Incentive problems and Reciprocal Behaviour in Venture Capital/entrepreneur Dyads: The effect of Inequity-aversion, Trust, and Social Norms

Transcription

1 Incentive problems and Reciprocal Behaviour in Venture Capital/entrepreneur Dyads: The effect of Inequity-aversion, Trust, and Social Norms Dr Richard Fairchild (School of Management, University of Bath) University of Bath School of Management, Working Paper Series This working paper is produced for discussion purposes only. The papers are expected to be published in due course, in revised form and should not be quoted without the author s permission.

2 University of Bath School of Management Working Paper Series School of Management Claverton Down Bath BA 7AY United Kingdom Tel: Fax: Androniki Apostolakou & Gregory Jackson 009 Corporate Social Responsibility in Western Europe: An Institutional Mirror or Substitute? Meryem Duygun- Fethi & Fotios Pasiouras Klaus E. Meyer Christos Ioannidis, Fotios Pasiouras & Constantin Zopounidis Assessing Bank Performance with Operational Research and Artificial Intelligence Techniques: A Survey Corporate Strategies under Pressures of Globalization: Globalfocusing Assessing bank soundness with classification techniques Klaus E. Meyer Corporate Strategies under Pressures of Globalization: Initiating a Forward-looking Debate Philip Cooper, Eleanor Dart Change in the Management Accountant s Role: Drivers and Diversity Gregory Jackson Actors and Institutions Richard Fairchild Incentive problems and Reciprocal Behaviour in Venture Capital/entrepreneur Dyads: The effect of Inequity-aversion, Trust and Social Norms

3 Incentive problems and Reciprocal Behaviour in Venture Capital/entrepreneur Dyads: The effect of Inequity-aversion, Trust, and Social Norms Author: Richard Fairchild, School of Management, University of Bath, UK. July 009 Abstract The performance of venture capitalist/entrepreneur dyads may be adversely affected by extreme incentive problems, which may affect the behaviour of both the entrepreneur and the venture capitalist (double-sided moral hazard). Recent research suggests that these problems may be mitigated by behavioural factors such as fairness and trust. In this study, we develop a game-theoretic model of venture capital contracting in which the behavioural factors of fairness and trust combine to reduce the double-sided moral hazard problem, hence enhancing equilibrium venture performance and value-creation. We conclude by considering the policy implications emanating from our model. 3

4 1. Introduction Entrepreneurs often obtain start-up finance from venture capitalists. However, the value-creating ability of venture capital-backed firms may be adversely affected by the complex relationships that exist between venture capitalists and entrepreneurs. For example, the venture capitalist/entrepreneur relationship may be fraught with extreme incentive (agency) problems (Amit et al 1990, Admati and Pfleiderer 1994, Bascha 000, Casamatta 003, Repullo and Suarez 004, Fairchild 004). Hence, venture capitalists and entrepreneurs have developed contracts that attempt to overcome these problems and align the parties interests (Klausner and Litvak 001, Triantis 001, Tykvova 007). More recently, it has been recognised that venture-backed performance may also be affected by behavioural and emotional factors. The performance of venture capitalist/entrepreneur dyads may be affected (positively or negatively) by reciprocal feelings of fairness, trust, empathy and spite (Busenitz et al 1997, Cable and Shane 1997, Sapienza and Koorsgard 1996, De Clerq and Sapienza 001, Shepherd and Zacharakis 001, Utset 00). Indeed, De Clerq and Sapienza (001) coin the term relational rents, referring to the value-creating potential of fairness and trust in venture-backed firms. However, these authors note that no in-depth analysis has been made of how relational rents might be created for both parties in the venture capitalist-entrepreneur dyad. This observation motivates our analysis. In particular, we develop a behavioural game-theoretic model of venture capital contracting in order to analyse the combined impact of fairness and trust on agency problems and venture performance. Our model develops the literature in two major ways. First, it presents a formal game-theoretic 4

5 analysis of the effects of fairness and trust on venture performance, hence addressing the lacuna identified by De Clerq and Sapienza (001). Second, we identify that fairness and trust may be related, and therefore, we provide the first formal analysis of the combined effects of fairness and trust. Our model consists of three stages; a financial contracting stage (in which a venture capitalist and an entrepreneur negotiate over their respective equity shares), followed by a performance stage (in which the entrepreneur exerts value-adding effort in developing the business). The entrepreneur s effort affects the probability of success of the venture. In the final stage of the game, the venture capitalist can force renegotiation of the financial contract (in order to reallocate the equity stakes in the venture capitalist s favour). When developing our analysis, we demonstrate that our model has the following ingredients; a) Double-sided moral hazard, in the form of the E s shirking incentives at the effort stage, and the VC s hold-up threat at the ex post stage. b) Fairness norms, which affect the VC s equity offer, and the E s effort incentives. c) VC trustworthiness, relating to the risk faced by the E that the VC will force ex post re-negotiation of the contract in her favour. We further demonstrate that factors b) and c) combine to mitigate the double-sided agency problem, and improve firm performance. Before proceeding to our model, we now provide a brief review of the key literature relating to each of our ingredients. 5

6 1.1. Double-sided Moral Hazard in Venture Capital. The early financial contracting models (e.g., Baker and Gompers 1999) assumed a pure principal-agent relationship in which the venture capitalist, as principal, suffers from moral hazard problems from the entrepreneur, as agent. However, Smith (1998) argues that, since both parties contribute wealth-creating efforts, the performance of the venture may be subject to double-sided moral hazard problems. Some researchers have developed double-sided moral hazard models (e.g., Houben 00, Casamatta 003, Fairchild 004, Repullo and Suarez 004) whereby the entrepreneur and the venture capitalist both supply value-adding effort. The doublesided moral hazard problem exists due to the parties incentives to shirk. In this paper, we consider a different form of double-sided moral hazard. In our model, the E contributes value-creating effort. Hence, the VC faces moral hazard in the form of entrepreneurial shirking. After the E has exerted effort, the VC can force ex post renegotiation of the equity stakes. Hence, the E faces moral hazard in the form of the VC s ex post hold-up threat. Indeed, Smith (1998) argues that the most prominent risk to entrepreneurs is opportunism. The potential for opportunism arises from the possibility that the VC will attempt to renegotiate with the entrepreneur at a point in the relationship when the entrepreneur has diminished bargaining power. The VCs ex post hold-up threat is receiving increasing attention in the literature (see eg; Smith 1998, Repullo and Suarez 003, Chemla et al 004, Skeie 004, Gebhardt and Schmidt 006, Bigus 00, Utset 00). These researchers note that certain 6

7 features of the VC contract provide the VC with increasing bargaining power over time 1. This provides the VC with leverage to force ex post renegotiation of the financial contract in her favour. The VC s ex post hold-up threat weakens the E s ex ante effort incentives. Some researchers (eg; Fairchild 004) have argued the VC may therefore voluntarily use contractual methods to commit not to act opportunistically ex post, in order to strengthen the E s ex ante incentives. In contrast to contractual solutions, our model analyses the role of VC trustworthiness in mitigating the VC hold-up problem. 1.. Emerging Research on Fairness and Trust in Venture Capital The existing financial contracting models assume that entrepreneurs and venture capitalists maximize utility based on narrow self-interest. However, behavioral economists recognize that relationships may be affected by psychological factors, such as feelings of fairness and reciprocity (Bolton 1991, Rabin 1993, Fehr et al 1999, 001, 00, 004, Bolton and Ockenfels 000, Anderhub et al 001), empathy (Sally 001), and trust (Berg et al 1995, Bolle 1995, Huang 000, Bacharach et al 001). Furthermore, these feelings may affect the outcomes of negotiations and performance. VC researchers are beginning to develop conceptual models of the impact of fairness and trust on VC contracting and performance. These researchers compare agency theory (Barney et al 1994) with procedural justice theory (Sapienza and Korsgaard 1996, Sapienza et al 000, De Clerq and Sapienza 001, Shepherd and Zacharakis 1 Smith (1998), Bigus (00), Repullo and Suarez (003), and Utset (00) all note that the VC s staging of financing enables the VC to increase her bargaining power over time, resulting in the ex post hold-up threat. In Skeie s (004) model, the VC s increased bargaining power, and the ex post threat, arises form the common feature of vesting of equity payments to the entrepreneur. 7

8 001). The agency theory approach emphasises mechanisms to control venture capitalist and entrepreneurial self-interested opportunistic behaviour. In contrast to agency theory, procedural justice theory focuses on the perceived sense of fairness in making decisions. Lehtonen et al (004) note that procedural justice theory may be particularly relevant to the VC/E relationship. The VC/E relationship may be subject to feelings of unfairness and mistrust. VCs often complain that entrepreneurs are reluctant to share information (Sapienza 1989). A person s willingness to share information and provide timely feedback may signal his openness and honesty (Sapienza and Korsgaard 1996). According to PJ, the more one party perceives a procedure to be fair, the greater they will trust the other party. Increased trust and communication between entrepreneurs and venture capitalists can create relational rents (De Clercq and Sapienza 001). Open and frequent communication between the entrepreneur and the venture capitalist may result in an increase in perceptions of fairness and trust, thus mitigating agency problems (Shepherd and Zacharakis 001). Procedural justice theory can also be employed to consider the VC s ex post hold-up problem. An increase in a person s perception of fairness may lead to an increase in commitment to decisions, performance, behaviour and attitude (Kim and Mauborgne 1991, 1993). This increased commitment, and feelings of fairness, may mitigate the venture capitalist s opportunistic post-investment threat of early exit (Shepherd and Zacharakis 001, Utset 00). 8

9 1.3. General Game-theoretic Research on Fairness and Trust In recent years, behavioural economists have begun the task of developing formal game-theoretic models of fairness (e.g, Becker 1974, Bolton 1991, Rabin 1993, Fehr and Schmidt 1999) and trust (e.g; Berg et al 1995). In general, these researchers have modelled fairness by assuming that agents not only consider their own payoffs in absolute terms, but also in relative terms (that is, compared to other agents payoffs). Formally, fairness is modelled by including both absolute and relative payoffs in the players utility functions. In this paper, we focus on Fehr and Schmidt s (1999) model of inequity-aversion. This approach combines envy and altruism; each player dislikes unequal payoffs, either in his own, or the other player s, favour. Fehr and Schmidt model this as follows. A player suffers a utility loss either if her opponent has a higher or a lower payoff. As a result, players take actions to equalise payoffs. Fehr and Schmidt develop a series of models of reciprocal fairness in a principalagent context, where reciprocal fairness may induce high wage offers from management to workers, and high effort levels by workers in response. Further, FS test their models experimentally. They provide evidence of reciprocal fairness. Models of fairness can be developed to include social norms. Huang (000) considers the effect of a fairness norm on the equilibrium of a two-player ultimatum bargaining Other fairness models include Becker (1974) who considers pure altruism, Bolton (1991) who analyses enviable fairness, and Rabin (1993), in which a player s reaction to another player s unfair behaviour depends upon whether that behaviour is intentional. He considers players reciprocation of perceived nice or nasty behaviour, and he derives a fairness equilibrium. 9

10 game. Interestingly, he relates fairness to anger as follows; In psychological games, the impact of anger on payoffs depends on endogenously determined equilibrium beliefs about behaviour. The less likely that a party believes that it is going to be offered the smaller part of an unequal division, the more anger that party will suffer if such an offer is made. Conversely, the more likely a party believes that it will be offered the lesser share of an unequal division, the less anger it will suffer if such an offer is made. Our modelling approach to fairness draws upon Fehr and Schmidt s inequity-aversion approach, together with Huang s (000) social norm methodology. Behavioural economists have identified several methods for modelling trust. The general trust game is based on reciprocity of behaviour (e.g, Berg et al 1995 and Maccabe et al) 3. In contrast, Al-Najjur and Casadessus-Masanell (001) develop a principal-agent model in which trust is a characteristic of a player, rather than a reciprocal response. Our approach to trust is closest to Al-Najjur and Casadessus-Masanell (001). That is, in our model, we simplify the approach to by not considering reciprocal trust. Instead, we focus on VC trustworthiness as a characteristic. That is the VC s behaviour depends on her type (trustworthy or untrustworthy), and is not a reaction to E s behaviour (that is, the VC does not exhibit reciprocal behaviour). 3 The reciprocal trust game, as developed by Berg et al (1995) and Maccabe et al, is designed as follows. The first player (the trustor) chooses whether to send money to the second player (the trustee). The second player then chooses whether to invest this money in a value-adding project. Finally, the second player then decides how much of the value created to return to the first player. Experimental evidence of the first player s willingness to send money, and the second player s willingness to return money, provides support of reciprocal trust. 10

11 Our model compares with Al-Najjur and Casadessus-Masanell (001) as follows. They develop a model of trust in agency contracts, whereby an agent chooses an effort level and then the principal chooses a wage from a bounded interval. The authors consider two types of principal; trustworthy and self-interested. In contrast to our model, the self-interested principal chooses the lowest possible wage in the interval, while the trustworthy principal chooses the lowest possible wage such that the agent s participation constraint is just satisfied. In contrast, in our model, the VC makes an initial equity proposal to the E, and then the E exerts effort. The trustworthy VC does not force ex post renegotiation of the equity allocation. The trustworthy VC does. Furthermore, in contrast to Al-Najjur and Casedessus-Masanell (001), we consider the combined impact of fairness and trust. The rest of the paper is organised as follows. In section 3, we present the model, and solve for the effect of the fairness norm and VC ability on the VC s equilibrium equity proposal and the players equilibrium effort levels. In section 4, we consider the effect of the fairness norm and VC ability on equilibrium venture performance. In section 5, we discuss policy implications emanating from our model. Section 6 concludes with a discussion of future research.. The Model We consider a venture capitalist/entrepreneur financial contracting game in which an entrepreneur (E) has an idea for an innovative project, and approaches a venture capitalist (VC) to obtain funding. The E and the VC firstly negotiate over the financial contract (specifically, they negotiate over their respective cashflow rights in the form of an equity allocation). After agreeing the contract, the E exerts effort in forming the 11

12 business and creating value. Specifically, the E s effort affects the probability of success for the business (hence, at this stage, we consider moral hazard in the form of E s shirking). After the project outcome is realised, the VC has an opportunity to force re-negotiation of the contract in order to re-allocate the equity stakes in her favour (the VC s hold-up threat). We model the negotiation stage as an ultimatum bargaining game in which the VC makes a take-it-or-leave-it offer of equity to the E (this embodies the idea that the VC has the bargaining power). The equity offer affects the entrepreneur s effort incentives. Furthermore, the VC s ex post hold-up threat reduces the E s ex ante effort incentives. We demonstrate that fairness and trust mitigate these moral hazard problems. Fairness induces the VC to offer a higher equity stake, which mitigates the E s shirking problem. VC trustworthiness reduces the ex post threat, which further strengthens the E s effort incentives. Our model incorporates the effects of fairness and trust as follows. The E and the VC are drawn randomly from a population consisting of r [0,1] inequity-averse (fair) VCs and 1 r self-interested VCs; r reciprocating Es (that is, these Es are concerned with fairness, and react to unfair treatment with anger or spite) and 1 r selfinterested Es. We denote these types as E and VC, with i { S, F}, where S refers i i to a self-interested type, and F refers to a fair type. In our model, fairness and trust are related as follows. We assume that all of the inequity-averse VCs are trustworthy. Furthermore, a proportion t of the selfinterested VCs are trustworthy; and a proportion 1 t are untrustworthy. We consider three possible levels of trust; none of the self-interested VCs are trustworthy, half of the self-interested VCs are trustworthy, or all of the self-interested VCs are 1

13 1 trustworthy; t {0,,1}. This enables us to consider the combined effects of increasing societal fairness and trust on venture performance in a clear and tractable way. Since the VC is drawn at random from a population of r fair and 1 r self-interested VCs, the probability of the VC being trustworthy is T = r + t( 1 r). We refer to T as the expected level of trustworthiness in society. The relationship between societal fairness and trust is as follows. If t = 0, T = r. Since none of the self-interested VCs are trustworthy, there is a direct, one-to-one, relationship between increasing societal r 1 fairness and increasing trust. If t =, T =, with T [,1] when r [0,1]. When t =1, T = 1 for any level of fairness r. Since all VCs (fair or self-interested) are trustworthy, the level of fairness is has no effect on societal trustworthiness. The timeline of the game is as follows; Date 0: The Matching Stage: A VC and an E are drawn and matched randomly from the population. Each player s fair or self-interested type is revealed to the other player. If the VC is revealed to be fair, then the E knows that the VC is trustworthy. However, we assume that if the VC is revealed to be self-interested, her trustworthiness is not revealed until date 4. Date 1: Financial Contracting Stage: The VC makes a take-it-or-leave-it proposal relating to the E s and the VC s relative equity stakes 4. Specifically, the VC offers the 4 Hence, we have chosen to model equity bargaining as an ultimatum game, with VC as proposer, and E as respondent. This captures the idea that the VC has the bargaining power. 13

14 ~ E a proportion α [0,1] of the date 3 realised project value V, with the VC retaining the balancing proportion 1 α. Date : Post-contractual Value-creating Stage: The E exerts effort e. Effort is costly, and we assume that the cost of effort is C = βe ; this represents increasing marginal cost of effort. The project has two possible date 3 outcomes, success, in which case the project provides date 3 income R, or failure, in which case the project provides income zero. The E s effort level affects the probability of success P. We model this as follows; the probability of success is P = γe, while the probability of failure is 1 P. Therefore, at date, the expected value of the project is V = γer. Date 3: Project outcome stage: At this stage, the outcome of the project (success or failure) is revealed. Date 4: The Ex Post Renegotiation Stage: After the project s outcome has been achieved, the VC has the opportunity to force renegotiation of the equity stakes in the financial contract, making a new take-it-or-leave-it offer. We assume that only the untrustworthy type carries out this threat, whereas the trustworthy type does not..1 Analysis of the Game We note that our game consists of three sequential decisions; a) the VC s date 1 equity proposal, b) the E s date effort decision, and c) the VC s date 4 renegotiation decision. We solve this game using the standard game-theoretic approach. That is, we employ backward induction, whereby we solve for the equilibrium decisions in reverse order, as follows. 14

15 First, we solve for the VC s ex post date 4 re-negotiation decision, taking as given the VC s date 1 equity proposal and the E s date effort decision. Then, we move back to date to solve for the E s effort decision, still taking as given the VC s date 1 equity proposal. Finally, we move back to date 1 to solve for the VC s equity decision, hence deriving the equilibrium of the entire game..: The VC s date 4 re-negotiation threat. At date 4, the E has already exerted effort, and the outcome of the project has been realised. If the project failed, realising income of zero, the VC s re-negotiation threat is irrelevant. If the project succeeded at date 3, then, by assumption, the trustworthy VC does not force re-negotiation, and the E and the VC receive their respective equity allocations, α R and ( 1 α ) R, as agreed at date 1. If the VC is untrustworthy, then, since the E has already exerted effort, and the outcome R has already been achieved, the VC will take all of the equity, providing the E and the VC with respective payoffs of zero and R...3: Date Effort Stage/date 1 Equity Stage. For presentation purposes, we present our analysis of these two decisions together in this sub-section (still using backward induction). Since the E and the VC are drawn randomly from the population at date 0, we note that there are 4 possible combinations of types in the VC/E dyad. We proceed to solve the game by considering the different combinations. Furthermore, we now formalise a payoff structure, which is necessary in order to solve the game. 15

16 .4. Benchmark Case: E S + VC S We refer to the dyad in which both players are self-interested as the benchmark case, since this reflects the standard view in economics that agents are strictly selfinterested (homo oeconomicus). We consider the following expected payoff 5 function for the self-interested E; = tαγer βe. (1) E The first term represents the E s equity share 6 α of the expected project income 1 V = PR = γer, multiplied by the probability t {0,,1} th at VC S is trustworthy. The E can observe that the VC is self-interested, but not whether the VC is trustworthy or untrustworthy. With probability t, the VC S is trustworthy, and, therefore, the E will retain his agreed equity share throughout the game. With probability 1 t, E has matched with an untrustworthy VC S who forces re-negotiation at date 4, taking all of the equity. The second term of the payoff is the E s cost of effort. In order to solve for the VC s optimal date 1 equity proposal, we define the VC s payoff, as follows; = α V = αγer. () VC Using equations (1) and (), we obtain our first result. 5 This represents the E s expected payoff as at date (when he makes his effort decision). At this stage, the E knows that the VC is self-interested, and so knows that she is trustworthy with probability t. 6 This equity share was already agreed at date 1. 16

17 Proposition 1: In the benchmark case, E + VC (both players are self-interested), S S the VC s equilibrium equity proposal i s α * = 0.5, and the equilibrium expected tγ R project value is V =. Therefore, V ( t = 0) = 0, 4β 1 γ R γ R V ( t = ) =, V ( t = 1) =. 8β 4β Proof: See Appendix. Proposition 1 reveals that, in the standard case where both players are self-interested, the VC offers half of the equity to the E, retaining the other half for herself. Intuitively, the VC faces a trade-off between giving equity to the E to motivate entrepreneurial value-creating efforts, and retaining equity for herself. If the VC kept all of the equity, the E would exert zero effort, and no value would be created (the VC would hold 100% of a project with zero value). If the VC gave all of the equity to the E, the E s effort levels would be maximised, but the VC would hold a zero percentage equity stake in a high-valued project. The VC maximises her payoff by providing 50% of the equity to the E, and retaining 50% herself. Note that this is the VC s optimal equity offer regardless of the level of trust. Increasing levels of trust simply increases the peak in the VC s payoff at α * = Either type of E ( or E ) matches with a fair VC VC ). E F F ( F W e define the fair equity proposal, α = α F, as the proposal that equalises expected payoffs. Furthermore, we define the fair (reciprocating) E s payoff as follows; 17

18 = [ t( α r( α α )) + (1 t)( rα )] γer βe. (3) E F F This payoff has been constructed as follows. First, consider the terms in the square brackets. The first term represents the E s equity stake when the VC S is trustworthy, minus a term representing the reciprocating E s disutility from receiving an equity allocation α that is different from the fair proposal α F. The sec ond term is the E s equity stake of zero minus the E s disutility from receiving an equity allocation less than the fair proposal. The term r represents the social fairness norm (as in Huang). The last term is the E s cost of effort. Using (1), () and (3), we obtain our second main result. Proposition : If the fair VC matches with either type of E ( VC +, or F E S VC F + E F ), the VC s equilibrium equity proposal is α F =, and the equilibrium 3 γ R expected project value is V =. By assumption, VC F is trustworthy. 3β Proof: See appendix. Comparison of propositions 1 and reveals that the fair VC offers more equity to the E than the self-interested VC. Recall that the fair VC acts to equalise the players payoffs. Intuitively, the fair VC s equity offer of α F = compensates the E for his 3 18

19 effort costs. Further comparison of the propositions reveals that expected firm value is higher when the VC is fair (combined with either type of E) than when both players are self-interested. VC fairness is indeed value-enhancing, since the higher equity offer increases the E s effort incentives..5. ( E + ). F VC S Finally, we consider the case where the VC is self-interested, but the E is a fair reciprocator; ( F S E + VC ). Our result is as follows; Proposition 3: In the E F + VC S dyad, the VC s equilibrium equity proposal is 1 r α * = min[ +,1], and the equilibrium expected project value is 3(1 + r) t V γ R 3 = max{ [ (1 + r) t r],0}. 6β a) When t = 0, α * is irrelevant, since V = 0 for any equity offer. Since all of the selfinterested VCs are untrustworthy, the E exerts zero effort for any equity offer, and no value is created b) When t =, α * [, ] when r [0,1]. Therefore, [, ] 6 V γ R γ R 8β 1β when r [0,1]. The expected value of the project decreases as societal fairness increases. 1 γ R γ R c) When t =1, α * [, ] when r [0,1]. Therefore, V [, ]. In contrast 3 4β 3β to b), the expected value of the project increases as societal fairness increases. Proof: see appendix. 19

20 Proposition 3 reveals that the self-interested VC s equity offer is affected by her anticipation of the reciprocating E s reaction (in terms of E s effort level), and is further affected by the extent of the societal fairness norm. If the fairness norm is zero ( r = 0), the self-interested VC offers half of the equity to the E (as in the E + dyad; see proposition 1) for any level of trust t. As the S VC S fairness norm increases towards 1, then (if t =1; all members of the population are trustworthy), the optimal equity offer incre ases towards the fair VC s equity offer α * =. For lower levels of t, the optimal equity offer increases more sharply, to 3 compensate the reciprocating E for higher re-negotiation risk). Similarly, in the E + VC dyad, the effect of fairness F S r on firm value is moderated by the level of trust t. If t =1, then firm value ranges from γ R V = when 4 β r = 0 (as in proposition 1), to γ R V = when r = 0 (as in proposition ). That is, if all 3β members of the population are trustworthy, then the self-interested VC s optimal equity offer to the fair E is merely affected by the societal fairness norm. When the fairness norm is zero, the self-interested VC acts as she would when matching with a self-interested E. As the fairness norm approaches unity, the self-interested VC s equity proposal is forced towards the fair VC s proposal. Comparison of proposition b) and c) provide interesting results regarding the relationship between fairness, trust and venture performance when the E is fair. If VC trustworthiness is too low, then no equity offer between zero and unity can induce the E to exert effort, due to the high re-negotiation risk. Hence, firm value is zero. 0

21 If the level of VC trustworthiness is in a medium range, firm value is actually reducing in the fairness norm. The intuition is that, as the fairness norm increases, the reciprocating E feels more indignation at the high renegotiation risk, and reduces his effort level. If the level of VC trustworthiness is sufficiently high, firm value is increasing in the fairness norm. The results of propositions 1, and 3 are compared in diagram 1. Fairness, Trust and Venture Performance 0.35 Venture Performance Fairness Diagram 1. The diagram shows the following. When E and VC match (proposition 1), or S S E S and VC F match, venture performance is independent of fairness (horizontal lines). Wh en E and VC match, venture performance is decreasing in fairness for low VC F S trustworthiness (proposition 3b), while it is increasing in fairness for high VC trustworthiness (proposition 3c). 1

22 .6. Expected Venture Performance. In order to consider the effects of societal fairness and trust on venture performance, we define an expected venture value prior to an E or a VC being drawn from e population. Hence, E(V ) th ],0} ) (1 3 [ 6 max{. ) (1 3 4 ) (1 ) ( r t r R r r R r R t r V E = β γ β γ β γ (9) Equation (9) may be thought of as an index of expected VC performance. From propositions 1, and 3, we note the following; Proposition 4: a) When, = 0 t. 3 ) ( β γ R r V E = Therefore, ] 3 [0, ) ( β γ R V E when ]. r [0,1 ; t 1 = b)when ],0}. ) (1 4 3 [ 6 max{. ) (1 3 8 ) (1 ) ( r r R r r R r R r V E = β γ β γ β γ Therefore, ] 3, 8 [ (V ) E ]. [0,1 β γ β γ R R when c) When r =1; t ],0} ) (1 3 [ 6 max{. ) (1 3 4 ) (1 ) ( r r R r r R r R r V E = β γ β γ β γ ] 3, 4 [ ) ( β γ β γ R R V E Therefore, when ]. [0,1 r

23 Hence, for a given level of trust, the performance index is positively related to societal fairness. For a given level of fairness, the performance index is positively related to trust. The performance index is maximised when fairness (and trust) is maximised. We present the effects of fairness r and trust t on expected venture value E(V ) in the following graph. Fairness, Trust and Performance Index x Performance Inde Fairness Diagram. The lowest line represents zero trustworthiness amongst self-interested VCs, t = 0. The middle line represents 1 t =. The top line represents maximum trustworthiness amongst self-interested VCs, t =1. Recall that societal trust is T = r + ( 1 r) t. Referring to equation (9), if t = 0, then, when r = 0, then T = 0, and E ( V ) = 0; that is, if all self-interested VCs are 3

24 untrustworthy, then a zero fairness norm (there are no fair VCs, and all VCs are selfesults in zero societal trust, and zero expected value. When r =1, interested) r γ R E ( V ) =. 3β If t =1, then T = 1 for all values of r ; all VCs (fair and self-interested) are γ R trustworthy. Now, referring to equation (9), when r = 0, E ( V ) = ; and when 4β r =1, E ( V ) = γ R 3β. Therefore; Proposition 4: a) For a given level of self-interested VC trustworthiness, an increase in the social fair ness norm increases expected venture value. b) For a given level of fairness norm, an increase in VC trustworthiness increases venture performance. c) Venture performance is increased by combined increase in fairness and trustworthiness. D) Venture performance is maximised at maximum fairness norm which maximises trustworthiness. We note that expected venture value is increasing a) in the level of the fairness norm (as a result of two effects; the self-interested VC is driven to offer a higher equity stake to the E, and the proportion of fair VCs in the economy increases and dominates (who both offer more equity, and are trustworthy for sure)), and b) in the level of VC trustworthiness. An interesting point to note is that, for any level of fairness below one, the expected venture value is increasing in the level of trust. This is the opposite 4

25 case to the VC s optimal equity offer, where the optimal equity offer was decreasing in the level of trust. Recall that th e VC needed to offer the E higher equity as There is much evidence (Camerer, 003) that fairness norms vary considerably across cultures. 7 Using experimental ultimatum bargaining games, Roth et al (1991) found much similarity in sharing norms amongst subjects in America, Israel, Japan and Yugoslavia. In contrast, Buchan et al s (1997) ultimatum bargaining experiment demonstrated a strong sharing norm amongst subjects in the collectivistic culture of Japan, compared with more self-interested behaviour in America. Finally, Henrich et al (001, 00) compared experimental ultimatum bargaining results for Machiguenga farmers in Peru with subjects in the US, and found much more self- interest in Peru. trustworthiness reduced in order to compensate the E for the increased re-negotiation risk. Diagram suggests that this increased offer is not sufficient to prevent expected value falling as trust diminishes. 3. Practical and Policy Implications. We now employ our game-theoretic model to consider practical and policy implications, with reference to existing research. We considered the combined effects of fairness and trust on venture performance. We obtained the following results; 3.1. Societal Fairness and Trust. First, we discuss the exogenous inputs into our model, societal fairness and trust. 7 For a detailed discussion of this point, see chapter of Camerer s (003) excellent text-book on behavioural game theory. We provide a brief review here. 5

26 Camerer (003) also discusses the trust experiment of Buchan et al (000). These authors found that Chinese investors were the most trusting and trustworthy, while the Japanese were the least. Therefore, the evidence suggests that fairness and trust is greatly affected by culture. Lehtonen et al (004) relate culture and venture capital performance, stating that we propose that a culture (geographical context), which does not call for high fairness and justice (of actions) will increased the risk of opportunistic behaviour on the part of the VC.. This could provide Es with a method to evaluate the risk of opportunistic behaviour based on geographical location. As a policy implication, our model demonstrates that measures to increase trust and fairness in society, and particularly between Es and VCs, would be beneficial for venture performance (and economic growth). As a practical example, we suggest that business school education (for example, at MBA level) of VCs and Es should emphasise the beneficial role of fairness and trust in negotiations and contracting 8. On an empirical level, researchers should test the relationship between different societies fairness-norms and venture capital contracting and performance. 3. Fair and unfair cashflow rights In our model, increasing societal fairness resulted in fairer cashflow terms (in the form of the equity allocation between the VC and the E). In this section, we consider specific examples of fair and unfair cash-flow rights. In our model, an increase in societal fairness resulted in the VC offering a higher (and fairer) equity stake to the E. In reality, the allocation of cash-flow rights subtly 8 Indeed, ethics courses at business schools often employ repeated prisoner s dilemma experiments to educate the students about fairness, trust and cooperation (Gibson 003). These could be extended to entrepreneurship classes. 6

27 depends upon the type of financial instruments employed (for example, equity versus convertibles), and the conditions that the VC may negotiate into the contract, such as punitive covenants (including anti-dilution and liquidation provisions) that subtly affect cashflow rights over time. Firstly, we consider the research into fair and unfair financial instruments (eg; Hoffman and Blakey 1987, Tucker 004, Cumming 005). Hoffman and Blakey (1987) argue that VCs usually invest in redeemable preferred stock or debentures. VCs like this type of instrument because it provides them with an equity stake in the case of success, and it provides them with debt (and first claim to the cash flows) if the company fails. The authors provide this advice to entrepreneurs, if you have a choice, try to get the VCs to accept a capital structure that consists entirely of common stock, because it is simpler and keeps the balance sheet clean. Chances are, however, that you won t get them to agree to this unless your bargaining position is extremely strong. Tucker (004) also argues that convertibles, widely used by the VC industry, represent an unfair financial instrument, and argues that Es should attempt to negotiate an equity-based financial structure. Cumming (005) demonstrates that US VCs make extensive use of convertibles, whereas Canadian VCs mainly hold equity contracts. The traditional agency argument is that convertibles can be used by VCs to control moral hazard problems relating to E s opportunistic and shirking behaviour. However, Cumming (005) suggests, as an interesting counter-argument, that the use of equity may signal a trusting and fair VC/E relationship. He states, The high proportion of straight common equity contracts in Canada suggests that relatively more trusts exist within Canadian VC markets when compared to the US. That is, common equity places the entrepreneur and the VC on the same team. 7

28 Next, we consider the unfair terms that may be attached to the parties cash-flow rights, such as anti-dilution and liquidation provisions (or ratchets ). VCs employ these provisions to protect their investment. Hoffman and Blakey (1987) advise entrepreneurs, VCs concern over preserving their capital is legitimate. At the same time, you should review these provisions carefully to make sure that they are fair to you and any other founders. Research has suggested (Hoffman and Blakey 1987, Tucker ) that such unfair punitive covenants should be excluded from any VC/E contract, since they may have the effect of reducing the E s equity stake over time, potentially damaging fairness and trust in the VC/E relationship. In summary, in terms of cash flow rights, it may be argued that performance- enhancing fairness and trust can be promoted by the VC s avoidance of convertibles, instead focussing on mutual allocation of equity to the VC and E, and by eliminating onerous punitive covenants, such as anti-dilution rights and ratchets. 4. Future Research Our model has focussed on the effects of social fairness and trust on the fairness of cash-flow rights and venture performance. However, Kaplan and Stromberg (000) identify that VCs and Es negotiate over both cash flow rights and control rights in the contract. Further, Cable and Shane (1997) argue that non-contractual factors, such as personal relationships and regular communication between VCs and Es could affect cooperation and performance. Therefore, we argue that increasing societal fairness 9 Tucker is an entrepreneur, who obtained VC funding for his venture. In 004, he conducted a small survey of Es regarding fair and unfair cash-flow and control rights. 8

29 may lead to fairer cash-flow and control rights, and fairer non-contractual behaviour, hence enhancing performance. We now discuss directions for future research, relating to fair control rights, and fair non-contractual factors in the VC/E relationship. 4.1: Fair and Unfair Control Rights. VCs and Es negotiate over their control rights in the venture. Researchers argue that VCs may need to protect their investment through such control provisions as performance/forfeiture provisions (Hoffman and Blakey 1987). Another control mechanism employed by the VC relates to staged financing, whereby VCs provide future rounds of financing if the E achieves certain milestones. Finally, VCs may negotiate a firing clause in the contract, whereby they can remove the E if the venture fails to meet certain targets (Hellmann 1998). Although these control rights may be seen as protecting the VC s investment, they may be destroy fairness and trust in the VC/E relationship. Indeed, a debate exists in the literature whether control rights and fairness/trust act as complements or substitutes in enhancing venture performance. Cable and Shane (1997) imply that they are complements (increasing control leads to increase in trust and venture performance). On the other hand, Shepherd and Zacharakis (001) argue that control and trust/fairness may be substitutes. An increase in control may actually destroy feelings of fairness and trust, adversely affecting venture performance. Tucker (004) argues that the E often signs over significant negative control rights, leaving him with a feeling of lack of control in the relationship. For example, the VC 9

30 may negotiate to include restrictive covenants in the contract. These ensure that approval must be sought from the VC when taking most operational decisions. This may damage trust, adversely affecting venture performance. Tucker also argues that a fair VC will produce detailed term-sheet running to several pages, so that much of the negotiation can be conducted before each party legally commits to each other. On the other hand, an unfair VC produces a brief term-sheet, leaving much room for future interpretation. The author further argues that VCs often employ over-zealous lawyers who build in investor protections that are not needed, or have been specifically excluded in discussions between the VC and the E. The result is a package of documentation which at best offers the investor all the protection they need while ensuring the E has the authority to run the business; and at worst, promotes a bureaucratic approach to operational management, reminiscent of many much larger corporations. 4.: Fair and Unfair Non-contractual Factors. We should consider non-contractual factors, not specifically related to cash-flow rights or control rights. According to procedural justice theory, procedures that increase the players perception of fairness can lead to increased cooperation. Cable and Shane (1997), and Lehtonen et al (004) place emphasis on relationship-building and open and frequent communication as methods of increasing perceptions of fairness and trust in the VC/E relationship. Communication provides a potent method of signalling fairness. VCs typically invest in new, highly risky ventures, characterised by large asymmetric information 30

31 problems (Houben 003). Es normally possess superior knowledge regarding the technical aspects of the venture, while the VCs have superior information regarding the potential market conditions. According to standard agency theory, VCs and Es possess incentives to mis-state or with-hold their private information 10. Therefore, timely and open communication of accurate information may increase the VC s and E s mutual perception of fairness Dynamic Fairness and trust VC/E Model. In our model, fairness and trust are exogenously-given inputs which induces fairness in cashflow terms, and possibly (but not modelled) fair control rights and fair noncontractual behaviour (such as increased communication and relationship-building), which, in turn, enhances cooperative efforts and performance. We do not consider the dynamic effects on perceptions of fairness and trust over time in the VC/E relationship. Utset (00) provides an inspirational basis for a future research agenda on this dynamic relationship. He provides a deep and extensive conceptual analysis of the effects of behavioural factors (particularly reciprocal fairness and anger) on the dynamic relationship between the VC and the E over time. Utset begins with the view that the VC, as principal, needs to protect her investment from the self-interested behaviour of the E. Hence, he argues that the VC structures the contract to retain control over the E s equity stake and the exit decision. However, Utset notes the dynamic and complex nature of the relationship. 10 See Cable and Shane (1997) and Lehtonen et al (004) for an extensive discussion of the parties incentives to mis-state or with-hold information. 11 We argue that this may be analogous to our cash-flow model. In our model, increased fairnessnorms resulted in a self-interested VC offering a more generous equity stake to the E, and the E responding with more cooperative value-adding effort. Similarly, increasing fairness-norms may induce self-interested VCs or Es to reveal more (truthful) information, with the same positive effects on cooperative efforts and performance. Future research will develop our model in this direction. 31

32 While the control mechanisms adopted by the VC play an important role in protecting their investment, they also give VCs great leeway to act opportunistically, and because Es make ex ante transaction-specific investments, the VC can threaten to fire the E or use some other form of leverage to appropriate a greater share of any potential surplus. This si the classic opportunism scenario. Utset incorporates behavioural considerations into the analysis. Firstly, he discusses how Es may enter into a VC contract with over-optimistic expectations of VC trustworthiness and fairness. As the relationship develops over time, the E may begin to observe increasing VC opportunism and unfair behaviour. As his initial, overconfident, expectations become deflated, the E may then retaliate with increasing anger and spite. As Utset states, The level of divergence of the E s original expectations matters, since it casts a shadow affecting VC/E relationships during the life of the venture... Under certain behavioural assumptions, the E will have an incentive to retaliate against the VC, even if doing so comes at a cost to the E. We believe that an important future step in the venture capital research is to develop a dynamic behavioural model that captures the E s retaliatory behaviour. 5. Conclusion 3

33 We have developed a behavioural game-theoretic model of venture capital/entrepreneur contracting and performance that demonstrates that stronger fairness-norms result in an increase in the equity stake offered by the VC to the E, which, in turn, induces higher entrepreneurial effort, and improved venture performance. We have also discussed the effects of other unfair contractual terms, such as unfair cash-flow provisions (anti-dilution provisions, ratchets, convertibles), and unfair control rights (such as performance/forfeiture provisions, short-term employment contracts, tranched financing). Further, we have noted that non- factors, such as regular communication and empathetic personal contractual relationships can promote fairness, trust and venture performance. Our model provides a basis for future research. The first challenge is to formally model these fair and unfair cash-flow and control rights. For example, how do we incorporate the fair and unfair features of equity and convertibles, as identified by Cumming (005) and Tucker (004), into a behavioural fairness model? Secondly, we have taken fairness as an exogenous initial input into the game, which subsequently affects equity allocation, effort levels, and performance. However, the VC/E relationship is dynamic and inter-temporal. As Utset (00) and Tucker (004) note, fairness and trust can be damaged over time. We suggest that future researchers develop a dynamic model, with fairness endogenously derived throughout the game. Finally, we have focussed on reciprocal fairness, taking trust and trustworthiness as an exogenous characteristic of the players. In future research, we will incorporate reciprocal trust into our behavioural game-theoretic approach to venture capital contracting. Appendix. 33

34 1. Proof of propositions: Proof of Proposition 1: We derive the date optimal effort level of E by solving S E = 0 in equation (1), e noting that E S knows that the VC is self-interested (and is therefore trustworthy with probability t ). We obtain the self-interested E s optimal effort level, given that he has matched with the self-interested VC, as follows; tαγr e * =. (A.1) β E W e note that, from equation (1), = β < 0 for all e. Therefore, the optimal e effort level given in (A.1) indeed maximises E. Given (A.1), the expected value of the project is tαγ R V = γer = β. (A.) We now move back to date 1 to solve for the trustworthy VC S ' s optimal equity proposal, given that VC S knows that she has matched with a self-interested E. VC Therefore, substituting e quation (A.1) into (), and solving = 0, we obtain the α 34