The Ties That Make Departure Costly

This chapter moves from the startup-level founder shock to the employees inside it. The question is why longer experience with a founder can make an employee more likely to leave after that founder disappears, even though long tenure usually predicts staying. A second comparison—whether an employee shares the founder’s sex—offers a different, more cautious window into social similarity. Together, these patterns make founder–employee complementarities more concrete while showing why a proxy is not the same as a measured relationship.

The distinction matters for both explanation and action. If employees leave simply because they have spent more time at the firm, the remedy would be ordinary retention and succession planning. If they leave because their most productive skills, relationships, or sense of fit were specific to the founder, then the startup faces a more concentrated transition problem. The Census evidence supports the second possibility, but it does not tell us exactly which tie each employee had.

Shared tenure is not ordinary seniority

Tenure usually has a stabilizing meaning. Employees who have stayed longer often know the organization better, have invested in relationships, and have accumulated firm-specific skills. They may have fewer reasons to leave or higher costs of starting over elsewhere. In a comparison group that did not lose a founder, the Census analysis finds the familiar pattern: longer tenure is associated with a lower risk of turnover.

The result changes for employees whose startup experiences the founder-loss shock. The study uses an individual-level Cox hazard framework, which models how the risk of an employee leaving changes over time. The key term is an interaction between treatment status—being in a startup that loses a founder—and shared tenure with the founder. That interaction is positive and highly significant, at p < 0.01. Among treated employees, more years working with the founder predict a larger increase in the risk of leaving after the founder’s premature death.

The important comparison is not simply “senior employees leave more.” It is that tenure has different implications under different founder conditions. In the control group, longer tenure is protective. In the treated group, longer co-working experience with the founder makes the post-loss disruption more severe. The evidence remains similar after adding demographic controls, year effects, and state and industry controls. This contrast is what gives the result its interpretive force.

Imagine two employees who have worked at the same startup for several years. One has spent most of that time operating through a founder’s priorities, translating broad decisions into practical work, and using the founder’s relationships to solve problems. Another has worked in a part of the company with more independent routines and less direct interaction. Both have the same firm tenure, but the first has greater founder-specific exposure. If the founder leaves, the first employee may lose more than a familiar manager: they may lose the context that made their own role effective.