Thursday 2 September 2010

Crash costs

I've apparently disappointed some folks by not having shouted about a front page story in Monday's Christchurch Press.

NZTA reckons the social cost of car crashes in Canterbury is around $500 million per year.

I actually have a lot less problem with this figure than I have with other social cost figures, if it's used properly.

What this number CAN be used for:
  • Evaluating whether a road safety improvement, like widening roads and putting in more passing lanes, passes cost-benefit analysis

What this number CANNOT be used for:
  • Evaluating whether a road safety campaign, like anti-speeding measures or banning heavy trucks from the road to Kaikoura, passes cost-benefit analysis

Why the distinction? The social cost figure includes a whole pile of private costs - the costs to occupants of a vehicle of being in a crash. Consider the driver of a sole-occupant vehicle that crashes due to his speeding. He must have viewed the speeding as being worth the increased risk, otherwise he wouldn't have done it. We'd overestimate the benefits of hitting more folks with speeding tickets if we counted the benefits of reduced car crashes without counting the costs of folks losing the valuable opportunity to drive fast - the vast majority of whom have no accidents. If we (very reasonably) want to worry about speeders hitting other people, then we need to count the costs to persons external to the speeder's vehicle of the speeding - we'd need a separate run-down of those costs. A legitimate worry, definitely. But we can't get it from the number provided. Similarly, if we wanted to ban long trucks from the windy road between Christchurch and Kaikoura, we'd need to pull out of the cost figure all of the crash costs imposed on the trucker himself.

On the other hand, consider a cost-benefit analysis of widening roads to prevent accidents. If we count up all of the costs, private and public, of widening the road - the money cost of the project, for the most part, along with the deadweight losses of taxation - we could reasonably weigh that against the benefits - the reduction in road accident probability multiplied by the full cost, private and public, of road accidents, along with reduced travel times.

It's because folks will always leave out the benefits of speeding or other unsafe behaviours that we need to be wary about using these figures to estimate the benefits of anti-unsafe-behaviour campaigns. But using it as part of a general assessment of whether it's worth widening roads and putting in passing lanes seems not at all unreasonable.

I haven't gone through the fine details of the project, so I'm certainly not giving it an unequivocal thumbs up. But it doesn't seem crazy.

Of course, Mayor Bob Parker's use of the study to argue for tougher anti-speeding and anti-drink-driving campaigns, noted in the Press article above-linked, is thoroughly inappropriate. If we're considering speeding and drink-driving, we have to net out the costs falling to the folks inside the speeder or drinker's car if we want to do things correctly, or we have to count up the losses imposed by the campaign on all the speeders and drinkers who don't have accidents but have reduced enjoyment because of the campaign.

Numbers can be fit for some purposes but not for others.

No comments:

Post a Comment