Experimental Economics, Entrepreneurs and The Entry Decision by Julie Ann Elston, Glenn W. Harrison and E. Elisabet Rutström † May 2006
Abstract. Do entrepreneurs exhibit more over-confidence in their own skills, leading to excess market entry? Building on previous laboratory experiments we propose experimental tasks to identify and test the hypothesized characteristic of overconfidence and its alleged impact on the entry decision. Taking these tasks into the field we identify and find striking differences between two types of entrepreneurs, which we call full-time and part-time entrepreneurs. Part-time entrepreneurs appear extremely reluctant to enter markets where profitability is based on their perception of their relative skill ability. On the other hand full-time entrepreneurs and non-entrepreneurs do not exhibit any systematic over-confidence in their relative skill abilities. Our results support the notion that entrepreneurs are rational and do not exhibit excess entry due to over-confidence as many have claimed.
†
Elston: Max Planck Institute of Economics (Jena, Germany) and Oregon State University (Bend, Oregon, USA). Harrison and Rutström: University of Central Florida (Orlando, Florida, USA). Email:
[email protected],
[email protected] and
[email protected]. Rutström thanks the U.S. National Science Foundation for research support under grants NSF/IIS 9817518 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.
1. Introduction What determines the decision of entrepreneurs to enter new markets? One answer provided by economic theory is that an excess level of profitability induces entry into an industry. This is why the entry of new firms is important: new firms provide an equilibrating function in the market, restoring price and profit to competitive levels. One natural question that has evolved from this theory is whether entrepreneurs exhibit more over-confidence than others, and therefore enter industries more quickly than others.1 If so, then there would be “excess entry” and the potential for an inefficient market outcome if exit is slow or costly.2 We make use of experimental economics to examine the impact of over-confidence on the entry decision for a group of high-technology entrepreneurs and a control group of nonentrepreneurs. Experimental methods have become standard in economics, as a way of generating data that allows the investigator to control many of the features of the environment that can only be proxied with naturally occurring data. A central feature of experiments in economics is that the design provides incentives for the subjects to make different decisions: there are always real consequences to taking one action or another.3 Our approach marries the rigorous control of an experimental laboratory environment with the relevance of having real entrepreneurs as subjects. We view the experimental data we obtained as methodologically complementary to naturally occurring data and previous surveys, recognizing that each has its strengths and limitations. There are several reasons to apply experimental methods to study entrepreneurial decision making. First, a well designed experiment can potentially elicit and measure the dynamic decision
1
De Meza and Southey [1996] present a formal argument to rationalize the oft-repeated claim that entrepreneurs have poor access to capital because there is a tendency for those who are excessively optimistic to dominate new entrants. They conclude that the tendency to unrealistic optimism on the part of entrepreneurs leads to excess entry and maximum use of self financing by a self selected group of risk-lovers. Hence banks should be applauded, so the argument goes, for “stemming” the rush for capital that would otherwise just be wasted by irrational entrepreneurs. In related studies, Bénabou and Tirole [2002], Dosi and Lovallo [1997] and Hoelzl and Rustichini [2005] all point to the role of self-confidence in influencing the entrepreneurial decision making process. 2 For example, Parker [2006] provides evidence that entrepreneurs in Britain do not rapidly update their beliefs about unobserved productivity in the light of new information. 3 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.
-1-
making of the entrepreneur. This can be done by controlling for or randomizing influences in the study environment, which cannot be effectively implemented with a static pencil and paper questionnaire. Thus we study how actual decisions are affected by the real-time flow of information. Second, the ability to accurately measure the impact of economic incentives on entrepreneurial decision making and performance is enhanced by the ability to use real cash incentives to induce incentive compatible behavior from subjects. Third, since it has historically been difficult to observe those who did not decide to become entrepreneurs, field experiments are ideal for accessing individuals who did and did not decide to become entrepreneurs, allowing potential differences to be directly observed and empirically tested.4 We define over-confidence as a judgmental error where the individual estimates their own skill or ability to be better than the average. Our tasks are structured to separate this decision from the entry decision which is based on the entrepreneur’s forecast of the competition -that is the entry decision of others. The empirical results of this study reveal that full-time entrepreneurs are no more likely to enter a market in which performance depends on perceived skill in relation to others. But we find that “wanna be entrepreneurs,” those that hold on to salaried employment while starting up a business venture, are significantly less likely to choose to enter such a market. These findings are consistent with Levesque and Schade [2005; p.314], who report survey evidence that the “...tendency to work too much in the wage job instead of entirely focusing on the [entrepreneurial] venture is found to be most pronounced for risk-averse individuals.”
4
Schade [2005; p.417] offers a survey of experimental studies in entrepreneurship and lists 14 studies, only 2 of which actually use real entrepreneurs. Of the remaining 12 studies, students are the most common type of experimental subject, which raises questions about the value of using students to generalize to real entrepreneurs raised by Robinson, Hueffner and Hunt [1991]. Moreover, all of the studies referred to use hypothetical surveys, and none of the tasks was incentivized in the sense that subjects earned more or less money depending on different choices. The use of real, controlled incentives has been a hallmark of experimental economics since Smith [1982] defined the “salience” and “dominance” precepts of an experimental micro-economy. The studies reviewed by Schade [2005] are better described as using what some have called the “questionnaire-experimental method,” where different hypothetical survey questions are exogenously posed to subjects with some experimental design. Amiel and Cowell [1999] illustrate the method, and clearly note (p.24) that their “... approach involves presenting individuals with questionnaires in a way that uses many of the features of experimental methodology.” The use of hypothetical questionnaires also has a long tradition in psychology and environmental valuation: see Harrison [2006a][2006b], Harrison and Rutström [2006] and Hertwig and Ortmann [2001] for extensive surveys of experimental evidence on the biases introduced by eliciting hypothetical responses.
-2-
Empirical evidence has been accumulating supporting the notion that over-confidence influences the entrepreneurial decision. For example, Cooper et al. [1988], Busenitz and Barney [1997], and Camerer and Lovallo [1999] conclude that the perception of both the risk associated with entrepreneurial activity as well as the entrepreneur’s own capabilities can result in what has been termed as over-confidence. Forbes [2005] reviews the literature, and presents evidence that many individual and contextual factors influence over-confidence. In particular, he shows that foundermanagers are more overconfident than comparable new-venture managers that played no role in the entry decision. Koellinger, Minniti and Schade [2005] use survey responses from the Global Entrepreneurship Monitor data base to show that perceptions of risk and own ability have a systematic influence on the decision for individuals to start a new business. Those individuals who suffer from over-confidence tend to have a greater propensity to enter into entrepreneurship. Their analysis spans many components of confidence. They ask questions about (a) whether the entrepreneur believes that they have sufficient skills, knowledge and ability to start a new business, (b) their perception of good business opportunities, (c) their optimism about their household financial security in the near future, (d) and their fear of failure. These surveys cover 18 countries, and focus on individuals in the process of starting a business venture. In section 1 we review the literature on the importance of over-confidence in the entrepreneurial entry decision, and suggest hypotheses to test with our experimental design. In section 2 we describe the experimental tasks we developed to study the entry decision, in section 3 we document the field experiments conducted, in section 4 we present the empirical results, and in section 5 we draw conclusions and make suggestions on directions for future research.
2. Experimental Tasks We build on experimental tasks which are well established in the literature, and have been applied to study the behavior of the traditional subject pool of college students. Specifically, we start with a market entry task developed by Camerer and Lovallo [1999] (CL), and then adapt it for our -3-
field application.
A. Previous Laboratory Experiments 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]. CL view these games as well-suited to study the behavior of potential entrepreneurs. They justify 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.5 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.6 They found “excess entry” with student subjects when profitability was
5
treatment.
They review a more extensive literature studying this basic entry game in various settings without the skill
6 Entry is also affected by risk attitudes, since entry is risky and 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. A similar design is developed by Hoelzl and Rustichini [2005] to identify over-confidence 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 over-confidence in their skill ability.
-4-
determined by perceived skill, and attribute this to over-confidence. The left panel of Figure 1 displays the essential results, in terms of average realized profit per entrant in the two main treatments.7 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 (and graphs) further shows that over-confidence 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.8 Figures 2 and 3 display results broken down by market capacity levels. If entrepreneurs exhibit more over-confidence than others and enter industries more quickly than others, then the implied outcome is that there will be relatively more subsequent failures and decisions to exit the market. However, this outcome is not one that is directly measured in this design. Such outcomes could be observed if we could run 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 we must 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
7 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. 8 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].
-5-
themselves as dumb blondes with no chance of being the most skilled of the entrants. By directly 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 1. 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 [2005] argue that one should think in more general terms of biases in comparative judgment that are contextual, rather than simply think in terms of over-confidence or under-confidence as generic traits of individuals. The reason is that there is (disputed) evidence from a range of tasks in psychology that the same person can be overconfident on simple tasks and underconfident on difficult tasks.9 Of course, what is simple for one person may be difficult for another
9 The disputes are well summarized by Juslin, Winman and Olsson [2000]. Over-confidence in the psychology literature refers to a specific concept and construct. Imagine that one has elicited answers to a two-alternative multiplechoice question, and that the proportion of correct answers is some fraction C. Then imagine that one has elicited a measure of the subjective probability that the answer a given subject has provided is correct, and that the average of these elicited probabilities is B. Then over-confidence is said to occur when B>C. To an experimental economist, one immediate methodological concern is that neither of the responses is typically generated with incentives for the subject to be accurate or thoughtful. Our experimental design does provide such incentives.
-6-
person, and vice versa. But assume that this difference in perceived difficulty can be made 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 [2005] essentially replicate the design of CL, who generally used simple trivia questions.10 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.
B. Our Experiments 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 to assist in recognizing the right answer. 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 also randomized the order and sub-set of questions from the larger test-bank.11 Market entry games are implemented in our experiments using the following instructions, which were given to each subject:
10 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 The original source of our questions was the commercial test bank http://www.funtrivia.com/.
-7-
Instructions to Participants 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 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 -8-
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 two treatments. In order to better identify the contribution of skill over-confidence 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. 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. It was important not to tie the answer to this question to the entry decision 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 -9-
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 in order to control for post-entry feedback.
3. Field Experiments A. Field Data In 2005 we conducted field experiments at both of the two biennial Small Business Innovation Research (SBIR) National Conferences. Appendix A lists the announcement used to inform entrepreneurs of the scope of the conference. These national conferences attract actual entrepreneurs from around the United States who attend to learn how to apply for SBIR funding for their entrepreneurial firm. The SBIR program is highly competitive and distributes about $2 billion dollars annually to small firms providing products and services to US government agencies. 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, and as such, is relatively representative of small high-technology firms in the US. Our sample of entrepreneurs included firms in business consulting (36%), engineering (21%), medical research & development (10%), information technology and software development (30%), and agriculture (3%).
B. Procedures Our field experiment was conducted from a booth set up in the exhibitors area of the national conference with a banner titled “Research Study: Case for Participation.” The booth was run by Elston, Rutström and two graduate students, all wearing matching university polo shirts and slacks in school colors. The picture below illustrates the environment. All tasks involved real economic consequences to the individual to provide motivation for subjects. 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 -10-
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 complete set of instructions for one treatment is in Appendix B. Each packet was housed in a folder so that participants could work privately given the tight spacing around the booth and to minimize subject interaction. The subject’s first task was to complete a survey of firm and individual characteristics before advancing to the decisions tasks, which were handled on a one to one basis. Subjects completed two
-11-
decisions tasks, only one of which we focus on here.12 The order of these two decision tasks was randomized, to control for possible order effects. 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. To ensure credibility we completed the other decision task and paid every subject in cash for their participation fee and earnings in that task. 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]). To facilitate the efficient conduct of the paperwork in the market entry game, we had a research assistant grading each quiz after it was completed and preparing the associated paperwork and payment records.
4. Empirical Results A. Data Description 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 report this as their sole business focus and state that they have no additional source of salaried income. We classify 38 (21%) as being Part-Time Entrepreneurs, since they report an entrepreneurial venture and state that they additionally have full-time salaried jobs. Thus we have information on 80 individuals that report some entrepreneurial experience. In each case we required that subjects provide detailed information on the nature of their entrepreneurial firm so that we could classify them as an entrepreneur or otherwise. 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
12
Elston, Harrison and Rutström [2005] describe and evaluate the other task, as well as other field experiments we have conducted with entrepreneurs.
-12-
groups of entrepreneurs, and the others to be our controls. Unless otherwise stated, the expression “entrepreneurs” refers to Full-Time and Part-Time Entrepreneurs as defined above, and the comparison group would 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 be 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. The mix of industries represents by the two groups is virtually identical, so there is no difference in the samples in terms of industry of the entrepreneurial activity.13
B. Analysis Figure 4 shows the results for the question eliciting beliefs about the number of potential entrants. We break the responses down according to the entrepreneurship categories used in our analysis. It is apparent that FT Entrepreneurs expected more entrants on average than the control group, although the latter has two modes. The striking result from Figure 3 is that PT Entrepreneurs as a group had such diffuse priors on the number of potential entrants. Roughly as many expected only 1 entrant as expected 5 entrants. Thus we see the most striking difference within the class of entrepreneurs. Figure 5 shows the results for the question eliciting beliefs about the skill that the subject perceived themselves as having. Since this is a critical variable for our analysis, consider again the 13 We test this hypothesis by means of a Fisher Exact Test of the null that the assignment of entrepreneurs by FT or PT status is not associated with their industry. The p-value on this null is 0.36, so we cannot reject the hypothesis of no difference.
-13-
question that elicited these responses: 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. Thus larger numerical responses indicated subjects that thought themselves less skillful in relation to the other subjects in the sample, and which they might expect to compete with in the market entry game. To display this response in the most transparent manner, we simply take it’s negative: hence in Figure 5 the subjects that are more to the right are those that have greater confidence in their relative skill. The results in Figure 5 are quite striking as they suggest: PT Entrepreneurs think very highly of themselves compared to all others. Putting the results from Figures 4 and 5 together, we can form some hypotheses about the forces leading to entry across the entrepreneurship categories. FT Entrepreneurs are less likely to enter since they expect more potential competitors and have no particularly crisp belief that their skills are better than others in this task. On the other hand, PT Entrepreneurs might be expected to enter more readily, since they have diffuse priors about the number of potential entrants, but perceive themselves as being more skilled than most. Figures 6, 7 and 8 contrast these elicited beliefs in skill level with actual outcomes. The actual outcomes are, of course, given by the score that the subject received out of 7 possible correct answers. To compare to the measure of confidence, we normalized this score by multiplying by 14.29 = 100÷7, and then taking the negative. Thus a normalized score of -100 indicated someone that did as poorly as possible (0 out 7 correct), and a normalized score of 0 indicated someone that did as well as possible (7 out of 7 correct). We should add that our quiz questions were not a pushover! The average score in the difficult frame was only 1.59 correct, and in the easy frame it was still only 3.99 correct. The results in Figures 6-8 are again striking: the PT Entrepreneurs were significantly over-confident in their skill ability, whereas the FT Entrepreneurs and Control group were quite accurate as a whole and exhibited no general tendency to over-confidence. -14-
Figure 9 brings us to the actual entry decisions, and compares them by market to the predictions. We observe significant entry into these markets. Over 73% of the markets had complete entry, with 5 out of 5 potential entrants choosing to enter. Around 18% of the markets had 4 entrants, and a handful of markets (around 4% each) had 2 or 3 entrants. The average number of firms to enter was 4.6. The distributions in Figure 9 are pooled across entrepreneurial categories, but it is an easy matter to gauge errors in belief for each category by comparison with Figure 3. Clearly FT Entrepreneurs do the best job of predicting entry, and PT Entrepreneurs the worst. To put these results in perspective, Table 1 spells out the arithmetic underlying this entry decision. In the final column we see that a risk-neutral subject, who only cared about average earnings, would be marginally inclined to enter if there were 3 entrants, since the average earnings would be $11 and the deterministic cost of entry is $10. So an average entry level of 4.6 is consistent with some slight risk-loving behavior, some over-confidence in ability to be the one entrant to earn the $35 jackpot, or simple mis-perception of the task (which can, of course, explain any outcome). Since the $10 from not entering is risk-free, and the $11 average earnings at an entry level of three has a considerable range, either $35 or $0, one would need to assume a considerable amount of risk loving to justify entry to that level. In Table 2 we evaluate the determinants of entry using a simple logit statistical model. Entry is the binary dependent variable. Our main focus is on the propensity of FT or PT entrepreneurs to enter compared to the control group. If we ignore all procedural differences in the various tasks, we calculate that FT Entrepreneurs are no more likely than the control group to enter, but that PT entrepreneurs are significantly less likely to enter. We calculate the estimated marginal effect of binary variables for each category.14 Specifically, FT entrepreneurs are 5.9% less likely to enter than the control, but the standard error on that estimate is 10.3 percentage points, so the p-value on the null hypothesis of no effect has a significance level of 0.569. Hence we cannot infer that there is any 14 These marginal effects are calculated at the means of all variables. For dummy variables they show the effect of a change from 0 to 1. Virtually identical effects are obtained if one calculates the average marginal effects over all observations, using procedures developed by Bartus [2005].
-15-
statistically significant difference between the control group and FT entrepreneurs. But PT entrepreneurs are estimated to be 30.8% less likely to enter, and the p-value on the null hypothesis is only 0.010 in this case. The 95% confidence interval on this estimated effect is between 54.3% and 7.4%. Are these conclusions robust to the various treatments and controls we have available? Essentially, yes. We obtain the same results if we add controls for whether the subject was in a multiple-choice quiz or not, whether they had heard about the task from a previous participant, whether they participated on the second day, whether the entry task was first or second in order, and an array of standard demographic characteristics. The marginal effect for FT (PT) Entrepreneurs is then estimated to be -0.02% (-41.8%), with a p-value of 0.80 (0.002). We do observe a striking effect from the subject being Black, associated with a reduction in the probability of entry of roughly 40 percentage points. Extending the robustness check to the inclusion of variables measuring individual confidence in skill levels, over-confidence in skill levels, and beliefs about likely entry, we find the same results.15 The marginal effects for FT (PT) Entrepreneurs are now 0.03% (-32.9%), with a pvalue of 0.43 (0.005). Those that predicted more entrants were more likely to enter, which is counter-intuitive unless subjects were risk-loving. Confidence in skill ability had no significant effect on entry, of course after all other factors are controlled for. But subjects that were over-confident in their skill levels were significant less inclined to enter, consistent with the finding that PT Entrepreneurs were particularly over-confident. This effect is consistent with the reduction in the marginal effect of being a PT Entrepreneur, from to -41.8% to -32.9% when these variables are controlled for. Thus we conclude that there is no effect of being a FT Entrepreneur on market entry relative to
15
The over-confidence measure does contain an element of endogeneity, since one of the entrants being counted in the actual entry for the market is of course the subject. Since the dependent variable is that entry decision, errors would tend to be correlated with the over-confidence measure. However, the entrant is typically only 1 of 5 (in 73% of cases) or 1 of 4 (in 18% of cases), so this effect cannot be quantitatively large.
-16-
those with no entrepreneurial experience. But there is a dramatic effect from being a PT Entrepreneur: these individuals are significantly less likely to enter. The difference between FT and PT Entrepreneurs is particularly telling because we have already controlled for differences in their individual characteristics and market beliefs which might independently explain the lower propensity of PT Entrepreneurs to enter.
5. Conclusions We examine the hypothesis, based on rich anecdote, that over-confidence in their own abilities leads entrepreneurs to excessively enter markets. We design and implement field experimental tasks to directly elicit choices of entrepreneurs among market entry alternatives. We control for beliefs about likely market entry, and their actual skill in determining the likelihood of success post-entry. We find robust evidence that full-time entrepreneurs are no more likely to enter a market in which performance depends on perceived skill in relation to others. But “wanna be entrepreneurs,” who simultaneously hold salaried employment, are significantly less likely to choose to enter such a market. Our approach offers a methodological bridge between early decades of entrepreneurial research into the individual traits of successful entrepreneurs, and the later decades of research into cognitive traits of successful entrepreneurs. The earlier literature found few characteristics of entrepreneurs that seemed to predict behavior with any reliability, leading Hatten [1997; p.40] to conclude that “The conclusions of 30 years of research indicate that there are no personality characteristics that predict who will be a successful entrepreneur. [...] Successful small business owners and entrepreneurs come in every shape, size, color, and from all backgrounds.” The latter literature found many cognitive characteristics that seemed to characterize entrepreneurs, as summarized by Baron [1998]. The bridge we offer is to use the methods of experimental economics, in which incentivized tasks are presented to subjects that bring the characteristics and context of the field to bear on responding to those tasks. Our experimental design illustrates how one can take -17-
well-posed laboratory instruments into the field, to the subjects that interest us.16 Our results on entry and over-confidence provide an obvious motivation for future experiments to study exit, given that we find that there is no lack of entry into our experimental markets. The key issue, then, is whether entrepreneurs behave differently when losses start piling up in these markets, and they have to decide whether to exit or keep producing. Our experiments did not extend beyond the one-shot entry stage, so we cannot address that issue within this design, although it raises fascinating issues such as the potential effects of differing access to credit, aversion to losses, and the horizon over which entrants frame their choices.
16
Harrison and List [2004] discuss the complementarity of laboratory and field experiments in greater detail.
-18-
Figure 1: Returns to Entry Average Profits of Entrants Data from rounds 1-12 of experiments of Camerer-Lovallo, AER 1999 Pooled over all market capacity levels Random Sample Selection
Skill Sample Selection
15
15
10
10 Skill Irrelevant
5
5
0
Skill Irrelevant
0
Skill Relevant
Skill Relevant -5
-5 1
2
3
4
5
6 7 Ro und
8
9
10 11 12
-19-
1
2
3
4
5
6 7 Round
8
9
10 11 12
Figure 2: Industry Profit By Market Capacity if Skill is Irrelevant Data from Random Selection Experiments of Camerer-Lovallo, AER 1999 Capacity = 2
Capacity = 4
Capacity = 6
Capacity = 8
40 20 0 -20 -40 40 20 0 -20 -40 1
2
3
4
5
6
7
8
9 10 11 12
1
2
3
4
5
6
7
8
9 10 11 12
Round
Figure 3: Industry Profit By Market Capacity if Skill Is Relevant Data from Random Selection Experiments of Camerer-Lovallo, AER 1999 Capacity = 2
Capacity = 4
Capacity = 6
Capacity = 8
40 20 0 -20 -40 40 20 0 -20 -40 1
2
3
4
5
6
7
8
9 10 11 12
Round
-20-
1
2
3
4
5
6
7
8
9 10 11 12
Figure 4: Elicited Beliefs About Entry Kernel density estimates, by Entrepreneurship category
Density
FT Entrepreneurs PT Entrepreneurs Control
1
2
3
Predicted Number of Entrants
4
5
Figure 5: Elicited Confidence in Skill Negative of the elicited percentile that the subject said that they were in Kernel density estimates, by Entrepreneurship category
Density
FT Entrepreneurs PT Entrepreneurs Control
-100
-75
-50
Confidence in Quiz Aptitude
-21-
-25
0
Figure 6: Over-Confidence of Part-Time Entrepreneurs
Density
Actual Predicted
-100
-75
-50
Quiz Aptitude
Figure 7: Over-Confidence of Full-Time Entrepreneurs
0
Figure 8: Over-Confidence of Control Group
Actual Predicted
Actual Predicted
Density
Density -100
-25
-75
-50
Quiz Aptitude
-25
0
-22-
-100
-75
-50
Quiz Aptitude
-25
0
Figure 9: Errors in Beliefs About Entry
Density
Actual Predicted
1
2
3
Predicted Number of Entrants
4
5
Table 1: 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
-23-
Table 2: Hypothesis Tests for Market Entry Decision Dependent variable is Entry, where entry is denoted by 1. Marginal effects from logit model, with market entry as the dependent variable. N=118. Variable
FT Entrepreneur PT Entrepreneur
FT Entrepreneur PT Entrepreneur Quiz_MC Heard Day2 OrderTask Age Female Black Asian HigherEd Married HighInc
FT Entrepreneur PT Entrepreneur Quiz_MC Heard Day2 OrderTask Age Female Black Asian HigherEd Married HighInc Predict Confident OverConf
Estimate
-0.059 -0.308
Std. Error
p-value
A. No Controls 0.103 0.569 0.119 0.010
95% Confidence Interval
-0.261 -0.543
Mean
0.143 -0.074
0.212 0.186
0.153 -0.150 0.152 0.079 0.160 0.239 0.006 0.070 0.080 0.181 0.141 0.227 0.123
0.212 0.186 0.500 0.441 0.475 0.500 43.740 0.364 0.051 0.076 0.500 0.754 0.407
C. Including Procedural Controls, Demographics, and Measures of Confidence 0.034 0.043 0.426 -0.050 0.117 -0.329 0.118 0.005 -0.560 -0.098 -0.076 0.099 0.444 -0.270 0.118 -0.040 0.046 0.378 -0.130 0.050 0.011 0.038 0.772 -0.064 0.086 0.097 0.050 0.055 -0.002 0.196 -0.001 0.002 0.560 -0.005 0.002 0.000 0.034 0.985 -0.072 0.073 -0.456 0.231 0.049 -0.909 -0.003 0.042 0.038 0.269 -0.033 0.117 -0.005 0.040 0.905 -0.084 0.074 0.037 0.062 0.548 -0.084 0.159 0.059 0.040 0.138 -0.019 0.136 0.075 0.026 0.003 0.025 0.126 -0.001 0.002 0.559 -0.005 0.003 -0.003 0.002 0.083 -0.007 0.000
0.212 0.186 0.500 0.441 0.475 0.500 43.740 0.364 0.051 0.076 0.500 0.754 0.407 3.550 -33.020 -6.813
B. Including Procedural Controls and Demographic Characteristics -0.023 0.090 0.789 -0.199 -0.418 0.138 0.002 -0.689 0.255 0.646 0.694 -0.101 -0.044 0.063 0.481 -0.168 0.034 0.064 0.593 -0.091 0.104 0.069 0.131 -0.031 0.000 0.003 0.868 -0.005 -0.057 0.065 0.379 -0.185 -0.398 0.243 0.102 -0.875 0.013 0.086 0.879 -0.155 0.016 0.063 0.794 -0.108 0.050 0.090 0.578 -0.127 0.000 0.063 0.993 -0.124
Definitions: 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; Quiz_MC is a dummy variable denoting the use of the multiple choice quiz questions; 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; OrderTask is a dummy variable measuring subjects who were given the market entry task before the risk attitude task; HigherEd indicates subjects that have completed a higher college degree (e.g., MBA, JD or PhD); Married indicates subjects that are currently married; HighInc indicates subjects whose household income in 2003 exceeded $100,000; 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); OverConf is Confidence minus the normalized score actually obtained by the subject, as explained in the text; other variables defined obviously.
-24-
References Amiel, Yoram, and Cowell, Frank A., Thinking About Inequality (New York: Cambridge University Press, 1999). Baron, Robert, Cognitive Mechanisms in Entrepreneurship: Why and When Entrepreneurs Think Differently from Other People,” Journal of Business Venturing, 13(4), 1998, 275-294. Bartus, Tamás, “Estimation of Marginal Effects Using margeff,” The Stata Journal, 5(3), 2005, 309329. Bénabou, Roland, and Tirole Jean, “Self Confidence and Personal Motivation,” Quarterly Journal of Economics, 2002, 871-915. Bridgeman, Brent, “A Comparison of Quantitative Questions in Open-Ended and Multiple-Choice Formats,” Journal of Educational Measurement, 29, 1992, 253-271. Busenitz, Lowell W., and Barney, Jay B., “Differences between Entrepreneurs and Managers in Large Organizations: Biases and Heuristics in Strategic Decision-Making” Journal of Business Venturing, 12(1), January 1997, 9-30. 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. Cooper, A.C., Gagman-Garcon, F.J., and Woo, C.Y., “Initial Human and Financial Capital as Predictors of New Venture Performance,” Journal of Business Venturing, 9, 1994, 371-395 de Meza, David, and Southey, Clive, “The Borrower’s Curse: Optimism, Finance and Entrepreneurship,” Economic Journal, 106, March 1996, 365-386. Dosi, G., and Lovallo, Dan, “Rational entrepreneurs or optimistic martyrs? Some considerations on technological regimes, corporate entries and the evolutionary role of decision biases,” in R. Garud, P. Nayyar & Z. Shapira (eds.), Technological Innovation: Oversights and Foresights (New York: Cambridge University Press, 1997). Elston, Julie; Harrison, Glenn W., and Rutström, E. Elisabet, “Characterizing the Entrepreneur Using Field Experiments,” Working Paper 05-30, Department of Economics, College of Business Administration, University of Central Florida, 2005. Forbes, Daniel P., “Are Some Entrepreneurs More Overconfident Than Others?” Journal of Business Venturing, 20, 2005, 623-640. Harrison, Glenn W., ““Hypothetical Bias Over Uncertain Outcomes,” in J.A. List (ed)., Using Experimental Methods in Environmental and Resource Economics (Northampton, MA: Elgar, 2006a). Harrison, Glenn W., ““Experimental Evidence on Alternative Environmental Valuation Methods,” Environmental & Resource Economics, 34, 2006b forthcoming. Harrison, Glenn W.; Lau, Morten I., and Williams, Melonie B., “Estimating Individual Discount -25-
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 Harrison, Glenn W., and Rutström, E. Elisabet, “Experimental Evidence on the Existence of Hypothetical Bias in Value Elicitation Methods,” in C.R. Plott and V.L. Smith (eds.), Handbook of Experimental Economics Results (North-Holland: Amsterdam, 2006) Hatten, Timothy S., Small Business: Entrepreneurship and Beyond (Upper Saddle River, NJ: Prentice-Hall, First Edition, 1997). Hertwig, Ralph, and Ortmann, Andreas, “Experimental practices in economics: A challenge for psychologists?” Behavioral & Brain Sciences, 24, 2001, 383-451. Hoelzl, Erik, and Rustichini, Aldo, “Overconfident: Do You Put Your Money On It?” Economic Journal, 115, April 2005, 305-318. Juslin, Peter; Winman, Anders, and Olsson, Henrik, “Naive Empiricism and Dogmatism in Confidence Research: A Critical Examination of the Hard-Easy Effect,” Psychological Review, 107(2), 2000, 384-396. Kennedy, Peter E., and Walstad, William B., “Combining Multiple-Choice and ConstructedResponse Test Scores: An Economist’s View,” Applied Measurement in Education, 10, 1997, 359-375. Koellinger, Phillip; Minniti, Maria, and Schade, Christian, “‘I think I can, I think I can’: Overconfidence and Entrepreneurial Behavior,” DIW Discussion Paper 501, German Institute for Economic Research, Berlin, July 2005. Levesque, Moren, and Schade, Christian, “Intuitive Optimizing: Experimental Findings on Time Allocation Decisions with Newly Formed Ventures,” Journal of Business Venturing, May 2005, 313-342. List, John A., and Lucking-Reilly, David, “Demand Reduction in Multi-unit Auctions: Evidence from a Sportscard Field Experiment,” American Economic Review, 90(4), September 2000, 961972. March, James, and Shapira, Zur, “Managerial Perspectives on Risk and Risk Taking,” Management Science, 33(11), November 1987, 1404-1418. Moore, Don A., and Cain, Daylain M., “Overconfidence and Underconfidence: When and Why People Underestimate (and Overestimate) the Competition,” Working Paper, Tepper School of Business, Carnegie-Mellon University, 2005. Parker, Simon C., “Learning About the Unknown: How Fast Do Entrepreneurs Adjust Their Beliefs?” Journal of Business Venturing, 21, 2006, 1-26. Robinson, P.B.; Hueffner J.C., and Hunt, K.H., “Entrepreneurial Research on Student Subjects Does Not Generalize to Real World Entrepreneurs,” Journal of Small Business Management, 29(2), 1991, 42-50. -26-
Roll, Richard, “The Hubris Hypothesis of Corporate Takeovers,” Journal of Business, 59(2), April 1986, 197-216. Schade, Christian, “Dynamics, Experimental Economics, and Entrepreneurship,” Journal of Technology Transfer, 30(4), October 2005, 409-431. 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).
-27-
Appendix A 2005 Spring National SBIR/STTR Conference: Omaha, NE Date: Monday, March 07, 2005 to Thursday, March 10, 2005 Web Site: www.sbirworld.com/omaha Conference Address: Hilton Omaha, 1001 Cass Street, Omaha, NE Description: The National Science Foundation, in association with the Small Business Administration and all 11 SBIR agencies, is sponsoring this 2005 National SBIR/STTR Conference. Annually, SBIR and STTR programs provide over $2 billion to small businesses through federal programs to help entrepreneurs take their ideas from conception to reality. This conference will give you the tools you need to obtain part of the $2+ billion available to small business innovators. • • • • • • • • • • •
Topics will include: Proposal Preparation -The Basics Business Basics Partnerships & Resources Accounting for SBIR/STTR Marketing: How to Sell Yourself and Your Idea Focus on Regional Projections & Trends Focus on Homeland Security Focus on Software Development Focus on BioMed Focus on Maximizing University- Business Relationships and the SBIR/STTR Program Corporations -Why They’re Interested and How to Access Them
Each participant will also have 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. Who Should Attend: Attendees include sales and marketing professionals, small business owners, entrepreneurs, university researchers, scientists seeking commercialization strategies, venture capitalists, and all small businesses seeking to secure federal funding. Email:
[email protected]
-28-
Appendix B Instructions for Experimental Sessions (Treatment REM) 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. 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.
Male
02
Female
Which of the following categories best describes you? (Circle one number.) 01 02 03 04 05
White African-American African Asian-American Asian
06 07 08 09 -29-
Hispanic-American Hispanic Mixed Race Other
4.
What is your current employment status, and that of your spouse or domestic partner? (Circle all that apply) You 01 02 03 04 05 06
5.
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.
01 02 03 04 05 06
What is the highest level of education you have completed? (Circle one number) 01 02 03
6.
Your spouse or domestic partner
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
$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 -30-
03
Occasionally
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
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? ____________________ Part 2 Decision Task: see Elston, Harrison and Rutström [2005]
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 -31-
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 highestranked 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 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 -32-
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 ______________________________________________
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. Last Name:_____________________
Subject ID:________
Second Payment Record No Entry Stake
$______________ -33-
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:___________________
-34-
