percent of the real
regional GDP per capita growth rate. A similar increase in the Herfindahl index is
associated with an increase of the real regional GDP per capita growth rate by 5.5
percent. Columns (2) and (3) show the estimated results when the measure of regional
openness (OPEN) and subsequently the dummy for regions bordering the EU
(BORDEU) are added. The knowledge spillover effects are still present. The elasticity
of real regional GDP per capita with respect to FDI intensity is around 1.6. In this case,
the effect of regional manufacturing specialization is not significantly different from
zero. Also, the coefficients of the additional control variables (OPEN and BORDEU)
are not significantly different from zero. The adjusted R2 in the cases of the last two
estimated models are lower compared to the adjusted R2 obtained for the first estimated
model. The results shown in Columns (1-3) indicate also that regions with high shares
of employment in agriculture and services are declining.
Column (4) shows the results of the panel model including unobserved time-
specific region-invariant characteristics. The value of the F-test for the joint significance
of the time dummies allows us to reject the null hypothesis of zero coefficients for the
time dummies. The results obtained in this case support the positive and significant
effects of the knowledge spillovers and regional manufacturing specialization. The
elasticity of real regional GDP per capita growth rate with respect to the FDI intensity
measure is 1.7 and with respect to regional manufacturing specialization, around 5
percent, results similar to those obtained in the previous estimations. The last two
columns in Table A2.2 show the results of the estimations when unobserved region-
specific time-invariant characteristics are controlled for. In both cases the values of the
F-test for joint significance of regional dummies do not allow the rejection of the
hypothesis that the coefficients of regional dummies are jointly zero. The values of
adjusted R2 are lower in comparison with those obtained when controlling for time-