The Impact of Public Infrastructure on Productivity: New Evidence for Australia

This paper presents new evidence on the impact of public infrastructure on the Australian economy. The contribution of the paper is three-fold. First, it estimates measures of multifactor productivity for each of the states and territories. Second, it employs a new data set on public infrastructure. Third, the paper applies detailed econometric investigations in an attempt to readdress the crucial econometric shortcomings of earlier studies. The analysis presented here is designed to investigate two widely-debated questions. First, whether aggregate time-series analysis is incapable of capturing infrastructure spillovers to productivity and, consequently, results in incredibly high estimates of infrastructure elasticity. Second, whether state-specific characteristics exhibit a significant role in explaining effects on productivity. To answer the first question, the study applies time-series regressions on both a national and state-by-state basis. Results from this approach confirm the implausibly large effect of infrastructure for the whole economy and four states. To examine the second issue, the paper develops a panel cointegration model which controls for state fixed effects. In sharp contrast with findings from aggregate time-series, results from the fixed effects approach are more plausible and robust to sensitivity tests. In another piece of evidence, estimation of an error-correction model reveals that a long-run identification and modelling of the relationship (i.e. a cointegration) re ects the important positive role of infrastructure on productivity. However, short-run dynamics provide no support for a positive effect which explains why earlier studies which employed differenced data found infrastructure has no discernible effect on productivity. In addition, applying a causality test suggests a long-run unidirectional causality running from public infrastructure to productivity.