The wage curve model with workforce underutilization and spatial and labor heterogeneity: New evidence using the Brazilian Continuous National Household Sample Survey

Abstract

This study examines the wage curve in Brazil, considering the unique characteristics of the country’s labor market, marked by significant regional and socioeconomic disparities. Using a robust econometric model with fixed effects and control for endogeneity, the analysis covers both the standard wage curve and an extended version that includes additional labor force underutilization measures, such as time-related underemployment (TRU) and potential labor force (PLF). The results indicate that a 10% increase in the unemployment rate (UR) results in a reduction of approximately 0.4% in wages. The wage elasticity is higher for male and young workers, while workers with a college degree show no wage sensitivity with respect to the unemployment rate. Interestingly, model estimation with regional versus group-specific labor underutilization reveals no cross effects of women’s UR on men’s wages, for example. However, there are cross-effects when the PLF and TRU variables are included in the model, as the estimated differences between the two models are statistically significant. Within this comparative exercise, cohorts of workers from low-density areas have their wages affected by labor underutilization observed in high-density areas. Our results also show significant differences for workers in either rural or non-metropolitan areas, while the converse is not true, that is, the wage elasticity in urban or metropolitan areas depend exclusively on the labor underutilization rate observed in the corresponding area, without any significant cross effects from the rural or non-metropolitan areas, respectively. Finally, there is no statistically significant difference between the model with group-specific versus the overall (regional) measure in the case of young workers. In other words, there is evidence that the salary of young workers is not affected by variations in the UR, PLF, and TRU of other age groups.