A bad time to believe in markets?
By Rob Knight on October 19th, 2008Given that ‘markets’ are currently deemed by many intelligent people to be responsible in some way for a string of global calamities, it might seem like an odd time to be promoting the virtues of markets. And yet, that’s what any prediction market advocate must do. How can we do this, and does this mean that we’re in favour of all markets everywhere?
Prediction markets work by simply assigning values to outcomes and allowing individuals to trade based on whether they believe that this outcome will occur. The market mechanism is a good way of focussing the mind, cutting out any need to explain or justify your opinion - if you’re right, you win and if you’re wrong, you lose. Where you correct the mistakes of others, you are rewarded. What this proves is that the market mechanism can usefully enable us to make better decisions in certain categories of problems.
So why do markets get such a bad rap? It seems to be a commonly-held view these days that something has gone badly wrong in the market for various financial products and some slightly more cavalier commenters claim that this is an inevitable flaw in the whole concept of ‘markets’. Dispersed, distributed individual decision-makers acting in their own interests can’t possibly be trusted with decisions of this importance!
The evidence, in the form of the financial crises of the last few months, seems to stack up in favour of this argument. But it’s my view that this evidence does not invalidate the concept of markets per se. Let’s look more closely at what’s actually been happening:
To simplify, the current ‘credit crunch’ is a result of the mis-pricing of assets, viz. mortgages. Various institutions loaned out large sums of money then sold these loans on to others. In theory, markets should be very good at dealing with this. Other agents in the marketplace can judge whether they think that the loan is likely to be repaid, and will pay accordingly for the right to collect the mortgage payments. If the mortgage is paid off in full, they make a profit. If not - if there is a default - then they lose. But that’s OK, because they price the likelihood of losing into their bidding price. It’s just like any classic prediction market in this respect.
But if that’s the case, things must be pretty bad for PMs, right? If the cause of the Nightmare on Wall Street is ‘just like’ a prediction market, then does this not suggest that PMs are tainted too? Well, not quite. The markets where mortgages were traded were not quite as simple as the example above. What happened was that the ‘bad’ (highly risky) mortgages were packaged up with perfectly good mortgages and sold together. The high ratings of the good mortgages obscured the risks of the bad mortgages and, after a while, nobody could be entirely sure of what was being traded. But there were well-respected and important-sounding institutions giving these mortgage assets their highest possible rating, so nobody worried much.
In the prediction market world, we can spot the problem here easily. Prediction market architects know that if we want to achieve accurate predictions, we have to be very clear about what’s being predicted. We can only ask people to predict things that they have some basis for knowledge of; for example, we can ask people to predict the outcomes of political elections because most people already have opinions on those things and can easily gather high-quality information from the public domain. But there would be no point in running a prediction market in, say, the number of pairs of socks that I own because that’s something that nobody has a reasonable basis for prediction of. I’m sure that we could get a market going on it, but the value of the predictions would be pretty low. The mortgage market has been behaving more like the Rob’s-socks-market recently (to stretch the metaphor even further, we might imagine that I’ve been inflating my pair-of-socks count by combining odd socks to make pairs), with people trading based only on guesswork. The ratings agencies whose job is to rate these assets actually made the situation a lot worse: by giving people the confidence to buy these assets, they allowed the problem to reach calamitous proportions before it became obvious (an example of the Tullock Effect in action).
In a prediction market, we’d find this kind of thing quite unacceptable. The idea of people trading based on what someone else is telling them (in fact, the ratings of the agencies were closer to a promise) is anathema because the whole idea is to capture diverse opinions and combine them into a price that reflects everyone’s knowledge. And the idea of people being asked to predict something so imprecise as an unknown bundle of mortgages would also set alarm bells ringing. In short, the problem was not the market but what was being traded on it and how little the traders understood this. The ratings agencies are largely to blame for assuaging the doubts of those who should have been refusing to touch these assets years ago, when the problem was only just beginning. But, egged on by AAA ratings and the promise of bonus payments, market traders stopped thinking and started buying.
The market did the best job it could with the information to hand. The information was useless, and that’s where the problem lies. The rating system and the obfuscation of the origins of certain assets combined to fool people into believing a falsehood, and they went out and bet everything on it. To find the problem, we need only to look at how the information about the value of mortgage assets was diluted. Market institutions may indeed need to regulate this practice in future - perhaps higher standards of information should be required which prevent this dilution. But the ratings agencies deserve their share of blame, and the whole concept of trading being conducted based on what a person who isn’t even participating in the market says about the value of assets is something that looks rather suspect now.
In conclusion, a market is only as good as the information available. Prediction markets, and other markets in general, retain a good record where the standard of information is high. There’s no reason to think that the credit crunch has invalidated the general principles behind prediction markets.
Tags: credit crunch