Wednesday, 9 May 2018

Mind the gap

We've been lucky, over the past year, to see two top-notch, data-heavy analyses of the gender pay gap appear in New Zealand. One, from March last year, is by AUT's Gail Pacheco, 'Empirical evidence of the gender pay gap in New Zealand', and was done for the Ministry for Women. The other, also last year, was by Motu Research's Isabelle Sin, Steven Stillman and Richard Fabling, 'What drives the gender wage gap? Examining the roles of sorting, productivity differences, and discrimination', and saw the light of day thanks to the Marsden Fund (and which I posted about in 'Competition is good for women's pay').

Gail reprised her research last night at the latest Auckland Law and Economics Association of New Zealand seminar, and if you're interested in the topic and you're based in Wellington, get along to Isabelle Sin's LEANZ seminar on Tuesday 22nd at 5.30pm (free registration here).

As it happens, and though they used different lines of attack, they both came out with much the same headline result, a gap against women of around 12-12.7% (Gail) or 16% (Isabelle et al). Gail used a battery of factors that might explain the gap - for example the industries people work in (if women were concentrated in low paying sectors, there might not be any gender difference in pay rates at all, even though there was a headline gap), or their levels of education. Here's how the battery of factors helped explain what is going on.

As you add more and more factors that might explain pay gaps, the amount of the gap that you can 'explain' goes up, but at the end of the day it remains remarkably small. When you try to adjust for the observable differences between men and women that might account for the overall 12.7% gap, you end up explaining only 2.1%, and not knowing where the other 10.6% came from. We're not unusual in that, by the way: as Gail mentions in her paper, there are sizeable pay gaps and sizeable proportions unexplained in many countries overseas, too.

It's the not knowing that Gail regards as significant. If you knew exactly what was going on, it's possible you mightn't be worried. Perhaps there's some extra factor, like the cost of potential interruptions to women's career paths to have children, that employers might be putting into the equation (no correspondence please, that's just a guess, not an endorsement). And if you were still worried (kinda reminded me of the old L V Martin line, "it's the putting right that counts"), you wouldn't know the right policies to fix it until you knew exactly how the problem arises. Motu, though, were less reticent about what's at play: "we blame sexism", as they put it in their executive summary haiku.

We may well get a better understanding from new work that Gail and Isabelle have in the pipeline, which looks at things like the impact of pregnancy career breaks, and which Gail said will be coming out at the end of this month.

It's possible, of course, that a bigger or different or finer set of explanatory variables might throw more light on the unexplained proportion. Gail suggested last night that degree subjects might matter, rather than just having a degree of any kind, and that's highly plausible. And there's also evidence that the more exactly you define comparable jobs, the smaller the pay gap gets.

Here, for example, are the key results of a very interesting global survey carried out in 2016 by Korn Ferry Hay, the recruitment people, based on a huge (8.7 million jobs) data base, which has info on exactly how 'big' each job is.

"Aha!", you might think, "there's no discrimination! Measure the nature of the job right, and gaps disappear!".  In New Zealand's case, a headline gap of 22.8% becomes almost nothing (0.9%) if the comparisons between the jobs held by men and women are as precise as possible.

But of course this raises a whole new issue: as Korn Ferry Hay put it, it's true that "men and women doing the same job, in the same function and company, get paid almost exactly the same". But "That’s because they still aren’t getting to the highest-paying jobs, functions, and industries, while men thrive in all three", and the question shifts to why they get (say) the Head of HR role rather than CIO or CFO or COO.

In any event, another excellent thought-provoking LEANZ seminar. Well done to Gail, to Richard Meade who organises the Auckland ones, to Andreas Heuser for organising the upcoming Wellington equivalent, and especial thanks to Bell Gully for hosting and sluicing both events. Without corporate support, they wouldn't happen. Though members' subs help too: here's the LEANZ subscription page.


  1. The gender pay gap rounds down to 0 for those at the 10th and 50th percentiles. Those in the top 10% have a pay gap of around about 20%. That suggests to me they are mixing career and family quite well. Having it all the top end of town

  2. Hi Jim - in the space I had I couldn't cover everything Gail found, but yes, the gaps widen as you go up the income distribution. There was a wee bit of discussion at the seminar about work/life balance but I suspect some clearer answers will emerge from the work that Gail and Isabelle have been doing since. I'm really looking forward to it

    1. Thanks, I think the data sources are just plain obsolete. Claudia Golding and others have shown that when you have detailed demographic data not only in the female worker but also the partner, and the number and spacing of the children, the gender gap falls away.