Strategic Management Journal Strat. Mgmt. J. , 31: 822–840 (2010) Published online EarlyView in Wiley InterScience (www. interscience. wiley. com) DOI: 10. 1002/smj. 833 Received 18 January 2007; Final revision received 14 November 2009 ARE INDIVIDUALS ENTERING SELF-EMPLOYMENT OVERLY OPTIMISTIC? AN EMPIRICAL TEST OF PLANS AND PROJECTIONS ON NASCENT ENTREPRENEUR EXPECTATIONS GAVIN CASSAR* The Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania, U. S. A. This research examines the rationality of the expectations of nascent entrepreneurs.
Consistent with conjectures regarding entry into self-employment, I ? nd substantial overoptimism in nascent entrepreneurs’ expectations, in that they overestimate the probability that their nascent activity will result in an operating venture. Further, for those ventures that achieve operation, individuals overestimate the expected future sales and employment. To explain variations in overoptimism, I posit that those individuals who adopt an inside view to forecasting through the use of plans and ? ancial projections, will exhibit greater ex ante bias in their expectations. Consistent with the inside view causing overoptimism in expectations, I ? nd that the preparation of projected ? nancial statements results in more overly optimistic venture sale forecasts. Copyright ? 2010 John Wiley & Sons, Ltd. INTRODUCTION Expectations of future returns from vocational activity underlie the choice of employment and trigger the decision to start a venture (Cassar, 2006; Shane and Venkataraman, 2000).
Individuals, when choosing to invest their time and capital in nascent entrepreneurial activity, do so with an expectation of success, such as the probability of achieving an operating business. Expectations also underlie the subsequent actions of individuals through the venturing process, such as whether to continue in self-employment or how much additional time and capital to invest (Gimeno Keywords: expectations; ? nancial projections; forecasting; nascent entrepreneur; overoptimism; planning; rationality Correspondence to: Gavin Cassar, The Wharton School, University of Pennsylvania, 1300 Steinberg Hall-Dietrich Hall, Philadelphia, PA 19104-6365, U. S. A. E-mail: [email protected] upenn. edu et al. , 1997; McCarthy, Schoorman, and Cooper, 1993). Given the importance of expectations on individual behavior, there are consequences for individuals, their stakeholders, and the economy to making inaccurate or systematically biased predictions related to entrepreneurial choices.
The rationality of nascent entrepreneurial expectations, that predictions of future outcomes related to venturing activity are made without bias, is often assumed when modeling entry into selfemployment and organizational creation (Evans and Jovanovic, 1989; Evans and Leighton, 1989). However, researchers acknowledge that individuals are bounded and open to potential in? uences that may lead them to have biased expectations (Kahneman and Lovallo, 1993; Kahneman and Tversky, 1973; Simon 1955).
For example, there is substantial evidence that individuals generally have overly optimistic expectations, in that their expectations of speci? c future outcomes are more favorable than what eventually occurs (Weinstein, Copyright ? 2010 John Wiley & Sons, Ltd. Nascent Entrepreneurial Expectations 1980). 1 The tendency of nascent and existing entrepreneurs to have overoptimistic expectations may explain why individuals enter and persist in self-employment even in the presence of lower returns from venturing (Camerer and Lovallo, 1999; Moskowitz and Vissing-Jorgensen, 2002).
While there are conjectures as to the presence or magnitude of overoptimism in the expectations of those entering or continuing in self-employment, surprisingly, there is little evidence on the rationality of entrepreneurial expectations. Simply put, there is no research that has investigated how ex ante expectations of individuals entering selfemployment, namely nascent entrepreneurs, map onto actual ex post realizations such as the creation of an operating business, achieved sales, or returns from self-employment. Consequently, our understanding of the nature of entrepreneurial xpectations and entrepreneurial entry is incomplete. I address this important gap by examining the rationality of the expectations of nascent entrepreneurs, namely those individuals who are currently involved in the process of starting a business, using data from the Panel Study of Entrepreneurial Dynamics (PSED). The distinctive feature of the PSED is that it identi? es and surveys nascent entrepreneurs who are in the process of starting new ventures, thereby overcoming potential survivorship and recall biases associated with surveying entrepreneurs already in business (Gartner et al. 2004). Further, the PSED longitudinally surveys nascent entrepreneurs throughout the startup process, allowing for the comparison of ex ante expectations with ex post realizations, and thereby overcoming the use of perceptual measures of predictions or outcomes. To understand potential causes for nascent entrepreneurial overoptimism, I examine the in? uence of plans and projections use by the entrepreneur on differences in overoptimism.
Individuals undertaking nascent venture activity provide a powerful sample for an investigation of overoptimism given the high uncertainty of the forecasting task and the increased likelihood that entrepreneurs will exhibit greater cognitive bias than the general population (Baron, 2004; Busenitz and Barney, 1997). While, 823 1 Consistent with observed overoptimism, some recent theoretical economic models have explicit modeling of optimistic or overcon? dent individuals entering into self-employment (Bernardo and Welch, 2001; Brocas and Carrillo, 2004). enerally, more comprehensive and detailed information analysis is considered to be bene? cial when formulating expectations (Durand, 2003; Simon, Houghton, and Aquino, 2000), several authors suggest that some information acquisition and use may, in fact, be ineffective in dampening, or may potentially exacerbate, the overoptimistic tendencies of individuals (Armor and Taylor, 1998; Kahneman and Lovallo, 1993; Zacharakis and Shepherd, 2001). In this article, I argue that overoptimism in nascent entrepreneurs is exacerbated by the adoption of an ‘inside view’ (Kahneman and Lovallo, 1993: 17) to forecasting.
Individuals form an inside view forecast by focusing on the speci? cs of the case, the details of the plan that exists, and obstacles to its completion, and by constructing scenarios of future progress (Kahneman and Lovallo, 1993: 25–26). Researchers have shown that individuals generally adopt an inside view to generate forecasts (Buehler, Grif? n, and Ross, 1994). The inside view is synonymous with the planning fallacy, which focuses the attention of decision makers on the speci? cs of the problem rather than on the outcomes of similar cases or base rate information (Baron, 1998; Kahneman and Tversky, 1979).
Consequently, the planning fallacy, realized through the inside view, is argued to cause the tendency for individuals to hold con? dent beliefs in the success of their own projects, even when knowing other individuals have failed in similar projects. I ? nd signi? cant overoptimism in the expectations of nascent entrepreneurs, in that they overestimate the probability that their nascent activities will successfully lead to an operating business. Further, of those nascent entrepreneurs who do achieve an operating venture, I observe that they overestimate both the future sales and number of employees in the ? st year of operation. These results are consistent with optimistic entry of individuals into self-employment. I also ? nd that formal business planning is associated with individuals having greater expectations of creating an operating business from their start-up activity; however, these beliefs appear justi? ed, in that formal business planning also increases the likelihood of creating an operating business. Consistent with the inside view exacerbating overoptimism in expectations, I ? nd that the preparation of projected ? ancial statements leads to overly optimistic sales forecasts for the venture. Overall, this study documents that nascent entrepreneurs exhibit Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Copyright ? 2010 John Wiley & Sons, Ltd. 824 G. Cassar THEORY DEVELOPMENT AND HYPOTHESES Overoptimism While rationality of entrepreneurial expectations is often assumed in economic models of entry into venturing, researchers acknowledge that individuals are bounded and open to potential in? uences that may lead them to have biased expectations (Kahneman and Lovallo, 1993; Kahneman and Tversky, 1973).
There is substantial evidence in many domains that individuals are optimistic, in that they have a generalized positive outlook toward the future (Scheier and Carver, 1985). This evidence shows not only that individuals perceive that favorable events are more likely to happen, but also that favorable events are more likely to happen to them rather than to their peers (Taylor and Brown, 1988). Similarly, individuals generally exhibit overcon? dence, in that they overestimate their ability to do well (Larwood and Whittaker, 1977). Such overcon? ence can be in regard to an individual’s knowledge, ability to predict the future, or general personal abilities (Hayward, Shepherd, and Grif? n, 2006: 162; Kruger and Dunning, 1999; Simon and Houghton, 2003). The presence of optimism about exogenous events and excessive self-con? dence results in decision makers generally having overly optimistic expectations, in that their expectations of the outcomes regarding their chosen actions exhibit a positive ex ante bias (Gervais, Heaton, and Odean, 2007; Heaton, 2002; Malmendier and Tate, 2005a). In other words, decision makers’ expectations of future outcomes are more favorable than what eventually occurs. Further, overoptimism tends to be exacerbated when tasks are perceived to be controllable and therefore is likely to be heightened if expectations are based upon planned activity (Weinstein, 1980). In this study’s setting, individuals are considered to be overly optimistic when they overestimate the likelihood 2 Some authors refer to overcon? ence to describe the overestimation of expected returns, such as personal wealth, from their decisions (Hayward et al. , 2006: 161; Malmendier and Tate, 2005b: 651). Kahneman and Tversky (1995: 46) use the term optimistic overcon? dence to describe individuals’ overestimation of success and their certainty to ensure success. I adopt the term overoptimism from the forecasting literature, which describes the properties of the forecast investigated that is argued to be an outcome of the decision-making process (Cassar and Gibson, 2007; Sedor, 2002). ubstantial overoptimism in their expectations and provides to what extent overoptimism is in? uenced by the use of plans and projections. The study’s ? ndings contribute to theory and practice in several ways. First, it is the ?rst study to provide longitudinal evidence of the presence and magnitude of overoptimism in individuals’ expectations during the venture evaluation process. The magnitude of overoptimism observed is substantial, with nascent entrepreneurs on average overestimating the likelihood of venture operation by 33. 2 percent.
By providing novel empirical support of overly optimistic expectations as an explanation for why individuals enter self-employment, this study contributes to the literature that seeks to understand the forces that underlie entry and persistence of entrepreneurial behavior, and consequently, organization creation. Second, this study contributes to understanding by being the ? rst study to investigate how the inside view can affect differences in forecast overoptimism in a ? eld setting. While the inside view has been conjectured to be associated with overoptimism, there is limited extant theory testing of this within a ? ld sample. The surveying of individuals undertaking start-up activity overcomes concerns from experimental data, which may overstate the role of decision-making biases in the ? eld (Camerer and Lovallo, 1999; Simon and Houghton, 2003). Third, this study contributes to management practice by providing a link between management practices and observed psychological bias. Understanding what in? uences forecasting bias is vital for all decision makers because expectations underlie future actions and behavior.
The novel contribution of this study is the theoretical development and empirical support that the use of plans and projections may have a non-neutral effect on the expectations of decision makers. Documenting a relationship between commonly used management practices and overly optimistic prediction is particularly important, given the fundamental role that ? nancial projections play in allowing decision makers to structure expectations and cope with the future, and because academics and stakeholders generally encourage the use of formal planning and ? nancial projections.
The evidence from this study suggests that the same management activities that are relied upon and encouraged to be adopted to cope with uncertainty are associated with overly optimistic expectations. Copyright ? 2010 John Wiley & Sons, Ltd. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Nascent Entrepreneurial Expectations that their nascent activity will result in an operating venture or when they overestimate the future sales and employment of their new ventures. However, the presence of overoptimistic beliefs in those that enter nascent entrepreneurship and self-employment does not require optimism, overcon? ence, or any presence of psychological bias. Entrepreneurs can hold overoptimistic beliefs simply as a consequence of acting rationally on noisy yet unbiased signals of returns from vocational activity (Van den Steen, 2004); whereby, those individuals who act on favorable signals from venturing are more likely to have observed signals with favorable errors (Brown, 1974; Harrison and March, 1984). Therefore, simple self-selection based on rational choices can result in overoptimism in those who choose to undertake nascent activity and become entrepreneurs. Given its de? ition and measurement, overoptimism is observed in this study regardless if it is a consequence of cognitive bias or hubris, of simple self-selection, or of a combination of these factors. While there is no empirical evidence of the magnitude, or even presence, of overoptimism in the expectations of nascent entrepreneurs, Cooper, Woo, and Dunkelberg (1988) observed that existing entrepreneurs reported the odds of their business ‘succeeding’ to be signi? cantly higher than historically observed and substantially better than the odds of success for other similar businesses.
They concluded that the observed differences in expectations was likely caused by ex post decision bolstering, in that once the entry into selfemployment had been made, entrepreneurs were likely to justify this choice by believing it would lead to success. In this study setting, ex post decision bolstering cannot be a cause of overoptimism, as the expectations observed are elicited before the venture is operational. Hypothesis 1: Nascent entrepreneurial expectations are overly optimistic.
Plans and projections I argue that overoptimistic forecasts of nascent entrepreneurs’ are exacerbated by the inside view whereby predictions are anchored on plans in which individuals have a vested interest. Individuals generate an inside view forecast by focusing on the case at hand, considering the plan and Copyright ? 2010 John Wiley & Sons, Ltd. 825 obstacles to its completion, and constructing scenarios of future progress (Kahneman and Lovallo, 1993: 25). The inside view, which is synonymous with the planning fallacy, draws on knowledge of the speci? s of the case, details of the plan, and ideas about how likely obstacles might be overcome. The inside view represents an individual’s attempt to come to grips with the complexities of the unique characteristics and circumstances of the task or project. In contrast, an outside view is statistical and comparative in nature and does not involve any attempt to divine the future at any level of detail (Kahneman and Lovallo, 1993: 25). For example, there is no detailed forecasting of future events that may in? uence the project’s course, or strategies and contingencies to deal with these events.
The outside view is a form of reference class forecasting where the success of the case at hand is evaluated through the observed success of other cases with similar attributes or circumstances (Lovallo and Kahneman, 2003). Consistent with an inside view approach, evidence from experimental research suggests that people typically make forecasts by constructing a mental scenario of how the project is likely to develop (Buehler et al. , 1994). Scenarios can be described as representations of temporally ordered sequences of events glued together by causal relationships (Grif? , Dunning, and Ross, 1990; Johnson and Sherman, 1990). The process of scenario formulation allows the decision maker to evaluate a given problem in a step-by-step manner by analyzing the outcome of a series of causal events (Hastie and Dawes, 2001). Experimental research has also shown that when individuals are provided with reasons or scenarios for why an outcome will be achieved, their expectation of that outcome occurring increases (Hoch, 1984; Levi and Pryor, 1987). Buehler, Grif? n, and MacDonald (1997) found that when participants were given incentives to complete a task, they exhibited greater optimistic bias.
Further analysis revealed that the incentives focused participants more narrowly on their plans, consistent with the inside view. Consequently, the inside view tends to exacerbate tendencies of optimism and overcon? dence that are observed in individuals’ decision making and formulation of expectations. The use of plans and projections by entrepreneurs is associated with scenario representation. For example, a formal business plan speci? es the ? rm’s goals and provides broad strategies through Strat. Mgmt. J. 31: 822–840 (2010) DOI: 10. 1002/smj 826 G. Cassar An important aspect of the arguments relating the inside view to forecasting bias is the context and setting of the forecast. As with decision making generally, the presence of nonrational expectations is more likely in situations of high uncertainty or complexity (Kahneman, Slovic, and Tversky, 1982). The tendency to rely on singular information while neglecting base rate or distributional information is enhanced by the perceived uniqueness of the problem (Kahneman and Tversky, 1979).
Consequently, the inside view arguments do not pertain to decisions that are routinely repeated, with the critical question being whether a particular problem or forecast is treated as unique, or as a problem similar to many others (Kahneman and Lovallo, 1993: 25). Grossly overly optimistic forecasts may be especially likely if the forecasts relate to a new technology or new type of project or otherwise places the forecaster in an unfamiliar or dynamic setting (Hayward et al. , 2006: 164).
In contrast, opportunities for learning and for statistical aggregation exist when closely related problems are frequently encountered (Kahneman and Lovallo, 1993: 18). All of this suggests that studying nascent entrepreneurs during the start-up stage provides a powerful setting to investigate the in? uence of the inside view on forecasting biases. Hypothesis 2a: Nascent entrepreneurs who adopt formal plans have expectations that are more overly optimistic. Hypothesis 2b: Nascent entrepreneurs who adopt ? nancial projections have expectations that are more overly optimistic. to operational details of how the ? m will achieve these goals. Similarly, ? nancial projections provide a monetary representation of the goals and expected performance of the ? rm. Business plans and ? nancial projections emphasize the interrelationship among various activities and actions of the ? rm. For example, strategic goals lead to operational goals, which lead to speci? c tasks to enable goal attainment. Financial projections are a function of strategies and planned behavior. Further, the relevance of outside sources can be masked by detailed acquaintance with the speci? c case or by intense involvement in it (Kahneman and Tversky, 1979).
As more scenario-based information is provided and processed by the decision maker, such as through formal documents detailing scenarios of goal achievement, the individual’s propensity to employ an inside approach should increase, reducing the consideration given to outside sources of information. Consequently, the development of business plans or ? nancial projections should encourage or be re? ective of an inside approach to the formulation of expectations. While entrepreneurs can make plans or set ? nancial targets without formally developing business plans or ? ancial projections, it is undertaking these management activities that encourage re? ection on the implementation of one’s actions, thereby making people more con? dent in the success of the plan or the probability of achieving the targets speci? ed (Armor and Taylor, 1998; Taylor and Gollwitzer, 1995). Social psychologists suggest that people become more con? dent in their beliefs when they generate explanations for their choices and beliefs (Heath and Gonzalez, 1995). Further, the inside view may exacerbate other biases in decision making that can contribute to overoptimism in expectations.
For example, when individuals focus on their own capabilities and actions, they may neglect the abilities and potential actions of competitors (Lovallo and Kahneman, 2003). Additionally, individuals may anchor on the positive goals formulated in plans and ? nancial projections, such as estimated revenue, and not place enough weight on new information, such as recent sales (Tversky and Kahneman, 1974). Consequently, the development or use of plans and scenarios by individuals with a vested interest in these planned behaviors is more likely to exacerbate tendencies to formulate overoptimistic expectations.
Copyright ? 2010 John Wiley & Sons, Ltd. METHOD Sample I used the PSED to investigate the rationality of entrepreneurs’ expectations. 3 The distinctive feature of the PSED is that it identi? es and surveys nascent entrepreneurs in the process of starting new ventures, thereby overcoming potential 3 Research using data, questions, or research designs developed from the PSED has been published in numerous mainstream management and entrepreneurship based journals including: Entrepreneurship Theory and Practice, Journal of Business Venturing, Journal of Management,
Journal of Small Business Management, Small Business Economics, Strategic Management Journal, and Management Science. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Nascent Entrepreneurial Expectations survivorship and recall biases, such as hindsight bias, self-justi? cation bias, and attribution bias, associated with surveying entrepreneurs already in business (Conway and Ross, 1984; Golden, 1992; Hawkins and Hastie, 1990). The PSED screened 64,622 individuals from the U. S. mainland, aged 18 or older, who were randomly selected using random digit dial sampling (Gartner et al. 2004). The screening phone interviewer asked if the respondent, alone or with others, was currently trying to start a new business. The respondents who answered yes were further asked if: 1) they expected to have at least some ownership in the new ? rm; and 2) they actively tried to start the new ? rm in the past 12 months. Those 3,592 respondents who answered yes to both the above questions were considered eligible for the nascent entrepreneur interview. Of those 3,592 respondents, 3,158 (87%) volunteered to participate in future interviews.
A random subsample of 1,164 respondents was selected, contacted, and asked a further screening question to ensure that their current activity was considered a nascent effort rather than an infant business. Speci? cally, they were asked: ‘Has the start-up had a positive monthly cash ? ow that covers expenses and the ownermanager salaries for more than three months? ’ All respondents who reported that their nascent venturing activity achieved positive cash ? ows for more than three months at the time of the phone survey were removed, resulting in 830 nascent entrepreneurs who were subsequently phone interviewed in the ? st wave of the survey. 4 Nascent entrepreneurs chosen for this study were chosen as those attempting to start a new business, but not on behalf of an employer. These two criteria ensure that the expectations and characteristics observed relate speci? cally to the respondent rather than occur as a consequence of undertaking entrepreneurial activities on another person’s behalf. 5 Further, I exclude nascent activity related 4 The rationale for adopting this de? nition was to be consistent with other PSED research that has investigated nascent venture activity. Under the PSED sampling, the chievement of positive monthly cash ? ows for more than three months was considered an infant business and not a start-up effort (Gartner et al. , 2004). To examine the concern that including ventures that were generating revenues in the sample was spuriously affecting the ? ndings, I removed all start-up activity that had any revenues at the time of the ? rst survey. The ? ndings for the remaining subsample were the same as the full sample. 5 A single respondent approach was used as the research focus is on expectations of individuals undertaking nascent activity. 27 to the purchase or takeover of existing businesses and franchise or multilevel marketing initiatives. Given the focus of the research on expectations before business operation, I remove all respondents who reported that their nascent activity had achieved positive monthly cash ? ow at the time of the ? rst phone interview. These additional criteria reduced the sample size to 592 respondents. All expectations used in this study were solicited from this nascent entrepreneur sample during the remainder of the ? rst phone interview.
Finally, I limit the sample to the 386 respondents who reported at least one expectation and subsequently reported the associated ex post realization, noting that the multivariate tests vary in sample size due to missing values in either the dependent or independent variables. An examination of potential sample bias between those who responded in future phone interviews and those who did not, revealed no statistically signi? cant differences with regard to respondent gender, age, formal business planning, ? nancial projections, expectations of venture operation, or ? st-year venture sales and employees. Variables Forecast bias Three expectations of the nascent entrepreneur were empirically examined in this study, namely: 1) likelihood of the start-up becoming an operating business; 2) future sales in the ? rst full year of operation; and 3) number of future employees, excluding owners, at the end of the ? rst full year of operation. To examine the expectation of the start-up becoming an operational business, the respondents were asked: ‘On a scale of zero to one hundred, what is the likelihood that this business will be operating ? e years from now, regardless of who owns and operates the ? rm? ’ The likelihood obtained is referred to as the nascent entrepreneurs’ expectation of operation. For expectations of future sales, nascent entrepreneurs were asked: ‘What would you expect the total sales, revenues, or fees to be in the ? rst full year of Given nascent activity that involves a start-up team comprises individuals, each with an expectation of success that underlies his or her decision to be involved and remain involved in the nascent activity, the study ? ndings are applicable to both individual- and team-based nascent activity.
In 36 percent of the nascent ventures studied, the entire start-up team consisted of a single founder. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Copyright ? 2010 John Wiley & Sons, Ltd. 828 G. Cassar Table 1. Future status of nascent venture activity Future status description Not operating business after 60 months Given up on start activity Business was shut down before 60 months Team members operating business that was given up by respondent Operating business at last phone interview Total Status Not operating Not operating Not operating Operating Operating n 4 168 29 20 165 386 peration? ’ While expectations related to ? fth-year sales were also elicited, given the temporal period of the survey, information regarding actual sales related to the ? fth-year operations was not obtained by the PSED. For employment expectations, the entrepreneurs were asked: ‘By the end of the ? rst full year of operation, about how many full-time employees, not counting owners, do you expect to be working for pay at this new business? ’6 Entrepreneurs self-reported whether the startup was an operating business in follow-up phone interviews.
These interviews were conducted in three waves at a mean (median) 14 (13), 33 (32), and 56 (56) months following the initial screening interview. Respondents were asked to classify the current status of the start-up effort as a(n): 1) operating business; 2) active start-up; 3) inactive start-up; or 4) no longer worked on by anyone. All those who reported the existence of an operating business, and who did not abandon the operating business at a future date, were coded as operating.
All those who reported that the business was no longer worked on by anyone were coded as not operational unless other team members had successfully started the venture. Those respondents reporting active and inactive start-up activity more than ? ve years from the original phone interview were classi? ed as not operational. All remaining start-ups that were not operational or abandoned, and for which a phone interview was not undertaken at least 60 months after the original phone interview, were classi? ed as missing. Table 1 provides a descriptive summary of the 386 respondents’ status obtained from the follow-up phone nterviews. 7 The degree of overoptimism in venture operation was determined as the difference between the nascent entrepreneur’s likelihood of venture operation and an indicator variable (status) that equals ‘1’ if the venture is operational and ‘0’ otherwise, as follows: Operational overoptimism = expectation of operation ? status (1) Sales and employment information were also obtained in the three follow-up phone interviews. Given that sales and employment expectations were framed on the assumption that the venture was operational, these analyses are only undertaken for those start-ups that became operating businesses.
For sales, entrepreneurs who were involved in a start-up that had completed its ? rst full year of operation were asked what sales or revenue was achieved in the ? rst full year of operation. I conducted validity checks on the timing of the ventures’ launch and the timing of the responses and removed cases where the responses did not represent ? rst-year operations. 8 For numbers of employees, entrepreneurs who were involved in a start-up that employed a nonowner employee were asked to provide the number of nonowner employees hired in full-time and part-time capacities.
The degree of overoptimism, for sales and employment, was determined 7 The ? ndings are quantitatively similar if: 1) the 20 ventures started by other team members’ ? rms are excluded from the sample; 2) the operating ventures classi? cation included the 29 ventures that operated and subsequently reported in future phone interviews that the business was shut down; and/or 3) a further 53 ventures that were active or inactive after 48 months but not phone interviewed afterwards were classi? ed as not operational instead of missing. These validity checks included removing ventures that were not operational for more than one year and removing responses in follow-up phone interviews that related to the second year’s, rather than the ? rst year’s, operations. 6 In addition to this measure, a further employment expectation measure was calculated to incorporate expectations of part-time employees. This alternative expectation measure is the sum of expected full time employment plus half the expected part time employment. The results are invariant to the alternative expectation measure and consequently are not reported.
Copyright ? 2010 John Wiley & Sons, Ltd. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Nascent Entrepreneurial Expectations as a function of ex ante expectations (F) and actual realizations (A), as follows: (Sales or employment) overoptimism = F ? A/(F + A) (2) Controls 829 variable (‘1’ yes, ‘0’ otherwise) based on the question: ‘Have projected ? nancial statements, such as income and cash ? ow statements or break-even analysis, been developed? ’11 The use of de? ators reduces heterogeneity in the sales and employment overoptimism measures obtained. Importantly, all measures of forecast bias in this study, through the utilization of ex ante expectations in comparison with ex post realizations, overcome the use of perceptual measures of expectations, such as ‘success,’ using speci? c traits or characteristics to proxy for overoptimism, or inferring overoptimism using observed overly optimistic biases in other domains (Cooper et al. , 1988; Lowe and Ziedonis, 2006; Puri and Robinson, 2007). Further, the con? dential nature of the PSED removes gaming concerns associated with forecasts that are observable to venture stakeholders.
For example, without con? dentiality, entrepreneurs may upwardly bias observable forecasts or actual realizations to present themselves or the venture in a more favorable light to stakeholders, or downwardly bias actual realizations to lower tax or regulatory costs. Entrepreneur behavior The behaviors empirically examined in this study are formal business planning and ? nancial projections. A business plan was de? ned as something that ‘usually outlines the markets to be served, the products or services to be provided, the resources required—including money—and the expected growth and pro? for the new business. ’ Formal business planning was determined to be present when nascent entrepreneurs responded that they had a business plan and that the current form of the plan was ‘formally prepared. ’ Those responding in the af? rmative were coded ‘1’ and ‘0’ otherwise. 10 The undertaking of ? nancial projections was represented by a dichotomous 9 Other approaches to reduce heterogeneity in the forecast error included using just the ex ante expectation or actual realization as the de? tor and trimming, Winsorizing, or log transforming the forecast error. The de? ator utilized was chosen as the skewness and kurtosis of this measure was substantially closer to zero, which is critical for parametric-based testing. Notably, the study ? ndings are invariant to alternative forecast error speci? cations discussed above. 10 Alternative de? nitions, such as de? ning business planning to represent all business plans regardless of whether the plan The characteristics of the entrepreneur may be associated with the overoptimism of their forecasts.
For example, research has shown that males are generally more optimistic and more likely to overestimate future performance than females (Henry, 1994; Puri and Robinson, 2007). Further, cognitive ability, knowledge, or experience of the forecasting task should reduce the error or bias in predicting future outcomes (Forbes, 2005). I control for three characteristics based on the demographics of the nascent entrepreneur, namely, education, gender, and industry experience.
Education was operationalized by a variable denoting the highest level of educational attainment of the respondent, being: 1) up to high school; 2) technical/vocational; 3) some college including associates degrees; 4) bachelor’s degree; and 5) post college. Gender was a dichotomous variable, coded was ‘informally written’ or ‘formally prepared,’ or adopting a hierarchical-based planning measure, were not utilized as the formal nature of planning is more likely to induce behavior consistent with the inside view (Buehler et al. , 1994; Kahneman and Lovallo, 1993).
Further, the use of dichotomous predictor variables or manipulations is consistent with experimental research in expectation formation and empirical research on business planning (Hoch, 1984; Levi and Pryor, 1987; Sedor, 2002; Delmar and Shane, 2003; Shane and Delmar, 2004). 11 Another entrepreneurial behavior that may be associated with overoptimism is the obtaining of market information by the venture team. Speci? cally, the PSED asked ‘Has an effort been made to de? ne the market opportunities by talking with potential customers or getting information about the competition? All the results were replicated including this independent variable, which was coded as ‘1’ for yes and ‘0’ otherwise. In unreported results, no support was found for an association between market information and overoptimism. However, very limited variation in this market information variable in the sample may also have resulted in a failure to ? nd an association with overoptimism. To further investigate if information acquisition was an omitted correlated variable, I also included a variable from the PSED mail survey that asked respondents: ‘How much new information was acquired in recognizing this new business opportunity? ) none; 1) some; 3) a moderate amount; and 4) a great deal. ’ Given this variable is from the mail survey, it was not available for the full sample, with 272 out of 368 available for the operational overoptimism model and 38 out of 50 available for the sales overoptimism model. The information acquisition variable was not signi? cant when included in the presented speci? cations, suggesting that it is not an omitted correlated variable. Given the loss of sample size, these speci? cations are not reported. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Copyright ? 2010 John Wiley & Sons, Ltd. 830
G. Cassar Again, a yes response was coded ‘1’ and ‘0’ otherwise. In addition, a variable representing whether the start-up had received revenue was solicited from: ‘Has the new business received any money, income, or fees from the sale of goods or services? ’ Those that had received revenue were coded ‘1,’ otherwise ‘0’. as ‘1’ for female and ‘0’ for male. Industry experience is the number of years the respondent has worked in the industry of the nascent venture. Several venture organizing activities were utilized to control for the stage of development of the nascent venture (Shane and Delmar, 2004).
Venture organizing activities may be associated with a nascent entrepreneur’s expectations, as the more activities undertaken, the greater the knowledge of the venture and its viability. Further, as more activities are undertaken, the greater in? uence of selection bias in the remaining set of nascent entrepreneurs, as those nascent entrepreneurs who obtain con? rming (discon? rming) information about the venture’s viability or success will be more likely to continue (disband) the nascent activity. Product development was determined from the response to: ‘at what stage of development is the product or service this (start-up/new ? m) will be selling: 0) still in the idea stage or no work has been done on a product or service; 1) a model or procedure is being developed; 2) a prototype or procedure has been tested with customers; or 3) completed and ready for sale or delivery? ’ This product development variable was de? ated by three to provide a variable similar to the other independent variables that were bounded between 0 and 1. 12 The extent to which the start-up had begun marketing and promotion activities was obtained from the dichotomous response to: ‘Have marketing or promotional efforts been started (for the product or service this (start-up/new ? m) will be selling)? ’ A yes response was coded ‘1’ and ‘0’ otherwise. The extent to which the start-up had begun to obtain raw inputs was solicited from: ‘Have any raw materials, inventory, supplies, or components for the new (start-up/business) been purchased? ’ A yes response was coded ‘1’ and ‘0’ otherwise. The extent to which the start-up had begun to obtain capital inputs was solicited from: ‘Have any major items ($1,000 or more) like equipment, facilities, or property been purchased, leased, or rented for the new (start-up/business)? 12 The decision to specify product development as an ordinal variable, rather than a series of categorical variables, was to avoid harmful multicolinearity while controlling for the predicted positive association between extent of product development and the dependent variables. In unreported results, the empirical ? ndings were found to be quantitatively similar when product development was speci? ed as a series of categorical variables. RESULTS Descriptive statistics Table 2 presents the descriptive statistics and correlations of the study variables.
Examining the descriptive statistics reveals substantial overoptimism in nascent entrepreneurs. First, signi? cant overoptimism is observed in the nascent entrepreneurs’ expectations of venture operation. At the mean (median), nascent entrepreneurs believe there is an 81. 0 percent (90 percent) chance that their nascent activity will result in an operating venture; however, only 47. 8 percent of ventures actually achieve operation, resulting in a mean bias in the expectations of venture operation of 33. 2 percent (t = 12. 33, p < 0. 0001). 13,14 Second, of those that achieved operation and had available ? rst-year sales information, signi? ant overestimation of projected sales is observed, with 62 percent overestimating ? rst-year sales as opposed to 34 percent underestimating sales, resulting in a mean (median) sales overoptimism of 0. 17 (0. 25) (t = 2. 27, p < 0. 03). Third, employment expectations were also overly optimistic, with 46 percent overestimating employment after one year of operations as opposed to 28 percent underestimating employment, resulting in a mean employment overoptimism of 0. 15 (t = 1. 79, p < 0. 08). Overall, the evidence presented is consistent with nascent entrepreneurs being overly optimistic, and therefore supports Hypothesis 1. 3 While substantial overoptimism was observed, the expectation of operation and the eventual status of the venture was signi? cantly positively correlated in the sample (? = 0. 14, p < 0. 01). Therefore, those individuals that predicted their nascent activity would become an operating venture were more likely to become operating ventures, suggesting that respondents, while biased, were careful when providing operational expectations for their start-ups. 14 Signi? cant overoptimistic expectations (p < 0. 05) are observed for all major Standard Industrial Classi? ation divisions with greater than 15 observations. Copyright ? 2010 John Wiley & Sons, Ltd. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Nascent Entrepreneurial Expectations 0. 35?? 12 831 Cross-sectional differences in overoptimism Table 3 reports a series of regressions reporting differences in expectations of venture operation and operating overoptimism. Given the bounds on the dependent variables, the assumptions that underlie the error distributions of ordinary least squares are not appropriate; consequently, I use generalized method of moments (GMM). 5 For descriptive purposes, the ? rst three columns report the coef? cients from models predicting the expectations of operation by nascent entrepreneurs. These regressions are analogous to empirical research that has examined entrepreneurial expectations without considering ex post outcomes. Reporting both the nascent entrepreneurs’ likelihood of success and the bias associated with their expectations allows comparison between those behaviors or attributes that are associated with positive expectations and those that are associated with overly optimistic expectations.
Requiring nonmissing values for all independent variables reduced the analysis to 368 observations. The coef? cient on formal business planning is 8. 42 (p < 0. 001), suggesting that undertaking this activity signi? cantly increases the nascent entrepreneur’s expectation of operation by 8. 42 percent. However, whether formal planning is associated with overoptimism is determinant on how planning is associated with subsequent venture operation. Examining variations in operational overoptimism reveals that the coef? cient on formal business planning for observed overoptimism is consistently negative and insigni? cant.
Therefore, while formal business plan preparation increases the expectation that the venturing activity will lead to an operating venture, such preparation is not associated with overly optimistic expectations, as planning is also positively associated with the likelihood that the nascent activity will become an operating venture. The coef? cients on capital inputs and money received are negative, suggesting that individuals whose nascent activities have purchased signi? cant capital inputs for the venture or who have received money have a lower bias in their expectations of venture operation than those who do not.
However, while this bias is lower, 15 Regression techniques that incorporate censored dependent variables, such as Tobit regression, are not appropriate, as the assumptions of self-selection and censorship that underlie these models are not present for the dependent variables investigated in this study (Maddala, 1983; Wooldridge, 2003: 553). Median Descriptive statistics and correlations 385 33. 17?? 50 0. 17? 46 0. 15† 380 0. 16 385 0. 33 386 3. 22 386 0. 51 386 9. 18 376 0. 63 386 0. 57 386 0. 73 386 0. 53 386 0. 1 Operational overoptimism Sales overoptimism Employment overoptimism Formal business planning Financial projections Education Gender Industry experience Product development Promotion Raw inputs Capital inputs Money received 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. n Mean 52. 78 10. 00?? 0. 53 0. 25? 0. 57 0. 00 0. 37 0. 00 0. 47 0. 00 1. 28 3. 00 0. 50 1. 00 10. 34 5. 00 0. 38 0. 67 0. 50 1. 00 0. 45 1. 00 0. 50 1. 00 0. 49 0. 00 † p < 0. 10; ? p < 0. 05; ?? s. d. 0. 20 ? 0. 22 ? 0. 02 ? 0. 03 ? 0. 07 0. 05 ? 0. 05 ? 0. 13? ?0. 13? ?0. 11 ? 0. 15?? ?0. 18?? p < 0. 01. Copyright ? 010 John Wiley & Sons, Ltd. Variables Table 2. 1 0. 49? 0. 03 0. 37?? 0. 04 ? 0. 12 ? 0. 01 ? 0. 17 ? 0. 00 ? 0. 05 0. 09 ? 0. 21 2 ?0. 13 ? 0. 21 0. 38 ? 0. 15 0. 06 0. 08 0. 05 0. 03 ? 0. 13? ?0. 17?? 0. 06 ? 0. 10† 0. 31? 0. 16?? 0. 03 ? 0. 08 0. 05 0. 05 0. 03 0. 01 0. 03 0. 11? 0. 15 0. 06 ? 0. 05 ? 0. 11 ? 0. 01 0. 12 ? 0. 01 0. 07 0. 20 0. 01 0. 12 ? 0. 05 ? 0. 08 ? 0. 01 0. 07 0. 07 0. 03 0. 05 3 4 5 6 7 0. 09† 0. 07 0. 00 0. 07 0. 08 8 0. 40?? 0. 34?? 0. 23?? 0. 44?? 9 0. 36?? 0. 23?? 0. 35?? 0. 38?? 0. 29?? 10 11 Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj 832 G. Cassar Copyright ? 010 John Wiley & Sons, Ltd. Table 3. Results of GMM regression analysis of operational expectations and overoptimism Expectation of operation Model 1 Coeff. (2. 85) (2. 91) ? 1. 48 ? 4. 60 (7. 40) (6. 07) s. e. Coeff. s. e. Coeff. s. e. Coeff. s. e. Coeff. Model 2 Model 3 Model 4 Operational overoptimism Model 5 s. e. Model 6 Coeff. s. e. Variables 9. 90??? ?0. 44 Formal business planning Financial projections Education Gender Industry experience Product development Promotion Raw inputs Capital inputs Money received Intercept Wald ? 2 R2 n (1. 65) ? 2. 68? 1. 66 0. 25? 2. 63 5. 55 ? 2. 04 ? 0. 37 1. 3 82. 42??? 18. 17? 0. 05 368 (1. 12) (2. 59) (0. 11) (3. 93) (3. 03) (3. 14) (2. 98) (2. 93) (5. 05) 34. 13??? 0. 81 0. 00 368 ??? 79. 40??? 14. 63?? 0. 02 368 p < 0. 001 for two-tailed tests. 8. 42?? 0. 11 ? 2. 75? 2. 45 0. 20† 2. 69 5. 00 ? 1. 78 ? 0. 12 1. 48 81. 42??? 31. 99??? 0. 06 368 (2. 91) (3. 11) (1. 14) (2. 69) (0. 11) (3. 92) (3. 03) (3. 16) (2. 99) (2. 93) (5. 10) ?3. 01 6. 22 ? 0. 19 ? 1. 39 ? 3. 69 ? 4. 28 ? 10. 27† ? 14. 28? (3. 50) 58. 27??? 27. 35??? 0. 06 368 (2. 19) (5. 60) (0. 26) (8. 67) (6. 52) (7. 20) (6. 16) (6. 41) (10. 24) 0. 97 0. 22 ? 3. 03 6. 34 ? 0. 19 ? 1. 37 ? . 78 ? 4. 27 ? 10. 25 ? 14. 33? 58. 11??? 27. 37??? 0. 06 368 (7. 40) (6. 10) (2. 22) (5. 70) (0. 27) (8. 73) (6. 57) (7. 28) (6. 22) (6. 45) (10. 34) † p < 0. 10, ? p < 0. 05, ?? p < 0. 01, Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Nascent Entrepreneurial Expectations individuals who have undertaken these activities still appear to exhibit overoptimism. Overall, the evidence presented for operational overoptimism does not support either Hypothesis 2a or Hypothesis 2b. Table 4 presents the GMM regression analysis for those ventures that achieved operation for one year or more.
Given the relatively small number of observations available for both sales and employment, formal business planning and ? nancial projections are included in separate models to avoid harmful multicolinearity (Kennedy, 2003). 16,17 Model 1 shows that formal business planning is not associated with sales overoptimism. However, in Model 2, it is observed that the preparation of projected ? nancial statements leads to overly optimistic forecasts in regard to venture sales (p < 0. 01). This result is consistent with the inside view exacerbating overoptimism in expectations.
In particular, the undertaking of ? nancial projections encourages the nascent entrepreneur to adopt an inside view in regard to the ? nancial performance of the venture, resulting in overly optimistic ? nancial expectations. This result is consistent with Hypothesis 2b. Further, the above association is invariant to the inclusion of venture organizing activities in the full model presented in the third column. The results from Model 3 also show that nascent entrepreneurs whose ventures have received money from the sales of goods or services have signi? cantly less overoptimism in their sales expectations.
This demonstrates the bene? t of actually making sales in improving the rationality of ? nancial sales expectations. 18 16 The condition index is advocated to be the best indicator of multicolinearity and is calculated as the square root of the ratio of the largest to the smallest characteristic root of X’X (Belsley 1991; Kennedy 2003). The condition index on the full models presented in Table 4 with the inclusion of both formal business planning and ? nancial projections was above 25, which is suggestive of harmful colinearity (Belsley, Kuh, and Welsch, 1980).
An explanation for the greater colinearity observed in these models, beyond the reduction in sample size, is the stronger correlation between formal business planning and ? nancial projections for the subsample of ventures that achieved operation for one year or more. 17 An example of the reduction in sample size for the tests reported in Table 4 is provided for the sales overoptimism tests. Of the initial sample of 386, only 321 respondents provided their expectation of total sales for the ? rst full year of operation.
Subsequent actual sales information was obtained from 93 of these respondents in follow-up phone interviews, of which only 50 respondents provided actual sales speci? cally from the ? rst year of operation. 18 To investigate the possibility that reported overoptimism in sales was driven by pressure placed on entrepreneurs by 833 Interestingly, for employment, the relationship between overoptimism and ? nancial projections is no longer found, with the coef? cients on Models 5 and 6 actually being negative and insignificant. A possible explanation for the difference between sales and employment is that ? ancial projections are the monetary representation of the ? nancial outcomes of intended behavior. Therefore, the nexus between sales and ? nancial outcomes is more direct than the nexus between employment and ? nancial outcomes. For example, additional employment may be associated with increased scale, which may represent larger ? nancial returns, but it may also bring increased costs, which are associated with lower ? nancial returns to the entrepreneur. In comparison, sales are speci? cally modeled in ? nancial projections, and greater sales are directly associated with positive outcomes for the entrepreneur.
Therefore, even though both sales and employment expectations were highly positively correlated in the sample (? = 0. 49, p < 0. 001), the undertaking of ? nancial projections may have different implications for the beliefs of nascent entrepreneurs in regard to sales and employment. DISCUSSION Implications for theory and practice The study ? ndings are of interest for several reasons. First, it is the ? rst study to provide longitudinal evidence of the rationality of individuals’ expectations during the venture evaluation process.
The advantage of the longitudinal approach is that it elicits expectations of nascent entrepreneurs and compares them to observed outcomes, thereby overcoming methodological concerns of extant research evaluating the expectations of decision makers. Extant research has demonstrated that individuals who are optimistic about their life expectancy or those who have higher selfef? cacy are more likely to be an entrepreneur than investors to facilitate investment, both the sales expectations and overoptimism of sales were compared for those who had asked ? nancial institutions and other people for funding and those who had not.
There were no signi? cant differences between these two groups (t = 1. 02, p = 0. 31) and (t = 0. 32, p = 0. 75), suggesting that this concern was not driving the presence of overoptimism. It should also be noted that given the survey is con? dential, there should be no incentive on the entrepreneurs to arti? cially in? ate their reported expectation to satisfy outside stakeholders. Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Copyright ? 2010 John Wiley & Sons, Ltd. 834 G. Cassar Copyright ? 2010 John Wiley & Sons, Ltd. Table 4. Results of GMM regression analysis of one-year sales and employment veroptimism Sales overoptimism Model 1 Coeff. 0. 04 0. 40?? (0. 14) (0. 18) ? 0. 15 (0. 16) s. e. Coeff. s. e. Coeff. s. e. Coeff. s. e. Model 2 Model 3 Model 4 Employment overoptimism Model 5 Coeff. ?0. 23 s. e. (0. 16) Model 6 Coeff. s. e. Variables Formal business planning Financial projections Education Gender Industry experience Product development Promotion Raw inputs Capital inputs Money received Intercept Wald ? 2 R2 n (0. 08) 0. 00 7. 88?? 0. 14 50 (0. 09) 0. 44? ?0. 03 0. 00 ? 0. 00 ? 0. 14 0. 12 ? 0. 01 0. 15 ? 0. 30† 0. 21 23. 79?? 0. 24 50 (0. 18) (0. 07) (0. 16) (0. 01) (0. 27) (0. 15) (0. 2) (0. 24) (0. 16) (0. 33) 0. 21 0. 85 0. 02 46 (0. 12) ??? 0. 16† 0. 04 0. 00 50 p < 0. 001 for two-tailed tests. 0. 26 1. 99 0. 04 46 (0. 12) ?0. 20 0. 00 ? 0. 03 0. 01 ? 0. 14 0. 07 ? 0. 14 0. 41† ? 0. 10 0. 00 12. 99 0. 23 46 (0. 16) (0. 08) (0. 17) (0. 01) (0. 40) (0. 25) (0. 32) (0. 23) (0. 18) (0. 52) † p < 0. 10, ? p < 0. 05, ?? p < 0. 01, Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj Nascent Entrepreneurial Expectations to be in career employment (Puri and Robinson, 2007; Zhao, Seibert, and Hills, 2005). Similarly, extant research has found that individuals who are overcon? ent in the certainty of their knowledge in that they overestimate the correctness or precision of their beliefs in answering challenging questions or predicting the future, are more likely to be entrepreneurs and more likely to undertake product introductions that are pioneering or risky (Busenitz and Barney, 1997; Forbes, 2005; Simon and Houghton, 2003). However, these research approaches cannot answer whether, or to what extent, individuals entering venturing are overly optimistic in regard to the venturing activities they undertake, as this can only be achieved through comparing ex ante expectations with ex post outcomes.
The magnitude of overoptimism observed is substantial with nascent entrepreneurs on average overestimating the likelihood of venture operation by 33. 2 percent. By providing speci? c evidence demonstrating that individuals entering venturing overestimate the bene? ts from undertaking nascent activity and subsequent entrepreneurship, this research contributes to the literature that seeks understanding of the forces that underlie entry into and persistence of entrepreneurial behavior. Cooper et al. 1988) observed that individuals who recently became entrepreneurs reported their odds of business success to be substantially higher than historically observed and greater than the odds of success for other similar businesses. Given their research design, the authors posited that these beliefs were caused by post-decisional bolstering. In the current study setting, ex post decision bolstering cannot be a cause of overoptimism, as the expectations observed are elicited before the venture is operational. The evidence from this study cannot con? m or reject that bias related to ex post decision bolstering is present in entrepreneurs; however, it does suggest that the presence of overoptimistic beliefs in entrepreneurs is not contingent on the presence of ex post decision bolstering. Second, this study contributes to theory. I argue that the use of plans and projections encourages re? ection on the implementation of one’s actions through scenario representation, consistent with an inside view to forecasting. Consequently, I contribute by proposing that the use of plans and projections in themselves may affect decision maker beliefs and forecasting biases.
Additionally, by using plans and projections, this is the ? rst study Copyright ? 2010 John Wiley & Sons, Ltd. 835 to empirically investigate how the inside view can affect expectations in a ? eld setting, thereby testing the bounds of the theory and the signi? cance of its hypothesized effect on expectations. While there is evidence of the presence of such views in forecasting behavior within a population generally, little research has investigated differences of inside view adoption and forecasting biases within sample in similar settings (Buehler et al. , 1994).
For example, while researchers conjecture that individuals are generally too optimistic, it is unclear whether those who adopt a more inside view to forecasting are more likely to have forecasting biases. Further, there is a lack of understanding about how the inside view in? uences expectations and beliefs in ? eld settings, with the limited evidence related to the inside view based on experimental settings. By using a ? eld design, I address concerns that experimental designs may overstate the presence of bias in decision making, given their reliance on salient signals, lack of meaningful incentives, or insuf? ient task orientation (Schwarz, 1994). For example, it is dif? cult to recreate with complete faithfulness the nature, tasks, and environment that entrepreneurs’ experience during new venture creation. Therefore, the experimental manipulation of ? nancial projections use may exaggerate the role of the inside view, which may be limited by the contingencies of daily life in entrepreneurial endeavor (Buehler et al. , 1997). Experiments that monitor subjects over short temporal periods may also overstate the inside view’s in? uence, as its effect may dissipate over the course of extensive, long-term projects (Buehler et al. 1997). Further, incentivizing subjects to complete tasks using nominal monetary incentives may overstate the inside view’s in? uence on task completion given the greater economic incentives that individual’s face when creating an operating business (Camerer and Lovallo, 1999). This study demonstrates that differences in management behaviors consistent with the inside view appear to be an economically signi? cant in? uence on expectation formation and contribute to understanding differences in forecast overoptimism in the ? eld.
Further, distinguishing between different types of expectations revealed that ? nancial projections, while being associated with overly optimistic sales expectations, were not associated with overly optimistic employment expectations. Such a ? nding may be explained by the compatibility principle, Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj 836 G. Cassar that are associated with positive expectations from those that are associated with unjusti? ed or overly optimistic expectations. More generally, the study ? ndings are of interest to decision makers.
While research has examined the forecast accuracy of decision makers, there is limited understanding of how particular behaviors in? uence forecasting bias (Durand, 2003). For example, there is little evidence on whether commonly used management practices moderate the level of biased or extreme prediction. Consequently, this study provides an important link between observed psychological bias and management practices, namely ? nancial projections. This documented relationship is particularly important, given the fundamental role that ? nancial projections play in allowing decision makers to structure expectations and to cope with the future.
The evidence from this study suggests that the same management activities that entrepreneurs rely on to cope with uncertainty appear to be exacerbating individuals to hold overoptimistic expectations. Acknowledging how management practices bias expectations may allow decision makers to use organizational or decision-making controls to reduce this in? uence. For example, generating reasons why the planned outcome may not be achieved, or consciously relating past experiences to the forecasting task at hand, are approaches individuals can take to reduce overly optimistic or overcon? ent forecasts (Buehler et al. , 1994; Koriat, Lichtenstein, and Fischhoff, 1980). In a group setting, the use of con? ict-based approaches whereby members critique proposals and make counter arguments may result in more realistic expectations (Schweiger, Sandberg, and Ragan, 1986). Another decision-making de-biasing technique is adoption of the outside view. Through incorporating relevant information that decision makers may otherwise neglect, the outside view should avoid the overoptimistic tendencies from scenario thinking (Kahneman and Lovallo, 1993; Kahneman and Tversky, 1979).
Speci? cally, with knowledge of the outcomes from similar previous endeavors, decision makers’ expectations can be formed through evaluation of how their present endeavor compares to previous endeavors. Within this evaluation, decision makers should also de-bias their expectations by factoring out historically observed forecasting errors made by similar endeavors and control for overoptimistic Strat. Mgmt. J. , 31: 822–840 (2010) DOI: 10. 1002/smj which has been observed in experimental settings, whereby the weight of any input component is enhanced by its compatibility with the output.
For example, when subjects were asked to provide a numerical output, such as a dollar amount, numerical data had a greater impact on their response than other data that was less compatible in content, scale, or display (Tversky, Sattath, and Slovic, 1988). As ? nancial projections are the monetary representation of the outcomes of intended behavior, they may be more likely to encourage re? ection on the causal events that lead to the ? nancial outcomes. Further, employment outcomes are more ambiguous in regard to the ? nancial performance of a venture than are sales outcomes.
For example, greater sales and growth may lead to greater need for the venture to employ more people; however, more employees also represent greater costs, thereby reducing the ? nancial returns from the venture. In other words, the undertaking of ? nancial projections is associated with overly optimistic expectations for ? nancial outcomes, but not necessarily for other outcomes. The varying implications for the expectations of decision makers from different management behaviors highlights the importance of the context of the inside view and the types of plans and projections used in shaping expectations.
Third, the study provides empirical evidence for differences in forecast overoptimism distinguishing between ex ante beliefs and ex post outcomes. In particular, the evidence suggests that formal