The Second Time Around? Repeat Entrepreneurs from MIT By: Charles E. Eesley and Edward B. Roberts October 25, 2006 Abstract In this paper we explore the factors that condition the likelihood that an entrepreneur starts a second firm. We use data from survey responses of 1,789 entrepreneurs to examine firm founding behavior. Results indicate that multiple entrepreneurs differ from single-firm entrepreneurs in certain demographic and educational characteristics prior to starting a first firm. The phenomenon of graduates embarking on careers of multiple entrepreneurship appears to be growing over time. The results also show that the first firms of eventual multiple entrepreneurs differ from the first firms of single-firm only entrepreneurs. The paper indicates that those entrepreneurs with the highest probability of starting a second firm have greater time and access to financial resources to undertake a new venture. Starting a first firm sooner after graduation, being divorced, the first firm being acquired, and raising initial capital for the first firm from angel investors all increase the probability that the entrepreneur will start a second firm. Authors are listed in alphabetical order. MIT Sloan School of Management, 50 Memorial Drive, Cambridge, MA 02142. eesley@mit.edu; eroberts@mit.edu 1 I. Introduction A growing literature in strategy and economics has noted that an important source of new entrants is incumbent firms in the same industry (Klepper 2001; Gompers, Lerner, and Scharfstein, 2005). However, these studies have developed and emphasized theories regarding employees leaving incumbent firms to start new ventures. Various lenses, such as agency theory, organizational capability theories, employee learning theories, and evolutionary theories have been used to evaluate why employees decide to leave their firms to spin-off new firms. Most of these theories emphasize some sort of conflict between the spin-off and the incumbent’s top management. On the other hand, in the finance literature, a growing group of scholars have examined the determinants of CEO turnover. However, these authors typically focus on forced CEO turnover in relation to firm and industry performance. CEOs who voluntarily leave and what they do after leaving are seldom studied (for important exceptions see Wasserman 2003; Bertrand and Schoar 2003). At the intersection is a relatively important gap in our understanding-- the founder or CEO who is pushed out or voluntarily leaves a firm and decides to start a new firm. The existing theories of spin-off activity (with the possible exception of evolutionary-based theories) do not explain or fit this phenomenon well, and the existing work on CEO turnover with an emphasis on corporate governance of large firms does not shed much light on the phenomenon either. As a result, it has proven difficult to determine how prior founding experience should be interpreted (i.e. as a measure of risk aversion, learning, or reduced asymmetric information on quality for investors). The importance of the entrepreneur who starts multiple firms over the course of a lifetime has been recognized in the academic literature as a potentially widespread aspect of new firm 2 creation for at least 35 years (Cooper, 1970). However, in relation to its likely significance for economic growth and for the field of entrepreneurship the phenomenon has been understudied (see Table 1). With some recent exceptions (Sarasvathy & Monon, 2005), almost no theoretical work has been done (Westhead, 1998). As a result, the few empirical studies use different definitions, populations, and control variables leaving little consensus on foundational questions about how important or how widespread is this phenomenon in the U.S. or internationally. This deficiency is especially true in regard to companies formed to exploit new technology. The purpose of this paper is to advance our understanding of what factors condition the likelihood that a founder will leave a firm and start a new firm. Is it differences in underlying individual demographic characteristics or differences in characteristics of the first firm experience or performance that motivate novice founders to initiate a new firm founding. To answer these questions, we use a founder-firm matched panel data set, where we track over time founders across different firms that they start or choose to remain with. Our cross-industry data also allow us to examine how industry characteristics may affect how likely a founder will be to try founding a second new firm. II. Background Literature / Theory Especially compared to the early years when entrepreneurs were seen as a homogeneous group, a number of different groupings of entrepreneurs has emerged in recent years. As Hsu (2006b) notes, an important recent theme in the literature is heterogeneity among entrants prior to the development of the venture, especially in terms of the experience of the founder or founding team. Compared to the literature on differences between entrepreneurs and nonentrepreneurs, scholarly attention to the heterogeneity among entrepreneurs has been very limited until recently. 3 Prior literature has examined the determinants of self-employment. Carroll and Mosakowski (1987) show that the transition to self-employment depends on factors including family background and prior experiences with self-employment. Our research differs in its focus on high technology entrepreneurs and their firms. As pointed out earlier (Shane & Khurana, 2003), self-employment does not necessarily involve the founding of a new firm, such as when an individual does independent consulting or contract work. Further, firms started to commercialize new technologies tend to be larger and have a strong impact on economic productivity. Therefore repeat entrepreneurs who focus on technology represent a particularly important and relatively unexplored area of study. Wasserman (2003) examines Founder-CEO succession in 202 venture-backed Internet start-ups. His study does not look explicitly at what Founder-CEOs do once replaced or after leaving the firm. He finds that many Founder-CEOs remain in the firm, even when they are replaced as CEO. Of importance for the current study, he finds that the likelihood of a FounderCEO being replaced increases with the achievement of critical milestones in building the company. Unlike the previous studies looking at the effects of prior career experiences or prior firm founding on the likelihood of subsequent firm founding, the present study examines individual and firm-level factors related to the transition from an active founding role in one firm to starting a second firm. In addition, our research takes advantage of data on the timing of the entire founding history and on the entrepreneur's personal success. III. THEORY AND HYPOTHESES In this paper we develop and test a theory of entrepreneurial exit and then re-entry into entrepreneurship. Through entrepreneurial experience the founder builds up social and financial 4 capital in addition to personal experience and human capital. As these three forms of capital are raised by the entrepreneur, he or she may choose to invest them in continuing and advancing the entrepreneurial career. If the entrepreneur chooses to “reinvest” these forms of capital produced by the first firm, one possibility is that the capital is invested in such a way as to maximize the number of high quality business opportunities available to the entrepreneur to pursue at minimal cost to the entrepreneur in terms of time, effort and/or money. This process of reinvesting the fruits of entrepreneurship can be conceptualized as a continuum from the initial startup to repeat entrepreneurship. The entrepreneur may be able to become involved in new firms because the older “new” firm has progressed to a more stable state where he/she does not need to worry about it as much and others, perhaps more experienced managers have taken over the day to day operations. While we portray the progress as a continuum, obviously steps could be skipped, perhaps indicating a more rapid personal development as a successful multiple entrepreneur. A nascent entrepreneur begins with relatively little human, social, and financial capital and access to many ideas, but probably only a few high quality entrepreneurial opportunities. If the entrepreneur succeeds in the first start-up and decides to start a second firm, the various forms of capital raised through the first startup will presumably tend to make this second effort somewhat easier. The buildup of human, social and financial capital with the entrepreneur is likely to occur with a substantial lag after the founding of the initial firm. The first firm of a nascent entrepreneur requires an incredible amount of effort, time and energy. As the firm gains its operational experience, the entrepreneur has increasing opportunity to reflect and begins to accumulate skills related to the entrepreneurial process along with important social contacts and, slowly, financial capital as well. As a result of the startup and operation of the first firm, the 5 entrepreneur may see various business opportunities or shadow options (McGrath, 1999) that he or she may not have been able or willing to pursue at the time, but which can now serve as the basis for a subsequent firm. An emerging literature has been developing to apply real options theory to strategy as a way of increasing flexibility (Bowman & Hurry, 1993; McGrath, 1996). As McGrath (1996, p. 103) points out, among several factors that may relate to the likelihood of recognizing viable new opportunities is previous entrepreneurial experience. While McGrath and MacMillan (2000) suggest that entrepreneurs with previous founding experience have developed a “mindset” or the ability and competence to pursue only the best opportunities, this may not be the entire story. It may be that while novice entrepreneurs only pursue what they see as the very best opportunities, repeat entrepreneurs may take a longer view. Repeat entrepreneurs may pursue a wider range of opportunities in round 1, allowing them to see many shadow options and choose only the best opportunities to invest in as real options for greater performance in round 2. Having reached the level of repeat entrepreneur, the individual will have built up higher levels of all types of capital and, also through the operation of the previous businesses, customer contacts or social contacts will have perceived a number of other business opportunities. At any point in this progression, as the entrepreneur builds sufficient financial capital, he or she may invest a portion of assets in early-stage ventures. In the investment community jargon, investments made by wealthy individuals have become known as angel investing and such individuals have been coined as angel investors or informal investors to distinguish them from venture capital institutions (Freear, Sohl, & Wetzel, 1994; Wetzel, 1983). These deals may be undertaken serially or in parallel and offer the entrepreneur multiple opportunities for entering into the venture as an active entrepreneur as well. In each stage of the continuum (from the 6 bottom left of Figure 2 to the top right), a greater number of business opportunities will arise (or perhaps be sought out by the entrepreneur) and will be available for pursuit or active involvement as a founder. Further, these opportunities become available to the entrepreneur at smaller and smaller cost in terms of relative time and effort necessary to encounter them and commercialize them. Indeed, other scholars have asserted that prior startup experience reduces the costs of entrepreneurial activity and increases the probability of acting on opportunities (Carroll & Mosakowski, 1987). Depending on personal preferences and style, an entrepreneur may not choose to reinvest her or his capital in further entrepreneurship. Especially if the first firm is still in operation and he is happy with his role there, the founder may choose to stay with the first firm and not continue down the path of multiple entrepreneurship. Our point is simply that this option becomes open to him or her and is easier to achieve after a first founding. One may also think of the progression described above as shifting into a role of opportunity recognition and generation, of “rainmaker” so to speak in the entrepreneurial process. The entrepreneur may then be in less of a day to day operational role in any one firm, but is engaged more fully in a business idea generation and evaluation role. These rare individuals may be playing a very valuable role in the entrepreneurial processes within one or more industries. Why Start a Second Firm? When asking who starts multiple firms and what conditions this choice, we are essentially asking about what motivates or enables an entrepreneur to start a second firm. At the individual level and at the first firm level, a number of factors may lead to the choice to start another firm. If starting a second firm is motivated by expectations of the returns to reinvesting financial, human and social capital back into further entrepreneurial activities, then factors that increase the 7 expected return should result in a higher likelihood for starting a new firm. Similarly, factors which decrease the expected return should decrease the probability the entrepreneur will decide to undertake a second firm. The “expected returns” to investing in entrepreneurial activities may be financial or non-financial. Nonetheless, the first firm experience is likely to condition the level of non-financial rewards anticipated as well as the relative importance of these compared to financial returns. During the first firm experience, the entrepreneur may discover, in fact, whether the non-pecuniary benefits to entrepreneurship make it more attractive than returning to regular employment (Barton 2002). First, simply having the time and energy to start another firm may be an important factor. If the entrepreneur is quite old, then he or she may expect that the fruits of a second firm creation will not accrue with enough lifetime left to enjoy them. Hypothesis 1: Entrepreneurs who are younger at the time of their first firm founding or otherwise have more time to spend on entrepreneurship will be more likely to start a second firm. Another possibility is that greater confidence leads to a higher likelihood of starting a subsequent firm. This would lead one to hypothesize that the better the outcome of the first firm or the better its performance relative to the entrepreneur’s expectations, the more likely the entrepreneur will undertake another start-up. Of course it is difficult to disentangle the effect that better first firm performance would have on confidence and the expectations of future success from the effect success has on the level of financial and social capital. The arguments outlined previously as to human, social and financial capital provide strong support that a successful first venture should increase the likelihood of starting a subsequent firm. Albeit, a very successful first venture could last so long as to diminish the time available for a second firm, or generate so much wealth or personal satisfaction as to diminish the motivation to do it again. 8 Higher financial performance of a first firm may also mean that the firm has progressed to a more steady state where the founder does not need to worry about its survival and has time to think about a new start-up. Either way would suggest that higher success would generally lead to more second firms. This is especially true since performance below a reasonable return to the founder’s human capital or attractive outside options for regular employment should lead to an exit from entrepreneurship (Gimeno, Folta, Cooper, & Woo, 1997). However, very high levels of success, especially financial success could result in the entrepreneur simply retiring and thus lower the chances of a second firm founding. Moderate levels of success should lead to higher probabilities of a second founding, since if the first firm was not particularly successful then the entrepreneur would have had a taste of success and may desire even more in a subsequent firm. Hypothesis 2: We anticipate an inverse U-shaped curve relating success of a prior new venture and the likelihood that a subsequent new venture will be undertaken. If it is true that the financial capital of the entrepreneur increases with multiple firm foundings, then we should see the net worth of the founders increasing with the number of prior firms. This statement may seem obvious. However, if multiple entrepreneurs are typically starting a series of small firms that are relatively unsuccessful, then the result would not necessarily hold. Hypothesis 3: The performance of the entrepreneur in terms of net worth (financial capital) will be higher with a greater number of prior firms. Alsos (1998) has investigated differences in firm formation processes between multiple entrepreneurs and first-time entrepreneurs. Few significant differences were found between novices and repeat entrepreneurs. Differences in motivation were found with repeat entrepreneurs more likely to indicate motivations related to instrumentality of wealth. Novice 9 entrepreneurs were more likely to stress a need for independence. This study could not discern whether the repeat entrepreneurs’ motivations changed between the first firm and later firms or whether these results may indicate that the multiple entrepreneur was destined to start multiple firms even prior to starting his/her first firm. Hypothesis 4a: Repeat entrepreneurs differ from single-firm entrepreneurs even prior to starting the first firm in terms of demographic variables. One motivation for starting a second firm may be that the entrepreneur is highly innovative and does not have the freedom to innovate as widely as he would like within the confines of the prior firm. This idea is supported by Klepper and Thompson’s (2006) model of spinoff activity resulting from disagreements within the existing firm. In this case, it is likely that those whose backgrounds suggest they may be highly innovative would be more likely to initiate a career of multiple foundings. Hypothesis 4b: Entrepreneurs with a background in the sciences or engineering will be more likely to start a second firm. At the first firm level, the tendency to be highly innovative should manifest in founding first firms that hold patents as opposed to first firms which do not hold patents. Hypothesis 4c: Entrepreneurs with patents in their first firm will be more likely to found a second firm. Aspects of the first firm may make it more likely for the founder to start a second firm. Alternatively, it is difficult to dismiss the possibility that the causation may be reversed, thus we propose the joint hypotheses 5a and 5b. Hypothesis 5a: Certain aspects of the first firm founding will condition the likelihood that a second entrepreneurial venture is started. Hypothesis 5b: Repeat entrepreneurs differ from single-firm entrepreneurs in the characteristics of the first firms they found. 10 Many prior studies have argued that access to financial capital is a limitation to entry, even if the level of personal assets are not predictive of who will enter self-employment (see Acs & Audretsch, 1989; Dunn & Holtz-Eakin, 2000 and others). Entrepreneurs entering industries with requirements for high amounts of initial capital should face a higher hurdle to start a second firm as well, resulting in decreased rates of second firm foundings. Alternatively, those founders in industries with low requirements for initial capital will be more likely to be able to fund a new venture themselves from the profits of the first firm. Hypothesis 5c: Entrepreneurs starting first firms in industries requiring high amounts of initial capital will be less likely to start multiple firms. In addition, if financial capital is a constraint for entrepreneurs, then those who have been successful in fundraising and in building links to sources of initial capital should have an advantage in raising money in the future. Indeed, Hsu (2006a) finds that experienced entrepreneurs have quicker access to venture capital than inexperienced founders. Hypothesis 5d: Entrepreneurs successful at raising initial capital from venture capital firms or angel investors will be more likely to start a second firm. Simply having access to fellow entrepreneurs who are willing and qualified to undertake the risks of a new start-up is likely to be a constraint for some entrepreneurs interested in starting a second firm. Therefore, we suggest that entrepreneurs who built larger teams for their first start-up should have more options for cofounders for a second start-up. This may be true in part because entrepreneurs with larger founder teams have been found to be more successful than those who are sole founders or have small teams (Roberts, 1991). Consequently, we hypothesize that holding the success of the first firm constant (controlling for first firm performance), those starting the first firm with a higher number of cofounders will be more likely to start a second firm. 11 Hypothesis 5e: All else equal, entrepreneurs who started a first firm with a higher number of cofounders will be more likely to start a second firm. IV. Methods Data and Sample The data come from a survey administered in 2003 to all living MIT alumni who had previously self-identified as founding at least one venture. The dataset of MIT alumni contains over 40,000 records including basic information on date of birth, country of citizenship, gender, major at MIT, highest attained degree, and new venture founding history. Out of 7,798 alumni indicating that they had founded a company, 2,111 founder surveys were completed, representing a response rate of 27.1%. Out of 3,156 alumni indicating that they had started multiple companies, 1,004 completed the survey for a multi-founder response rate of 31.8%. A total of 1,107 single-firm founders responded to the survey giving a 21.8% response rate out of the 5,086 single-firm alumni founders. Some of these 1,107 single-firm founders may later go on to become multiple entrepreneurs, however we can isolate 728 single-firm founders over the age of 59 (95% of firms are started by entrepreneurs under the age of 60) and be reasonably confident in having eliminated such right-side censoring problems. One of the key features of this dataset is its long time horizon in the cross section (1930s-2001) which allows us to analyze entire entrepreneurial careers of a large number of founders. In addition, this represents what we believe is one of the first datasets which includes both orderings and the timings of the firms of multiple entrepreneurs, but also individual level performance data in addition to firm level performance measures. To address the distinction between self employment and new ventures, we eliminate any 12 firms identified by the respondents as architecture firms1, consulting firms, and law/miscellaneous firms from the database.2 Eliminating these firms brings the sample size to 1,789 firms. Measures Dependent Variables. Second firm founded indicates the year that the entrepreneur’s second firm was founded and is our primary dependent variable. Independent Variables. The independent variables pertain to: 1) the individual level, and 2) characteristics of the start-up process and outcomes of the first firm. Academic major is categorized according to degrees (Bachelor’s, Master’s, or Doctorate) in engineering, management, social science and the humanities, architecture, and natural science. We also have data on the highest degree attained by the founders. Country of citizenship is categorized into Latin America, Asia, Europe, the Middle East, Africa, and North America.3 Graduation year indicates the year of the founder’s highest degree from MIT. First start-up founded (or second, and third) is the year of the founding of the first (or later) firm. The entrepreneurs indicated on the survey their current marital status. To measure net worth, the founders were asked to indicate their net worth within five categories ranging from $0 to $100,000, $100,000 to $500,000, $500,000 to $1 million, $1 million to $10 million and over $10 million. The mean in each range was used and although $10 million is likely a low upper range, this was used for the fifth and highest category. Note that the net worth data are not ex ante, but rather the cumulative There is a very high rate of “serial” entrepreneurship for architects, but this is likely due to the flexible partnership nature of the business and low technology nature of these firms. 2 The analyses were run with a cutoff of 10 employees or more, 5 employees or more, and all firms included. Since the results and conclusions were largely similar for all three, only results for ‘all firms’ are reported. The only variable that changes significance is management degree which becomes significant when firms below 5 employees are included. The most likely interpretation is that this is capturing some services/consulting firms that were selfcategorized under a different industry and thus not eliminated. 3 We use responses for country of citizenship since only 182 of the founders provided information on country of origin, compared to 366 with information on country of citizenship. In only 14 cases does the information on country of origin differ from the corresponding country of citizenship data. 1 13 net worth evaluated as of the time of the survey (2003). We also use data on the total lifetime of firms founded by the entrepreneur. Respondents indicated a coarse industry segment within which the firm was founded.4 The industry categories represent check boxes on the survey in which the entrepreneurs were asked to classify their firms. While these might be thought of as control variables, there may be relevant differences within these areas that influence how easy it is to start multiple ventures within that industry grouping. Lag to first firm measures the lag in years from graduation to the founding of the entrepreneur’s first firm. Held patents is coded as a 0 if the firm never patented and as a 1 if the firm was granted patents by the USPTO. # Cofounders is a count of the number of cofounders listed for the first firm. Team from Social Activities and Team from Work indicate the source of the co-founding team and Team from Research is the excluded category in the regression models. Idea from Research, Idea from Work, and Idea from Networking are 0/1 variables capturing the original source of the idea, and Idea from Extracurricular Activities is the excluded category. Respondents indicated all sources of initial capital as well as the total amount (initial capital) for the firm including VC funded and Angel funded. Acquired and Public capture firm outcomes (as of 2003). Employees and Revenues are the number of employees at the firm and revenues (for a specific year of operation). A first founding during an economic recession might force the entrepreneur to be careful about costs and be more disciplined but the struggle might make him/her less likely to start a second firm. A first firm founding during an economic expansion might bolster the founder’s confidence and make subsequent startups easier. Recession Year is a dummy variable obtained from the National Bureau of Economic Research (NBER) chronology of the U.S. business cycle. Finally, operating years is a count of the 4 These include aerospace, architecture, biomedical, chemicals, consumer products, consulting, electronics, energy, finance, law/miscellaneous, machine tools, publishing, software, telecommunications as well as other services, and other manufacturing. 14 number of years the firm has been in operation or the number of years from founding to closing. The variable log revenues is the natural log of the revenue for the firm as reported by the entrepreneur. The respondents were asked to report revenues for the most recent fiscal year available; therefore these firms are not necessarily still in operation. Since revenues show an upward trend over the age of the startup, we normalize the revenue data by dividing by the age of the firm at the year for which revenues are reported. We adjust for inflation (2001 $) and because of the skewed distribution take the natural log of this measure for our dependent variable in the multivariate analysis. We also look at acquisitions. Founders were asked to indicate what role they played at the time of the founding of the company and what role they play now.5 If they were not currently active with the firm, founders were asked to indicate whether they were: consulting, involved with another startup, employed elsewhere, retired, working in another company that they had founded, or other. V. Results We first present descriptive statistics followed by the regression results. Figure 1 shows a histogram of the number of 1st firms, 2nd firms, 3rd firms, 4th firms, and 5th firms and greater (subsequently referred to as graphed by firm ranking) by their founding year. First firms are the most prevalent and the number of firm foundings increases over the years. Separate from any trends, we expect this increase due to our panel data set and the fact that each year adds another year of graduates at risk for entering entrepreneurship. Figure 3 plots the percentage of founders starting a second firm by deciles of first firm revenues (divided by the operating year of the firm). The line appears to be concave and begins decreasing again for the top decile of revenues. Figure 4 shows the percentage of novice founders and repeat founders with Bachelor’s, Master’s, 5 Possible roles include President/CEO, Chairman (if not also CEO), Chief Operating Officer, Chief Scientist/Chief Technology Officer/Chief of R&D, Chief Finance Officer, Chief Information Officer, Vice President, Outside Director, Consultant or Advisor, Other employee. 15 and Doctorate degrees. Proportionately more of the repeat founders are non-U.S. citizens and a slightly higher proportion of the repeat entrepreneurs hold Master’s degrees. Figure 5 demonstrates that novice entrepreneurs differ significantly from the other categories of founders (p<0.001) in being older and having a longer lag from graduation to the founding of their first firms. Repeat entrepreneurs enter entrepreneurship much sooner, which likely gives them time for subsequent firms. Next, we move from the individual level to examining characteristics of the first firm experience. Figure 6 shows that the types of entrepreneurs show differences in their first startup experiences in terms of the amount of capital raised and the level of revenues achieved. Repeat founders raise significantly higher amounts of initial capital than novice-only entrepreneurs (p<0.05). Figure 7 graphs that more of the repeat founders had first firms which were acquired or went out of business. In Figure 8 we see that higher levels of net worth are associated with repeat entrepreneurs. Figure 9 shows the roles the entrepreneurs played in their firms at founding. Figure 10 shows the roles that the founders currently play in their firms. Over 60% of the founders were President/CEO of the firm at founding, however the most likely roles currently are COO or not active. Table 1 contains the estimated economic impact of the novice and repeat entrepreneurs in terms of firm foundings, revenues and employees. Only the revenues and employees are estimated since we do not currently have these figures for the prior firms of the multiple entrepreneurs. Therefore we extrapolate by multiplying the number of these prior firms by the mean revenues and mean employees for those firms founded by that category of entrepreneur in our sample. These estimates are likely to be quite conservative relative to the (real data) for novice entrepreneurs since we included novice entrepreneurs of all ages and some of the repeat 16 entrepreneurs may start still more firms.6 Thus, the percentage of novice founders is surely an overestimate since many of these novice entrepreneurs will go on to found multiple firms over their entrepreneurial careers. From these estimates we see that repeat entrepreneurs have a substantial economic impact relative to their percentage of total entrepreneurs. Table 2 presents by their decade of graduation the number of entrepreneurs founding first firms all the way up to 11th firms. The row labeled “% serial” is the percentage of founders from that decade of graduation who have started more than one firm. Across the decades, from 3050% of the founders have chosen to found multiple start-ups.7 The decrease in the percentage from the 1980s on is likely due to right-censoring in that the most recent graduates have not yet had time to start a second firm and may do so in the future. Table 3 examines the industry distribution across firm rank. Here we include architecture, consulting, and law firms in order to show the complete data. Elsewhere these firms are deleted from the sample because we are interested in new firm creation and the flexible nature of partnerships or self-employment in architecture, law, and consulting is likely to be of a different nature or motivated by entirely different dynamics. Software and electronics represent the largest proportion of the sample and we find that across multiple ventures the percentages in each industry grouping remain nearly constant. In Table 4 we present summary statistics and variable definitions. As expected second and third firms are on average founded in more recent years than first firms. We use Cox (1972) hazard regression models for our analyses. First, the model estimates the probability of founding a second firm in a given year conditional on not having founded a 6 One approach to this problem would be to perform these calculations only for entrepreneurs over age 60 who are unlikely to found more firms. Since the phenomenon of multiple entrepreneurship is increasing with each graduation year, we do not take this approach as it would produce a drastic underestimate by eliminating many of the multiple entrepreneurs and their firms. 7 This statistic uses the total number of firms the entrepreneur claimed to have started. For the remainder of the analyses, we use the number of firms that that they listed company names and founding dates for in the survey. The latter is more reliable and conservative but was capped by the survey instrument at five. 17 (second) firm up until that time period. Thus, it adjusts for the right-censoring of the data by explicitly taking the timing of events into account. Second, the model is semi-parametric, which allows us to estimate the impact of independent variables on the hazard of founding a firm while leaving the baseline hazard function unspecified. Founders start being “at risk” of founding a second firm at the time of their first firm founding, and a “failure” event occurs the year the individual founds a subsequent firm (otherwise, the founding year is considered censored for that individual as of the year 2003).8 Reported coefficients are hazard ratios and are in comparison to the omitted group. Values above 1.0 represent increases in the hazard of founding a subsequent firm and values below 1.0 represent decreases in the hazard. Statistically significant estimates are indicated through asterisks. The specification of the Cox (1972) model is as follows: t | X o (t ) exp X (1) where the vector X includes our founder and first firm experience characteristics. t | X is the rate at which founders will start a second firm at any particular date, given that they have not founded a second firm up until that point in time. Equation (1) specifies the hazard rate as the product of two components: a function of the spell length (i.e. delay time since founding the first firm), o (t ) or baseline hazard, and a function of the observable individual and first firm characteristics, denoted by the vector X. Thus, each of our models predicts the likelihood that an entrepreneur will found a second firm. The Cox nonparametric estimation allows us to estimate without needing to make a distributional assumption about o (t ) . 8 We have also run these analyses with individuals becoming at risk of founding a firm at their year of university graduation. The results are stronger than those reported here, which is easy to understand since the same number of firm founding events are being predicted over a shorter time horizon (graduation versus birth year). We choose to report the more conservative birth year entry estimates because we are agnostic as to when an individual might start a venture. As well, the graduation event depends on the degree the individual received. 18 Table 5 shows the regression results using the year that a second start-up was founded as the dependent variable and characteristics of the founder as the independent variables. In Model 5-1 we find that more recent graduates, males, and those who are divorced are significantly more likely to found a second firm. Model 5-2 tests the academic majors and finds that those with management degrees are more likely to start a second firm. Model 5-3 tests differences in country of citizenship and finds that citizens from Latin American and Asian countries are more likely to start a second firm. In the fully specified model (5-4) the graduation year, gender, divorced, management degree, Latin American, and Asian citizen variables remain significant. The regressions in Table 6 use the year a second start-up was founded as the dependent variable and show the effects of first firm characteristics as the independent variables. In Model 6-1 variables capturing the initial conditions at founding are included. We find that a longer lag from graduation to the start of the first firm decreases the likelihood by 5.2% (for each year) that a second start-up will be founded. When the first firm was started during a period of economic recession, the entrepreneur is not less likely to attempt a second firm. Each additional cofounder involved in the first experience increases the chance of a second firm by 24.2% (p<0.001). If the source of the first firm idea came from research, then this is associated with a decrease in the chance that a second start-up will be attempted. The industry average R&D to sales ratio is significant and indicates that those in higher R&D intensity industries are 32.1% more likely to do another start-up. Model 6-2 examines the fundraising and outcomes of the first firm to see if these play a role in the decision to start a second firm. If the first firm was acquired, this is associated with a 50.1% higher likelihood of a second start-up attempt. While we are not confident the IPO coefficient is different from zero, we expect that if the IPO valuation was taken into account then 19 we may find significant effects. The number of years in operation and whether the first firm went out of business are not associated with higher or lower likelihoods that the founder will attempt a second founding. Model 6-3 estimates a model with the previous characteristics and examines whether entrepreneurs starting first firms within certain industries have a higher likelihood of starting a second firm. Entrepreneurs with first firms in biomedical industries or in telecommunications showed a significantly higher likelihood for a second founding. In the fully specified model (6-4), we drop the industry dummy variables and include instead industry average R&D to sales ratio and average industry fixed assets. Shorter lags from graduation to the first firm continue to increase the chance of a second firm. We are no longer confident that first firm ideas from research are associated with fewer second firm attempts. We find that a second firm is 80.6% more likely to be founded if the first firm was funded by angel (but not VC) investors. In addition, first firms in R&D intensive industries continue to be associated with greater likelihood of a second founding. VI. Conclusion Overall, the data support the idea that both first firm level factors and individual level characteristics influence the entrepreneur’s decision to start a second firm. In addition aspects of the business environment at the time of the first start-up (such as recessionary economy) do not appear to influence repeat entrepreneurship. However, it is difficult to disentangle the direction of causation from first firm experience characteristics to the decision to start a second firm. It may be that these types of founders have differences in style from the very beginning, or that characteristics of the first experience allow the founders to pursue involvement in the founding of multiple firms versus ceasing new firm founding after the first firm. 20 Hypothesis 1 was that younger first time entrepreneurs and those with more time in general will have a higher chance of founding a second firm and it is supported. Shorter lags from graduation to the first firm founding are associated with higher hazard rates of founding a second firm. Furthermore, those who are divorced presumably have fewer family commitments and thus more time to devote to founding a second firm. (Perhaps the formerly married are more inclined to manifest their need for achievement in building companies instead of building families. We are not aware of any prior work examining this question.) Both of these factors appear to make an entrepreneur more likely to start a subsequent firm. From a rational expectations perspective, it is interesting that having a first firm go out of business is not associated with entrepreneurs who are less likely to found second firms. One explanation may be that the entrepreneurs who have higher human capital are more likely to close a low performing business and expect to do better if they try again. The results indicate that age and time are important factors in the decision to enter a career of multiple entrepreneurship. Hypothesis 2 was that we would see an inverse-U relationship between first firm performance and the likelihood of founding another start-up. Looking at revenues, this hypothesis was not supported. The coefficients on the revenues and revenues squared variables were not significantly different from zero. If the first firm underwent an IPO, then this has no significant impact on the likelihood, which may be because the value of the company at IPO or beyond is unknown and could range from quite low to very high. However, first firms that were acquired are associated with higher likelihood of a second founding. Hypothesis 3 was that greater numbers of start-ups would be associated with higher financial capital for the entrepreneur. The data, especially in Figure 8, support the hypothesis. 21 Hypothesis 4a was that repeat entrepreneurs differ from single-firm entrepreneurs even prior to starting the first firm. This hypothesis is supported. When repeat entrepreneur is defined as those starting at least a second firm, we see that males, those who are divorced, and those who are citizens of Latin American or Asian countries are more likely to be repeat entrepreneurs. Hypothesis 4b was not supported since entrepreneurs with a background in engineering or the sciences are no more likely to start a second firm. Also, hypothesis 4c was not supported since first firms with patents are no more likely to be followed by a second founding. However, these variables may be poor proxies for the level of innovativeness of the entrepreneur. Hypotheses 5a and 5b were the joint hypotheses that aspects of the first firm experience would condition the likelihood of the founder starting a second firm versus multiple entrepreneurs choosing to start different types of first firms or going through a different startup process from those destined to only found a single firm. With these data we are not able to ascertain which direction the causality is flowing, however we can say with certainty that the firms of single-firm entrepreneurs do differ from the first firms of repeat entrepreneurs. The first firms of those who will eventually start a second firm are started sooner after graduation, and are funded by angel investors (Hypothesis 5e). Other factors associated with higher likelihood of a second startup include if the first firm was acquired (giving the entrepreneur time to do a second start-up), and being in an R&D intensive industry. The hypothesis that entrepreneurs starting first firms in high capital industries will be less likely to start multiple firms was not supported (hypothesis 5c). Industry fixed assets for the first firm appears to have no impact. First firms in biomedical and telecommunications industries are associated with an increased likelihood of the founder initiating a career of multiple entrepreneurship. First firms in software do not result in an increased likelihood of a subsequent 22 firm founding. While prior studies have noted that initial financial capital constraints may not inhibit many aspiring entrepreneurs from attempting to start a business (Dunn and Holtz-Eakin, 2000), such constraints appear not to cause entrepreneurs to think twice about a second attempt. Hypothesis 5d is related in that it proposes that entrepreneurs who have raised initial capital from VCs or angel investors will show an increased hazard for starting a second firm due to easier access to capital. This hypothesis is partially supported. Future work should explore more deeply why funding from angels but not from VCs increases the chances for a second start-up. Previous financing from angels may also increase the probability of a second startup due to the founder receiving mentoring from these investors and this may foster the ability or inclination to do another startup. Robustness and Limitations. This study is not without limitations. Year and sector dummy variables help alleviate concerns that various sources of unobserved heterogeneity drive our findings. The entrepreneurs were not randomly assigned to different conditions and thus the choice to start a new firm is endogenously determined by the entrepreneur. One might be concerned that the decision to found another firm is not based on the factors we propose, but rather from idiosyncratic or lucky private decisions by the entrepreneur or by others who approach the entrepreneur about joining their new start-up. Many factors are likely to have an impact on the decision to found a firm, however the individual level characteristics analyzed would not be expected to have any explanatory power if this was entirely true. The current data also may suffer from a self-report bias since both the dependent and some of the independent variables were reported by the entrepreneur. Net worth data in particular are likely to suffer from response bias. In general, while we observe a wide range of outcomes and firm sizes, it is likely that we do not observe every startup firm attempted by this 23 group of entrepreneurs. It could be that the single-firm entrepreneurs attempted to start other firms and failed or did not report them. We also cannot ascertain their view for the reasons for deciding to start a new firm. VII. Discussion The results on net worth indicate that a more subtle question needs to be asked about whether entrepreneurship pays (Barton, 2000). Entrepreneurship may indeed pay relative to regular employment for multiple entrepreneurs, though possibly not for single-firm entrepreneurs. In this paper we explore the factors that condition the likelihood that an entrepreneur starts a second firm. We use data from survey responses of 2,111 entrepreneurs to examine firm founding behavior. Results indicate that multiple entrepreneurs differ from single-firm entrepreneurs in certain demographic and educational characteristics prior to starting a first firm. The phenomenon of graduates embarking on careers of multiple entrepreneurship appears to be growing over time. The results also show that the first firms of eventual multiple entrepreneurs differ from the first firms of single-firm only entrepreneurs. The paper indicates that the factors associated with the highest probability that the entrepreneur will start a second firm are that they have greater time and access to financial resources to undertake a new venture. Starting a first firm sooner after graduation, in an R&D intensive industry, being divorced, having a first firm undergo an acquisition, and raising initial capital for the first firm from angel investors all increase the probability that the entrepreneur will start a second firm. The results all indicate factors which increase the expected returns (financial or non-financial) and reduce the costs to reinvesting time and money back into gaining access to new business opportunities via entrepreneurial activity. 24 Future work in this area could investigate the dynamics of the relative importance of financial versus non-pecuniary incentives for engaging in entrepreneurship as the entrepreneurial career progresses. Our results clearly indicate that financial incentives and expected financial returns appear to be important in deciding whether to engage in subsequent entrepreneurial activities. However, more detailed data on the entrepreneur’s perceptions of non-pecuniary motivations at various levels of prior entrepreneurial experience would provide the basis for an interesting study for future researchers in this area. These results have important implications for entrepreneurs and for those who invest in entrepreneurs. For venture capitalists and investors, our results could be used to identify who among the first time entrepreneurs they see are likely to become repeat entrepreneurs. Then, they could give more attention, funding or mentoring to these more promising first time entrepreneurs and put themselves and their firms in a position to be the ones that get to fund their subsequent firms as well. VC firms benefit since they can be more likely to pull eventual repeat entrepreneurs into their network before competing firms have a chance to do so (Gompers et al., 2006). Managers of large firms could use these results when attempting to hire individuals who are most likely to lead growth or new initiatives within the firm. For entrepreneurs, our results could imply greater mentoring for those who are most promising and encouragement to embark on a career of entrepreneurship sooner after graduation. The results also provide some guidance on which industries an entrepreneur might choose if becoming a repeat entrepreneur is the goal. Furthermore, for policymakers, our results could be used to structure programs to increase the incentives to engage first time entrepreneurs in becoming repeat entrepreneurs. The results also provide evidence of the important and significant economic and job growth impacts of repeat entrepreneurs relative to single-firm only entrepreneurs. 25 We see this paper as greatly advancing the literature by making significant strides toward both theory and empirical evidence on when first-time entrepreneurs will become repeat entrepreneurs. The results show factors that make an individual more likely to decide to reinvest their personal benefits from entrepreneurship back into new firm creation activities. Such a decision is clearly an important and overlooked aspect of entrepreneurship and this paper lays important groundwork for future researchers in this area. 26 References Acs, Z. J., & Audretsch, D. B. 1989. Small firm entry in manufacturing. Economica, 56: 255265. Alsos, G. A. 1998. The business gestation process of novice, serial, and parallel business founders. Entrepreneurship: Theory & Practice, 22(4): 101-14. Barton, H. H. 2000. Does entrepreneurship pay? An empirical analysis of the returns to selfemployment. The Journal of Political Economy, 108(3): 604. Bowman, Edward H., & Hurry, Dileep. 1993. Strategy through the option lens: An integrated view of resource investments and the incremental-choice process. Academy of Management.The Academy of Management Review, 18(4): 760-782. Carroll, Glenn R., & Mosakowski, Elaine. 1987. The career dynamics of self-employment. Administrative Science Quarterly, 32(4): 570-589. Cooper, Arnold C. 1970. Entrepreneurial environment. Industrial Research, 12. Dessein, W. 2002. Authority and communications in Organizations. Review of Economic Studies, 69: 811-838. Dunn, Thomas, & Holtz-Eakin, Douglas. 2000. Financial captial, human capital, and the transition to self-employment: Evidence from intergenerational links. Journal of Labor Economics, 18(2): 282-305. Freear, John, Sohl, Jeffrey E., & Wetzel, William E. 1994. Angels and non-angels: Are there differences? Journal of Business Venturing, 9(2): 109-123. Gimeno, Javier, Folta, Timothy B., Cooper, Arnold C., & Woo, Carolyn Y. 1997. Survival of the fittest? entrepreneurial human capital and the persistence of underperforming firms. Administrative Science Quarterly, 42(4): 750-783. 27 Gompers, Lerner, Scharfstein Gompers, P., Anna Kovner, Josh Lerner, David Scharfstein. 2006. Skill vs. Luck in Entrepreneurship and Venture Capital: Evidence from Serial Entrepreneurs. NBER Working Paper. Available at http://papers.nber.org/papers/W12592 Hayek, F. A. 1945. The use of knowledge in society. The American Economic Review, 35(4): 519-530. Hsu, David H. 2006a. Experienced entrepreneurial founders and venture capital funding. Wharton Management Department Working Paper: Wharton School, University of Pennsylvania-http://www-management.wharton.upenn.edu/hsu/files/Hsu%20%20Experienced%20Founders%208-25-05.pdf. Hsu, D.H. 2006b. Technology-based entrepreneurship. In S. Shane (Ed.), Blackwell handbook on technology and innovation management. Oxford, UK: Blackwell Publishing. Klepper, Steven, & Thompson, Peter. 2006. Spinoff entry in high tech industries: Motives and consequences. Working Paper. Klepper, 2001 McGrath, R.G., & MacMillan I. 2000. The entrepreneurial mindset. Boston, MA: Harvard Business School Press. McGrath, Rita G. 1996. Options and the enterprise: Toward a strategic theory of entrepreneurial wealth creation. Academy of Management Proceedings: 101-5. McGrath, Rita G. 1999. Falling forward: Real options reasoning and entrepreneurial failure. Academy of Management.The Academy of Management Review, 24(1): 13. Roberts, E.B. 1991. Entrepreneurs in high technology: Lessons from MIT and beyond. New York, NY: Oxford University Press. 28 Sarasvathy, Saras D., & Monon, Anil D. 2005. Failing firms and successful entrepreneurs: Serial entrepreneurship as a temporal portfolio. Working Paper Series, University of Virginia, Paper 04-05: available at http://ssrn.com/abstract=571921. Shane, Scott, & Khurana, Rakesh. 2003. Bringing individuals back in: The effects of career experience on new firm founding. Industrial and Corporate Change, 12(3): 519-543. Westhead, Paul. 1998. Novice, portfolio, and serial founders in rural and urban areas. Entrepreneurship: Theory & Practice, 22(4): 63-38. Wetzel, W. E. 1983. Angels and informal risk capital. Sloan Management Review, Summer: 2334. 29 Figure 1 Year of Founding 50 45 40 35 2nd firms 25 3rd firms 4th firms 20 >=5th firms 15 10 5 2002 1999 1996 1993 1990 1987 1984 1981 1978 1975 1972 1969 1966 1963 1960 1957 1954 1951 1948 0 1945 # of firms 1st firms 30 30 Figure 2 Model of the Entrepreneurial Career 31 6 3 -8 -4 -2 8, 44 4 5, 25 4 98 ,7 2 79 ,3 3 6 24 ,5 83 -9 -4 9 79 -1 4, 79 -7 6 6 -1 1, ,8 81 14 4, ,4 42 21 1 -5 5, , 25 25 1, 1, 24 24 8 8 an d ab ov e 89 8, 72 47 9, 33 22 4, 58 98 ,4 44 45 ,2 54 14 ,7 99 77 9 0 % Figure 3 Proportion of Founders Starting a 2nd Firm 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 Revenues of 1st Firm / Firm Age 32 Figure 4 Education and Citizenship 50 45 40 35 % Bachelor's Degree % 30 % Master's Degree 25 % Doctorate Degree 20 Non-U.S. citizen % 15 10 5 0 Novice (0 prior, lifetime) Repeat Entrepreneurs 33 Figure 5 Differences in Age at First Founding 45 40 35 # Years 30 25 Lag to First Firm Age founded 20 15 10 5 0 1st firm – novice only 1st firm – eventual repeat entrepreneur 34 Figure 6 First Firms of Different Entrepreneur Types 1000.0 900.0 Thousands of $ 800.0 700.0 600.0 Median Revenues / Operating Year (in 2001 $) 500.0 Median Initial Capital (2001 $) 400.0 300.0 200.0 100.0 0.0 1st firm – novice only 1st firm – eventual repeat entrepreneur 35 Figure 7 First Firm Outcomes 60 50 40 % Acquired (%) 30 Public (%) Out of Business (%) 20 10 0 1st firm – novice only 1st firm – eventual repeat entrepreneur 36 Figure 8 Net Worth Adjusted for Age 90.0 80.0 net worth / age 70.0 60.0 Net worth / age 50.0 40.0 30.0 20.0 1 2 3 4 5 6 or more Lifetime Firms 37 Figure 9 Founding Roles 60 CFO COO Chairman not also CEO 40 % Chief Scientist/CTO/Chief of R&D Consultant or Advisor President/CEO 20 VP 0 1st firms 2nd firms 3rd firms 4th firms or higher 38 Figure 10 Current Roles 45 40 COO 35 Chairman (if not also CEO) Scientist/CTO/Chief of R&D 25 Consultant or Advisor % 30 Not Active 20 Other employee 15 Outside Director President/CEO 10 VP 5 0 1st firms 2nd firms 3rd firms 4th firms or higher 39 Table 1 Estimated Economic Impact of Novice and Repeat Entrepreneurs Category of Entrepreneur Total Revenues (in '000 $2001) Total Employment Total # of Firms Founded Total Founders in the Sample Percentage of Entrepreneurs Percentage of Firms Percentage of Total Revenues Percentage of Total Employment Novice entrepreneurs (all ages) $20,736,000 1,353,420 954 954 72.1 22.7 11.6 51.3 Repeat Entrepreneurs $157,029,700 1,285,886 3254 878 27.9 77.3 88.2 48.7 40 Table 2 MIT Repeat Founders by Decade of Graduation Decade Total # firms = 1 % serial =2 =3 =4 =5 =6 =7 =8 =9 =10 =11 Totals 1930s 6 33.3 0 0 1 2 0 0 0 0 0 0 9 1940s 75 39 14 11 10 3 4 0 2 0 0 4 123 1950s 161 44.5 60 29 19 10 4 0 5 0 0 2 290 1960s 231 46.2 85 47 28 14 7 5 2 5 2 3 429 1970s 238 52.3 115 79 30 16 7 2 3 2 1 6 499 1980s 242 43.3 92 49 23 9 3 3 0 1 1 4 427 1990s 182 38.9 69 27 9 6 0 3 0 0 1 1 298 2000s 19 40.6 9 3 1 0 0 0 0 0 0 0 32 41 Table 3 Industry Distribution of Founders # total firms Manufacturing Aerospace Drugs & medical Chemicals & materials Consumer Products Electronics Machinery Other Manufacturing Manufacturing total Services Architecture Energy & utilities Telecommunications Finance Management & finance consulting Publishing & schools Software Law & accounting Other services Services total 1st firm % of (include 1108 1st s novices firms 25 51 23 2nd firm 2.2 3 4.6 21 2.1 7 % of 345 2nd 3rd firms firm 8.7 0 6.1 9 2.0 1 % of 3rd 4th firms firm 0.0 1 5.6 5 0.6 0 % of 4th firms 5th firms and higher % of 5th firms and higher Total # firms 2.0 1 10.0 1 0.0 3 3.3 30 3.3 87 10.0 34 29 181 14 30 353 2.6 16.3 1.3 2.7 31.9 9 55 3 11 109 2.6 15.9 0.9 3.2 31.6 7 33 0 5 55 4.3 20.5 0.0 3.1 34.2 2 8 2 2 20 4.0 16.0 4.0 4.0 40.0 0 2 1 2 10 0.0 6.7 3.3 6.7 33.3 47 279 20 50 547 53 29 34 41 89 4.8 2.6 3.1 3.7 8.0 13 14 7 19 28 3.8 4.1 2.0 5.5 8.1 6 5 11 6 15 3.7 3.1 6.8 3.7 9.3 2 0 2 2 5 4.0 0.0 4.0 4.0 10.0 1 1 2 1 3 3.3 3.3 6.7 3.3 10.0 75 49 56 69 140 21 196 47 245 755 1.9 17.7 4.2 22.1 68.1 12 78 8 57 236 3.5 22.6 2.3 16.5 68.4 1 36 6 20 106 0.6 22.4 3.7 12.4 65.8 0 9 3 7 30 0.0 18.0 6.0 14.0 60.0 1 3 2 6 20 3.3 10.0 6.7 20.0 66.7 35 322 66 355 1147 42 Table 4 Summary Statistics and Variable Definitions VARIABLE9 First start-up founded Second start-up founded DEFINITION Year in which first firm was founded (censored if not observed by 2003) Year in which second firm was founded (censored if not observed by 2003) MEAN SD 1985.10 12.30 1989.96 10.43 Individual Characteristics Graduation year Male Academic major Country of citizenship Year of MIT graduation 1973.25 Dummy = 1 if the individual is male 0.93 Set of dummies for academic major: engineering (53%), management (14%), social science (5%), architecture (4%), and natural science (the excluded category) Set of dummies for country of citizenship: Latin America (2%), Asia (7%), Europe (6%), Middle East (1%), Africa (1%) or North America (the excluded category) 15.04 0.26 First Firm Level Characteristics Age at 1st firm founding Recession Year Lag to First Firm Held Patents Team from Research Team from Work Idea from Research Idea from Work Idea from Networking VC funded Angel funded L Initial Capital Acquired Public L Revenues # Cofounders Out of Business Operating Years Industry 9 Age of the entrepreneur the year the first firm 37.5 10.27 was founded Dummy = 1 if the firm was founded during a 0.22 0.41 recession year as categorized by the NBER. Lag (in years) from graduation to the first firm 14.28 9.95 founding Dummy = 1 if the firm held patents 0.14 0.35 Dummy = 1 if the team met doing research / in a 0.11 0.32 lab Dummy = 1 if the team met via work 0.21 0.41 Dummy = 1 if the source of the idea was research 0.14 0.35 Dummy = 1 if the source of the idea came from work 0.62 0.49 Dummy = 1 if the source of the idea came from networking -0.26 0.60 Dummy = 1 if the firm received venture capital funding 0.14 0.35 Dummy = 1 if the firm received funding from angel investors 0.09 0.29 Capital “raised to get the company off the ground” 11.97 2.71 Dummy = 1 if the firm was acquired 0.21 0.41 Dummy = 1 if the firm had an IPO 0.13 0.33 Firm revenues for a specific year 14.24 3.03 Number of cofounders 0.15 0.36 Dummy = 1 if 15.27 12.01 Number of years the firm has been in operation 1.17 0.78 Set of dummies for self-reported industry: aerospace, architecture, biomedical, chemicals, consumer products, consulting, electronics, energy, finance, law/miscellaneous, machine tools, publishing, software, telecommunications as well as other services, and other manufacturing. (see Table 3 for percentage in each category.) “L” preceding the variable name in the regression tables denotes natural log. 43 Table 5 Entrepreneurship Cox Hazard Rate Regressions (Individual level of analysis) Dependent Variable = Second start-up founded (subjects start being at risk at year of first firm founding) Note: reported coefficients are hazard ratios Independent Variables (5-1) (5-2) (5-3) (5-4) Demographic vars. Education vars. Citizenship vars. Full model Graduation year 1.028*** 1.027*** (0.003) (0.003) Male 1.593* 1.578* (0.307) (0.308) Single 1.027 1.072 (0.165) (0.174) Divorced 1.396+ 1.399+ (0.286) (0.288) Bachelor’s degree 0.954 1.037 (0.084) (0.097) Doctorate degree 0.909 0.967 (0.110) (0.122) Engineering major 0.908 1.112 (0.088) (0.189) Management major 1.236+ 1.373+ (0.156) (0.257) Science major 0.861 1.081 (0.112) (0.211) Architecture major 0.769 0.875 (0.188) (0.246) Latin American citizen 1.877*** 1.606** (0.333) (0.297) Asian citizen 1.757*** 1.486* (0.299) (0.259) European citizen 1.327+ 1.265 (0.200) (0.198) Middle Eastern citizen 0.868 0.716 (0.309) (0.257) African citizen 1.929 1.619 (0.309) (0.670) Log likelihood -4,065.23 -5,040.05 -4,281.47 -4,055.56 Number of obs. 1338 1559 1383 1338 Note: 630 failures; 15,099 years at risk; ***, **, *, + indicate statistical significance at the 0.1%, 1%, 5%, and 10% levels, respectively. 44 Independent Variables Table 6 Cox Hazard Rate Regressions Dependent Variable = Second start-up founded (subjects start being at risk at year of first firm founding) Note: reported coefficients are hazard ratios (6-1) (6-2) (6-3) (6-4) Initial Conditions Performance All + Industries All Effects 0.995 (0.023) 1.003 (0.026) 1.007 (0.025) 0.749 (0.143) 0.745 (0.168) 0.719 (0.159) 0.948* (0.023) 0.936* (0.025) 0.935** (0.025) 0.681+ (0.137) 0.730 (0.190) 0.680 (0.167) 0.979 (0.230) 0.833 (0.238) 0.799 (0.225) 1.021 (0.187) 1.092 (0.239) 1.099 (0.239) 0.474* (0.156) 0.561 (0.229) 0.552 (0.222) 0.805 (0.198) 0.795 (0.239) 0.744 (0.235) 0.707 (0.219) 0.893 (0.344) 0.914 (0.341) 0.991 (0.273) 1.127 (0.345) 1.122 (0.334) 1.549 (0.442) 1.725+ (0.525) 1.806* (0.536) 0.969 (0.036) 0.942 (0.043) 0.945 (0.042) 1.501* (0.284) 1.630* (0.335) 1.597* (0.319) 1.098 (0.271) 1.025 (0.270) 1.072 (0.270) 1.005 (0.028) 0.899 (0.128) 0.924 (0.129) 0.999 (0.006) 1.004 (0.006) 1.003 (0.005) 1.324 (0.354) 1.144 (0.344) 1.180 (0.351) 1.003 (0.028) 1.025 (0.032) 1.025 (0.033) 0.999 (0.001) 0.999 (0.032) 0.999 (0.001) 0.681 (0.377) 2.362* (1.000) 1.467 (0.727) 2.105* (0.808) 0.948 (0.284) 1.304 (0.690) 1.234 (0.662) 0.500 (0.512) 0.972 (0.257) 1.321*** (0.119) 1.114 (0.107) -1.239+ (0.139) Age at 1st firm founding Recession Year Lag to First Firm Held Patents Team from Research Team from Work Idea from Research Idea from Work Idea from Networking VC funded Angel funded L Initial Capital Acquired Public L Revenues L (Revenues)2 Out of Business Operating Years Operating Years2 Aerospace Biomedical Chemical Telecommunications Electronics Energy, Elec. Utilities Finance Machinery Publishing Software Industry R&D to sales ratio Industry Fixed Assets 1.503 (1.629) 0.741 (0.845) -1.735 (2.110) # Cofounders 1.242*** (0.084) 1.181 (0.082) 1.129 (0.094) 1.129 (0.092) Log likelihood -991.55 -822.47 -723.43 -728.62 Number of observations 719 320 482 480 Note: 756 failures; 17,947 years at risk; ***, **, *, + indicate statistical significance at the 0.1%, 1%, 5%, and 10% levels, respectively. 45