Self-containment of employment in Shenzhen : a mobile phone location data approach

Abstract

Self-containment of employment (SCE) measures journey-to-work trips by the percentage of workers who work locally. High SCE encourages non-motorized transport and reduces transport related energy consumption. Previous studies of SCE were mostly based on travel survey data and the spatial analysis unit was traffic analysis zone, which were of different sizes and shapes. The results may however be biased because of the modifiable areal unit problem (MAUP). In this study, mobile phone location data are used instead to examine work trips on a large scale and at a finer geographical scale. Spatial variations of SCE at multiple scales are explored, using Shenzhen as a case study. It is found that SCE in the suburbs is significantly higher than that in the central areas, and SCE decreases from a macro scale to a micro scale. Studies on SCE should pay attention to the MAUP, and use uniform spatial analysis units if possible. Urban planning policies aimed at SCE should also pay attention to the scale issue. SCE and excess commuting are two closely related concepts. SCE mainly reflects intra-zonal commuting at the local level, whereas excess commuting measures commuting efficiency at the city level. Their relationships are explored in this study. Workers identified from the mobile phone location data are disaggregated into secondary sector workers, tertiary sector workers and mixed-type workers at the 1 km grid level according to job types in their work trip destination grid cells. A disaggregated linear programming model is proposed to provide an expected scenario that total commuting cost is at the minimum and SCE is at the maximum. Expected SCE in the expected scenario provides a benchmark for the actual SCE. Expected SCE in the suburbs is significantly higher than that in the central areas. The discrepancy between actual SCE and expected SCE demonstrates how far actual SCE deviates from expected SCE. Studies on the relationship between mixed land uses and SCE are few in literature. In this study, a mixed use index (MUI) in terms of employment is examined to measure land-use mixture for employment. The relationships between MUI and SCE are examined in both industrial and commercial areas, to understand the effect of industrial-residential mix and commercial-residential mix on SCE. The correlation analyses show that industrial-residential mix has a stronger positive effect on SCE than commercial-residential mix. In addition, the correlation analyses show that MUI has a generally positive effect on SCE, but the relationship is marginal in mixed-use areas. As MUI moves from less mixed-use areas to more mixed-use areas, SCE shows a diminishing rate of change. Thus, mixed land uses policies should pay more attention to less mixed-use areas than to areas that are already sufficiently mixed. This study will enhance our understanding of the spatial variations of SCE at multiple scales, as well as its relationships with excess commuting. In addition, an investigation on the effect of mixed land uses on SCE will further enhance our understanding of the relationship between land use and journey-to-work patterns.