Glass Ceiling Index
- Introduced by
- David Cotter, Joan Hermsen, Seth Ovadia & Reeve Vanneman
- Year
- 2001
- Type
- Distributional gender-gap criterion / index
- Family
- Labour-market segregation indices
What does it do?
The glass ceiling index and related distributional measures quantify the 'glass ceiling' — the tendency for gender disadvantage to intensify toward the top of a wage distribution or organisational hierarchy.
Cotter and colleagues (2001) set out formal criteria distinguishing a true ceiling from a general gap, while labour economists operationalise it as a widening female–male gap at high quantiles of earnings, and popular indices (such as The Economist's) rank countries by women's representation in senior roles, pay, and leadership.
The intuition behind the method
A glass ceiling is an invisible barrier that lets women rise only so far. Statistically, this means the gender gap is small near the bottom and middle of the ladder but grows sharply at the top: women are well represented among workers but scarce among executives, and the pay gap is widest among high earners. The key analytic move is to look not at the average gap but at how the gap changes as you move up the distribution — a gap that accelerates toward the top is the signature of a ceiling, as opposed to a uniform disadvantage at every level.
History
The metaphor of a glass ceiling entered popular use in the 1980s and a US federal Glass Ceiling Commission reported in the 1990s. David Cotter, Joan Hermsen, Seth Ovadia, and Reeve Vanneman gave it analytic precision in their 2001 Social Forces article, specifying criteria to distinguish a ceiling from a general gap. In parallel, labour economists — Albrecht, Björklund, and Vroman for Sweden and Arulampalam, Booth, and Bryan across Europe — operationalised it through the gender gap at high wage quantiles, and media indices such as The Economist's glass-ceiling index turned the concept into an annual country ranking.
When to use it
Use glass-ceiling measurement when the question is whether gender disadvantage is concentrated at the top — in executive representation or among high earners — rather than spread evenly. Distributional (quantile) methods are appropriate with individual wage data; representation-based indices suit comparisons of leadership across firms or countries. The approach is less informative when disadvantage is uniform or worst at the bottom (a sticky floor), and the Cotter criteria warn against labelling any high-level gap a ceiling without ruling out compositional explanations.
Limitations and considerations
- Distributional analysis requires individual-level data and methods beyond ordinary mean comparisons.
- Quantile gaps describe where the gap is largest but do not by themselves identify its cause.
- Composite leadership indices depend on the choice and weighting of indicators, which vary across publishers.
- A widening gap at the top can reflect selection into top jobs rather than a barrier within them.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Distinguishes top-concentrated disadvantage from uniform gaps, capturing a distinct and policy-relevant phenomenon.
- Cotter et al.'s criteria give a disciplined definition that prevents loose use of the metaphor.
- Quantile-based operationalisation reveals where in the distribution the gap arises, complementing mean decompositions.
- Representation indices provide intuitive cross-national and cross-firm rankings of women's access to leadership.
Applications
- Testing whether the gender wage gap widens at the top of the earnings distribution within a country.
- Distinguishing glass-ceiling from sticky-floor patterns in comparative labour-market studies.
- Ranking countries or firms on women's access to senior management and board positions.
- Evaluating whether promotion and advancement chances deteriorate for women at higher organisational levels.
- Informing corporate diversity policy and board-quota debates with representation and pay evidence.
How it works
How it works
Cotter and colleagues propose four criteria a pattern must meet to count as a glass ceiling rather than ordinary inequality: the disadvantage must (1) not be explained by other job-relevant characteristics, (2) be greater at higher levels of the outcome, (3) apply to the chances of advancement, not just current standing, and (4) increase over the course of a career. These criteria discipline the term so that not every gender gap is labelled a ceiling.
Empirically, the gender gap is measured across the distribution rather than at the mean. At each quantile τ, the difference between the male and female outcome (for example log wages) is computed, tracing how the gap varies from low to high earners. This requires distributional methods such as quantile regression rather than ordinary least squares, which would report only the average.
Formula
\Delta(\tau) = Q^{M}_{\tau} - Q^{F}_{\tau}A glass ceiling shows up as a gap that rises as τ approaches the top of the distribution — the female–male difference is larger at the 90th percentile than at the median. Studies such as Albrecht and colleagues' analysis of Sweden and Arulampalam and colleagues' European comparison test exactly this: an accelerating gap at high quantiles is read as evidence of a ceiling, while a gap that is flat or larger at the bottom indicates a 'sticky floor' instead.
Formula
\dfrac{\partial \Delta(\tau)}{\partial \tau} > 0 \;\text{near}\; \tau \to 1In organisational and cross-national settings, glass-ceiling indices instead use representation at successive hierarchy levels. A typical index compares women's share among senior managers or executives to their share in the workforce (and combines this with pay and promotion ratios). The Economist's glass-ceiling index, for example, aggregates indicators such as the gender pay gap, women's share of senior and board positions, and labour-force participation into a single country ranking.
Formula
GCI = \dfrac{w^{F}/w^{M}}{\text{share}^{F}_{\text{senior}}/\text{share}^{F}_{\text{workforce}}}Common pitfalls
- Calling any gender gap among high earners a glass ceiling without checking it grows toward the top and is not compositional.
- Confusing a glass ceiling (worst at the top) with a sticky floor (worst at the bottom); the two have opposite distributional signatures.
- Comparing composite glass-ceiling rankings across publishers that use different indicators and weights.
- Ignoring sample selection into senior positions, which can mimic or mask a ceiling in observed wages.
Frequently asked questions
How is a glass ceiling different from a sticky floor?
Both describe where the gender gap is concentrated in the distribution, but in opposite places. A glass ceiling means the gap is largest at the top — women fall furthest behind among high earners and in senior roles. A sticky floor means the gap is largest at the bottom — women are disproportionately stuck in the lowest-paid positions. Quantile analysis distinguishes them: the gap accelerates toward the top under a ceiling and toward the bottom under a floor.
Why use quantile regression rather than a mean decomposition?
A mean decomposition like Oaxaca-Blinder reports the average gender gap and cannot reveal whether that gap is uniform or concentrated at one end of the distribution. Quantile regression estimates the gap at different points — the 10th, 50th, 90th percentiles — so it can detect a gap that widens at the top, which is the defining feature of a glass ceiling. Distributional decompositions such as recentered influence function methods extend this to attribute the quantile gaps to characteristics and returns.
Are the country 'glass-ceiling index' rankings the same as the academic measure?
No. Media indices, such as The Economist's, are composite rankings that combine indicators like the pay gap, women's share of senior and board roles, and labour participation into one number per country. The academic glass-ceiling concept is narrower: it specifies criteria (Cotter et al.) and is typically tested as an accelerating gap at high wage quantiles. They share a theme but differ in construction, so they should not be conflated.
Related methods
Sources
Cotter, D. A., Hermsen, J. M., Ovadia, S., & Vanneman, R. (2001). The glass ceiling effect. Social Forces, 80(2), 655–681.
Open DOI referenceCopied.Arulampalam, W., Booth, A. L., & Bryan, M. L. (2007). Is there a glass ceiling over Europe? Exploring the gender pay gap across the wage distribution. ILR Review, 60(2), 163–186.
Open DOI referenceCopied.Albrecht, J., Björklund, A., & Vroman, S. (2003). Is there a glass ceiling in Sweden? Journal of Labor Economics, 21(1), 145–177.
Open DOI referenceCopied.