How Productive is Public Capital? A Meta-Regression Analysis

The last two decades have witnessed a great deal of effort devoted to measuring the private output elasticity of public capital, but remarkably little consensus has emerged. This paper estimates the meta-output elasticity of public capital and explains the heterogeneity of primary estimates around this value. We develop a simple approach to measure and correct meta-estimates for bidirectional publication bias. Using meta-regression analysis applied to a sample of 67 studies for the 1983{2008 period, we derive an unconditional meta-output estimate of 0.146 and show that the `true' effect is rather heterogeneous. The intraregional short-run output elasticity amounts to 0.085. In the long run, and after accounting for interregional spillover effects, the contribution of public capital to output increases by a factor of three. The results suggest that the high output elasticities found in the early time-series literature are compatible with long-run (cointegrating) estimates found more recently. We also show that the observed estimates are significantly inflated by bidirectional publication bias.