I just read an interesting paper about the venture capital industry titled “Learning to Govern: Venture Capital Firms and the Replacement of Founder-CEOs”, written by Timothy Pollock of Penn State University , Ted Baker of North Carolina State University, and Bret R. Fund of Penn State University. This paper was presented on April 15, 2006 at the Entrepreneurship Research Conference sponsored by the Robert H. Smith School of Business at the University of Maryland[click on prior highlighted text for link to the conference and to the paper].
This research document made several hypotheses about venture capitalist behavior patterns associated with the decision to replace or retain Founder-CEO’s in portfolio companies. The report tests these hypotheses through mathematical models of investment performance that analyze causality and signs of adaptive learning in a sample of about 400 Internet IPO’s from 1995 – 2000. I thought that the most useful conclusion from the paper was that it questioned whether the sample data, and therefore its conclusions, made any sense given the anomalous period of time chosen for the sample:
“Overall, our results provide limited evidence of experience-based learning about the fundamental governance question of whether to replace founder-CEOs in Internet companies by VCFs. We found little evidence that VCs engage in any adaptive learning … of the sort that might affect future decisions making about whether to retain or replace founder-CEOs. VCs’ experience observing their own prior decisions about whether to replace CEOs in portfolio firms and observing the subsequent performance of these firms had no apparent effect on later replacement decisions. Contrary to our expectations, we found that the VCFs’ prior post-IPO investment performance experience in the new sector did not help, and frequently may have hurt performance on later deals. Apparently, experience with success did not provide lessons in future success. This may be due in part to regression to the mean, but because our investment performance measure was averaged across prior Internet sector deals, this is likely not a complete explanation. We speculate that this finding may result in part from a pattern in which some VCs, and especially those that experienced success with a prior investment, may have become progressively less choosy about their investments, especially as the mania in this market increased. This pattern would be consistent with contemporary reporting by the business press about VCF practices during this time, suggesting that firms which during other times would not have received the time of day from a VCF were showered with cash and quickly rushed to IPO during the Internet Bubble…"
The researchers’ speculation is correct. In fact, the risk-reward equation was inverted during this period and risk capital was, for a very brief period of time, effectively free for VC-backed Internet ventures. So this wasn’t a good sector or a good time from which to be drawing conclusions as to how VC’s were or weren’t making CEO replacement decisions. Unfortunately this means that the sample data is irrelevant and the conclusions of the report have no basis for support.
A more useful point that the researchers could have made is that IPO’s typically account for less than 20% of venture-backed company exits (current statistics are around 8:1 in favor of M&A), with mergers and acquisitions accounting for the vast majority of VC liquidity over the past 40 years. Consequently, it would be flawed to ever use IPO-based metrics regarding investment performance to correlate any meaningful relationship with regard to the propensity for VC’s to replace founder-CEO’s. Recognizing that it is difficult for researchers to access reliable data outside of SEC-filings such as IPO’s, I would suggest that a more extensive broad sampling of actual practitioners, with the methodology vetted by an organization such as the NVCA, could be far more reliable.
More troubling and inaccurate is the paper’s assertion that VC’s have a mental default setting to always replace Founder-CEO’s. The authors reference my paper “Rites of Passage: Managing CEO Transition in Venture-Backed Technology Companies”, as well as Professor Noam Wasserman’s 2003 paper on Founder-CEO replacement to support this flawed assumption:
“Wasserman (2003) described rationalized myths adopted by VCFs and expressed in the form of largely untested “rules of thumb,” including the rule that all-else-equal, the founder-CEOs should be replaced. This belief is consistent with a great deal of the practitioner and academic literature on life-cycle models of firms (e.g., Flamholz & Randle, 2000). Indeed, Wasserman quotes a partner of one VCF he studied as noting that, “Our default assumption when we first look at a company is that the founder-CEO can’t lead this company going forward (2003:154-5).” Wasserman pushed for an explanation of how the VC “had arrived at this rule of thumb” and the partner “admitted that it was a widely held belief within the venture capital industry and one that he had not questioned (Wasserman, 2003: 167).” Because increasing uncertainty makes it more likely that firms will follow institutionalized rules, we expect that as firms learn about the uncertainty of the internet startups they have funded, they will be more likely to follow the “replace-the-founder” myth, and therefore hypothesize that, H1: The greater the uncertainty perceived by a VCF, the less likely the VCF is to retain a founder as CEO . . .”
“Given that replacing founders is considered the default action in the venture capitalist industry (Levensohn, 2006; Wasserman, 2003), there are likely to be industry-wide routines available for managing this process. In contrast, there is less likely to be widely-held knowledge about how to most effectively keep and work with a founder-CEO. The important, hard to predict and nuanced tradeoffs among the risks and benefits of retaining the founder-CEO suggest that firms may benefit from the development of experience-based competencies at managing the retention and behavior of founder-CEOs in a way that minimizes issues of narrow CEO experience and maximizes the benefits of founders’ idiosyncratic skills, knowledge and continuity of leadership.”
I will assert unequivocally that replacing founders is NOT considered the default action by venture capitalists in the industry.
I called Professor Noam Wasserman at the Harvard Business School this afternoon and asked him if he felt that the authors of this paper correctly interpreted his research, and he also disagreed with their conclusion quoted above.
The more relevant point made in both “Rites of Passage” and in “Founder CEO Succession and the Paradox of Entrepreneurial Success” is that there are structural elements particular to the venture capital industry that account for the frequency of Founder-CEO replacement and of CEO replacement in general—these elements are driven by the material difference in requirements from a CEO between a seed stage R&D company and a company experiencing rapid acceleration of revenues in its early commercialization.
A recent NVCA survey of 375 VC directors also highlights various skill set mismatches as the top reasons for CEO replacement. In “Rites of Passage”, one of the basic propositions is that VC’s can do a better job of identifying the shortcomings of CEO’s earlier, whether they are founders or not, and take specific actions to manage the situation constructively. The reality of the CEO replacement process is that it is always highly disruptive and can be crippling when managed poorly.
The cost of a CEO replacement to venture capital funds and employee shareholders easily reaches 10% dilution, even if things go relatively well. It is in the interests of all shareholders to minimize those costs and to strive to avoid them altogether. Statistics show that the majority of venture-backed CEO’s are replaced at least once. Asserting that this means replacing founders is “the default action” for VC’s is simply wrong.
To conclude this post, I will quote the report once more:
"Together, these factors suggest that VCs are much better positioned than most firms to learn under dynamic and ambiguous conditions: they have strong incentives, clear and consistent goals and measures, and repeated opportunities to try their hands and observe the results. From this perspective, we find it remarkable how little learning we observed. As we noted, if we had not taken a very broad perspective on learning, we would have likely concluded that no performance-enhancing learning took place at all. This suggests that learning under highly dynamic and ambiguous conditions is even more difficult and unlikely than prior research has indicated, and warns against underestimating the role of randomness and luck in firm survival and success in such environments.
I conclude that the authors should have written a paper on a different subject. I find it remarkable that they persisted in publishing these results and drawing these conclusions when they acknowledge that their data set could be meaningless and that the results don’t make intuitive sense. I am also disappointed that they misread basic tenets from both my work and that of Professor Wasserman’s. For a more deeply informed commentary on VC’s and Founders, I do recommend that people read Noam Wasserman’s Founder Frustrations Blog at http://founderresearch.blogspot.com/
You must be logged in to post a comment.