ECONOMIC ASPECTS OF COMMUTING
This dissertation studies economic aspects of commuting. It explores, on one hand the mechanism of self-selection into long-distance commuting, return to commuting and, on the other hand, the factors that determine exits from commuting. After investigation of the main bulk of literature about commuting, the research addresses the selectivity of commuters from ex-ante earnings and ability distributions (Chapter 1), monetary return to the commuting distance (Chapter 2) and factors that affects the probability of various exits from commuting spells (Chapter 3) with particular focus on the role of commuting distance and earnings. The analysis uses extensive longitudinal dataset with the precise geocoded information on the individuals' places of work and residence which is based on the administrative registers of Statistics Sweden. The first research paper, titled "Self-selection into long-distance commuting on earnings and latent characteristics", focuses on understanding the nature of selectivity, as it is important factor in interpretation the results of empirical research. In our study we consider two potential dimensions of self-selection: the selection based on latent characteristics and the selection based on the measured earnings before starting long distance commuting. Both dimensions are captured using single model allowing identification of testable hypothesis about the simultaneous selection based on the previous earnings and latent characteristics. In order to conduct our analysis, we apply extensive administrative geocoded dataset with precise individual information including the coordinates of the places of residence and work. We demonstrate the negative selection of commuters from the ex-ante earning distribution. In the same time, our results indicate that the individuals with unobserved traits associated with higher earnings are also more likely engage into the long distance commuting. The second research paper, titled "Return to commuting distance in Sweden", aims to estimate the magnitude of the economic return to commuting and compare the relative returns received by men and women. We apply fixed effect models to deal with individual unobserved heterogeneity that could potentially generate an endogeneity issue. We use a large dataset based on Administrative Registers for Sweden, which gathers detailed information on residential and job location, and indirectly on commuting. Results indicate that individuals receive relatively small compensations for commuting, with higher returns in agglomerations. Moreover, the relative return as a fraction of hourly wage is approximately similar across genders. This last finding provides evidence of similar bargaining powers for both men and women. In our third paper, titles "Hazard from commuting: the role of earnings and distance. The case of Sweden", we estimate the effect of earnings and commuting distance on the probability of exiting from a duration spell of commuting using a discrete time competing risk model. The data set, used in analysis, is based on the Swedish administrative registers from Statistics Sweden and the Swedish Tax Board and covers the period between 2000 and 2009. The problem of endogeneity of individual earnings and commuting distance in determining the length of work-related commuting spells is addressed using two-stage residual inclusion (2SRI). The estimates reveal that the earnings paid by firms have a positive impact on the probability of migration and a negative impact on the probability of job separation. At the same time, greater distance increases the probabilities of migrating closer to the place of work, re-employment closer to the place of residence and separation to non-employment while decreasing the probabilities of migration further away from the place of work and re-employment further away from the place of residence. The results are revealed to be robust in the samples of married and unmarried individuals.