When a skilled job becomes vacant, the firm can hire externally or promote internally and backfill from the entry-level market. We extend the vacancy-chains framework of Elsby et al. (2025) to incorporate this endogenous port-of-entry choice. The model reveals two opposing forces: rank-varying hiring costs create congestion externalities that absorb most of the amplification in the original framework, while the promotion margin partially restores it by rerouting replacement demand away from congested skilled markets. Quantitatively, the model generates substantial amplification of labor market fluctuations, accounting for roughly a third of the empirical volatility of job-finding rates, even though over half of skilled placements occur internally
and are invisible to conventional vacancy measures. This measurement wedge is procyclical. The cyclical promotion share generates differential hiring volatility across job levels, providing a mechanism for slack entry-level markets during recessions. The endogenous composition of new hires also dampens measured new-hire wage procyclicality through a between-firm channel. Using resume and job-posting data, we document that approximately half of skilled openings are filled internally, with systematic heterogeneity across job types.

Using matched employer-employee data from Brazil, we study how firms’ occupational composition changes as they grow. We show that management and professional employment shares decline with within-firm growth. A broader class of occupations that we classify as overhead (occupation-by-industry cells with within-firm elasticities below one) exhibits the same pattern. Despite this within-firm decline, larger firms in the cross section employ disproportionately more overhead workers, and overhead share is positively associated with firm growth, survival, and proxies for productivity. These facts are consistent with a production function in which overhead labor is complementary to persistent firm type or organizational capital, but each unit of overhead supports more activity at larger scale, allowing firms to decrease overhead share as they grow. Consistent with this interpretation, overhead share measured at founding is more strongly associated with subsequent performance than later overhead share, while firms that increase their overhead share have worse outcomes. Exploiting a minimum-wage law that increased the relative cost of non-overhead labor, we find that exposed firms shrink and raise overhead share, but continue to move along the same within-firm scaling relationships estimated in the pre-period. Firms entering the event with higher overhead share contract less and are less likely to exit. These results are consistent with the view that the overhead share signals, rather than determines, productivity.