Different to Begin With: The Resources Behind the Premium

The Portuguese study finds a performance gap between portfolio serial-entrepreneur firms and other firms, but a gap is a symptom, not a diagnosis. This chapter asks what happens to that gap when researchers account for the people who own the firms, their financial position, and the workers they employ. The result helps separate observable starting advantages from possible learning, while also showing why a decomposition of an association cannot be treated as proof that any one resource caused the premium.

What the decomposition is really measuring

A comparison between portfolio serial-entrepreneur firms and other firms can establish that their average outcomes differ. It cannot, by itself, explain why. The portfolio firms may be larger because their owners have more education or managerial experience. They may grow faster because they face fewer financial constraints. They may survive more often because they employ a different workforce. Or all of these differences may be expressions of another condition that the data does not observe directly.

The researchers approach this problem by estimating the portfolio-firm premium first and then adding groups of explanatory variables. The outcome can be firm size, exit, growth, or labor productivity. The initial estimate is the observed or unconditional gap after the study’s baseline controls. A later estimate conditions on entrepreneur, workforce, or financial characteristics. The reduction between the two estimates is then decomposed into contributions from individual factors using the Gelbach method.

“Contribution” needs careful translation. If adding education and past wages makes the gap smaller, the observed difference in those variables accounts for part of the gap under the model. It does not mean that education caused a fixed percentage of the firm’s success. The variables may be proxies for other resources, may have been selected by the owner’s earlier history, or may be correlated with an omitted factor. The decomposition tells us which measured differences are compatible with the premium; it does not turn an observational comparison into an experiment.

This distinction is especially important for the title’s contrast between learning and being different at the beginning. The decomposition mostly identifies differences that are already visible in the data when the firm is observed. It cannot always tell whether those differences were present before the first business, acquired through earlier employment, or produced by an earlier venture. A past wage is used as a proxy for ability, for example, but it can also reflect occupation, opportunity, family circumstances, or outside options. The variable is informative without being a transparent window into innate talent.

The same caution applies to education. Education can supply knowledge, analytical tools, credentials, and access to networks. It can also signal selection into environments where opportunities and capital are easier to reach. If portfolio entrepreneurs are more educated on average, the result supports a human-capital or selection explanation for part of their firms’ performance. It does not imply that a degree mechanically produces a larger, safer, or more productive business, nor that owners with less formal education lack the relevant capability.

The Portuguese results nonetheless provide a meaningful answer to the starting question. Entrepreneur characteristics alone account for about 20% to almost 50% of the estimated portfolio-firm premia, depending on the outcome. The largest contributors are education, past wages, and prior managerial experience. Whether owners are also classified as managers helps explain lower exit rates, but it is not quantitatively important for most of the other reported premia. A substantial part of the difference is therefore associated with who the owners are and what they had done before, while a substantial remainder is not explained by those measured characteristics.