Preinvestment Negotiation Characteristics and Dismissal in Venture Capital-Backed Firms

From management turnover studies, we know that inadequate performance and management inability may result in the dismissal of chief executive officers (CEOs). With regard to new ventures, however, the issue of dismissal is more controversial. The dismissal of new venture team (NVT) members in knowledge-intensive firms is equivalent to dismissing a significant part of the firm because key tacit knowledge is often embedded in NVTs and their networks. We then regard NVT as the (top) management team in a new business venture. Hence, it is in the interest of the venture capitalist (VC) to avoid situations that could lead to dismissals that are detrimental to the performance of the portfolio company. NVTs will obviously seek to avoid being dismissed from their positions in the firm when they derive a personal benefit from continuing.

Within venture capital research, empirical investigation of the VC–NVT relationship has primarily focused on the contractual and structural aspects of venture capitalists' governance of their portfolio companies. This research has focused on incentive schemes and contractual issues and has often used insights from agency theory (Jensen and Meckling 1976; Fama and Jensen 1983) and principal–agent theory (Holmström 1989).

Another recent stream of research, explicitly pronounced in several studies carried out by Harry Sapienza and his colleagues (2000), has examined the procedural and relational aspects of VC governance. Rooted in procedural justice theory, this research has suggested that the relational processes and perceived fairness of the parties' behavior influence the levels of cooperation between venture capital firms and NVTs. As in the literature on interfirm cooperation in general (Das and Teng 1998), VC research has begun to explore the complementarities of relational and structural governance (Sapienza et al. 2000). In particular, James Fiet and his colleagues (1997) investigated the dismissal of venture managers and found that both structural and relational issues were involved in their dismissal.

In this study, we examine how two main governance mechanisms, contractual governance and relational governance, transform practice. Research focusing on contractual issues is often rooted in agency theory (Jensen and Meckling 1976), transaction cost economics (Williamson 1975, 1985, and 1988), or in incomplete contracting theory (Grossman and Hart 1986; Hart 2001). Contracts represent the structural dimension of governance and represent promises or obligations to perform particular actions and deliver certain outcomes in the future. In other words, contractual governance involves the structuring of contracts or rewards and punishment so that individuals behave in a prespecified manner (Bhattacharya, Devinney, and Pillutla 1998).

Relational governance, on the other hand, involves norms of cooperation and obligation that govern interfirm relationships (Ring and Van de Ven 1992 and 1994). In particular, a growing body of research on social control theory suggests that the social embeddedness of interfirm cooperation may also establish behavioral norms that act as implicit controls governing the relationship (Larson 1992). In this literature, the role of trust is especially critical for discouraging malfeasance (Granovetter 1985). Indeed, “in an embedded logic of exchange, trust acts as the primary governance structure” (Uzzi 1997: 61). Hence, relational governance is a mechanism largely based on trust.

Negotiation scholarship and governance scholarship share some key concepts, especially when it comes to the implementation of governance logics in practice. A focus on contractual governance, as outlined above, encourages positional bargaining, whereas an emphasis on relational governance encourages more collaborative negotiations.

Research indicates that VCs frequently use their power to dismiss managers in their portfolio companies (Rosenstein 1988; Gorman and Sahlman 1989; Schefczyk and Gerpott 2000). For instance, Fiet and his colleagues (1997) applied agency, power, and procedural justice theories to develop a complementary theoretical perspective on dismissals. In particular, they found that contractual covenants that limit salaries paid to venture managers reduced their likelihood of dismissal and that contractual covenants effectively aligned managerial financial incentives with those of the investors and the board of directors. They also found that dismissal covenants that forced a change in the management team were ineffective means of preventing dismissal: specifying the conditions that would lead to a change in management was an ineffective mechanism for aligning the investor's interests with those of the NVT. Furthermore, they identified a negative relationship between revenue growth (per employee) and dismissal, suggesting that weak performance may indeed lead to dismissal. They further found that the number of seats on the board of directors was negatively related to management dismissal (e.g., the more board members there were, the less likely that managers were to be fired), but the effect was reversed with an increasing number of investors on the board.

Moreover, Gary Bruton and his colleagues (1997) found that it is CEO failure in strategic matters, and not operational ones, that led to dismissal. They also found that management replacement typically improved new venture performance. In a follow-up article, (Bruton, Fried, and Hisrich 2000) they reported that the primary reason for dismissing managers was agent inability followed by good faith disagreements between principals and agents (e.g., owners and managers), and managerial opportunism ranked last. Beyond the above findings, little is known about what other factors may trigger dismissal in new ventures.

Hans Landström and his colleagues (1998) were among the first scholars to apply negotiation theory to the domain of entrepreneurship. They argued that the initial negotiation agreements between investors andentrepreneurs may indeed shape their relationships. For instance, they suggested that future studies should use agency and social exchange theories. Accordingly, we consider contractual and relational governance as they represent two salient perspectives within the literature on governance of interfirm relationships (Poppo and Zenger 2002; Ness and Haugland 2005; Lee and Cavusgil 2006).

One major focus of negotiation theory has been the distinction between (zero sum) fixed-pie and variable-pie perceptions or what has been labeled as distributive and integrative negotiations. The former is also sometimes known as positional bargaining and the latter as collaborative bargaining. Depending on each parties' interests (preferences and priorities) and whether the parties' interests overlap or not, each will most likely pursue the negotiation strategy that he or she believes is most beneficial for reaching his or her own goals. A high level of concern for one's own interests may be combined with high or low levels of concern for the other party's interests (Ness and Haugland 2005). In situations in which a negotiator has high levels of concern for his own interests combined with low levels of concern for the other party's interests, the negotiator is most likely to choose a positional bargaining strategy. Positional bargaining is a rights-based negotiation process in which the focus is on securing one's own interests. Such a negotiation strategy has similarities with some of the key features of contractual governance.

In negotiations in which both parties pursue shared interests and feel mutual concern for each other's interests, collaborative negotiations are more likely to be their preferred strategy (Ness and Haugland 2005). Collaborative bargaining and relational governance share some features, including the central role of trust.

The strategies chosen in a preinvestment negotiation may thus signal concern for the other party's interest and also indicate the level of the parties' goal congruence: a high level of goal congruence between the investor and the investee will reduce the agency risk (Eisenhardt 1989), for example, the risk that the manager may be bargaining for himself, and less for the firm. Accordingly, collaborative strategies may indicate less conflict of interest and thus reduce the probability that venture capital firms will dismiss managers of the companies in their portfolio.

To avoid agency problems such as moral hazard and adverse selection, investors may use two major approaches to contractual governance: legal protection and concentration of ownership. These two common approaches both rely on giving investors power (Shleifer and Vishny 1997). With the first approach, investors construct contracts to protect against expropriation by managers, while with the latter approach, they may seek a substantial portion of control rights incorporated in the ownership. Although both mechanisms enhance investor power, the logic of power inherent in the mechanisms is distinct. Robert Kahn (1964) distinguished between two categories of power: absolute power and relative power — the concentration of ownership gives the investors relative power, while legal protections give them absolute (specific, definable, and limited) power. For instance, the very process focusing on the allocation of rights and contractual covenants may indicate concern for one's own interests at the sacrifice of the other party's interests. Fiet and his colleagues (1997: 348) concluded from their study of CEO dismissals that “specifying the conditions that will lead to a change in management is an ineffective mechanism for aligning the VCs interests with those of the NVT.”

Jay Barney and his colleagues (1994) noted that contractual covenants are costly to write and enforce, and thus will only be included in a contract if their benefits are greater than the cost of writing and enforcing them. Hence, the very process of focusing on the allocation of rights and of including contractual covenants in the negotiations may signal an increased probability that these rights will be enforced.

Furthermore, as suggested by those scholars who espouse the procedural justice perspective, the subsequent behavior of the parties depends on their perception of the decision process rather than on decision outcomes (Sapienza and Korsgaard 1996). Hence, in addition to the outcome in terms of the allocation of rights and contractual covenants, the negotiation process itself may frame the subsequent behavior of the parties in the relationship. According to Lowell Busenitz and his colleagues (1997), contractual covenants can adversely affect perceptions of whether the actions undertaken by venture capitalists are just. That is, placing too many restrictions on new venture management via the use of contractual covenants may damage the relationship (Busenitz et al. 1997). Consequently, an overemphasis on contractual covenants — being too positional — in the preinvestment phase leads us to our first hypothesis:

As a counterbalance to positional bargaining, we also explored the impact of collaborative bargaining in the preinvestment phase and its likely effect on postinvestment dismissal. Collaborative negotiations are known as integrative or problem-solving negotiations and have also been characterizedas variable-sum negotiations or as “win–win” negotiations. This type of negotiation is not only interest based, but relational. That is, the negotiator displays a high level of concern for the other party's interests, as well as for his or her own. This relational-based view emphasizes the expectation of behavior irrespective of one's ability to monitor or control the other party (Mayer, Davis, and Schoorman 1995). In fact, it reflects the concept of relational governance with all its ingredients and facets, including the building and maintaining of trust. Thus, a relational orientation in the preinvestment negotiations will signal a reduced likelihood of post-investment dismissal. Accordingly, we have developed the following hypothesis:

It could also be argued that collaborative negotiation represents an “internalization” of an agreement, where the agreement is governed by confidence in a partner relationship, whereas the results of a negotiation process characterized as positional bargaining often have to be “externalized” in a contract to maintain the negotiated outcomes. This is a critical distinction between these two approaches. Under conditions associated with the first hypothesis, the behaviors that reflect a mindset that focuses on rights will likely be accompanied by a tendency to enforce these rights when those conditions emerge. Under the conditions affiliated with the second hypothesis, we would expect that the parties together would handle new challenges based on a set of norms of cooperation developed through an understanding of each party's interests. Typically such a process is characterized by mutual trust, reasonable behavior, and consensus-driven solutions.

To accumulate our data, we surveyed the CEOs of 240 current and newly exited portfolio companies held by members of the Norwegian Venture Capital Association as of March 2004. We included only “classic” venture funds, leaving out firms involved in buy-outs or funded by half-public seed funds. Seventy companies returned their questionnaires, resulting in a response rate of 29 percent.

Most of the companies that responded were in their expansion phase. Fifty-four of the respondents were CEOs; sixteen held other management-team positions (e.g., chair, chief financial officer, chief technology officer). Only twenty-seven of the respondents identified themselves as the founding managers.

We checked for nonresponse bias by comparing respondents and nonrespondents according to business sector and sales. We found no significant differences across these variables. The average amount invested in these companies was 34 million Norwegian kroner (approximately $5.6 million). The ownership stake of the venture capital firms ranged from 10 percent to 90 percent, with almost one-third holding between 40 percent and 60 percent of the equity of the companies.

Our dependent status variable in this analysis was how many managers had been dismissed. In the Cox regression model we used, however, the dependent variable was the hazard rate, which was the probability that an event (dismissal) would occur within a particular time interval to a particular firm at risk during that time interval.¹ Thus, the hazard rates represent the longitudinal risk profiles for the managers in the sample. The hazard rate is constructed from three recorded variables. The investment year and the censor year² are used to construct the number of years that the managers stayed in their positions, which defines the hazard rate together with the status variable NVT member dismissal.³ To determine the status variable dismissal we used two sources. First, we asked the respondents whether management turnover had occurred in their company after VC investment. Then, we asked whether this was actively triggered by the VC. Based on responses from sixty-three managers (answering on behalf of their management teams, which totaled 262 NVT members), thirty-five answered that management turnover had occurred and twenty of these stated that this was actively triggered by the VC. That means that 57 percent of the turnover was actively triggered by the VC. This is strikingly similar to the findings of Michael Schefczyk and Torsten Gerpott (2000) who reported that 58 percent of CEO turnover in their German study was actively triggered by the VC. In our study, only those considered actively triggered by the VC were defined as dismissals. We used data from the national database to verify the replacements and recorded the time of dismissal.

The two negotiation strategies used by the parties in the preinvestment process were labeled as either positional bargaining or collaborative bargaining. We obtained these measures by analyzing eight statements regarding the prenegotiation process as perceived by the respondents. Specifically, we asked the respondents to agree or disagree with the eight statements according to a seven-point scale, with 1 indicating strong disagreement and 7 indicating strong agreement.

Because one statement applied to both the styles, it was removed, and the results of the new factor analysis (with seven items) are shown in Table One. The table illustrates that all the remaining items load appropriately on their respective factors.

As a measure of the VC firm's relative power, we used its proportion of board representation (the firm's number of seats divided by total number of seats). We then assumed that their shares of the cash flow rights are represented at the company's board of governance. We also controlled for venture cycle stage (transformed with the natural logarithm) and the size of the venture management team.

Longitudinal data analyses create certain statistical challenges. Problems arise when the individual cases are tracked over different time periods and when the event of interest (in this case CEO dismissal) does not occur. Using the amount of time before the dismissal as the dependent variable, the analysis would exclude data from those cases in which the executives were not dismissed.⁴ The exclusion of censored cases can produce large biases (Sørensen 1977; Tuma and Hannan 1978). Furthermore, using thestatus variable of dismissal/nondismissal would exclude data about dismissals on either side of the defined study period. This means that if the study period were five years, the method would neither distinguish between a non-dismissed CEO after one year and those avoiding dismissal for four years.

Event history models deal with these kinds of problems and, in the case of our study, actually make use of both dismissal and years of non-dismissal in constructing the hazard rate.⁵ The hazard rate is then used as the dependent variable in the regression (in our case a Cox regression). Thus, the method is advantageous because it both provides efficient information and avoids the biases associated with “censoring.” That is, the event history model does not exclude information from nonevent cases, which is the case with logit regressions; rather, the model makes use of the information from the time until event in the event cases.

We measured the data in this study on a discrete basis (years), but the empirical model we employed, a Cox regression, is a continuous-time hazard model, assuming that events occur at any point in time.⁶Table Two lists the correlations, means, and standard deviations for the variables used in this study.

We used Cox regression to conduct an event history analysis of dismissals. Because some data were missing from seven of the responses, we only had complete data regarding dismissal from sixty-three companies. Out of these, eighteen had experienced management change actively triggered by the VC. Using a cross-tabulation with a dichotomized positional variable, we found that fourteen of the eighteen dismissals were exposed to various degrees of positional bargaining preinvestment. For the main analysis, wemade three sets of models to separate the effects of the control variable and the two foci variables. The results of the Cox regression analysis are shown in Table Three.

We found support for the first hypothesis, which tested the effect of positional bargaining on the likelihood for postinvestment dismissal, finding strong support for the relationship between positional bargaining during the preinvestment period and the likelihood of postinvestment dismissal. The data, however, failed to support the second hypothesis, which tested the effects of collaborative negotiations on subsequent dismissal. That is, we found no significant relationship between collaborative negotiations preinvestment and the likelihood of subsequent dismissal. An insignificant relationship in the hypothesized direction is present, however (−0.121 in Block 3). Finally, we found no relationship between whether or not the VC firm was represented on the board or the size of the NVT and dismissals, neither between cycle stage and dismissal.

The results indicate that using a positional negotiating style in the pre-investment phase can have a strong significant effect on dismissal when things get tough postinvestment. This is a valuable and significant finding: it suggests that entrepreneurs should avoid investors who focus too much on their own rights in the preinvestment negotiation phase because they will have a tendency to demonstrate the same behavior after the investment. That is, if the negotiation process in such an important — and fragile — development stage is characterized as positional, entrepreneurs should expect that things will get equally tough, or even tougher when unexpected events occur after the investment.

Preinvestment collaborative bargaining apparently has no bearing on postinvestment dismissal. We hypothesized that collaborative pre-negotiation behavior would decrease the likelihood for subsequent post-investment dismissal. One possible explanation for this lack of support is that a collaborative attitude may change if unforeseen events occur during the development process — for example, if the company does not develop as expected. Although collaborative strategies should suggest less conflict of interest, and thus less agency risk, the results of this do not support that idea. In other words, the ability of the parties to solve good faith disagreements through negotiations does not appear to have a bearing on the likelihood of subsequent dismissals.

As we have seen, positional negotiation behaviors are synonymous with rights-based claims. To investigate this further, we also asked about the inclusion of various mechanisms in the negotiated agreement. Specifically, we asked “to what extent are the following mechanisms important in the contract between the management team and the venture capitalist?” (See items 3 through 16 in Table Four.) Respondents were asked to rate their answers on a seven-point scale with 1 meaning “to a very small extent” and 7 indicating “to a very large extent.”

The results shown in Table Four provide additional support for the idea that positional bargaining behaviors are associated with tougher measures in the negotiated agreement. That is, positional negotiations correlate significantly with most of the mechanisms in the contract. As such, it provides additional validity to the positional bargaining construct. None of the items correlates significantly with collaborative bargaining, also reflecting additional support for that construct.

To ease the interpretation of this correlation matrix, we ran an explorative principal component analysis with varimax rotation, and the method extracted four components with eigenvalues greater than one. They are: innovation protection, economical restrictions on management, other restrictions on management, and employee incentives. When we controlled for these four components, the effect of positional bargaining improves.

Moreover, we also asked what difficulties the managers perceived during the preinvestment negotiation process. Specifically, we asked them “How difficult was it to negotiate with the venture capitalist about the following items?” They again answered on a seven-point scale, with 1 meaning “very difficult' and 7 meaning “very easy” (see Table Five).

Again, to ease the interpretation of this matrix, we ran an explorative principal component analysis with varimax rotation, and our method extracted three components with eigenvalues greater than one. They were: difficulties negotiating performance milestones, difficulties affiliated with negotiating company valuation, and difficulties negotiating shareholder agreements and investor exits. Controlling for these items did not significantly change our results with regard to positional bargaining, indicating that our results are robust across multiple context variables.

Clearly our results indicate that the parties in a funding negotiation should write covenants cautiously, as claim-based negotiation processes are strongly related to the presence of various mechanisms in the contracts, as Table Four indicates. Insights from negotiation theory prescribe a relational orientation in durable relationships. If parties need to be positional, it is important that they conduct their initial and subsequent negotiations in a procedurally just — and collaborative — manner. Collaborative negotiations are best conducted by making trade-offs of issues of concern to both parties, a process also known as logrolling (Froman and Cohen 1970). If the parties become too positional, they may signal a concern for their own interests at the expense of the other party, risking a reduction, rather than an expansion, of the “pie.”

Some would argue that the managers participating in this suffer from what we may call retrospective rationality, that is, the tendency to form a perception after the fact — a form of hindsight bias. It is the NVT's retrospective reflection of the negotiation process, however, that is the appropriate antecedent.

Also, we do not intend to imply that VCs typically have a rights-based position. It may well be that it is the members of the NVT who are too concerned about their rights. At least, that appears to be the case with regard to Table Four where “innovation protection” reflects a separate component in the preinvestment negotiation. Normally, if even only one party is overly concerned about his or her rights, it is sufficient to facilitate an unfavorable negotiation atmosphere that can turn the negotiation into a positional battle.

Future research could seek to assess in more detail what other factors may trigger the decision to dismiss managers in portfolio firms. Studies of the effect of forcing strategies and difficult contractual covenants — with regard to innovation protection and various incentives for and restrictions on the management team, for example — would also be worthwhile.

In this study, we have demonstrated that positional bargaining in the pre-investment phase may indicate the probability that there will be managerial turnover postinvestment. Our findings suggest that entrepreneurs should avoid business partners who are too positional in the preinvestment negotiation phase because they are likely to exhibit the same tendency after the investment. In other words, the long-term effect of contractual governance may be management replacement when the venture capitalist encounters bumps in the road.

Our results also have implications for investors: they should be cautious when negotiating with managers lest they send the wrong signal by being overly focused on rights and contractual covenants. Although positional bargaining may help secure a larger share of the pie, it may also limit the growth of that pie because subsequent interactions may be hampered by mutual mistrust. In the fragile and often turbulent world of new ventures, mistrust could prove costly for everyone involved.