Geographically Weighted Regression Analysis of Abandoned Farmland based on the Census of Agriculture and Forestry in Japan

Abstract

Reducing abandoned farmland is an urgent policy issue in Japan. Using the 2015 Census of Agriculture and Forestry in Japan, we explore spatial patterns of abandoned farmland for 1,667 municipalities in 45 prefectures in Japan. Further, we examine geographic factors of abandoned farmland and their regional differences, employing both non-spatial regression and Geographically Weighted Regression (GWR) models. The results reveal the existence of spatial clustering of high and low rates of abandoned farmland. The regression results indicate that higher rates of noncommercial farm households, Kengyo-farm households, lending farmland, next generation of commercial farm households, farmers under 45 years old, and producers with more than 3 ha are significantly associated with lower rates of abandoned farmland. The degrees of these associations vary by region, suggesting the importance of considering regional specific features in advancing agricultural policies.