Wednesday 6 September 2017

Competition is good for women's pay

Motu's recent paper 'What drives the gender gap', has rightly got a lot of attention: full marks to its authors Isabelle Sin, Steven Stillman and Richard Fabling. Motu has gathered a broad selection of the coverage here and if you haven't yet read the piece for yourself then here's a longish executive summary and the whole caboodle. And if you want the whole thing boiled down to 17 syllables, Motu's executive summary haiku said
Women are paid less,
but aren’t less valuable.
We blame sexism.
There's one aspect that hasn't caught much of the headlines, however, and that's the link between how competitive a marketplace a business is in, and the extent of gender discrimination it goes in for. In sum, the link is strong, and it means that if an industry is more down the monopoly end, women get treated even worse than usual.

As the paper reminds us (page 27),
Starting with [Nobel Prize winning economist Gary] Becker (1957), the argument has been made that taste discrimination [i.e. discrimination in the negative sense we use in everyday English] cannot persist in a perfectly competitive product market because firms that discriminate will lose money compared to those that do not and will be driven out of the market. This has led a number of papers to focus on the relationship between product market competition and discrimination.
The corollary to that however is that if markets aren't competitive, there aren't the same pressures on employers to make most efficient use of their staff, and can afford to pander to whatever prejudices they've got without taking much of a hit to the bottom line.

Does this happen in real life? When I was a financial journalist in Tokyo, one American banker told me that he had the pick of the Japanese labour market, because Japanese banks strongly preferred to hire men for the important jobs, leaving him a clear run at the best women graduates. Conversely I remember a Japanese banker proudly showing off his state of the art foreign exchange dealing room, and telling me that "Yes, we've got 23 people here - 17 dealers, and six women".

In New Zealand, Motu devised a measure of how competitive each industry is (if you're of the wonkish tendency, I'm about to add a technical footnote - here it is * - and the rest of us can now carry on). While they were at it, they also devised measures of how much skilled labour each industry uses, and how tight the labour market was for each industry at any point in time, which they needed to try to sort out different explanations for the gender wage gaps.

And with that out of the way, here's what they found (page 31):
There are a number of key findings. First, industry-years with a one standard deviation more skilled workforce have a gender wage-productivity gap that is 19.2 percentage points higher if they have the mean level of product market competition and difficulty hiring. Second, this gap is doubled if the industry-year is one standard deviation less competitive, or is eliminated if the industry-year is one standard deviation more competitive than average. Third, this additional effect of lower levels of competition is eliminated if the industry-year has a one standard deviation higher difficulty in hiring. Overall, we find that the gender wage-productivity gap is larger in industry-years with higher skilled workers, lower levels of product market competition, and more competitive hiring markets ['competitive' in this sentence means lots of people looking for jobs].
Let's unpack this a bit. Firms with an unusually high level of skilled workforce pay men a stonking 19.2% more than women for the same productivity contribution to the business. That's on the basis that the firm is in an industry that is about average for the level of competition going on in the sector, and also when the labour market in that sector at the time is nothing unusual. That's a whole story in itself.

But look again at that second finding. That already large pay difference is doubled - doubled! - if there's lots less competition among businesses in the sector. However the large difference goes away completely - to be consistent, completely! - if there's lots more business competition. It also goes away completely if a tight labour market is holding employers' feet to the fire and forcing them to make gender-blind hiring decisions, which is what you'd expect. We routinely see employers, for example, hiring more people from minority groups when there's been a sustained business cycle and hirers can no longer pick and choose the way they might have done.

There are people who don't like competition - the hand-wringing types who don't like the Schumpeterian real world and who'd prefer collaboration or cooperation. Get real, folks: if women want fairer pay, one highly effective approach would be to use markets to work for them. Insist on gales of competition in every industry (and, incidentally, support initiatives like the Commerce Commission being allowed to look at the competitive state of play). That way, there'll be fewer guys with cozy jobs in dozy industries ripping you off - because you'll have the real choice of going to his competitor and getting what you're worth.

* The measure of competition comes from a principal components analysis (love it as a technique) run over four measures of competition from the Business Operations Survey (eg firms reporting no competition, or only one or two competitors), plus a capital/labour ratio. Personally I can't see the relevance of the capital/labour ratio to competition or (excess) profitability - airlines for example might well have a high capital/labour ratio because of the planes but I'm not sure that tells me a lot about whether the airline game is competitive or hyperprofitable - but in any event their measure of competition (the first component) has stronger links with the competition measures than with the capital/labour ratio, so that's all right.


  1. Reading it out to my husband, a carpenter. He says, "Because there isn't an open pay policy, a company can hire a girl 'chippy' (carpnter) for 2-3 dollars less an hour and that's that." She doesn't know what he earns, he doesn't know what she earns, company can keep their bias. What's your opinion on that?

  2. Thanks for the comment, it's a good one.

    My response is that an employer of chippies can only get away with this if he holds all the cards - only he knows what the 'proper' rate for a chippy is, and he can undercut it because chippies don't know what they're really worth, and he faces no consequences if he stiffs a woman with a below-par offer.

    In a workably competitive market - where there are plenty of businesses looking for chippies and plenty of chippies looking for jobs - he's not going to get away with it. A chippy who's talked to a variety of potential employers will be able to throw away the lowball offer. And businesses who keep making lowball offers won't be able to hire anyone. So both sides of the market will end up only considering the fairer deals.

    You might well think, what if all the employers are in it together? And they think they can't be undercut, because they're all going to offer the 2-3 dollars less to women and there's nowhere else for women to turn to?

    That comes back to my Japanese example. There's now an incentive for a business to renege on the industry sexism, pay a decent rate, and get good people (women) that other firms have been shafting.

    Your husband would be right if every industry was an employer-friendly rort, no business competed hard with an other business, and all employers could rip off all employees. Maybe that's true of parts of the economy. But that's my point: if we can manage things so that less employers have that cushy life, there wouldn't be the opportunity to rip women off. The ratbags lose out because their competitors get better people.

    As you may gather from my posts, I'm very much in favour of more competition rather than less, and I've mostly been of that view from a consumer perspective. The nice thing, I thought, about the Motu paper was that it went beyond what you can buy in the shops and showed that more competition can mean better social as well as better material outcomes.

    Thanks for the feedback - happy to discuss further if you like.

    1. I'll have to think about it, and talk to my husband again. But another thing that evens out pay differences: open pay policy. In Norway, everyone's income is publicly available to everyone. That, supposedly, is one of the factors that helps even out the payscale.

    2. Added to clarify: in your example of competitiveness, you have to wait until the economic cycle gets competitive enough for women to get the advantage of lessened sexism; which also means that when economy eases, sexism can return again.

      With open pay policy, it works regardless of where economy's at: if there's sexism in hiring or pay decisions, it's out in the open, and therefore up for scrutiny. And as much as some industries are prone to sexism - most people don't like being *blamed* of being sexist.

    3. Re your point about sexism re-emerging in a slower economy, that's right. As the Motu paper said, there are two things going on at the same time. One is whether there is more competition between businesses (good for women) or less (bad). But the other is whether the labour market is short of people (good) or has lots of people looking for work (bad). So yes, you could get things turning worse during downturns in the labour market. One of the things I bang on about in this blog from time to time is that there are a lot of social costs to slow economic growth, and this is yet another of them.

      I don't know anything much about Norway's open pay policy, though I have a vague memory that anyone's income (and tax returns?) can be looked up at will. It's an interesting idea, though (and this is more a guess than anything) I doubt if there is much popular support for it at the moment in 'Anglo-Saxon' cultures like ours. I can see that it could be helpful along the lines you suggest of 'naming and shaming', but whether it's a practical political reality here - dunno.

  3. This comment has been removed by the author.

  4. Thanks Donal for a nice summary of our paper.

    1. You're welcome. It was well worth giving another extra nudge along. Though 'summary' is too charitable: to be honest, I only picked up on a small sub-set of the many conclusions (competition being one of my 'things'). I also found the other patterns the paper identified fascinating - maddening, too, of course, and still puzzling in many respects, but interesting throughout. The paper identified a whole new research agenda. And it's probably the most significant use to date of the LEED database.

      And it's neither here nor there really but I especially liked the principal components application. It's a technique I learned in my first job (at a research institute) but it went somewhat out of fashion since. It still makes occasional appearances (eg the RBNZ's 'LUCI' indicator of labour market tightness) but it deserves more outings.


Hi - sorry about the Captcha step for real people like yourself commenting, it's to baffle the bots