NBER Working Paper 35265, issued in May 2026 and circulated in the newsletter received on 1 June 2026, revisits the employer size wage effect. Using linked German data on firms, establishments and employees, the researchers find that the usual relationship between establishment size and wages disappears across establishments within the same large firm after controlling for the firm.

The result distinguishes two comparisons that can otherwise be confused: employees working at different firms and employees working at differently sized establishments of one firm. It concerns wages in the studied setting. It is not an assessment of employee benefits, nor a finding that establishment size never matters in any country or employer.

Ask what the benchmark compares For remuneration teams, the study supports a closer look at the structure of a comparison group. This editor recommends documenting whether a benchmark groups establishments, legal employers or wider corporate organisations. A relationship visible across employers may not describe the differences within a particular organisation.

The same care applies when teams interpret local market information. A large employer may operate many sites with different functions and workforce compositions. A broad size category can hide those differences. The benchmark should help the team understand the relevant comparison rather than imply that one observable characteristic explains the whole pay outcome.

Keep pay evidence distinct from benefits evidence An employer reviewing total reward should identify which parts of its package are actually measured. A wage study does not establish the value employees place on healthcare, retirement provision or time off. Nor does it show that benefits explain a wage difference. Those questions require their own information and a clear account of the population concerned.

Internal review can combine a documented job structure with external comparisons and employee feedback. The purpose is to understand a proposed decision, including the limits of the available evidence. A change should not be justified by a general relationship if the comparison is poorly matched to the organisation’s circumstances.

The working paper adds a useful qualification to a familiar labour-market pattern. Employers can use that qualification to ask better questions about remuneration data and the units being compared. A defensible benchmark makes its boundaries explicit and leaves room for the organisation’s own employment practices to explain differences that a simple size measure cannot.

Sources: Source de référence