The glass ceiling, a metaphor turned measure

The phrase appears in the American business press in the late 1980s. Its rigorous formulation, proposed by David Cotter and colleagues in 2001, imposes four criteria, which turn a striking image into a testable hypothesis.

The most demanding is that the gap is not explained by observable characteristics and that it widens as one moves up the hierarchy. A constant gap at all levels is a different phenomenon: a general inequality, real, but with nothing specific to the top.

That requirement has a practical consequence for reading the figures. Observing that there are fewer women among executives is not enough to establish a ceiling, since the observation is compatible with several mechanisms located elsewhere along the career path.

Analyses of pay distribution find in several countries a gap widening at the top of the distribution, a result compatible with the criterion. The work of James Albrecht and colleagues on Sweden, published in 2003, is an often cited example.

Analyses of access to executive positions find substantial gaps, part of which is explained by the composition of the pools and by earlier trajectories, which makes verifying the criterion more delicate. The pool itself results from earlier decisions, so that controlling for it sometimes amounts to neutralising part of what one is looking for.

Longitudinal work on the effect of a birth, including that of Henrik Kleven, Camille Landais and Jakob Søgaard on Danish data published in 2019, finds a divergence of trajectories occurring early in the career and accumulating thereafter.

That mechanism produces a widening gap at the top of the hierarchy without requiring a barrier at that level: a gap formed early is enough to produce the appearance of a ceiling. The two explanations are not mutually exclusive, and the available data do not always allow the gap to be apportioned between them.

Several European countries have imposed minimum proportions on company boards, Norway as early as 2003, France since 2011. Evaluations find a direct and rapid effect on board composition.

Effects at lower management levels are much weaker in most evaluations, including that of Marianne Bertrand and colleagues on the Norwegian case, published in 2019. The result is often described as a ceiling displaced rather than removed, and supporters of quotas reply that their indirect effects require a longer observation period.

Speaking of a ceiling suggests a single obstacle located at the top. The data rather describe an accumulation of small gaps spread across the whole path, which other images, such as the sticky floor or the leaky pipeline, attempt to capture.

That difference in representation has practical consequences: it locates action at the moment of final promotion rather than at the earlier stages where the gap forms.