Is There Long Run Industrial Convergence in Europe?

This paper examines whether convergence is occurring at the industry level in 11 EU countries from 1960 to 1993. Both time series and non-parametric (σ-convergence) methods are applied. Using time series analysis we test whether there is within sector convergence towards a common steady state. Although we adopt a very flexible definition of steady state, allowing for two stochastic trends, given by the sector-specific technology and by the country-specific characteristics, as well as a time trend, we find very little evidence for convergence. Results indicate that the deviation of the national trends from the European average trend are persistent, and that EU countries do not respond to the same long run driving processes, but only to the same short run shocks. From σ-convergence analysis we find that convergence in aggregate productivity does not hold uniformly at the industry level: some sectors, such as communications, distribution and non-market services, present strong evidence of convergence, and some others, mainly manufactures, show substantial divergence. However, interestingly, we find that productivity was strongly converging within all sectors in 1960-73, whereas it shows a general tendency to diverge after 1985. We interpret the switch from convergence to divergence as the effect of a process of radical change, driven by information technologies, affecting all industries in the last two decades. We arrive at the general conclusion that convergence is not an inevitable process, but rather a cyclical phenomenon alternating with divergence.