Public infrastructure and regional growth: Lessons from meta-analysis

The aim of this study is to synthesize the current literature on infrastructure and growth by determining sources of variation in empirical results by means of a meta-analysis. We use an ordered probit model for investigating changes in the probability of finding negative, positive, and insignificant impacts. The total data base consists of 912 observations from 42 studies conducted between 1995 and 2014. The meta-analytical results show that study characteristics do matter for the magnitude and sign of the variables concerned. We find that studies which employ data from the US are more likely to register a negative impact of public infrastructure on regional growth. We also find that type of infrastructure, research methodology, time span, type of infrastructure measure, and geographical scale affect the outcomes of the primary studies. Studies that take into account interregional, interstate and interprovincial relations have a higher chance of finding negative effects, which gives an idea about the spillover effects of these investments. In contrast, some characteristics like output measure and selection of a particular sector appear to have no effect on obtaining positive, negative or insignificant outcomes. The findings of this study offer new insights to policy makers on the variation in empirical results regarding the relationship between public investment infrastructure and regional growth. (C) 2016 Elsevier Ltd. All rights reserved.