Estimating a matching function and regional matching efficiencies: Japanese panel data for 1973-1999

In estimating matching functions by using aggregate time series data, an implicit assumption is imposed that search efficiencies are common to all regions in a country. This paper estimates the matching function using annual panel data covering 47 Japanese prefectures from 1972 to 1999, which allows variation of matching efficiencies over regions. We find that the matching function exhibits mildly but statistically significantly decreasing returns to scale regardless of whether unobservable regional heterogeneity be controlled or not. Further, we find a statistical evidence that estimated matching efficiency is negatively correlated with population density and per capita income. This contradicts previous finding that, ceteris paribus, matching is better for higher population density area. We give a verbal interpretation of this finding.