Real convergence in Europe. How robust are econometric inferences?
The paper examines the convergence properties of productivity in fifteen European countries over the period 1960-1997. European productivity is estimated using modified translog production systems with a completely general intertemporal structure. To capture the notion that trade may have an influence on aggregate convergence, imports are included in the production function. Stationarity tests are used to test for convergence around a common long-run trend, including unit root tests, the KPSS and Bayesian tests, as well as testing based on fractional differencing parameter estimation for long memory models. It is found that convergence inferences are sensitive to the type of test employed.