Osaka University Knowledge Archive : OUKA
Social and economic problems taking place in one region are likely to be connected to those in other regions and vice-versa. If this connection is ignored, the conducted fact-findings might be substantially erroneous. It is important for researchers to account for the regional dependencies in dealing with social and economic issues. This thesis applies the techniques of Spatial Econometrics to several topics in Labor Economics. In Chapter 1, we summarize the four empirical articles included in the thesis, emphasizing the importance of the application of Spatial Econometrics. Chapter 2 deals with spatial dependencies to estimate a job matching function using five different spatial weight matrices. The results indicate a significant spillover effects, regardless of the fitted spatial weight matrix. That is, job creation depends on the stock of unemployed and vacancies both within and across prefectures in Japan. It is also shown that the spillover effects are confined within local labor markets and best approximated using the contiguity weight matrix. The estimation of the matching function shows that job seekers in Japan have significantly shifted their search radius across prefectures. Chapter 3 applies Spatial Econometrics to trade, with the World Input-output Tables, and examines the impact of the participation of the global value chains (hereafter, referred as GVCs) on employment within and across industries in the case of Turkey. We find that job creation depends not only on the GVCs integration within their own sectors but also on changes in the GVCs participation within sectors connected via the weight matrix. This indicates the existence of significant spillover effects across sectors. Moreover, employment in the manufacturing sector benefits most from the GVCs integration via backward linkages. However, employment declines with higher backward linkages in neighboring sectors. The GVCs integration via forward linkages appears to displace labor in the service sectors. Chapter 4 investigates the inter-industry diffusion in productivity, allowing the impact of the GVCs participation on productivity in Turkey. The chapter uses the same World Input-Output Tables as is used in Chapter 3 to model inter-industry input-output relations. It is shown that a stochastic shock in productivity in one industry significantly transcends and boosts productivity in other industries through input-output dependencies. Moreover, considering all industries together, productivity significantly declines with backward linkages within own sectors. However, productivity increases with forward linkages both within own sectors and across sectors. A sectoral analysis of the GVCs effects on productivity reveals that, in a manufacturing industry, backward-linkage effects are negatively associated with productivity not only within own sectors but also across sectors, whereas productivity in service sectors rises together with forward linkages within and across service sectors. Chapter 5 investigates the determinants of regional homicide rates in Japan, utilizing Spatial Econometrics to account for the regional correlations. We find that deteriorating labor market conditions are significantly related to the homicide rate. Particularly, the regional homicide rate rises with its unemployment rate. Moreover, the demographic composition plays a substantial role in explaining the level of homicide in Japan. Specifically, an increase in the share of the elderly group and a decrease in the share of the youth group is significantly associated with crimes of homicide in Japan. In addition, we find that there exists a positive total effect of the number of police stations per capita on the homicide rate. The estimates of the Spatial Econometric model convey a negative spatial diffusion, which supports the hypothesis of Spatial Crime Displacement in the case of Japan. That is, policies aiming at tackling crimes in one region are not necessarily efficient, displacing crimes to other regions rather than eradicating them. The whole chapters show that we cannot overlook the existence of spatial correlations, in dealing with social economic issues, and that ignoring these spillover effects across regions and industrial sectors cause the estimates to be biased.