The Subsidized Dynamic Pari-Mutuel Market
An additional technical challenge is the combination of the market with a subsidy mechanism for cases in which a subsidy is desired. Note that simply placing the subsidy in the pool, to be allocated among the winning shares, will not do. That would give individuals an incentive to invest simultaneously in both market outcomes, possibly using separate accounts, a combination of transactions that ordinarily would have no net economic effect but would entitle the trader to a share of the subsidy. In effect, the problem is the same as the one that complicated the design of a subsidy in Chapter 2: traders might enter into wash transactions. One possible solution to this problem would be to charge commissions, which would also be added to the pool. Suppose that $100 total has been invested in a market with a $1 subsidy, and a 5 percent commission is charged. Investing another $100 in direct proportion to the earlier investments just before the market ends would cost the investor $5, though the investor would expect to receive $2.50 of this back when final payouts are issued, plus $0.50 of the subsidy, for a total of $3, a losing proposition. The trick is to identify a commission level that will be high enough given the amount of legitimate trading expected. An alternative possibility is to build on the subsidy mechanism described in Chapter 2, in which traders receive proportional shares of a fixed subsidy depending on the amount of time that they expose themselves at the front and the back of the bid and ask queue. As Pennock argues, a benefit of the dynamic pari-mutuel market is that it can be presented in a manner similar to a continuous double auction. A complication is that transactions are often fulfilled through the automated market maker rather than by other participants, unless a participant has offered a better deal then the market maker has offered. The subsidy should not be wasted on an offer that is inferior to the market maker’s. A solution is for the subsidy to be distributed only to those who have offered third parties better deals than the automated market maker would offer. The result of this subsidy scheme is that prices would move more slowly for traders making purchases on the market, thus allowing greater profits. A final possibility, which Pennock appears to advocate, requires the market sponsor to place a seed wager, for example, distributing the subsidy randomly over the different possible outcomes or distributing it in accordance with a model of the relevant events developed by the market sponsor. A problem with the seed wager approach is that the exact amount of the subsidy is not fixed in advance. If placed haphazardly, seed wagers will generally be losing propositions, thus providing a subsidy, but sometimes they might produce positive returns for the market sponsor. For example, if the seed wager is distributed unequally to a favorite and a long shot but the long shot wins, then the market sponsor will have made money. The seed wager at least provides an upper limit on the amount of money that the market sponsor will lose, certainly no more than the entire amount of the seed wager. If the market sponsor, meanwhile, distributes the subsidy in accordance with its own model, and that model is relatively good, then the market sponsor may be providing only a very small subsidy to the market. In some cases, of course, this will be exactly what the sponsor intends, because it will be providing higher subsidies when its own model is weaker and it thus needs more help, and yet it does not need to figure out in advance whether its own model is strong or weak. A separate potential problem with this approach is that most of the benefit of the subsidy will apply at the very beginning of the market. So, for example, the first trader to correct a blatant mispricing at the beginning of the market would receive most of the benefit of the subsidy. This will be a particular danger if the subsidy is distributed randomly or if the market sponsor’s model is weak. Of course, the first bettor’s investment would provide some additional incentive for others to fix any remaining mispricing, but still much of the subsidy might in effect be wasted on relatively easy fixes, at least unless the seed wager is set in a relatively sophisticated way. To some extent, this mechanism might counteract another problem with the dynamic pari-mutuel market: bettors with relatively good information in the absence of subsidy might have an incentive to wait to trade until there is already some amount invested in the market, in order to get the maximum profit possible from the available investment, at least if no one else is likely to develop the same information. The subsidy counteracts this tendency in an imprecise way, however, so that there might be strong incentives to trade at the very beginning and the very end of the market and less incentive during the bulk of the market.
Leave a Reply