Financing the Entrepreneurial Venture

Transcription

1 Financing the Entrepreneuria Venture Jean-Etienne de Bettignies y This Draft: November, 2005 Abstract This paper is about nancia contracting choices for the entrepreneur. In an incompete contracts mode, the entrepreneur can design contracts contingent on three possibe contro right aocations: entrepreneur-contro, investor-contro, and joint contro, with each aocation inducing di erent e ort eves by both the entrepreneur and the investor. Four types of contract emerge as potentiay optima: debt with iquidation, debt with reorganization, equity-ike nancing, and preferred-type (convertibe and straight) nancing. The mode: a) determines optimaity aong two dimensions: the convexity of output as a function of incentives, and the investor s opportunity cost of capita; b) highights the importance of ex-ante and ex-post e ciency in contracting; c) generates empirica predictions about the determinants of nancia contracts. Keywords: Entrepreneuria nance, incompete contracts, debt versus equity. JEL Codes: M3, G32. I thank Martin Boyer, Jim Brander, Gies Chema, Matt Cements, Doug Cumming, Murray Frank, Keith Head, Bob Hesey, Thomas Hemann, Gordon Phiips, John Ries, Tom Ross, and Raph Winter, for hepfu comments. Abe research assistance was provided by Chris Bidner. A errors are mine. y Sauder Schoo of Business, University of British Coumbia. Te: (604) Emai: bettignies@sauder.ubc.ca.

2 Introduction Entrepreneuria ventures pay a crucia roe in the economy: in the U.S. sma rms represent more than 99.7 percent of a empoyers, [...] empoy more than haf of a private sector empoyees, [...and] create more than 50 percent of the nonfarm private gross domestic product. 2 The vast majority of the 550,00 new businesses in the U.S. in 2002 required a capita investment to get started. How did the founders nance these initia requirements? The theory of nancia structure of the rm has been one of the centra themes of corporate nance since Modigiani and Mier s (958) irreevance resut. This paper addresses this theme in the context of sma business: it anayses optima contracting for the entrepreneur. The internet boom of the ate 990s spurred much research on the nancing of new ventures. 3 That iterature has focused mainy on venture capita (VC) nancing, 4 and has o ered many insights into the use of convertibe preferred equity and its ubiquity in VC deas in the U.S. 5 However venture capitaists nance ony a sma fraction of entrepreneuria ventures: Davis (2003) reports that venture capitaists nance ess than 0 percent of startups in Canada, and that in 2000 Canadian venture capita represented 4 percent of the doar investment in sma rms. Moreover, the prevaence of convertibe preferred equity in VC deas appears to be particuar to the U.S. In Europe in genera, The O ce of Advocacy of the U.S. Sma Business Administration (SBA) de nes a sma business as an independent business having fewer than 500 empoyees. 2 See SBA (2003). These statistics can be obtained from the frequenty asked questions on how important are sma businesses to the U.S. economy? at 3 See, for exampe, Bascha and Waz (200), Bergöf (994), Bergmann and Hege (998), Casamatta (2003), Cornei and Yosha (997), Gompers (997), Gompers and Lerner (999), Hemann (2000), Kapan and Strömberg (2003), Marx (998), Repuo and Suarez (998), Schmidt (2002), Trester (998), Yerramii (2004). 4 Four exceptions come to mind which do not focus on convertibe preferred contracts. Garmaise (200) show that when investors are better informed than entrepreneurs, the entrepreneur tends to prefer junior equity to debt. Landier (2002) deveops a mode where the choice between bank debt and venture capita nancing depends on the entrepreneur s exit option: a good (bad) exit option tends to favor venture capita nancing (bank debt). Ueda (2002) argues that the tradeo between bank and venture capita nance is that athough banks are better at project evauation, they are more ikey to expropriate the entrepreneur. Finay, Winton and Yerramii (2004) show that both uncertainty in continuation strategy choices, and skewness of cash ow distributions (ow probabiity of success, high returns when successfu), tend to favor the use of convertibe debt instead of pure debt. 5 Evidence of the prevaence of convertibe preferred equity in VC nancing in the U.S. can be found in Bergmann and Hege (998), Gompers (995, 997), Gompers and Lerner (999), Kapan and Strömberg (2003), Lerner (994), and Sahman (990). 2

3 as we as in speci c countries such as Canada, Germany, Finand, Taiwan, and Austraia, 6 evidence points to the use of a variety of securities. The main contribution of this paper is to o er a unifying theory of the entrepreneur/investor reationship, which can be appied to entrepreneuria ventures in genera rather than speci cay to VC nanced ventures, and where the di erent types of contracts commony used in new venture nancing, such as straight debt, common equity, and preferred equity, emerge as potentiay optima. Entrepreneuria ventures have distinctive characteristics. i) They are subject to far ess restrictive discosure aws reative to arge, pubicy hed corporations, and their income and accounts are not easiy veri abe by a court of aw. ii) Both the entrepreneur and the investor may pay an active roe in the management of the venture, but these investments in e ort are di cut to contract upon. The nonveri abiity of pro ts impies non-contractabiity: parties do not want to make contracts contingent on pro ts or cash ows as they woud have no recourse in court in the event of disagreement. The contract signed at the beginning of the game thus ony speci es the aocation of contro rights over the venture. As argued in Grossman and Hart (986) and Hart and Moore (990) (GHM), contro rights - board rights, voting rights, veto rights, iquidation rights, etc. - pay an important roe because they confer bargaining power in the negotiation over the pro ts generated. This in turn provides anticipating agents with incentives to invest in e ort. In our mode, for exampe, Entrepreneur-contro aocates a contro rights and ownership rights to the entrepreneur; in that case the entrepreneur has a bargaining power and can extract a rent in negotiation. She 7 thus has high incentives to exert e ort. The investor on the other hand anticipates he wi get nothing and does not participate in the management of the venture. Conversey, investor-contro gives a contro to the investor: he exerts high e ort whie the entrepreneur does not participate. Finay, joint-contro aocates enough contro rights to the investor to provide him with some bargaining power, and consequenty with the 6 See Cumming (2002), and the references cited therein. 7 Throughout the paper we refer to entrepreneurs as femae, and to investors as mae. 3

4 abiity to extract some rents. The investor may for exampe have enough contro to in uence the entrepreneur s use of rst-cass airine tickets or her access to the company car. As ong as the investor can interfere with the entrepreneur s abiity to enjoy the surpus, he can extract a fraction of that surpus by threatening interference. With joint contro, both the investor and the entrepreneur participate in management, but anticipating that they wi have to share the surpus, they have ow incentives and exert ow e orts. We show that when the entrepreneur is weath constrained, 8 this simpe framework generates severa resuts:. Ony four types of contract are potentiay optima for the entrepreneur to o er to the investor. 9 Debt with iquidation assigns a contro rights to the entrepreneur, which revert to investor-contro in case of defaut on the debt repayment. Debt with reorganization is simiar to the rst contract except that defaut eads to joint contro. The entrepreneur may aso o er a simpe joint contro contract to the investor. This aternative, despite non-contractibiity of pro ts, yieds an expected cash ow to the investor which ooks much ike that of equity: the investor has enough contro to interfere with the entrepreneur s abiity to enjoy the rents, and can thus extract a fraction of the surpus. Therefore we ca this equity-ike nancing. Finay, the entrepreneur may o er a contract which assigns joint contro conditionay on debt repayment, with defaut resuting in investor-contro. This contract ooks very much ike the preferred equity contracts observed in practice. For this reason we ca it the preferred-type contract. In the paper we distinguish between two sub-categories: straight preferred-type contracts and convertibe preferred-type contracts. 2. The optima contractua form is mainy determined by two factors: a) agents motivationa margina e ciency reative to participatory margina e ciency, or convexity of output with respect to incentives; 0 and b) the investor s opportunity cost of capita. We nd that for ventures in which 8 If the entrepreneur were not weath constrained, then, as shown in GHM, the optima aocation of contro rights woud be the one which maximizes tota surpus. See our discussion in section 5. 9 The entrepreneur initiay has a the bargaining power in negotiation: she makes a take-it-or-eave-it o er to the investor. 0 The three contro rights aocations yied three types of incentives for the agents: a) no incentives (zero expected payo ): in that case they don t participate; b) ow incentives: enough to participate, but ow e ort exerted; c) high incentives: participation and high e ort exerted. We de ne participatory margina e ciency as the increase in output 4

5 motivation - rather than participation - matters, debt with iquidation is optima. In contrast, when participation - rather than motivation - matters, equity-ike nancing dominates when cost of capita is ow, whie the preferred-type contract is optima when it is high. 3. We use the resuts of the mode to address interesting issues in entrepreneuria nance: the prevaent use of debt contracts to nance ifestye venture, and of equity contracts (common equity, straight and convertibe preferred equity) in cassic startup ventures; investors outside options, investor sophistication, and tempora/geographica di erences in VC contracts. Much of the work in the security design 2 iterature anayzes rms which share some of the characteristics described in the third paragraph of this introduction. As noted in Fuck (998), this iterature su ers from the drawback that it cannot simutaneousy assume characteristics i) and ii) 3 and incorporate outside equity nancing. Fuck soves this probem by arguing that equity-hoders who have the right to dismiss the manager, and have an unimited time horizon, can discipine the manager into paying out dividends with a credibe threat of dismissa. In contrast, our contribution to that iterature is the foowing. We get around Fuck s critique by arguing that in entrepreneuria ventures the crucia components of equity contracts are contro rights, such as board rights, voting rights, veto rights, etc. which are typicay associated with it. These contro rights give ex-post bargaining power to the investor, who is thus abe to extract some rents. We incorporate equity not directy through its caim on cash ows 4, but indirecty through the contro rights associated with it, and the bargaining power that they confer in negotiation. Like Fuck, Myers (2000) and Dybvig and Wang (2002) aso use the threat of dismissa to introduce equity in a mode with non-veri abe cash ows. These two resuting from a switch from a) to b); and motivationa margina e ciency as the increase in output resuting from a switch from b) to c). Lower participatory margina e ciency reative to motivationa margina e ciency impies a more convex output as a function of incentives. In other words, ventures in which agents motivationa margina e ciency is high reative to their participatory margina e ciency: output is highy convex with respect to incentives. 2 See for exampe Aghion and Boton (992), Boton and Scharfstein (990, 996), Chang (992), DeMarzo and Fishman (2000), Dewatripont and Tiroe (994), Diamond (984), Gae and Hewig (985), Hart and Moore (998), Fuck (998), Pove and Raith, (2003), Townsend (979), Wiiams (989). 3 We refer the reader to the third paragraph. 4 This caim is irreevant when cash ows are not veri abe. 5

6 papers are somewhat coser to ours, in that they take incentives into account. Dybvig and Wang in particuar compare debt and equity and argue that whereas debt generates higher (e cient) e ort exertion, it aso give the manager incentives to defaut. Our paper however di ers not ony in context - our focus is on entrepreneuria ventures where the investor, as we as the entrepreneur, exert e ort - but aso in modeing structure, contracts, and resuts. Dybvig and Wang s tradeo described above, for exampe, does not necessariy hod in our mode, where tota investment in e ort may be higher in with equity than with debt. Our modeing structure buids on the work of Boton and Scharfstein (990, 996) and Hart and Moore (998) (BSHM), as we as Aghion and Boton (992). We show that resuts can be quite di erent if we introduce a key variabe that is absent in their modes: e ort. Gertner et a. (994) aso introduce e ort in a nancing mode. The two modes remain quite dissimiar, 5 however, and the papers address two very di erent questions: their focus is on the tradeo between interna and externa nancing choices for the rm, whereas ours is on optima security design for the entrepreneur. The paper is structured as foows: section 2 presents the basic mode. Section 3 describes the contractua choices for the investor, and section 4 anayzes optima contracts. Section 5 discusses the main empirica predictions of the mode, and concudes. 2 Mode 2. Basic Setup Consider a weath constrained entrepreneur who has a two-period, positive net present vaue project in mind, which requires an initia investment k. The timing of the game is as foows. At date 0, the entrepreneur makes a take-it-or-eave-it contractua o er to an investor 6. The investor s next- 5 For exampe, whereas they focus on two contro aocations (investor contro and entrepreneur contro), we aow for a third feasibe aocation, joint contro, which enriches consideraby the set of contractua possibiities. 6 We impicity assume that there are many more nanciers wishing to invest than there are good entrepreneurs (good projects to be funded). 6

7 best opportunity yieds an expected net return r over two periods. We ca r the investor s (doar) opportunity cost of capita. If the investor accepts the entrepreneur s o er, then at date, the project produces a payo x with probabiity p, and zero payo with probabiity p. Both x and p are exogenous. At the beginning of the second period, the entrepreneur decides whether to ) quit the venture, 2) stay invoved and exert ow e ort e at cost c (e ) = c, or 3) stay invoved and exert high e ort e h at cost c (e h ) = c h, with c h > c. Simiary, at the same time the investor decides whether to quit the venture, stay invoved and exert ow e ort f at cost c, or stay invoved and exert high e ort f h at cost c h. Whie the entrepreneur s e ort is manageria and technoogy improving in nature, e ort by the investor mainy takes the form of manageria hep to the entrepreneur (genera manageria guidance, marketing, access to incubator services, etc.). Both e and f are assumed to be non-veri abe - and thus non-contractibe - investments embodied in physica (rather than in human) capita. 7 Moreover, e h > f h and e > f. This assumption distinguishes the entrepreneur from the investor. Whie the entrepreneur devotes a her time to the project and has no other activities, the investor typicay is invoved in severa ventures at the same time and thus has a higher opportunity cost of e ort. Equivaenty, for an equa persona cost, he exerts ess e ort than the entrepreneur. At date 2, the project yieds a positive payo v (e; f) = e + f, which depends on the investments made at the beginning of the second period. We make this separabiity assumption for simpicity. Simiar resuts hod when technoogy is not additivey separabe. 7 Investments in human capita increase the size of the end-of-period payo ony if the agent remains invoved in the venture. If the agent eaves the venture, he/she takes his/her human capita with her. In contrast, investments embodied in physica capita increase the size of the end-of-period payo whether or not the agent remains invoved with the venture. The reason for our focus on physica capita investments shoud become cear beow. 7

8 2.2 Incompete Contracts and Nash Bargaining We assume that payo s x and v are not veri abe by a court (though they are observabe to entrepreneurs and investors), and thus cannot be contracted upon. To quote Boton and Scharfstein (996, p.5), this assumption is meant to capture the idea that managers have some abiity to divert corporate resources to themseves at the expense of outside investors, and that such perk consumption and investment may be di cut to distinguish from appropriate business decisions and thus impossibe to contro through contracts. Because of non-contractabiity, at the end of each period the entrepreneur and the investor bargain over reaized payo s. Let z (e; f) be the tota payo reaized if negotiations succeed (i.e. x or v (e; f)). Let z e (e; f) and z f (e; f) denote the payo s to the entrepreneur and investor, respectivey, shoud bargaining negotiations break down, i.e. their outside option. 8 In equiibrium the negotiations aways succeed and the entrepreneur keeps tota payo z (e; f), minus an equiibrium transfer t (the Nash bargaining soution), de ned as: t 2 arg max [z (e; f) t z e (e; f)] t z f (e; f), or t = 2 z (e; f) ze (e; f) + z f (e; f). Bargaining in period Two striking features characterize the entrepreneur in the eary stages of a new venture. First, the entrepreneur is indeed indispensabe to the success of the venture. If this were not the case, many entrepreneurs woud simpy se their ideas to investors, and move on to another project. Second, the entrepreneur s indispensabiity typicay comes from her human capita, e.g. her idea, and unti speci c investments are made, her idea tends to be highy portabe. For that reason investors try to incude goden handcu s covenants in nancing contracts. These covenant however cannot perfecty immobiize the entrepreneur, who coud, in many cases, take her human capita esewhere. To re ect these features of eary stage ventures in a stark way, we assume that in period, a) the entrepreneur is absoutey indispendabe to the venture, in the sense that the investor cannot generate 8 In the rst period there are no investments e and f, so z (e; f) = z = x, z e (e; f) = z e and z f (e; f) = z f. 8

9 any rents on his own, i.e. z f = 0; and b) the potentia payo x is based entirey on the entrepreneur s human capita, and coud be repicated esewhere at no extra cost, i.e. z e = x. These assumption have the advantage of making rst period bargaining very simpe, as they impy t = 0: the entrepreneur aways extracts the entire payo x at date. 9 Investments and Bargaining in Period 2 The indispensabiity and portabiity of the entrepreneur s human capita become much ess important in the medium run, as two changes graduay occur. First, as time goes by the entrepreneur s human capita becomes more and more speci c to the venture, and hence ess and ess portabe. Second, the success of the venture, as we as the abiity to extract rents from it, starts to depend reativey more on physica assets and access to them. In period 2 in our mode, this access is determined by the aocation of contro rights, which can be of three types. Joint Contro. With this aocation of contro rights, the investor contros enough board rights, voting rights, inspection/monitoring rights, etc., such that used in combination, these rights provide him with some bargaining power. Despite the fact that perk consumption by the entrepreneurs is not veri abe in court, these contro rights may enabe the investor to interfere with it. Even the abiity to interfere ony sighty with the entrepreneur s perks gives the investor bargaining power, and enabes him to extract a part of the surpus. For carity purposes, we make the strong assumption that joint contro gives the investor enough contro rights for maximum interference: the investor can bock competey the entrepreneur s access to perks. If he does that, however, trade breaks down, and he can get nothing either. More formay, z e (e; f) = z f (e; f) = 0. Under Nash bargaining, this eads to the two parties spitting the payo in equiibrium: the entrepreneur and the investor each obtain 2v (e; f) at date 2. 9 If instead of assuming that z e = x, we assumed that the entrepreneur coud not exacty repicate x, and that her outside option were 0 < z e < x, a resuts of the mode woud sti hod, despite the fact that date bargaining woud ead to x being spit between the two parties. If we assume that the entrepreneur is not indispensabe in the rst period, the mode predicts what we woud expect to see in practice: the entrepreneur woud se the venture to the investor (perhaps with equity vesting in the second period). 9

10 Entrepreneur-Contro gives the entrepreneur compete authority over the venture, and the investor cannot interfere with her access to rents. If negotiations break down, the entrepreneur continues to enjoy a rents (z e (e; f) = V (e; f)), whie the investor gets nothing (z f (e; f) = 0). This eads to a bargaining equiibrium in which the entrepreneur gets a of the rents and the investor gets nothing at date 2. Investor-Contro is the ip-side of entrepreneur-contro: the nancier has 00% of the contro rights over the venture and fu bargaining power in negotiation: z f (e; f) = z (e; f) and z e (e; f) = 0 impy that the investor gets a of the rents, and that the entrepreneur gets nothing at date 2. Stricty Dominating Strategies. For simpicity, we focus on vaues of f, f h, e, e h, c, and c h, such that there exists a stricty dominating strategy for each payer. Ceary the entrepreneur wi exit the venture under investor contro: she anticipates no reward at date 2, and even ow participation e ort is costy. Simiary, the investor wi exit under entrepreneur contro. It is natura to expect that for both the entrepreneur and the investor, the higher the expected fraction of the surpus, the higher the equiibrium eve of e ort. This impies: Entrepreneur-contro Joint contro Investor-contro Entrepreneur High e ort Low e ort No participation Investor No participation Low e ort High e ort Date 2 Exp. Output V Y = v (e h ; 0) c h V Q = v (e ; f ) 2c V L = v (0; f h ) c h Thus, with joint contro the entrepreneur and the investor must share the surpus, and even though they both participate, their individua incentives are ow and they exert ow e ort. With entrepreneurcontro, a ex-post bargaining power is transferred to the entrepreneur: he has high incentives and exerts high e ort, whie the investor does not participate. Conversey with investor-contro, the investor has high incentives and exerts high e ort, but the entrepreneur does not participate. More formay we focus on vaues of f, f h, e, e h, c, and c h, such that: ) 2 v (e ; f) c > 2 v (e h; f) c h ; 0

11 2) 2 v (e; f ) c > 2 v (e; f h) c h ; 3) 2 v (e ; f) c 0; 4) 2 v (e; f ) c 0; 5) v (e h ; f) c h > v (e ; f) c ; 6) v (e; f h ) c h > v (e; f ) c : Conditions -4 refer to joint contro. They impy rst that ow e ort is the optima choice both agents (cond.,2); and second they impy that both agents choose to participate in the venture (cond. 3,4) in the second period. Conditions 5 and 6 refer to entrepreneur-contro and investorcontro, respectivey. They impy that whenever an agent has fu contro over the venture, he/she has an incentive to make the arge investment. Conditions 3 and 4 aso impy that participation of the entrepreneur with entrepreneur-contro, and of the investor with investor-contro, are optima in equiibrium. There exists a (arge) set of vaues for parameters f, f h, e, e h, c, and c h, such that conditions )-6) in hod. We show in the appendix that for given vaues of f, f h, c, and c h, with f h > 2c h, there exist variabes e min f 2 f min ; e max ; f min ; and f max ; f max. We de ne VQ min = v e min V Q 2 (V Q min ; V Q max ) such that conditions )-6) hod for a e 2 e min ; f min ; e max and 2c, and V Q max = v(e max ; f max ) 2c. Then, Note that due to entrepreneuria superiority, we must have V Y > V L, and thus tota output is aways higher with entrepreneur-contro than with investor-contro. Note aso that here joint contro may we generate the argest tota output, provided V Q V Y. This is unike the GHM standard framework in which joint contro is never an e cient contro aocation. Our resut comes from our assumption that e and f represent investments embodied in physica capita. 20 If instead we assumed investments in human capita, as in GHM, then entrepreneur contro woud aways dominate joint 20 This point was made previousy by Hart (995, p. 68). Recenty, other authors have described situations in which joint e ort provision may be optima. See for exampe Casamatta (2003), De Meza and Lockwood (998), Haonen (2002), Rajan and Zingaes (998), Yerramii (2004).

12 contro. 2 3 Contractua Choices for the Entrepreneur Let us describe the rst-best (FB) scenario as benchmark. First-Best Outcome. In the FB, both the entrepreneur and the investor exert e ort to maximize the socia surpus: Max e;f px + v (e; f) c e (e) c f (f) : Conditions 5) and 6) in assumption 3 impy that that the rst-best eves of e ort for the entrepreneur and the investor, respectivey, are e (v) = e h, and f (v) = f h. The resuting equiibrium socia surpus can be expressed as: R F B = px + V F B, where V F B = v(e h ; f h ) 2c h. 3. Debt with Liquidation BSHM have shown that when investor-contro and entrepreneur-contro are the ony two feasibe contro right aocations, the optima contract assigns entrepreneur-contro if a debt repayment d is paid out to the investor at date ; and investor-contro in the event of defaut. 22 This debt contract - which we denote debt with iquidation (DL) because in the event of iquidation the investor obtains fu contro and either manages the venture himsef or iquidates it - is sti potentiay optima in our framework. The expected return for the entrepreneur, from a date 0 perspective, foows directy from the form of the contract and from the above anaysis: R e DL = p (x d + V Y ). If the projects was successfu in period and generated payo x (which occurs with probabiity p), the entrepreneur pays d to the investor, keeps x d for hersef, and expected net payo V Y at date 2. I. If the project was not successfu in the rst period and no cash was generated, the entrepreneur must defauts and gets nothing. The expected return for the investor is R f DL = pd + ( p) V L k. In the good state 2 This is easiy shown by repacing z f (e; f) by z f (f), and z e (e; f) by z e (e) in the Nash bargaining soution. 22 The repayment d is assumed to be veri abe. 2

13 he receives d from the entrepreneur, and in the bad state he obtains contro of the assets, with an expected payo V L at date 2. To fuy characterize DL, we must determine a) the equiibrium vaue of d, and b) when this contract is feasibe. Since by assumption the entrepreneur makes a take-it-or-eave-it o er to the investor at date 0, whenever possibe she wi set d such that R f DL = r, i.e. such that the investor is indi erent between investing or not. However d must be ow enough to be incentive compatibe. Indeed the entrepreneur may be tempted to strategicay defaut in the good state, anticipating that since iquidation is ine cient (V Y > V L ), renegotiation woud take pace. Under Nash bargaining, the renegotiation outcome woud be a payment s DL = 2 V L + 2 V Y from the entrepreneur to the investor. 23 For the debt payment to be incentive compatibe, we must thus have d s DL ; 24 s DL is the maximum debt repayment that the entrepreneur can commit to make. in other words, It then foows that the maximum return that the investor can expect to receive is R f DL (max) = p 2 V L + 2 V Y + ( p) (V I f I ) k. Thus, as ong as the investor s opportunity cost r of investing in the venture is sma enough for R f DL (max) r to hod, there exists an equiibrium debt repayment d s DL such that R f DL = r, and the entrepreneur extracts a rents from a date 0 perspective: R e DL = p (x + V Y ) + ( p) V L k r. 25 If r is so arge that R f DL (max) < r, then DL is not feasibe. Note that this contract is ony feasibe if and ony if (henceforth i ) 2 v (e ; f ) c v (0; f h ) c h, i.e. i V Q 2V L ; otherwise, renegotiation woud aways occur in the bad state: the investor woud aways propose to switch from investor-contro to joint contro, and the entrepreneur naturay woud accept. To avoid this possibiity and keep the mode as simpe as possibe, we assume that 2V L V Q max. The consequences of reaxing this assumption are discussed in the appendix. In contrast to BSHM, in this paper entrepreneur-contro and investor-contro are not the ony two 23 Using the Nash bargaining soution, s DL = arg max (V Y z) (z V L). 24 Otherwise it is in the interest of the entrepreneur to defaut, renegotiate, and pay out s DL. 25 In the specia case when k and r are so ow that ( p) V L k + r, the entrepreneur set d = ", with "! 0, and requests a transfer t 0 from the investor at date 0 such that ( p) V L k t 0 = r. 3

14 possibe aocations of contro rights. Indeed, joint contro is aso avaiabe here, and as a resut three other types of contract are potentiay optima for the entrepreneur. 26 We anayze each one in turn. 3.2 Debt with Reorganization Another possibe debt contract is to have entrepreneur-contro if the entrepreneur makes the debt repayment d at date, and joint contro if she defauts. In that case the debt contract is in some sense converted into equity in bankruptcy, and reorganization occurs, with both agent sharing contro. We ca this debt with reorganization (DR). 27 This contract is very simiar to the debt with iquidation contract, the main di erence being the aocation of contro rights in bankruptcy. This di erence in contro rights a ects the entrepreneur s bargaining power in renegotiation, and hence the maximum payment she can commit to repay at date. Here the entrepreneur can commit to pay at most 28 s DR = 2 V Y, and the maximum return the investor can receive is: R f DR (max) = p 2 V Y + ( p) 2 V Q If r is such that R f DR (max) r, there exists an equiibrium d s DR such that R f DR = r, and the entrepreneur extracts a rents from a date 0 perspective: R e DR = p (x + V Y ) + ( p) V Q k r. 29 If r is so arge that R f DR (max) < r, then DR is not feasibe. k. 26 In addition to these three contracts, severa other types of contracts may be feasibe. As shown in the appendix, however, these other contracts are a suboptima: they are aways dominated by one of the contracts presented in the text. 27 The distinction we make between debt with iquidation and debt with reorganization is not as cear in the rea word. Indeed most contracts tend to ook ike the DL contract in that in the event of defaut the investor takes contro over the assets. Ony after taking contro the investor decides whether to a) manage the venture himsef or iquidate it, or b) reorganize, keep the entrepreneur on board, and manage the venture jointy. We coud thus have set up the probem with one type of debt contract, which in the bad state sometimes eads to iquidation, other times to reorganization. From a date 0 perspective, the two ways to present are identica: the date payment d aways re ects the agents expectation about whether iquidation or reorganization wi occur. We present the two contracts separatey for carity purposes. 28 With Nash bargaining, we have: s DR = arg max (V E e E) z 2 VJ ej z 2 VJ fj, which yieds s DR = [(VE ee) + (ej fj)] In the specia case when k and r are so ow that ( p) VQ k r, the entrepreneur set d = ", with "! 0, and 2 requests a transfer t 0 from the investor at date 0 such that ( p) VQ k t0 = r. 2 4

15 3.3 Preferred-Type Contracts: Redeemabes and Convertibes Preferred-type (P) contracts assign joint contro conditionay on a prespeci ed debt payment d at date ; and investor-contro in the event of defaut. If the entrepreneur strategicay defauts in the good state of the word, renegotiation occurs, and in equiibrium she transfers an amount 30 s P = 2 V L to the investor. This is thus the maximum debt payment d she is wiing to forfeit at date, and consequenty the most the investor can receive is: R f P (max) = p 2 V Q + 2 V L + ( p) VL k. If r is such that R f P (max) r, there exists an equiibrium d s P such that R f P = r; and the entrepreneur extracts a rents and expects R e P = p [x + V Q]+( p) (V L ) k r. 3 In the specia case when p 2 V Q + ( p) VL k > r, the entrepreneur sets d = ", with "! 0, and requests a transfer t 0 from the investor at date 0 such that p 2 V Q + ( p) VL k t 0 = r. Finay, if r is so arge that R f P (max) < r, then P is not feasibe. Note that, as in the case of DL, and for the same reasons, the P contract is feasibe i V Q 2V L. This is aways the case when, as assumed, 2V L V Q max. There is a cear resembance between this contract and the two main categories of preferred equity nancings observed in practice: Redeemabe ( straight ) preferred contracts (SP) typicay specify the redemption vaue of the investment (say d), the redemption date (say date ), and the amount of common stock to be issued in combination with the preferred stock. This common stock gives the investor some rights over the future cash ows, in addition to the pre-speci ed redemption vaue; this is the so-caed doube-dipping. If the company cannot make the redemption payment, the assets are iquidated, with the proceeds accruing to the investor rst ( iquidation preference ). Convertibe preferred contracts (CP) give the investor the choice between redeeming his stock at the pre-speci ed redemption vaue d, and converting it into common stock. In the bad state, the investor does not convert, thus forcing iquidation and making use of the iquidation preference attached to his 30 With Nash bargaining, we have: s P = arg max 2 VQ z 2 VQ + z VL, which yieds sp = VL. 3 2 In the specia case when k and r are smaer than p 2 VQ + ( p) VL, the entrepreneur set d = ", with "! 0, and requests a transfer t 0 from the investor at date 0 such that p 2 VQ + ( p) VL k t 0 = r. 5

16 security to extract as much out of the iquidation vaue as possibe. In contrast, in the good state he converts, provided that the caim on a fraction (say 2 ) of the proceeds he can get with the common stock after conversion is higher than the redemption vaue. The reason why we ca this contract the preferred-type contract is this: for sma investments, when k is so ow that in equiibrium d tends to zero, our contract mimicks the payo of convertibe preferred stock. In contrast, when k is arger and the equiibrium date payment d is stricty positive, our contract mimicks the payo of redeemabe preferred stock: doube-dipping occurs as the payment d is foowed by joint-contro and a sharing of the surpus. Thus we ca this preferred-type contract convertibe preferred when equiibrium d tends to zero, and redeemabe (or straight) preferred when d > Equity-Like Financing This contract assigns joint contro unconditionay for the second period. Even though in our mode an equity contract per se woud not yied any payo to the investor (since by assumption cash ows are not veri abe, the entrepreneur woud divert them), joint contro (with equity) generates a stream of payo s to the investor which is simiar to the one typicay obtained in a standard equity contract. For that reason we ca this equity-ike nancing (E). Focusing on the contro rights associated with equity (joint contro) rather than the cash ow rights, 32 we are abe to introduce an equity-ike contract in the mode and to compare with other contracts, a the whie keeping our assumptions of non-veri abiity of cash ows and performance, necessary for debt to be emerge as potentiay optima. Contract E is very simpe: as ong as the investor s opportunity cost of investing, r, is sma 32 In an interesting study of venture capitaists, Kapan and Strömberg (2003) found evidence that the contro rights aocated to venture capitaists through covenants are independent of the nancia contracts o ered. One may thus argue that this impies that contro rights cannot be used as a good instrument for equity since the two are assigned independenty. However Kapan and Strömberg ony ook at entrepreneuria ventures nanced with venture capita, and thus where private equity is used. They do not ook at entrepreneuria ventures in genera, incuding the ones nanced with debt by commercia banks. We argue that even if contro rights are separatey aocated within the cass of VC equity nancings observed by Kapan and Strömberg, on average the VC has much ess contro than woud a creditor in a debt contract foowing defaut, and much more contro than the creditor in a debt contract where the contracted debt repayment has been made. These are the important di erences in our mode. 6

17 enough for R f Q (max) = 2 V Q k r to hod, the entrepreneur can give joint contro to the investor in exchange for (k and) a date 0 transfer t 0 such that the investor s expected return exacty equas r, i.e. such that R f Q = 2 V Q k t 0 = r. The entrepreneur extracts a ex ante surpus and expects R e Q = px + V Q k r. If r is such that 2 V Q k < r, then contract E is not feasibe. We show in the appendix that these 4 types of contract, namey debt with iquidation, debt with reorganization, preferred-type contracts, and equity-ike nancing are the ony ones which can potentiay be optima. In this section we have determined the optima contractua form within each type. We now determine which type is optima. 4 Optima Contracting Determinants of the Optima Contract Convexity of Output as a Function of Incentives: Motivation Versus Participation. Let us treat e h, f h, and c h as constant. A change in e can be interpreted as a change in participatory margina e ciency reative to motivationa margina e ciency for the entrepreneur. When contro rights switch from investor contro to joint contro, bargaining power is transferred from the investor to entrepreneur, from no bargaining power at a (investor contro) to some bargaining power (joint contro). Consequenty the entrepreneur s incentives increase enough for her to participate, but not enough to be highy motivated, and her e ort/output, net of e ort cost, rises from 0 (no participation) to e c (participation with ow e ort). The entrepreneur s participatory margina e ciency in that case is (e c ) 0 = e c. When the aocation of rights switches from joint contro to entrepreneur contro, more bargaining power is transferred from the investor to the entrepreneur. The entrepreneur s incentives increase even more and she becomes highy motivated: her motivationa margina e ciency is (e h c h ) (e c ). 7

18 Thus, for a given e h, c h, and c a ow (high) e impies a ow (high) participatory margina e ciency reative to a high (ow) motivationa margina e ciency for the entrepreneur. Simiary for the investor, for a given f h, c h and c, a ow (high) f impies a ow (high) participatory margina e ciency, (f c ), reative to high (ow) motivationa margina e ciency, (f h c h ) (f c ). We measure this reative margina e ciency for each agent with the vaues of e and f, and consequenty with V Q = e +f 2c. 33 We divide the set of vaues for V Q 2 (V Q min ; V Q max ) into three regions, represented on the vertica axis in gures and 2. Region : V Q 2 [V Q min ; V L ). When V Q is sma, the entrepreneur and the investor do not create much vaue by merey participating in the venture with ow incentives. Much more vaue can be added (comparativey), by motivating them with high incentives. Thus, when V Q is sma, output as a function of incentives is highy convex, 34 and we say that motivation, rather than participation, is the primary driver of success in the venture. Firms in region are motivation ventures. Region 3: V Q 2 [V Y ; V Q max ). At the other end of the spectrum, when V Q is arge, the entrepreneur and the investor create a ot of vaue even with ow incentives, simpy by participating in the venture. Motivating them with high incentives yieds itte more vaue added. In that case output is a highy concave function of incentives, and we say that participation, rather than motivation, is what reay drives success in the venture. Firms in region 3 are participation ventures. Region 2: V Q 2 [V L ; V Y ). Finay, in between regions and 3 is, the intermediate region where output is neither highy convex nor highy concave. Opportunity cost of capita, and ex-ante versus ex-post (in)e ciency. We compare contract A s (A = DL, DR, E, P ) tota expected payo - net of cost of e ort and initia capita requirement - to a pie. Note that: a) since the entrepreneur aways extracts a expected payo s from 33 Keeping V Y = e h c h, V L = f h c h, and c, xed, the ower V Q, the higher (e h c h ) (e c ) reative to (e c ), and the higher (f h c h ) (f c ) reative to (f c ). 34 We (perhaps crudey) use convexity to refer to reative margina e ciencies, even though the output function is ony de ned in three points. We coud convexify the function by inking the three points. In that case, moving the midde point up (down) corresponds to making the convexi ed function ess (more) convex. 8

19 a date 0 point of view, the entrepreneur s expected return RA e is an appropriate measure of the pie associated with contract A. b) Contract A yieds a pie RA e that is smaer than the FB pie, R F B. Conditionay on being nanced, the contract does not yied the highest possibe overa payo ; and this so-caed ex-post ine ciency is denoted XP A = R F B RA e. The contract yieding the argest pie is ex-post optima. The part of the tota pie that the investor receives, R f A, must be superior to his opportunity cost of capita r for him to nance the venture. However, R f A (max) < R F B, and hence some projects which shoud be nanced because R F B > r, are not undertaken because R f A (max) < r. Fewer projects are nanced reative to the rst-best, and this ex-ante ine ciency associated with contract A is denoted XA A = R F B R f A (max). Two factors a ect our measure of ex-ante e ciency, Rf A (max) = RA e R f A (max). i) The pie e ect, or ex-post e ciency e ect, RA e, simpy means that ceteris R e A paribus, the arger the pie, the arger the part of it that the entrepreneur can give to the investor. ii) The sice e ect, RA e R f A (max), represents the smaest sice of the pie that the entrepreneur can commit to keep to hersef in renegotiation. The smaer that sice, the arger the part of the pie that the investor can expect to get, and and the more ikey that he wi agree to nance the project. The opportunity cost of capita r pays an important roe in determining the importance of ex-post versus ex-ante e ciency. When the investor s opportunity cost of capita is ow, R f A (max) is ikey to be superior to r for many di erent types of contracts, and ex-post e ciency matters most: the entrepreneur chooses amongst the many feasibe contracts the one which yieds the argest pie. In contrast, when the opportunity cost of capita is high, ony very ex-ante e cient contracts are feasibe, and must be chosen even if they are ex-post ine cient. In other words, the pie and sice e ects may work in opposite directions: the optima contract may yied a sma pie, but may aow the entrepreneur to commit to keep ony a very sma sice, eaving enough to the investor for the project to be nanced. Graphica Representation of Optima Contracts. To understand how the optima contract depends on convexity of output and the opportunity cost of capita, we anayze how r a ects optima 9

20 contracting in each region in turn. Figure (at the back of the paper) iustrates our resuts by depicting optima contracts in a (r; V Q ) space. 35 Each ine represented on the gure represents an iso-return ine for the investor, drawn at R f A (max) = 0: it is the ocus of points for which that contract yieds a zero return to the investor. A points to the eft of that ine yied a positive return to the investor, and a points to the right of the ine yied a negative return. It is easy to understand which contract is optima from gures and 2: at each point on the gure, the optima contract maximizes the entrepreneur s return among a contracts whose iso-return ine for the investor is to the right of that point. (A the resuts of this section are aso proven agebraicay in the appendix.) Region : V Q 2 (V Q min ; V L ) - Motivation Ventures In motivation ventures, joint contro yieds a comparativey sma output. Mere participation in the venture with ow incentive is reativey unproductive; but much vaue can be added by motivating the agents with high incentives. As a consequence, contracts which provide ow incentives (by assigning joint contro), such as E, DR, and P, tend to yied a sma pie. In contrast, DL never assigns joint contro: with high incentives, it motivates the entrepreneur in the good state and the investor in the bad state, respectivey. For that reason, DL yieds the argest pie and is ex-post optima: it is the entrepreneur s preferred choice as ong as R f DL (max) r. Figure shows that DL is aso ex-ante optima: its iso-return ine is the furthest to the right in region. As a resut, there are no projects that coud not be nanced with DL, but that coud be nanced with another contract. DL makes the project most ikey to be nanced as we as yieding the argest pie: hence it must be optima contract for a points in region. DL s ex-ante optimaity can be expained in terms of pie and sice e ects. We iustrate this by comparing maximum returns 35 Note that gure is drawn for a probabiity of date success p =. We do this for simpicity but the resuts of 2 the paper hod for a 0 < p <, uness stated otherwise. Note aso that gure exhibits a reativey high vaue of investor-contro payo, compared to that of the entrepreneur-contro payo : 2V L > V Y. This is the most interesting case because the convertibe preferred contract can be optima under this condition. 20

21 to the investor for DL and DR. The di erence R f DL (max) Rf DR (max), can be expressed as: [( p) (V L V Q )] ( p) 2 V Q + p 2 V L > 0: () The rst square-bracketed factor represents the di erence in pie e ects, which here favors DL. In the bad state DR yieds a smaer tota pie than DL: V Q 2c < V L c h. In the good state the two contracts yieds the same expected tota output, and the di erence between the two cances out. The second square-bracketed factor represents the di erence in sice e ects, which aso favors DL. In the bad state, with DL the entrepreneur can commit not to interfere in period 2 by reinquishing a contros to the investor in case of defaut: the investor gets a of V L and there is no sice e ect. In contrast, with DR the entrepreneur cannot commit not to extract a sice 2 V Q of the tota pie, at the detriment of the investor. In the good state, with DL the entrepreneur can commit to keep a smaer sice in renegotiation, thus eaving more to the investor. The net sice e ect is 2 V L in favor of DL. This is reated to Nash bargaining. DL puts the entrepreneur into a worse bargaining position in renegotiation by worsening her situation if no agreement is reached (investor contro yieds a ower expected payo to the entrepreneur than does joint contro). Because of that inferior bargaining position, the entrepreneur has ess incentive to defaut and is ess ikey to renegotiate on the debt payment d: the maximum renegotiation-proof payment the entrepreneur can commit to repay at date through Nash bargaining is higher by an amount 2 V L. In region the pie and the sice e ects both favor DL, thus expaining why this contract is ex-ante optima. The main resut in region is the foowing: For ventures in which motivation - rather than participation - matters, debt with iquidation is ex-ante as we as ex-post optima. It is the optima contract regardess of the investor s opportunity cost of capita. Region 2: V Q 2 [V L ; V Y ) - Intermediate Region 2

22 In intermediate ventures, mere participation with ow incentives yieds a payo V Q that is sti too ow reative to that generated by high entrepreneuria incentives: V Q < V Y. Due mainy to this pie e ect, 36 contracts that provide high incentives to entrepreneurs (through entrepreneur contro) tend to dominate contracts that do not. Indeed we show in the appendix that DR and DL dominate E and P, respectivey, in terms of both ex-ante and ex-post e ciency. As a consequence, in region 2 ony DR and DL are potentiay optima. Low opportunity cost of capita. The pie e ect is aso the reason why DR is ex-post superior to DL. In the good state of the word, both contracts yied the same expected payo associated with entrepreneur contro. In the bad state however, DR eads to reorganization (joint contro), which in region 2 is superior to the iquidation associated with DL. Thus, debt with reorganization is the entrepreneur s preferred choice when R f DR (max) r. High opportunity cost of capita. Looking at gure, we see that DL is ex-ante superior to DR. Let us go back to condition (). The sign of the second square-bracketed term does not change in region 2: just as in region, DL has a sice e ect advantage over DR; it gives the investor a arger sice of the surpus/pie. In contrast, the sign of the rst square-bracketed term does change in region 2: it becomes negative. Our ex-post anaysis shows that here debt with reorganization yieds a higher overa surpus: it generates the argest pie. In other words, in region 2 the pie e ect and the sice e ect work in opposite directions: they favor di erent contracts. We show in the appendix that the sice e ect dominates in this region, 37 making DL ex-ante superior. It must be, therefore, that when r is too high for DR to be feasibe, DL becomes the optima contract. In sum: For intermediate ventures, the debt with reorganization contract is ex-post optima; it dominates when the investor s opportunity cost of capita is ow. Debt with iquidation is generay ex-ante optima, and dominates when the investor s opportunity cost of capita is high. 36 The sice e ect aso pays a imited roe in these resuts. See appendix for detais. 37 This hods under the assumption that 2V L V Q max. 22