45th Congress of the European Regional Science Association: "Land Use and Water Management in a Sustainable Network Society", 23-27 August 2005, Amsterdam, The Netherlands
Although urban economics theory predicts that households with higher incomes have different commuting time patterns than low income households, the direction of the effect is ambiguous. From a "value of time” perspective, one can argue that high income households may have shorter commuting times because their time is more precious; thus they choose to live closer to the job and are willing to pay for faster modes of travel. However, it can alternatively be argued that they may have longer commuting times, because they desire and can afford more living space and a higher quality of housing than is available or affordable in closer proximity to employment centres. Empirical testing of these alternative hypotheses is not straight forward because income, itself, is determined by commuting time as workers are willing to travel further for higher wages. This reverse causation must be taken into account in the estimation, but this is often problematic due to a lack of good control variables. In the current paper, we employ panel data to overcome these problems. To deal with reversed causation we only select workers who did not change workplace location during the period. For these workers, changes in commuting time resulting from an income change come about either through a change in residential location or a change in travel mode. In addition, by estimating the effect of annual income changes on commuting time changes, the necessity of good control variables is avoided.