Entrepreneurship, institutions and growth in European regions: a uniform mechanism

In this paper we present an alternative empirical strategy that sheds light on the importance of institutional quality for regional economic growth. The fundamental problem in this type of research is that institutional quality cannot be measured directly. Existing proxies are typically highly correlated with and endogenous to our dependent variable. We therefore propose a method that is akin to growth accounting, where we refer to the unexplained residual of a production function as total factor productivity. First we run a standard growth regression, including the usual suspects, on 90 European NUTS-2 and 3 regions. The residual variation in regional growth then includes the effect of the fundamentally latent variable that is institutional quality. If this is the case, then variables for which we may assume the effect on growth is strongly moderated through institutional quality, should have a different marginal effect on residual growth at the aggregation levels at which institutions operate. Theory and preliminary empirical evidence suggests that the effect of entrepreneurial activity is strongly moderated by institutional quality. We can thus test the hypothesis that the marginal effect of entrepreneurship on growth differs across regions and countries in our data. We indeed establish that entrepreneurial activity has a positive effect on growth. That is, entrepreneurial activity helps explain the residual variance in our standard growth regressions. However, this effect is not significant when we control for unobserved country differences in a multilevel analysis. We find that most of the cross regional variation is due to between country variation in average entrepreneurial activity, and that there is no longer an effect at the within-country regional level. In a latent class analysis, however, we do not find statistically significant differences in the coefficient across (groups of) regions after controlling for country effects. That is, our data suggests that country level institutional quality is most important for the effect of entrepreneurial activity on growth. After controlling for those differences, regions do not cluster in different classes. This also suggests that the difference between for example rural and urban regions in our sample is not significant, as such regions do not cluster in different groups. Before we draw strong conclusions, however, we note that this may be due to the administrative nature of our regional classification.