An Analysis of Industry Economic Performance in Search Funds
Daniel Lazier, Jacob Thomas, A. J. Wasserstein. (2026)
Yale Case, no number printed, March 13, 2026
Not an independent sample: "this note is effectively an addendum to the demographics note ... we are working from the same dataset, using the same methodology, and the same limitations still exist." (p1). Same 155 exited traditional search funds; cite together with the demographics note, where the inclusion criteria are stated. Headline numbers. Industry counts (figure, p14): software 44, business services 52, healthcare 23, consumer services 12, food and beverage 4, blue-collar services 4, education 3, unidentified 13. "healthcare has a higher median IRR at 39%, compared to 24% for non-healthcare." (p5). "Food and beverage has negative mean and median IRRs, -33% and -49%, respectively" (p7), on four observations. "Our coefficient of determination (R2) is 0.06, meaning only 6% of the dependent variable (IRR) is explained by industry segment. However, the joint F-test is not statistically significant (p = 0.14)" (p18). Healthcare coefficient 0.21, p 0.08; food and beverage -0.48, p 0.06 in the multiple regression. Limits. Authors: classification "involves a combination of art and science"; "Food and beverage and education have too few observations to draw conclusions" (p16); "we are not claiming that industry is irrelevant". Reader's notes: this note defines the top IRR decile as 15 observations while the demographics note used 12 on the same data; no demographic or geography controls; software is 28% of the sample, which matters for readers buying non-software businesses.