Economic Structure and Regional Performance in Germany, 2002–2007

Abstract

This paper explores the impact of industrial clusters on regional growth at the German labour market region level using a regional convergence model. Based on the results of an exploratory study of the geography of German industrial clusters, we are able to differentiate the impact of industrial clustering from a horizontal and a vertical perspective while taking regional convergence into consideration. The results indicate that in addition to an all-German process of convergence, a specific East German one can be identified. The different types of industrial clusters show mixed effects within this framework. While vertically isolated industrial clusters have a negative impact on regional growth in this period, positive growth effects can be identified when industrial clusters show an intra-regional vertical interconnectedness.

Notes

One drawback of using input–output methods for cluster identification is the limited availability of disaggregated input–output tables at the regional level. This leads to the assumption that inter-sectoral relationships that are similar to those at the national level exist at the regional level (Spencer et al., 2010; Titze et al., 2011).

In contrast to administrative units such as districts, labour market regions are defined on the basis of economic interdependencies and commuter flows. This allows regions of this type to be seen as relatively independent economic units (Eckey et al., 2008).

Longer time periods are often characterized by structural breaks (in the case of Germany, for example, the fall of the wall in 1989). Usually, periods of at least 5 years are used to carry out studies on the convergence of regions at the regional level (Eckey et al., 2008).

While the input–output framework used involves the German classification scheme from 2003 (which is equivalent to NACE rev 1.1), the employment and establishment statistics used go back to the German classification scheme from 1993. At the aggregation level of 59 industrial sectors, the differences are marginal between these two classification systems. Moreover, the input–output table for 2002 is not available for the older classification scheme.

The intercept dummy is 1 for a region within East Germany, otherwise zero. Berlin is attributed to West Germany.

The slope dummy results from the interaction between the GVA per capita in 2002 and dummy East.

The estimation of the labour market region of Berlin does not correspond with its spatial effects on surrounding regions because of its NUTS 1 classification and requires, therefore, a dummy Berlin, which controls these functional relationships. The dummy is 1 if the labour market regions are subareas of the functional area of Berlin, otherwise zero.

The number of companies in a region represents a similar indicator.

This can lead to the insignificance of the correlated explaining variables.

Dynamic aspects can be examined only to some extent with the applied approach. Furthermore, dynamic information about industrial clusters is not available at the moment. Given this, we are not able to include a panel structure in the econometric model about industrial clusters.