Saturday, September 10, 2005

Skill of futures markets

Steve Bloom asked some time ago: "Also, how good are futures markets generally at accurately forecasting future prices, and especially peak prices, while at a peak?"

In order to begin to address this, I'd better start with the "Efficient Market Hypothesis". The most widely-quoted version of the EMH basically states that the current market price reflects all current information (there are also "strong" and "weak" versions which handle technical nuances).

Time for an economics joke. Two economists are walking down the road, and one says to the other "Hey, do you see that? It looks like a 10 pound note in the gutter over there. I'm going over to get it". The second replies "Don't bother, if it really was a tenner someone would already have picked it up". The point is that in a truly efficient market, there would be no point in doing any analysis. But the market price does not exist in a vacuum, and it can only reflect current information if analysts put in the effort to work out the implications of this information, which they would not do if there was no chance of a return! The solution to this paradox is to accept that the market is probably not always absolutely perfect, but it is usually pretty good (very few analysts can reliably beat the market over the long term). EMH is not axiomatically or inevitably true, but is instead a reasonable approximation to what is observed.

The very act of taking advantage of mispriced markets will move the price to reduce the error. So as soon as an inefficiency is spotted, it is liable to be eliminated. However, if the (real or perceived) entry barriers are high enough, and the reward insufficient, it is quite possible that errors will remain over the longer term. For example, the longstanding differential in returns between stocks and bonds takes some explaining. There are also obviously mispriced claims in the FX market - coupons with a guaranteed but distant $1 pay-off trade at a modest but significant discount to that value (which also means that people are paying non-zero prices for coupons with a guaranteed long-term zero pay-off - this can only be because they hope to sell to a "greater fool" within the time frame of the investment). This long-term discounting due to speculation makes it difficult to assess what the market really "thinks" about claims like Slv1 - is its high price really the market players' consensus estimate of the chance of a catastrophe, or just a long-term speculative discount of a claim that is virtually equivalent to "False in 2030"? It seems to be an open question (ie, I don't know) as to how far ahead a futures market can be expected to provide useful information, due to this speculative noise. But over a few months or years in the oil market, I would expect that it shouldn't be a huge problem.

However, even in the case that the market is truly efficient, it still cannot predict an uncertain future, but can only reflect the consensus likelihood of different possible outcomes. If all the players are badly informed, the market will not correct them. Its function is to act as an efficient and credible mechanism for aggregating the opinions of many, and perhaps the real question should not be "is the market accurate" but "is there a better mechanism for prediction"?

Anyway, what about oil prices? All the market traders can do is try to take account of their estimates of the likelihood of various future price-changing events - they have no crystal ball to actually predict the future accurately. The market might well account for the possibility of some disuption to supply in hurricane season, but could not in advance have priced Katrina into the forecasts. Although I would never claim that the market is always right, it is not likely to be far wrong for a long time, and is especially unlikely to be regularly wrong in the same manner. Such a failure would be spotted by traders who would be able to make a profit every time this situation occurred. With hindsight, the market is proven wrong time after time. But at any time, the current prices should simply reflect current expectations. Anyone who thinks otherwise is welcome to re-mortgage their house and make a killing...

2 comments:

Anonymous said...

Thanks for that; I learned some things I hadn't known. I guess the speculation inherent in my question was that market peaks, and especially ones like we have now that are the result of an unprecedented combination of factors, would be situations where the players would indeed not be well informed. If so, I suspect that the oil futures market might be slow to realize that peak oil has occurred (assuming it happens during a peak market). Things would certainly undergo an adjustment after that, but the adjustment might be in the form of a financial panic.

I'm probably a little prone to see these markets as less stable than they perhaps are since I live and work at the heart of the current speculative housing market in the SF Bay Area. There's a "greater fool" factor at work here for sure.

James Annan said...

Housing is complicated, because the assets are huge, indivisible, and physically fixed. Plus, we have to have _somewhere_ to live. The housing boom in the 80s in the UK was largely triggered by changing loan policy (offering a multiple of the combined salaries of a couple, rather than just that of the "breadwinner") and a fear that anyone who did not buy NOW would never again get the chance.

Having now spent 4 years in Japan with UK house prices rocketing and my salary dropping in UK terms, I know how they felt!