Economica (2009) 76, 219224

doi:10.1111/j.1468-0335.2008.00753.x

The Growth of Gambling and Prediction Markets:

Economic and Financial Implications
By DAVID PATONw, DONALD S. SIEGELz and LEIGHTON VAUGHAN WILLIAMSzz
wNottingham University Business School
zUniversity at Albany, SUNY
Business School

zzNottingham

Final version received 2 June 2008.

In recent years, there has been a substantial global increase in gambling and prediction markets, including
casinos, sports betting, lotteries, elections and wagering on nancial instruments. This trend has
heightened interest in numerous economic and nancial issues related to this sector. These include
questions relating to the efciency of these markets, heterogeneity in risk attitudes among economic
agents, and the use of prediction markets in policy analysis. The papers in this special issue provide a mix
of theoretical and empirical evidence on these issues.

INTRODUCTION
In recent years, there has been a substantial global increase in gambling and prediction
markets, including casinos, sports betting, lotteries, elections and wagering on nancial
instruments. This trend has heightened interest in numerous economic and nancial
issues related to this sector. These include questions relating to the efciency of these
markets, heterogeneity in risk attitudes among economic agents, and the use of
prediction markets in policy analysis.
We have also witnessed major changes in the taxation and regulation of gambling in
the UK and other nations, as well as technological innovations that inuence this sector
(e.g. the internet). These changes have stimulated numerous research and consultancy
projects related to this rapidly growing sector. Such projects include analyses of the
economic and social impacts of gambling, as well as examinations of the manner in which
betting markets can be used as a tool for aggregating decentralized information so as to
produce efcient and unbiased forecasts. This emerging interest in what have come to be
known as prediction markets has generated considerable interest among academics in
determining how best to design and utilize these new markets.
A convenient denition of a prediction market is a speculative market that is created
for the purpose of making a prediction. In such a market, an asset is created whose nal
cash value is tied to a particular event (e.g. who will win the US Presidency) or key
economic indicator (e.g. a rms total sales revenue in the next quarter). The market price
of the asset is then interpreted as a prediction of the probability of the event or the
expected value of the parameter. Traders who have bought low and sold high are
rewarded for improving the market prediction, while traders who buy high and sell low
are penalized for reducing the accuracy of the market prediction.
There is now a strong body of evidence suggesting that prediction markets may be
more accurate than other mechanisms (e.g. election polls), in terms of predicting
outcomes. At the very least, they elicit a different perspective on how people actually
believe the future will play out compared to conventional surveys. As noted by Levitt
(2007), corporations are now using prediction markets as a decision support tool.
The link between gambling and prediction markets can be traced in the UK to the
abolition of a turnover tax on betting in 2001, and its replacement with a much lower
r The London School of Economics and Political Science 2008

220

ECONOMICA

[APRIL

effective incidence of taxation on the gross prots of companies offering betting services
(Paton et al. 2002, 2004). In concert with the recent rapid expansion of person-to-person
betting exchanges, with relatively low margins, the extent to which price biases may exist
in these markets, and more generally their ability to forecast a range of outcomes, has
been addressed and tested across a range of published outputs in the recent academic
literature (e.g. Smith et al. 2006; Borghesi 2007).
The rapid growth in gambling and prediction markets has heightened interest in a
variety of economic and nancial issues related to these activities. Policy-makers seek
guidance on how to tax and regulate these markets at the state, regional, national and
cross-country levels. Furthermore, betting and prediction markets provide a unique and
convenient framework within which to examine fundamental issues relating to traditional
areas of economics.
Our objective in this special issue is to help ll various gaps in this emerging
literature. We issued an open call for papers (which appeared in Economica and also was
distributed to a wider audience) via the Financial Economics Network, Economics
Research Network, TIMS/ORSA and the Royal Economic Society. The leading
contributors to this burgeoning literature were also directly solicited. We received 52
manuscripts. To aid in the development of these papers, a special issue workshop was
held at the UC-Riverside Palm Desert Graduate Center on 2122 May 2007. Each paper
was allocated an assigned discussant who provided feedback to the authors. After the
workshop, each author was invited to resubmit his or her paper. The papers were then
sent to external reviewers, following standard procedures in Economica. Five papers were
ultimately accepted for inclusion in the special issue.

PAPERS

IN THE

SPECIAL ISSUE

The articles in the special issue address three key themes:

1. Information efciency in betting and prediction markets.
2. Theoretical and empirical evidence on the favourite/longshot bias in betting markets.
3. The use of prediction markets for policy analysis.
In the remainder of this introductory essay, we provide focused summaries of the
articles in the special issue and then attempt to place them in a broader context.
Hanson and Oprea
The paper by Robin Hanson and Ryan Oprea is a theoretical analysis of the efciency of
low volume information markets (e.g. prediction markets and idea futures). It is
interesting to note that some politicians, journalists and economists have raised concerns
about manipulation in these thinly-traded markets. The purpose of the Hanson and
Oprea paper is to assess the impact of manipulators on the efciency of information
markets. In modelling the behaviour of agents who participate in these markets, the
authors identify three types of traders: liquidity traders, informed traders, and
manipulators.
In their model, manipulators function as noise traders, in the sense that they induce
more traders to become better informed. Other features of the model are that there are
clues provided regarding the true asset value by a manipulator, and the authors also
allow for irrational choices, provided by the quantal response equilibrium (McKelvey
r The London School of Economics and Political Science 2008

2009]

THE GROWTH OF GAMBLING AND PREDICTION MARKETS

221

and Palfrey 1995; Goeree and Holt 2001). Their key theoretical result is that
manipulators play an important social role by increasing the efciency of information
accuracy. Specically, their actions result in an increase in the variance of the target
price. This, in turn, increases the accuracy of the average price, by increasing the returns
to informed trading and thus the incentives traders have to become informed. We believe
that the Hanson and Oprea result constitutes fertile ground for experimental research.

Wolfers and Zitzewitz

The paper by Justin Wolfers and Eric Zitzewitz advances traditional retrospective
analyses of the expected effects of policy changes (see Wolfers and Zitzewitz 2004;
Snowberg et al. 2005) and the forecasting efciency of betting markets (e.g. Strumpf and
Rhode 2004; Vaughan Williams 2005), by demonstrating how prediction markets can be
used to prospectively estimate policy effects. The example they use to illustrate this is
market trading in contracts tied to the removal from power of Saddam Hussein, which
provides information regarding nancial market participants expectations of the
consequences of the Iraq war. They conducted an ex ante analysis, disseminated before
the war, which revealed that a 10% increase in the probability of war was accompanied
by a $1 increase in spot oil prices in the futures markets. Equity price movements implied
that the same shock led to a 1.5% decline in the S&P 500.
More generally, Wolfers and Zitzewitz used the existence of widely-traded equity
index options to back out the entire distribution of market expectations of the wars
near-term effects, nding that the ow of war-related news throughout their sample was
able to explain a large proportion of daily oil and equity price movements. Their
subsequent analysis suggested that these relationships continued to hold out of sample.
Most interesting is the implication that this type of analysis can allow us to characterize
which industries and countries are likely to be more or less sensitive to particular types of
breaking news event. Indeed, the authors focus on the more general lessons that can be
learned from their paper by highlighting those features of the case study that make it
particularly amenable to policy analysis, and by discussing some of the issues in applying
this method to other policy contexts.

Peel and Law

Many authors have attempted to reconcile observed behaviour in betting markets with
standard models of attitudes to risk. The paper by David Peel and David Law attempts
to provide a more general non-expected utility model that is capable of explaining what
appears to be risk-seeking behaviour in gambling and risk-averse behaviour in other
contexts such as insurance.
The authors emphasise the existence of heterogeneity in risk attitudes and individual
probability distortions. Allowing for such heterogeneity enables a single model to
encompass a range of observed outcomes. The insight that risk attitudes may differ in
systematic ways associated with cultural or institutional factors is likely to be of
considerable benet to researchers who have long puzzled over why, for example, we nd
a standard favourite-longshot bias in some betting markets such as the UK (see, for
example, Vaughan Williams and Paton 1997) but a reverse bias in others. Peel and Law
demonstrate one application of their model by re-examining existing literature on the
relationship between Tote and bookmaker returns. We believe, however, that the model
r The London School of Economics and Political Science 2008

222

ECONOMICA

[APRIL

proposed in this paper will generate a number of further empirical applications both in
betting markets and in other contexts.
Bruce, Johnson, Peirson and Yu
The theme of differences in underlying characteristics of bettors as a way of explaining
why we observe varying levels of the favourite-longshot bias in different contexts is
continued in the fourth paper in the special issue. Alistair Bruce, Johnnie Johnson, John
Peirson and Jiejun Yu introduce a theoretical model in which bettors differ according to
(i) the level of information they possess, (ii) their motivations for gambling and (iii) the
level of transactions cost. Previous authors have used each of these in isolation as
possible reasons as to why we observe the favourite-longshot bias (see Vaughan Williams
1999). The contribution of this paper is to bring the three streams of literature together
into a single encompassing model.
The authors test their model using a dataset covering betting activity on the same
outcomes by three distinct groupsFbettors at the racetrack, off-track bettors at licensed
bookmaking premises and off-track bettors using the telephone. In this data, they nd
signicant differences in the extent of the favourite-longshot bias that are generally
consistent with their theoretical model.
Link and Scott
The special issue concludes with a paper by Al Link and John Scott, which evaluates the
potential impact of a prediction market on a key public policy initiative: the Small
Business Innovation Research (SBIR) programme (Siegel et al. 2003). SBIR was
established in 1982 by the US government as a set-aside programme. In its current
version, SBIR requires all federal research and development (R&D) funding agencies
sponsoring extramural research to allocate 2.5% of their extramural research budgets to
fund, via a peer-review process, R&D in small (less than 500 employees) rms and
organizations. A key goal of the SBIR programme is the private-sector commercialization of these research projects by these entrepreneurial companies. Indicators of
successful commercialization include the creation and sale of new products, job growth,
patents, copyrights, trademarks, and technology licensing agreements.
Given the early stage nature of much of this research, and the fact that commercialization
involves small rms, who typically have a high failure rate, the likelihood of successful
commercialization is quite low. Link and Scott view these public investments in
commercialization projects as gambles, and conjecture that a prediction market could be
used to enhance the rate of successful commercialization of SBIR projects. Using project-level
data, they present an econometric analysis of whether outside private investors have useful
information about proposed SBIR projects prospects for commercialization. Their ndings
suggest that private investors indeed have such information, which provides support for the
view that a prediction market could improve the performance of the SBIR programme.

CONCLUSION
Many scholars have analysed the efciency of betting markets. While most of these
studies agree that it is difcult to earn abnormal returns in these markets, there exist
established biases across a range of these, perhaps the most famous of which is the
r The London School of Economics and Political Science 2008

2009]

THE GROWTH OF GAMBLING AND PREDICTION MARKETS

223

favouritelongshot bias, i.e. the tendency for wagers placed at shorter odds to yield
higher expected returns than wagers placed at longer odds. The explanation for this bias,
and why it persists in most arenas, but not all (sometimes it is reversed), has long been the
topic of academic scrutiny. The papers by Peel and Law and by Bruce et al. in this special
issue go some way to resolving this puzzle.
The value of betting markets as a forecasting tool is also of considerable current
academic interest, and a number of recent papers have addressed this from various
angles, including the aggregation of information from different markets (Paton and
Vaughan Williams 2005), the inuence of market ecology (Sung and Johnson 2007) and
optimal design (e.g. Servan-Schreiber et al. 2004). The key questions are whether these
markets are able to add value to traditional forecasting methodologies, and if so to what
extent and with what caveats. A signicant amount of this research effort has to date
focused on the value of these markets in predicting the outcomes of political elections
(see, for example, Strumpf and Rhode 2004), although the focus of other studies ranges
from the sales of Hewlett-Packard printers to the probability of meeting project deadlines
at Google (Leigh and Wolfers 2007). There is also growing interest in the value of
contingent markets (If event X occurs, how will this affect outcome Y?).
As these markets become more prevalent, it is inevitable that questions will be raised
regarding whether these markets can be manipulated. The paper by Hanson and Oprea,
included in this volume, is timely in addressing this important issue.
Another key issue is how prediction markets can be used as an independent estimator
of the impact of a unique event, such as the outcome of an election, on key economic
variables (see Snowberg et al. 2007). The paper by Wolfers and Zitzewitz makes a
valuable contribution to this aspect of prediction markets. Most importantly, their article
provides us with several general lessons that can be applied across a range of policy
contexts.
Finally, the paper by Link and Scott provides an interesting perspective on this very issue
of policy context, by evaluating the potential impact of a prediction market on a key public
policy initiative, namely the Small Business Innovation Research (SBIR) programme.
Future research can usefully build on all these areas of current public discussion to
generate ideas as to how prediction markets can best be used to add maximum value to
existing forecasting methodologies as well as to most fully inform the widest range of policy
contexts. We believe that this special issue of Economica makes a useful contribution to
improving our understanding of the policy implications of gambling and prediction markets.
ACKNOWLEDGEMENTS
We thank Steve Levitt, William Eadington and participants at the workshop The Growth of
Gambling and Prediction Markets: Economic and Financial Implications held at the UCRiverside Palm Desert Graduate Center on 2122 May 2007, for comments and suggestions. We
are especially grateful to Carolyn Stark, Jessica Enders, Ron Willison and Toni Lawrence of the
UC-Riverside Palm Desert Graduate Center for helping us to organize the workshop. Financial
support from the A. Gary Anderson Graduate School of Management at UC-Riverside, the
Nottingham University Business School, Nottingham Trent University, and the University of
Buckingham Press is also greatly appreciated.
REFERENCES
BORGHESI, R. (2007). Price biases in a prediction market: NFL contracts on tradesports. Journal of Prediction
Markets, 1(3), 23353.
GOEREE, J. K. and HOLT, C. A. (2001). Ten little treasures of game theory and ten intuitive contradictions.
American Economic Review, 91(5), 140222.
r The London School of Economics and Political Science 2008

224

ECONOMICA

[APRIL

LEIGH, A. and WOLFERS, J. (2007). Prediction markets for business and public policy. Melbourne Review, 3(1),
715.
LEVITT, S. (2007). Keynote Address, Workshop on The Growth of Gambling and Prediction Markets:
Economic and Financial Implications, UC-Riverside Palm Desert Graduate Center, 21 May, Palm Desert,
CA (http://palmdesert.ucr.edu).
MCKELVEY, R. D. and PALFREY, T. (1995). Quantal response equilibria for normal form games. Games and
Economic Behavior, 10, 638.
PATON, D. and VAUGHAN WILLIAMS, L. (2005). Forecasting outcomes in spread betting markets: can bettors
use quarbs to beat the book? Journal of Forecasting, 24(2), 13954.
FFF, SIEGEL, D. and VAUGHAN WILLIAMS, L. (2002). A policy response to the e-commerce revolution: the
case of betting taxation in the UK. Economic Journal, 112(480), F296F314.
FFF, FFF and FFF (2004). Taxation and the demand for gambling: new evidence from the United
Kingdom. National Tax Journal, 57(4), 84761.
SERVAN-SCHREIBER, E., WOLFERS, J., PENNOCK, D. and GALEBACH, B. (2004). Prediction markets: does
money matter? Electronic Markets, 14(3), 111.
SIEGEL, D. S., WESSNER, C., BINKS, M. and LOCKETT, A (2003). Policies promoting innovation in small rms:
evidence from the US and U.K. Small Business Economics, 20(2), 1217.
SMITH, M., PATON, D. and VAUGHAN WILLIAMS, L. (2006). Market efciency in person-to-person betting
exchanges. Economica, 73(292), 67389.
SNOWBERG, E., WOLFERS, J. and ZITZEWITZ, E. (2005). Information (in)efciency in prediction markets. In L.
Vaughan Williams (ed.) Information Efciency in Financial and Betting Markets. Cambridge: Cambridge
University Press, pp. 36686.
FFF, FFF and FFF (2007). Partisan impacts on the economy: evidence from prediction markets and
close elections. Quarterly Journal of Economics, 122(2), 80729.
STRUMPF, K. S. and RHODE, P. W. (2004). Historical presidential betting markets. Journal of Economic
Perspectives, 18(2), 12741.
SUNG, M. and JOHNSON, J. E. V. (2007). The inuence of market ecology on market efciency: evidence from a
speculative nancial market. Journal of Gambling Business and Economics, 1(3), 18598.
VAUGHAN WILLIAMS, L. (1999). Information efciency in betting markets: a survey. Bulletin of Economic
Research, 51(1), 130.
VAUGHAN WILLIAMS, L. (2005). Models, markets, polls and pundits: a case study of information efciency. In
L. Vaughan Williams (ed.) Information Efciency in Financial and Betting Markets. Cambridge: Cambridge
University Press, pp. 193214.
VAUGHAN WILLIAMS, L. and PATON, D. (1997). Why is there a favouritelongshot bias in British racetrack
betting markets? Economic Journal, 107, 1508.
WOLFERS, J. and ZITZEWITZ, E. (2004). Prediction markets. Journal of Economic Perspectives, 18(2), 10726.

r The London School of Economics and Political Science 2008