Joyce E. Berg

Prediction Markets as Decision Support Systems

Valuations from “prediction markets” reveal expectations about the likelihood of events. “Conditional prediction markets” reveal expectations conditional on other events occurring. For example, in 1996, the Iowa Electronic Markets (IEM) ran markets to predict the chances that different candidates would become the Republican Presidential nominee. Other concurrent IEM markets predicted the vote shares that each party would receive conditional on the Republican nominee chosen. Here, using these markets as examples, we show how such markets could be used for decision support. In this example, Republicans could have inferred that Dole was a weak candidate and that his nomination would result in a Clinton victory. This is only one example of the widespread potential for using specific decision support markets.

What Makes Markets Predict Well? Evidence from the Iowa Electronic Markets

We use the data from the Iowa Electronic Markets to study factors associated with the ability of markets to predict future events. These are large-scale, real-money experimental markets with contract payoffs determined by political election outcomes. They provide data about individual trader characteristics and market micro-behavior which is not available from larger exchanges. In this study we find that market characteristics motivated by financial theory and previous experimental research account for most of the variance in predictive accuracy across sixteen markets. Three variables are particularly important: 1) the number of contract types traded, 2) pre-election market volumes and 3) differences in election eve (weighted) market bid and ask queues.

Chapter 80 Results from a Dozen Years of Election Futures Markets Research

Introduction and description of election futures markets The Iowa Electronic Markets are small-scale, real-money futures markets conducted by the University of Iowa College of Business. In this review we focus on the best known of these markets, The Iowa Political Markets. Contracts in these markets are designed so that prices should predict election outcomes. The data set contains the results of 49 markets covering 41 elections in 13 countries. The Iowa Markets operate 24-hours a day, using a continuous, double-auction trading mechanism. Traders invest their own funds, make their own trades, and conduct their own information search. The markets occupy a niche between the stylized, tightly controlled markets conducted in the laboratory and the information-rich environments of naturally occurring markets. By virtue of this design, the Iowa Markets provide data to researchers that is not otherwise available.

The Effect of Electronic Markets on Forecasts of New Product Success

In this paper, we extend field experiments of real money prediction markets to the problem of forecasting the success of a new product. We collect forecasts using a traditional survey mechanism and a market mechanism. Our results suggest that market prices summarize the information contained in survey forecasts and improve those forecasts by reducing the variability of the forecast. However, we find no evidence of a “crystal ball” equilibrium. Our markets have considerable variability and predict only as well as the public signal provided by the HSX movie market game.

Searching for Google's Value: Using Prediction Markets to Forecast Market Capitalization Prior to an Initial Public Offering

We conducted prediction markets designed to forecast post-initial public offering (IPO) valuations before a particularly unique IPO: Google. The prediction markets forecast Googles post-IPO market capitalization relatively accurately. While Googles auction-based IPO price was 15.3% below the first-day closing market capitalization, the final prediction market forecast was only 4.0% above it. The forecast also accorded with the level of over-subscription in the IPO auction. Evidence available to both outsiders (from the prediction market forecasts) and insiders (through the orders in Googles auction) predicted similar degrees of underpricing. We argue that, with repetition, such markets could provide useful information for understanding the IPO process.

Judgment and decision-making research in accounting and auditing: The individual versus the aggregate

Introduction To what degree do individual decision biases affect aggregate behavior? This question was introduced, and “answered, ” in the accounting literature twenty years ago when Gonedes and Dopuch (1974) argued that market efficiency necessarily precluded any impact of individual bias on aggregate capital–market behavior (that is, price). We know now that this claim need not be true. Recent advances in both theoretical and empirical research open the door for the influence of individual bias on aggregate–level behavior in capital markets as well as other aggregate settings. Experimental methods enhance our ability to pinpoint when biases do occur, measure the cost of bias, and examine what factors extinguish biases. In this chapter, we review the historical development of the issue of individual and aggregate behavior and develop a framework to systematically advance our knowledge in this area. Since no generally accepted theory linking individual behavior to aggregate level behavior exists, we develop a framework enumerating the observable factors that distinguish individual decision–making settings from aggregate decision–making settings. Since these factors transcend theoretical paradigms, they form the basis for dialog between those that draw theory from economics and those that draw theory from psychology. In the spirit of enhancing such a dialog, we use this framework to examine several streams of research, and begin to address how changes in observable factors affect aggregate behavior. In examining these settings, we ask not only whether individual biases persist at the aggregate level, but also whether aggregate settings introduce “new” biases of their own.

Longshots, Overconfidence and Efficiency on the Iowa Electronic Market

A basic proposition of behavioral finance is that individual decision-making biases affect financial markets. We examine this proposition using data from the Iowa Electronic Market (IEM), a real money market conducted for teaching and research purposes. We ask whether the longshot bias or an overconfidence bias affects market prices. In our market context, these two biases yield opposing predictions. The IEM is ideal for answering these questions because it mixes many desirable research features from betting markets (where the longshot bias is observed) with a closer parallel to naturally occurring financial markets (where researchers look for evidence of overconfidence). While the longshot bias affects many sports betting markets robustly, no such bias appears here. Nor does overconfidence influence prices at short horizons. If there is a bias, it results from overconfident traders at long horizons. While the markets incorporate information efficiently at short horizons, non-market data indicate some long-horizon inefficiency. When markets appear inefficient, we calculate Sharpe ratios for static trading strategies and document returns for dynamic trading strategies to show the economic content of the inefficiencies.

Preference Reversals: The Impact of Truth-Revealing Monetary Incentives

Researchers vigorously debate the impact of incentives in preference reversal experiments. Do incentives alter behavior and generate economically consistent choices? Lichtenstein and Slovic (1971) document inconsistencies (reversals) in revealed preference in gamble pairs across paired choice and individual pricing tasks. The observed pattern is inconsistent with stable underlying preferences expressed with simple errors. Lichtenstein and Slovic (1973) and Grether and Plott (1979) introduce incentives, but aggregate reversal rates change little. These results fostered numerous replications and assertions that models of non-stable preferences are required to explain reversals. Contrary to this research, we find that incentives can generate more economically consistent behavior. Our reevaluation of existing experimental data shows that incentives have a clear impact by better aligning aggregate choices and prices. The effect is sufficiently large that, with truth-revealing incentives, a stable-preferences-with-error model not only explains behavior, but fits the data as well as any model possibly could.

Preference Reversals and Induced Risk Preferences: Evidence for Noisy Maximization

We combine two research lines: preference reversal research (Lichtenstein and Slovic, 1971) and research on lottery-based risk preference induction (Roth and Malouf, 1979). Our results are informative for both research lines. We show that inducing risk preferences in preference reversal experiments has dramatic effects. First, while our subjects still display reversals, they do not display the usual pattern of “predicted” reversals suggested by the compatibility hypothesis. By inducing risk averse and risk loving preferences, we can dramatically reduce reversal rates and even produce the opposite pattern of reversals. Our results are consistent with the assumption that subjects maximize expected utility with error. This provides evidence that Camerer and Hogarths (1999) framework for incentive effects can be extended to include the risk preference induction reward scheme.

Memoirs of an indifferent trader: Estimating forecast distributions from prediction markets

Prediction markets for future events are increasingly common and they often trade several contracts for the same event. This paper considers the distribution of a normative risk-neutral trader who, given any portfolio of contracts traded on the event, would choose not to reallocate that portfolio of contracts even if transactions costs were zero. Because common parametric distributions can conflict with observed prediction market prices, the distribution is given a nonparametric representation together with a prior distribution favoring smooth and concentrated distributions. Posterior modal distributions are found for popular vote shares of the U.S. presidential candidates in the 100 days leading up to the elections of 1992, 1996, 2000, and 2004, using bid and ask prices on multiple contracts from the Iowa Electronic Markets. On some days, the distributions are multimodal or substantially asymmetric. The derived distributions are more concentrated than the historical distribution of popular vote shares in presidential elections, but do not tend to become more concentrated as time to elections diminishes.