Skip to main content

20 November 2020

Discrimination lies behind the gender pay gap – and we need to investigate it further

Almudena Sevilla

Economists are collecting hard evidence that discriminatory behaviour is widespread

Gender pay gap

Today is Equal Pay Day in the UK – the point in the year at which women, on average, stop earning relative to men. The Fawcett Society predicts that, at the current rate of progress, it will take around 60 years to close today’s gender pay gap of 14.6%. This slow pace is mirrored by the incremental nature of the research aiming to understand pay disparities.


After half a century of research, we are yet to find a single factor that can finally explain why women earn less then men. Although part of the gender wage gap can be explained by men and women having different characteristics (such as working in different occupations and industries), another part cannot be explained by these factors – that’s down to discrimination.


Discrimination can be overt prejudice against women – so-called “taste-based discrimination” – or result from employers lacking information. Because employers do not know enough about a potential applicant or existing employees, they use the group’s characteristics to infer the applicant’s qualities. Women are more likely to stop working when they have children, so an employer assigns a lower probability of job commitment to all women (regardless of whether a particular woman wants to have kids or would stop working after having kids).


Taste-based discrimination calls for interventions to address the structures that perpetuate discriminatory behavior in the first place, or to address it directly through legislation such as the 1970 UK Equal Pay Act. The information problem – or “statistical discrimination” – calls for policies to challenge the blanket characterisation of whole groups of workers.


Whereas the theoretical underpinnings and possible policy solutions of discrimination have been well-understood for some time, data limitations have made it difficult to distinguish which type of discrimination is at play. As a result, economic research and policy carried on under the (more politically correct, yet unrealistic) assumption that taste-based discrimination did not exist.


New ways of collecting, storing, and analysing rich data from images, text, and other forms of media have enabled economists to produce hard quantitative evidence that systematic discriminatory behavior is ubiquitous. For example, clever analysis of online video content reveals that female economists are asked more questions – and more questions that are deemed to be unfair – than male economists. Another study, of words used anonymously by economists in an online discussion forum, found that the set of words most associated with female subjects are “hot, hotter, attractive, gorgeous, pregnant, tits, lesbian, bang, horny”, while for men the top words were “homo, motivated, keen, slides, textbook, Nordic”.


This type of new evidence has spurred a wave of measures aimed at tackling discrimination. Explicit regulation and punishment of overt prejudice remain hard to implement as it is very costly for an employee to make a discrimination case. Recent pay transparency legislation constitutes one of the efforts to lower the “transaction costs” of implementing these earlier policies. Alternatively, lighter-touch interventions, such as statements about a company’s culture and accepted behaviour, send a strong signal about what will be tolerated and accepted (the Royal Economic Society, for instance, approved its code of conduct in December 2019).


While we are still unable to fully explain why women earn less than men, the fact that economists are finally taking taste-based discrimination seriously may bring us closer to an answer.

Almudena Sevilla is a Professor in Economics and Public Policy at UCL. This piece is based on her forthcoming article, “Gender economics: an assessment”, which will be published in the Oxford Review of Economic Policy, vol. 36 no. 4a.