Road infrastructure and TFP in Japan after the rapid growth: A nonstationary panel approach

This study investigates the relationship between road infrastructure stock and total factor productivity (TFP) using R-JIP2017, a database of productivity by industry for each prefecture in Japan, which allows us to estimate TFP with considering the quality of inputs. Specifically, using the growth accounting method, we estimated TFP for each industry in each prefecture from 1972 to 2012, after the period of high economic growth. Afterwards, we conducted a panel data analysis to explain the estimated TFP by road stock. The results of a panel unit root test indicated the existence of unit roots in the road infrastructure stock. Therefore, unlike many previous studies, a panel autoregressive distributed lag (ARDL) model was used as the empirical model, considering the nonstationarity of the variables. The results of the analysis indicated that road stock had a positive and significant relationship with TFP at the 5% level in the majority of industries, even after the period of rapid economic growth. Further, we found that the two-way fixed effects model, which does not consider the non-stationarity of road infrastructure stock, could produce misleading results.