MEASURING INFRASTRUCTURE PRODUCTIVITY WITH REFERENCE TO LONG PERSISTENT EFFECTS

Abstract

The productivity effects by infrastructure would appear with time delays. The long persistent effects of infrastructure on production output are worth explicitly considering in the econometric measurement of infrastructure productivity. In this paper, the long persistent effects of infrastructure are modeled in the form of the fractional integration of the time series of infrastructure stocks in the past. The regional production function is formulated by an ARFIMAX (Auto-Regressive Fractionally Integrated Moving Averaged model with eXogenous variables) model, and the estimation method is also presented. Our empirical study shows that the existence of long persistent effects in the production function cannot be statistically rejected. The productivity of infrastructure could be underestimated if the long persistent effects are ignored in econometric measurement of infrastructure productivity.