Lee et al. (2017), using the method of Combes et al. (2008), investigated the drivers of local wage inequality in Korea. They found that in big cities with a large diversity of local industrial structures, workers' wage is low, and they concluded that Korean workers seem to accept low wage rates in big cities because they appreciate the high quality of life that big cities can provide. In this paper, we extend Lee et al. (2017) in several important ways.

First, in calculating dependent and independent variables in our two-step regression model's first and second stages, we will utilize administrative data such as Job Administrative Statistics and Business Register Database rather than survey data. This study first utilizes administrative data sets and then uses survey data if some establishments are missing in the aforementioned administrative data sets.

Second, we will control for the labor market structure on the supply side when we estimate the agglomeration economies. Since considerable monetary and non-monetary compensation gaps exist between regular and irregular workers, estimating wage equations without controlling for the labor market duality can produce misleading results. We intend to investigate whether wage rates would still be negatively correlated with the index of urbanization economies even after labor market duality is properly controlled in the estimations.

Third, we will deal with the endogeneity problem relating to agglomeration economies by constructing ideal instrumental variables (lagged variables of agglomeration economies) pertaining to the periods 2007-2014. For example, in Korea, the Innovative City Plan was initiated in 2007 and had been continuously implemented up to 2014. Constructing the lagged variables of agglomeration economies variables may render us an ideal instrument because the Innovation City Plan is exogenous to wage rates for workers at the private sectors, but would influence the level of agglomeration economies and the level of local amenities (Combes and Gobillon, 2015, pp.288).

Fourth, we consider a more refined index of urbanization economies by considering the composition of sub-level industries within one-digit industries. We will consider how diverse two-digit or three-digit industries are distributed across the one-digit industry in the different local areas.