C:\papers\ee\entrepreneur\papers\Characterizing the ... - ITD

Characterizing the Entrepreneur Using Field Experiments by Julie Ann Elston, Glenn W. Harrison and E. Elisabet Rutström † October 2005

Abstract. We use field experiments to provide operationally meaningful measures of several important characteristics of entrepreneurs. We examine four characteristics of entrepreneurs that have been variously alleged: that they are risk loving, that they have a joy of winning that is independent of expected profit, that they suffer from judgmental errors that err on the side of optimism, and that they exhibit overconfidence in their own skills leading to excess market entry. In each case we propose an experimental task that can be used to identify evidence for or against these hypothesized characteristics. We then take these tasks into the field and collect responses from small-business entrepreneurs and a control group of nonentrepreneurs. We find striking differences between two types of entrepreneurs, which we call full-time and part-time entrepreneurs. The full-time entrepreneurs are less risk averse than non-entrepreneurs, and the part-time entrepreneurs are more risk averse than non-entrepreneurs. We also find that full-time entrepreneurs exhibit a significant joy of winning. On the other hand, there is no evidence that any of our entrepreneurs exhibited any systematic judgmental errors about the profitability of a bid. Finally, we find that part-time entrepreneurs are extremely reluctant to enter markets where profitability is based on their perception of their relative skill ability, even after controlling for their higher aversion to risk. Full-time entrepreneurs and non-entrepreneurs do not exhibit any systematic over-confidence in their relative skill abilities.

†

Max Planck Institute of Economics and Oregon State University (Elston) and Department of Economics, College of Business Administration, University of Central Florida, USA (Harrison and Rutström). E-mail: [email protected], [email protected] and [email protected]. Rutström thanks the U.S. National Science Foundation for research support under grants NSF/IIS 9817518, NSF/MRI 9871019 and NSF/POWRE 9973669. We are grateful to Mark Schneider and Ryan Brosette for research assistance. All data, statistical code and experimental instructions are available at the public ExLab Digital Library at http://exlab.bus.ucf.edu.

The entrepreneur plays a major role in public policy debate, and is increasingly appearing in formal models of economic growth and development. As with many other words, the term has evolved from originally being a neutral descriptor of someone who manages a public musical institution1 to become something of a pejorative. In fact, in economics the word has undertaken an even less direct path, with many of the early economists taking a decidedly one-sided view of the entrepreneur. As Evans [1949; p. 337] puts it, “Cold-blooded appraisals of the role of the entrepreneur in economic development are rare: glorification is usual.” In fact, there is a faint hint of entrepreneurs being regarded as irrational in the face of adversity. In a passage describing the role of the entrepreneur, Marshall [1927; p.23] could write as follows: “For just as a racehorse or an athlete strains every nerve to get in advance of his competitors, and delights in the strain; so a manufacturer or a trader is often stimulated much more by a hope of victory over his rivals than by the desire to add something to his fortune.” And Schumpeter [1934; p.93] would famously write that entrepreneurs were largely motivated by “... the dream and the will to found a private kingdom ... [and by] the impulse to fight, to prove onself superior to others, to succeed for the sake, not of the fruits of success, but of success itself ... and [by] the joy of creating, of getting things done, or simply of exercising one’s energy and ingenuity.” We propose that one substitute cold-blooded experimental tasks for the use of anecdote to characterize entrepreneurs. We begin by examining four characteristics of entrepreneurs that have been variously alleged in the entrepreneurship literature: that they are risk loving, that they have a joy of winning that is independent of the expected profit, that they suffer from judgmental errors that err on the side of optimism, and that their overconfidence in their own abilities leads to excess entry. In each case we propose an experimental task that can be used to identify evidence for or against these hypothesized characteristics. We then take these tasks into the field and collect responses from

1 The Oxford English Dictionary (Second Edition) dates the use of the word from 1828, although it derives from the French “entreprendre” (to undertake) as early as 1475 in the form of the related word “entreprenour.” The economic use of the term to mean “one who undertakes an enterprise; one who owns and manages a business; a person who takes the risk of profit or loss” originated soon after in 1852. Curiously, the use of the expression “entreprenurship” to denote the characteristics of an entrepreneur only dates from a Webster’s entry in 1934.

-1-

small-business entrepreneurs and a control group of non-entrepreneurs.2 Each task involves real economic consequences to the individual, to provide motivation.3 We certainly do not claim to have exhaustively characterized entrepreneurs,4 that the tasks we use are the only ones that could be used for these purposes, or that our tasks apply beyond their domain in terms of economic consequences. Instead, we see our approach as a methodological and substantive start. We also view the experimental data we obtain as complementary to naturally occurring data of the type reviewed by Evans and Leighton [1989]. Our results are striking. We indeed find that full-time entrepreneurs are significantly less risk averse than others, but do not exhibit risk-loving behavior. This evidence supports the conventional wisdom held by many economists, such as Kihlstrom and Laffont [1979], that less risk averse individuals become entrepreneurs while others less able to bear risk avoid entrepreneurial ventures. However, it is not the case that entrepreneurs are characterized by risk-loving tendencies. We also find that full-time entrepreneurs do indeed exhibit a significant joy of winning compared to others in competitive settings. Nonetheless, entrepreneurs do not make systematic judgmental errors when evaluating the statistical value of an object. Finally, we find that full-time entrepreneurs are no more likely to enter a market in which performance depends on perceived skill in relation to others, but that “wanna be entrepreneurs” that hold on to salaried employment are significantly less likely to

2 Our experiments are “artefactual field experiments” in the terminology of Harrison and List [2004]: we take laboratory experiments out into the field, to study a field population of interest. Our experiments also differ from the artefactual field experiments of Harrison, Lau and Williams [2002], who attempt to characterize a national population. 3 An obvious extension of our work would be to compare our findings to those that have used hypothetical surveys to identify one or more of the characteristics we consider, or who used psychological constructs to proxy the formal concept we try to measure. For example, Brockhaus [1980] employed standard hypothetical survey questions from psychology to elicit attitudes towards risk of entrepreneurs. 4 The interdisciplinary entrepreneurship literature is replete with characterizations and definitions of entrepreneurial traits drawn from psychology, sociology, management, finance and economics. An excellent overview of the field is provided by Acs and Audretsch [2003]. In a book-length survey Davidsson [2004] concludes that “there is no shortage of suggestions as to what the phenomenon entrepreneurship really consists of.” He cites no less than 22 possible characteristics of the entrepreneur from the literature including risk taking, optimism, firm size, business cycle state, firm age, manager’s (age, sex, shoe size…) business education, taxes, education level, locus-of-control, regional characteristics, industry experience, location networking, role models, industry structure, labor market legislation, industry growth rate, and management experience. He also notes that none of the existing constructions of language used to define these traits seems to have achieved dominance over others. He sensibly concludes that “entrepreneurship research should deductively test theory from psychology, sociology, and economics and various other business disciplines” in order to gain structure and insight into the entrepreneurial process. The economics tradition stems primarily from Knight [1921], Schumpeter [1934] and Hayek [1945].

-2-

choose to enter such a market. In section 1 we describe the experimental tasks we developed to operationalize and examine those traits, in section 3 we document the field experiments conducted, and in section 4 we present the results.

1. Experimental Tasks We use three experimental tasks which are established in the literature, and have been applied to study the behavior of the traditional subject pool of college students. One is a risk aversion task developed by Holt and Laury [2002], another is a bidding task developed by Holt and Sherman [1994], and the last is a market entry task developed by Camerer and Lovallo [1999]. We discuss each task, explain the hypotheses that each tests, and discuss our field adaptation.

A. Risk Aversion Holt and Laury [2002] (HL) introduced a simple experimental measure for risk aversion. Each subject is presented with a choice between two lotteries, which we call A or B. Table 1 illustrates the basic payoff matrix presented to our subjects. The first row shows that lottery A offered a 10% chance of receiving $20 and a 90% chance of receiving $16. The expected value of this lottery, EVA, is shown in the third-last column as $16.40, although the EV columns were not presented to subjects. Similarly, lottery B in the first row has chances of payoffs of $38.50 and $1.00, for an expected value of $4.80. Thus the two lotteries have a relatively large difference in expected values, in this case $11.60. As one proceeds down the matrix, the expected value of both lotteries increases and the expected value of lottery B eventually exceeds the expected value of lottery A. The subject chooses A or B in each row, and one row is later selected at random for payout for that subject. The logic behind this test for risk aversion is that only risk-loving subjects would take lottery B in the first row, and only risk-averse subjects would take lottery A in the second last row. A risk neutral subject should switch from choosing A to B when the EV of each is about the -3-

same, so a risk-neutral subject would choose A for the first four rows and B thereafter. Responses to this task may be analyzed using a constant relative risk aversion (CRRA) characterization of utility, employing an interval regression model. The CRRA utility of each lottery prize y is defined as U(y) = (y1- r )/(1- r), where r is the CRRA coefficient.5 The dependent variable in the interval regression model is the CRRA interval that subjects implicitly choose when they switch from lottery A to lottery B. For each row of Table 1 one can calculate the implied bounds on the CRRA coefficient, and these intervals are shown in the final column of Table 1. Thus, for example, a subject that made 5 safe choices and then switched to the risky alternatives would have revealed a CRRA interval between 0.14 and 0.41, a subject that made 7 safe choices would have revealed a CRRA interval between 0.68 and 0.97, and so on.

B. Judgmental Error and the Joy of Winning Holt and Sherman [1994] (HS) introduced a class of bidding games that elegantly allow one to identify both the “joy of winning” and evidence of certain types of judgmental errors. Their design had its origins in an effort to identify cleanly the source of a behavioral phenomenon known as the Winner’s Curse (WC). This Curse refers to the belief that some people lose money when bidding in common value auctions. In particular, the winner in a first-price auction is the person that bid the highest, and that winner often ends up regretting the victory when it means that they over-estimated the value of the object. The WC appears to be a robust phenomenon in controlled laboratory settings unless the subjects are very experienced in tasks of this kind (e.g., Kagel and Levin [2002]). HS point out that there are two confounded explanations of this outcome in the standard bidding experiments: a judgmental error and the joy of winning. The previous empirical literature on the WC had focused exclusively on judgmental errors, implicitly assuming that there was no joy of winning.

5

With this parameterization, r = 0 denotes risk neutral behavior, r > 0 denotes risk aversion, and r < 0 denotes risk loving. When r =1, U(y) = ln(y).

-4-

Judgmental error arises from bidders failing to recognize that they should “shave” their estimate of the value of the object conditional on their bid winning, since that conditional event is informative when it happens. Specifically, it informs them that they had the highest bid, and if their information set was identical to others and everyone adopted a symmetric bidding function, that their estimate was the highest amongst all the bidders. Hence they have some probability of having over-estimated the object if they are the winner. The amount by which they should reduce their estimate, and hence their bid, is complicated to calculate, but the qualitative logic is not. The observed behavior may also be consistent with some subjects simply having a joy from winning the auction.6 Since the winners are the ones most likely to lose, if subjects have a greater utility from winning conditional on the profit they make then somebody with a joy of winning might bid higher just to increase the chance of winning. This holds even if they make the correct estimate of the value of the object conditional on them winning, and exhibit no judgmental error. To tease these two explanations apart HS devise a bidding game in which avoidance of the judgmental error involves the bidder increasing their estimate of the object and their bid, in contrast to the standard design in which it involves the bidder decreasing their estimate of the object and their bid. If subjects in the traditional design were led to the WC because of the joy of winning, and not the existence of judgmental errors, their design would reveal this. Instead, they find that subjects in both settings bid in a manner that points squarely to the judgmental error explanation, and not the joy of winning explanation. Thus the confound in the traditional design appears not to play any role for college students in conventional laboratory experiments. Their bidding games involve three cases: •

The Winner’s Curse (WC) treatment in which the rational bidder that does not suffer from a judgmental error will bid lower than a naive bidder that does suffer from the judgmental error.

6

Cox, Smith and Walker [1988; p.91ff.] developed and tested a model of the possible joy of winning in firstprice sealed bid auctions with independent private values.

-5-

•

The No Curse (NC) treatment in which the rational and naive bidder will bid the same amount as the naive bidder, so that the standard judgmental errors7 play no role in bidding.

•

The Loser’s Curse (LC) treatment in which the rational bidder that does not suffer from a judgmental error will bid higher than a naive bidder that does suffer from the judgmental error.

The NC treatment serves to isolate the joy of winning, along with any general confusions surrounding bidding in games such as these. If one assumes that general confusions affect bidding behavior identically across these three treatments, the NC treatment can be used to measure the joy of winning whenever observed bids exceed the predicted bid. The LC treatment can be used to identify the judgmental error, since it is not confounded with any joy of winning as the WC treatment is. If one has some independent estimate of the extent of the joy of winning, such as from behavior in the NC treatment, one can infer the extent of the judgmental error from behavior in the WC and LC treatments. These bidding games are implemented in our experiments using these instructions, which were given to each subject on a single sheet of paper: In this task you will have the opportunity to bid for the chance of getting an additional sum of money. Your bid can be any number in even quarters as long as it is not so large that it will lead to a loss for sure. We will tell you what that maximum is. You will draw a card from this deck to determine a value. If the value is less than your bid you will get the extra money. The money you get will be equal to 1.5 times the value, but we will then subtract your bid from that. You will be given three opportunities to bid, but we will only select one for payment at the end. This will be done randomly. Since it is possible to make a loss, we will give you a sum of money up front in addition to your earnings. This will be an amount between $5 and $15 and you will draw a card to determine that now. Any loss will be subtracted out of this sum, but you will be paid the remainder. If you do not make a loss you will be paid both your earnings and this sum of money. The three opportunities differ in the range of values that can be drawn from the deck of cards: • • •

The first time the value will be drawn between $3.50 and $10.50. This will be done using the deck of cards. The second time the value will be drawn between $4 and $15. The third time the value will be drawn between $3 and $6.

7 Of course, there are a myriad of judgmental errors that are possible. In context, we are referring to the one that motivated the WC literature: failure to correct for the statistical implications of the informative event of winning the auction.

-6-

Additional amount of money: ____________________ Please write down your three bids here: Bid1: Bid 2: Bid 3:

_____________ (maximum bid is $15.75) _____________ (maximum bid is $22.50) _____________ (maximum bid is $9)

We will fill in the remainder after you have drawn the cards. First we will go on to the other short task. Value 1: Value 2: Value 3:

Bid1 > Value1? Bid2 > Value2? Bid3 > Value3?

[1.5 × Value1] - Bid1 = [1.5 × Value2] - Bid2 = [1.5 × Value3] - Bid3 =

In these instructions the first task is the NC treatment, the second task is the WC treatment, and the third task is the LC treatment. There were actually several variants of these instructions in which the order of the three bidding treatments was randomly ordered, and in which the order of the risk aversion task in relation to the bidding task was varied. These procedural specifics are discussed in the next section. Following the derivations for risk-neutral bidders by HS (p. 645), the predicted bids in the NC treatment are $7.00 whether or not the subject exhibits the judgmental error. Predicted bids in the WC treatment are $8.00 if the subject bids rationally, and $9.12 if the subject bids naively in the sense of mis-judging the value of the object. Hence we see that the naive bidding prediction in the WC treatment exceeds the rational bidding prediction, and can be confounded by the joy of winning. Finally, predicted bids in the LC treatment are $6.00 if the subject bids rationally, and $4.87 if the subject bids naively in the sense of mis-judging the value of the object. So we see the point of the HS design of the LC treatment: bidders suffering from the judgmental error will bid lower than those that do not suffer from it. Assuming a CRRA specification for utility, Figure 1 displays predicted bids for the general case in which bidders may be risk-averse or risk-loving. The same qualitative predictions hold as for the risk-neutral case, providing bidders are not so risk averse or risk loving that they bid at the natural boundary points (no more than the maximum possible value, or below the minimum

-7-

possible value). In particular, the naive and rational bids for the NC treatment are identical, which provides an additional reason for using this treatment as the basis for isolating the individual’s joy of winning. In the WC and LC case we would expect to see bids exceed the risk-neutral prediction whenever subjects are risk averse, even if the critical qualitative properties designed by HS (the ordinal ranking of the rational and naive bids) are preserved.

C. Overconfidence and Entry Camerer and Lovallo [1999] (CL) consider a class of entry games in which the subjects have to decide whether to enter a market or not. Their payoffs, if they decide to enter, depend on how many others enter the market and their relative skill. They motivate this experimental design by a discussion of hypotheses advanced by March and Shapira [1987] to explain why most new businesses fail within a few years: the “hubris hypothesis,” to borrow the expression coined by Roll [1986]. And they see these games as well-suited to study the behavior of potential entrepreneurs, justifying their use of graduate and undergraduate students from business schools as follows: “Business students, especially MBA’s, are an appropriate sample because many go on to start businesses or participate in corporate entry decisions (e.g., entrepreneurship is the fifth most popular major among Wharton MBAs).” (p.310). We extend this line of research into field settings with actual entrepreneurs. The key insight in their design is to allow relative profitability in the market to be determined by a skill task involving a series of quiz questions.8 Thus subjects that had a higher skill ranking would earn more money if they entered, and subjects that had a lower skill ranking would earn less money and might even lose an initial stake provided by the experimenter. Subjects that chose not to enter would be allowed to keep the initial stake. No subject knew their score on the skill test prior to entry, so entry was determined in part by the subject’s belief about their skill level in relation to the others that might enter. Of course, entry is also affected by risk attitudes, since entry is risky and 8

They review a longer literature studying this basic entry game in various settings without the skill treatment.

-8-

non-entry generates a non-stochastic payoff (the initial stake). CL used a within-subjects control treatment, in which rankings were determined by a random device instead of by their skill ranking, as a control on risk attitudes. By comparing differences between the random and skill treatments for the same subject, they could discern the pure effect of the skill treatment.9 They found “excess entry” with student subjects when profitability was determined by perceived skill, and attribute this to overconfidence. The left panel of Figure 2 displays the essential results, in terms of average realized profit per entrant in the two main treatments.10 The initial stake for each subject was $10, so the horizontal line at $10 shows the certain opportunity cost of choosing to enter. Excess entry occurs whenever a subject’s risk-adjusted return from entering is lower than $10. If we assume for the moment that subjects were risk-neutral or risk averse, any average profit less than $10 is inefficient. And we see that apart from the first few rounds in the random rank treatment in which expected skill was irrelevant, average profits were below $10, signaling excess entry. The differential between the two treatments further shows that overconfidence in skill generates even more entry, particularly for early rounds. These results are pooled over different capacity levels in their experiments, but the qualitative result of excess entry is generally the same across those levels.11 One natural hypothesis is that entrepreneurs exhibit more overconfidence than others, and will therefore be expected to enter industries more quickly than others. The implied outcome, although not one that is directly measured, is that there will be relatively more subsequent failures and decisions to exit the market. In this setting such an outcome would be observed if we could run 9 One might question this inference, since the subjective lottery that the subject faced in the two treatments is not identical because the probabilities of each final outcome varied. But these differences are likely to be of second order since the final outcomes were identical in the two treatments. A similar design is developed by Hoelzl and Rustichini [2005] to identify overconfidence in skill abilities: subjects vote on being rewarded either by a 50/50 lottery, or being rewarded with the same monetary prize by scoring in the top 50% of a skill test. If a subject votes for the latter option, they exhibit overconfidence in their skill ability. 10 We only examine the random sample selection treatment (experiments 1-4), and the first 12 periods. The skill sample selection treatment raises additional issues of control, and behavior in periods 13-24 must be evaluated conditional on the history that preceded it. The balanced in-sample design of CL allows these controlled comparisons to be made at the level of individual responses, but not at the level of aggregate market responses. Furthermore, behavior in these initial periods amply illustrates their main qualitative conclusion. 11 The literature generally shows excess entry for games with low levels of capacity and under-entry for games with higher levels of capacity, as reviewed by Camerer [2003; §7.3].

-9-

this market for a number of periods, and observe if some early entrants decide later to stay out of the market. Of course, before one can say that excess entry has occurred one has to adjust for risk, hence we pair a market entry game with a risk aversion elicitation task for the same subject. We also elicit information to test for alternative explanations for entry. Some subjects might think of themselves as invincible in skill quizzes of this kind, resulting in what might be called a “Ken effect” after the über-champion of the popular U.S. TV show Jeopardy! who won in 74 straight episodes and amassed winnings of over $2,520,700. Of course, others might have the opposite “Barbie effect,” in which they think of themselves as dumb blondes with no chance of being the most skilled of the entrants. By ascertaining how well they see themselves ranking in skill tasks of this kind we can proxy for subjective beliefs of perceived skill at answering quiz questions. An additional treatment employed by CL was the use of deliberate sample recruitment procedures in which subjects were told that their earnings would depend in part on how good they were at sports trivia. Subjects were reminded of this when they turned up, and told that all others in the experiment that day had been similarly recruited. The results are shown on the right panel of Figure 2. The clear effect from this treatment was to make the average profits from the game in which skill was irrelevant even lower than when subjects were randomly recruited. Of course, this treatment might also have encouraged differences in sample composition which could account for the difference: if there is a gender bias in perceived skill at answering sports trivia, such that more males were recruited when this was announced beforehand, then this could just be a gender effect rather than an effect from a failure to account for the sample being skewed towards those that perceive themselves to be skillful. Despite these concerns, this treatment is pregnant with implications for the study of entrepreneurship: a natural question to ask, when starting a new business, is whether the people that enter at the same time are just as smart at this niche as I believe that I am. Moore and Cain [2003] argue that one should think in more general terms of biases in -10-

comparative judgement that are contextual, rather than simply think in terms of overconfidence or underconfidence as generic traits of individuals. The reason is that there is evidence from a range of tasks in psychology that the same person can be overconfident on simple tasks and underconfident on difficult tasks. Of course, what is simple for one person may be difficult for another person, and vice versa. But assume that this difference in perceived difficulty can be make operational. They argue that it could rationalize why one might have excess entry in some industries where profitability is viewed as a relatively simple goal (“you just have to work hard”) and yet also have insufficient entry where profitability is viewed as a relatively difficult goal (“many people have tried to make money here before, and all have failed”). Moore and Cain [2003] essentially replicate the design of CL, who generally used simple trivia questions.12 They then varied the difficulty of the questions, identifying one set as relatively difficult and providing some sharply contrasting examples. They observe slightly higher rates of entry in the simple skill treatment compared to the difficult skill treatment. In our design we operationalize the difficulty of the skill questions by using general knowledge questions from a large test-bank. In the “difficult” treatment the questions have to be answered in an open-ended way, with no prompts. In the “simple” treatment the same questions are multiple-choice, so the subject sees the correct answer along with several incorrect answers. There is considerable evidence to support the intuition that the multiple-choice questions are easier (Bridgeman [1992], Kennedy and Walstad [1997] and Snow [1993]). We randomize the order and sub-set of questions from the larger test-bank. Market entry games are implemented in our experiments using these instructions, which were given to each subject: In this task you will have the opportunity to earn more money, by deciding if you want to enter a competitive market. We will determine your earnings from this task at 5.15pm today, here at our booth. You can come at any time from 5.15 pm until 6 pm and collect the remainder of your earnings. If you prefer, we can also mail them to you. You will be given a stake of $10 at the outset of this task. This task involves a decision as to whether to enter

12

In two sessions CL (p.308) used quizzes consisting of 10 “logic puzzles,” and in the remaining six sessions they used quizzes consisting of 10 “trivia questions about sports or current events.”

-11-

or not enter a market. 4 other people will be invited to enter or not enter this same market, but you will not know in advance how many of them have accepted. We are going to match people into markets randomly at the end of the day. The capacity of each market is 1 - only one person can make a profit on each market. If you decide to enter, your success will be under your control and will depend on your skill in answering some questions. We will rank all entrants according to how well they answer these questions. The highest ranked entrant will receive $35. Nobody else in the same market will earn anything. In order to enter you have to give up the $10 stake. This is the fee for entering the market. If you do not enter the market, you keep your $10 stake. The ranking system. The way that entrants will be ranked in this market is on the basis of a 7-item quiz of general knowledge. The questions will cover topics such as movie trivia, world history, geography, science, pets & animals, and “the world around us.” Those with higher scores will be ranked higher. If there is a tie for the top rank we will flip a coin to choose one person. You will take the quiz after you make your decision to enter the market. The questions will either be open-ended or multiple-choice. The open-ended questions will require you to write down the correct answer. The multiple-choice questions will require that you pick the correct answer from three alternatives. Many people believe that multiple-choice questions are much easier than open-ended questions. You will be given multiple-choice questions, and you will be competing against people who also have multiple-choice questions. To summarize: If you decide not to enter, you will keep the $10 stake. You will earn nothing beyond that from this task. Therefore, to guarantee that you will not lose, simply do not enter the market. If you decide to enter, you will receive $35 instead of the $10 stake if you are the highest-ranked out of those who enter. You will receive nothing from this task if you are ranked lower. At the most 5 people can enter each market. Of course, whatever you decide here, and whatever the earnings in this task, you will also be paid $10 for completing the task, plus the earnings from the first task as soon as you are done. You will have to return between 5.15 pm and 6 pm to claim the earnings from this task, or they will be mailed to you. DO YOU CHOOSE TO ENTER THE MARKET AND COMPETE OR DO YOU CHOOSE TO STAY OUT?

G

Enter the market and give up the $10 stake

G

Stay out of the market and keep the $10 stake

The text that tells the subject that they have been given multiple-choice questions was replaced with the obvious variant for those subjects given the open-ended questions. There were no other changes in the instructions between these treatments. In order to better identify the contribution of skill overconfidence to excess entry, we followed CL by asking some additional questions to gauge other possible reasons for entry: We have three final questions for you to answer. 1. Here is one more opportunity to earn money. Please answer this question: Across all of today’s markets that have multiple-choice questions, how many people do you think will enter on average? You will receive $10 if you estimate the number exactly. You will receive $8 if you are off by 1, $6 if you are off by 2, and so on. Please round off to an integer value. PICK ONE:

0

1

2

3

4

5

2. We would like to get your estimate of how confident you are about your ability to answer these skill questions. We are not paying you for your answer to this question, but would appreciate you thinking about it carefully. Specifically: If we compared your quiz answers with those of 100 other people picked at random from this conference, what rank do you think you would have? A rank of 1 means that you answered the questions better than anyone else, a rank of 50 that you think you answer better than half but not as well as the other half, and a rank of 100 means that you think everyone else does better than you.

-12-

ANSWER: ____________ 3. Finally, How did your hear about our booth? 01 02 03

Walk-by Heard from somebody that participated Other ______________________________________________

The first question elicits beliefs about the expected number of entrants, with a simple reward for accuracy. We did not want to tie the answer to entry in the specific game that the subject was playing, to avoid obvious endogeneity issues. The second question assesses the subject’s perceptions of their relative ability at answering skill questions, where relative ability is deliberately measured in comparison to the people that they would be competing with. The third question tries to identify if the subject could have known anything substantive about the tasks from prior contacts. We also kept track of the order in which subjects answered these questions, to further identify subjects that might have self-selected into this task.

2. Field Experiments A. Location In April 2004 we conducted field experiments in Atlanta, Georgia at a convention designed to attract aspiring small business entrepreneurs, and in March 2005 we conducted field experiments in Omaha, Nebraska at a comparable convention. In each case we retained a booth at the Small Business Tech Expo (SBTE), formally attached to the Small Business Innovation Research (SBIR) National Convention. These conventions attract potential entrepreneurs, as illustrated by the public invitation to the 2004 conference: The National SBIR/STTR Conference And Small Business Tech Expo (SBTE) Technology Connections: Funding, Strategies, and Partnerships April 26–29, 2004 Hilton Atlanta Hotel Atlanta, GA This year, the Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) programs will provide $2 billion to small businesses through federal programs to help entrepreneurs take their ideas from conception to reality. This conference gives you the tools you need to obtain part of the $2 billion available to small business innovators. CONFERENCE TOPICS INCLUDE:

-13-

! ! ! ! ! ! ! ! ! ! ! ! ! ! !

SBIR/STTR 101: An Overview of the SBIR and STTR Programs Federal Agency Overviews The Do’s and Dont’s of Proposal Writing Writing a Cost Proposal Identifying Your Market Opportunities Increasing Your Chances with SBIR/STTR Partners (STTR): University & Federal Labs Partners: Utilizing Incubators Leveraging & Protecting Your Intellectual Property Managing SBIR/STTR Projects—The Basics What the Agencies Look For & How They Do It Phase III Government Contracts—What’s Relevant Now? Exploring Alternative Financing Programs Accessing the Other 97% Federal R&D Funding Corporate Alliances—Overview

NETWORKING OPPORTUNITIES This conference provides each participant with multiple opportunities to meet and network with SBIR and STTR Program Managers, and fellow attendees, including SBIR/STTR award winners, speakers, and experts from businesses and the government willing to work with you to move your business ahead. Extra Bonus: Conference participation guarantees entry into the co-located Small Business Tech Expo (SBTE) sponsored in part by NASA. The expo will showcase new technologies, support services available to small business and technology development, and commercialization opportunities. For more details and exhibiting opportunities visit: www.barayevents.com/sbte FEDERAL AGENCIES PARTICIPATING IN THE SBIR/STTR PROGRAMS Department of Agriculture, Department of Commerce, Department of Defense, Department of Education, Department of Energy, Department of Health and Human Services, Department of Homeland Security, Department of Transportation, Environmental Protection Agency, National Aeronautics and Space Administration, National Science Foundation If you think your small business could benefit from more than $2 billion in government grants and contracts, this is the opportunity you can’t afford to miss.

The general introduction to SBIR on the SBIR web page13 further identifies it as being targeted to entrepreneurs: SBIR is a highly competitive program that encourages small business to explore their technological potential and provides the incentive to profit from its commercialization. By including qualified small businesses in the nation's R&D arena, high-tech innovation is stimulated and the United States gains entrepreneurial spirit as it meets its specific research and development needs. Competitive Opportunity for Small Business: SBIR targets the entrepreneurial sector because that is where most innovation and innovators thrive. However, the risk and expense of conducting serious R&D efforts are often beyond the means of many small businesses. By reserving a specific percentage of federal R&D funds for small business, SBIR protects the small business and enables it to compete on the same level as larger businesses. SBIR funds the critical startup and development stages and it encourages the commercialization of the technology, product, or service, which, in turn, stimulates the U.S. economy. 13

Http://www.sbirworld.com.

-14-

Since its enactment in 1982, as part of the Small Business Innovation Development Act, SBIR has helped thousands of small businesses to compete for federal research and development awards. Their contributions have enhanced the nation's defense, protected our environment, advanced health care, and improved our ability to manage information and manipulate data.

B. Procedures Our field experiment was conducted from a booth set up in the exhibitors area of the national conference with a banner which read “RESEARCH STUDY: CASH FOR PARTICIPATION.” The booth was run by two faculty members (Elston and Rutström) and one male graduate student, all wearing matching university polo shirts and slacks in school colors. We started each cohort by telling people that this was a study on economic decision making and if they were interested in participating it would take about 15 minutes for which they would receive $10 or possibly more. If they were interested we told them that we needed them to first read and sign a release form for the study. We then proceeded to explain the general information on the study provided on the cover page of the participant packets, while they read through the sheet themselves. A copy of the complete set of instructions is in the Appendix. Each packet was housed in a folder so that participants could work privately given the tight spacing around the booth. The subject’s first task was to complete a survey of firm and individual characteristics before advancing to the decisions tasks 1 and 2, which were handled on a one to one basis. In the 2004 experiments the first decision task was either the risk aversion instrument or the bidding instrument: the order was randomized. In turn, the order in which the lotteries were presented to subjects in the risk aversion task was randomized: for the X treatment they were in the order shown in Table 1, and for the Y treatments they were in the reverse vertical order so that the bottom row in Table 1 was the first row, and so on. Similarly, the order in which the WC, NC and LC tasks were presented to the subject was randomized within the overall bidding task. Each bidding task was presented on the same page, so the randomization of order only refers to the physical listing of the three tasks.14 14 The sheets in the Appendix indicate the order, using a simple key. The label XRW means that the X version of the risk aversion instrument was presented, and that the risk aversion instrument was presented before the bidding instrument (the “W” for Winner’s Curse). The XWR label means that the bidding instrument was presented before the

-15-

Observation suggests that most subjects in fact worked through the bidding tasks in the order presented, but this is anecdotal. In the event that the first decision task was the risk instrument, we proceeded to show them the 10-sided dice, and had them complete this task explicitly leaving out the final steps of throwing the dice to determine their final payout until their second decision task was completed. With the bidding task, and the final dice rolls determining their payout completed, we calculated and assembled their total cash payout while they decided whether to fill out the optional request for study results. In the 2005 experiments we followed the same basic procedures over two full days. The market entry task replaced the bidding task. Since the market entry task required that we group subjects, we “closed” each market at the end of the day (5.45pm), when we knew that all conference activities were over. Subjects were told that they could come back then to collect their earnings from that task, or we would mail them out.15 To ensure credibility we completed the risk aversion task and paid every subject in cash for their participation fee and earnings in that task.16 To facilitate the efficient conduct of the paperwork in the market entry game, we also had an undergraduate research assistant grading each quiz after it was completed and preparing the associated paperwork and payment records.

3. Results Data were collected from 182 individuals. Responses to the questionnaires allowed us to stratify the sample into Full-Time Entrepreneurs, Part-Time Entrepreneurs, and Salaried NonEntrepreneurs. We classify 42 (23%) as being Full-Time Entrepreneurs, in the sense that they state

bidding instrument. Similarly for the YRW and YWR labels, using the reverse vertical order of the risk aversion decision table. In the bidding instructions, label WLR indicates that the WC treatment came before the LC treatment and that these bidding tasks preceded the risk aversion task. So WLR would have been paired with either of the XWR or YWR instructions. 15 The use of delayed closings is standard practice in field experiments of this kind that involve groupings in auctions or markets (e.g., List and Lucking-Reilly [2000; p. 965]). 16 We again randomized the order of the risk aversion task and the other task, since our analysis of the Atlanta results indicated that this task order did have a statistically significant effect on elicited risk attitudes (discussed below). Our analysis of the Atlanta results showed no effect from the row orderings of a given risk aversion task, so in the interests of simplicity of execution we dropped this treatment from the Nebraska experiments.

-16-

this as their sole business focus. We classify 38 (21%) as being Part-Time Entrepreneurs, since they have one foot in the self-employed and salaried occupational worlds. Thus we have information on 80 individuals that report some entrepreneurial experience. In each case we required that they provided some information on the nature of their entrepreneurial firm before we classified them as an entrepreneur. We classify 92 (51%) as being Salaried Non-entrepreneurs, and 10 (5%) as being none of these categories. We therefore consider the first two groups to be our treatment groups of entrepreneurs, and the others to be our controls. Unless we say otherwise, the expression “entrepreneurs” will refer to Full-Time and Part-Time Entrepreneurs as defined above, and the comparison group with be all others. The characteristics of the two groups differ slightly. The entrepreneurs in our sample are roughly the same age (44 versus 42), the same fraction are female (34% versus 37%), and are just as likely to have some higher education (49% versus 51%). But they are more likely to Asian (13% versus 5%) and have a higher income (48% versus 42%). In short, in terms of major demographic characteristics this sample of entrepreneurs is relatively similar to the sample of non-entrepreneurs. We do have two noteworthy differences between the two types of entrepreneurs: the full-time entrepreneurs have more females (44%) and blacks (14%) compared to the part-time entrepreneurs (23% and 3%, respectively), and these differences in means are significant at the 6% level.

A. Measures of Risk Attitudes Figure 3 displays the raw responses elicited from our subjects. The vertical axis shows the faction of choices of the safe lottery, option A, and the horizontal axis shows the problem sequence corresponding to the rows of Table 1. Thus problem sequence 1 is the choice in row 1 of Table 1, problem sequence 2 is the choice in row 2 of Table 1, and so on. Risk neutral subjects would exhibit the pattern shown by a solid line: picking option A for the first four rows in Table 1, and then switching sharply to option B. The observed data, broken down by type of subject, is much smoother than the risk neutral prediction. This reflects the fact that it is averaged over 182 subjects, as well as -17-

the fact that some (few) subjects switched back and forth between options A and B.17 The observed choices indicate that some subjects exhibited risk loving behavior, since they switched to option B before the risk neutral benchmark would predict. On the other hand, somewhat more subjects exhibited risk aversion, since they switched to option B after the risk neutral benchmark would predict. Thus we see some evidence of risk loving behavior, but a general tendency for risk aversion. Each of the observed choices in Figure 3 represents the average behavior of a different type of subject. There is some tendency for the FT Entrepreneurs to be less risk averse than the PT Entrepreneurs or Others, since the line showing their observed choices generally lies below the other lines. Similarly, there is some indication that the PT Entrepreneurs are more risk averse than the nonentrepreneurs, particularly for the last three problems measuring extreme risk aversion. Thus the raw responses, which are not conditioned on observed differences between the samples, suggests that there may be some differences in the risk attitudes of entrepreneurs and others. Although the responses in Figure 3 have the advantage of representing raw responses, they fail to condition on observed differences in the samples of entrepreneurs and others. For example, if sex makes a difference to risk attitudes, then the fact that we have sampled more female FT Entrepreneurs than female PT Entrepreneurs might be masking an effect from entrepreneurial type. Similar differences arise between the entrepreneurs and the control group. In addition, the raw responses in Figure 3 only reflect averages, and one must take into account sampling uncertainty before drawing firm conclusions. Another source of uncertainty is that some responses fall into an open CRRA interval, as indicated in the far right column for the first and last rows in Table 1. These are valid responses, but should be viewed as less precise than responses corresponding to finite upper and lower CRRA intervals. An alternative procedure for measuring risk attitudes from these choices employs an interval 17 One of these subjects was extremely risk-loving, two were extremely risk averse, and seven exhibited erratic responses that defy classification. This fraction of erratic responses is not particularly high, by comparison with some lab samples in the United States.

-18-

regression model, as explained earlier. This statistical model has the advantage of accounting properly for the wider response interval of those subjects that switched back and forth, or who expressed extreme risk-loving or risk aversion. This model can also provide estimates of the marginal effect of the subject being an entrepreneur, controlling for other characteristics of the individual. We can also correct for the possibility that the residual variance of the process generating risk attitudes amongst entrepreneurs may not be the same as for others, implying that one should correct for heteroskedasticity with respect to observable characteristics of the sample. We do so, using maximum-likelihood estimation, and predict the risk attitude of the individual based on this model.18 Table 2 reports the estimates of the interval regression model of risk attitudes. The most important effect is that FT Entrepreneurs have a significantly lower aversion to risk than Part-Time Entrepreneurs and non-entrepreneurs (the default alternative). The difference is estimated to be 0.202, and is statistically significant with a p-value of 0.068. We also observe that PT Entrepreneurs are not significantly different in terms of risk attitudes than non-entrepreneurs. Figure 4 reports kernel densities of these estimated risk attitudes across the sample, stratified by the type of entrepreneur.19 We observe a striking pattern, consistent with the estimates in Table 2. Most subjects are risk averse, but FT Entrepreneurs stand out as being less risk averse than either PT Entrepreneurs or non-entrepreneurs. The contrast between the two types of entrepreneurs is particularly clear. In fact, Figure 5 shows what would have been inferred if we had pooled these two types: the data would suggest that entrepreneurs simply have a wider variance in risk attitudes. Instead, we have been able to identify two distinct types of entrepreneurs, and see that they have different risk attitudes.

18 Nine subjects out of 62 are lost in this estimation exercise since they failed to report their sex or age, or provided responses that were incoherent in the risk aversion task. 19 These densities are kernel density estimates, which may be viewed as generalizations of the histogram as a way of visualizing continuous univariate data. We employ the common Epanechnikov kernel function, and the so-called “optimal bandwidth.” The latter is actually calculated as the bandwidth that would be optimal in a well-defined sense, akin to minimizing mean square error, if the empirical distribution were Gaussian and the Gaussian kernel function had been used. It will be approximately optimal if the estimated density function using other kernel functions are unimodal and symmetric, and will tend to excessively “smooth” densities that violate these conditions. Silverman [1986] provides a rich discussion of density estimation issues, and StataCorp [2005] documents the specific procedures used.

-19-

B. Measures of the Joy of Winning Figure 5 displays histograms of observed bids in the three treatments, along with the riskneutral bid predictions shown as vertical lines. The first reaction from observing these data is that there is too much variation in bids to worry about testing any of the predictions, since the differences in the predictions is tiny in relation to the variation in bids. However, this confuses the fact that the predictions shown here are for riskneutral agents, and we know from our analysis of the responses to the risk attitudes instrument that we have considerable heterogeneity in the sample in terms of risk attitudes. So the relevant prediction for each subject is actually one that is based on “their CRRA” from the risk aversion instrument and the corresponding predictions from Figure 1. Thus somebody bidding in excess of their risk-neutral predictions in Figure 5 could be someone that is risk averse and behaving in accord with the (naive or rational) prediction from Figure 1 for a risk averse individual. Similarly, somebody bidding below their risk-neutral predictions in Figure 5 could be someone that is risk loving and behaving in accord with the (naive or rational) prediction from Figure 1 for a risk loving individual. There are, of course, other explanations, such as confusion or joy of winning that could explain deviations; our point is that one has to take risk attitudes into account before drawing any conclusions about bidding behavior in this setting. To identify the joy of winning we focus on the No Curse bidding treatment, since judgmental errors play no role in explaining different bidding behavior here. The average bid in this treatment is $9.42, the standard deviation is $3.82, and a 95% confidence interval places bids between $8.45 and $10.40. There is a clear mode at $10.20 Without correction for heterogeneous risk attitudes, this average is 35% above the risk-neutral prediction of $7, providing an initial measure of the joy of winning. Corrections for differences in risk attitudes are critical to proper identification of the joy of

20

The bars in these histograms refer to the amount shown at the left of the bar. These data are quite lumpy, with bids often stated in exact dollar amounts.

-20-

winning, particularly if we are interested in differences between entrepreneurs and others. Using the mid-point of the CRRA interval elicited for each subject, we can directly test if observed bids in the NC treatment differ from the predicted bid in a systematic manner. To do so we examine the ratio of the observed bid to the predicted bid for that individual given the individual’s CRRA value.21 Evidence for a joy of winning would come from this ratio being greater than one and statistically significant. Figure 6 displays the kernel density of this ratio for all subjects that participated in the bidding task, as well as a Normal density for reference. The average ratio is 1.23, with a standard deviation of 0.50 for the ratio. Thus the corrections for risk attitudes in the predicted bids reduce the average estimate of the joy of winning from 35% to 23%, but there remains considerable noise in the data. Figure 7 displays the kernel density of the joy of winning ratio for each type of subject. These results suggest that the joy of winning for FT Entrepreneurs is larger than the joy of winning for PT Entrepreneurs and non-entrepreneurs, particularly when we allow for the evidence that PT Entrepreneurs and non-entrepreneurs exhibit some “fear of winning.” But there is no clear separation of the densities. Table 3 reports the estimates from a regression of the joy of winning ratio against the treatment conditions in the experiment and individual characteristics of the respondents. The striking result is the finding that FT Entrepreneurs do exhibit a statistically significant joy of winning, after correcting for all other determinants. Again, we find a stark difference between FT Entrepreneurs and PT Entrepreneurs, with the latter exhibiting no significant joy of winning. Table 3 also shows that there were significant treatment effects on the joy of winning, as might be expected from the manner in which these were presented to subjects on one sheet. When the LC treatment was listed first, and presumably was answered first by the subject, bids in the NC 21

Since we use the directly elicited CRRA intervals here, rather than values predicted by some statistical model, there is no need to allow for the fact the CRRA estimate is an estimate subject to sampling error. One might argue that we do have to estimate the mid-point implicitly using an assumption of a uniform distribution of true values within the interval. But this simply adds noise to the explanatory variable that is exogenous, and not correlated with the risk aversion choices of the subject. Our qualitative results are considerably stronger if one uses the lower bound CRRA instead, and neither sub-sample exhibit a statistically significant joy of winning if one uses the upper bound CRRA instead. These differences point again to the need to obtain estimates of the risk attitudes of the subjects in order to evaluate their bidding behavior.

-21-

treatment were 51 percentage points higher on average.22 In addition, higher random endowments slightly muted the average joy of winning, as well as the residual variance. Age has a significant effect on the joy of winning, with older subjects exhibiting more joy. Every additional year adds about 1.6 percentage points to the average bid from this effect, and this estimate is statistically significant with a p-value of 0.001. The only other demographic to show some effect is whether the subject was Asian. Although the p-value on the estimated effect is only 0.131, the effect is large (-0.293).

C. Measures of Judgmental Error Judgmental error in the Winner’s Curse and Loser’s Curse treatments arise when the subject bids closer to the predicted naive bid than the predicted rational bid. When subjects have heterogeneous risk attitudes or joy from winning, the relevant versions of the predicted bids must take these into account. We specify a likelihood function that jointly estimates CRRA risk attitudes from responses to the risk aversion task, the joy of winning from bids in the NC treatment, and separate judgmental errors from bids in the WC and LC treatment. The likelihood function consists of three components. The first part estimates the CRRA coefficient r using an interval regression specification and the choices in the risk aversion task. This part replicates the specification used in §3.A to study risk attitudes. The second part assumes a multiplicative “joy of winning” model, paralleling the regression analysis of §3.B. The observed bid in the NC treatment is assumed to be some scalar 8 of the predicted bid.23 The third part of the likelihood function estimates a multiplicative judgement error term for the predicted WC and LC

22 This is consistent with the information in the LC treatment having an effect on bids in the NC treatment, and effectively causing an overshooting as suggested by the location of the predictions in Figure 1 for a given risk attitude; or, equivalently, that having the WC treatment come first led to a hysteresis effect on bids, keeping them higher than they would otherwise have been. In future work it will be valuable to undertake between-subject comparisons in which order does not play a role, although larger samples will be needed to maintain statistical power. This is not a flaw in our design, which is constrained by the sample sizes and time available in a field setting such as this, but points to the importance of building in the simple order-effect controls we had. 23 Again, the predicted rational bid is the same as the predicted naive bid in the NC treatment, where “rational” and “naive” are defined in terms of the judgmental error.

-22-

bids. The latter predictions are conditional on the estimated risk attitude of the subject, as well as the joy of winning that the subject has.24 Since we do not view a joy of winning as irrational (it is just an argument of some utility function), we refer to these as the predicted rational bids. The model here is therefore observed bid in LC case = " × 8 × predicted rational bid in LC case and observed bid in WC case = $ × 8 × predicted rational bid in WC case, where the predicted rational bid depends on the estimated CRRA for this subject. Since “rational” here means “no judgment errors,” the estimates of " and $ depend on the estimates of the joy of winning term 8 and the CRRA coefficient r. The likelihood function is defined over observed choices in the risk attitudes task, bids in the NC task, and bids in the LC and WC tasks, and we obtain estimates of ", $, 8 and r jointly. We allow each parameter to be a linear function of the observed individual characteristics of the subject, as well as treatment effects. The estimates of each parameter in the above likelihood function entails estimation of the coefficients of a linear function of these characteristics. So if sex and age were the only two characteristics used, the estimate of r, ^r , would actually be ^r = ^r 0 + (r^FEMALE × FEMALE)+ (r^AGE × AGE), where ^r 0 is the estimate of the constant. If we collapse this specification by dropping all covariates, we would simply be estimating the constant terms for each parameter. The estimates allow for the possibility of correlation between responses by the same subject, so the standard errors on estimates are corrected for the possibility that the 4 responses are clustered for the same subject. The use of clustering to allow for “panel effects” from unobserved individual effects is common in the statistical survey literature.25 24

We are assuming that the same multiplicative joy of winning applies in the NC, WC and LC cases. It is not obvious in our setting why there should be some interaction effect, although one could easily imagine field settings where there might be (e.g., winning a PGA golf tournament versus winning a local club competition, or turning a profit in a small industry versus turning a profit in a large industry). 25 Clustering commonly arises in national field surveys from the fact that physically proximate households are often sampled to save time and money, but it can also arise from more homely sampling procedures. For example,

-23-

The results for the estimates of judgmental errors are displayed in Table 4 for two model specifications. The estimates in Panel A report unconditional estimates, in which there are no covariates at all. The coefficients reported here measure the association of observed bids to the predicted rational bid, corrected for the estimate of CRRA and the joy of winning. Thus we estimate in Panel A of Table 4 that bids were 98% of the predicted rational bid in the LC treatment, and 117% of the predicted rational bid in the WC treatment. Recalling the predictions from Figure 1, these are each in the direction that one would expect if there was some judgmental error: if there is such an error, bids should be below the predicted bid in the LC case and above the predicted bid in the WC case. Thus it would appear that there is no judgmental error of any substance in the LC case, but that there is in the WC case. In fact, tests of the hypothesis that there are no judgmental errors, which is to say that "=1 and that $=1, have p-values of 0.76 and 0.06 in the LC and WC cases, confirming these conclusions. Panel B reports the estimation results of real interest, since they include controls for types of entrepreneurs, experimental treatments, individual characteristics, and multiplicative heteroskedasticity for each of the estimated error terms.26 The first result is that we observe a significant difference between non-entrepreneurs and FT Entrepreneurs with respect to judgmental errors in the LC case: controlling for all other characteristics and treatments, the FT Entrepreneurs have an estimate of " that is 0.283 higher, and this is a significant difference (p-value = 0.031). However, it does not follow that FT Entrepreneurs exhibit a significant deviation from unity. Adding this estimate of the incremental " to the estimate for the constant, we estimate the judgmental error for FT Entrepreneurs to be 1.22 (=1.015+0.283) but with a standard error of 0.39, resulting in a pvalue of only 0.37 on the null hypothesis that "=1, and a 95% confidence interval between 0.45 and Williams [2000; p.645] notes that it could arise from dental studies that “collect data on each tooth surface for each of several teeth from a set of patients” or “repeated measurements or recurrent events observed on the same person.” The procedures for allowing for clustering allow heteroskedasticity between and within clusters, as well as autocorrelation within clusters. They are closely related to the “generalized estimating equations” approach to panel estimation in epidemiology (see Liang and Zeger [1986]), and generalize the “robust standard errors” approach popular in econometrics (see Rogers [1993]). Wooldridge [2003] reviews some issues in the use of clustering for panel effects, in particular noting that significant inferential problems may arise with small numbers of panels. 26 The heteroskedasticity is multiplicative with respect to the type of entrepreneur.

-24-

1.99. Furthermore, in results not reported in Table 4, we find that there is a significant interaction between FT Entrepreneurs and the NCfirst treatment, such that the estimated " is actually 0.80 when FT Entrepreneurs face the NC task first. But a test that this coefficient is different from 1 cannot be rejected with a p-value of 0.53. Thus there is simply too much noise to claim that this point estimate of " $100k Georgia Georgia conference

0.293 -0.202 0.094 -0.014 0.017 0.013 0.002 -0.018 0.225 -0.095 -0.088 0.044 -0.144 0.234

0.237 0.110 0.112 0.022 0.087 0.132 0.005 0.101 0.223 0.192 0.089 0.112 0.109 0.265

0.216 0.068 0.403 0.534 0.843 0.919 0.727 0.859 0.314 0.619 0.325 0.697 0.187 0.377

-0.172 -0.418 -0.126 -0.057 -0.154 -0.246 -0.009 -0.216 -0.213 -0.471 -0.262 -0.175 -0.358 -0.285

0.758 0.015 0.314 0.030 0.188 0.273 0.012 0.180 0.663 0.280 0.087 0.262 0.070 0.754

0.000 0.597 0.922 0.735 0.903 0.508 0.000

-0.911 -0.286 -0.349 -0.043 -0.285 -0.725 1.201

-0.403 0.497 0.386 0.030 0.323 0.359 3.281

B. Multiplicative Residual Variance Effects

Entrepreneur Both Endowment OrderTask OrderRow Erratic

Constant FT Entrepreneur PT Entrepreneur Random endowment Risk elicitation was second task Reverse order of MPL table Some erratic responses

-0.657 0.106 0.018 -0.006 0.019 -0.183 2.241

0.130 0.200 0.188 0.019 0.155 0.276 0.531

Legend: Unless otherwise stated, all variables are binary dummy variables in which 1 denotes true and 0 denotes false. Age is measured in years.

-33-

Figure 4: Kernel Density of Estimated Risk Attitudes Data consists of predicted CRRA from interval regression model Risk Averse 4 Risk Loving

Density

3

2

1

0 -.2

-.1

0

.1

.2

.3 CRRA

FT Entrepreneurs

.4

.5

.6

PT Entrepreneurs

.7

Others

Figure 5: Kernel Density of Estimated Risk Attitudes With Full-Time and Part-Time Entreperenurs Pooled Data consists of predicted CRRA from interval regression model Risk Loving Risk Averse 4

Density

3

2

1

0 -.2

-.1

0

.1

.2

.3 CRRA

Entrepreneurs

-34-

.4

.5 Others

.6

.7

Figure 5: Observed Bids Vertical lines are risk-neutral bid predictions Winner's Curse

.2 .15 .1 .05 0

No Curse

.2 .15 .1 .05 0

Loser's Curse

.2 .15 .1 .05 0

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

Bids in Dollars -35-

15

16

17

18

19

20

21

22

23

24

25

Figure 6: Joy of Winning Bids from the No-Curse Treatment 1

Fear of Winning

Joy of Winning

Density

.8 .6 .4 .2 0 0

.5 1 1.5 2 Ratio of Actual Bid to Risk-Adjusted Predicted Bid Kernel density estimate

2.5

Normal density

Figure 7: Joy of Winning By Subject Type 1

Fear of Winning

Joy of Winning

Density

.8

.6

.4

.2

0 0

.5

1 1.5 2 Ratio of Actual Bid to Risk-Adjusted Prediction

FT Entrepreneurs

PT Entrepreneurs

-36-

2.5 Others

Table 3: Regression Model of Joy of Winning Dependent variable is ratio of actual bid in No Curse treatment to CRRA-adjusted predicted bid. Variable

Description

Estimate

Std. Error

p-value

95% Confidence Interval

A. Average Effects Constant Entrepreneur FT Entrepreneur Both PT Entrepreneur Endowment Random endowment NCfirst No Curse treatment listed first LCbeforeWC LC treatment before WC treatment Age Age Female Female Black Black Asian Asian HigherEd Completed higher college degree Married Current married HighInc Household income > $100k

0.989 0.482 -0.046 -0.072 -0.151 0.508 0.016 0.092 0.007 -0.293 0.054 0.082 0.015

0.331 0.160 0.255 0.023 0.147 0.213 0.005 0.170 0.147 0.194 0.131 0.112 0.152

0.003 0.003 0.856 0.002 0.307 0.017 0.001 0.588 0.963 0.131 0.680 0.468 0.919

0.340 0.169 -0.546 -0.118 -0.439 0.091 0.006 -0.240 -0.282 -0.672 -0.202 -0.139 -0.283

1.637 0.796 0.454 -0.026 0.138 0.925 0.025 0.424 0.296 0.087 0.310 0.302 0.314

0.345 0.418 0.370 0.002 0.998 0.166

-0.696 -0.409 -1.112 -0.216 -1.633 -1.285

1.991 0.984 0.414 -0.046 1.628 0.221

B. Multiplicative Residual Variance Effects

Entrepreneur Both Endowment NCfirst LCbeforeWC

Constant FT Entrepreneur PT Entrepreneur Random endowment No Curse treatment listed first LC treatment before WC treatment

0.648 0.288 -0.349 -0.131 -0.002 -0.532

-37-

0.686 0.355 0.389 0.043 0.832 0.384

Table 4: Estimates of Judgmental Errors Maximum likelihood estimates of model of risk attitudes, joy of winning and judgmental errors, correcting for clustered errors by individuals. Only estimates of the judgmental error parameters listed below. All variables are defined in Tables 2 and 3, and text. Core Parameter

Variable

Estimate

Std. Error

p-value

95% Confidence Interval

A. No Covariates Loser’s Curse (") Winner’s Curse ($)

Constant Constant

0.977 1.166

0.076 0.090

0.000 0.000

0.827 0.990

1.127 1.342

B. Adding Covariates Loser’s Curse (")

FT Entrepreneur PT Entrepreneur Endowment NCfirst LCbeforeWC Age Female Black Asian HigherEd Married HighInc Constant

0.283 -0.023 0.027 -0.876 -0.197 0.009 -0.115 -0.185 0.139 -0.115 -0.123 -0.049 1.015

0.131 0.188 0.021 0.201 0.122 0.007 0.134 0.223 0.248 0.163 0.180 0.160 0.342

0.031 0.904 0.195 0.000 0.108 0.206 0.390 0.408 0.574 0.482 0.494 0.761 0.003

0.026 -0.391 -0.014 -1.269 -0.436 -0.005 -0.378 -0.623 -0.346 -0.434 -0.476 -0.363 0.346

0.540 0.346 0.067 -0.483 0.043 0.023 0.148 0.253 0.625 0.205 0.230 0.265 1.685

Winner’s Curse ($)

FT Entrepreneur PT Entrepreneur Endowment NCfirst LCbeforeWC Age Female Black Asian HigherEd Married HighInc Constant

0.154 -0.105 -0.009 0.532 -0.232 -0.003 0.182 0.223 0.034 -0.019 0.143 -0.032 1.066

0.179 0.207 0.022 0.164 0.142 0.009 0.243 0.291 0.305 0.150 0.217 0.194 0.416

0.390 0.613 0.692 0.001 0.102 0.723 0.454 0.442 0.911 0.900 0.511 0.869 0.010

-0.197 -0.511 -0.051 0.210 -0.509 -0.020 -0.294 -0.346 -0.563 -0.313 -0.283 -0.412 0.251

0.505 0.302 0.034 0.854 0.046 0.014 0.657 0.793 0.631 0.275 0.569 0.348 1.881

-38-

Table 5: Profitability of Market Entry Number of Entrants

Total Industry Profit ($)

Total Non-entrant Profit ($)

Total Earnings ($)

Average Earnings Per Entrant ($)

Average Earnings Per Subject ($)

0 1 2 3 4 5

0 35 35 35 35 35

50 40 30 20 10 0

50 75 65 55 45 35

75 32.5 18.33 11.25 7

10 15 13 11 9 7

Table 6: Hypothesis Tests for Market Entry Decision Marginal effects from logit model, with market entry as the dependent variable. N=118. Variable

Estimate

Std. Error

p-value

95% Confidence Interval

Mean

CRRA Quiz_MC

0.002 0.062

0.054 0.053

0.969 0.242

-0.104 -0.042

0.108 0.165

0.217 0.500

Predict Confident

0.090 0.002

0.028 0.001

0.001 0.014

0.035 0.000

0.144 0.004

3.551 -33.017

Heard Day2

-0.078 0.012

0.062 0.047

0.205 0.794

-0.199 -0.080

0.043 0.105

0.441 0.475

FT Entrepreneur PT Entrepreneur

0.002 -0.351

0.062 0.117

0.975 0.003

-0.119 -0.580

0.123 -0.122

0.212 0.186

OrderTask Age Female Black Asian HigherEd Married HighInc

0.092 -0.001 -0.025 -0.347 0.060 -0.007 0.006 0.052

0.055 0.002 0.046 0.215 0.045 0.048 0.060 0.048

0.095 0.828 0.587 0.107 0.189 0.893 0.924 0.277

-0.016 -0.005 -0.116 -0.768 -0.029 -0.102 -0.111 -0.042

0.201 0.004 0.066 0.075 0.149 0.088 0.122 0.145

0.500 43.737 0.364 0.051 0.076 0.500 0.754 0.407

Definitions: CRRA is the elicited CRRA mid-point, with extreme risk-lovers and risk-averse individuals replaced with values -0.95 and 1.37, respectively; Quiz_MC is a dummy variable denoting the use of the multiple choice quiz questions; Predict is the number of entrants the subject predicted; Confident is a measure of how confident the subject was of their quiz aptitude (it is the negative of the elicited percentile that they believed they were in); Heard is a dummy variable denoting someone that had heard about this task from another person at the conference; Day2 is a dummy variable for subjects responding on the second day of the conference; FT Entrepreneur is a dummy variable for subjects classified as being Full-Time Entrepreneurs; PT Entrepreneur is a dummy variable for subjects classified as being Part-Time Entrepreneurs; OrderTask is a dummy variable measuring subjects who were given the market entry task before the risk attitude task; other variables defined earlier.

-39-

References Acs, Zoltan J., and Audretsch, David B., Handbook of Entrepreneurship Research (Boston: Kluwer, 2003). Bridgeman, Brent, “A Comparison of Quantitative Questions in Open-Ended and Multiple-Choice Formats,” Journal of Educational Measurement, 29, 1992, 253-271. Brockhaus, Robert H., Sr., “Risk Taking Propensity of Entrepreneurs,” Academy of Management Journal, 23(3), 1980, 509-520. Camerer, Colin, Behavioral Game Theory (Princeton, NJ: Princeton University Press, 2003). Camerer, Colin F., and Lovallo, Dan, “Overconfidence and Excess Entry: An Experimental Approach,” American Economic Review, 89(1), March 1999, 306-318. Cox, James C; Smith, Vernon L., and Walker, James M., “Theory and Individual Behavior of FirstPrice Auctions,” Journal of Risk & Uncertainty, 1, 1988, 61-99. Davidsson, Per, Researching Entrepreneurship: International Studies in Entrepreneurship (Boston: Springer Science, 2004). de Meza, David, and Southey, Clive, “The Borrower’s Curse: Optimism, Finance and Entrepreneurship,” Economic Journal, 106, March 1996, 365-386. Evans, George Heberton, Jr., “The Entrepreneur and Economic Theory: A Historical and Analytical Approach,” American Economic Review (Papers & Proceedings), 39(3), May 1949, 336-348. Evans, David S., and Leighton, Linda S., “Some Empirical Aspects of Entrepreneurship,” American Economic Review, 79, June 1989, 519-535. Harrison, Glenn W.; Lau, Morten I., and Williams, Melonie B., “Estimating Individual Discount Rates for Denmark: A Field Experiment,” American Economic Review, 92(5), December 2002, 1606-1617. Harrison, Glenn W., and List, John A., “Field Experiments,” Journal of Economic Literature, 42(4), December 2004, 1013-1059. Hayek, F.A., “The use of knowledge in society,” American Economic Review, 35(4), September 1945, 519-530. Hoelzl, Erik, and Rustichini, Aldo, “Overconfident: Do You Put Your Money On It?” Economic Journal, 115, April 2005, 305-318. Holt, Charles A., and Laury, Susan K., “Risk Aversion and Incentive Effects,” American Economic Review, 92(5), December 2002, 1644-1655. Holt, Charles A., and Sherman, Roger, “The Loser’s Curse,” American Economic Review, 84(3), June 1994, 642-652. Kagel, John H., and Levin, Dan, Common Value Auctions and the Winner’s Curse (Princeton: Princeton University Press, 2002). Kennedy, Peter E., and Walstad, William B., “Combining Multiple-Choice and Constructed-Response -40-

Test Scores: An Economist’s View,” Applied Measurement in Education, 10, 1997, 359-375. Kihlstrom, R. E., and Laffont J. J., “A general equilibrium theory of firm formation based on risk aversion,” Journal of Political Economy, 87, 1979, 719-748. Knight, F.H., Risk, Uncertainty and Profit (New York: Houghton Mifflin,1921). Liang, K-Y., and Zeger, S.L., “Longitudinal Data Analysis Using Generalized Linear Models,” Biometrika, 73, 1986, 13-22. List, John A., and Lucking-Reilly, David, “Demand Reduction in Multiunit Auctions: Evidence from a Sportscard Field Experiment,” American Economic Review, 90(4), September 2000, 961-972. March, James, and Shapira, Zur, “Managerial Perspectives on Risk and Risk Taking,” Management Science, 33(11), November 1987, 1404-1418. Marshall, Alfred, Principles of Economics (London: Macmillan, 8th Edition, 1927). Moore, Don A., and Cain, Daylain M., “Myopic Biases in Comparative Judgement and Entrepreneurial Entry,” Working Paper, Tepper School of Business, Carnegie-Mellon University, 2003. Rogers, W. H., “Regression standard errors in clustered samples,” Stata Technical Bulletin, 13, 1993, 1923. Roll, Richard, “The Hubris Hypothesis of Corporate Takeovers,” Journal of Business, 59(2), April 1986, 197-216. Schumpeter, Joseph A., The Theory of Economic Development (Cambridge, MA: Harvard University Press, 1934). Silverman, Bernard W., Density Estimation for Statistics and Data Analysis (London: Chapman & Hall, 1986). Snow, Richard E., “Construct Validity and Constructed-Response Tests,” in R.E. Bennett and W.C. ward (eds.), Construction Versus Choice in Cognitive Measurement (Hillsdale, NJ: Lawrence Erlbaum, 1993). StataCorp, Stata Statistical Software: Release 9 (College Station, TX: Stata Corporation, 2005). Thaler, Richard H., The Winner’s Curse: Paradoxes and Anomalies of Economic Life (New York: The Free Press, 1991). Williams, Rick L., “A Note on Robust Variance Estimation for Cluster-Correlated Data,” Biometrics, 56, June 2000, 645-646. Wooldridge, Jeffrey, “Cluster-Sample Methods in Applied Econometrics,” American Economic Review (Papers & Proceedings), 93, May 1993, 133-138.

-41-

Appendix A: Instructions for Georgia Sessions (NOT FOR PUBLICATION)

WELCOME TO THE RESEARCH STUDY DESCRIPTION This is a study of economic decision making. We think you will find it interesting, you will be paid $10 for your participation and you could earn additional money. How much you earn will depend partly on chance and partly on the choice you make in decision problems which you will be presented with. The instructions are simple and you will benefit from following them carefully. The problems are not designed to test you. What we want to know is what choices you would make in them. The only right answer is what you really would choose. That is why the problems give you the chance of earning real money. You will be paid in cash as soon as you finish the task. The task will proceed in three short parts. The first part consists of a few questions about your firm and you. This information is for research use only. The published results of our research will not identify any firm or individual, or the choice he or she made in any way. Nor will we give this identifying information to anyone else. The second and third part are short decision problems in which chance may play a part. Each decision-problem requires you to make a choice. This is described in more detail when you have completed the first part. Both of these parts may result in additional earnings over and above the $10 participation fee. We expect the entire task to take less than 15 minutes.

-42-

ID: _______

Part 1: Some Questions About You and Your Firm

In this survey most of the questions asked are descriptive. We will not be grading your answers and your responses are completely confidential. We will not be recording your name or the name of your firm on this sheet. Please think carefully about each question and give your best answers. Questions About Your Firm 1.

How old is your firm, in years? ____________

2.

What type of product or service do you provide? _____________________________

3.

Have you ever experienced a shortage of capital in running your firm? (Circle one number) 01 02 03 04 05

4.

Do you have a shortage of capital now? 01

5.

02

No

Inheritance Gift Credit cards Earnings from another job

05 06 07

SBIR grant Private loan from a bank or person Other

Have you ever applied for or received an SBIR grant? Applied: Received:

8.

Yes

How did you primarily finance your firm’s start up? (Circle all that apply) 01 02 03 04

6.

Never Rarely Occasionally Often Always

01 01

Yes Yes

02 02

No No

How do you finance your firm now? Enter rough percentages for each: 01 02 03 04 05 06 07

Government loans or grants Private loans from banks or people Credit cards Earnings from another job Cash from operations Equity capital Other

__________ __________ __________ __________ __________ __________ __________

9.

What would you estimate to be the annual revenue of your firm? _________________

10.

What would you estimate to be the value of the assets of your firm? _________________

11.

What is the state and ZIP code of the main location of your firm? ____________________

-43-

Questions About You 1.

What is your AGE? ____________ years

2.

What is your sex? (Circle one number.) 01

3.

01 02 03 04 05

06 07 08 09

Hispanic-American Hispanic Mixed Race Other

Your spouse or domestic partner 01 02 03 04 05

Self-employed only Part-time employment in another firm Full-time employment in another firm Actively seeking employment Unemployed

Less than high school GED or High School Equivalency High school

04 05 06

Vocational or trade school Bachelor’s degree at college Higher degree at college

03 04

Separated or divorced? Widowed?

Are you currently... 01 02

7.

White African-American African Asian-American Asian

What is the highest level of education you have completed? (Circle one number) 01 02 03

6.

Female

What is your current employment status, and that of your spouse or domestic partner? (Circle one number for each) You

5.

02

Which of the following categories best describes you? (Circle one number.) 01 02 03 04 05

4.

Male

Single and never married? Married?

How many people live in your household? Include yourself, your spouse and any dependents. Do not include your parents or roommates unless you claim them as dependents. ___________________

8.

Please circle the category below that describes the total amount of INCOME earned in 2003 by the people in your household (as “household” is defined in the previous question). [Consider all forms of income, including salaries, tips, interest and dividend payments, scholarship support, student loans, parental support, social security, alimony, and child support, and others.] 01 02 03 04 05

$25,000 or under $25,001 - $45,000 $45,001 - $65,000 $65,001 - $85,000 $85,001 - $100,000

06 07 08 09 10

-44-

$100,001 - $125,000 $125,001 - $150,000 $150,001 - $175,000 $175,001 - $200,000 over $200,000

XRW

Decision Task

Your decision sheet shows ten decisions listed on the left. Each decision is a paired choice between “Option A” and “Option B.” You will make a choice on each row and record these in the final column. Here is a ten-sided die that will be used to determine payoffs. The faces are numbered from 0 to 9, and we will use the 0 face of the die to serve as 10. Look at Decision 1 at the top. Option A pays $20.00 if the throw of the ten sided die is 1, and it pays $16.00 if the throw is 2-10. Option B yields $38.50 if the throw of the die is 1, and it pays $1.00 if the throw is 2-10. The other Decisions are similar, except that as you move down the table, the chances of the higher payoff for each option increase. In fact, for Decision 10 in the bottom row, the die will not be needed since each option pays the highest payoff for sure, so your choice here is between $20.00 or $38.50. After you have made all of your choices, you will throw this die twice, once to select one of the ten decisions to be used, and a second time to determine what your payoff is for the option you chose, A or B, for the particular decision selected. Even though you will make ten decisions, only one of these will end up affecting your earnings, but you will not know in advance which decision will be used. We will not perform the dice roll until after you have completed the next short task.

-45-

XWR

Decision Task

Your decision sheet shows ten decisions listed on the left. Each decision is a paired choice between “Option A” and “Option B.” You will make a choice on each row and record these in the final column. Here is a ten-sided die that will be used to determine payoffs. The faces are numbered from 0 to 9, and we will use the 0 face of the die to serve as 10. Look at Decision 1 at the top. Option A pays $20.00 if the throw of the ten sided die is 1, and it pays $16.00 if the throw is 2-10. Option B yields $38.50 if the throw of the die is 1, and it pays $1.00 if the throw is 2-10. The other Decisions are similar, except that as you move down the table, the chances of the higher payoff for each option increase. In fact, for Decision 10 in the bottom row, the die will not be needed since each option pays the highest payoff for sure, so your choice here is between $20.00 or $38.50. After you have made all of your choices, you will throw this die twice, once to select one of the ten decisions to be used, and a second time to determine what your payoff is for the option you chose, A or B, for the particular decision selected. Even though you will make ten decisions, only one of these will end up affecting your earnings, but you will not know in advance which decision will be used.

-46-

ID: X______________

Decision

Option A

Option B

Your Choice (Circle A or B)

1

$20.00 if throw of die is 1 $16.00 if throw of die is 2-10

$38.50 if throw of die is 1 $1.00 if throw of die is 2-10

A

B

2

$20.00 if throw of die is 1-2 $16.00 if throw of die is 3-10

$38.50 if throw of die is 1-2 $1.00 if throw of die is 3-10

A

B

3

$20.00 if throw of die is 1-3 $16.00 if throw of die is 4-10

$38.50 if throw of die is 1-3 $1.00 if throw of die is 4-10

A

B

4

$20.00 if throw of die is 1-4 $16.00 if throw of die is 5-10

$38.50 if throw of die is 1-4 $1.00 if throw of die is 5-10

A

B

5

$20.00 if throw of die is 1-5 $16.00 if throw of die is 6-10

$38.50 if throw of die is 1-5 $1.00 if throw of die is 6-10

A

B

6

$20.00 if throw of die is 1-6 $16.00 if throw of die is 7-10

$38.50 if throw of die is 1-6 $1.00 if throw of die is 7-10

A

B

7

$20.00 if throw of die is 1-7 $16.00 if throw of die is 8-10

$38.50 if throw of die is 1-7 $1.00 if throw of die is 8-10

A

B

8

$20.00 if throw of die is 1-8 $16.00 if throw of die is 9-10

$38.50 if throw of die is 1-8 $1.00 if throw of die is 9-10

A

B

9

$20.00 if throw of die is 1-9 $16.00 if throw of die is 10

$38.50 if throw of die is 1-9 $1.00 if throw of die is 10

A

B

10

$20.00 if throw of die is 1-10

$38.50 if throw of die is 1-10

A

B

DECISION ROW CHOSEN BY FIRST THROW OF THE DIE: _________ THROW OF THE DIE TO DETERMINE PAYMENT: ____________ EARNINGS: ________________

-47-

YRW

Decision Task

Your decision sheet shows ten decisions listed on the left. Each decision is a paired choice between “Option A” and “Option B.” You will make a choice on each row and record these in the final column. Here is a ten-sided die that will be used to determine payoffs. The faces are numbered from 0 to 9, and we will use the 0 face of the die to serve as 10. Look at Decision 10 at the bottom. Option A pays $20.00 if the throw of the ten sided die is 1, and it pays $16.00 if the throw is 2-10. Option B yields $38.50 if the throw of the die is 1, and it pays $1.00 if the throw is 2-10. The other Decisions are similar, except that as you move up the table, the chances of the higher payoff for each option increase. In fact, for Decision 1 in the top row, the die will not be needed since each option pays the highest payoff for sure, so your choice here is between $20.00 or $38.50. After you have made all of your choices, you will throw this die twice, once to select one of the ten decisions to be used, and a second time to determine what your payoff is for the option you chose, A or B, for the particular decision selected. Even though you will make ten decisions, only one of these will end up affecting your earnings, but you will not know in advance which decision will be used. We will not perform the dice roll until after you have completed the next short task.

-48-

YWR

Decision Task

Your decision sheet shows ten decisions listed on the left. Each decision is a paired choice between “Option A” and “Option B.” You will make a choice on each row and record these in the final column. Here is a ten-sided die that will be used to determine payoffs. The faces are numbered from 0 to 9, and we will use the 0 face of the die to serve as 10. Look at Decision 10 at the bottom. Option A pays $20.00 if the throw of the ten sided die is 1, and it pays $16.00 if the throw is 2-10. Option B yields $38.50 if the throw of the die is 1, and it pays $1.00 if the throw is 2-10. The other Decisions are similar, except that as you move up the table, the chances of the higher payoff for each option increase. In fact, for Decision 1 in the top row, the die will not be needed since each option pays the highest payoff for sure, so your choice here is between $20.00 or $38.50. After you have made all of your choices, you will throw this die twice, once to select one of the ten decisions to be used, and a second time to determine what your payoff is for the option you chose, A or B, for the particular decision selected. Even though you will make ten decisions, only one of these will end up affecting your earnings, but you will not know in advance which decision will be used.

-49-

ID: Y______________

Decision

Option A

Option B

Your Choice (Circle A or B)

1

$20.00 if throw of die is 1-10

$38.50 if throw of die is 1-10

A

B

2

$20.00 if throw of die is 1-9 $16.00 if throw of die is 10

$38.50 if throw of die is 1-9 $1.00 if throw of die is 10

A

B

3

$20.00 if throw of die is 1-8 $16.00 if throw of die is 9-10

$38.50 if throw of die is 1-8 $1.00 if throw of die is 9-10

A

B

4

$20.00 if throw of die is 1-7 $16.00 if throw of die is 8-10

$38.50 if throw of die is 1-7 $1.00 if throw of die is 8-10

A

B

5

$20.00 if throw of die is 1-6 $16.00 if throw of die is 7-10

$38.50 if throw of die is 1-6 $1.00 if throw of die is 7-10

A

B

6

$20.00 if throw of die is 1-5 $16.00 if throw of die is 6-10

$38.50 if throw of die is 1-5 $1.00 if throw of die is 6-10

A

B

7

$20.00 if throw of die is 1-4 $16.00 if throw of die is 5-10

$38.50 if throw of die is 1-4 $1.00 if throw of die is 5-10

A

B

8

$20.00 if throw of die is 1-3 $16.00 if throw of die is 4-10

$38.50 if throw of die is 1-3 $1.00 if throw of die is 4-10

A

B

9

$20.00 if throw of die is 1-2 $16.00 if throw of die is 3-10

$38.50 if throw of die is 1-2 $1.00 if throw of die is 3-10

A

B

10

$20.00 if throw of die is 1 $16.00 if throw of die is 2-10

$38.50 if throw of die is 1 $1.00 if throw of die is 2-10

A

B

DECISION ROW CHOSEN BY FIRST THROW OF THE DIE:________ THROW OF THE DIE TO DETERMINE PAYMENT: ____________ EARNINGS: ________________

-50-

ID: WLR _________

Decision Task

In this task you will have the opportunity to bid for the chance of getting an additional sum of money. Your bid can be any number in even quarters as long as it is not so large that it will lead to a loss for sure. We will tell you what that maximum is. You will draw a card from this deck to determine a value. If the value is less than your bid you will get the extra money. The money you get will be equal to 1.5 times the value, but we will then subtract your bid from that. You will be given three opportunities to bid, but we will only select one for payment at the end. This will be done randomly. Since it is possible to make a loss, we will give you a sum of money up front in addition to your earnings. This will be an amount between $5 and $15 and you will draw a card to determine that now. Any loss will be subtracted out of this sum, but you will be paid the remainder. If you do not make a loss you will be paid both your earnings and this sum of money. The three opportunities differ in the range of values that can be drawn from the deck of cards: •

The first time the value will be drawn between $3.50 and $10.50. This will be done using the deck of cards.

•

The second time the value will be drawn between $4 and $15.

•

The third time the value will be drawn between $3 and $6.

Additional amount of money: ____________________ Please write down your three bids here: Bid1: _____________ (maximum bid is $15.75) Bid 2: _____________ (maximum bid is $22.50) Bid 3: _____________ (maximum bid is $9) We will fill in the remainder after you have drawn the cards. First we will go on to the other short task. Value 1:

Bid1 > Value1?

[1.5 × Value1] - Bid1 =

Value 2:

Bid2 > Value2?

[1.5 × Value2] - Bid2 =

Value 3:

Bid3 > Value3?

[1.5 × Value3] - Bid3 = -51-

ID: RWL _________

Decision Task

In this task you will have the opportunity to bid for the chance of getting an additional sum of money. Your bid can be any number in even quarters as long as it is not so large that it will lead to a loss for sure. We will tell you what that maximum is. You will draw a card from this deck to determine a value. If the value is less than your bid you will get the extra money. The money you get will be equal to 1.5 times the value, but we will then subtract your bid from that. You will be given three opportunities to bid, but we will only select one for payment at the end. This will be done randomly. Since it is possible to make a loss, we will give you a sum of money up front in addition to your earnings. This will be an amount between $5 and $15 and you will draw a card to determine that now. Any loss will be subtracted out of this sum, but you will be paid the remainder. If you do not make a loss you will be paid both your earnings and this sum of money. The three opportunities differ in the range of values that can be drawn from the deck of cards: •

The first time the value will be drawn between $3.50 and $10.50. This will be done using the deck of cards.

•

The second time the value will be drawn between $4 and $15.

•

The third time the value will be drawn between $3 and $6.

Additional amount of money: ____________________ Please write down your three bids here: Bid1: _____________ (maximum bid is $15.75) Bid 2: _____________ (maximum bid is $22.50) Bid 3: _____________ (maximum bid is $9) We will fill in the remainder after you have drawn the cards. Value 1:

Bid1 > Value1?

[1.5 × Value1] - Bid1 =

Value 2:

Bid2 > Value2?

[1.5 × Value2] - Bid2 =

Value 3:

Bid3 > Value3?

[1.5 × Value3] - Bid3 =

-52-

ID: RLW _________

Decision Task

In this task you will have the opportunity to bid for the chance of getting an additional sum of money. Your bid can be any number in even quarters as long as it is not so large that it will lead to a loss for sure. We will tell you what that maximum is. You will draw a card from this deck to determine a value. If the value is less than your bid you will get the extra money. The money you get will be equal to 1.5 times the value, but we will then subtract your bid from that. You will be given three opportunities to bid, but we will only select one for payment at the end. This will be done randomly. Since it is possible to make a loss, we will give you a sum of money up front in addition to your earnings. This will be an amount between $5 and $15 and you will draw a card to determine that now. Any loss will be subtracted out of this sum, but you will be paid the remainder. If you do not make a loss you will be paid both your earnings and this sum of money. The three opportunities differ in the range of values that can be drawn from the deck of cards: •

The first time the value will be drawn between $3.50 and $10.50. This will be done using the deck of cards.

•

The second time the value will be drawn between $3 and $6.

•

The third time the value will be drawn between $4 and $15.

Additional amount of money: ____________________ Please write down your three bids here: Bid1: _____________ (maximum bid is $15.75) Bid 2: _____________ (maximum bid is $9) Bid 3: _____________ (maximum bid is $22.50) We will fill in the remainder after you have drawn the cards. Value 1:

Bid1 > Value1?

[1.5 × Value1] - Bid1 =

Value 2:

Bid2 > Value2?

[1.5 × Value2] - Bid2 =

Value 3:

Bid3 > Value3?

[1.5 × Value3] - Bid3 =

-53-

ID: LWR _________

Decision Task

In this task you will have the opportunity to bid for the chance of getting an additional sum of money. Your bid can be any number in even quarters as long as it is not so large that it will lead to a loss for sure. We will tell you what that maximum is. You will draw a card from this deck to determine a value. If the value is less than your bid you will get the extra money. The money you get will be equal to 1.5 times the value, but we will then subtract your bid from that. You will be given three opportunities to bid, but we will only select one for payment at the end. This will be done randomly. Since it is possible to make a loss, we will give you a sum of money up front in addition to your earnings. This will be an amount between $5 and $15 and you will draw a card to determine that now. Any loss will be subtracted out of this sum, but you will be paid the remainder. If you do not make a loss you will be paid both your earnings and this sum of money. The three opportunities differ in the range of values that can be drawn from the deck of cards: •

The first time the value will be drawn between $3.50 and $10.50. This will be done using the deck of cards.

•

The second time the value will be drawn between $3 and $6.

•

The third time the value will be drawn between $4 and $15.

Additional amount of money: ____________________ Please write down your three bids here: Bid1: _____________ (maximum bid is $15.75) Bid 2: _____________ (maximum bid is $9) Bid 3: _____________ (maximum bid is $22.50) We will fill in the remainder after you have drawn the cards. First we will go on to the other short task. Value 1:

Bid1 > Value1?

[1.5 × Value1] - Bid1 =

Value 2:

Bid2 > Value2?

[1.5 × Value2] - Bid2 =

Value 3:

Bid3 > Value3?

[1.5 × Value3] - Bid3 = -54-

Request for information on the survey and experiments – VOLUNTARY If you would like to receive a copy of a brief summary of the survey and our experiments, please provide your name and contact information below. We will not be using this information for any other purpose than to send you the summary. If you do not want to receive this information, then do not write anything. Note that we will not connect your name with the ID you used for the responses you gave us. NAME:

____________________________________________________

ADDRESS: ____________________________________________________ ____________________________________________________ ____________________________________________________ ____________________________________________________ E-MAIL:

____________________________________________________

FAX:

____________________________________________________

-55-

Appendix B: Instructions for Nebraska Sessions (NOT FOR PUBLICATION)

WELCOME TO THE RESEARCH STUDY DESCRIPTION This is a study of economic decision making. We think you will find it interesting, you will be paid $10 for your participation and you could earn additional money. How much you earn will depend partly on chance and partly on the choice you make in decision problems which you will be presented with. The instructions are simple and you will benefit from following them carefully. The problems are not designed to test you. What we want to know is what choices you would make in them. The only right answer is what you really would choose. That is why the problems give you the chance of earning real money. You will be paid in cash today. Part will be paid right after you have finished the task and the rest you can collect from 5.15 pm to 6 pm tonight, or else we can mail it to you. The task will proceed in three short parts. The first part consists of a few questions about your firm and you. This information is for research use only. The published results of our research will not identify any firm or individual, or the choice he or she made in any way. Nor will we give this identifying information to anyone else. The second and third part are short decision problems in which chance may play a part. Each decision-problem requires you to make a choice. This is described in more detail when you have completed the first part. Both of these parts may result in additional earnings over and above the $10 participation fee. We expect the entire task to take 20 - 30 minutes.

-56-

ID: _______

Part 1: Some Questions About You and Your Firm

In this survey most of the questions asked are descriptive. We will not be grading your answers and your responses are completely confidential. We will not be recording your name or the name of your firm on this sheet. Please think carefully about each question and give your best answers. Some Questions About You 1.

What is your AGE? ____________ years

2.

What is your sex? (Circle one number.) 01

3.

01 02 03 04 05 06

06 07 08 09

Hispanic-American Hispanic Mixed Race Other

Your spouse or domestic partner 01 02 03 04 05 06

Self-employed part-time Self-employed full-time Part-time employment in another firm (or government) Full-time employment in another firm (or government) Actively seeking employment Unemployed

Less than high school GED or High School Equivalency High school

04 05 06

Vocational or trade school Bachelor’s degree at college Higher degree at college

03 04

Separated or divorced? Widowed?

Are you currently... 01 02

7.

White African-American African Asian-American Asian

What is the highest level of education you have completed? (Circle one number) 01 02 03

6.

Female

What is your current employment status, and that of your spouse or domestic partner? (Circle all that apply) You

5.

02

Which of the following categories best describes you? (Circle one number.) 01 02 03 04 05

4.

Male

Single and never married? Married?

How many people live in your household? Include yourself, your spouse and any dependents. Do not include your parents or roommates unless you claim them as dependents. ___________________

OVER

-57-

8.

Please circle the category below that describes the total amount of INCOME earned in 2003 by the people in your household (as “household” is defined in the previous question). [Consider all forms of income, including salaries, tips, interest and dividend payments, scholarship support, student loans, parental support, social security, alimony, and child support, and others.] 01 02 03 04 05

$25,000 or under $25,001 - $45,000 $45,001 - $65,000 $65,001 - $85,000 $85,001 - $100,000

06 07 08 09 10

$100,001 - $125,000 $125,001 - $150,000 $150,001 - $175,000 $175,001 - $200,000 over $200,000

IF YOU SELECTED “SELF-EMPLOYED” IN QUESTION 4 THEN PLEASE COMPLETE THE QUESTIONS BELOW. OTHERWISE, YOU MAY PROCEED TO PART 2. Questions About Your Firm 1.

How old is your firm, in years? ____________

2.

What type of product or service do you provide? _____________________________

3.

Have you ever experienced a shortage of capital in running your firm? (Circle one number) 01 04

Never Often

02 05

Rarely Always

03

4.

Do you have a shortage of capital now?

5.

How did you primarily finance your firm’s start up? (Circle all that apply) 01 02 03 04

6.

Inheritance Gift Credit cards Earnings from another job

05 06 07

Yes

02

No

SBIR grant Private loan from a bank or person Other

Have you ever applied for or received an SBIR grant? Applied: Received:

8.

01

Occasionally

01 01

Yes Yes

02 02

No No

How do you finance your firm now? Enter rough percentages for each: 01 02 03 04 05 06 07

Government loans or grants Private loans from banks or people Credit cards Earnings from another job Cash from operations Equity capital Other

__________ __________ __________ __________ __________ __________ __________

9.

What would you estimate to be the annual revenue of your firm? _________________

10.

What would you estimate to be the value of the assets of your firm? _________________

11.

What is the state and ZIP code of the main location of your firm? ____________________ -58-

RE

Part 2: Decision Task

Your decision sheet shows ten decisions listed on the left. Each decision is a paired choice between “Option A” and “Option B.” You will make a choice on each row and record these in the final column. Here is a ten-sided die that will be used to determine payoffs. The faces are numbered from 0 to 9, and we will use the 0 face of the die to serve as 10. Look at Decision 1 at the top. Option A pays $20.00 if the throw of the ten sided die is 1, and it pays $16.00 if the throw is 2-10. Option B yields $38.50 if the throw of the die is 1, and it pays $1.00 if the throw is 2-10. The other Decisions are similar, except that as you move down the table, the chances of the higher payoff for each option increase. In fact, for Decision 10 in the bottom row, the die will not be needed since each option pays the highest payoff for sure, so your choice here is between $20.00 or $38.50. After you have made all of your choices, you will throw this die twice, once to select one of the ten decisions to be used, and a second time to determine what your payoff is for the option you chose, A or B, for the particular decision selected. Even though you will make ten decisions, only one of these will end up affecting your earnings, but you will not know in advance which decision will be used. We will not perform the dice roll until after you have completed the next short task. After you have finished all the tasks we will pay you the fee for completing the study and for this task. They payment for the next task will be ready this evening from 5.15 pm - 6 pm.

OVER

-59-

ID: RE______________

Decision

Option A

Option B

Your Choice (Circle A or B)

1

$20.00 if throw of die is 1 $16.00 if throw of die is 2-10

$38.50 if throw of die is 1 $1.00 if throw of die is 2-10

A

B

2

$20.00 if throw of die is 1-2 $16.00 if throw of die is 3-10

$38.50 if throw of die is 1-2 $1.00 if throw of die is 3-10

A

B

3

$20.00 if throw of die is 1-3 $16.00 if throw of die is 4-10

$38.50 if throw of die is 1-3 $1.00 if throw of die is 4-10

A

B

4

$20.00 if throw of die is 1-4 $16.00 if throw of die is 5-10

$38.50 if throw of die is 1-4 $1.00 if throw of die is 5-10

A

B

5

$20.00 if throw of die is 1-5 $16.00 if throw of die is 6-10

$38.50 if throw of die is 1-5 $1.00 if throw of die is 6-10

A

B

6

$20.00 if throw of die is 1-6 $16.00 if throw of die is 7-10

$38.50 if throw of die is 1-6 $1.00 if throw of die is 7-10

A

B

7

$20.00 if throw of die is 1-7 $16.00 if throw of die is 8-10

$38.50 if throw of die is 1-7 $1.00 if throw of die is 8-10

A

B

8

$20.00 if throw of die is 1-8 $16.00 if throw of die is 9-10

$38.50 if throw of die is 1-8 $1.00 if throw of die is 9-10

A

B

9

$20.00 if throw of die is 1-9 $16.00 if throw of die is 10

$38.50 if throw of die is 1-9 $1.00 if throw of die is 10

A

B

10

$20.00 if throw of die is 1-10

$38.50 if throw of die is 1-10

A

B

DECISION ROW CHOSEN BY FIRST THROW OF THE DIE: _________ THROW OF THE DIE TO DETERMINE PAYMENT: ____________ EARNINGS: ________________

-60-

ID: REM _________

Part 3: Decision Task

In this task you will have the opportunity to earn more money, by deciding if you want to enter a competitive market. We will determine your earnings from this task at 5.15pm today, here at our booth. You can come at any time from 5.15 pm until 6 pm and collect the remainder of your earnings. If you prefer, we can also mail them to you. You will be given a stake of $10 at the outset of this task. This task involves a decision as to whether to enter or not enter a market. 4 other people will be invited to enter or not enter this same market, but you will not know in advance how many of them have accepted. We are going to match people into markets randomly at the end of the day. The capacity of each market is 1 - only one person can make a profit on each market. If you decide to enter, your success will be under your control and will depend on your skill in answering some questions. We will rank all entrants according to how well they answer these questions. The highest ranked entrant will receive $35. Nobody else in the same market will earn anything. In order to enter you have to give up the $10 stake. This is the fee for entering the market. If you do not enter the market, you keep your $10 stake. The ranking system. The way that entrants will be ranked in this market is on the basis of a 7-item quiz of general knowledge. The questions will cover topics such as movie trivia, world history, geography, science, pets & animals, and “the world around us.” Those with higher scores will be ranked higher. If there is a tie for the top rank we will flip a coin to choose one person. You will take the quiz after you make your decision to enter the market. The questions will either be open-ended or multiple-choice. The openended questions will require you to write down the correct answer. The multiplechoice questions will require that you pick the correct answer from three alternatives. Many people believe that multiple-choice questions are much easier than open-ended questions. You will be given multiple-choice questions, and you will be competing against people who also have multiple-choice questions. To summarize: If you decide not to enter, you will keep the $10 stake. You will earn nothing beyond that from this task. Therefore, to guarantee that you will not lose, simply do not enter the market. If you decide to enter, you will receive $35 instead of the $10 stake if you are the highest-ranked out of those who enter. You will receive nothing from this task if you are ranked lower. At the most 5 people can enter each market. PLEASE TURN OVER -61-

Of course, whatever you decide here, and whatever the earnings in this task, you will also be paid $10 for completing the task, plus the earnings from the first task as soon as you are done. You will have to return between 5.15 pm and 6 pm to claim the earnings from this task, or they will be mailed to you. DO YOU CHOOSE TO ENTER THE MARKET AND COMPETE OR DO YOU CHOOSE TO STAY OUT?

G

Enter the market and give up the $10 stake

G

Stay out of the market and keep the $10 stake We have three final questions for you to answer.

1. Here is one more opportunity to earn money. Please answer this question: Across all of today’s markets that have multiple-choice questions, how many people do you think will enter on average? You will receive $10 if you estimate the number exactly. You will receive $8 if you are off by 1, $6 if you are off by 2, and so on. Please round off to an integer value. PICK ONE:

0

1

2

3

4

5

2. We would like to get your estimate of how confident you are about your ability to answer these skill questions. We are not paying you for your answer to this question, but would appreciate you thinking about it carefully. Specifically: If we compared your quiz answers with those of 100 other people picked at random from this conference, what rank do you think you would have? A rank of 1 means that you answered the questions better than anyone else, a rank of 50 that you think you answer better than half but not as well as the other half, and a rank of 100 means that you think everyone else does better than you. ANSWER: ____________ 3. Finally, How did your hear about our booth? 01 02 03

Walk-by Heard from somebody that participated Other ______________________________________________ -62-

ID: REO _________

Part 3: Decision Task

In this task you will have the opportunity to earn more money, by deciding if you want to enter a competitive market. We will determine your earnings from this task at 5.15pm today, here at our booth. You can come at any time from 5.15 pm until 6 pm and collect the remainder of your earnings. If you prefer, we can also mail them to you. You will be given a stake of $10 at the outset of this task. This task involves a decision as to whether to enter or not enter a market. 4 other people will be invited to enter or not enter this same market, but you will not know in advance how many of them have accepted. We are going to match people into markets randomly at the end of the day. The capacity of each market is 1 - only one person can make a profit on each market. If you decide to enter, your success will be under your control and will depend on your skill in answering some questions. We will rank all entrants according to how well they answer these questions. The highest ranked entrant will receive $35. Nobody else in the same market will earn anything. In order to enter you have to give up the $10 stake. This is the fee for entering the market. If you do not enter the market, you keep your $10 stake. The ranking system. The way that entrants will be ranked in this market is on the basis of a 7-item quiz of general knowledge. The questions will cover topics such as movie trivia, world history, geography, science, pets & animals, and “the world around us.” Those with higher scores will be ranked higher. If there is a tie for the top rank we will flip a coin to choose one person. You will take the quiz after you make your decision to enter the market. The questions will either be open-ended or multiple-choice. The openended questions will require you to write down the correct answer. The multiplechoice questions will require that you pick the correct answer from three alternatives. Many people believe that multiple-choice questions are much easier than open-ended questions. You will be given open-ended questions, and you will be competing against people who also have open-ended questions. To summarize: If you decide not to enter, you will keep the $10 stake. You will earn nothing beyond that from this task. Therefore, to guarantee that you will not lose, simply do not enter the market. If you decide to enter, you will receive $35 instead of the $10 stake if you are the highest-ranked out of those who enter. You will receive nothing from this task if you are ranked lower. At the most 5 people can enter each market. PLEASE TURN OVER -63-

Of course, whatever you decide here, and whatever the earnings in this task, you will also be paid $10 for completing the task, plus the earnings from the first task as soon as you are done. You will have to return between 5.15 pm and 6 pm to claim the earnings from this task, or they will be mailed to you. DO YOU CHOOSE TO ENTER THE MARKET AND COMPETE OR DO YOU CHOOSE TO STAY OUT?

G

Enter the market and give up the $10 stake

G

Stay out of the market and keep the $10 stake We have three final questions for you to answer.

1. Here is one more opportunity to earn money. Please answer this question: Across all of today’s markets that have multiple-choice questions, how many people do you think will enter on average? You will receive $10 if you estimate the number exactly. You will receive $8 if you are off by 1, $6 if you are off by 2, and so on. Please round off to an integer value. PICK ONE:

0

1

2

3

4

5

2. We would like to get your estimate of how confident you are about your ability to answer these skill questions. We are not paying you for your answer to this question, but would appreciate you thinking about it carefully. Specifically: If we compared your quiz answers with those of 100 other people picked at random from this conference, what rank do you think you would have? A rank of 1 means that you answered the questions better than anyone else, a rank of 50 that you think you answer better than half but not as well as the other half, and a rank of 100 means that you think everyone else does better than you. ANSWER: ____________ 3. Finally, How did your hear about our booth? 01 02 03

Walk-by Heard from somebody that participated Other ______________________________________________ -64-

ER

Part 3: Decision Task

Your decision sheet shows ten decisions listed on the left. Each decision is a paired choice between “Option A” and “Option B.” You will make a choice on each row and record these in the final column. Here is a ten-sided die that will be used to determine payoffs. The faces are numbered from 0 to 9, and we will use the 0 face of the die to serve as 10. Look at Decision 1 at the top. Option A pays $20.00 if the throw of the ten sided die is 1, and it pays $16.00 if the throw is 2-10. Option B yields $38.50 if the throw of the die is 1, and it pays $1.00 if the throw is 2-10. The other Decisions are similar, except that as you move down the table, the chances of the higher payoff for each option increase. In fact, for Decision 10 in the bottom row, the die will not be needed since each option pays the highest payoff for sure, so your choice here is between $20.00 or $38.50. After you have made all of your choices, you will throw this die twice, once to select one of the ten decisions to be used, and a second time to determine what your payoff is for the option you chose, A or B, for the particular decision selected. Even though you will make ten decisions, only one of these will end up affecting your earnings, but you will not know in advance which decision will be used. We will perform the dice roll after you have completed this task. We will then pay you the fee for completing the study and for this task. They payment for the previous task will be ready this evening from 5.15 pm - 6 pm.

OVER

-65-

ID: ER______________

Decision

Option A

Option B

Your Choice (Circle A or B)

1

$20.00 if throw of die is 1 $16.00 if throw of die is 2-10

$38.50 if throw of die is 1 $1.00 if throw of die is 2-10

A

B

2

$20.00 if throw of die is 1-2 $16.00 if throw of die is 3-10

$38.50 if throw of die is 1-2 $1.00 if throw of die is 3-10

A

B

3

$20.00 if throw of die is 1-3 $16.00 if throw of die is 4-10

$38.50 if throw of die is 1-3 $1.00 if throw of die is 4-10

A

B

4

$20.00 if throw of die is 1-4 $16.00 if throw of die is 5-10

$38.50 if throw of die is 1-4 $1.00 if throw of die is 5-10

A

B

5

$20.00 if throw of die is 1-5 $16.00 if throw of die is 6-10

$38.50 if throw of die is 1-5 $1.00 if throw of die is 6-10

A

B

6

$20.00 if throw of die is 1-6 $16.00 if throw of die is 7-10

$38.50 if throw of die is 1-6 $1.00 if throw of die is 7-10

A

B

7

$20.00 if throw of die is 1-7 $16.00 if throw of die is 8-10

$38.50 if throw of die is 1-7 $1.00 if throw of die is 8-10

A

B

8

$20.00 if throw of die is 1-8 $16.00 if throw of die is 9-10

$38.50 if throw of die is 1-8 $1.00 if throw of die is 9-10

A

B

9

$20.00 if throw of die is 1-9 $16.00 if throw of die is 10

$38.50 if throw of die is 1-9 $1.00 if throw of die is 10

A

B

10

$20.00 if throw of die is 1-10

$38.50 if throw of die is 1-10

A

B

DECISION ROW CHOSEN BY FIRST THROW OF THE DIE: _________ THROW OF THE DIE TO DETERMINE PAYMENT: ____________ EARNINGS: ________________

-66-

ID: EMR _________

Part 2: Decision Task

In this task you will have the opportunity to earn more money, by deciding if you want to enter a competitive market. We will determine your earnings from this task at 5.15pm today, here at our booth. You can come at any time from 5.15 pm until 6 pm and collect the remainder of your earnings. If you prefer, we can also mail them to you. You will be given a stake of $10 at the outset of this task. This task involves a decision as to whether to enter or not enter a market. 4 other people will be invited to enter or not enter this same market, but you will not know in advance how many of them have accepted. We are going to match people into markets randomly at the end of the day. The capacity of each market is 1 - only one person can make a profit on each market. If you decide to enter, your success will be under your control and will depend on your skill in answering some questions. We will rank all entrants according to how well they answer these questions. The highest ranked entrant will receive $35. Nobody else in the same market will earn anything. In order to enter you have to give up the $10 stake. This is the fee for entering the market. If you do not enter the market, you keep your $10 stake. The ranking system. The way that entrants will be ranked in this market is on the basis of a 7-item quiz of general knowledge. The questions will cover topics such as movie trivia, world history, geography, science, pets & animals, and “the world around us.” Those with higher scores will be ranked higher. If there is a tie for the top rank we will flip a coin to choose one person. You will take the quiz after you make your decision to enter the market. The questions will either be open-ended or multiple-choice. The open-ended questions will require you to write down the correct answer. The multiple-choice questions will require that you pick the correct answer from three alternatives. Many people believe that multiple-choice questions are much easier than openended questions. You will be given multiple-choice questions, and you will be competing against people who also have multiple-choice questions. To summarize: If you decide not to enter, you will keep the $10 stake. You will earn nothing beyond that from this task. Therefore, to guarantee that you will not lose, simply do not enter the market. If you decide to enter, you will receive $35 instead of the $10 stake if you are the highest-ranked out of those who enter. You will receive nothing from this task if you are ranked lower. At the most 5 people can enter each market. PLEASE TURN OVER -67-

Of course, whatever you decide here, and whatever the earnings in this task, you will also be paid $10 for completing the task, plus the earnings from the next task as soon as you are done. You will have to return between 5.15 pm and 6 pm to claim the earnings from this task, or they will be mailed to you. DO YOU CHOOSE TO ENTER THE MARKET AND COMPETE OR DO YOU CHOOSE TO STAY OUT?

G

Enter the market and give up the $10 stake

G

Stay out of the market and keep the $10 stake We have three final questions for you to answer.

1. Here is one more opportunity to earn money. Please answer this question: Across all of today’s markets that have multiple-choice questions, how many people do you think will enter on average? You will receive $10 if you estimate the number exactly. You will receive $8 if you are off by 1, $6 if you are off by 2, and so on. Please round off to an integer value. PICK ONE:

0

1

2

3

4

5

2. We would like to get your estimate of how confident you are about your ability to answer these skill questions. We are not paying you for your answer to this question, but would appreciate you thinking about it carefully. Specifically: If we compared your quiz answers with those of 100 other people picked at random from this conference, what rank do you think you would have? A rank of 1 means that you answered the questions better than anyone else, a rank of 50 that you think you answer better than half but not as well as the other half, and a rank of 100 means that you think everyone else does better than you. ANSWER: ____________ 3. Finally, How did your hear about our booth? 01 02 03

Walk-by Heard from somebody that participated Other ______________________________________________ -68-

ID: EOR _________

Part 2: Decision Task

In this task you will have the opportunity to earn more money, by deciding if you want to enter a competitive market. We will determine your earnings from this task at 5.15pm today, here at our booth. You can come at any time from 5.15 pm until 6 pm and collect the remainder of your earnings. If you prefer, we can also mail them to you. You will be given a stake of $10 at the outset of this task. This task involves a decision as to whether to enter or not enter a market. 4 other people will be invited to enter or not enter this same market, but you will not know in advance how many of them have accepted. We are going to match people into markets randomly at the end of the day. The capacity of each market is 1 - only one person can make a profit on each market. If you decide to enter, your success will be under your control and will depend on your skill in answering some questions. We will rank all entrants according to how well they answer these questions. The highest ranked entrant will receive $35. Nobody else in the same market will earn anything. In order to enter you have to give up the $10 stake. This is the fee for entering the market. If you do not enter the market, you keep your $10 stake. The ranking system. The way that entrants will be ranked in this market is on the basis of a 7-item quiz of general knowledge. The questions will cover topics such as movie trivia, world history, geography, science, pets & animals, and “the world around us.” Those with higher scores will be ranked higher. If there is a tie for the top rank we will flip a coin to choose one person. You will take the quiz after you make your decision to enter the market. The questions will either be open-ended or multiple-choice. The openended questions will require you to write down the correct answer. The multiplechoice questions will require that you pick the correct answer from three alternatives. Many people believe that multiple-choice questions are much easier than open-ended questions. You will be given open-ended questions, and you will be competing against people who also have open-ended questions. To summarize: If you decide not to enter, you will keep the $10 stake. You will earn nothing beyond that from this task. Therefore, to guarantee that you will not lose, simply do not enter the market. If you decide to enter, you will receive $35 instead of the $10 stake if you are the highest-ranked out of those who enter. You will receive nothing from this task if you are ranked lower. At the most 5 people can enter each market. PLEASE TURN OVER -69-

Of course, whatever you decide here, and whatever the earnings in this task, you will also be paid $10 for completing the task, plus the earnings from the next task as soon as you are done. You will have to return between 5.15 pm and 6 pm to claim the earnings from this task, or they will be mailed to you. DO YOU CHOOSE TO ENTER THE MARKET AND COMPETE OR DO YOU CHOOSE TO STAY OUT?

G

Enter the market and give up the $10 stake

G

Stay out of the market and keep the $10 stake We have three final questions for you to answer.

1. Here is one more opportunity to earn money. Please answer this question: Across all of today’s markets that have multiple-choice questions, how many people do you think will enter on average? You will receive $10 if you estimate the number exactly. You will receive $8 if you are off by 1, $6 if you are off by 2, and so on. Please round off to an integer value. PICK ONE:

0

1

2

3

4

5

2. We would like to get your estimate of how confident you are about your ability to answer these skill questions. We are not paying you for your answer to this question, but would appreciate you thinking about it carefully. Specifically: If we compared your quiz answers with those of 100 other people picked at random from this conference, what rank do you think you would have? A rank of 1 means that you answered the questions better than anyone else, a rank of 50 that you think you answer better than half but not as well as the other half, and a rank of 100 means that you think everyone else does better than you. ANSWER: ____________ 3. Finally, How did your hear about our booth? 01 02 03

Walk-by Heard from somebody that participated Other ______________________________________________ -70-

ID: _____

Address for Payment of Earnings

Please provide your name and the address that we should mail your earnings to, in the event that you are not able to return tonight at from 5.15 pm until 6 pm to collect them in cash: NAME:

____________________________________________________

ADDRESS: ____________________________________________________ ____________________________________________________ ____________________________________________________ ____________________________________________________

If you would like to receive a copy of a brief summary of the survey and our experiments, please check the box below:

G Yes, please send me the summary If you prefer e-mail or FAX for the summary, please provide this information (otherwise we will mail it to you if requested): E-MAIL:

____________________________________________________

FAX:

____________________________________________________

-71-

Nebraska Sessions Subject ID:________ First Payment Record Task 1 (decision with dice) First Die Throw (Decision Row Chosen)

________________

Second Die Throw (Outcome)

________________

Earnings

$______________

Participation Fee

$10.00

Total payment in this part:

$______________

Print Name:

_________________________________________

SSN:

_________________________________________

Your SSN is required by the University of Central Florida Accounting in order for us to get reimbursed from our research funds for these payments.

Signature that you have received the TOTAL earnings indicated above in cash:

Sign here:

_________________________________________

You may also receive additional earnings from Task 2 plus you will get your participation fee when you return this evening.

-72-

Nebraska Sessions Subject ID:________

Last Name:_____________________

Second Payment Record No Entry Stake

$______________

Task 2 (market entry decision) Your score in the quiz: Your rank in the group: Earnings for Task 2

$______________

Task 3 (prediction) Your prediction of entrants: Actual entrants on average: Earnings for Task 3

$______________

TOTAL Earnings from these two tasks:

$______________

Print Name:

_________________________________________

SSN:

_________________________________________

Your SSN is required by the University of Central Florida Accounting in order for us to get reimbursed from our research funds for these payments.

Signature that you have received the TOTAL earnings indicated above in cash, in addition to the earnings received earlier for participating and from Task 1:

Sign here:

_________________________________________

Market:___________________

-73-