Behavioral response to promotion-based public transport demand management: longitudinal analysis and implications for optimal promotion design

Increasing ridership in the urban rail systems in major cities is outpacing their designed capacity. Promotion based demand management can facilitate better utilization of the available capacity of the existing system when the investment and opportunity to expand the system are limited. While several studies address short-term behavioral responses to such promotions using before and after analysis, how behavioral changes are sustained in the long run is also very important as well as differences in response among different user groups. Using an extensive smart card dataset over two years from Hong Kong's urban heavy railway system, this paper explores the longitudinal behavior of passengers in response to a promotion aiming at changing passengers’ travel period from peak to the pre-peak. The approach uses customer segmentation to understand the heterogeneous response of different groups. Users who have high flexibility in schedule and familiarity with the system and travel long distances tend to be more likely to change their travel periods to take advantage of the discount. The longitudinal promotion analysis reveals that 35–40% of passengers who initially adopted the promotion will eventually revert to their previous travel time periods. The results suggest that the promotion designs should be adjusted/renewed over time to sustain the initial response given the attrition of early adopters. Based on the behavioral analysis, an ‘optimal’ promotion design approach is applied to examine the effectiveness of promotion strategies given different behavioral responses over time, heterogeneous group behavior, and constraints on the investment budgets and performance requirements. The promotion design using group-specific response can better target price-sensitive users, hence improves its effectiveness over time, while the design based on the average response shows a significant performance decrease. However, the optimal design schemes using different behavioral responses are relatively consistent in terms of the selected stations for promotion, though some differences exist in the discount levels and time periods for the areas where there can be more early morning travelers. From a design perspective, there is not much difference in the promotion effectiveness regardless of the behavioral response assumed for the design.