Abstract:
Empirical studies analysing German import demand functions traditionally report implausibly high income and relativly low price elasticities. Furthermore, estimation results strongly depend on the observation period. Minor variations in the estimation period typically lead to insignificant price terms often displaying the wrong sign. Based on an extensive econometric analysis, the author shows that these problems are caused by the use of highly aggregated activity variables (GDP or total demand). The problem is easily solved if GDP components, namely exports and investment, are used to model domestic economic activity. We find that imports, exports, investment, and a relative import price form highly stable cointegration relationships. The corresponding activity elasticity is clearly below 1 and the price elasticity is highly significant. Changes in the estimation period neither change the impact nor the significance of the determinants of imports.