Trade, FDI and Cross-Variable Linkages: A German (Macro-)Regional Perspective

We analyse the evolution of German Trade and FDI activity within the EU27 using a simultaneous equation gravity approach for imports, exports, in- and outward FDI stocks based on German regional data (NUTS1-level) for 1993-2005. Our approach seeks to explore the main long-run driving forces of both trade/FDI and identify the likely linkages among them. Our motivation for a joint system estimation rests on the observation of a significant cross-equation residual correlation for single equation trade/FDI gravity models, which in turn opens up the possibility for enhancing estimation efficiency in a full information approach. 'On the fly' the simultaneous equation model also allows us to derive a measure for trade/FDI linkages based on the variance-covariance matrix of the system's error term. Adopting both a Hausman-Taylor (1981) IV approach (3SLS-GMM) and a rival non-IV estimator (the system extension to the Fixed Effects Vector Decomposition model recently proposed by Plümper \& Tröger, 2007) our main results are: We find empirical support for the chosen gravity setup as an appropriate framework in explaining German trade and FDI patterns with a prominent role given to trade costs (proxied by geographical distance). Looking at cross-variable linkages we find a substitutive link between trade (both ex-/imports) and outward FDI for the average of German states in line with earlier evidence for Germany, while imports and inward FDI are found complement each other. We also analyse the sensitivity of the results for regionally disaggregated sub-aggregates among the total pool of German state - EU27 country pairs. The results hint at structural differences among the trade and FDI activity of the two German Eastern and Western macro regions on the one hand, and also their interaction with the 'core' EU15 member states opposed to the overall EU27 aggregate on the other hand. Taking the West German - EU27 trade \& FDI relationship as an example, the identified pairwise linkages between the four variables closely ...