Independent Directors and Shared Board Control in Venture Finance

Several papers specifically model the allocation of control in VC-backed firms (Berglof, 1994; Hellmann, 1998, 2006; Kirilenko, 2001), and VC contracts are one of the few empirical settings used to test financial contracting theory (Kaplan and Strömberg, 2003; Cumming, 2008; Bengtsson and Sensoy, 2009, 2010).

Not all firms that share control with an independent director are under ID-arbitration. The board may have more than three directors or multiple representatives from each group. I focus on ID-arbitration, however, because it is the simplest form of shared control. ID-arbitration applies to firms with multiple entrepreneurs or VC investors sitting on the board, provided the entrepreneurs as a group have similar interests and the investors as a group have similar interests. While this is generally a reasonable assumption there are cases where the interests of early investors may diverge from later round investors (Bartlett, 2006).

Board authorization applies to a much broader range of corporate actions than shareholder voting (Clark, 1986), and the allocation of board seats can be decoupled from the distribution of shareholder votes (Kaplan and Strömberg, 2003). Consequently, even if a firm’s VCs control the vote of shareholders they do not necessarily control the board of directors.

Two notable exceptions are (i) Yerramilli (2006), who shows that joint control coupled with a harsh penalty if the entrepreneur and VC fail to reach an agreement may be preferable to unilateral control, and (ii) Meyersson Milgrom et al. (2007), who model joint control arising from equityholder veto rights over specific classes of decisions. These studies, however, apply to a form of shared control which requires consent of both parties, and they do not address the use of third party independent directors.

This distinction is particularly relevant whenever the independent director prefers an action that neither the entrepreneur nor the investor would select if given control. Under this scenario ID-arbitration creates an incentive for compromise that is not present under state-contingent control.

To further illustrate this distinction note that ID-arbitration and state-contingent control are not mutually exclusive. The financing contract may call for ID-arbitration as the normal board arrangement but specify that the VCs will acquire additional board seats if the firm fails to meet certain performance targets, effectively shifting from ID-arbitration to VC-control in bad states of nature. Kaplan and Strömberg (2003:288) find evidence supporting the simultaneous use of ID-arbitration and State-contingent control: shared board control declines (61–52%) and VC control increases (25–36%) in the adverse state. Approximately 10% of VC-backed firms appear to use both ID-arbitration and State-contingent control simultaneously.

In addition to mechanisms discussed above, contracting parties can prevent holdups though an ex post message game (Maskin and Tirole, 1999) or renegotiation design (Aghion et al., 1994).

As noted by Maskin and Tirole (1999) this type of contractual incompleteness could in theory be circumvented by ex post message games. While I acknowledge this limitation of the incomplete contracting literature, the structured bargaining game that I model in Section 4 generates the first-best outcome under various parameter values, removing any need for ex post messaging.

This assumption, also used in Aghion and Bolton (1992), can affect the ex ante feasibility of each governance arrangement; however, this paper’s main results do not depend on the distribution of bargaining power or even the availability of renegotiation. In fact, if renegotiation were unavailable or costly, the benefits of ID-arbitration would be increased relative to E-control and V-control (see discussion in Section 4).

My analysis throughout this paper assumes that each non-independent director will act in the interests of the constituency – investor or entrepreneur – that he represents. This view is potentially at odds with corporate law, under which directors have a fiduciary obligation to serve the best interests of the corporation and its stockholders. In VC-backed firms, however, fiduciary obligations place little constraint on opportunistic behavior (Fried and Ganor, 2006), suggesting it is reasonable to ignore this legal constraint. Alternatively, the action set A can be thought of as the set of actions consistent with these legal obligations.

If E were allocated all of the monetary returns (i.e. if then and E would choose the optimal action; however, this would necessarily violate V’s individual rationality constraint.

The parties cannot contract over future actions at date 0. Thus technically, they are negotiating the choice of action for the first time after date 1. However, following the literature I refer to this as “renegotiation” throughout the paper, emphasizing that this bargaining occurs after the original contract and after the state of nature is revealed.

In practice VCs typically hold convertible preferred stock, while entrepreneurs hold common stock (Kaplan and Stromberg, 2003). Potential conflicts can arise between preferred stock and common stock (Fried and Ganor, 2006; Hellmann, 2006; Broughman and Fried, 2010), and a tiebreaking independent director may be desirable in this context as well. However, because preferred stock is endogenous to the financing contract I focus instead on the more basic tradeoff between private benefits and financial returns. The same intuition, however, should extend to other conflicts that may arise in a startup firm.

For ease of presentation I drop most references to θ and π from the notation in the following sections, except where needed for clarification. For example, will be used instead of .

We assume that investment is feasible for at least some allocation of cash-flow and control rights. Since maximizes the Firm’s monetary returns it is equivalent to say that .

I do not explicitly model the cost of adding an independent director to the board. Rather, the model below implicitly assumes zero cost. Even though the cost of an independent director would generally be quite small, this is an arguably unrealistic simplification. Alternatively, one could refine the model by stating that ID receives a fixed payment equal to C for agreeing to serve on the board. Because, the firm needs additional funds to pay the ID it follows that V’s investment at Date 0 would increase to K+C. Similarly, under ID-arbitration V would demand a larger share of the cash flow rights, sufficient to give V an expected payoff of . This refinement would not change the analysis associated with Propositions 3 and 4 below (only change is that “K” would be replaced by “” throughout). This refinement would, however, complicate the comparative static results discussed in Section 4.5 and illustrated in Figure 2. The choice between ID-arbitration and V-control would not simply depend on the amount invested, but also on the cost of adding an independent director. See discussion in footnote 22 below for more details.

Collusion with another party would violate the director’s fiduciary duty of loyalty.

It should be noted that a particularly robust (and unrealistic) conception of fiduciary obligations would prevent all instances of ex post opportunism (Hart, 1993; Broughman, 2010). In practice, however, the business judgment rule prevents judicial review of most operational decisions, unless the decision-making process was tainted by self-dealing (Clark, 1986).

This example is suggested by McCarty and Meirowitz (2007) at pages 105–107.

In this example, the ID’s preference ordering (i.e. g) is symmetric about its optimum (i.e. for any ). Consequently, we can simply consider the distance between each proposal and to determine the selected proposal.

We can compare the renegotiation payments under E-control and ID-arbitration. To reach the first-best under E-control the investor must pay the entrepreneur an amount equal to ; however, under ID-arbitration the investor only needs to pay , where and are the cash-flow rights awarded to E under each governance arrangement respectively. The second renegotiation payment is smaller since and .

If an ID brings additional cost or uncertainty the scope of Range 2 may be somewhat smaller (see discussion in Section 4.4).

I am assuming that there is no uncertainty regarding ID’s preferred outcome. I am also assuming that there is no cost to adding an independent director. As noted in footnote 16, we could alternatively model ID-arbitration as fixed cost “C” that V would come out of V’s initial investment. This refinement would introduce a tradeoff between (i) the cost of adding an independent director to the board, and (ii) the risk of ex post inefficiency under V-control. Range 2 would effectively become “Medium K and Low C”, while Range 3 would become “High K and High C”, and the comparative statics would be adjusted accordingly. If we introduce cost or uncertainty associated with ID-arbitration, the scope of Range 2 may be somewhat smaller, pushing some firms to use V-control rather than ID-arbitration.

Meyersson-Milgrom et al. (2007) addresses the use of targeted shareholder veto rights.