Legal Markets
As long as prediction markets remain illegal in the United States, U.S. policy analysis organizations’ options are considerably restricted, but there might still be some means of taking advantage of the information aggregation power of prediction markets. One approach is to enter into a contract with an organization abroad that runs prediction markets. I know of no examples of this type of arrangement, but sufficient demand would presumably bring supply. An organization such as TradeSports, after all, can add markets quite easily and presumably would be willing to host a market for a fee. TradeSports might even be willing to implement subsidized markets, provided the market sponsor pays the subsidy and some fee that would at least cover the hosting services that TradeSports provides. This approach reflects a form of regulatory arbitrage, taking advantage of the legality of TradeSports in the jurisdiction in which it exists. As long as the sponsor of a prediction market retains no control and no knowledge of who participates in the market, it seems unlikely that The prediction market sponsor could instead provide money to each of a number of selected participants with the stipulation that this money could be used to bet on a prediction market but that no participant would be allowed to bet any more money than initially provided. For example, a trader who receives one hundred dollars might lose that money but could not lose any more than that. Perhaps in part because of legal restrictions, PAM was to have been structured this way. Conceivably, prosecutors might claim that such an arrangement violates gambling laws, though the impossibility of losing any money from the market as a whole would seem to make such a prosecution unlikely. To improve appearances and decrease the chance of such a prosecution, the prediction market might run on the basis of points, with points convertible into dollars only after the market closes. It seems unlikely that the CFTC would seek to prevent such an arrangement, because the individual traders receiving subsidies would not likely be seen as ordinary investors. A danger in taking this approach to subsidizing a market is that some traders might not participate in evaluation and trading. Instead, they might simply pocket the money provided to them. Sponsors may be tempted to require that funds be traded at least a certain number of times, but there are hazards to this approach. In particular, traders might seek to evade the limitation by executing the same kind of wash transactions that, we saw above, could frustrate a simplistic subsidy design. A single trader presumably would not be allowed to open multiple accounts, but traders might work together to enter wash transactions. For example, one trader might agree with another to buy a number of shares from the other at a price somewhere in the bid-ask spread, with the traders then immediately executing the same set of transactions in reverse. The market sponsor would need to find some way of policing such transactions, as well as more subtle variations. If it were possible to do so, individual traders might respond to the requirements by executing only the minimum number of trades and hedging risk with offsetting transactions over time. An alternative approach might be to use the decentralized subsidy system (described in Chapter 2) to provide incentives for individuals to place their money at risk. For example, suppose that a would-be sponsor of a prediction market would like to invest ten thousand dollars to derive probability estimates from one hundred individuals who seem likely to have information or to be able to obtain information about a particular prediction. Instead of simply giving each of these individuals one hundred dollars that they might trade, the prediction market sponsor might give each trader fifty dollars and then offer five thousand dollars as a subsidy, to be distributed among those who place the money provided to them at risk by placing bid or ask orders. The market sponsor might run a points-only prediction market in which points would not be redeemable for dollars but then distribute the entire ten-thousand-dollar subsidy in proportion to the number of points that participants put at risk. Those who seek a portion of the subsidy by placing points at risk could expect to earn profits, and more knowledgeable participants should expect to earn higher profits than less knowledgeable ones. Although the points in this system have no direct cash value, they are valuable as means of obtaining subsidies, and so market participants will have incentives to be careful about the bid and ask offers that they make. Usually, however, it will be preferable to allow points to be cashed in at some value, in order to provide robust incentives for individuals to accept bid and ask offers when they believe that there has been mispricing. Although allocating a greater portion of a subsidy to encourage trading reduces the risk that individuals will simply pocket the trading funds, there is a significant drawback if too large a portion of the subsidy is allocated in this way. In a market in which points initially allocated are not redeemable for dollars or are redeemable for only a small portion of the subsidy, someone who does have a better probability estimate might not be able to invest any more than individuals with worse probability estimates. Such a market design is likely to be successful only if the appropriate probability estimate will tend to become more obvious to all participants over time. In that case, an individual who makes a sensible trade will earn points while the market is still operating and will thus have more points to place at risk in order to obtain a higher portion of the subsidy. If predictions are not likely to improve markedly over the life of the market, then someone who has good information will not accumulate points much faster than others and therefore will not receive a particularly large portion of the subsidy. The remedy is to allocate a meaningful portion of the subsidy to initial funds, so that trading can directly lead to profits. This analysis suggests that careful design can improve traders’ incentives in a market in which no participant can lose money. That does not mean that such markets will be just as good as subsidized markets in which traders can put their own money at risk. The markets will be particularly poor at encouraging evidence aggregation. Someone who has a specific piece of information indicating that the market price is off by a significant amount will be able to push the market in the right direction using only whatever dollars or points are initially assigned to the trader’s account. Even if the market succeeds in encouraging traders to place large amounts of money at risk in the bid and ask queues, the increased liquidity will not greatly increase the amount of money that someone with information will be able to earn by trading. Suppose that the total subsidy to a market with one hundred participants is ten thousand dollars. It will be difficult for any participant to earn a significant portion of that amount, and the market will provide only very limited incentives for individuals to conduct complicated and time-consuming analysis. If the goal is simply to obtain a crude form of assessment aggregation, however, the market is likely to be more successful. In this case, though, the market will not weigh individual traders’ confidence in their predictions as effectively as a market in which traders can invest their own funds. One approach to making markets in which participants cannot lose more like real prediction markets is to provide participants with accounts that can be used in each of a variety of prediction markets. Suppose, for example, that a policy analysis organization is interested in sponsoring ten different prediction markets in related topics and that it has identified a group of individuals that it wishes to encourage to trade in these markets. It could provide each individual with ten separate accounts, each assigned to a separate market. But pooling the accounts gives each trader an incentive to decide which market or markets to focus on. A trader, for example, who has private information relevant to only one of the markets might invest all of the funds allotted to trading in that market. Other traders will tend to invest in the markets in which they have information. Thus, each trader is likely to move prices more in markets in which the trader might have information, providing some assurance that market prices will be weighted by the self-confidence each trader has for that market.
Leave a Reply