Nearly all the research detailed below relies on the generous access to the VentureOne database provided by VentureSource and Correlation Ventures.
Working Papers
A New Model of Venture Capital Risk and Return (supplementary appendix)
I formulate a model and estimator of venture capital (VC) returns motivated by the entrepreneurial firm life-cycle and the extreme return outcomes of typical venture capital investments. The model incor- porates tail events and the estimator corrects for sample selection bias and endogenous investment holding periods. I find that an asymmetric three-state mixture distribution is a better characterization of returns than the standard single-state model. Mixture states mimic typical VC outcomes: “winners,” “break- even” and “failures.” Imposing normality on venture capital investment returns understates downside risk and kurtosis. In contrast to earlier studies, the mixture model reveals a leptokurtic, negatively-skewed returns distribution. Two new implications follow from the results. First, volatility as an estimate of risk underestimates the frequency and magnitude of large, negative VC returns. Investors in venture capital may need to incorporate additional moments or semivariance into their allocation decisions. Second, a microcap index benchmark previously shown to mimic the means and CAPM alphas of VC returns lacks the downside risk or fat tails of the VC mixture distribution. Thus, VC investments offer some risk and return features unavailable in publicly traded equities.
Entrepreneurial Activity of Venture Capitalists
Each year, about 15% of new venture capital firms are formed by venture capital partners at existing firms. This paper investigates the consequences of these employee spinoffs for parent firms and the characteristics of their founders. As predicted by the spinoff formation model of Cabral and Wang (2009), high type partners and low type partners at failing parent firms are the most likely to leave and start a new venture capital firm. After accounting for the endogeneity of the timing of spinoff formation, estimates reveal a large, negative impact of spinoffs on parent firm performance. The direction of the endogeneity bias suggests that spinoff founders time their exits to coincide with success at the parent firm. The negative impact of spinoffs shows that individual venture capital partners can play a crucial role in the performance of their firms. Thus, findings on venture capital firm characteristics such as performance persistence (Kaplan and Schoar (2005)) or social networks (Hochberg, Ljungqvist and Lu (2007)) may be a partner-level phenomenon.
The Transfer of Social Networks and Entrepreneurial Spawning
The source of the competitive advantage of employee spinoff firms and their formation’s negative impact on parent firms is often attributed to a transfer of knowledge from parent firm to spinoff. This paper studies one potential asset transfer in the venture capital industry that can explain both patterns: social networks. Results comparing the social network evolution of venture capital spinoffs and other new firms shows that the former have more central network positions and gain those positions more quickly. Non-spinoffs are more likely to join smaller, more isolated cliques in the network. Spinoffs also have higher survival rates, co-invest with more experienced investors and produce more positive exit outcomes for their investments. The results suggest that parent and employee relative network status in other human capital-intensive industries predicts spinoff formation. The conclusions also extend the Ljungqvist, Hochberg and Lu (forthcoming) study of entry costs and social networks in the venture capital industry by detailing the differences in social network evolution of new entrants.
Estimating Statistical and Preference-based Racial Discrimination in the US Apartment Rental Market (with Bryan Tomlin and Choon Wang)
We test for statistical discrimination in the apartment rental market by responding via email to landlords postings their vacancies on craigslist.org. in 35 U.S. cities. By manipulating the level of positive or negative information included in over 14,000 fictitious inquiries with randomly assigned black or white sounding applicant names we are able to evaluate the effect of information on differential treatment by race. We find evidence of preferential treatment of whites when no additional information is provided in an inquiry, both in terms of the likelihood of receiving a response and the likelihood of receiving a positive response. The gap in differential treatment is unchanged when positive information is added, though the gap diminishes in the presence of negative information for males and increases in the presence of negative information for females. These results suggest that landlords have a race-based prior expectation of applicant quality that is (dis-)confirmed by the presence of negative information, and curiously uncorrelated with positive information. Our findings challenge the theory of statistical discrimination and suggest that preference based discrimination is likely to play a strong role in this market.
Works in Progress
These papers are in various stages of development and ordered by the likelihood of completion with an interesting result.
Encouraging Cooperation through Network Punishment: The Case of VC Follow-on Investments
Venture capital syndicates rely on a set of both formal and informal agreements for effective capital allocation and monitoring decisions. One important informal rule determines whether syndicate members should continue to invest in subsequent financing rounds. I study if and how venture capitalists punish deviations from this rule. Preliminary results suggest that investors who fail to follow past investments exhibit below-average growth in deals post-deviation. I hypothesize that punishment through restricted deal flow is used by both syndicate members and their network connections. Discovery of such relationships would show how information about deviations spreads through social networks in order to enforce informal rules.
Social Capital in Venture Capital Syndicates
I use person-level data to test how social networks in early-stage venture capital syndicates impact performance. The structure of the database allows me to control for important VC firm fixed effects and thus reduce endogeneity problems. The model posits that the internal network of the syndicate and its “external reach” are directly correlated with important outcomes like IPOs and bankruptcies.
Disentangling Performance Persistence in VC Fund Returns with Partner-level Performance Measures