Introduction to crowding

As anyone who has traveled on a crowded metro can tell, it is one of the biggest detractors of travel comfort that you can have. It leads to more fatigue, less comfort, loss of polychronic time-use (you cannot work or read a book), and leads to social security concerns. Often, it is known that many earlier trains might have lower occupancy, but departing an hour earlier is most of the time not an option. 

In modern transport science, peak-hour overcrowding in public transport is also increasingly recognized as a significant problem. It not only contributes to urban externalities that reduce the utility of public transport users but also increases the generalized costs which are the agency costs and user costs. It reduces the attractiveness of public transport and increases the opportunity costs due to the underutilization of transport resources during off-peak hours. The crowd levels should be considered in transport policy and cost-benefit analysis (Haywood, & Koning, 2013). During the busiest part of rush hour or the “hyper-peak”, these issues are exacerbated (Keolis, 2022). 

The rest of this white paper follows the following structure. We start with a quick scientific background on crowding and nudging. We then show how nudging and pricing incentives work in other public transport networks. After this, we apply the previously collected details on the case study of Lyon’s metro line A, showing potential cost-savings in societal- and welfare benefits, and additional revenue from increased ridership. These numbers are just to give an idea, and more detailed vehicle-specific occupancy can give more realistic numbers. 

Quick scientific background on crowding

Rush-hour overcrowding in public transport is a significant issue that can lead to various negative effects on passengers and the transportation system. In-vehicle crowding has been linked to stress, health problems, and increased anxiety. Overcrowding can also lead to a higher perception of risk to personal safety and security. Overcrowding can also be exacerbated by and create further service disruptions, delays, or cancellations, leading to increased waiting times for passengers and reducing service quality. This can have effects on job quality and life quality (Metis, 2010). During special occasions and big events such as concerts, football games, and festivals, the adverse effects of overcrowding could get far worse. 

Outside of personal perceptions, over-dimensioning a transport network to cope with peak demand is increasingly perceived as inefficient. Bus and train drivers often have a minimum shift length of 4 hours, sometimes longer than the rush-hour exists. The fleet size and network capacity also have to be scaled up to accommodate peak-hour demand, which leads to an over-supply of empty seats being moved around during peak shoulders and off-peak hours (Eriksson 2023). 

With a growing trend in public transport ridership as a result of the collective effect towards promoting sustainable mobility, this increase in ridership is outpacing capacity expansions in many transit systems. In Paris, the amount of metro ridership has increased by 10%, while supply has stayed the same (Haywood & Koning, 2013) while it also increased by 9.9% in Lyon from 2014 to 2019. While important to existing commuters, finding a method of travel with low crowding is said to be one of the most important factors determining potential public transport users, instead of existing transport users (Olio, Ibeas, and Cecin 2011).

With some public transportation systems now facing significant crowding and thus, issues with bottlenecked capacity, transit agencies are showing an increased interest in the (traditionally car-focused) concept of travel demand management. The term “travel demand management” is used to describe strategies that increase transportation system efficiency by altering demand patterns, rather than increasing the supply of service.

Travel demand management strategies encourage demand to spread more evenly throughout the network or over the day, allowing agencies to make better use of their resources and improve their customers’ experiences. In many cases, effective travel demand management policies and strategies are also quicker and less expensive than adding capacity via new vehicles or rail infrastructure. Both theory and practice show people are sensitive to temporal fare distinction, where off-peak travels are encouraged with lower fares. For the Sydney rail case, the most effective combination was a 30% discount during the peak shoulders and a 30% surcharge during the peak hours, forecasted to reduce peak loading by 10%. A survey commissioned by Passenger Focus found that over 40% of commuter rail riders could arrive outside of the peak, with almost all preferring to arrive earlier. Over half of respondents said that reducing fares by 25-30% could encourage them to travel earlier (Halvorsen et al., 2016). 

Managing price based on demand and crowding levels, thereby making a distinction is however politically unpopular, and sometimes impossible. When surcharging the price during the hours is not a viable solution, we have to consider alternative ways of changing behaviour. This means we have to step away from the assumption that people only care about price and time and make irrational decisions based on their beliefs, knowledge, and routines: we get into behavioural transport science. 

In many earlier transport models, passenger decisions between various multi-modal trip options are calculated using just a combination of travel time and end-user price (Small and Verhoef, 2007). Increasingly, however, it is known that comfort, security, and entertainment can make a large difference in how people decide to travel. So much in fact, that transport behaviour science has created a time multiplier (TM), indicating how much worse time feels when standing in a crowded carriage, compared to an empty one. A TM changes from place to place and depending on the level of crowdedness, but indicates that time standing in crowds can feel between 1.7 to 2.6 times longer than time seated comfortably (Wardman & Whelan, 2010; Haywood & Koning, 2013).

Instead of price incentives, providing the user with information and rewards can be an effective method for travel demand management. In behavioural science, rewarding good behaviour has worked much more effectively than punishing bad behaviour. In our case, we define good behaviour as people previously making their journey during peak times, moving their trip to peak shoulders. 

Scientific background on nudges 

Nudging is a method of behavioural intervention that seeks to change the way options for decisions are presented within the environment in which we make these decisions. It was famously coined by Sunstein and Thaler in 2008, in their book Nudgd, where they propose a patriarchal liberal idea of choice architecture. 

Nudging does not restrict an individual's freedom to choose and does not operate with threat or severe economic consequences. Instead, nudging tries to guide behaviour in a certain direction by reframing, supplying new information, or providing social feedback. It states that there are always choices to be made, and not choosing is also a choice, only the choice by the choice architect. Sunstein and Thaler break down Nudges into the following six steps.

1. Incentives Getting a reward for making a decision, any decision

2. Understand Mappings A choice architecture assumes to know the consequences for the user, which is important additional information not always directly implied by the choice. This can be an early appointment where you don’t care about other factors, but you still change behaviour.

3. Defaults A choice architect makes the default choice, such as the recommended route, or the routine train. 

4. Give Feedback In order to make a correct choice, the feedback loop needs to be short enough to see the results of your actions.

5. Expect Error Humans are fallible: errors should never lead to critical unwanted situations, such as not arriving at the destination.

6. StructureTo make a decision, information has to be easily digestible and readily available, in a fixed structure.

One example of combining these factors into a single nudge could be as follows: 

Nudging in sustainable mobility refers to the use of behavioural interventions to encourage people to make more sustainable mobility choices, such as walking, cycling, or using public transport or changing their travel behaviour to contribute to the overall sustainability of the system. The use of nudging in sustainable mobility has evolved, with the advent of digital tools and the collection of user data making personalized nudges possible. For example, data concerning people's past choices, preferences, and current environment or situational components can be integrated to create situation-aware digital nudges. 

Overall, nudging has become an increasingly important tool in promoting sustainable mobility, as it can help remove barriers and address psychological factors that hinder people from making more sustainable choices. By providing personalized and situation-aware nudges, nudging can be an effective way to encourage sustainable mobility choices and promote positive change in urban environments.

The case of DyMoN

The Sustainable Mobility Handbook developed as a result of the research project DyMoN (Dynamic Mobility Nudge) provides examples of potential behavioural interventions for sustainable mobility, targeting capability, opportunity, and motivation.

1. Targeting Capability: This involves providing resources and support to enhance individuals' ability to engage in sustainable travel. Examples include increasing the availability of bike-sharing services or making public transport schedules easier to understand.

2. Targeting Opportunity: “Opportunity” describes factors that lie in one’s environment that facilitate sustainable behaviour, including both an individual’s social and physical opportunities.  Opportunity is not necessarily within the person’s control; for example, in the context of peak-hour travel, this can be the inflexibility in work timings or living in suburban areas where there is much less frequency of public transport, etc.

3. Targeting Motivation: These interventions aim to motivate individuals to choose sustainable mobility options or change their travel behaviour by offering incentives or valuable real-time information for avoiding peak hours or using shared bikes, such as valuable information or incentivized rewards. This is where Mobility Meters provides the service, targeting users with valuable real-time information on crowding and delays and offering alternate suggestions to target the user’s motivations. 

Understanding the users and their current travel behaviour would be key in identifying users who can be nudged and over time, by understanding what types of interventions or nudges they are sensitive to, a personalized and effective nudging technique can be used on the passengers to effectively reduce overcrowding and improve sustainable mobility choices in general.

In addition to these specific examples, the DyMoN handbook emphasizes the importance of understanding the barriers that hinder individuals from executing sustainable mobility behaviours and designing interventions to address these barriers. Overall, behavioural interventions work by addressing capability, opportunity, and motivation to promote sustainable mobility choices and create a positive impact on urban environments.

Prior Strategies to Reduce Overcrowding

Transport planners are well aware of the need to manage both public transport supply (e.g., the number of buses, and trains) and demand (e.g., commuters) in concert. As long-term demand increases, supply needs to increase to keep pace.  For transient peak demand, it can be very expensive to add supply and, should supply be increased, latent demand moves in to fill it. Transport planners, therefore, understand that peak demand must be actively managed (Small and Verhoef, 2007). Transit agencies worldwide have tried several demand-management techniques to reduce peak crowding, including peak pricing or off-peak discounts, better information, or employer partnerships to encourage more schedule flexibility among workers. Pricing is considered a particularly effective method of reducing peak period congestion but may be politically difficult to implement (Greene-Roesel et al., 2018).  

Another approach, that of using incentives and gamification to reduce peak travel, is a potentially more palatable alternative (or complement) to peak pricing because, rather than penalizing people for traveling during peak times, it rewards them for traveling during less congested times. One prominent example is the Singapore Land Transportation Authority’s (LTA) INSINC program (now TSR—Travel Smart Rewards) (SLTA, 2023). Since implementing various Travel demand management strategies, the LTA has seen significant shifts in travel patterns during peak hours. 

One game-changing policy is offering free travel before 7:45 AM on select routes which reduced the MRT urban rail trips between 8:00 and 9:00 AM by about 7.5% with most of this shift seeming to have been to the few minutes just before the peak hour. The Travel Smart Rewards started in 2014 program alone had encouraged a 7.5% trip adjustment from peak to off-peak among its users. The behaviour remained consistent among reward program members who were already traveling during off-peak times. The evaluation also concluded that shift effects were greater for those with friends in the program and that the special offers had a positive effect in encouraging additional shifts. These well-orchestrated efforts of Singapore's LTA show that good policy design, combined with incentives and gamification, can lead to sustainable commuting practices (Greene-Roesel et al, 2018; Halvorsen et al., 2016).

Melbourne’s Free Before 7 campaign which made rail trips before 7:00 am free to us had an effect of increased ridership of 41% during the peak shoulders (the hours climbing towards the peak hours) and the program found that the program led to enough savings in capital and operating costs to cover its costs incurred. 

The Tokyo Metro has also implemented an incentive program to encourage off-peak commuting. During certain times of the year, travelers earn points based on how many hours they take the train before or after peak hours. Once the traveler has earned sufficient points, they can enter a lottery to win cash rewards.

In Hong Kong, to reduce the peak-hour overcrowding, they introduced an Early bird proportion. This fare differential strategy offers a 25% discount for trips to 29 heavy rail stations that end in the pre-peak hour of 7:15 to 8:15 am. This strategy decreased the proportion of peak-hour trips by about 3%. The effect of the Early bird promotion is shown below in the figure.

Figure 3. Effects of the peak-shoulder discounts in Hong Kong

Halvorsen (2016) argues that these results show that rather than using a single discount at a specific time slot, an agency could gradually taper its fare differential, giving a higher discount farther from the peak. This could better discriminate between time- and cost-sensitive users and encourage more shifting from the peak-of-the-peak. An agency could also provide more personalized information to its users. However, providing such dynamic fare differentiation would not be practical and feasible as it reduces transparency and increases uncertainty among travelers; however, a micro-incentive system like Mobility Meters can have a dynamic reward differentiation and they can be personalized as Halvorsen recommends.

Expected benefits for the City of Lyon in France

These three factors are estimated with the reduction in overcrowding.

1. Societal and Welfare Benefits

2. Increased Ridership

3. Increased Revenue

Social and Welfare Benefits

The Welfare costs refer to the negative impacts or costs associated with public transport (PT) crowding. The study estimated the social benefits of reducing PT crowding in terms of reduced disutility of time and increased passenger comfort, which can lead to increased PT usage and reduced congestion on roads. The disutility of time refers to the negative impact or cost associated with the time spent on a particular activity. The disutility of time is used to quantify the impact of crowding on passengers' perceived travel experience and costs. The disutility of time, influenced by in-vehicle density, was found to increase the value of travel time by 34% during peak crowding periods in Paris subways. At the most crowded conditions of 6 persons per square meter, one minute traveling in this condition is equally perceived as traveling for 1.6 minutes. This ratio of 1.6 is defined as a time multiplier for crowding in Haywood & Koning 2013.

In the research by Haywood & Koning 2013, around 69% of people in Paris are willing to increase their travel time by up to 8.3 minutes to avoid a crowded metro. On average, people are willing to travel 3.3 minutes longer per trip to avoid crowding. This was used to calculate the willingness to Pay (WTP) which resulted in a value of 0.66€ per trip according to this paper. Adjusting this number to inflation, it is valued at €0.904 in today’s value.

Using another dataset collected by a consulting firm, Factual last year, which is shown in the figure below, we calculated the willingness to pay for avoiding crowded trips in Lyon. The data collected shows the minimum discount people expect to have a 50% or higher probability to accept an incentivized off-peak hour trip.

Using this data, adjusting for GDP per capita and average ticket prices of the city, we performed linear regression to arrive at the approximate value of willingness to pay for the City of Lyon. As you can see in the linear regression below, the predicted willingness to pay per trip is 0.9€ which is very similar to the willingness to pay calculated for Paris. Please note that the methodology used by  Haywood & Koning 2013 takes into account the income levels of regular commuters in Paris to arrive at the value. Assuming the income levels of regular commuters remain similar, this value of 0.9€ per trip is validated confidently.

Using this value and the same methodology, the welfare cost savings for the entire transport network of Lyon is calculated if 10% of all peak hour trips are displaced which is approximately 5% of total trips made by 2.5% of the total users. The total welfare cost savings are € 22,600,000. 

The figure above shows how the graph looks like when just 5% of the trip is distributed from peak hours to off-peak hours gradually in the metro line A at the stop ‘Charpennes’ in Lyon. 

Increase in Ridership and Revenue

With displacing 5% of the trips and a 10% reduction in peak-hour trips and keeping the linear relation between willingness to pay and occupancy rate, 10% of the willingness to pay contributes to the attractiveness of public transport. Since willingness to pay is € 0.9/trip, we assumed a 0.09€ discount on public transport tickets due to the improved crowd levels and the increased attractiveness. Considering the average ticket price of €1.6 in Lyon, the perceived ticket price would be €1.51 due to increased attractiveness. This perceived discounted ticket price results in an increase in ridership of 2.8% in public transport. This is based on the demand elasticity value determined by Kholodov & Jenelius 2021 on public transport fares. An increase in ridership will directly contribute to the increase in revenue which will also be a 2.8% increase.

Summary

With just 5% of trip displacements, performed by 2.5% of the users,

• Social cost reduction = € 22,600,000 per year

• Increase in peak-hour capacity = 10%

• Estimated Increase in Ridership = 2.8%

• Estimated Revenue growth = 2.8%

Other KPI Improvements

Other factors that will improve as a result of reduced overcrowding can be analyzed and quantified with further research and more data.

1. Costs saved due to delaying fleet and infrastructure investments.

2. Modal Shift from Cars to Public Transport and the results.

3. Improvements in overall occupancy rate within the buses or trains.

4. Opportunity cost reduction due to improved utilization during off-peak hours.

Mobility Meters

Commuting data reveals that over 76% of morning peak trips are due to just 20% of commuters (6). By targeting these commuters with personalized nudging and micro incentives, we can effectively reduce peak-hour travel gradually. A survey conducted by our team and peer-reviewed Research studies show that 20% of the users can be nudged to change their travel time by 5 to 6 minutes or alternate routes just with information and gamification techniques without the need for discounts or rewards. Therefore, basically with a fraction of the cost of implementing our solution, there are potential benefits in millions of euros every year.

Mobility Meters, a demand management and analytics solution that is currently under development by Urban Vind helps cities like Lyon reduce rush hour overcrowding with gamification and behaviour analytics techniques. Mobility Meters product development process incorporates state-of-the-art research on nudging, gamification and transport demand management. Currently, in the final stages of the development of the minimal viable product, we are looking for cities to collaborate with and test our solution. In addition to just the reduction of overcrowding, the solution considers the goals and KPIs of the city when it comes to sustainable mobility to help them steer their citizens towards sustainable behaviour.

References

dell’Olio, L., A. Ibeas, and P. Cecin (2011): “The Quality of Service desired by Public Transport Users,” Transport Policy, 217–227. https://econpapers.repec.org/RePEc:eee:trapol:v:18:y:2011:i:1:p:217-227

Greene-Roesel, R., Castiglione, J., Guiriba, C., & Bradley, M. (2018). BART Perks: Using Incentives to Manage Transit Demand. Transportation Research Record: Journal of the Transportation Research Board, 2672(8), 557–565. https://doi.org/10.1177/0361198118792765

Halvorsen, A., Koutsopoulos, H. N., Lau, S., Au, T., & Zhao, J. (2016). Reducing Subway Crowding: Analysis of an Off-Peak Discount Experiment in Hong Kong. Transportation Research Record: Journal of the Transportation Research Board, 2544(1), 38–46. https://doi.org/10.3141/2544-05 

Haywood, L., & Koning, M. (2013). Estimating Crowding Costs in Public Transport. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2256332

Keolis. 2022, June 7. Easing rush hour crowding in the Netherlands’ public transportation. Keolis Innovation. Retrieved from https://innovation.keolis.com/en/projects-news-insights/easing-rush-hour-public-transit-crowding-in-the-netherlands-by-reaching-out-to-stakeholders-in-education/

Kholodov, Y., Jenelius, E., Cats, O., van Oort, N., Mouter, N., Cebecauer, M., & Vermeulen, A. (2021). Public transport fare elasticities from smartcard data: Evidence from a natural experiment. Transport Policy, 105, 35–43. https://doi.org/10.1016/j.tranpol.2021.03.001 

Metis, P. 29-10-2012. Etude d’impact des transports en commun de région parisienne sur la santé des salariés. Accessed on 14-01-2024. https://www.metiseurope.eu/2012/10/29/etude-dimpact-des-transports-en-commun-de-rgion-parisienne-sur-la-sant-des-salaris/

Pluntke, C., & Prabhakar, B. (2013). INSINC: A Platform for Managing Peak Demand in Public Transit.

SLTA (Singapore Land Transportation Authority). 2023.. Travel Smart. https://www.simplygo.com.sg/about-travel-smart-journeys/. Accessed January 11, 2024.

Small, K. and E. Verhoef (2007): The Economics of Urban Transportation - 2nd Edition, Routledge.