The impact of metropolitan structure on commute behavior in the Netherlands: a multilevel approach

The effect of decentralization of land uses on travel behavior remains an unresolved issue in the academic literature. Some US researchers argue that a tendency towards polycentrism is associated with decreasing commute times and distances. Others have, however, suggested and shown the opposite commute times and distances tend to be longer in polycentric than in monocentric urban areas. Using this controversy as a starting point, we analyse how monocentric and polycentric urban structures affect commuting in the Netherlands with data from the 1998 National Travel Survey. Four kinds of urban systems are distinguished: one monocentric and three polycentric systems. In contrast to most previous work, we use multilevel regression analysis to take adequate account of the effects of individual and household attributes. The results indicate that urban structure influences most dimensions of commuting considered here. Yet, individual and household level variables are more important determinants than characteristics of the residential and workplace environment. Gender, household type and their interaction effects remain important determinants of commute behavior in the Netherlands; particularly women in two-earner households commute less than average. Education and income are both positively related to the amount of commuting. Further, the effects of mono- and polycentrism on commuting are more complicated than the literature makes us believe. When individual and household level factors are taken account of, polycentrism does not always result in more efficient commute patterns than monocentric urban structures: in most polycentric urban areas commute distances and times are longer than in monocentric ones. Only when polycentric regions consist of several relatively independent and self-contained development nodes are commute distances shorter than elsewhere. Commute times are in that case comparable to those in monocentric urban areas. The impact of urban structure disappears when commute time is related to the time spent on work activities; the ratio between commute time and work duration is not much affected by the type of urban system in which workers reside. The fact that commute times and distances are not lower in polycentric urban areas may be attributed to the specific situation in the Netherlands: strong spatial planning policies may have obstructed the relocation of employment and housing in close proximity of each other. However, the longer commute in most policentric regions may also indicate that workers and their households not always behave as urban economic theory predicts. In any case, the results show that it is necessary to distinguish several types of polycentric systems instead of merely using a dichotomy of monocentric and polycentric in the analysis of commuting.