The effect of individual tolerance threshold of commuting time on commute mode choice: a case study of Kunming, China

Commuting tolerance constrains people's choice behavior of residential location, workplace location or travel mode. However, the characteristics of tolerance threshold of commuting time (TTCT) and its impact on commuting mode choice have not received enough attention. To reveal the influence of TTCT on commuting mode choice, the semi-compensatory decision-making mechanism was used to modify the commuting utility function. Based on the multinominal logit model (MNL), the constraint characteristic and heterogeneity characteristic of TTCT were considered, therefore, the constrained multinomial logit model (CMNL) and the mixed piecewise logit model (Mixed-PMNL) were established, respectively. The results of an empirical study of 823 commuters in Kunming, China, showed that: Respondents' TTCT is mainly distributed in 30-50 minutes, with a sample mean of 40 minutes. Respondents' commute mode, workplace location, and educational background have a significant impact on their TTCT. The goodness of fit of the CMNL model is higher than that of the MNL model, indicating that integrating TTCT into the commute utility function could help to improve the fitting effect of the commute mode choice model. Furthermore, the goodness of fit of the Mixed-PMNL model is higher than that of the CMNL model, proving that considering the heterogeneity of TTCT into the utility function could help to improve the explanatory power of the commute mode choice model. The utility loss functions of both CMNL and Mixed-PMNL illustrate the penalty effect of TTCT on commuting utility. When the actual commuting time exceeds TTCT, the commuting utility will lose, and the magnitude of the utility loss of different commute modes are different. The utility loss of active commuting is the most severe, while that of car commuting is the least. The more respondents' acceptable commuting cost, the more likely they are to choose cars to go to work. We conclude that it is necessary to consider the impact of the individual commute tolerance on mode choice when formulating the policy to guide car commuters transfer to active modes.