What is actually going on with the gender pay gap?
The most recent publication of gender pay gap data has generated a flurry of articles — as it should. Given that the metric being used (difference between the average earnings of men and women across an entire workforce) is a simple one, it reveals relatively little about the underlying dynamics. Moreover, only organisations with a headcount of 250 or more (representing around 50% of total UK employment) need to report this information. Therefore the mandatory requirement has, in my view, always been mostly about paying attention to the issue, not so much about what the data reveal.
I was therefore curious to look at another, more comprehensive and more granular, dataset: the annual survey of hours and earnings (ASHE). The downside is that the most up-to-date data is only for 2020, not 2021. But the upside is that the data is easily* available (in comparable format) from 2004 onwards. [*Easily may be an exaggeration, unless you are an academic with direct access to the data. For us mere mortals, it involves downloading the relevant zip file for each year, extracting the particular Excel files that one is interested in, extracting the specific sheets within the Excel files, and then combining and cleaning the data. Thank goodness for Python!]
While lots of people tend to look at the information by sector, I actually find it more instructive to look at occupations. Yes, certain occupations (e.g., nurses) are concentrated in certain sectors (e.g., health care), but most organisations employ a mixture of different types of people, from customer service and other front-line roles, to more technical occupations (say, accountants), to middle and senior management. It is at this occupational level that some of the more revealing differences in pay emerge. This time, I was also interested in whether the pay gaps differ by age cohort.
There seem to be some discontinuities in the dataset in 2011 (I suspect some occupations were reclassified), so I have focused on data since then. The left hand panel of the chart shows the change in the gender pay gap (the percentage amount by which women are paid less than men, in this case based on hourly wages) since 2011. Quite a few of the squares are either blue or beige, indicating that the gap between men’s and women’s pay has at least not increased for most age groups in most occupational groups. There are a few exceptions, but even these seem to be somewhat driven by some volatility in the data. Eyeballing a more detailed graph, the ones that would seem to be “signal” rather than “noise” are the increased pay gaps for young workers in caring and leisure occupations.
Despite those positive trends, though, men remain better paid in most age groups in most occupations*, as indicated by the preponderance of organge or red squares in the right hand chart. Here, too, the odd blue square seems to be mostly noise in the data, with the exception of customer service occupations where young and old (but not middle aged) women have been better slightly paid than men for a while.
In interpreting all of this it is important to remember that these numbers are not controlled for all the various reasons why pay is going to differ for individuals, and for women and men, within each group. Indeed, the occupational groups shown here are all relatively large: each includes between 300,000 to 2.7 million employees, so you would expect to see some significant variation inside each group. The kinds of things that typically mean some people are paid more and others less include skill levels, length of experience, length and flexibility of work week, and productivity.
According to one recent study, years experience in particular is a key driver of the gender pay gap. That would also explain why the right hand panel gets progressively more red or orange as you move to the right. The older the age cohort, the bigger the difference in accumulated years of experience between men and women. The main reason for this is of course the fact that women tend to take (more) time off to look after children or, even if they continue working, their total number of hours per week tends to be, on average, lower. [Note: that might be true for women even without children, given that household duties are are still very unequally distributed.]
It is such broader societal (and biological) factors that make eradicating gender pay gaps so challenging. But there must be a lot more scope to do so. For example, while I’m sure years of accumulated experience are important for productivity, economic theory would also suggest that someone’s first 3–4 days per week are much more productive than the last 1–2 days. So why do part-timers not get paid more rather than less per hour?
