Not very useful data at all

This dropped into my email inbox today from Statistics New Zealand (full release here)

The gender pay gap was 9.2 percent in the June 2018 quarter, Stats NZ said today.

This is the second-smallest gap since the series began 20 years ago. In comparison, the gender pay gap was 9.1 percent in 2012 (the lowest on record) and 9.4 percent last year.

The gender pay gap shows the difference in median hourly earnings for men and women. For the second year in a row, the gap reflected that median hourly earnings for women, from wages and salaries, increased faster than for men, up 3.2 and 2.9 percent, respectively.

gender pay
“The gender pay gap is a useful measure when trying to understand differences in pay between men and women, due to its simplicity. But this measure is limited. It doesn’t account for men and women doing different jobs or working different hours. It also doesn’t account for personal characteristics that can influence pay, such as qualifications and age,” Mr Broughton said.

In other words, it is almost certainly of very little use at all.

There does seem to be an increasing trend for our supposedly independent statistics agency to be working in league with government departments championing particular causes.  This seems to be another example.  Thus on the SNZ email we read this advert

Renee Graham, Chief Executive of the Ministry for Women, discusses the gender pay on YouTube from 11am.

It is to be expected that the Ministry for Women will have a view on such data, and causes to champion.  But what is Statistics New Zealand doing promoting the views of an advocacy agency?  Doesn’t it risk compromising confidence in SNZ’s statistics?  No doubt the Governor of the Reserve Bank has a view on the CPI numbers, or indeed the GDP or employment numbers, but it would be a worry if SNZ were using their platforms to promote the Governor’s efforts to spin things his way.

But as it is, surely the chart above points to something quite important.  There is next to no difference in the average hourly wages of men and women in the under 24 age group, and even the gap in the 25-29 age group is probably barely statistically significant (look at how much the measured gaps in individual age categories can jump around from year to year). Not much sign of the much-vaunted conscious or unconscious discrimination.  It seems quite likely that the subsequent gaps mostly reflect (the lagged effect of) intra-family choices about childcare etc, which have no obvious policy implications.  In such a context neither a widening nor a narrowing of the gap is something to be welcomed; it just is.

But I guess that wouldn’t fit the desired political and bureaucratic narrative.

21 thoughts on “Not very useful data at all

  1. It seems reasonable for SNZ to advertise govt ministries that will or should use their statistics where relevent. I would expect them to similarly add references to INZ discussions of Immigration and RBNZ discussions. From the Ministry of Women website “The Ministry provides accessible, evidence-based information to inform and assist others.” Surely SNZ provides that evidence.

    More interesting is an argument I read recently that although median intelligence is the same for men and women the bell curve is a slightly different shape with more men mentally limited and more with exceptional gifts. I suppose much depends on what is meant by intelligence and nurture effects probably swamp comparisons however the theory is if you are looking for a chess master then you will naturally have more male candidates. Maybe the same for applies to some of the women nominees for appointment to state sector boards and committees?

    Historically men dominate classical composers, painters, jazz innovators, mathematicians, physicists but not literature (which is the most human of these endevours). Of course history has given women a raw deal but there is always a possibility that there really is a difference at the extremes of ability. Certain there is considerable prejudice against women in many fields but I honestly and possibly naively believe most serious mathematicians and jazz innovators are totally blind to anything other than their obsession. That there has only ever been one female Fields medalist is surprising and maybe significant. Note that this argument does not support any significant gender pay gap – here are very few paying jobs that demand extreme intelligence.


    • I disagree with the first paragraph. It is still very unusual (except for prominent political cause numbers) and should remain that way. SNZ should be there primarily as an independent collector and publisher of stats for all – not primarily for govt agencies. There are all sorts of users of all sorts of data and SNZ should not be preferencing the perspectives/interpretations of govt agencies.

      Re your final sentence yes I agree, and note that the data released yesterday are medians so should not be influenced by a few extreme observations at either end of the distribution.

      i’m tempted to dispute “history has given women a raw deal”, but will leave that for another (far distant) day.


      • Yes there is a study that shows that women have become less happy as their opportunities expand. The underlying idea of FOMO almost certainly applies to men too. But maybe best to leave it to Katherine Moody to dissect.

        With reference to the gender pay gap I wonder if they allow for prisoners. That is an example of a comparatively large number of extreme cases (dumb men incarcerated) that might affect the figures if they are left out. If you leave out 10,000 men earning effectively nothing it would make those who do work appear to be better paid on average.
        It must be difficult measuring a gender pay gap since we bring up girls and boys differently so they have different asperations. My daughter chose to work as a receptionist and my son as a builder – fairly typical but how do you compare the work effort, risk, talents required before measuring hourly pay rate? It is not an issue that will go away soon since I note my 19 month old grand daughter who has the innate personality of a lion tamer or rugby forward has parents who dress her in pink and gave her a pink pony rocking horse. We start brain washing our chldren very early.


      • Interesting point re prisoners. They won’t be included. That said, even in NZ, prisoners are well under 1% of the employed population.

        Recall that pink used to be a boy colour and blue a girl one only a hundred years ago.

        On doubt there is an element of dressing to culture (after all, at 19 mths parent determine clothes) but I’m not sure any such “brainwashing” lasts much longer in most families.


      • Bob’s points regarding “brainwashing” (I’d call it gender roles) are worth thinking about.

        There is a heck of a lot of it, and it begins early in life. I got sick of reading kids books with male protagonists, and started seeking out books with more girls in them. And I have had to give a hard stare or two to local parents who have let slip messages about boys being more interested in math and alike.

        That’s not to say that there aren’t differences between young boys and girls, on average. But we certainly should not send messages or reinforce stereotypes that ultimately constrain what a young person can do with their life.


      • Prisoners – however I cook the figures I can’t make it more than a 1% effect on gender pay gap (probably about half that).

        Gender roles: somewhere on the internet is a site with a mother and her daughter and a library of the 400 most popular young childrens books. First they remove all the books with no female character, then the books where the female character never speaks, then the books where the female character is a princess and the bookcase becomes progressively emptier. Apparently the first year at school is critical. At the begining of the year you read an exciting book where the lead character is not identifiably male or female and then ask the child “do you think that character was a boy or a girl?” and at that stage they are as likely to say girl as boy; then at the end of a year’s schooling you repeat and there is a preponderance for ‘boy’. So society ends up with women taking what are considered women’s roles: such as prime minister as a matriarchal leader using platitudes to control the giant family of all Kiwis. But it needs a man to be Governor of the Reserve Bank who can do dashing things such as change the value of the money in your pocket with a single slash of his pen.

        This subject might depress me if it wasn’t for the exceptions: Mary Lou Williams and Maryam Mirzakhani matched any man in their chosen fields.


  2. Hi Michael:

    You appear to be in two minds about whether the pay gap is a useful measure of (workplace) discrimination.

    On the one hand, you (rightly IMO) point out that it does not control for personal/career choices, etc. But then you say:

    >Not much sign of the much-vaunted conscious or unconscious discrimination.

    If the wage gap is a poor measure of discrimination, then it is a poor measure regardless of how big or small it is. In other words, a wage gap of zero would not be evidence that there is no discrimination going on (in some average sense).


    • If we think about business being primarily motivated by profit then the reason for the pay gap becomes clearer. Generally women are more passive in the work place and men are more aggressive. That is pretty much inbred into their psyche.

      The intent is to keep the best people you have irrespective of whether they are men or women. The NZ workplace since I have been in NZ now for 30 years working in many different companies does not discriminate. It is totally wrong in my personal experience that Gender bias is even an issue. But to keep men you have to pay. They have a tendency to ask for it or they would outperform to get it. Even if they do not ask for a pay increase they would behave as if they are on the market and could leave at any time. It is this on edge performance by men that puts the employer under pressure to pay more to men. Women have so many self esteem issues to deal with they get behind in their day to day knowledge or even to build a sufficient depth of knowledge to feel confident about leaving.

      A business being profit motivated would pay the least amount to get the maximum profit. It is attitude that makes the difference between who gets paid how much.


  3. Fair point, although my implicit (but unstated) assumption is that the personal choices that materially affect the medians later in life don’t cut in (for most families) until after age 24.


    • Thanks Andrea. I hadn’t previously read it (only reports of it). From a look at the Exec Summary it isn’t clear that they do control for time out of the workforce, or self-selecting to a less demanding job (perhaps they do, but i couldn’t see it, and can’t see how they would get such data).

      Also interesting that in only about a third of industries is there a statistically significant difference at all.

      Jim Rose here summarises Claudia Goldin’s work in this area, amd story still rings true to me.

      If there were genuine discrimination going on, on any widespread scale, I’d have expected it to show up among the young, who typically have the least effective bargaining power, the least knowledge of the labour market etc. But on the SNZ numbers such a gap doesn’t show up.


      • Hi Michael:

        Here again you are using the gap as a (good) measure of discrimination. We should always control (or try to control) for things like sector, education, etc. We agree on that.

        That the wage gap is zero for young people tells us very little about workplace discrimination. It could be, for example, that women have to attain higher levels of education in order to be paid the same as less-educated men, on average. (Women have been attending tertiary education institutions in greater numbers than men since the 1970s/80s or so.)

        I am not saying that this is what is going on. Far from it. It could be the opposite. What I am saying is: these simple averages tell us nothing much about discrimination.


      • IIRC the recent paper by Gail Pacheco, Isabelle Sin (and one more person – apologies) shows that mums that spend a year or more out of the workforce to look after kids earn 8.5% less (on average) ten years later compared to mums that stay out less than three months (or perhaps women that do not leave at all – I do not remember). The latter is not an ideal control group, since it could be that well-paid mums choose to re-enter quickly (seems likely). But 8.5% is a lot in the NZ context, IMO.

        You’d be interested in that paper, Michael. It gels with what Goldin is saying about the US.


      • I went to a LEANZ talk by her earlier this year. She pretty much controls for all of those issues and answered all her challengers. From what I can tell it is super orthodox econometrics too.


      • I don’t think Sin, Stillman and Fabling (2017) directly control for time out of the workforce; rather, they lean on modeling assumptions to estimate individual level productivity, and make inferences from there. Or such is my understanding from a brief reading. With the IDI data you should be able to directly observe individuals dropping in and out of employment due to parenthood (as in Sin, Dasgupta and Pacheco, 2018).


  4. This seems like fomenting discord down among the lower echelon of fomentables.

    As we watched the passing parade of the Russell McVeagh shenanigans followed by the #metoo movement we learned that more than 50% of all practitioners before the bar are female – they are intelligent, up and comers who have the power to congregate and form their own power structures and firms. The only one I am aware of that has done that is Mai Chen of Chen Palmer


    If you watch the nightly news and all the calamitous government goings on guess who is sent out to cop the flak – females – In a lot of cases it is self inflicted, Clare Curran, Tracy Martin, Meteria Turei, Crusher Collins, Jan Logie, Marama Davidson, Eugenie Sage


  5. This is how SNZ describe their recommended measure on their web site:

    “If we want to understand the fairness of pay (do males and females get equal pay for equal work?) the hourly pay measure is the best. It allows us to compare male and female pay for a fixed amount of work (one hour).

    “In an ideal world, we would also match males and females on characteristics that influence pay, and see if there is any remaining difference. For example, we expect occupation and qualifications to affect pay. So we would compare the difference in pay for males and females within the same occupations, and holding the same qualifications.

    “However, we don’t do this analysis because it isn’t possible to control for all factors that influence pay (and we don’t measure all factors). We are also limited by our surveys’ sample size.”

    This seems to suggest, as you say, that the data they are releasing is pretty useless for the one purpose it is presumably supposed to be used for: identifying systemic discrimination against women. Given the inherent measurement difficulties and the resulting paucity of hard data to back up claims of discrimination it appears unwise for a supposedly impartial official statistical agency to be siding so strongly with another department that is clearly not impartial. And what is more SNZ is now advising other government departments how to measure their so-called gaps.
    They appear to assume (ideologically) that conscious and unconscious bias is an unmeasured factor in the gender pay gap measure.

    There are also studies that find the pay gap statistically disappears when relevant factors are accounted for,e.g. The BBC had a number of independent studies done that suggested the pay gap doesn’t exist when legitimate explanatory factors are accounted for, and yet they capitulated to the concept anyway after complaints from some female reporters that male reporters (with more experience and doing more hours) were being higher paid for the same job. I believe Audi claimed they have no gender pay gap for similar reasons.

    My worry is that the diversity and inclusion maelstrom that is riding through the public sector (and the private sector too) that facts and meaningful statistics could be ignored while at the same time all but useless statistics are used to justify an ideological view that results in discriminatory pay and hiring policies when the underlying assumptions should at the very least be recognized as debatable (if we are still allowed to debate such things!).


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s