What does the “cents on the dollar” pay gap actually measure?

Each version of the figure divides women’s typical (median) earnings by men’s, for a defined group of workers. Which earnings, which workers and which year all change the answer.

In Canada, Statistics Canada reports that in 2024 the median hourly wage of women aged 25 to 54 working full-time was 90% of men’s, up from 75% in 1981. The ratio differs by age: 96 cents for every dollar among 25 to 34 year olds, and 83 cents among 45 to 54 year olds.

In the United States, the Census Bureau’s annual figure for full-time, year-round workers was 82.7% in 2023 and 80.9% in 2024. The Bureau of Labor Statistics’ weekly figure for full-time wage and salary workers was 83.6% in 2023.

The figure is often quoted as if women are paid less than men for doing the same job. That is a different question. The raw figure compares all women with all men in the group, whatever their occupation, industry, hours, experience or job level.

Unequal pay for the same work is a separate thing, and it is unlawful in both countries.

Supported: there is a difference between women’s and men’s median earnings in both countries, and it has narrowed substantially since the early 1980s.

Not established by the raw figure: that the difference is caused by employers paying women less for identical work, or that it is caused entirely by choices. Answering either needs studies that compare similar workers.

Those studies find that measured factors explain part of the gap and leave part unexplained. Statistics Canada found that differences in industry, occupation and part-time work accounted for some of the 2018 hourly gap, and that nearly two-thirds of it was not explained by the factors in its model. In the US, Blau and Kahn found that after accounting for occupation, industry, education, experience and other measured factors, 38% of the 2010 gap remained unexplained.

An unexplained gap is not the same as discrimination. It includes the effect of anything the model could not measure, such as detailed work experience or job demands, which could make it overstate discrimination. It can also understate it, if discrimination shapes the factors being controlled for, such as occupation. Blau and Kahn discuss both possibilities.

The Canadian figure above uses hourly wages for workers aged 25 to 54. The US figures use weekly or annual earnings for all ages. Weekly and annual measures are affected by differences in hours worked even among full-time workers, so they tend to show a larger gap than hourly measures. Comparing 90 cents in Canada with 81 cents in the US mostly compares two different measures.

Full-time is defined differently: 30 or more hours a week in Statistics Canada’s figures, and 35 or more in the BLS and Census figures. Full-time workers do not all work the same hours, and weekly or annual earnings reflect that. BLS reports that in 2023 women earned 87% as much as men among full-time workers with a 40-hour workweek, compared with 83.6% across all full-time workers.

The figures cover employees, not the self-employed, and medians, not averages. Neither shows how gaps differ within the same occupation or employer.

Year-to-year changes can reflect who is in work, not only what people are paid. The fall in the US annual ratio in 2023 and 2024 needs that kind of analysis before drawing conclusions.