Current practices for investing in transport are flawed, due to forecasts of future travel used to inform important transport investment decisions.
Unfortunately, future estimates are uncertain, as forecasts are influenced by some factors that are unexpected, unpredictable, difficult to predict, and outcomes are not easily quantifiable.
Forecasting Transport Using Models
Transport models are used to forecast future travel. But, as the saying goes… ‘all estimates are wrong, some are useful’. Models are simplified representations of the real world, built on algorithms using parameters such as economic conditions, land use, population, employment, related projects and more.
Basically, the model predicts the future based on calibrated relationships, such as the growth in population results in growth in travel. Predicting the future based on the past is problematic.
The conclusions in the recent NCHRP Report on travel forecasting accuracy [1] indicated that patronage on major passenger transport projects often fell short of forecasts, the post-opening traffic on toll roads was lower on average than forecast and actual traffic volumes in general, were lower than forecast on average. Forecasts over five years into the future were less accurate, and all forecasts critically depended on macro-economic projections.
Facing Uncertainty in being able to Forecast Effectively
Today, there are several disruptions and radical changes occurring that impact the effectiveness of transport forecasts, and thus investment decisions. For example:
- Online shopping has, and will continue to, result in many more individual deliveries, rather than travel to a shopping centre or precinct to purchase a range of goods or groceries.
- Demand responsive transport, such as Uber, has changed travel patterns, and this includes ancillary services such as Uber Eats. The younger generation of travellers are not getting driving licences or buying cars at the same level as previous generations.
- The coronavirus pandemic has had a dramatic short-term effect on travel – reduced public transport patronage, increased car traffic, increased active transport, move to remote work – and the longer-term impacts are uncertain.
- Connected and autonomous vehicles are expected to change travel patterns and traffic volumes. Flying taxi services are also being developed.
How to Deal with Uncertainty
In developing a travel forecast, it is important to agree on the assumptions up front, preferably with key stakeholders and funders. Don’t expect transport modellers to have to make all the decisions.
Present a range of potential outcomes, rather than a single forecast outcome, by conducting sensitivity assessments of critical components.
To make forecasts more believable and useful, acknowledge the uncertainty, as this will help decision makers understand the context.
How should we deal with uncertainty when investing in transport?
[1] NCHRP 934. 2020. Traffic Forecasting Accuracy Assessment Research. Washington, DC: The National Academies Press.