MARKETS AS AN INFORMATION AGGREGATION MECHANISM FOR DECISION SUPPORT

Abstract

In almost all walks of life, predicting uncertain future events plays an essential role in decision-making processes. However, information related to future events frequently exists only as dispersed opinions, insights, and intuitions of individuals. Each individual only knows a little, but aggregating the dispersed information together may make considerable contribution to decision making. This is typical in many domains including business, politics, and entertainment. Therefore, how to aggregate such dispersed information for useful decision support is a crucial task.

Markets have shown great potential as one of the most effective mechanisms for gathering distributed information and generating accurate forecasts, often surpassing many existing methods in practice. This research studies information markets, markets that are specially designed for information aggregation and forecasting, from four different perspectives: theoretical examination, experimental evaluation, empirical analysis, and design.

With the ultimate goal of better understanding information markets as a forecasting device, this thesis makes four contributions to the field of information markets. The first contribution is a theoretical model of information markets that generalizes an existing model to situations with aggregate uncertainty, which is ubiquitous in the real world. It helps answering the question of why information markets work, by modeling how information flows from traders to the market and back again, and characterizing convergence properties of information markets.

The second contribution is an experimental evaluation of several theoretical models of information markets. Because theoretical models often have to make simplified assumptions about human behavior for tractableness, we use human subject experiments to test them, while still maintaining close parallel settings with the theoretical models. Results of this part demonstrate whether and to what extent theoretical models are supported in a more realistic environment and point out important areas to be improved by theoretical models.

The third contribution is an initial attempt to compare the prediction accuracy of information markets and opinion pools using real-world market data. The results provide insights into the predictive performance of information markets, and the relative merits of selecting among various opinion pooling methods.

The last contribution of the thesis is a generic framework of information market development. Although evidence has shown that information markets can make accurate predictions, there are certainly cases that markets fail. How to design an information market for accurate predictions in practice remains an open question. To facilitate the development process, the proposed framework illustrates the life cycle of information market development and explains issues to be considered at each stage.

Chapter 1 Introduction

1.1      Problem Statement

Forecasting seems to be an ubiquitous endeavor in human societies. For instance, governments conjecture the effects of alternative policies; businesses project product sales revenue; meteorologists forecast future weather conditions; financial analysts predict stock price trends; and individuals bet on outcomes of sport games. In almost all walks of life, predicting uncertain future events plays a crucial role in decision support.

Many forecasting problems have at least two characteristics in common. First, the uncertainty of the problem changes as time goes by and new relevant information appears. Second, information related to the forecasting problems frequently only exists as the dispersed opinions, insights, and intuitions of individuals. Each individual knows very little, but aggregating the dispersed information together can make considerable contribution to decision-making. This is especially typical in situations such as supply chain management, business forecasting, new product development, policy analysis, and sports betting. Thus, how to make timely and accurate predictions that make use of such bits and pieces of information is very important, which is also the scope of the thesis.

1.2     Motivation

For decades, scientists have devoted themselves to developing and exploring various forecasting methods, which can be roughly divided into statistical and non-statistical approaches. Statistical methods, including econometric models and some machine learning techniques, are based on historical data. Non-statistical methods frequently rely on expert judgment and opinions. However, both of these approaches have limitations. Statistical methods require not only the existence of enough historical data but also that past data contain valuable information about the future event. Eliciting expert opinions means identifying experts, soliciting their participation, and determining how to combine different opinions when experts are not in agreement, which are often not easy [4, 6, 28, 35] .

With the fast growth of the Internet, information markets have recently emerged as a promising alternative forecasting tool. Also called prediction markets, decision markets, event markets, or virtual stock markets, information markets are markets that are specially designed for aggregating information and making predictions about future events. Such markets are becoming very popular online. The Iowa Electronic Markets (IEM) [50] are real-money futures markets to predict economic and political events such as presidential elections. Hollywood Stock Exchange (HSX) [21] trades securities to forecast future box office proceeds of new movies. Tradesports.com [72] , a betting exchange registered in

Ireland, hosts markets for sports, political, entertainment, and financial events. Foresight

Exchange (FX) [20] allows traders to wager on unresolved scientific questions or other claims of public interest. NewsFutures.com’s World News Exchange [54] has very popular sports and financial betting markets. MIT’s Innovation Futures [29] predict important business and technology trends. Tech Buzz Game [30] aims at both forecasting high-tech trends and testing a new market mechanism.

Information markets as a forecasting method have many advantages. Compared with statistical forecasting methods, information markets can incorporate real-time information, which was not contained in historical data. Compared with eliciting expert opinions, information markets are less constrained by space and time; they eliminate the effort of identifying experts and soliciting their participation, and hence are often less expensive in practice; and they do not need to deal with conflicting opinions. More importantly, information markets can potentially make real-time predictions that take advantage of the dispersed information, which are sometimes hard to capture using other forecasting methods.

Despite merits and popularity of information markets, why they work, how well they perform, and how to design effective information markets are still open questions to a large extent. If information markets are to be used to assist businesses, universities, and governments in making critical decisions in the real world, investigating these questions are imperative. The thesis is an effort on this track. It aims at providing a comprehensive understanding on properties and performance of information markets, through rigorous theoretical, experimental, and empirical examinations, and obtaining an initial framework to guide information market design and development.

1.3       Purposes of the Study

We investigate information markets from four related approaches: theoretical examination, experimental evaluation, empirical analysis, and design. Table 1.1 shows the general and specific research questions that we intend to address with each approach. Theoretical examination can help understand the general question of why information markets work. This is achieved through developing computational models of information markets. Experimental evaluation uses human subjects to test theoretical models in a controlled laboratory environment and identify where to improve. Empirical analysis using real-world datasets aims at investigating the actual predictive performance of information markets. Based on previous results, we then investigate issues of information market design and development. We will discuss our specific research questions in later sections when we actually address them.

1.4      Thesis Organization

The remainder of the thesis is organized as follow. Chapter 2 introduces basics of information markets and reviews related work. Chapter 3 covers the theoretical examination of information markets, in which we present a theoretical model of information markets, and discuss properties of information markets based on the model. Experiments to evaluate several theoretical models are the theme of Chapter 4. Chapter 5 compares information markets and various opinion aggregation methods in terms of prediction accuracy. Based Table 1.1.  Research Approaches and Research Questions

Research Approaches	General Research Questions                  Specific Research Questions
Theoretical Examination	Will an information market converge to a consensus equilibrium? If yes, how fast is the convergence process? Why do information market work? What         is             the         best    possible equilibrium? Will an information market always converge to it?
Experimental Evaluation	To what extent, are theoretical models of information markets valid?	Are properties derived from theoretical models supported by experiments? Are assumptions of theoretical models supported by experiments? Which aspects of theoretical models should be improved?
Empirical Analysis	How well do information market work?	How well do information markets perform compared with other forecasting methods, especially opinion pools?
Design & Development	How to develop an effective information markets?	When to choose information markets over other forecasting methods? What are issues to be considered when developing an information market? What are the options for each issue? Is there a generic framework for information market development?

on the evidence of our studies and previous research, Chapter 6 proposes a framework for information market development and identifies issues to be considered in each step of the development process. Chapter 7 concludes the thesis.

MARKETS AS AN INFORMATION AGGREGATION MECHANISM FOR DECISION SUPPORT

Sharing is caring!