Principal investigator: Bruno Benevit
Authors: Anupriya, Daniel J. Graham, Daniel Hôrcher, Richard J. Anderson and Prateek Bansal
Original title: Quantifying the ex-post causal impact of differential pricing on commuter trip scheduling in Hong Kong
Location of the Intervention: Hong Kong
Sample Size: 1.720.976 trips
Sector: Transportation Economics
Primary Variable of Interest: Travel schedule
Type of Intervention: Price discrimination
Methodology: DID
Summary
With population growth and urban expansion, overcrowding and a lack of adequate infrastructure have exacerbated the problems of congestion and delays in current public transportation systems. In this sense, this study examined the impact of differentiated pricing on the travel schedules of recurring passengers in... Mass Transit Railway (MTR) in Hong Kong. Using the difference-in-differences method, the study identified modest effects of pricing intervention on passenger departure times, with a trend toward earlier travel. Empirical results suggest that fares and occupancy levels are the main factors influencing passenger response to the policy.
- Policy Problem
Congestion in public transportation systems is a growing challenge in urban areas, directly impacting service quality and the efficiency of daily commutes. To mitigate these problems, incentive policies through differentiated pricing have been adopted as demand management strategies. These interventions seek to influence passenger behavior, either by regulating the aggregate volume of trips or by altering the temporal distribution of journeys.
As in many regions around the world, the subway system Mass Transit Railway The Hong Kong Municipal Transport Network (MTR) faces this challenge (ANUPRIYA et al., 2020). To mitigate overcrowding issues, the MTR has adopted a policy... Early bird discount (EBD), providing price incentives for travel prior to the morning peak hours.
Several factors influence passenger response to fare incentive programs. The cost of travel is one of the main mechanisms, with passengers traveling longer distances tending to adjust their schedules to take advantage of discounts. Flexibility in work schedules also plays a significant role, as individuals with less rigid schedules have a greater capacity for adaptation. Reduced occupancy during off-peak hours can represent an attractive option for travelers, as it offers greater comfort and service quality. Finally, it is worth highlighting the existence of a medium-term consolidation effect, which suggests that the aggregate impacts of changes in travel habits tend to increase over time.
With population growth and urban concentration, overcrowding and a lack of adequate infrastructure have exacerbated problems of congestion. Although the effects of fares on aggregate demand are widely studied, there are still gaps in the literature on how these policies affect individual travel time decisions. Understanding these impacts is essential for the development of more effective policies capable of improving demand management and the user experience in public transport systems.
- Policy Implementation Context
In 2014, the MTR in Hong Kong was the ninth busiest metro system in the world, with an annual demand of 1,8 billion trips. During the morning rush hour, most lines operated at maximum capacity. To reduce overcrowding during this period, the MTR implemented EBD in September 2014, offering a 25% discount to passengers disembarking at 29 specific stations between 7:15 am and 8:15 am.
Prior to its implementation, the operator estimated that approximately 105 users would benefit daily, with an expectation that 5 to 7 passengers would avoid traveling during peak hours. Furthermore, it was projected that 2,5% to 3,5% of users would change their travel times, avoiding the critical peak between 8:15 am and 9:15 am.
All stations on the MTR network are equipped with turnstiles. The MTR uses a smart card system that records all trips made on the network. Fares are calculated based on the distance traveled, ranging from HK$3,5 to HK$27,6 for most trips, with additional costs for journeys that cross the port or depart from stations near the border.
- Evaluation Details
The study uses smart card data provided by Hong Kong's MTR, a closed system that records all trips made on the network. The database covers millions of trips, allowing for the analysis of travel patterns and the effects of EBD. The data includes detailed information about each transaction, such as card number, user type (student, senior citizen, etc.), date, time, origin and destination stations, card balance, and fare paid.
For the analysis, the period was divided into pre- and post-intervention phases. The pre-intervention phase included data from July and October 2013, while the post-intervention phase covered July and October 2014. Given that the EBD program was implemented in September 2014, October was chosen as the representative month of the post-intervention period to verify the short-term effects of the policy. According to the authors, this choice is justified by the literature, which indicates that the first month of implementation of fare policies is usually a transition period, while the second month better reflects adjustments in travel habits.
The treatment and control groups were defined based on the travel patterns of regular passengers, considering those who frequently travel between the same origin and destination pairs on 80% of weekdays. The control group consisted of trips made in July 2013 and 2014, while the treatment group included trips from October 2013 and 2014. This division allowed for a comparison of travel patterns before and after the implementation of the EBD, ensuring that the observed differences were attributable to the fare policy.
- Method
The study adopted the difference-in-differences (DID) method to assess the impact of EBD on passenger behavior. The DID method requires the existence of parallel trends in the arrival time distributions in the control and treatment groups in the pre-intervention period, an assumption validated by the authors. DID was applied at both aggregate and disaggregate levels. At the disaggregate level, origin-destination (OD) pairs were observed. The disaggregate analysis focused on identifying which ODs showed significant changes in arrival times after the implementation of EBD, considering the heterogeneity in passenger responses. In addition, specific fixed effects for each passenger were included in the model in order to control for unobserved heterogeneities and increase the precision of the estimates.
Additionally, a detailed analysis of the disaggregated DID results was conducted, investigating the distribution of significant effects among the OD pairs. Factors such as the proportion of regular trips in each OD and the magnitude of changes in arrival times were considered. Spatial analysis was also performed, mapping the geographical distribution of the EBD effects on arrival and departure times. This focused on the most congested stations and lines. This approach allowed for an assessment of whether the policy achieved its objective of reducing overcrowding in the most critical sections of the network.
The study also explored the factors that influenced passengers' response to EBD, using a heterogeneous effects version of the DID model. Variables such as travel cost, average pre-intervention arrival time, and previous occupancy levels, obtained from smart card and train movement data, were included. The analysis sought to understand how these characteristics affected passengers' willingness to adjust their travel schedules.
Finally, the study conducts an analysis focusing on specific variables, such as the cost of the trip, average occupancy, and the presence of routes that cross the port of Hong Kong, which involve higher fares and greater occupancy. These variables were used to explain the heterogeneity in the effects of the EBD, highlighting how factors such as cost savings and expectations of reduced occupancy influenced passengers' decisions.
- Main results
The results of the DID model indicated that the EBD program had a statistically significant, albeit small, impact. Passenger arrival times decreased by an average of approximately 25 seconds. Disaggregated analysis revealed that 58,4% showed some significant change in arrival times, while only 34,9% of OD pairs showed significant reductions. This indicates that the effect of EBD was heterogeneous, concentrating on a specific segment of passengers.
Detailed analysis of the disaggregated results showed that the significant effects of EBD were not limited to high-demand OD pairs. Many pairs with lower travel volumes also showed relevant changes, suggesting that insensitivity to EBD was not directly linked to passenger numbers. Furthermore, the greatest impacts were observed on trips with average arrival times between 8:10 and 8:40, indicating that the policy was most effective at the beginning of the peak period. Conversely, some passengers postponed their trips to the peak period, possibly due to the expectation of reduced occupancy after the implementation of EBD.
Regarding the factors that influenced the response to the EBD (Emergency Daily Travel), it was observed that the cost of the trip and previous occupancy levels were decisive. Outbound (OD) pairs with higher costs and greater occupancy showed more positive responses to the policy, with greater reductions in arrival times. Trips crossing the port, which involve higher fares and greater occupancy, had an average reduction of 19 seconds more compared to other trips. These results suggest that cost savings and the expectation of lower occupancy were the main motivators for passengers to adjust their travel schedules. The policy proved more effective on more congested routes, partially achieving its objective of redistributing demand.
- Lessons in Public Policy
In this article, the authors conducted empirical analyses to evaluate the impact of the EBD program on the passenger behavior of the Hong Kong metro, focusing on the redistribution of demand during peak hours. The results indicated that EBD generated a small but significant reduction in arrival times, averaging 25 seconds. Specifically, disaggregated analysis showed that only one-third of OD pairs experienced significant reductions. Furthermore, the authors identified that factors such as travel cost, crowding levels, and specific route characteristics, such as harbor crossings, influenced passenger response to the policy.
The evidence from this study helps to understand how fare incentive policies can affect travel patterns and crowding in public transport. The results suggest that interventions such as EBD (Economically Differentiated Bus) are more effective on more congested routes and for passengers who value cost savings and reduced crowding. This information is useful for public policy makers, highlighting the importance of considering passenger heterogeneity and the specific characteristics of the transport network when designing interventions. Consolidating policies that reduce crowding and optimize travel schedules can contribute to improving the efficiency and quality of urban transport systems.
References
ANUPRIYA et al. Quantifying the ex-post causal impact of differential pricing on commuter trip scheduling in Hong Kong. Transportation Research Part A: Policy and Practice, v. 141, p. 16–34, nov. 2020.
