Abilio Silveira Garcia Neto

This paper aims to study the determinants of formal employment in the southern region of Brazil and to verify if there is any spatial effect on the behavior of employment for the years 2003 and 2013. Based on the theoretical framework of Marston (1985), it is estimated a model for the selected years, where the dependent variable is the employment-to-population ratio of the municipalities, estimated with RAIS data. The Exploratory Analysis of Spatial Data (ESDA) confirms the presence of spatial dependence of the employment-to-population ratio for the two years. Spatial Autorregressive (SAR), Spatial Error Model (SEM), Spatial Durbin Model (SDM) e Spatial Durbin Error Model (SDEM) models were estimated for 2003 and SEM and SDEM models for 2013. The SDEM model had the best performance for the two years. Based on the estimated models, it is estimated that in 2003, the industrial and services sector has positive effects on the employment-to-population ration of municipalities, while the agricultural sector has a negative impact. Average remuneration also has negative impacts on the level of formal employment. The proportion of literate people in the potentially active population and the population density did not present statistically significant results for 2003. In relation to the spatially-lagged variables, only the variables referring to the industrial sector, average remuneration and density were significant, and the latter two presented negative signs. In 2013, the literate population and density are statistically significant and have positive effects on the level of occupation, evidencing the importance of the human capital in the labor market in the recent period. In 2013, all spatially-lagged variables were statistically significant, indicating a greater interaction among the municipalities of this region.