Sunday 25 January 2015

Davos diversity

This year, as is my custom, I did not go to Davos. I do not have any animus against the place itself, but I have never been drawn to the conference known by that name, nor have I been invited. For the avoidance of doubt, I do not object to rubbing shoulders with the Great and Good. Thompson, with a p. I assume I would enjoy it, on the basis of that famous protest banner held above a baying mob: “They promised us there would be biscuits”.

One of the things I missed by failing to attend, breathlessly reported by the unfailing BBC, was a debate on the “diversity dividend”. This is the notion that diversity brings advantages, a broader range of outlooks which translates into real world gains in productivity. For example, including Keynesians and Hayeksians in central banks, centrists and localists in governments; and fascists and communists in academia. I suppose, if diversity is in fact the key to success, that there should also be gains to be made by hiring employees of very diverse levels of intellect, lest the too clever fail to notice insights more readily available to the dull.

In fact, Davos seems to regard diversity purely in terms of sex and race, particularly which sex you are and which sex you prefer, with preferring your own sex being an indicator of diversity. I do not regard these matters as being any of my damn business, unless a friend decides to confide in me about these matters. Privacy is not what it was, and the size, shape and above all the uses to which genitalia are put are now matters of common conversation.

The item begins with a strong evidential claims:

There is evidence, for instance, that boards of directors with greater diversity generate more dividends. That means greater returns to shareholders and fewer bonuses paid to managers.

With respect to gender, there are numerous studies that show that adding women to the labour force increases national output, or gross domestic product

The panel referred to evidence that makes the business case that a more diverse workplace boosted the bottom line. For instance, a Massachusetts Institute of Technology study has found that changing from all-male or all-female workforces to equal numbers of both sexes could raise revenues by around 40%.

No references are given. “Trust Aunty” seems to be the mantra.

About a year ago I corresponded with people who were making such claims and I was sent one bit of evidence, suggesting that in one period stock market gains had been higher for diverse as opposed to non-diverse companies, but I did not find this compelling. Proofs of this sort are always being proposed by companies and investment groups, and generally depend on choice of time periods and comparators. That is not to say that one could not do it properly, over a longer time period, but I haven’t seen any such studies.

However, I have found the MIT study to which the BBC was referring.

Diversity, Social Goods Provision, and Performance in the Firm
Ellison, Greenbaum, Mullin. April 2011

There are 4 mentions of profit, but no figures. That is to say, no data on whether the profits made by these individual offices (all members of the same company) differ according to diversity. There are 32 mentions of revenue and here is the key phrase which has been trumpeted by the BBC:

Turning to additional results in Table 5, higher levels of GendDiversity are positively
and significantly associated with office revenue in our base specification (1).
The estimated coefficient of 0.41 implies that going from an office that is either all
male or all female to an office split equally between the sexes would be associated
with a revenue gain of 41%(!).

On your behalf I bothered to look at Table 5, and the best predictor of office revenues, at a stunning 0.465, which they would no doubt describe as “boosting office revenues by 46.5% is…….. the number of years the office has been open!

Amusingly, the BBC report (see above) also makes the error of regarding revenues as “the bottom line”! No, dear reporter, revenues are the top line: the bottom line (profit) is what is left after all the costs have been removed, and this paper has nothing to say on that subject.

The authors of the paper then immediately go on to admit:

In column (4), we also report the results of a specification with office-level fixed
effects. When we control for the office fixed effects, the estimated contribution of
GendDiversity to office level revenue is no longer statistically significant.

So, the authors were right to put in their exclamation mark, knowing that the apparent effect was not a real one once local office variables were taken into account. The paper actually reports that diversity does not even boost revenues.

So, can anyone help me where the BBC cannot? Can you send me good studies showing that when the balance of a workforce is brought into line with the demographic composition of a nation it does better in terms of bottom line profits than when employees are selected solely on competence?

Finally, I may not totally follow capitalist ideology, but I had understood that the purpose of forming a company was to make profits by competing in a free market, honouring contracts drawn up freely between employer and employee, and between company and customers. Competition is the name of the game.

Suppose it was the case that having a sexually and racially diverse workforce led to larger profits, why on earth would any company admit this, and give away a useful business secret to competitors?

I hope you understand why I prefer the beach to Davos.


  1. I'd bet money that female labor force participation correlates at least moderately with IQ and other well-being variables when using the 50 states as the level of analysis. I also suspect black LFP would correlate, but in the inverse direction (and that there may even be suppression effects with these two variables predicting IQ).

    I'd be skeptical of causality, though. I predict the correlation would exist because almost any well-being variable-- especially those co-varying with liberalism/conservatism-- correlates with nearly all other variables.

    The only exceptions I've seen so far (i.e., variables that don't predict much) are state personality measures, state penis size (which correlates -.30 with IQ but nothing else), and the sex ratio.

    I'll see if I can get lfpr data now.

  2. "There is evidence, for instance, that boards of directors with greater diversity generate more dividends"

    I imagine it's the other way round - that very successful firms (who ipso facto become political players, with a Corporate and Government Affairs Department) will appoint tokens, to gain brownie points from the politicians.

    "With respect to gender, there are numerous studies that show that adding women to the labour force increases national output, or gross domestic product"

    It certainly does. A mother raising children at home adds nothing to GDP, because there's no explicit cash payment involved. But the same woman raising other people's kids at a nursery generates taxable revenue. With this in mind it's interesting that the ONS has started adding the (estimated) value of drugs and prostitution to our GDP. They must look wistfully at all the sex that ISN'T being paid for (not directly at any rate).

    "Suppose it was the case that having a sexually and racially diverse workforce led to larger profits, why on earth would any company admit this, and give away a useful business secret to competitors?"

    If The Powers That Be really believed this, there would be no laws against sex and race discrimination in employment, because all the sexist and racist employers would be outsold and out-competed by the diverse ones, and the laws would be unnecessary. My work involves frequent (at least annual) diversity training, and while the (assumed) corporate benefits of diversity are highlighted, "remember - it's the law" seems to be heavily emphasised as well.

    You'll note that while diversity is seen as a strength for the UK, it's not seen as such for places where the Evil Colonialists laid their heavy hands e.g. Syria after the Sykes-Picot agreement. Such places aren't "diverse", they are "divided".

  3. Good point about diversity not needing to be enforced by law if it was intrinsically more productive, because companies would want to do it to gain greater profits.

  4. If you'd actually read Hayek you would know that communists and fascists are essentially the same thing.