The DfT and their car traffic forecasts

This graph, from the Department for Transport’s 2013 Road Transport Forecasts (which summarises the results from their National Transport Model) has been doing the rounds on social media this week.

It shows that the amount of distance we are travelling by car, per capita, in Britain has fallen consistently since the early-2000s; and yet their model predicts that this decline will reverse, and car miles per person will increase by 15% by 2040.

What is just as remarkable, however, is the Department for Transport’s own analysis of this graph –

Figure 16 below shows that, according to our forecast, miles per person will increase by 15% percent by 2040 (9% above pre-recession levels) despite an increase in GDP per capita of 66% and fuel cost decreasing by 24%. [my emphasis]

The key word here being ‘despite’.

The DfT believe that increases in GDP per capita, and falling fuel costs, should really push car miles per person even higher than the projected 15% increase. Coupled with a projected 20% increase in English population by 2040, the DfT are forecasting that overall road traffic will be 46% higher in 2040 than 2010.

They acknowledge that the effect of their ‘key drivers’ on road traffic levels (GDP per capita, population, and fuel prices) is becoming less elastic, as the market becomes saturated –

As explained in section 2, the elasticity of miles per person to key drivers is falling over time, and will keep falling into the future as the market moves further towards saturation.

However, they still think that this 15% rise in car miles per person will happen, principally because of falling costs per mile, meaning people will be incentivised to travel further.

This increase in miles per person [15% on 2010], however smaller than it would have been in the past, reflects the fact that people will be able to travel longer distances with their cars, as the cost per mile will decline sharply compared to ability to pay.

Whether people will actually want to do this – to spend more time stuck in cars – appears to simply be assumed.

The other interesting detail from this report is… London. This document essentially acknowledges that the National Transport Model has failed to predict that the amount of car traffic in London would fall as much as it has –

… analysis of our forecast from 2003-2010 shows that although the NTM predicts a fall in London car traffic of 1.5%, this was not as great as the actual 7.8% fall in traffic count statistics.

What’s the explanation?

We believe that the reason for this short-term model error and long-run discrepancy with other forecasts is due to:

Car Ownership – the number of cars per person in London has been relatively flat over the last decade. While we have different car ownership saturation levels for different area types, including London, these may need to be re-estimated.

Public Transport – London has seen high levels of investment in public transport, capacity and quality improvement on buses and rail based public transport. London will continue to see high levels of investment in public transport with increase in capacity into the future, e.g. Cross Rail. We will need to revisit our modelling on the impact this may have on car travel.

Road capacity, car parking space cost and availability – There is evidence to suggest that In recent years London road capacity has been significantly reduced due to bus lanes, congestion charge and other road works. There is also a significant constraint and cost to parking in London which would reduce the demand to travel by car. We will need to revisit our modelling on the impact this may have on car travel.

On each of these three factors, the DfT are admitting that their model needs to be ‘revisited’ – their model simply hasn’t correctly taken into account the effect of public transport, and reallocation of road space, on the amount of car traffic that might be on the roads.

It’s also worth noting this ‘London’ example appears to show that levels of car ownership – which the DfT tie closely with GDP per capita – might be much more strongly affected by these other two factors assessed here, public transport and use of road space. Again, a challenge for the DfT modellers.

It seems that the DfT are admitting that their model doesn’t accurately take into account factors beyond income, population and fuel costs, their ‘key drivers’ – which is hugely significant if, as is likely to be the case, urban areas (in particular) in England continue to reallocate road space to other modes of transport, and prioritise these other modes, ahead of car travel.

Certainly, planning for future growth in car travel using a model that the DfT itself admits isn’t properly reflecting other factors on car demand looks increasingly silly.