Gambling, prediction markets and public policy.

Paton, David ; Siegel, Donald S. ; Williams, Leighton Vaughan 等

1. Background

The recent worldwide increase in gambling and prediction markets, including casinos, sports betting, lotteries, elections, and wagering on financial instruments has stimulated an important debate regarding the public policy implications of these activities. Some critical research questions concern the efficiency of such markets, heterogeneity in risk attitudes among agents engaged in these activities, the factors that influence performance in gambling, and the desirability of using prediction markets.

Prediction markets are essentially wagering markets created for the purpose of making predictions. The theoretical underpinning for the use of these markets stems from the view that relevant information concerning the likelihood of future events is dispersed among many people (i.e., the crowd), and that prediction markets allow for the aggregation of this information. The design of the incentive mechanism is critical, since people may invest more thought and energy into expressing their opinion when they have a strong incentive to do so. While the effective use of prediction markets has the potential to help forecasts at a macro level, they may also assist corporations in providing a good estimate of, for example, the launch date of a new product. These markets have, for these reasons, generated substantial interest among social scientists, policy makers, and the business community.

The insights gained have many potentially valuable applications for policy more generally. In particular, quantifiable targets are ideal candidates for operating a trading market. In this context, the value of the information provided by prediction markets will come primarily from the advance warning that managers will be given of weak performance in particular areas. This can help improve resource allocation.

Prediction markets have been used to provide forecasts of the probability as well as the mean and median outcomes of future events, and these markets have been used to forecast outcomes ranging from vote shares and election outcomes (e.g., Rhode and Strumpf 2004; Wolfers and Zitzewitz 2004; Snowberg, Wolfers, and Zitzewitz 2005), to the probability of meeting project deadlines at Google (Leigh and Wolfers 2007). Prediction markets have also been used to forecast the timing of an outbreak of bird flu (Wang et al. 2009), and more generally in health-related markets (e.g., Polgreen, Nelson, and Newmann 2007). Prediction markets may also be used as a mechanism to help market participants hedge their exposure to risk (e.g., Athanasoulis, Shiller, and Rietz 1999; Shiller 2003).

However, some researchers have questioned how far prediction markets are able to outperform other means of forecasting (e.g., Erikson and Wlezien 2008). It is also suggested that prediction markets have a number of limitations (e.g., Green, Armstrong, and Graefe 2007). In particular, they may be open to manipulation (e.g., Wolfers and Leigh 2002), though this might actually aid prediction market accuracy (Hanson and Oprea 2009). Again, they may not provide efficient forecasts of low probability events (e.g., Wolfers and Zitzewitz 2004; Smith, Paton, and Vaughan Williams 2006) and may be open to systematic biases, such as optimism bias (Cowgill, Wolfers, and Zitzewitz 2008) and the favorite-longshot bias (e.g., Vaughan Williams and Paton 1997).

The role and value of prediction markets is one of the key areas of inquiry addressed in this special issue. The other is gambling more generally, and its relationship to public policy. For instance, since gambling is highly taxed, a key research question is how output and performance respond to changes in taxation. Also, since the industry is highly computerized, it would also be useful to analyze the relationship between investment in information technology and performance.

2. Papers in the Special Issue

To address these topics, we issued a general Call for Papers, which appeared on the Financial Economics Network, Economics Research Network, TIMS/ORSA, and the Royal Economic Society. Selected papers that were submitted in response to this call were presented at a special issue conference held in September 2008 at Nottingham Business School, Nottingham Trent University, devoted to the theme of "Gambling, Prediction Markets and Public Policy." Sixteen of the papers presented at the conference were Submitted to be considered for publication in the special issue. These were externally reviewed according to standard policies of the Southern Economic Journal. Five of these papers were eventually selected for this special issue.

The articles in the special issue address four key themes: (i) the relationship between information, market efficiency, and scope for regulation; (ii) assessing the "favorite-longshot bias"; (iii) evaluating the potential for manipulation of gambling and prediction markets; and (iv) how does information technology influence productivity in gambling? In the remainder of this essay, we provide focused summaries of the articles in the special issue.

Chung and Hwang

There have been numerous studies of the efficiency of sports betting markets (e.g., Sauer 1998; Vaughan Williams 1999; Paton, Siegel, and Vaughan Williams 2009). Many of these studies have documented the existence of a "favorite-longshot bias," which is an apparent inefficiency in the sports betting market (e.g., Thaler and Ziemba 1988; Vaughan Williams and Paton 1997). Most of these studies have focused on horse racing.

The paper by Chung and Hwang is an empirical test of the favorite-longshot bias based on non--horseracing gambling activity. The authors analyze data from the sports lottery market in Korea, where betting is based on the outcomes of soccer matches in the English Premier League. They compare winning payoffs in this pari-mutuel-type sports lottery against corresponding payoffs from established U.K. bookmakers. An interesting finding of the study is the existence of a favorite-longshot bias in sports betting, even though the threat of insider trading is smaller than it is horse racing. On the other hand, Korean participants in the parimutuel sports lottery market demonstrate a reverse favorite-longshot bias, in contrast to the U.K. bookmaker market.

Johnson, O'Brien, and Sung

The paper by Johnnie Johnson, Raymond O'Brien, and Ming-Chien Sung examines the basis for regulating the flow of information in betting markets. The authors focus in particular on the case in which the set of information available to bettors changes as a result of an intervention by racetrack managers. A common example is a deliberate alteration to the ground conditions between races. The authors categorize this type of information as "dynamic and implicit". Now, if bettors are unable to incorporate such changes to information effectively in their probability judgements, there is a case for regulatory intervention either to limit such practices or to require full public disclosure.

Note that whether or not bettors are able to efficiently process dynamic information sets is not clear a priori. It may be, for example, that bettors can use heuristics based on information they already possess or on historical data in such a way that the effects of managed changes to information are minimized. Consequently the authors proceed to explore these possibilities on a large sample data drawn from horse-race betting with U.K. bookmakers at a single racetrack between 1995 and 2000. They focus on the possible impact of dynamic information regarding the horse's post position (distance from the barrier, denoted as post position, or PP) at the start of a race.

The results are striking. Bettors appear to be very adept at factoring changes in track management practice into their probability calculations, even where the changes are not announced publicly. Learning through experience appears to be a driving factor in this efficient processing of complex dynamic information. This result limits the case for regulatory intervention to force race-track managers to disclose particular management practices. Clearly, although the particular analysis in this paper provides focused lessons for regulation of betting, there are more general lessons that can be learned from the approach employed here. Regulators of financial markets often seem to face a conflict between improving market efficiency (through regulating for better information flows) and reducing the costs of regulatory burdens on markets. Identifying contexts within which market participants are able successfully to process information even when this is not made explicitly public may help to reduce this apparent conflict.

Marginson

A key component of the recent rise in gambling (see Siegel and Anders 2001) is an increase in person-to-person internet betting (O'Connor 2007). A critical research question concerning this activity is whether these new Internet betting exchange markets are efficient and whether they can be manipulated.

The article by Marginson constitutes the first study to apply Shin's measure of market efficiency to these exchanges. A key result of this paper is that such markets may be subject to more manipulation than previously reported in a study conducted by Smith, Paton, and Vaughan Williams (2006), who concluded that these markets have fewer biases than bookmaker-dominated markets.

Marginson analyzes data from 1165 horse races run in the United Kingdom over a two-year period, as well as bets that were placed on these races on Betfair, the world's leading person-to-person internet betting exchange (O'Connor 2007). Based on these data, he concludes that insider trading is likely to be a serious problem in internet markets, since activity aimed at profiting from "known losers" is more likely to occur on such exchanges.

Paton, Siegel, and Vaughan Williams

Although gambling is one of the fastest-growing service industries, there have been no studies of total factor productivity (TFP) in this sector. The paper by Paton, Siegel, and Vaughan Williams fills this gap, based on an analysis of U.K. establishment-level data and a stochastic frontier production function framework, which they use to construct estimates of TFP. The authors also consider several key measurement issues relating to the estimation of gambling productivity, focusing in particular on the impact of information technology on labor productivity and TFP.

The empirical results suggest that the production function models fit well, generating plausible elasticity estimates of the production function parameters. An interesting finding is that productivity increased following major reforms to gambling taxation in 2001. There also appear to be limited evidence of regional variation in TFP. Another key result is that internet operations appear to be associated with higher relative efficiency.

Peirson and Smith

The paper by John Peirson and Mike Smith presents both theoretical and empirical evidence on pricing by fixed-odds bookmakers. A key insight of this paper is that anomalous pricing patterns such as the favorite-longshot bias in fixed odds betting can be explained not only by the presence in the market of bettors with privileged (or inside) information, but also by the presence of bettors who are experts in the field but who do not have privileged information. The authors point out that previous work such as that of Shin (1993) has tended to conflate the impact of both these groups on odds-setting. In the empirical section of the paper, the authors attempt to isolate the impact of inside information from publicly available expert information by estimating anomalous pricing on a sample of races involving two-year-old racehorses in the United Kingdom. It is reasonable to assume these horses, most of which have never raced before, will be subject to a greater degree of insider trading than those that have already raced. Accordingly, the authors split their sample along these lines expecting to find a greater odds bias for unraced two-year-old horses. The empirical estimates strongly support the authors' hypothesis.

The estimates of insider trading derived from the authors' model also have clear implications for regulatory policy in this area. Price biases due to insider trading impose clear and unfair costs on bettors without insider information. In contrast, price biases due to efficient use of publicly available information represent a fair reward for the (presumably costly) accrual of expert knowledge. The authors urge regulators seeking to restrict the consequences of insider trading to take care so that in so doing they do not eliminate the rewards to superior processing of public information.

References

Athanasoulis, S., R. Shiller, and E. T. Rietz. 1999. Macro-markets and financial security. Economic Policy Review 5(1):21-39.

Cowgill, B., J. Wolfers, and E. Zitzewitz. 2008. Using prediction markets to track information flows: Evidence from Google. Dartmouth College: http://www.bocowgill.com/ GooglePredictionMarketPaper.pdf.

Erikson, R. S., and C. Wlezien. 2008. Are political markets really superior to polls as election predictors? Public Opinion Quarterly 72(2):190-215.

Green, K. C., J. S. Armstrong, and A. Graefe. 2007. Methods to elicit forecasts from groups: Delphi and prediction markets compared. Foresight: The International Journal of Applied Forecasting 8:17-20.

Hanson, R., and R. Oprea. 2009. A manipulator can aid prediction market accuracy. Economica 76(302):304-11.

Leigh, A., and J. Wolfers. 2007. Prediction markets for business and public policy. The Melbourne Review 3(1):7-15.

O'Connor, N. A. 2007. Betting exchanges--disruptive innovation at work. Available from: Bettingmarket.com. Accessed 3 January 2008.

Paton, D., D. S. Siegel, and L. Vaughan Williams. 2009. The growth of gambling and prediction markets: Economic and financial implications. Economica 76(302):219-24.

Polgreen, P. M., F. D. Nelson, and G. R. Newmann. 2007. Use of prediction markets to forecast infectious disease activity. Clinical Infectious Diseases 44:272-9.

Rhode, P. W., and K. S. Strumpf. 2004. Historical presidential betting markets. Journal of Economic Perspectives 18(2):127-41.

Sauer, R. 1998. The economics of wagering markets. Journal of Economic Literature 36:2021-64.

Shiller, R. 2003. The new financial order: Risk in the twenty-first century. Princeton, NJ: Princeton University Press.

Shin, H. S. 1993. Measuring the incidence of insider trading in a market for state contingent claims. Economic Journal 103:1141-53.

Siegel, D., and G. Anders. 2001. The impact of Indian casinos on state lotteries: A case study of Arizona. Public Finance Review 29(2):139-47.

Smith, M., D. Paton, and L. Vaughan Williams. 2006. Market efficiency in person-to-person betting exchanges. Economica 73(292):673-89.

Snowberg, E., J. Wolfers, and E. Zitzewitz. 2005. Information (in)efficiency in prediction markets. In Financial and betting markets, edited by L. Vaughan Williams. Cambridge, MA: Cambridge University Press, pp. 366-86.

Thaler, R., and W. T. Ziemba. 1988. Anomalie--Parimutuel betting markets: Racetracks and lotteries. Journal of Economic Perspectives 2:161-74.

Vaughan Williams, L. 1999. Information efficiency in betting markets: A survey. Bulletin of Economic Research 51:1-30.

Vaughan Williams, L., and D. Paton. 1997. Why is there a favourite-longshot bias in British racetrack betting markets? Economic Journal 107:150-8.

Wang, S.-C., J.-J. Tseng, S.-P. Li, and S.-H. Chen. 2009. Prediction of bird flu A (H5N1) outbreaks in Taiwan by online auction: Experimental results. New Mathematics and Natural Computation 2:271-80.

Wolfers, J., and A. Leigh. 2002. Three tools for forecasting federal elections: Lessons from 2001. Australian Journal of Political Science 37(2):223-40.

Wolfers, J., and E. Zitzewitz. 2004. Prediction markets. Journal of Economic Perspectives 18(2):107-26.

David Paton, * Donald S. Siegel, ([dagger]) and Leighton Vaughan Williams ([double dagger])

* Nottingham University Business School, Nottingham NG8 1BB, United Kingdom; E-mail David.Paton@ Nottingham.ac.uk.

([dagger]) School of Business, University at Albany, SUNY, 1400 Washington Avenue, Albany, NY 12222. USA: E-mail DSiegel@uamail.albany.edu.

([double dagger]) Nottingham Business School, Nottingham Trent University, Burton Street, Nottingham NG1 4BU, United Kingdom; E-mail Leighton.Vaughan-Williams@ntu.ac.uk; corresponding author.

We thank participants at the "Gambling, Prediction Markets and Public Policy" workshop held at Nottingham Business School, Nottingham Trent University in September 2008 for comments and suggestions. Financial support from the Nottingham University Business School, Nottingham Trent University, Intrade, and the University of Buckingham Press is also greatly appreciated.