Friday, 9 April 2010

Missing markets

It's looking more likely that National will be pushing the drink drive blood alcohol limit down from 0.08 to 0.05. I'd noted the insufficiency of current data for making that call here; David Farrar calls for more research here. But I fear that National's preferred research method's already been conducted: a poll shows majority support for such a reduction.

I'd really love to see a set of conditional iPredict markets on this one.

Market 1: Contract pays $1 if the drink driving limit is lowered to 0.05 by 1 Jan 2011, $0 otherwise.

Market 2: A set of bundles on the 2011 road toll relative to the average number of road deaths from 2004-2009 CONDITIONAL on no changes to the drink drive limit. 401 was the average over the period, standard deviation 25. So have the first bundle 388-412, then a couple of bundles with 25-count intervals either side. Maybe cut it to half-SD interval width for a finer measure.

Market 3: As above, but CONDITIONAL on the limit dropping to 0.05.

I'd expect a slight reduction in the road toll with the limit dropping. Not enough to make it a worthwhile policy, but I'd still expect a small reduction. Price differences between markets 2 and 3, conditioned by prices in market 1, would give a market-based estimate on the effect of lowering the drink driving limit.

iPredict does operate under regulation though, and running markets having the potential to be viewed as distasteful could make a securities commission re-authorization more difficult. And so a cautious market would refrain from running socially useful contracts. Recall that the DARPA markets were aborted after a couple of the worst senators in the US characterized them as allowing for betting on terrorism. Shame. I'd reckon that the odds on the drink drive limit dropping are close to 50/50. If it were a 90-10 thing, the conditional markets probably wouldn't work well - we had a lot of nonsense in the DOW conditional markets on Bernanke's reappointment when the odds of reappointment were upwards of 95%.

A good conditional market will require a policy that's definitive, that's likely to have an effect in the short term (within a year or so) on a regularly measured and hard to manipulate outcome, an initial probability of the policy's adoption that's neither a sure thing nor a complete longshot, and a long enough interval between the launch of the policy contract and the policy becoming either a sure thing or a lost cause. This would have been a decent one but for the fear it could bring down the market.



    Don't know if you've seen this. Market haters are everywhere it seems

  2. Cowen had good comments over on Marginal Revolution yesterday....