Where Stability Matters Most—and Where the Evidence Stops

The average effect of losing a founding-team member is not the whole story. This chapter asks which member leaves, how much human capital they bring, how concentrated the team is, and whether the result changes across industries or research settings. Those distinctions support a more useful judgment than “teams should stay together” or “founders are irreplaceable.” They show where continuity appears especially valuable, where redundancy can change vulnerability, and where the evidence is not strong enough to choose between retaining a person and adding a new capability.

Importance is graded, not binary

The U.S. Census results first distinguish key founding-team members from non-key members. The KFT category is an earnings-based approximation of likely key positions: the three highest earners in a corporation, or the owner and two highest-earning workers in a sole proprietorship. It is not a verified list of founders. That qualification matters because the category is useful for testing relative importance, but it should not be turned into a claim about who truly founded every firm.

The loss of a KFT member has a much larger adverse effect than the loss of a non-KFT member. In the reported inverse-hyperbolic-sine examples, the effect of losing a non-KFT member is still negative and statistically significant: about −0.08 for employment, −0.16 for revenue, and −0.09 for the revenue-minus-employment productivity measure. The additional effect associated with losing a KFT member is large enough that the total impact is nearly three times as severe in the examples. The pattern is consistent with concentration of leadership, relationships, or decision authority among key members.

But the non-key result is just as important conceptually. If non-KFT departures had no measurable effect, a founder-centered explanation could carry most of the burden. Instead, the evidence indicates that members outside the likely top group still participate in the productive system of the young firm. They may carry less authority or less prior earnings, but they can still hold customer knowledge, process knowledge, or complementary relationships. The team matters as a graded structure: roles differ, yet contribution is not limited to the most visible people.

Relative human capital produces a similar pattern. The study measures the departing member’s prior earnings relative to the average prior earnings of the active founding team immediately before the death shock. Losing a member at the team’s average level still produces large reductions in employment, revenue, and productivity. Losing someone with higher relative human capital produces a larger decline. The paper reports that a member with 25 log points more relative human capital is associated with an additional revenue reduction of about 0.18 in the relevant inverse-hyperbolic-sine estimate, for a total effect around −0.48, and an additional productivity reduction of about 0.12, for a total around −0.31.

These estimates do not mean that prior earnings are the value of a person or that a lower-paid member is disposable. Prior earnings are a proxy for human capital, not a direct measure of trust, compatibility, or team-specific knowledge. Nor do the results establish that a team should retain its highest-paid member regardless of fit.