Chris Masse has taken the CEO of Tradesports/Intrade to task for recent prediction market failures. As many have noted, outcomes which depend on a small group of decision-makers and involve a semi-open list of candidates are especially difficult to predict. One of Chris' points is that it is far better for exchanges to foster markets with "social relevance", where the desired knowledge is more widely dispersed. I agree with this, but I prefer to stress the risk-sharing and hedging potential of these markets more than their predictive abilities per se. Consider the Nobel Prize markets. It is unclear to what extent aggregating opinions would have been able to produce accurate predictions, but
the potential hedging needs of the candidates and their publishers is clear and somewhat quantifiable. Chris' main theme in his analysis was "better marketing", and I recall from my internet start-up days that one of the things marketers like to do is to identify the potential size of markets. This is where the hedging perspective helps. For example, in a Nobel prediction market, there is at least $1.3 million on the line that could possibly be hedged by the candidates and their publishers. On the surface, this would already be a large market in Tradesports terms. The actual formation of a Nobel insurance/prediction market is unlikely for a few reasons, however: 1) There is an insider problem (although I don't think this is insurmountable), 2) The candidate list is essentially open, 3) Since the candidate list is relatively large and open, candidates would have to sell themselves short at low prices, which might be problematic since the magnitude of their selling interest would probably constitute a supply imbalance and push the price significantly lower than the actual mean perceived probability of their reward. (although the "longshot effect" should mitigate this somewhat)
In general though, I don't consider the demand/supply imbalance by hedgers as an obstacle to prediction market formation. In fact, I think it is a key to their viability. Successful futures markets tend to also be insurance markets. This is because risk-sellers (hedgers) are compelled to enter the market, and risk-buyers (speculators) are attracted to the market because the hedgers' imbalance creates exploitable value for them. The fact that a prediction market is also a peer-to-peer insurance market creates liquidity, which in most cases improves its predictive ability and reduces the ease of manipulation. The contract design question can then be seen as identifying a field
in which certain markets will be "fit", in the evolutionary sense. Which markets will naturally compel participation?
The field I stress here is politics, specifically legislation that impacts taxes and subsidies. Legislation-linked prediction markets also seem politically "fit" in that they would quicken the cutting of pork from the federal budget. I am in the process of compiling a list of possible tax and subsidy markets. I have already
cited the estate tax issue as the basis for one possible market. If the estate tax were to be permanently repealed, this would cost the government $70 billion per year starting in 2011. Even limiting ourselves to a time-horizon of one generation, the present value of wealth impacted by tax and subsidy legislation surely runs into the trillions. This then is my "marketing" contribution to the future of insurance, er, prediction markets.
Read Part 1