A Decentralized Subsidy Approach
A decentralized strategy for dispersing the risk assumed by market makers would borrow from the TradeSports policy of offering lower commissions to traders who make bid or ask offers that are later accepted. One possibility is to distribute a fixed subsidy to traders who offer the most generous bid and ask prices, in proportion to the time and volume of the traders’ exposure. (Higher subsidies might be offered for exposure at times of day in which trading is more likely to occur.) For example, suppose that the current bid and ask prices are 28 cents and 30 cents, respectively. If I offer to buy up to 100 shares at a price of twenty-nine cents, and if that offer stays at the front of the bid queue for ten minutes, I would receive 100 <mul> 10, or 1,000 credits.9 The same would be true if I offered to sell up to 100 shares at a price of twenty-nine cents, if that offer stayed at the front of the ask queue for the same length of time. If over the duration of the market 1,000,000 credits were awarded, then 1,000 credits would be worth 0.1 percent of the total fixed subsidy amount. This system would provide traders with incentives to make generous bid and ask offers, allowing those with information to profit on that information. In effect, this system improves the incentives for individuals to serve as market makers willing to buy and sell tradable contracts and for these individuals to maintain a small spread between the prices at which they are willing to buy and sell. The system is not easily manipulated. If, in the above example, I simultaneously offered to buy at twenty-nine cents and to sell at twenty-nine cents, those orders would be fulfilled immediately, and so, having exposed myself to no risk, I would receive no credit. This might not be the only subsidy system available, but it should help convert what otherwise would be a zero-sum (or, with commissions, negative-sum) game from the perspective of traders into a positive-sum game, and it therefore should help enhance market participation even if the topic is of intrinsic interest to only a few individuals. Someone who feels relatively confident in a prediction, regardless of the extent to which that prediction is the result of private information, may be able to profit in two ways. First, if the trader’s prediction is below the bid price or above the ask price, the trader can accept the corresponding offer. The subsidy should result in more favorable terms’ being available to that trader than otherwise would exist, thus increasing trading profits. Second, after any such transactions clear, the trader might seek to make more generous offers to buy and sell than those currently in the queue, thus directly earning a portion of the subsidy. This approach provides a reward not only for a trader who finds that a tradable contract is mispriced but potentially also for a trader whose efforts provide additional confidence that a tradable contract is correctly priced. Will subsidies be sufficient to allow a prediction market effectively to perform the task of information aggregation? The answer depends both on the amount of the subsidy and on the nature of the prediction market. A million-dollar subsidy, after all, would attract interest in virtually any topic, and though the number might seem impossibly large, some policy analysts or philanthropists might think such a subsidy would be well spent. Of course, for some issues, no subsidy at all is necessary, and indeed TradeSports shows that prediction markets can flourish with the opposite of subsidy: commissions. A relatively small subsidy, say, one thousand or one hundred dollars, might be enough to accomplish the task of assessment aggregation on an issue regarding which relatively few people have opinions. Many of these individuals ordinarily might not be interested in participating a game in which they might be as likely to lose as to gain money, but they might participate when making money is the more likely outcome for a skilled participant. Subsidies will be particularly helpful if a prediction market might aggregate not only assessments but also evidence. If private information is likely to be useful in generating the predictions, then individuals who do not have such information ordinarily might be unwilling to trade. The possibility of evidence aggregation makes it more difficult to aggregate assessments, because individuals who have assessments based on available evidence will be less likely to trade if others can take advantage of the assessors’ relative lack of knowledge. Where the goal of a prediction market is simply assessment aggregation, that goal will be more difficult to achieve when some participants might hope to profit from evidence aggregation. In general, the greater the degree of private information in a market, the larger the bid-ask spread in that market. Subsidies will tend to narrow the spread between the bid and ask prices, and thus if few trades occur, the bid-ask midpoint can serve as an effective prediction.
One Response to “A Decentralized Subsidy Approach”
Leave a Reply
January 28th, 2008 at 9:42 pm
[…] response: […] The incentives provided by two of my technical proposals (the decentralized subsidy approach and the nobody-loses prediction market) are sufficiently straightforward to me that math seems […]