Transport-related decision-making for a post-COVID-19 world

The Problem

As well as the direct terrible impacts of the Covid19 pandemic on those who catch the virus &/or lose loved ones to its effects, the pandemic and the ‘lock-down’ measures designed to tackle it have had unprecedented impacts on the economy, travel demand and travel behaviour – unprecedented in terms of the scale of the reduction on total travel & the duration of this slump in travel demand, drop-off, combined with unusual impacts on mode choice, particularly a reduction in the attractiveness of (crowded) public transport.

Previous events (like the ‘Great Storm’ of October 1987, the Hatfield train crash (2000), the banking crisis (2008) and subsequent recession, the ‘Beast from the East’ (February 2018) etc created impacts significant enough to spot in travel time series (&/or be remembered by the travelling public), but none of these changed as many of the assumptions underlying traditional transport models as the present pandemic and the measures being used to tackle it.

Even before Covid19, transport planners in the UK [1], [2], [3], [4] were conscious of the increasing difficulty of predicting future travel behaviour, with influences such as the climate emergency (and the commitment for the country to achieve ‘net zero’ emissions by 2050 (2045 in Scotland), Intelligent Mobility, changes in licence-holding and car ownership, the rise of on-line shopping, etc, all being cited as being sufficiently ‘disruptive’ to travel trends to warrant some consideration of alternatives to essentially trend-based forecasting of travel behaviour.

The crux of the problem is not quantifying how much travel (and emissions and air quality and public transport revenue and activity levels) changed during ‘lock-down’, but trying to predict how that ‘new normal’ (perhaps defined as the point when an effective vaccine is widely available?) will differ from what have been predicted by transport planners in the pre-Covid19 world, using trend-based ‘Business as Usual’ forecasting approaches.

Some of the travel-related factors which might be affected in the post-Covid19 world are listed below:

Transport Factors Likely to Change in the Post-Covid19 World

Social Distancing on PT – effective capacity of individual public transport vehicles reduced (short-term effect ?)
Mode choice parameters – (active travel more popular?, public transport less popular?, fuel price =??, changes in parking charges?) and how long will these changes persist – cycling in November is less attractive than cycling in April & May!

Travel demand to/from existing premises

• commute (reduced employment and more home working),
• business travel (economic downturn and more ‘virtual’ meetings),
• shopping (economic downturn & more on-line shopping),
• other leisure (reduced levels of disposable income and loss of leisure facilities),
• universities (restrictions on overseas travel, global recession, Brexit, etc)

Changes to public transport services (and fares) – reductions in services due to losses incurred during lock-down & changes in post-Covid19 demand, possible increases in frequency to cope with the capacity constraints of social distancing, changes to franchises, level of resource available to subsidise services (and concessionary travel schemes), less frequent travel reducing applicability of season tickets etc.

Age profile of vehicle fleets – reduced vehicle purchases during lock-down and subsequent economic conditions important for air quality

Car Ownership – reduced (eg. due to economic down-turn) or increased (eg replacing use of public transport)?

New Developments – delay in future-year land-use developments coming on-stream (eg due to lock-down), lower rents in existing premises, potential permanent changes in the future land-use -&/or Local Plans

Delay in transport schemes previously assumed in the Reference Case in Year X (direct impacts of the lock-down and diversion of Government/scheme promoter resources)

Public/political acceptability of emissions-related measures – may drop significantly as a result of the list above

Knowing where we are now is the easy bit!

In addition to identifying all of the aspects of travel behaviour that have been changed by the Covid19 pandemic, can we predict how quickly these impacts will return to ‘normal’ in the post-Covid19 world, ranging from behaviour which rebound to normal almost as soon as ‘lock-down’ ends, to aspects which remain permanently changed.

Too many unknowns?

• There are currently too many unknowns to be able to rely on a single ‘best guess’ forecast when making significant transport investment decisions
• And too many variables to test all combinations (especially given the complex interactions between them)
• The simple approach of including a ‘Worst Case’ Sensitivity Test is likely to give too much weight to an unlikely combination of inputs
• And we should be trying to influence some or all of these changes, rather than assuming they are all ‘a given’ and beyond our control
• Or, at the various least, identify which of the unknowns are most likely to influence to outcomes we are trying to achieve, so that we can collect/monitor the relevant data, to reduce this uncertainty going forward

Possible Scenario-planning-based Approaches

The simplest approach is to identify a set of X ‘equally-likely’ scenarios (typically 5 or 6) and test our investment decisions against these – often using a ‘least-regret’ approach (ie identify how the options we are considering perform in each of the X scenarios and choose the scheme whose worst performance in any of the scenarios is the best of the alternative schemes being considered).

An alternative is to develop scenario prediction tools which use outputs from a manageable number of scenario tests of an existing transport model (often using automated generation of the various input combinations) to populate a database of results which can then be used to synthesise the outcomes for other combinations of the input assumptions – examples include the Scenario Planning Tool which SYSTRA developed recently for Transport Scotland [5] , our meta-modelling of Northern PowerHouse Rail and the Lookup Tables of Benefits of Overtaking Provision which MVA/SYSTRA created for the DRD in Northern Ireland in the late 1990s).

A Third approach is so-called Monte Carlo analysis, where the user attempts to predict the statistical likelihood of the various inputs (and any correlations between them) and the analysis tool (typically an add-on in Excel), randomly generates multiple future scenarios using these distributions, to provide statistical estimates of the range of the various outputs – the snag with this approach is that it can be difficult to predict the relevant distributions to apply to the input variables, particularly when the disruption to the Business as Usual conditions make it difficult to apply historic time series.

The Need for Coherent Groupings of Input Assumptions

If adopting the simple ‘X Equally-Likely Scenarios’ approach, it is important to group the ‘umpteen’ input parameters into coherent groups, to inform the understanding of which category of input assumptions has the most impact on the relevant outputs &/or to help understand when combinations of inputs interact in significant ways.

It is also likely to be helpful to use coherent and internally-consistent groupings of the input assumptions when describing the results, rather than arbitrary and random combinations of these inputs.

In the case of the post-Covid19 predictions, one possible approach to this grouping would be to distinguish between impacts associated with the economy (eg increased unemployment, reduction in bus services, reduced car ownership etc), impacts associated with individuals’ travel preferences (more home working, altered mode choice preferences etc) and attributes associated with Government &/or operator policy/interventions (requirement for social distancing on PT, support for PT services etc).

An example of this approach is illustrated below:

Suggested Scenarios

3 Dimensions

• Economic Impacts
• Behavioural Change
• Policy Interventions

Suggested Scenarios
0. BaU (=Current Business as Usual)
1. Slump (=All of the relevant economic impacts, no long-lasting behavioural change)
2. I’ve Never Looked Back (=The economy bounces back to pre-Covid19 levels, but all of the behavioural changes are maintained)
3. Coping as Best We Can (= Combination of 1 & 2)
4. It Could Have Been Worse (= Government interventions reduce/avoid some of the negative economic impacts, no long-lasting behavioural change)
5. Brave New World (= 4 plus the behavioural change)

How the Scenario-based Analysis Would Be Done

The X ‘Equally-Likely’ Scenarios would be converted into (evidence-based) assumptions for the various impacts:

• change in employment (by economic sector)
• increase in working and shopping from home
• effective capacity of public transport vehicles
• changes in Reference Case public transport services
• change in the effective capacity of public transport vehicles (eg due to ongoing social distancing)
• mode choice parameters by journey purpose, car availability etc)

and the relevant Do Something schemes tested under each set of scenarios.

How the Results of the Scenario-based Analysis Might Be Used

The results and the associated analysis would then help to identify:

• How resilient the various schemes are to the variation in the input assumptions
• What are the short, medium and long-term impacts on public transport patronage and revenue
• How the long-term impacts of Covid19 affect our investment decisions
• Which of the Covid19-related impacts we need to monitor, in order to reduce the most significant uncertainties
• Which of the possible future worlds we want to achieve and which we want to avoid (and therefore which of the scenario inputs do we need to try to influence)

References

1. Guidance for Transport Planning and Policymaking in the face of an uncertain future, Lyons and Davidson, 2016, https://doi.org/10.1016/j.tra.2016....
2. Embracing Uncertainty and Shaping Transport for Scotland’s Future, Lyons, Cragg & Neil, https://aetransport.org/
3. FUTURES - Vision-led strategic planning for an uncertain world, Mott MacDonald and University of West of England, www.mottmac.com/futures
4. Opening out and closing down: the treatment of uncertainty in transport planning’s forecasting paradigm , Lyons and Marsden, 2019 https://doi.org/10.1007/s11116-019-...
5. Developing A Scenario Planning Tool, Cragg and Johansson, Modelling World, 2019