Reading the Evidence: What Syndicate Diversity Really Predicts
The empirical question here is how confidently we can interpret a reported relationship between venture stage and VC syndicate resource diversity. The study follows financing rounds and uses measured proxies for prior industry experience, lead-VC breadth, and co-investment ties; it does not observe a randomized partner choice or the quality of advice delivered. Reading the evidence well means understanding what the comparison actually is, what the estimates show, and how much remains uncertain before turning a composition pattern into a business recommendation.
What the design compares
The study’s unit of analysis is a financing-round syndicate around a venture. That choice matters because the same venture can appear in several rounds, allowing the researchers to examine how its syndicate composition changes over time rather than comparing entirely different companies once each. The data cover US ventures receiving VC funding from 1990 through 2019, with earlier investment activity used to construct rolling measures of experience.
The starting data contain 7,650 VC firms, 57,434 new ventures, and 287,122 investments. The final analytical sample contains 32,805 round-level observations from 13,191 ventures and 3,340 VC firms. Reaching that sample required excluding undisclosed entities, several nontraditional VC categories, buyout and other deal stages, ventures with only one funding round, rounds with a single investor, and observations missing the histories or attributes needed for the measures. The exclusions make the longitudinal comparison possible, but they also define its population.
The retained deals are not a random slice of all venture finance. The article reports that they are better documented and more likely to involve younger ventures, larger syndicates, or more established investors. Older deals, smaller investors, and less professionally documented transactions are more likely to lack the information needed for analysis. The result is a study of documented, repeatedly financed syndicates, not a census of every startup that sought money or every investor relationship that ended after one round.
The key variables are proxies with specific meanings. Venture stage comes from VentureXpert’s startup/seed, early-stage, expansion, and later-stage categories, coded from 1 to 4. Syndicate resource diversity is calculated from the members’ prior five-year investment experience across 69 VEIC 3-digit industry categories using an entropy measure. A wider and more balanced spread of prior industry experience produces a higher diversity value. The measure describes the breadth of the group’s accumulated knowledge and social capital; it does not directly record whether a particular investor gave useful advice, activated a customer relationship, or helped resolve a problem.
Lead-VC diversification is constructed with the same general logic, using the lead’s prior investment experience across industries. The lead is identified primarily as the investor with the largest cumulative investment in the company for the round, with network centrality used to break equal-investment ties. Co-investment experience captures the average number of prior co-investments among members within a five-year rolling window. It is intended to represent familiarity among partners, not to measure trust or communication quality directly.
The estimation design then asks whether diversity changes with stage within a venture, after accounting for other factors and fixed effects. Venture fixed effects absorb characteristics that do not change over the observed rounds, such as persistent aspects of the venture’s identity.