# Blackwell Publishers Ltd and the Board of Trustees of the Bulletin of Economic Research 1999. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA Bulletin of Economic Research 51:3, 1999, 0307-3378 REGULATING INSIDER TRADING IN BETTING MARKETS David Paton, Leighton Vaughan Williams and Stuart Fraser ABSTRACT Although trading in securities in conventional financial markets on the basis of inside information is restricted by law, the rules against such trading in betting markets are rather more ambiguous. It is argued in this paper that, since insider trading in betting markets imposes a cost on the great majority of bettors, tighter strictures against such trading would benefit all but the insiders. This case is supported by the use of empirical evidence which shows that betting markets which are characterized by tighter controls against insider activity are also characterized by a significantly lower incidence of such activity. I. INTRODUCTION There have been long-running concerns over the use of insider information in UK sports betting markets. Of particular concern is the fact that the costs of trading on the basis of insider information are borne, either directly or indirectly, by ordinary bettors. In this paper we seek to investigate the impact on insider trading of different regulatory structures. We do this by analysing the extent of insider trading in two parallel betting markets that are distinguished by the legal strictures to which they are subject. II. INSIDER TRADING IN BETTING MARKETS In recent months there has been a great deal of public attention focused on alleged irregularities in football and horse race betting. Most notable We would like to thank Wendy Chapple for her assistance with the data, and Andy Cooke, Steve Heasell and an anonymous referee for helpful comments. 237
238 BULLETIN OF ECONOMIC RESEARCH in this regard have been the two criminal trials during 1997 involving professional footballers accused of match fixing and a 1998 civil case involving a high-profile jockey accused of race fixing. The problem in football was emphasized by Sir John Smith's final report to the FA on Betting on Professional Football within the Professional Game. Sir John Smith admitted that there were difficulties associated with determining the precise extent of betting by players and club officials, but found no evidence to `conclude that it is anything other than a widespread practice' (Smith, 1997, p. 15). Furthermore, he argued that, `using knowledge which is not generally available to the public at large... in their own interest and, therefore, against the interests of others would clearly be quite wrong. It would be a practice as unfair as trading in shares having insider knowledge' (ibid., p. 17). Attempts to date to quantify the extent of insider trading in betting markets have, however, tended to focus on horse race and greyhound betting. Shin (1993), for example, concludes that about 2 per cent of all money bet on horse racing in the UK is placed by insiders, an estimate supported by the work of Vaughan Williams and Paton (1997). Other studies, based on related methodological approaches, by Fingleton and Waldron (1996), for Irish horse racing, and Cain et al. (1997), for UK greyhound racing, find the figure to be even higher in those markets. Such analyses suggest that upwards of 300 million may be wagered by bettors in UK sports betting markets on the basis of inside information. The implication of any reduction in the incidence of insider trading is that bookmakers would be able to offer better odds to all gamblers (Bird and McCrae, 1994). III. THE GROWTH OF SPREAD BETTING MARKETS An interesting feature of the betting industry in recent years has been the development of spread-betting markets. In these markets, a commodity may be bought at one price, or sold at another (lower) price. Returns or losses are equal to the difference between the traded price and the actual outcome. For example, a football supremacy bet operates as follows. If the market-maker expects the home team to win by one goal (a supremacy of 1), potential bettors would be offered a spread either side of this mid-point, for example 0.9 to 1.1. Bettors may `buy' at the upper end of the spread (1.1 in this case) or `sell' at the lower end (0.9). Buyers win (lose) if there is a positive (negative) difference between the actual supremacy and the buying price. The converse applies to sellers. These markets often operate in parallel to traditional fixed-odds markets, meaning that it is possible to place a bet on the same event in different markets. The interesting feature of these two markets from the
INSIDER TRADING IN BETTING MARKETS 239 point of view of investigating insider trading is that they operate under different regulatory structures. Fixed-odds betting is regulated by the 1963 Betting, Gaming and Lotteries Act (as subsequently amended) and trading on the basis of inside information is not in itself illegal. Spread betting is subject to the 1986 Financial Services Act and is treated in the same way as trading on financial markets. Betting in these markets is limited to registered clients and the identity of bettors is known to the spread-betting companies. Consequently, trading on the basis of privately held information is extremely difficult. This provides a convenient (and possibly unique) opportunity in which to analyse the impact of regulation on insider trading. IV. COMPARING DIFFERENT BETTING MARKETS In an influential paper, Shin (1993) presents a theoretical model of the odds-setting process by bookmakers who are faced by an unknown percentage of bettors with insider information. The prediction of this model is that the bookmakers' margin (overround 1 ) is directly related to the percentage of money wagered by insiders (denoted by Shin as `z'). We consider here bets on the results of football matches. Fixed odds are offered on the favourite winning, the draw and the longshot winning. The equivalent spread bet is the supremacy bet described above. In order to estimate Shin's model on the two football betting markets in question, prices in both of the parallel markets need to be translated into probabilities. In the fixed-odds market this can be done directly as each odds value has associated with it an implied probability. In the spread market, the procedure is somewhat more complicated. However, Jackson (1994) shows that the implied probabilities can be estimated by assuming that the number of goals scored by each team can be approximated by Poisson distributions where the parameter of the distribution for each team is given by the mean rate of goal scoring per match. Specifically, consider a match between the favourite, team X, and the longshot, team Y. Let the expected number of goals scored by X be G x and that by Y to be G y. These are the parameters of the two Poisson distributions. Knowledge of the Poisson distribution probabilities and of G x and G y allows computation of the probabilities of the number of goals scored by each team and, consequently, the probabilities of each match outcome. 2 The implied values of the goal rates, G x and G y can be calculated by using two common forms of spread bets: supremacy and total goals. A 1 In a game with n outcomes, the overround (OR) is calculated as P n i ˆ 1 p i, where p i is the probability of outcome i occurring as implied by its odds, where i ˆ 1... n. 2 The probability that X scores r goals is given by e Gx G r x =r!.
240 BULLETIN OF ECONOMIC RESEARCH supremacy bet is on the difference in the number of goals scored by the teams, whereas a total goals bet is on the sum of the goals scored by each team. In each case, a spread is given, bounded by a buying and a selling price. We assume here that the mid-point of the total goals spread (given by M) is the expected total number of goals in the match. If the buying price of the supremacy spread is given by BUY, then M ˆ G x G y and BUY ˆ G x G y. Knowledge of M and BUY allows us to solve for G x and G y and thus to calculate the probabilities of each match outcome based on the supremacy buying price. In the same way, the selling point of the spread (given by SELL) yields the probabilities of the match outcome based on the supremacy selling price. We use data on 257 football matches from the 1996=97 English Premier League. The optimum fixed odds and spread-betting odds offered by all the leading national bookmakers are taken for each match. In each case, we compute six probabilities for each match: favourite win, longshot win and draw based on the fixed odds; favourite win, longshot win and draw based on the appropriate spread prices. Jullien and Salanie (1994) demonstrate how exact nonlinear estimation techniques, specifically the method of moments and nonlinear least-squares, may be used to calculate Shin's `z' value. Applied to our data, the method of moments approach yields a value for z of 3.46 per cent in the fixed odds market and 1.72 per cent in the spread market. Using nonlinear least-squares, we estimate the mean value of z to be 3.12 per cent (standard error 0.078) in the fixed-odds market compared with a value of 1.51 per cent (standard error 0.092) in the spread market. 3 In fact out of the 257 matches, z is larger for the fixed-odds bets in 222 cases for the method of moments estimates and 220 cases for the nonlinear least-squares estimates. A t-test of the null hypothesis that the fixed odds z equals 1.51 per cent yields a test statistic of 20.64 with 770 degrees of freedom. The null is therefore rejected at all levels of significance. 4 Thus, the level of insider trading is estimated to be twice as great in the fixed-odds market as in the spread market. We interpret this as providing support for our hypothesis that the regulatory framework can have a real impact on the extent of insider trading in financial markets. 3 Details of the algorithms used in these calculations are available from the authors on request. 4 In the case of a bet with a constant number of possible outcomes (in this case three), the value of z is likely to be highly correlated with the bookmakers overround. Thus a further test is to compare the overround in each case. In the fixed odds market, the mean overround is 1.069 (standard error 0.0013) whilst in the spread market it is 1.037 (standard error 0.0025). A t-test of the null hypothesis that the overround is equal in both markets is rejected at all levels of significance.
INSIDER TRADING IN BETTING MARKETS 241 V. DISCUSSION Sporting authorities, and to an extent the betting industry, have been somewhat ambivalent about the issue of insider trading. For example, the placing of bets on their own horses by trainers and owners is seen as perfectly acceptable by the internal authorities, although betting by jockeys is not. Yet if bookmakers set odds to take account of any insiders, then ordinary bettors are disadvantaged. The analysis presented here suggests that the regulatory framework has an important influence on the extent of insider trading. Specifically, the incidence of insider trading is significantly lower in the market in which regulation imposes higher costs on insider activity. If the betting industry and sporting authorities are serious about taking steps to cut down on such activity, then one possible approach is to support a change in the law so that insider trading in conventional betting markets would be subject to penalties sufficiently serious that they constitute an effective deterrent. Department of Economics and Politics, Received March 1998 Nottingham Trent University Final version accepted October 1998 REFERENCES Bird, R. and McCrae, M. (1994). `Tests of the efficiency of racetrack betting using bookmaker odds', in: Hausch, D. B., Lo, V. S. Y. and Ziemba, W. T. (eds), Efficiency of Racetrack Betting Markets. Academic Press, London, pp. 593±603. Cain, M., Law, D. and Peel, D. (1997). `Is one price enough to value a statecontingent asset correctly? Evidence from The British Greyhound Gambling Market', Salford Papers in Gambling Studies, vol. 97±102. Fingleton, J. and Waldron, P. (1996). `Optimal determination of bookmakers' betting odds: theory and tests', Trinity Economic Papers, technical paper 9, December. Jackson, D. A. (1994). `Focus on sport: index betting on sports', The Statistician, vol. 43, pp. 309±15. Jullien, B. and Salanie, B. (1994). `Measuring the incidence of insider trading: a comment on Shin', Economic Journal, vol. 104, pp. 1418±19. Shin, H. S. (1993). `Measuring the incidence of insider trading in a market for state contingent claims', Economic Journal, vol. 103, pp. 1141±53. Smith, Sir John (1997). Betting on Professional Football within the Professional Game. Football Association, London. Vaughan Williams, L. and Paton, D. (1997). `Why is there a favourite±longshot bias in British racetrack betting markets?', Economic Journal, vol. 107, pp. 150±8.