PhD Seminar Series: Pietro Morinio and Claudia Frosi

Using Machine Learning to Estimate the Effect of General Partners on Venture Capital Performance

Speaker: Pietro Morino (Bocconi University)

Abstract: Despite extensive research on performance persistence in the venture capital (VC) industry, evidence supporting the common assumption that general partners drive such persistence remains limited. In this paper, we employ machine learning methods to quantify the effect (if any) of general partners on performance persistence across multiple VC funds. Analyzing 29,021 quarterly observations from 722 funds managed by 811 general partners between 1997 and 2022, we document statistically significant, albeit modest effects of general partners on performance persistence. These magnitudes are substantially smaller than those found in the literature, highlighting the limited external validity of traditional methods. We also find that general partner effects consistently exceed VC firm effects, suggesting that individual-level analyses provide greater insights than firm-level ones. Broadly, our results suggest that most of the variation in VC performance is not attributable to the organizational characteristics of VC firms and can be explained only partially by the individuals managing them.

Selection in Strategy Formulation: The Impact of Domain Salience

Speaker: Claudia Frosi (Bocconi University)

Abstract: Does increasing the salience of a knowledge domain affect how entrepreneurs formulate strategies and how their ventures perform? While theory suggests that salience shapes strategic reasoning, experimental evidence on whether making a latent knowledge domain salient influences strategy formulation and performance is lacking. We address this question through a randomized controlled trial embedded in a startup training program across 3 European countries (N ≈ 450). Treated entrepreneurs are exposed to one domain - i.e., environmental sustainability - via examples of startups that integrate sustainability into their business models, while control entrepreneurs are exposed to comparable startup examples, with the exclusion of such elements. Using human-coded and NLP-based classifications of startups’ theories of value (over 12 months), we test whether induced salience increases incorporation of sustainability into theories, prompts domain exploration among previously non-sustainability-oriented entrepreneurs, and differentially affects those with lower initial confidence. We also examine whether start-up performance follows a hypothesized U-shaped trajectory over time.