This paper has provided a comprehensive review of the multiple effects that the crowding of passengers in public transport
systems has on the quality and comfort of travelling, waiting and riding times, travel time variability, passenger wellbeing,
vehicle and route choice, and the optimal value of a service frequency, size of vehicles and fares. A summary follows:
(i) the impact of crowding on travel time through friction effects between passengers when boarding and alighting has been
usually estimated using regression models that include the number of standees inside vehicles or at stations as an explanatory
variable of dwell times. (ii) High average occupancy levels also increase the probability of vehicles circulating full, and
therefore, not being able to pick up passengers waiting at stops and stations, which increases waiting time and travel time
variability. (iii) Amongst the impacts of the crowding phenomenon on passengers’ health and wellbeing, authors have documented
increased anxiety, stress and feeling of exhaustion, perceptions of risk to personal safety and security, feelings of
invasion of privacy, propensity to arrive late at work and a possible loss in productivity for passengers that work while sitting
on a train. (iv) These and other factors are likely to be behind the negative valuations that users have of experiencing high
occupancy levels at stations, transfers and vehicles, which is obtained in demand models that account for a crowding effect
Fig. 9. Train modal share for different crowding levels and travel times.
A. Tirachini et al. / Transportation Research Part A 53 (2013) 36–52 49
on passengers’ choices, most commonly through an effect on the valuation of travel time savings. (v) Different crowding levels
between competing routes and unbalanced vehicle loads are also found to affect passengers’ choices of route and vehicle.
(vi) Finally, because the crowding externality increases the marginal cost of travelling, it should be accounted for in the design
process of public transport systems, in particular in the determination of frequencies, vehicle size and fare, as shown in
the public transport economic literature.
The second part of the paper is concerned with the effects on the valuation of travel time savings and estimation of demand
of alternative assumptions regarding how sensitive users are to a crowding externality, in particular, of the minimum
occupancy level that triggers a crowding effect on travel utility. Using data from Sydney, we have estimated crowding cost
functions that depend on the availability of seats, the density of standees per square metre or the occupancy rate of vehicles.
Two main findings are obtained that reveal the potential problems of omitting people’s perception of crowding when estimating
demand for public transport: (i) a model that assumes users as indifferent to the occupancy levels of vehicles overestimates
the value of travel time savings (VTTS) for low load factors and underestimates VTTS for high load factors, and
likewise (ii) a model that is insensitive to crowding levels underestimates demand if vehicles are uncrowded and overestimates
demand if vehicles are crowded. The generalisability of these findings is not proven; however, assuming a multinomial
logit model for mode choice, a load factor threshold that marks the underestimation or overestimation of demand when
ignoring crowding is analytically found. More research is needed to explore if these findings hold with more complex choice
models and in other contexts.
Regarding alternative crowding disutility specifications, we found that alternative assumptions concerning the threshold
load factor that triggers a crowding effect do have an influence on the resulting VTTS. The effect is especially noticeably for
low occupancy levels (all passengers sitting); however, for high occupancy levels, alternative crowding models tend to estimate
similar VTTS. In other words, if a system is operated in highly crowded conditions, it makes little difference how sensitive
were people to crowding while everyone is sitting.
The implications of the findings of this paper for cost benefit analysis and public transport policy are clear. The impact of
crowding on demand and supply should be considered from the early stages of the appraisal of public transport projects, as
the design of the system and the estimation of demand and social benefits rely on whether or not the multiple dimensions of
the crowding phenomenon are accounted for in the formal assessment of projects. For example, where projects are marginal
on benefit-cost ratios in the absence of allowing for crowding impacts, the inclusion of crowding can tip the balance (or at
least significantly improve the benefits) in supporting public transport investments that struggle to compete in benefit-cost
terms with road investments.
Several directions of further research to obtain a more comprehensive understanding on the implications of passenger
cro