Statistical power and productivity effects of transport investments: A critical review

Abstract

Numerous studies have attempted to identify how transport investments affect the overall economy. The results of these studies vary considerably. This variation has been explained by factors such as differences in estimation methods as well as differences among countries and sectors. This paper considers another explanation: low statistical power. By revisiting a previous meta-analysis, this paper shows that accounting for statistical power substantially reduces the estimate. Accounting for statistical power also reduces the relevance of estimation methods as well as country and sectoral differences. These results indicate that a larger data set than typically available is required to overcome the issue of low power. The implication of this is that the existing and, perhaps, future studies using macro data are of limited use to policymakers.