While the discussion of changes in financial accounting properties over time is already well-established, there is a lack of evidence whether changing firm compositions in empirical samples might bias cost stickiness research. We document that with each additional listing cohort, the U.S. public firm universe becomes more knowledge-intensive and, at the same time, more cost sticky. Higher reliance on temporary labor by newer listing cohorts partly mitigates this development. Our results call for the use of listing cohort-specific slopes to allow for cohort-specific estimates of cost stickiness in future research.