Do you commute further for extra money? : region specific income effects on commuting distances

We start referring to the striking phenomena that over the past decades commuting distances in Germany have steadily risen, although commuting costs increased over-proportionally. This is surprising, as urban economic theory predicts increasing commuting distances especially for higher income, which indicates that either other factors must drive the increase in commuting distances or the income effect may not be as general as so far assumed. We pick up urban economic theory and assume sequential optimization of place of work and residence, occupational specialization and a structure mono- and polycentric sub-labor-markets. Based on these assumptions, we obtain differential predictions on regional income effects and overall commuting distances, which can as well be an explanation for the increasing commuting distances on an aggregate level. To test our assumptions we use both--process data from the German Federal Employment Agency and retrospective life-course data from the German ALWA-ADIAB survey. We draw on the former to show that indeed macro developments of commuter shares and distances match our assumptions and the latter to analyze the determinants of commuting distance on an individual level, focusing especially on the income effect. Identifying the income effect is not straightforward, as employers may reimburse employees for commuting longer distances, thus leading to a causal effect of commuting distance on income rather than vice versa. We account for endogeneity with sensitivity analyses excluding the group of employees who have the highest bargaining power and show that reverse causality does not seem to be a major issue. Income effects do not substantially vary excluding or including this specific group. Descriptively, we can show that commuting distances rise from urban to rural areas. Using a mixed-effects design to disentangle time variant and constant effects for both, individuals and regions, we show that first employees with higher income reside further away from their workplace. Second, we show that a possible change in income results in a small further increase in commuting distance--supporting the assumption of sequential optimization. Third, we show that predominantly employees residing in urban areas increase their relative commuting distance for higher income. In rural areas, where appropriate jobs are rare, commuting longer distances into economic centers seems to prohibit a positive income effect. The results suggest that the positive income effect on commuting, which is found in other studies, does not exist for everyone and heavily depends on the residential location.