Intuitive Optimizing for Time Allocation Decisions in Newly Formed Ventures

Moren Lévesque, CWRU
Christian Schade, Humboldt-Universität zu Berlin

CHAPTER MENU

ABSTRACT
INTRODUCTION
RELATED STUDIES
ANALYTIC FRAMEWORK
EXPERIMENTAL DESIGN AND FINDINGS
BEHAVIORAL INTERPRETATION OF FINDINGS AND CONCLUSIONS
CONTACT
REFERENCES
TABLE 1
TABLE 2

Abstract

We analyze a time allocation situation typically faced by entre­preneurs in new ventures: how to allocate time between a newly formed enterprise and a wage job. First, we formulate a formal model and derive predictions on optimal behavior in different decision situations. Second, we test these predictions with real decision makers in questionnaire experiments. Third, we compare model’s predictions to experimental data and provide behavioral interpretations of discrepancies.

Introduction

There are many decision situations where individuals have to intuitively optimize on economic variables. By intuitive optimization we refer to a method where an individual makes a decision based on intuition rather than using a formal procedure that identifies the optimal solution, and thus leads to perfect optimization. Perfect optimization may not always be possible because there is not enough information available to formulate a complete model, real decision makers may be boundedly rational, and decision makers do not always have enough time for determining the optimum. These three conditions apply to most entrepreneurial decision situa­tions since these situations are by and large new and unique, entrepreneurs may not always have received a formal training in optimization, and there is serious time pressure involved.

In this paper we analyze a time allocation situation typically faced by entre­preneurs in newly formed ventures: how to distribute time between their own newly formed risky enterprise and a risk-free wage job. First, we formulate a formal model and derive predictions on optimal behavior in different decision situations, assuming that individuals’ preferences obey certain rationality axioms and that these individuals perfectly optimize. Second, we test these predictions with real decision makers using questionnaire experiments on simple situations where respondents must detect corner solutions in different experimental conditions, i.e., solutions where they allocate either the maximum or the minimum number of hours to the venture. To test the sensitivity of behavior with respect to risk propensity, we also measure this construct on each indivi­dual. Third, we compare model’s predictions to experi­men­tal data and provide behavioral inter­pretations of discrepancies. We find that, al­though respondents are relatively close to the nor­mative predic­tions, deviations are systematic and worth exploring. Many individuals seem to utilize an anchoring and adjustment procedure and seem to be strongly influenced by an affect heuristic recently introduced to the decision making literature.

Related Studies

There are relevant empirical studies involving the allocation of time in new ventures. These studies include the work of McCarthy, Krueger, and Schoenecker (1990) who study the relationship between a firm’s early stages of development and its founder’s time allocation pat­tern. More recently, Cooper, Ramachandran, and Schoorman (1997) conclude that craftsmen-entrepreneurs devote less time to administrative activities while entrepreneurs with mana­gerial experience are less likely to follow the objectives of craftsmen-entrepreneurs and devote more time to administrative activities.

There also is fundamental analytical research in career choice to draw upon in developing our model. Lévesque and MacCrimmon (1997) offer a dynamic model that addresses the question of when the best time is for an entrepreneur to leave a wage job and become a full-time entrepreneur. They show that the optimal time-allocation policy is driven by the entrepreneur’s tolerance for work and by how returns from time invested behave with respect to time allocated to that venture. Others have investigated the choice of moving in and out of self-employment (e.g., Lévesque, Shepherd and Douglas, 2002). 

Campbell (1992) also uses an economic decision model to study the decision to become an entrepreneur as an alternative to wage labor. Although he does not investigate the dynamics of career choice, his framework offers a more explicit consideration of risk and implies that risk-takers should be less inclined than risk-averters to increase entrepreneurial activity when the probability of success increases. 

The entrepreneurship literature, however, mostly deals with risk propensity as a discriminator between entrepreneurs and non-entrepreneurs. There has been no evidence for risk-neutrality or even risk prone behavior of entrepreneurs as compared with others. In fact, entrepreneurs have a tendency to risk aversion (e.g., Brockhaus, 1982). The impact of risk propensity on entrepreneurial success has also been studied from a developmental psychological perspective, but no conclusive findings have been drawn. Risk propensity should not impact the predicted optimal solutions according to the model we offer, but respondents may be boundedly rational and hence risk propensity may matter. 

We build upon the above literature in three ways. First, our model is closer to reality by taking into account the riskiness of most entrepreneurial decision situations and diffe­rences in the “stakes” of certain newly formed ventures. Second, our model is “testable” since it is built upon a preference calculus that is consistent with rational decision making under risk. Third, we test in a questionnaire experiment whether respondents are able to intuitively optimize—and thus behave in a rational manner—and how close actual behavior is to the normative prediction.

Analytic Framework

The Entrepreneur’s Decision and Ventures’ Characteristics

We offer an effort-allocation model where the entrepreneur’s decision-making problem is based on a utility function consistent with rational decision making according to expected utility theory. Assume that the total work tolerance of an entrepreneur is given. That is, there is a maximum of working hours that can be devoted either to the current wage job or to developing a new venture. The key decision is how many hours, h, to allocate to the new venture; then  hours will be devoted to the wage job. Time allocation h is restricted to be above a certain minimal threshold, , for the new venture to stay alive. Thus, along with the investment into the new venture, the entrepreneur commits to a minimal effort . Moreover, h must be below the work tolerance of the entrepreneur,.  

Different ventures can be described by different combinations of risk and return. Let s (> 0) be the stakes of the venture. s enlarges both payoff mean and variance, and thus captures the stakes (or “leverage”) of the venture. One can think in terms of a high stakes venture for start-ups in the high-tech industry where enormous payoff may be expected, but also low payoff can be encountered as the uncertainty surrounding the purchase of expensive setups and equipment may result in an unprofitable venture. Low stakes ventures are associated with less ambitious start-ups such as those in the service industry. 

Let V be the marginal payoff from the increase in the stakes of the venture. This margi­nal payoff is not only affected by the entrepreneur’s effort into that venture, but also by exoge­nous risk. Let the ran­dom variable X be this risk and V = f(h,x), where x is a realization of X. Note that the only input upon which the decision is made is the entrepreneur’s time allocation to the new venture, h. Any other inputs are assumed fixed, or their quantities already established beforehand.

Therefore, the entrepreneur decides upon her/his time allocation by facing a wealth, denoted by W, of . The first term on the right-hand side represents the (risk-free) payoff associated with the wage job. Since the marginal payoff is uncertain, the wealth of the entrepreneur will be uncertain at the time of the time-allocation decision. The welfare derived from wealth W can be formalized by the expected utility E[U(W)], or by the entrepreneur’s welfare U which is expressed in terms of the certainty equivalent U(h,p) = u-1(E[u(W)], where u is the entrepreneur’s Neumann-Morgenstern utility function. We suppose that the entrepreneur is risk neutral or risk averse, and thus u concave. Risk-loving behavior in the normative sense is atypical among regular subjects and, based on the empirical studies referred earlier, we argue that it hardly ever applies to entrepreneurs—entrepreneurs are no “gamblers,” they are just more or less risk averse.

We are able to study how an entrepreneur’s work tolerance, wage rate and characteristics of the new venture affect the time allocation that maximizes welfare by further specifying functions and variables as that utilized by the Linear-Exponential-Normal-Model (e.g., Spremann, 1987). The relevant assumptions underlying this model for our analysis are that: (a) the marginal payoff from increased stakes of the new venture V is linear in the risk X, and consequently payoff linear in V; (b) the entrepreneur’s utility function u is exponential; and (c) the risk from the new venture X follows a normal probability distribution. Assumptions (a) and (c) imply that the entrepreneur’ wealth is normally distributed. That, in conjunction with assumption (b), implies that the certainty equivalent U can be expressed as expected value minus half the variance times risk aversion (Bamberg and Spremann, 1981). 

Therefore we select X normally distributed, , and the entrepreneur’s utility to be , with . The entrepreneur selects a time allocation h* that maxi­mizes the certainty equivalent, which now reads as  =  = .  measures the entrepreneur’s risk propensity. We next characterize a time allocation strategy that optimizes the entrepreneur’s wealth for decreasing, increasing and constant returns from time invested in the new venture. We also investigate how this strategy is affected by changes in key model parameters. 

Time Allocation Strategies and their Sensitivity Analysis

Ventures differ in the amount of payoff that results from different levels of time allocated to them. Some new ventures consume a lot of effort before they begin to pay off but when they do the payoff (e.g., financial) may become very large. This characterizes the case of increasing returns to entrepreneurial effort or when viewed in terms of the functional relationship between the payoff and time allocated, we say that the function g is convex. An example is a business where reputation builds up such as consulting. Other new ventures give significant immediate payoffs when effort is first invested but then give decrea­sing marginal payoffs. This is a case of diminishing returns to entrepreneurial effort and is represented by a concave function g. Examples for such ventures include babysitting and catering in a small village where willingness to pay decreases with the number of customers already reached.   

Decreasing Returns to Time Allocated in the New Venture.  We first investigate the case where g is strictly concave in h, and more specifically where , and a > 0 is the non-random part of V from the first hour allocated to the venture. We select an exponent function because by varying the value of the exponent n one can “swipe” an infinite set of possible curves, making empirical validity of this functional form more likely. How, then, will the entrepreneur choose to allocate time between a wage job and developing a new venture? It is straightforward to verify that sufficient and necessary conditions for optimality are satisfied for . Therefore, since ,

Proposition 1 (corner solutions).  When returns to time allocated in the new venture are decreasing with , it is optimal to allocate h* = e hours in to this venture (the survival constraint of the venture) if , and  hours in to this venture if .

Proposition 2 (sensitivity of the corner solutions).  When returns to time allocated in the new venture are decreasing with , and , a change in the wage rate , the stakes of the new venture (s), or the non-random portion of V from the first hour allocated to the new venture (a) does not affect the time allocated to the new venture, unless (a) the wage rate becomes large enough, (b) the stakes of the new venture becomes small enough, or (c) the non-random portion of V from the first hour allocated to the venture becomes small enough for to exceed .

Proposition 3 (invariance of optimal solutions to risk propensity).  Optimal solutions as stated in Proposition 1 are independent of the entrepreneur’s risk propensity . 

Increasing Returns to Time Allocated in the New Venture.  With increasing returns to time allocated in the new venture, the optimal time is always a corner solution. The minimum or the maximum number of hours is allocated to the venture depending on how the average rate of returns compares to the wage rate. It is straightforward to verify that the following results hold. Note that Propositions 3 and 5 are direct implications of a basic feature of our simple model, as it does not exploit the potential tradeoff between the mean and variance of a business opportunity: time allocations do not affect the variance term.

Proposition 4 (optimal solution).  When returns to time allocated in the new venture are increasing with g strictly convex in h, the optimal time allocation takes the form of a corner solution where  hours in to this venture if  and   hours in to this venture if .

Proposition 5 (invariance of optimal solution to risk propensity).  The optimal solution as stated in Proposition 4 is independent of the entrepreneur’s risk propensity . 

EXPERIMENTAL DESIGN AND FINDINGS

Research in entrepreneurship has begun to build up from (beha­vioral) decision theory and to carry out questionnaire experiments on entrepreneurial behavior (e.g., Forlani and Mullins, 2000). We chose to conduct an experimental study because of the likely difficulty in a field study to control for the broad array of factors believed to influence risky new venture decisions in natural settings (Baird and Thomas, 1985). This approach has advantages in internal validity for theory testing purposes, but may be criticized on the basis that the experimental task is not a real one with real payments. Given the newness of this research area, we believe this tradeoff is acceptable. 

Design of the Questionnaire Experiment

In the questionnaire experiment we selected values for the model parameters in such a way that only corner solutions had to be detected by the respondents. We compared the optimal time allocation of situations with increasing versus decreasing returns to time allocated in the newly formed venture, low versus high stakes for the venture, and low versus high wage rate. We also manipulated the tolerance for work by initially fixing it in the different groups of respondents on different levels (i.e., 8 and 12 hours per day) and allowed for later adjustments on that tolerance for work. While some of these combinations should lead to the lower bound solution ( ), others should lead to the upper bound solution .

We therefore empirically investigate whether respondents are able to intuitively optimize in a manner such as that suggested by Propositions 1-5.  Table 1 presents predictions according to Propositions 1 and 4, along with experimental findings and numerical values utilized on the model parameters.

Respondents were asked to imagine being in an entrepreneurial decision situation and were provided with basic information such as how a normal distribution looks like. Each group of respondents (group 1: 12 hours work tolerance; group 2: 8 hours work tolerance) had to decide on time allo­cation in the venture for different decision situations. Exact condi­tions, including payments, probability distributions (figures), earnings for an (addi­tional) hour spent on the venture, and total earnings with specific time allocations, were dis­played in the questionnaire for each of the 16 cases. These 16 cases are referred to as “condition 1” to “condition 16” in Table 1. 

In each of the high work tolerance cases the subject was asked the following: “We now want to know how many working hours you would allocate to your own venture in this situation, daily. Remember that you are willing to work a total of 12 hours a day. The rest will auto­ma­tically be allocated to your wage job. You can report fractions of hours.” In each of the low work tolerance cases the same text was given with 12 hours being replaced by 8 hours. Later in the questionnaire, the respondent was asked for each situation to state the total amount of time (in number of hours) s/he would be willing to work as well as the time (in number of hours) s/he would allocate to the new venture. More specifically we asked: “Imagine now that you could change the total hours you work daily (that you can move it upwards or downwards). How many total hours would you have liked to work in this situation, and how would you have liked to distribute these hours between your wage occupation and your own venture?” 

Finally, to verify whether or not risk propensity has an influence on actual behavior—according to Propositions 3 and 5 it should not—we measure each respondent’s risk propensity via the McCord and Neufville (1986) lottery comparison approach. After pre-testing the experiment with a few master students, we ran the experiment on a total of 112 students of Humboldt-Universität zu Berlin, Germany. In general these students had a good formal education in decision sciences and game theory, and were stemming from different fields including business, economics, computer scien­ce, and pedagogy. The entire experimental design may be seen in Table 1, where the expe­rimental findings are compared to the normative predic­tions of Propositions 1 and 4.

Unless stated differently, each reported p-level below is from a two-sided non-parametric (e.g. Wilcoxon) test.

Testing Propositions 1, 2 and 4

12 Hours Work Tolerance Group.  We first examine the results of the 12 hours work tolerance group. Since all answers are usable, subsequent calculations are based on the entire group of 52 respondents. Respondents do meet the normative prediction from interpreting predictions of 12 versus 1 hour spent in the venture as directional hypo­theses. From Table 1, hours allocated to the venture are “high” when they should be, i.e., under conditions 1-4, & 7; whereas hours allocated to the venture are “low” when they should be, i.e., under conditions 5, 6 & 8. All relevant pair wise comparisons are highly significant (.000).

However, the interpretation of how close decision makers are to the absolute normative prediction—the actual 12 or 1—must involve a discussion of deviations. Looking at individual data and referring to Table 1, 57.7% of the respondents selected the correct corner solution in condition 1, 28.8% in condition 2, 75.0% in condition 3, 55.8% in condition 4, 42.3% in condition 5, 51.9% in condition 6, 44.2% in condition 7, and 43.1% in condition 8 (one missing value). Based on a confidence interval analysis, under all conditions deviations from the correct corner solu­tions are significant at a 1%-level (one-sided). 

If respon­dents’ behavior differs significantly from the normative prediction, are the deviations syste­matic? Indeed they are. In the high-stakes/high-salary cases (conditions 1 and 2), the optimal solution is 12 hours whether returns from time allocated to the venture are decrea­sing or increasing (.000). However, many individuals rectified the number of hours allocated to the venture by adjusting that number downwards when returns from time allocated to the venture were decreasing as compared to when they were increasing. The same holds in the high-stakes/low-salary cases (conditions 3 and 4) (.000). In the low-stakes/high-salary cases (conditions 6 and 7), both increasing and decreasing returns should lead to a corner solution of 1 hour, but again the number of hours allocated to the venture was adjusted downwards in the decreasing returns case (.000). 

One the other hand, low wages make the individuals adjust the hours allocated to the venture upwards, even though it should have no impact according to the optimization model. This holds for the high-stakes/increasing-returns cases (conditions 1 and 3; .000) and for the high-stakes/decreasing-returns cases (conditions 2 and 4; .000), which should all lead to a corner solution of 12 hours regardless of wages. The same holds for the low-stakes/decreasing-returns cases (conditions 6 and 8; .001), were both wage rates should lead to a 1-hour solution. 

Finally, high stakes of the venture also make the individuals adjust the hours allocated to the venture upwards, even though it should have no impact according to the optimization model. Time allocation into the venture should be 12 hours for the increasing-returns/low-salary/high-stakes case and for the increasing-returns/low-salary/low-stakes case (conditions 3 and 7). The difference under these conditions is in favor of the high stakes venture where more time is allocated (.000). In summary, in the 12 hours work tolerance group we find weak support for Propositions 1 and 4, and Proposition 2 clearly has to be rejected.

8 Hours Work Tolerance Group.  We next look at the same data for the 8 hours work tolerance group. Two respondents opted for more hours in their own venture then they were “allowed” to and one stated zero hour. These three respondents were eliminated from the following analyses (and Table 1), leading to a total of 57 usable respondents. As in the previous group, hours allocated to the venture are “high” when they should be (under conditions 9-12 & 15); whereas hours allocated to the venture are “low” when they should be (under conditions 13, 14 & 16). All relevant pair wise comparisons are highly significant (.000).

There is again a fraction of individuals in each condition who “detected” the correct corner solutions: 38.6% in condition 9, 28.1% in condition 10, 47.4% in condition 11, 49.1% in condition 12, 49.1% in condition 13, 58.9% in condition 14 (one missing value), 33.9% in con­dition 15 (one missing value), and 44.6% in condition 16 (one missing value). However, as in the 12 hours work tolerance group, we find an effect of normatively “irrelevant” conditions. Low stakes make working in the venture less attractive in the increasing-returns/low-salary case (.000) while low salaries make it more attractive in the high-stakes/increasing-return, high-stakes/decreasing-returns, and low-stakes/decreasing-returns cases (all .000). Also, decreasing returns make the enterprise less attractive in the high-stakes/high-salary cases (.007), high-stakes/low-salary cases (.018), and low-stakes/high-salary cases (.001). We thus find in the 8 hours work tolerance group some support for Propositions 1 and 4, whereas Proposition 2 has to be rejected.

Testing Propositions 3 and 5

Table 2 reports on r2 values and significance levels of linear OLS regressions. These regressions are run under the 16 conditions associated with both work tolerance groups. Risk propen­sity is the independent variable whereas number of hours allocated to the venture is the dependent variable. Surprisingly, and in direct contradiction with Propositions 3 and 5, six out of eight regressions (75%) in the 12 hours work tolerance group are at least marginally signi­ficant. In the 8 hours work tolerance group, only three out of eight regressions (37.5%) are at least marginally significant. In other words, the number of hours allocated to the venture de­pends on the risk propensity under multiple conditions, but it should not according to the nor­mative model.

Furthermore, the signs of the regression coefficients in Table 2 offer interesting conclu­sions. First notice that based on our basic exponential utility function, the larger the risk pro­pensity  the more risk averse the decision maker will be. Second, according to our regression model , the larger the risk propensity  the larger the time allocated to the venture (h) whenever the regression coefficient  is positive, but when that coefficient is ne­gative it is a small time allocation to the venture that is associated with a large risk propensity. Therefore, in the 8 hours work tolerance group where the sign of the regression coefficient is always negative, the more risk averse the decision maker, the less hours s/he will allocate to the venture. In the 12 hours work tolerance group, the regression coefficient is negative in the cases where the upper bound on time allocated to the venture was predicted as optimal, but it is positive when the lower bound was predicted as optimal. 

Behavioral Interpretation of Findings and conclusions

Optimal Solutions.  Respondents behave according to the normative predictions when interpreted as directio­nal hypo­theses. That is, when a corner solution with the maximum available time allocated to the venture was predicted, respondents did in fact allocate a high number of hours to the venture, whereas when a corner solution with the minimum input (1 hour) was predicted, they chose to allocate little effort into the venture. Nevertheless, only a subset of the respondents opted for the corner solution, and this percentage varies between experimental conditions. From a more care­ful analysis of the data, we find that our propositions are unsupported under each of the 16 experimental conditions of Table 1. Namely, our results are not consistent with the respondents selecting the corner solution, and deviations from optimal behavior are not only due to random errors. We detected a strong effect of factors that should have no impact: low wage rates, high stakes or increasing returns to time allocated in the venture lead to more hours spent on the venture. Without knowing the model’s predictions, these deviations may appear reasonable at first sight. 

What does it mean for our respondents’ decision processes? A subgroup of respondents does in fact detect the corner solution as optimal. We argue that the remaining respondents make use of an anchoring and adjustment procedure (Tversky and Kahneman 1974). In such a proce­dure, respondents select a preliminary level of time allocated to the venture—the anchor. This anchor would start high in situations where the venture seems profitable and/or the wage job relatively unattractive, but low in situations where the venture appears relatively poor and/or the wage job relatively attractive. The anchor is then adjusted where a high stakes ven­ture would justify more effort into the venture than a low stakes venture, a high wage job lead to a lower time allocation into the venture than a low wage job, and increasing returns to time allo­cated to the venture provide better incentive for more hours spent on the venture than decreasing returns. Respondents do not optimize and hence are unaware that some of these con­ditions require the same amount of hours spent on the venture, even if initially these conditions may appear to differ in their attractiveness.

Risk Propensity.  More surprisingly, risk propensity plays a role in the time allocation of our respondents. Indivi­duals have been demonstrated in beha­vioral deci­sion theory to disobey normative models of choice (e.g., Hershey, Kunreuther, and Schoemaker, 1982; Tversky and Kahneman, 1992). Also, recent studies demonstrate that affect and emotion may be more important than the objective analysis in many decision situations. According to the affect heuristic, certain alternatives may intuitively appear riskier although this is not warranted by an objective analysis (Slovic, Finucane, Peters and MacGregor, forthcoming). Whilst adjusting the hours allocated to the venture upwards and downwards, respondents may take into con­sideration subjective risk differences. If a venture is just perceived as a more risky occupa­tion than a wage job, the subjective risk may appear higher when allocating more hours to the venture, while that risk appears lower when allo­cating less hours to the venture. 

This interpretation is consistent with observed behavior under all condi­tions in the 8 hours work tolerance group, and under all conditions where the upper bound is optimal in the 12 hours work tolerance group. Since the entire payoff distribution is shifted upwards with every hour spent in the venture (under the con­ditions where the maximum number of hours should be allocated to the venture) our respondents seem to violate the principle of sto­chastic domi­nance. In fact, the risk of a loss diminishes as more time is devoted to the venture! It is only if the enter­prise appears to be a poor alternative and the respondent is allowed to work 12 hours that risk aversion leads to more hours spent on the venture. Only here, the respondents may be forced to realize how close they are to incurring losses with their enterprise and opt to spend more hours on it to move the distri­bution up­wards—without being forced to significantly reduce the hours allocated to the safe wage job. In the 8 hours work tolerance cases, on the other hand, this relationship is not found since they prefer to allocate very little away from their safe wage job.

Implications for Entrepreneurial Behavior and Education.  Does it matter to teach optimization to entrepreneurs? Optimization methods can help to answer the “what should be” questions, and guide those entrepreneurs who wish to improve their decisions. Especially if systematic deviations from rational behavior—i.e., behavior predicted from the model—are observed empirically, decision makers may want to learn which mistakes are typical and how to steer against potentially inefficient decisions. 

Are we successful in teaching optimization to entrepreneurs? The answer is probably not. What can we do to help entrepreneurs overcome inefficient decisions? We argue that a star­ting point is to teach them how to maneuver against these typical mistakes. In our decision situations, decision makers had a tendency to violate the principle of stochastic dominance by not realizing that working more in their enterprise did not make their position more risky but moved upwards the entire payoff distribution. We suspect that it would be easier to train people to become aware of such situations—and avoid perceiving every hour spent on their own busi­ness as more risky than spending that hour on a wage job—than making them apply a closed form solution from a mathematical model.

Implications on Incentives to Work in the New Venture.  This work should be of interest to banks and venture capitalists in better understanding the disincentives to work in the new firm. As long as every hour worked in these firms is perceived as more risky than an hour spent in a wage job, entrepreneurs may opt to dedi­cate an insufficient amount of work to their company. Perhaps it would help to make potential entrepreneurs aware of a recent trend in some industries: their wage jobs may also be unsafe, that is, they may be laid off.

Implications on Labor Market Regulations.  In Europe, the labor market is highly regulated so that time allocation in the investigated sense is difficult for the potential entrepreneur. An entrepreneur may face the alternatives of either working half time in the wage job, full time, or not at all. This makes starting one’s own busi­ness even riskier. In turn findings that demonstrate the relevance of these speculations to real (or at least experimental) entrepreneurs may convince politicians and labor unions of needed changes in the institutional environment to encourage entrepreneurship. Therefore, the design and experimental testing of a more refined model where the possibility to allocate time away from the wage job is limited would be desirable. 

Conclusions.  We offered a formal model where an entrepreneur must decide on how to allocate time between a wage job and a new venture. Predictions were derived on optimal behavior in 16 different decision situations, assuming individuals’ preferences obey certain rationality axioms and they perfectly optimize. These predictions were tested with real decision makers using simple que­stionnaire experiments where respondents had to detect corner solutions in the 16 experimen­tal conditions. These questionnaires were also utilized to test the sensitivity of the decisions with respect to risk propensity. 

Model’s predictions were compared to experimental data and found unsupported. Respondents behaved according to the normative predictions when interpreted as directional hypo­theses, but only a subset of the respondents opted for the corner solution, and this percen­tage varied between experimental conditions. Also, the number of hours allocated to the venture depended on the risk propensity under multiple conditions, although it should have not according to the optimization model. We detected a strong effect of factors that should have had no impact, namely low wage rates, high stakes or increasing returns to time allocated in the venture led to more hours spent on the venture. 

Our experimental findings are consistent with most individuals utilizing an anchoring and adjustment procedure and being influenced by affects, rather than a fully rational optimizing approach. The use of heuristics in decision making under risk is a relatively well-known fact in behavioral decision theory, and many recent studies deal with the impact of affects on choice. Here, many individuals seem to be misled by their subjective risk perception (i.e., by affects) to a degree that they violate the principle of stochastic dominance. An additional hour invested in the own enterprise appears riskier than investing that same hour in a wage job, although this was not the case.

CONTACT: Christian Schade, Institute for Entrepreneurial Studies and Innovation Management, School of Business and Economics, Humboldt-Universität zu Berlin, Spandauer Str. 1, 10178 Berlin, Germany;  (T) + 49 30 2093 5904; (F) + 49 30 2093 5918; schade@wiwi.hu-berlin.de

REFERENCES

Bamberg, G., & K. Spremann. (1981) “Implications of Constant Risk Aversion.” Zeitschrift für Operations Research 25: 205-224.

Baird, I.S., & H. Thomas. (1985) “Toward a Contingency Model of Strategic Risk Taking.” Academy of Management Review 10(2): 230-243.

Brockhaus, R.H. (1982) “The Psychology of the Entrepreneur.” In Encyclopedia of Entrepreneurship, ed. C.A. Kent, D.L. Sexton, & K.H. Vesper, 39-56. Englewood Cliffs NJ: Prentice Hall.

Campbell, C.A. (1992) “A Decision Theory Model for Entrepreneurial Acts.” Entrepreneurship Theory and Practice 17(1): 21-27.

Cooper, A.C., M. Ramachandran, & D. Schoorman. (1997) “Time Allocation Patterns of Craftsmen and Administrative Entrepreneurs: Implications for Financial Performance.” Entrepreneurship Theory and Practice 22(2): 123-136.

Forlani, D., & J.M. Mullins. (2000) “Perceived Risks and Choices in Entrepreneurs’ New Venture Decisions.” Journal of Business Venturing 15(4): 305-322.

Hershey, J. C., H. Kunreuther & P. J. H. Schoemaker. (1982) “Sources of Bias in Assessment Procedures for Utility Functions.” Management Science 28: 936-954.

Lévesque, M., & K.R. MacCrimmon. (1997) “On the Interaction of Time and Money Invested in New Ventures.” Entrepreneurship Theory and Practice 22(2): 89-110.

Lévesque, M., D.A. Shepherd, & E.J. Douglas. (2002) “Employment or Self-Employment: A Dynamic Utility Maximizing Model.”  Journal of Business Venturing 17: 189-210.

McCarthy, A.M., D.A. Krueger, & T.S. Schoenecker. (1990) “Changes in the Time Allocation Patterns of Entrepreneurs.” Entrepreneurship Theory and Practice 15(2): 7-18.

McCord, M., & R. de Neufville. (1986) “Lottery Equivalents”: Reduction of the Certainty Effect Problem in Utility Assessment.  Management Science 32: 56-60.

Slovic, P., M. Finucane, E. Peters, & D. MacGregor. (Forthcoming) “The Affect Heuristic.” In Intuitive Judgment: Heuristics and Biases, ed. T. Gilovic, D. Griffin, & D. Kahneman. Cambridge University Press.

Spremann, K. (1987) “Agency Theory and Risk Sharing: Agent and Principal.” In Agency Theory, Information, and Incentives, ed. G. Bamberg & K. Spremann, 3-37. Berlin: Springer Verlag.

Tversky, A., & D. Kahneman. (1974) “Judgment under Uncertainty: Heuristics and Biases.” Science 185: 1124-1131.

Tversky, A. & D. Kahneman. (1992) “Advances in Prospect Theory: Cumulative Representation of Uncertainty.” Journal of Risk and Uncertainty 5: 297-323.

2002 Babson College. All Rights Reserved. Last Updated March 2003.