The Gender Pay Gap Myth – How controlling for factors that affect income eliminates the gender pay gap

The gender pay gap – a social phenomenom that the International Labour Organization (ILO)called “one of today’s greatest manifestations of social injustice”. There is a lot of confusion and misinformation about the gender pay gap. The average gender wage gap based on US data, was 79.3% – women earned on average 79.3% of what men earned. Looking across 70 countries the ILO found that on average the gender pay gap was 20% – meaning that women across the world on average earn 20% less than men. Now most people mistakenly think this means that women get paid 20% less than men for the same work – hence eNCA for example ran a story citing this statistic as proof that there is no equal pay for equal work currently and hence a lot of institutional sexism still present.

However the gender pay gap does not mean that men and women are not paid equally for equal work. You have to compare apples with apples, you have to compare men and women in same occupations, with same experience, education; job title and roles. You would not think it is unfair if someone with 20 years’ experience in a particular profession is paid more than someone with 5 years’ experience in the same profession. And yet the gender pay gap does exactly that – it compares men and women with vastly different work profiles and then finds a difference or a gap and mistakenly concludes an injustice is occurring.

Comparing apples with apples

Numerous studies have been conducted to compare men and women with very similar work profiles. These studies control for various factors that affect people’s incomes. Controlling for factors simply means adjusting the wage to gap to take into account differences in those factors which affect people’s incomes.

A study by Blau and Kahn, The Gender Wage Gap: Extent, Trends, and Explanations, found that once you control for (1) education, experience, race/ethnicity region and metropolitan area the pay gap drops slighty. But once you control for (2) the fact that men and women are not distributed equally in different industries and occupations the wage gap drops to 8.4%. (This means that occupational and industry differences explains almost half of the pay gap.)

To put it differently, it means that men and women are separated into different occupations and industries which compensate them differently. Men tend to be concentrated in occupations and industries with higher compensations. However even after controlling for those factors there was still a significant gap (8.4%).

Figure 1 – Change in wage gap over time and effect of adding controls to gender wage gap

Another study by Glassdoor went further and added more detailed controls such as company specific controls and job title controls (accounting for differences in companies and job titles) and when they did that the U.S. wage gap declines to 4.9% for base pay. That means 78% of the initial wage gap can be explained and it is not due to discrimination where women are paid less for the same work.

Figure 2 – Effect of adding controls to gender wage gap

So how then do we explain the remaining 4.6% wage gap? Are there other unknown factors or is it discrimination against women. I think the answer is likely to be due to differences in hours worked. If you look at studies that zoom into specific occupations and those that measure the hours that men and women work differences emerge which I think should explain the remaining pay gap.

Men tend to work longer hours

A Harvard study looked at bus and train operators in the Massachusetts Bay Transportation Authority (MBTA) and found that weekly earnings gap between men and women was due to the fact that: “…men take 48% fewer unpaid hours off and work 83% more overtime hours per year than women. The reason for these differences is not that men and women face different choice sets in this job. Rather, it is that women have greater demand for workplace flexibility and lower demand for overtime work hours than men. ” So the exact same job still had gender differences in pay because of differences in hours worked by men and women.

One particularly interesting study, Overwork and the Slow Convergence in the Gender Gap in Wages, found that there are gender differences in what they define as “overwork” (working for 50 or more hours per week). More men are willing to overwork; women are less likely to enter jobs requiring extremely long hours; women are also less likely to stay in these jobs. They also found that there has been an increase in earnings for overwork which further increases the wage gap. The type of occupations that are likely to require and reward overwork are the professional and managerial occupations.

Figure 3 – Proportion of men and women willing to “overwork” over time.

Figure 4 – Proportion of men and women will to “overwork” broken down into different occupation groups (professional, managerial, and other occupations)

Some occupations have high penalties for work interruptions and high nonlinear rewards for long hours

Blau and Kahn cite research that shows the role of work force interruptions in lowering female wages. There are disproportionate rewards for long working hours in certain occupations. In other words this means that difference in pay between working 40 hours and 50 hours can be significantly higher than 25% even though the hours worked differ by 25% – these are occupations which show non-linearity. Women place more value on flexible schedules than men. Different occupations and workplaces face different costs for providing flexibility. The wage penalty for flexibility will be high in occupations where there is “meeting deadlines (time pressure), being in contact with others to perform the job, maintaining and establishing interpersonal relationships, adhering to preset schedules, and doing work for which other workers are not close substitutes”.

One study found that male and female graduates from a law school who started with similar earnings had diverged after 15 years due to the greater willingness by women to work shorter hours and to have worked part time in past. Similar findings were found by a longitudinal study which tracked male and female graduates over 16 years found that men earned more due to differences in weekly hours worked and actual post-MBA work experience. These occupations showed nonlinearity; meaning the more hours you work, the higher your hourly rate.

Gender pay gap in STEM

When looking particularly at STEM fields – similar results have been found. In STEM fields the average gap by various researchers has been found to range from 14% to 28%. A number of studies cited by Kahn & Ginther (2017) including their own research demonstrates that once you control for a number of variables such as age, race, level of highest degree, marital status, presence of children, current job and work time, employment sector, years worked: the gap disappears and reverses for single women without children and declines significantly for married women with children.

Single women without children have a pay gap advantage of 5.4% and married women with children have a gender pay gap disadvantage of 12.4%. Kahn and Ginther (2017) also found that research by the US commerce department showed that the gender salary gap in STEM is lower than in non-STEM occupations. A surprising finding if we expected that STEM fields have higher levels of bias against women which leads to them exiting STEM fields.

In the American academic science and engineering fields Ceci, et al. (2014) cite studies that generally agree with the above finding, although there were cases that differed. In general studies found when you control for rank, type of research institution, field, productivity and then include the effect of marriage and children; the gender gap declines significantly. Women with children with engineering PhDs earned less than single women; a similar result is noted for biomedicine field.

Lubinski & Benbow (2006) conducted a study where they tracked for four decades the educational, occupational, life outcomes and preferences of a group of intellectually talented individuals who were in the top 1% of mathematical ability when they were 13 years old (Study of mathematically precocious youth –SMPY). They also found that controlling for hours worked per week amongst this group eliminated sex differences in income

One of the most dominant fallacies accepted today in society is the disparity fallacy – the claim that all disparities, differences in outcomes are the result of systemic bias and discrimination. A widely held but deeply mistaken idea. Society is far too complex for one single variable to be able to explain everything. The reality is far more nuanced and complex than that. The gender pay gap is a good illustration of that. There are too many factors, interacting in hidden and non-obvious ways to produce the gender pay gaps we see.

However, when researchers have tried to make apple to apple comparisons – the gender pay gap largely disappears. This should be good news and welcomed by everyone because it means then that the outcomes we see are not driven by systemic biases and discrimination against women. The research also highlights that there are career-family tradeoffs which are not going to be solved simply.