Corporate Venture Capital: The Upside of Failure and Competition for Talent

We consider the motives for a firm to engage in corporate venturing. We argue that, in case of failure of a new venture, corporate venture capitalists (CVC) have a strategic advantage relative to traditional venture capitalists (VC) in creating rents after rehiring or refinancing the entrepreneurs. Hence, corporate venturing induces the would-be entrepreneur to exert an effort that is higher than within the corporation, but lower than under traditional venture capital financing. Ceteris paribus, the entrepreneur, ends up with fewer shares and less control under CVC financing than under traditional VC financing. Competition from venture capitalists increases corporate venturing activity, the salaries of potential entrepreneurs, and total economic output. Our results are consistent with the observed pro-cyclicality of corporate venture capital activity with venture capital activity.

. Does this simply suggest that corporations attempt to obtain their share of the potential pro…ts associated with …nancing new ventures, or is there more to it?
The second observation is that corporate venturing seems to yield signi…cantly lower returns relative to venture capital investments, unless investments are in related lines of business (Gompers, 2002). 2 Indeed, venture capitalists claim that corporations do not have the competencies for such investments (Chesbrough, 2002). So why do corporations engage in intrapreneurship?
In this paper we provide a simple explanation for these two empirical regularities.
We argue that one of the crucial characteristics of intrapreneurship is that, by funding ideas from their own employees, corporations can generate a rent by rehiring a valuable manager into their own ranks (or …nancing another project of his) even when the new 1 In the United States, over 200 corporations were listed in the 2002 Directory of Corporate Venturing as investing as active corporate venture capitalists. Corporations also invest in venture capital through specialized institutions such as venture capital funds. At the end of 2001, corporations were the second largest source of capital to venture capital funds, after endowments and foundations, with total commitments of about $35 billion (Goldman and Russell, 2002). Some of these investments, often organized through partnerships, are sometimes included in the de…nition of corporate venture capital, but not in that of "intrapreneurship". Throughout the paper, we do not distinguish between corporate venturing and intrapreneurship. 2 See also Gompers and Lerner (1998 2 venture is not successful. 3 The CVC obtains higher rents after the failure primarily because his past experiences as an employee and as an entrepreneur provide an informational and/or matching advantage to the CVC compared to the VC. We do not formally specify whether these higher rents take the form of the …nancing of a second project from the failed entrepreneur or a return to employment. Although we say that the CVC "rehires" his former employee, the rents can take either form. During the late 1990s many companies, such as Procter and Gamble and Dannon, organized contests among their employees and funded the most promising business plans. The employees that won the contest then became entrepreneurs that were …nanced by the corporate venture capitalist. Interestingly, the winning projects were not necessarily directly related to the CVC's lines of business, and in case of failure the entrepreneur could have his/her old job back. Introducing this idea in an otherwise standard principal-agent model is su¢ cient to explain the second observation. Speci…cally, we show that: Other things being equal, corporate venturing is likely to yield lower returns than independent venture capital investing: The rent created in case of failure o¤ers insurance to the corporation and/or the manager, thus reducing incentives, e¤ort, and expected performance.
Unless the VC has a signi…cant expertise advantage, the rehiring gain dominates the e¢ ciency loss due to lower incentives. Despite a lower expected return on investment in the new venture, the total return to the investor/principal is higher with intrapreneurship than with venture capital, and it is still optimal to invest in corporate venturing rather than venture capital.
To answer the …rst question we must understand why employing the star manager as part of an intrapreneurship project rather than keeping him as a regular employee becomes more attractive as entrepreneurial activity increases. 4 We argue that as entrepreneurial activity increases, competition for star managers intensi…es as the managers 3 If the venture is successful the corporation bene…ts for obvious reasons. 4 Throughout the paper we view the principal as a female and the agent as a male.
3 become aware of, and attracted to, the life of an entrepreneur. Indeed, at the peak of the entrepreneurial boom in the late 1990s, retaining talented employees proved di¢ cult: " reasoning relies on the assumption that the entrepreneur's e¤ort has more impact on overall performance in an entrepreneurial project than as an employee of a corporation.
The driving forces of our analysis work as follows: 1. Without competition the corporation often decides to keep the potential entrepreneur as an employee. The reason for this is that the …rm appropriates rents from the employment relationship that often exceed the rents that it could obtain by …nancing a venture in which his limited ownership would restrict the entrepreneur's incentives.
2. As competition from venture capitalists for promising entrepreneurs intensi…es, the manager's market wage becomes higher, and so do the rents to the corporation if the manager remains an employee, and the manager's ownership and e¤ort under both VC …nancing and CVC …nancing. 4 3. The cost to the investor of the increase in market wage is mitigated by the manager's increase in e¤ort, and this mitigating e¤ect is larger with the CVC and the VC as the manager's increase in e¤ort is larger. This increases the attractiveness of the CVC and the VC relative to retaining the manager within the …rm.

4.
A VC competing with a corporation for a manager can o¤er him an expected compensation that is high enough to steal the employee from the corporation.
Hence, keeping the potential entrepreneur as an employee is no longer possible.
Competition may prompt the corporation to engage in corporate venturing in cases where the manager would have remained an employee otherwise.

5.
Unless the VC's expertise advantage is signi…cant, CVC …nancing yields a higher total payo¤ than VC …nancing. The CVC can therefore outbid the VC and retain the star manager.
6. In our model, pro…table projects are not …nanced, as outside …nancing leads to the potential entrepreneur to exert a suboptimal e¤ort level, and hence leads to an expected return lower than that obtained with the employment relationship.
Competition from a VC provides entrepreneurs with more opportunities to be …nanced and a higher ownership in their project. This leads them to exert a higher e¤ort, and generates more entrepreneurship in equilibrium.
In other words, competition from the VC induces the …rm to o¤er higher salaries to their most promising employees and to engage in corporate venturing. Hence corporate venturing and venture capital activities are highly correlated. Apart from being consistent with the …rst observation, our results also have a number of implications. For example, we expect that the salaries of potential entrepreneurs (young managers may be a proxy for possible entrepreneurs) may move together with the entrepreneurial cycle, and that high levels of entrepreneurship, through higher managerial compensation, can improve incentives within organizations.
Earlier papers argue that strategic motives for intrapreneurship may help answer our second question, and that rates of return are high when these strategic factors are 5 taken into account. For example, Gompers (2002) and Gompers and Lerner (1998) document that the projects …nanced by the CVC exhibit higher success rate when the projects are related to the CVC's main line of business, but signi…cantly lower returns on "non-strategic" investments. Hellmann (2002) argues that corporate venturing is more likely to take place when the project being …nanced is more complementary with the corporation's existing lines of business. Although product-market interactions are undoubtedly a factor, existing papers do not explain the presence of "non-strategic" investments by CVCs. We suggest that retaining star managers and rehiring them when ventures fail may play an important role. This suggests that the positive spillovers of corporate venturing to other activities of the corporation may be observed when the project fails, rather than when the project succeeds as suggested by the complementarity hypothesis. 5 The two papers that are closest to ours are probably Gromb and Scharfstein (2003) and Amador and Landier (2003). Gromb and Scharfstein also consider the "safety net" that intrapreneurship provides to entrepreneurs. However, our approach and our results di¤er from theirs in several respects: Gromb and Scharfstein focus on the choice between traditional entrepreneurship …nanced by venture capital and intrapreneurship …nanced by CVCs, and on the interactions between the expected rents to failed entrepreneurs and the equilibrium level of entrepreneurship. As in our paper, intrapreneurship provides lower incentives than entrepreneurship because of the safety net, but it comes with an informational advantage about the quality of the failed intrapreneur. They do not address the e¤ect of competition from traditional venture capitalists on the decisions to engage in corporate venture capital as opposed to retaining managers as employees. In contrast, the central features of our analysis are the choice between the corporate structure and the CVC, and how this choice is a¤ected by competition from venture capitalists for star managers. This is the driving force behind our result that traditional venture capital and corporate venture capital move together, which as argued above is consistent with empirical evidence. 6 Amador and Landier (2003) developed (independently) a model in which competition from venture capitalists sometimes prompts …rms to adopt innovations that they would not adopt otherwise. As in our paper, competition makes it more likely that the manager's idea will be …nanced elsewhere (e.g. by a VC), and it transfers rents from the investor to the manager. However, the modeling structures and the results are di¤erent.
For example, in Amador and Landier, the …rst e¤ect increases the value of …nancing the manager's idea, but the second one reduces it, thus making the overall e¤ect of competition ambiguous. In our model, with the CVC the cost of transferring rents to the manager is mitigated by e¢ ciency gains, but not (or at least less so) in a corporation.
Here the second e¤ect reinforces the …rst one, and competition unambiguously increases the value of …nancing new ideas. Furthermore, they do not distinguish between corporate venturing and an innovation adopted within the corporation, and hence they do not discuss the interactions between corporate venturing activity and venture capital activity, which are the focus of our paper. 7 Section 2 describes our simple model. Section 3 derives the subgame equilibrium when the …rm retains the manager as an employee, and when the manager's project is …nanced by a corporate venture capitalist and a traditional venture capitalist. Section 4 develops a comparative analysis of corporate venture capital …nancing and traditional venture capital …nancing. Section 5 examines how a change in the manager's market compensation a¤ects the attractiveness of corporate venturing relative to traditional venture capital …nancing. Section 6 analyzes the e¤ects of competition between a CVC 6 In addition, project failure plays a role in our paper that is di¤erent from Gromb and Scharfstein (2003) and Landier's (2003) paper on the e¤ect of failure on the equilibrium level of entrepreneurship. In their papers, the market perception of the reasons for failure plays a central role in reaching the equilibrium level of entrepreneurship. For instance, in Landier's paper, if the market believes that failed entrepreneurs ask to be …nanced a second time mostly because of their …rst failure, then …nancing will be expensive, and a low entrepreneurship equilibrium will obtain. But if the market believes that the entrepreneurs that ask to be …nanced after a failure have an improved project, then this will lead to cheaper …nancing, and hence a high entrepreneurship equilibrium. 7 In this paper, we adopt a conceptual view of entrepreneurship that may di¤er from earlier papers. First, we view managerial talent and entrepreneurial talent as requiring similar skills, which appears to be consistent with the fact that …rms lose a large number of successful managers during times of high entrepreneurship. In contrast, earlier papers focus on the di¤erences between managerial talent and entrepreneurial talent. For instance, Lazear (2003) argues that entrepreneurial talent requires a broad of skills while managerial talent requires perhaps more outstanding, but also more specialized skills. 7 and a traditional VC for promising would-be entrepreneurs. Section 7 concludes.

The Model
Consider the relationship between an investor and a penniless project manager. The manager has a market salary w 0 , and he is risk-neutral. His involvement in the project requires an unveri…able e¤ort p that a¤ects the probability of success. The project requires an investment I and generates high return h with probability p and low return l with probability (1 p). The cost of e¤ort to the manager is c(p) = k 2 p 2 . We consider this principal-agent relationship in three types of environment: Corporation (C), venture capital (VC), and corporate venture capital (CVC). We consider each of them in turn.
The investor and the manager can simply interact in a "corporate" relationship in which the investor is a CEO and the manager an employee. We argue that an employee in a corporation has a much lower impact on the overall surplus than does an entrepreneur starting a new venture. For simplicity, and without loss of generality, 8 we make the extreme assumption that the employee's job is routine and he has no impact on the surplus: The surplus in a corporation is the same in both states of the world, i.e. h c = l c = c . 9 In the case of corporate venture capital, the investor represents corporate headquarters, the manager is an employee who becomes an entrepreneur, and the project itself is the entrepreneur's idea. The particularity of corporate venturing is that if the project fails, the …rm can re-employ the project manager in the job he had before starting the project, and the …rm generates a rent y in addition to l . Without loss of generality, we assume that the …rm appropriates this entire additional rent y, e.g. it has all bargaining power when negotiating this (non-contractible) rent.
With venture capital …nancing, the investor is a venture capitalist, and the project manager is an entrepreneur. In our standard case throughout the paper we assume 8 What is important is that the manager's marginal product is smaller in a corporation than in a new venture, not that it is zero. All results in the paper still hold as long as h c l c < h cvc l cvc y. 9 Despite being a routine job, an employment relationship could yield a high return c , e.g. because of the …rm's market power and accumulated experience and physical assets. that the payo¤s are exactly the same as with corporate venture capital, except that the VC cannot generate any rent by rehiring the entrepreneur if the venture fails. When relevant, we also explicitly consider the di¤erence between the VC's higher ability to manage new ventures, and the CVC's advantage in strategic product development. To that end, we assume that the payo¤ in the good state under the VC is h vc = h cvc + s, where s is positive when the VC's higher e¢ ciency dominates, and negative when the CVC spillover e¤ect dominates. The e¤ect of s is in many respects formally equivalent to the e¤ect of y, so for the sake of clarity we assume s = 0 in most of the paper and we discuss how s 6 = 0 a¤ects our results when relevant.
Since e¤ort is unveri…able, payment to the manager is contingent on the return realized: The project manager receives a fraction h of the payo¤ h in the good state of the world, and a fraction l of l in the bad state of the world. These di¤erent fractions will capture option-like features that are traditional in venture capital contracts (de Bettignies, 2003, Chemla, Habib, and Ljungqvist, 2003, Schmidt, 2003. 10 The timing is as follows: At date 0, the investor chooses the organizational form F 2 fC; V C; CV Cg.
At date 1, the investor makes a take-it-or-leave-it contractual o¤er to the manager.
The manager decides whether to accept or reject the o¤er. If the o¤er is rejected, the game ends and both parties obtain zero. If the o¤er is accepted, the investor makes an investment in a new venture when applicable.
At date 2, the manager exerts e¤ort.
At date 3, the project's return is generated and the payo¤s are distributed.

Corporation
In a corporation, the manager has no impact on the payo¤: h c = l c = c . The investor does not need to provide the manager with incentives, and the equilibrium is trivial: She o¤ers him the market wage w 0 in both states of the world, he exerts zero e¤ort, the surplus is c , of which the investor keeps c w 0 .

Corporate Venture Capital
The corporate headquarters'program can be written as follows: subject to the manager's incentive compatibility (IC) constraint: the individual rationality (IR) constraint: and wealth constraints: We call w 0 the market wage: It re ‡ects the compensation that obtains if the manager does not participate and seeks another job in the labor market. As we shall see below, this market wage may be determined endogenously.
As a benchmark, we …rst describe the …rst-best (FB) outcome, in which the investor can observe the manager's e¤ort. Since the above program then reduces to maximizing (1) with respect to p, subject to the manager's participation constraint (3), it follows that the FB optimal e¤ort level is: For ease of exposition, we assume that h ; l ; y and k are such that 0 < p < 1. In equilibrium, the manager exerts e¤ort p , and he obtains w 0 + k 2 p 2 . In the second-best (SB), when the e¤ort is non-contractible, the IC and the IR constraints reduce to: and: respectively. The e¤ort level and fractions of payo¤ to be given to the manager in equilibrium depend on the market wage w 0 . Let Z = p cvc ; h cvc ; l cvc be the set of equilibrium e¤ort and fractions of payo¤ in the good and the bad states, respectively.

Proposition 1
The second-best equilibrium e¤ort and compensation variables with the CVC can be described in three regions: Proof: See the Appendix.
We will call 1 cvc , 2 cvc , and 3 cvc the regions identi…ed from top to bottom by the second column. We assume that k > 2w 0 , so that in equilibrium, 0 < p cvc < 1.
In order to provide the manager with appropriate incentives, the investor sets l cvc < h cvc , and l cvc = 0 as long as the e¤ort exerted is not optimal. The fractions above can be interpreted as a fraction l cvc of the shares plus a fraction ( h cvc l cvc ) of call options allocated to the entrepreneur, or as a fraction h cvc of the shares, plus a fraction ( h cvc l cvc ) of call options to the CVC (or put options to the entrepreneur). As documented in Chemla, Habib, and Ljungqvist (2003) and Schmidt (2003), such clauses appear to be frequently used in venture capital contracts. Since CVC …nancing shares many features with traditional venture capital …nancing, it is natural that it comes with similar contractual clauses.

Venture Capital
The same reasoning as in the previous subsection applies. This leads to a …rst-best equilibrium e¤ort is p vc = h l .

Corollary 1
The second-best set of equilibrium e¤ort and fractions of payo¤ in the good and bad states Z = p vc ; h vc ; l vc can be written: where the second column represents regions 1 vc , 2 vc , and 3 vc from top to bottom respectively. Figure 1 graphically depicts some of the information in (8) and (9) (8) and (9) and from Figure 1, the equilibrium level of e¤ort p by the project manager, and consequently the expected return on the project, are higher with venture capital than with corporate venture capital, so p vc > p cvc and To explain this, it is convenient to re-write the investor's program as follows: where h (p), and l (p) are equilibrium fractions of payo¤. Expression (11) re ‡ects the corporate venture capitalist's program. For the venture capitalist the program is the same, but with y = 0. The …rst and second squared brackets represent the bene…t and cost to the investor of inducing e¤ort p, respectively. In equilibrium, the payment variables h (p), and l (p) are the same for the VC and the CVC as functions of p.
Consequently the marginal cost of inducing e¤ort, d[p h (p) h +(1 p) l (p) l ] dp , is the same for the CVC and the VC (taking p as given).
The di¤erence between the VC and the CVC comes from the marginal bene…t of inducing e¤ort. For the corporate venture capitalist this marginal bene…t is h l y, which is lower than that of the venture capitalist ( h l ). This occurs because the latter does not have the opportunity of hiring the manager back into its core business in the event that the venture fails. We then obtain the following proposition. This could explain why corporate venture capitalists tend to perform worse than venture capitalists: They provide too much insurance to the entrepreneur by garanteeing his job back in case of failure. The manager/entrepreneur is thus given lower incentives, which in turn leads to lower expected returns.
Even though, as depicted in (10), the expected return on the project is higher with venture capital, the overall return to the investor is higher with the CVC than with the VC: In the good state, the surplus is h for both the CVC and the VC, and in the bad state the surplus is y more for the CVC. The two types of investor are otherwise identical. Thus, as long as the CVC does not relinquish all of that surplus advantage y to the manager -and she designs the contract such that he does not -she will be better o¤ than the VC. In sum: Proposition 3 When project-speci…c payo¤s h and l are the same for the CVC and the VC, the overall return to the investor is higher with corporate venturing than with venture capital.
After Proposition 2 we may be tempted to ask: "if corporate venturing yields lower returns than traditional venture …nancing, then why do …rms invest in corporate venturing rather than in a venture capital …rm?" This question focuses on the advantage of the VC, namely a higher probability of success, and ignores the CVC's advantage, a 13 higher payo¤ in the bad state. In fact, this higher success probability with the VC is the result of the CVC's advantage. As such the cost of having a lower likelihood of success is always more than o¤set by the bene…t of having a higher payo¤ in the bad state.

Specialization E¢ ciency Versus Strategic Product Development
In Proposition 3, the VC's higher e¢ ciency exactly o¤sets the CVC's spillover e¤ect: As discussed in section 2, s = 0. We now extend the results in i.e. if s is su¢ ciently negative, the marginal bene…t from managerial e¤ort is actually higher with the CVC, which implies higher managerial e¤ort and higher expected return in equilibrium relative to the VC. This is consistent with the results in Gompers and Lerner (1999) and Gompers (2002) that CVC returns are not lower than VC returns for strategic investments. In our model with s < 0, Proposition 2 still holds provided y + s 0, 11 i.e. if s is not too low. In addition, Proposition 3 still holds for all values of s < 0: If the CVC was better o¤ than the VC with s = 0, then she must also be better o¤ with s < 0. capitalist of inducing highe e¤ort is even higher, and the di¤erence in e¤ort p vc p cvc > 0 is even larger. On the other hand, Proposition 3 no longer necessarily holds with s 0.
Indeed, for a given rehiring advantage y there is a threshold of net VC e¢ ciency level s > 0 such that if s > s there exists a set of feasible values of w 0 over which the investor's return as a VC is higher than that as a CVC. 12;13 However, when the venture capitalist's e¢ ciency advantage s is low, corporate venturing still dominates venture capital (the return to the venture capitalist is denoted R vc ) in abolute terms (see Figure 3). Since we focus on corporate venture capital, the remainder of the paper focus on cases where s is (close to) zero.

Organizational Choice and Managerial Compensation
In this section, we aim to analyze how a change in the market wage a¤ects the relative attractiveness of one organizational form over the others. Proof: Follows directly from (8) (see Figure 1).
The intuition behind the result above is simple. For the …rst-best level of e¤ort to be implemented, the di¤erence between the high payo¤ and the low payo¤, h h l l , must be large enough to provide the manager with appropriate incentives. However, the investor is constrained because i) l l cannot be too low, as l must remain positive, and ii) h h cannot be too high, as the higher h the higher cost to the investor. Thus in equilibrium the optimal incentives to the manager generate an equilibrium e¤ort that is weakly lower than the …rst-best. As the manager's market wage w 0 increases, the investor must pay him more to induce him to become an entrepreneur. The investor increases the manager's compensation by increasing the payo¤ in the good state, h h , because this leads to an increase in h h l l , and thus in the manager's incentives and equilibrium e¤ort level. Thus, the higher cost to the investor of inducing the manager to become an entrepreneur is partly compensated by the manager's higher incentives.
When market wage is high enough (w 0 The payo¤ to the investor is always higher in a corporation. With R iii cvc , the situation is reversed: The investor is always better o¤ with corporate venturing. The most interesting case is that of R ii cvc . In that case, a corporate structure dominates at low levels of w 0 , but as w 0 increases the attractiveness of corporate venturing relative to a corporate structure Proof: Similar to that of Proposition 5.
From Propositions 5 and 6 we obtain the following corollary: The payo¤ to a venture capital investor relative to a corporation increases with w 0 .

Competition for "Star Managers"
The investor must choose one of the three possible organizational structures around her manager. The standard corporate structure, in which the investor keeps the manager 14 In contrast, changes in parameters of the production function may increase the attractiveness of CVC relative to VC. An increase in y unambiguously favors CVC over both VC and C, despite an adverse e¤ect on the manager's e¤ort under CVC …nancing. The reason for this is that a higher y increases the CVC's advantage of creating rents after a failure. Similarly, an increase in h and, equivalently, a decrease in l , both have a positive e¤ect on the entrepreneur's incentives both under VC and CVC …nancing. The convexity of the cost of e¤ort implies that when there is an insu¢ cient provision of e¤ort under CVC, an increase in h and a decrease in l both favor CVC over VC. as an employee; corporate venturing, where the project is …nanced, with the possibility of rehiring (or re-…nancing) the manager and generating a rent y if the venture fails; or an investment through an independent venture capital partnership, where it is not possible to generate rents by rehiring the manager in bad times, but in which there may be an e¢ ciency advantage s in good times due to specialization. Section 4 examined the choice between corporate venturing and venture capital …nancing. We now examine organizational choice between corporate structure and corporate venturing, i.e. we retain the assumption that s is (close to) zero.
A central question in this paper is how competition for star managers a¤ects this choice. Given our foregoing analysis, the answer to this question is simple: Competition for star managers tends to increase their market wage, consequently making corporate venturing a relatively more attractive opportunity.
For ease of exposition, we restrict our analysis to the most interesting case, depicted in …rgure 4, where the value of c is such that R c R cvc at low values of the market wage w 0 , and such that R c < R vc at high values of w 0 . 15 When there is no competition for star managers, and when the market wage equals zero (w 0 = 0), a corporate structure is more pro…table for the investor: The R c line is above R vc and R cvc .
We focus on the case where the investor competes with a venture capitalist to keep her star manager. Once the organizational structure is chosen, the investor competes with the VC on the market wage. In the unique equilibrium in the market wage, the payo¤ to the weaker competitor is zero. If the investor chooses a corporate structure, an increase in the manager's market wage weakens her position (from Corollary 2). In Figure 4, she is the weaker competitor: The equilibrium is w 0 = w 0 0 where R c (w 0 0 ) = 0. She will lose her manager to the venture capitalist who will be able to lure him away by o¤ering w 0 0 + ". If, on the other hand, the investor chooses corporate venturing, the equilibrium wage is w 0 = w 00 0 such that R vc (w 00 0 ) = 0. Our corporate venture capitalist wins the competition for the star manager by o¤ering w 00 0 + ". Thus, choosing corporate venturing rather than keeping a manager as an employee may allow the investor to retain that manager when there is such competition: Proposition 7 Competition for star managers increases their market wage. This may prompt the investor to choose corporate venturing when she would otherwise have preferred to keep the manager as an employee.
In our model, there is a suboptimal level of venture …nancing: whenever c is higher than the total expected output generated by the CVC with …rst best e¤ort, the …rm can always retain the manager by increasing his compensation beyond the level that the competing VC could o¤er. In that case the investor's organizational choice is also the social optimum. Now consider the situation where c is lower than the total expected output generated by the CVC at the …rst best level of e¤ort, but not at the second best e¤ort level. In that case corporate venture capital is socially optimal, but is not the organizational structure chosen by the investor, due to agency costs. Competition for star managers alleviates agency costs, increases the equilibrium e¤ort level towards the …rst best, and that may induce the investor to choose the socially optimum structure, which is corporate venturing in this example. Hence, competition for star managers increases total economic output: By increasing both e¤ort and the fraction of the surplus created by the project that is allocated to the manager under both VC …nancing and CVC …nancing, and By prompting the …rm to …nance the project in cases where she would have retained the manager as an employee in the absence of competition. This is summarized in the following proposition: Proposition 8 Competition for star managers i) prompts investors to …nance entrepreneurial projects, ii) increases entrepreneurial e¤ort under both VC …nancing and CVC …nancing, and iii) increases total economic output. 3. As the market wage goes up, the investor under both VC and CVC …nancing can incentivize the manager more easily than in a regular …rm, because the manager's e¤ort has a higher marginal impact. More incentives imply that the employee increases e¤ort more as part of a CVC or VC than as an employee 16 . 4. The cost to the investor of the increase in market wage is mitigated by the manager's increase in e¤ort. This mitigating e¤ect is larger with CVC and VC …nancing as the manager's increase in e¤ort is larger.
5. This makes corporate venturing and traditional venture capital attractive options relative to keeping the manager as an employee.
6. Thus a VC competing with a corporation for a star manager can o¤er a market wage high enough to steal the star manager from the corporation. Keeping the manager as an employee in the face of competition by venture capitalists becomes unfeasible.
7. In the absence of a high s corporate venturing yields a higher total payo¤ than venture capital …nancing, and the CVC can outbid the VC and retain the star manager.

Competition increases total output by increasing both managerial e¤ort and in-
vestors'incentives to …nance new ventures.
Item 7 may be a¤ected in a case where the VC's valuation for the project in case of success is signi…cantly higher than the CVC's valuation. If s > s, there is a threshold 16 In our model, the manager's e¤ort in a corporation remains constant at zero. However, an analysis similar to that made in that paper implies that if h c > l c , competition from VC may lead the …rm to keep the manager as an employee, but to increase the incentive component of his package. Hence, competition from VC may also improve incentives within organizations. 20 level of w 0 = w 0 such that it is optimal for the investor to choose VC …nancing over CVC …nancing for w 0 > w 0 . From Proposition 6 venture capital becomes relatively more attractive an option compared to corporate venturing, as w 0 increases. When s is large, this relative advantage can turn into an absolute advantage. Thus, Proposition 7 can be amended as follows for large values of s: When s is large, competition may prompt the investor to choose venture capital (and not CVC …nancing) when they would otherwise prefer to have the manager as an employee. This may explain some …rms' decision to set up an independent venture capital fund.

Discussion and Concluding Remarks
In this paper we have provided simple answers to two questions that are central to the empirical literature on corporate venturing. The …rst question is, "Why do …rms continue to invest in corporate venture capital when they could make a higher return by investing in independent venture capital?" We have argued that the CVC's cost of yielding lower expected returns than the VC is more than o¤set by the superior ability to rehire entrepreneurs as employees when their venture fail, thus making CVC …nancing an attractive alternative for …rms.
The second question that we tried to answer is, "What happened in the late 90s?
Why was there such a dramatic increase in corporate venturing, which vanished as quickly when the e-commerce bubble burst?" We have argued that one of the explanations for this pro-cyclical investment in corporate venturing comes from competition for star managers. The media buzz around entrepreneurship in the mid to late 90s made employees more aware of their potential as entrepreneurs and became more open to such a career.
Consequently, VC …rms and corporations started to compete more …ercely for such managers. This resulted in an increase in managers'compensations, and in a strategy switch, by corporations, into corporate venturing.
This new interest from agents to become entrepreneurs may have come from more than just a media buzz. An increase in payo¤s and returns may have increased their interest in the …rst place. How does an increase in the project's good payo¤ ( h ) a¤ect our model? 21 An increase in project returns reinforces our main result. It increases the attractiveness of corporate venturing relative to corporate organization through both direct and indirect e¤ects. The direct e¤ect is simple: As the return rises, so does the expected payo¤ from the venturing project relative to that of the corporation, which remains constant. There are also two indirect e¤ects. First, ceteris paribus a higher return ( h ) generates a higher di¤erence between good and bad outcomes, and thus higher equilibrium e¤ort with CVC …nancing relative to within a corporation. Second, it implies that the VC, who also bene…ts from this increase, has an increased advantage relative to the corporation when it comes to competing for the manager, and hence the opportunity cost of not choosing corporate venturing, in terms of risk of losing good managers, is higher.
In our paper, there is too little entrepreneurship in equilibrium from a social viewpoint. Entrepreneurial activity generates competition for star managers, which in turn induces e¢ ciency and output gains. This result comes about because competition, by forcing the CVC to forfeit more rents to the manager, also allows her more freedom to provide incentives, and this in turn generates the aforementioned e¢ ciency and output gains. This is to be contrasted with Gromb and Scharfstein, where suboptimal entrepreneurial activity arises because would-be entrepreneurs do not internalize the positive externality that they have on the labor market. Further analyses of the equilibrium level of entrepreneurial activity and the way it compares with the level that would be best for the economy will undoubtedly be the subject of much future research. 22 optimal e¤ort level, the investor must o¤er a spread h h l l that is large enough. This is achieved by setting l to zero and h to a level that is high enough to generate the required incentives. The manager receives a compensation that resembles an "e¢ ciency wage". An increase in w 0 a¤ects neither the shares allocated to the manager nor the equilibrium level of e¤ort, and the investor's payo¤ remains constant. An increase in w 0 reduces the payo¤ to the corporation, but not to the corporate venture.
In region 2 cvc , the market wage is so high that the IR constraint (3) is binding: The investor pays the manager his market wage plus his cost of e¤ort. As the market wage goes up, the payment that the investor must make to the manager increases, and the investor responds by increasing the fraction of payo¤ h . This in turn increases the spread between the good and bad outcome for the manager, and hence his e¤ort (from (6)). Unlike in a corporation where e¤ort does not matter, with corporate venturing the negative e¤ect of an increase in w 0 on the investor's payo¤ is mitigated by the increase in managerial e¤ort and in the expected payo¤.
In region 3 cvc , the market wage is so high that for the manager's participation constraint to hold, h must be su¢ ciently high to generate the …rst best e¤ort level p cvc = p cvc (see ICC (6)). Since the investor does not want the manager to exert an e¤ort higher than p cvc , as w 0 goes up the investor increases h and l to ensure that the equilibrium e¤ort is p cvc = p cvc , and that the manager's participation constraint still holds, and is binding. This implies that the expected return on the project is constant and equal to p cvc h +(1 p cvc ) l , and that the cost to the investor is equal to k 2 p 2 cvc +w 0 and increases at rate 1 with w 0 Thus, in region 3 cvc the marginal cost to the investor of an increase in w 0 equals 1 both in the corporation and with corporate venturing.  As the reservation wage increases, the CVC return always rises relative to that of the corporation. As compared to the return from corporation, the return from CVC may be i) always below, ii) below for low levels of reservation wage, and above for high levels, or iii) always above. In the paper we focus on the most interesting case ii).

Figure 3: Return from CVC versus return from VC.
As the reservation wage increases, VC becomes more attractive relative to CVC, but in our base case, the return from CVC remains higher throughout. When there is little competition from VCs, the reservation wage is low and the corporate structure yields the highest return. When competing for star managers with VCs, they lose the employee if they keep the corporate structure (NE at w'0), but keep the manager and obtain a higher return if they choose intrapreneurship (NE at w"0).