Spatial aggregation and resampling expansion of big surveys: An analysis of wage inequality

Abstract

Income inequality is becoming a growing concern, worldwide, with wage inequality being the root cause of its recent escalation. With the aim of adding to the knowledge on this subject, this paper focuses on the spatial dimension of the problem, an aspect which has received less attention in the literature. We identify the determinants of inequality in wage distribution in Spain at a provincial level using the microdata of the Structure of Earnings Survey (N = 216,769) and estimate their impact from a spatial perspective. Spatial computation of wage concentrations, however, reduces the sample size to just 52 observations, leading to model challenges. To overcome this problem, we adopt a super-population approach and, by exploiting the rich dataset available, increase the instances of the model variables by replicating the actual sampling design employed to collect the data. We apply a re-sampling method that gives us the opportunity to test the impact of spatial dependence and recover provincial fixed effects. Our best model is a spatial-lag of X fixed-effect SLX model. We analyse the impact of workers' personal and employment characteristics, their workplace and their membership of a province on the concentration of wages. Our study finds that a more equitable distribution of wages could be brought about by reducing the impact of workers' training and level of responsibility on their wages and by the promotion of small and medium-sized enterprises. These results highlight the importance of the labour and salary structure in the equitable development of societies.