Road infrastructure and TFP in Japan after the rapid growth period: A non-stationary panel approach

This study investigates the relationship between road infrastructure capital stock and total factor productivity (TFP) in Japan using the R-JIP 2017 database, allowing us to estimate TFP by considering the input quality. Using the growth accounting method, we estimate the TFP of each industry in each prefecture from 1972 to 2012 and conduct a panel data analysis to explain the TFP by road capital stock. The second-generation panel unit-root test results indicate the possibility of unit roots in road capital stock. Therefore, we use a panel autoregressive distributed lag model, considering non-stationarity and the specific type of reverse causality. The empirical analysis results show that the elasticity of aggregate TFP to road stock is 0.05 and has positive effects for 11 out of 18 industries, even after the period of rapid economic growth. In particular, the effect of infrastructure tends to be positive during periods of increased value added and TFP. The strongest impact is found for the transport equipment sector, followed by government services, transport, and communications. Furthermore, we find that the two-way fixed effects model and the first-difference estimates could produce misleading results.