Rice productivity tends to become similar among neighboring regions because of spillover effects, but few previous studies considered such spatial dependence in Japanese rice production. Measuring such spatial influences is important for policy making. This study analyzed the socioeconomic and climate factors for rice total factor productivity (TFP) by considering spatial dependence caused by technological spillover and other latent factors. Results of the spatial auto-regressive (SAR) model and spatial error model (SEM) showed the following. First, effects of spatial dependence in rice TFP level remained even when climate factors, which affect TFP in broader regions than prefectural areas, were controlled. Hence, some biases occur in ordinary least square estimations commonly used in the panel data analysis. Second, the elasticity of TFP with respect to temperature via rice yield was positive until 2000 in most prefectures where summer temperatures were under the threshold level to cause changes in yield. The elasticity of TFP was high and remained positive until the 2050’s in northern Japan. However, in middle and western regions and Kyushu, the elasticity was predicted to become negative after 2030. The same tendency was found in TFP elasticity for effects of temperature on rice quality. The threshold value for temperature in rice quality was lower than rice yield, so an increase in temperature depresses TFP in middle and western regions and Kyushu even under the present situations. Only a few prefectures in northern Japan can achieve an increase in TFP at present. The TFP elasticity against rainfall is constantly negative. Impacts of long rain were stronger than floods. As shown by these tendencies in TFP elasticity, future climate change negatively affects rice TFP via temperature and rain. Third, the ratios of indirect effects from neighboring prefectures by causative factors against total effects on TFP changes, measured by the SAR model, were high in northern Japan where neighbor TFP levels were relatively high. Contrarily the ratio was low in middle-and-western Japan where neighbor TFP levels were low. Even so, the percentage of indirect effects compared with total effects was approximately 10 to 25% in all regions and throughout all periods. As such, spatial econometric models are useful to show results without spatial biases. Based on these results, technological development of rice production and provision of precise climate prediction to farmers are important to mitigate the influences of climate change.
- Indirect effects from neighboring prefectures
- Spatial Auto-Regression（ SAR） model
- Spatial error model（ SEM）
- Technological spillover
ASJC Scopus subject areas
- Environmental Science(all)
- Social Sciences(all)