Most studies on gender wage gaps look at labor market outcomes through the lens of participation rates. A common finding is that participation rates for women lag participation rates for men, sometimes by large amounts. The implication is that economies are both leaving money on the table and leaving potential workers unfulfilled by underutilizing women. Here we take a different and complementary approach using sectoral labor force data. Specially, we look at:

• The composition of employment across sectors (the number of men and women workers in each sector of the economy), and
• The output per worker across sectors broken down between men and women (the GDP for each sector from the national accounts, divided by the number of workers in that sector).

Based on 2019 data for the U.S., Canada, and Mexico, we find relatively large implicit gender wage gaps of 10%-20% using our approach. Due to data limitations, we use average output per worker as a wage proxy, which could vary significantly from actual wages. (We looked at 2019 pre-pandemic data, to avoid the large swings in the data during and after the pandemic.) We conclude that the participation rate approach alone is unlikely to paint a complete picture and that we also need a sectoral lens in any gender inequality analysis.

As with all national accounts data, we are measuring the market value of the goods and services that firms and workers produce. We are not making any inferences on the societal value of the output of workers using our dataset. For example, we cannot compare the societal value of a worker in the mining sector versus a worker in the education sector. The societal value may differ, perhaps substantially, from the market value.

In our view, addressing the labor supply imbalance across sectors--by tackling biases or structural impediments that prevent both women and men from reaching their full potential--would be key to reducing the gender wage gap.

U.S.: Manufacturing Is The Largest Contributor To The Implied Wage Gap

Looking at employment shares and output per worker by sector, we find that the average labor productivity (implied wage) gap--across the whole economy--between men and women workers in the U.S. was about $50,000 in 2019, or around 21% (see chart 1). For both populations, we take a weighted average labor productivity across all sectors. That is, we weight the output per worker in each sector by the share of men and women workers in that sector.

Given the wide differences in employment shares between men and women across sectors combined with the even wider differences in the output per worker across sectors, our result is not too surprising.

Chart 1

In most sectors in the U.S., the gap between the men and women labor force shares is quite small (see chart 2). Ten of 16 sectors have a differential of less than 2%. However, in construction and manufacturing, the share of employees who are men is much higher, and in health care and education, the share of workers who are women is much higher. These differences are the drivers of our main findings. Among the major sectors of employment, retail trade stands out with almost identical labor force shares for men and women--serving as a sectoral equality benchmark.

Chart 2

Like the employment shares, the output per worker across sectors is also highly uneven for men and women. (Output is measured as the value of total production of goods and/or services in a given sector.) The output per worker is highest in the mining sector and lowest in the education sector (see chart 3). And the gap is huge: the former is 42x greater than the latter.

This result is intuitive, at least in terms of the ordering. In the capital-intensive sectors such as mining, there is a relatively large amount of capital (cumulative net investment) per worker, which raises output per worker. Conversely, the education and leisure sectors are less capital intensive, implying a lower amount of market-valued output per worker.

Chart 3

We do not assume any difference in output per worker between men and women within each sector. This is more for analytical simplicity rather than as a depiction of reality. A more complete picture would adjust the sectoral data for seniority as well as specific job types. As a result of this assumption, any difference in the economy-wide output per worker will come from the distribution of men and women across sectors, not from any differences between men and women within sectors.

A way to visualize how the sectoral composition of employment in the U.S. generates this large implied gap is to plot the data as a waterfall (see chart 4), showing the sectors contributing positively and negatively to the gap. The numbers for the individual sectors shown in chart 4 are the combination of the employment share gap between men and women in that sector and the average output per worker in that sector.

Chart 4

The largest contributor to the U.S. implied gap in 2019 is the manufacturing sector, at $27,000. This is twice the contribution of the runner-up construction sector. While not the sector with the largest men-women gap or the largest output per worker, manufacturing was relatively large on both measures. On the other end of the spectrum, health care had the largest employment share gap of any sector and a relatively small output per worker, leading to an $18,000 negative contribution. This more than doubled the negative contribution of financial activities.

As noted above, retail trade can be used as an equality benchmark since the shares of men and women working in that sector are almost identical. Not surprisingly, retail trade has a contribution of zero in our waterfall chart. Our economy-wide gap is therefore independent of the average productivity in the retail sector. That is, whether a sector has high or low output per worker, if the share of men and women is equal, then its contribution to the gap will be zero.

If the share of employment in each sector were equal for women and women (that is, if all of the men-women dots lined up perfectly in chart 1), there would be no waterfall since the gap would be zero in each sector and, therefore, zero for the economy as a whole.

Canada: A Stark Gender Divide Between Employment In Capital-Intensive And Labor-Intensive Industries

In Canada, the gender gap in the sector-weighted average of output per worker was over C$23,000 in 2019 (see chart 5). This is smaller than the U.S. using both market and purchasing power parity exchange rates, but a better comparison would be to look at the percentage difference between men and women. Here the gap is 22.5%, larger than in the U.S.

Chart 5

We derive a sectoral labor productivity (wage proxy) gap for Canada along the lines just illustrated for the U.S. We begin with the sectoral employment breakdown in 2019. Most sectors show a little gender disparity in their employment shares (see chart 6). However, specific sectors have relatively large imbalances. Construction and manufacturing are skewed toward men, where the disparity was 12% and 8%, respectively. Conversely, in health care, women predominate with a share differential of 19%.

As in the U.S., there is a stark divide between employment in capital-intensive industries, mostly dominated by men, and labor-intensive fields such as health care and education, where women make up the largest share. In contrast to the U.S., administration and information are the sectoral equality benchmarks, with gaps of less than 0.5%.

Chart 6

In terms of output per worker, or productivity levels, across these sectors (see chart 7), average labor productivity in Canada varies considerably, as in the U.S. The ratio of average labor productivity between the highest (mining) and lowest (leisure) is almost 16--smaller than the U.S. but still sizable. The ordering generally aligns with the U.S. and broadly lines up with capital intensity. But we would caution against direct comparison across countries because the definition of sectors across economies may not be completely aligned.

Chart 7

The sectoral contributions to Canada's aggregate per-worker output gap are more one-sided than in the U.S., particularly regarding the positive contributors to the gap (see chart 8). Construction, mining, and manufacturing all have an outsize impact. In contrast, one sector has a material negative contribution: health care. As expected from our equality benchmarks, the contributions of public administration and other services are effectively zero.

Chart 8

Mexico: The Smallest Implied Wage Gap In North America, But With The Lowest Productivity

When looking at the breakdown of per-worker labor productivity for Mexico, we initially see the familiar gap between men and women (see chart 9). But, Mexico's percentage gap in average labor productivity between men and women is significantly lower than the U.S. and Canada. In 2019 U.S. dollar terms, the difference is roughly $1,500, and in percentage terms it's about 10%. By comparison, the gaps in the U.S. and Canada are 21%-22%.

Chart 9

Before discussing the components of our implied wage inequality measure, a few observations are in order, considering that Mexico is the sole emerging market in our group:

• First, Mexico has significantly lower average output per worker than the U.S. and Canada. This is true in both levels and growth terms. Therefore, the starting size of the productivity pie is smaller, which makes the potential gains of reallocating labor across sectors also smaller in absolute terms. Also, average labor productivity growth in Mexico is 0.6%, about half of the U.S.
• Second, Mexico's economy has high levels of informal employment. According to most estimates, this is over 50% of the labor force, keeping productivity low, and introducing data coverage issues.

We again start with the sectoral composition of labor (see chart 10). In Mexico--similar to the U.S. and Canada--men have a significantly higher share of labor force participation in key, relatively capital-intensive sectors: agriculture, construction, and transport/warehousing. Similarly, women have higher participation than men in most of the services-related sectors.

Importantly, manufacturing is the equality benchmark sector in Mexico, with a gap of less than 1% between the labor force shares of men and women. This reflects the prominence of factories along the border with the U.S., which were boosted by NAFTA (and continued under the USMCA trade agreement).

Chart 10

Turning to output per worker, Mexico's ordering is broadly consistent with the U.S. and Canada (see chart 11). However, the levels of labor productivity are much lower, reflecting its lower per capita GDP. Another unique feature of Mexican per-worker output is the dominance of mining and utilities, which is 8x higher than the second-most productive sector: transport and warehousing. Consistent with the U.S. and Canada, labor participation among men in Mexico skews toward the higher average productivity sectors, while labor force participation among women skews toward the lower average productivity sectors.

Chart 11

The construction sector explains most of Mexico's men-women output per-worker gap (see chart 12). Although, noticeably, even sectors with small differences in men versus women participation, such as the mining sector, account for a large part of the output per-worker gap. This is due to the outsize level of capital intensity and, therefore, high average output per worker in that sector. Conversely, retail and wholesale trade, and social services help narrow the output per sector gap, although both of these sectors have output per sector close to the average for the whole economy.

Chart 12

North America Comparison: Significant Improvement In Mexico, But More Work To Be Done

We also compared economies--in terms of percentage deviations between men and women--and looked at progress (or lack thereof) over time: 2007 (before the global financial crisis), 2019 (before the COVID-19 pandemic), and 2022 (the most recent observation). Because the sectoral breakdowns are slightly different across our sample of countries, a direct sector-to-sector comparison is not possible.

The most striking result is the outperformance of Mexico (see chart 13). First, the size of the output per worker (implied wage gap) is smaller in Mexico than in the U.S. and Canada in percentage terms. And the margin is not small. In the most recent observation in 2022, the gap in Mexico is less than one half of the others: 9.9% versus 20.6% in the U.S. and 22.4% in Canada.

But again, the high levels of labor informality in Mexico introduce data issues, which reduce the robustness of our conclusion. That said, the gap in Mexico has fallen by about one-third from 2007, while the gaps in the U.S. and Canada have been flat, at best.

Chart 13

Our tentative explanation for Mexico's outperformance is the structure of its economy. As noted, manufacturing is the sectoral equality benchmark in Mexico, meaning its shares of employment among men and women in that sector are roughly equal. Moreover, manufacturing has a relatively large and fast-growing share of output. According to the latest data, manufacturing constitutes 19% of GDP, compared with 12% in the U.S. and 10% in Canada. However, this explanation would need to be backed up by future research.

To Equality And Beyond?

Progress in reducing the wage gap requires addressing the labor supply imbalance across sectors. Unfortunately, defining the ideal state is difficult. There is no reason for this allocation to be equal across all sectors, even if eliminating all biases and structural impediments. The preferences of men and women to work in various sectors may be different. Moreover, there is no reason that this unbiased allocation would favor either gender. We could end up with sectoral equality, or men could retain their current overrepresentation in the higher-productivity sectors, or women could become overrepresented in the higher-productivity sectors.

Rather than have equality as a goal, a more fruitful approach would be to attack any biases or structural impediments that prevent both women and men from reaching their full potential. This could take several forms. Institutionalized notions that girls and boys should have different career paths are ripe for change. More proactive options include encouraging and supporting young women to go into fields that are historically heavily skewed toward men. For example, "Girls Who Code" aspires to build a large pipeline of women engineers.

Finally, narrowing the labor force participation gap between men and women will not necessarily reduce the wage gap (see chart 14). This will only happen if the distribution of labor across sectors improves as a result of higher participation.

Chart 14