Risk Markets And Politics

Wednesday, January 25, 2006

Hedging Against Education Inflation

Some costs such as education and healthcare are difficult to hedge against because there are few willing to take the other side of the trade. Take education, one of the very best investments. The average college graduate in the United States makes about $20,000, or over 60% per year more than those who stopped with high school diplomas. It is easy then to understand why college costs can continue to advance at roughly twice the overall rate of inflation. Education is a major cost for many families, sometimes comparable to housing.

Consider a market that tracked the level of the "College Tuition and Fees" component of the CPI. This market could work very much like the upcoming CME housing futures, with the monthly CPI release acting as a discontinuous "spot" price. Although the current CME CPI derivative operates as a parimutuel auction, a standing continuous double auction would probably be better here. If the futures had a long term length, this would mitigate the occasional hassle (and cost) of having to "roll" them to the next expiration. Remember, the goal is to provide a hedging tool to members of the public who wouldn't otherwise trade futures.

The fly in the ointment of course is that there would be a massive demand imbalance — there would be many buyers, and few natural sellers. (A parimutuel auction wouldn't really address the dearth of sellers, as buyers would have to commit to a certain price range, when really they just want protection from any upward movement.) The basic supply problem is that colleges aren't commodity producers, and barring potential legislation, they aren't in need of a hedging vehicle.

Aside from legislation, is there anything that might eventually threaten to bring education costs down, making such a market viable? Robert Shiller has suggested that communications technology will decrease the demand for professors. Arguably, this trend is underway in the form of some for-profit universities, but any price competition is not yet material — and rightly or wrongly, academic accreditation does not seem to be just a transfer of knowledge.

What about demographic trends? Perhaps climbing retirement ages will help by increasing the supply of working intelligence, but with the constant explosion of knowledge, this is unlikely. It is more likely that lower birth rates in the developed world will do the trick.

Now, if one believes Ray Kurzweil, intelligence might literally become a commodity by 2029. He made a "$10,000" (the effect of interest rates on prediction market prices is almost never discussed — more on this soon) bet to this effect with Mitchell Kapor on longbets.org. Well, it is doubtful whether a machine passing the Turing Test by being able to simulate a human conversation would actually possess the kind of intelligence that, say, employers would pay for. For the sake of argument, let's assume that it would, because this leads to a good example of market design. The Kapor/Kurzweil bet was a one-off, but imagine a standing Turing Test futures market. Now re-consider the education cost futures. The education futures would then be correlated to the Turing Test market, but the former would compel liquidity because of its double hedging function. One should always think about how changing the explicit content of a market will affect hedging demand, liquidity and balance. Often it might be worthwhile to dial the "beta" of a prediction market tied to a specific event below 100% in order to capture more general concerns (and in some cases, make it less controversial.) Even if there is a huge potential hedging function, there may not be enough balance in the market to realize volume, especially if there are existing ways to express correlated views. It will be very interesting to observe the materialization of the new housing futures market, which should be correlated to bonds.