Accounting for the location and allocation of working hours throughout the working week: A discrete-continuous choice model
本文构建了一个离散-连续选择模型，分析一周七天中个体选择工作地点（办公室、居家、混合或休息）及相应工作时长，并考虑通勤时段，为交通规划提供行为依据。
As COVID-19 becomes a close distant memory for many, we are seeing the impact it has had on where working hours throughout the week are being undertaken. It is reasonable to assume that the support for greater flexibility in where work is completed compared to pre-COVID-19 is here to stay and that transport planning needs to move this new pattern of location behaviour centre stage in the revision of strategic transport models. Throughout a seven-day week, we are seeing days in which an individual goes to the main office all day or works from home all day, or undertakes a blended location workday, or does not work at all on a particular day. These four alternatives for each day of the week define a discrete choice model setting which together with the actual hours worked at each location on each day represents a discrete–continuous modelling setting. The paper is interested in identifying where work takes place and the committed hours for each day of the week and treats the seven days of the week as an instantaneous panel. For days where there is commuting involved, we split the discrete alternatives to account for whether commuting occurs during the peak or off-peak period of a day, which is important in terms of the commuting activities in the transport network. We account for the presence of error correlation between the discrete (mixed logit with error components) and continuous (seemingly unrelated regression equations) choices through a selectivity correction for each alternative where it is shown to be statistically significant. A series of direct and cross elasticities provide behaviourally informative evidence on the key drivers of the choice amongst the discrete location alternatives and the continuous choice of hours associated with each. The model system has a very practical feature, in the sense that it can be easily programmed into a strategic transport model system in order to adjust commuting travel activity by mode and time of day in the presence of a more flexible and hence less rigid profiling of when and where work takes place.