TransPosition to present two presentations at the online-only 2020 AITPM conference
20 Aug 2020
Like many things in 2020, the AITPM conference has shifted to an online-only format. TransPosition has prepared two presentations- one on income effects on travel behaviour, and one on microsimulation of an autonomous fleet.
Rags and Riches: How income influences travel behaviour and project evaluation
Peter Davidson and Tom McCarthy
It is well understood that travel patterns significantly differ for individuals with different incomes due to a range of factors including housing choice, car availability, relative cost of travel by mode and job locations. Conventional wisdom suggests that low-income individuals are more reliant on public transport while high-income individuals are more likely to use private vehicles. We conducted an analysis of 2016 Census Journey to Work (JTW) across Australia and Household Travel Survey (HTS) data in Brisbane, Melbourne and Sydney to examine travel behaviour across the income distribution. Analysis of trip rates, trip distance and mode choice was performed across income as well as other demographic factors. Results indicate that both low- and high-income individuals are less likely to be vehicle drivers than those in other income groups but public transport is a more viable option for high-income workers than low-income workers.
Analysis of work trip lengths finds that high-income workers also live further away from work than low-income workers. In fact, total daily travel (measured in time and in distance) generally increases with income. This fact is somewhat surprising, as the opportunity cost of travel is higher for wealthier individuals, and they have greater freedom in choosing their residential location, affording greater opportunity to trade property price for travel time. The paper explores the reasons for this apparent contradiction, and concludes that the net benefit of travel must increase with income. This has implications for economic evaluation of transport initiatives, particularly when willingness to pay is considered. The paper concludes with some suggestions on how income effects can be better incorporated into project evaluation.
Understanding how low- and high-income individuals differentially access the transport network allows for the planning of services to help those who need it most. Conventional wisdom suggests that low-income individuals have limited access to private vehicles and are more reliant on public transport while high-income individuals use private vehicles instead of public transport. This is also sometimes used as justification for government subsidy of public transport, using it to provide mobility for those who are economically disadvantaged. While public transport does fill this function, a substantial proportion of travellers are relatively well paid CBD workers.
But the impact of income on travel behaviour has wider implications. Behavioural transport models seek to understand the differences in how different sectors of society perceive transport costs. This can be critical when alternatives allow travellers to trade time for money, such as a toll road or differential parking costs. Income is usually considered in these models, as the opportunity cost of travel is assumed to be correlated with income. However the evidence suggests that income also affects the benefit that users accrue from travel. Understanding this relationship will improve our ability to estimate accessibility improvements and evaluate the user benefits of transport investment.
Analysis of Census JTW data, for individuals aged 20-80 who travelled to work, found that low-income individuals are less likely to drive to work (64% of all low-income) than middle-income workers (77%). However, the same is also true of high-income individuals (66%), resulting in an inverted-U distribution of driver mode share across income.
Mode share and travel distance also varies between Regional Australia and capital cities. In regional areas, vehicle reliance is higher for commuters of all incomes, but high-income individuals are far more likely to utilise trains and buses (7% share for high income vs 4% for low income).
In terms of trip length, there is an overall positive relationship between income and travel. This is true for commuting, where Journey-to-work data shows that high income commuters travel almost one-third further than low income commuters. The relationship also holds for travel times from household travel surveys, both for commuting and for total daily travel.
Given the higher opportunity cost of travel that high income people experience, the benefits of travel must also scale with income. This also means that higher income travellers enjoy a higher consumer surplus from their travel. Economic analysis of user benefits from transport investment often uses changes in consumer surplus as a key measure. This work would suggest that a behaviourally sound utility model would end up with higher costs and benefits for those with higher incomes. The paper shows that ignoring the effects of income will result in unrealistic behaviour, but using constant values of time will lead to invalid estimates of benefits, in some cases showing a negative result for a change that would improve everyone’s perceived benefit. The problem can be resolved by including an explicit treatment of the marginal utility of income, that most economic models assume is constant.
Automating the Last Mile - Simulations of autonomous feeders to public transport and the future of vehicle occupancy
Morgan Weston and Peter Davidson
Autonomous vehicles (AV) have the potential to significantly disrupt transport systems. Previous work by the author finds that while operational improvements may be possible with AV, the demand side effects associated with cheaper and more attractive car travel will dominate, leading to higher road congestion. The nature of public transport is also likely to change, with reduced need for local bus services, and more emphasis on line-haul mass transit.
But some problems can be mitigated through the use of AV as fleet vehicles and as feeders to public transport. By providing high quality door-to-station pickups the “first/last mile” problem of mass transit can be solved. This paper addresses the feasibility of providing autonomous feeders to public transport, and the benefits of subsidising these services. It also presents a new fleet microsimulation model that has been developed by the authors and used to determine likely fleets sizes, average occupancy, waiting times and the cost of providing these services. The simulation model also shows some of the practical difficulties with achieving high vehicle occupancy, even when taking passengers to a centralized location such as a train station. This has consequences for the future of vehicle occupancy, and the possibility that even shared autonomous vehicles will spend most of their time with zero or one passenger.
An advanced transport demand model (such as TransPosition’s 4S model) can be used to investigate demand side effects of automation, and the author has published a number of previous papers exploring these effects. This paper seeks to consider one particular aspect - the use of autonomous feeders to public transport, and explore how this impacts the wider transport system. As the roads become more congested, the role of mass transit vehicles with a dedicated way (such as rail, light rail and busways) will increase. PT feeders unlock these services and ensure that they can be widely used. A different type of model is needed to examine the operation of these feeder services, since they rely on a complex interaction between traveller demand and responsive services. This paper presents a new operational microsimulation model, that integrates closely with the previous demand modelling. Questions such a fleet size, average occupancy, kilometres travelled, and average waiting are addressed.
By combining the simulation model with the demand model it is possible to look at the trade off between single occupancy travel and multi-occupancy travel, which provides a cheaper service to customers at the expense of increased waiting and diversion time, and a generally less pleasant time in vehicle. This makes it possible to identify the key factors that will influence vehicle occupancy in an autonomous fleet deployment scenario.
The authors have analysed a number of aspects of the impacts of autonomous vehicles and have become increasingly convinced that rather than solving all transport problems, AV’s will make many existing problems more difficult. In particular, the demand-side impacts of reducing the cost of car travel (both real and perceived costs) will outweigh the supply-side improvements that the new technology may enable. A key objective is finding ways to mitigate the worst problems. One promising avenue is in combining the strengths of AV with the demonstrated efficiency of line-haul public transport through the use of automated PT feeders.
The demand-side impacts of this new solution to PT’s “first/last mile” problem can be tested in a transport demand model, such as TransPosition’s 4S model. The initial assessment of this shows that PT feeders can lead to significantly higher public transport demand and a corresponding drop in urban congestion. But a different tool is needed to examine the feasibility/cost of providing these services. We have developed a new AV fleet microsimulation model that allows these operational issues to be addressed.
This paper shows that autonomous PT feeder services would make a very significant contribution to the urban transport task, and do so in a way that is practical and affordable. The number of required vehicles in an area would be manageable, and the waiting times for travellers would remain low. However one of the less desirable outcomes is that it is difficult to encourage high occupancy without high priced services, and without causing significant delays to travellers. This has lead the authors to be sceptical that vehicle occupancies will be higher in an autonomous future, even if people can be encouraged to shift from private vehicle ownership.