The data used for the generation of the empirical research, was gathered by a matched employer-employee dataset, maintained by Statistics Sweden. The data from this institute was composed by all the employees in Sweden during the period 2002 to 2008. When constructing the data, the authors assigned the employees to their work establishment regarding the sector, occupation and location. This information gives about 2.4 million employees with a mean population size of just short of 2 million yearly observations. The data included different characteristics of each employee and their employer. To define the spatial economic density, the basic spatial unit used in the analysis is the municipalities in Sweden, which there are 290 in total. Specifically, the data informs where every workers it’s situated along the municipalities. The municipalities have a limited size and there is an important interaction between the borders. With this context, they think of the total density of a municipality r as the sum of municipal, regional and extra-regional accessibility to total wage-earnings, W (See exhibit 1: explanation of the variables):
Here, the total amount of wage earnings reflects the degree of economic activities and the accessibility to economic activity is the measure of spatial economic density. The authors explain that municipal density is simply each municipality’s total wage earnings weighed exponentially with travel time distances by car between zones within the municipality. When it comes to establish controls, they include a set of characteristics that may affect the worker’s wage, such as experience, schooling, education specialization, immigrant status, sex, tenure, job change, sector affiliation, municipal and regional density, among other (See exhibit 2). Another important factor included is the employment size of the establishments in Sweden at which the employees work. To differentiate the level of skills on the workers they used the classification scheme developed by Becker (2009), which reports the fraction of non-routine job tasks associated by each ISCO-88 occupation (See exhibit 3). Here they give a list of the occupation titles according to the scheme and expose a percentage of the non-routine tasks that are necessary for each occupation.
After this, they present data about the mean wage, fraction of graduates, meaning experience and the fraction of workers working in any of the three largest regions in Sweden for all workers as well as for occupations with high and low fractions of non-routine job tasks, respectively to analyze whether the occupations with higher necessity of non-routine skills are more attractive to be done in metropolitan areas or not (See exhibit 4).
In the next part of the empirical analysis, the main target is the relationship between density as measured by accessibility to total wage earnings and the wages for the workers. Exhibit 5 plots the logarithmic relationship between mean wages of the workers and the sum of the density measure. Here we see that workers in denser municipalities have higher average wage. Once the analysis of the empirical data is done, they apply the empirical model that will be show next