Madness behind the method in racial pay gap stats

Yesterday the front pages were filled with the news that the Prime Minister was going force all companies with more than 250 employees to disclose how many of each census ethnicity they have, and what different ethnicities get paid at the median.

This builds on the gender pay gap stats that Kate Andrews at the Institute of Economic Affairs has so roundly rebutted before, but companies nonetheless have to report. A totally useless set of statistics used to bash businesses and one which does not solve the issues it purports to.

Leaving aside the fact that there is nothing to force employees to disclose their race to their employer, or that there is something weirdly sinister the government demand companies do so, I’m convinced this crude measure will add yet another unnecessary burden to business and do nothing but sow division without reason.

As Tim Worstall, a Senior Fellow here at the Adam Smith Institute, said in a letter to the Times today: these figures are “misleading because the age structure of the population differs by formally defined race. From the 2011 census, the whole population median age was 39, that of the white population 41, Asian, black and other, 30, 30 and 29 respectively, and mixed 18. Pay rises with age, as promotions to better-paid positions are earned through experience.”

To put it bluntly, in the United Kingdom we’ve seen a large change demographic change in the past 50 to 60 years. In the UK, you’re like to earn the most between the ages of 40 to 54. If the white population’s median age is at that point and at peak earning potential at present, while other groups are younger and so behind in peak income, then any figure that does not control for this but is used to highlight a supposed ‘gap’ will be misleading at best, or otherwise downright dishonest.

But all this is bizarrely arbitrary anyway. Are we going to be so crude as to use ‘White’ as a broad brush signifier? Or will we split it into White British, White Irish and ‘Other White’ as we do at the census? If that’s the case, will it show up any racism or prejudices that exist towards various ethnic groups from the continent? Which one do British people of Irish origin choose?

Do we really expect media to split out all their headlines to explain the minutiae of British Indians pay being higher than the median pay for white Britons or lower than those from an East Asian background? Or will it be lumped together as ‘BAME’ and ‘White’?

I think we all know the answer to that: it’s the latter. That’s not journalists fault per se, columns are only a certain number of words long and headlines have to catch your attention. But it will be cheap and it will be nasty. And ultimately crude comparisons will not bring down barriers, tackle ongoing injustices in our society, or even free up time for HR departments to tackle actual issues of explicit or tacit discrimination in their companies.

Let’s take a dive into this proposal:

These figures won’t reflect reality

How the figures are presented matters. Usually when this kind of survey analysis is done, it gets split up into regions too. We’ll see some very odd results, not least because ethnicity gap statistics, as proposed, will not control for country of birth or fluency of language.

Even on a micro level, a firm may have a large manufacturing base in a part of the country with low diversity, but a small office in diverse London with higher than average wages for that company. Their pay gap will say nothing of whether there is discrimination within their London office, between the offices, or if there is a racial gap between their company and any others in either region.

But let’s say they don’t and said firm only has a base in one region. Let’s say for the sake of argument that region is Lincolnshire.

Lincolnshire has 92.9% born in the UK, EU 27 citizens make up 4.7% of the population with 3% being from the EU accession states of the past two decades. Just 2.4% of the population come from a BAME background. The BAME group are more educated and qualified than the average citizen of the area, are more likely than those from a white immigrant background to be full-time employed and more likely to be first language English.

It’s likely that the racial pay gap in the region will be low, and lower than a city like London, but would we really say that Lincolnshire is a better place to be someone from an ethnic minority than our capital city? It might be, but reality as shown by the real experiment of where people choose to live, says otherwise.

In fact, that will play out across the country. EU migrants have a higher tendency to be in low skill employment. Combined with the average age profile of UK migrants in recent years as being very young, there is a skew if we lump together ‘white’ workers of all backgrounds.

Skill level of citizen groups in work (UK, EU, non-EU)

Skill level of citizen groups in work (UK, EU, non-EU)

Screen Shot 2018-10-12 at 17.48.50.png

There are natural barriers to progression and regulatory barriers

Where you get people who don’t speak the language of the majority, they don’t tend to perform as well in labour markets. One study suggested that fluency in English in the UK increases employment probabilities by about 22% and proficiency in English is associated with 18-20% higher earnings.

Occupational licensing is perhaps one of the biggest barriers that people from all backgrounds face. As far back as 2013 the number of jobs in the UK ‘requiring’ a degree overtook those without. It hasn’t changed since. And even those without this requirement specify licenses from governmental or non-governmental agencies.

Take a security guard for example. Most contracting jobs and most agency work requires individuals to have undertaken a Security Industry Authority licence from the Home Office. This is required for any work that takes place outdoors, and is often demanded by unions for any work taking place in-house to protect their members’ interests. Oddly enough the £220 licence and up to £400 training cost acts as a barrier to entry, especially for those unemployed or without ready savings. There’s little to suggest that on the job training or agency training couldn’t get people up to speed with security work, and that the spend on licence is really impactful on what is being guarded.

Take another. If you take an entry-level job at an insurer in the city, say doing secretarial or copy work, you may find later down the line that you’ll understand the business pretty well. You might want to move over into doing base level and then more complex but now understood underwriting or account management. These companies often require internal testing (which some charge employees for) and many require a qualification from the Chartered Insurance Institute. Without it many will watch young graduates go past them, and this has, for quite some time meant those from a white and middle class background. But will that last?

What matters most is integration of second generation and beyond

This I think is key. In the UK we’ve prided ourselves on bringing down institutional and legal barriers over woke measures and positive discrimination. Where ethnic groups are found in productive cities or are personally mobile to move for work, their access to higher salaries and the world of employment more generally is close to the average worker, and often higher.

Within the UK, it is again the return of class and IQ that has the biggest impact on avoidance of unemployment. When in employment, salariat requires being geographically mobile and there are cultural norms that discriminate. Where they are not mobile, there is an increase of their chances of being unemployed and having a lower salary.

Screen Shot 2018-10-12 at 17.50.03.png

They also, importantly, outperform their first-generation parents. This matters enormously. While in 1979 only 0.3% of the working age population was composed of British born ethnic minority individuals, this rose to 2.9% by 2009. The test for discrimination and success of integration within the educational system will be if these people perform worse than their white peers.

Knowing that university is used as a signifier for employers, is correlated to higher earnings, and that ethnic minority students now outperform their white British peers, we should expect them to outperform earnings-wise in future years.

Prejudice has a price

As legal and non-legal barriers to getting a full education and qualifications have fallen we’ve seen more people from minority backgrounds applying for opportunities. As legal penalties have come in to protect applicants from minority backgrounds from being discriminated against in employment processes, we’ve seen an increase in labour market participation and pay.

Quite frankly, if you’re a company that chooses to discriminate on the colour of skin or cultural background of an applicant, you are costing yourself talent. In an open market companies compete on profit. If they’re refusing talent based on an arbitrary factor, they will be making themselves uncompetitive. Those that do so will harm their productivity and will fall behind those without such inclinations. There’s some evidence to suggest this is already happening.

It is talent that matters though, which education and qualifications can help or hinder to signal, and which language barriers hinder being used to full potential. When the IZA investigated diverse workforces they found no correlation between diversity of ethnicity and firm performance, but they did for educational diversity.

Crude measures lead to crude outcomes

You might have noticed in this article, but this policy announcement raises more questions that it can possibly seek to answer. And each one is complex and, as the lingo goes, intersectional.

Crude race stats mask class differences between North and South and both of these mask gender differences. Ultimately, we also have to ask why we stop at race or gender. We know that height and mature facial features are highly correlated to earnings, blondes earn more tips than brunettes, why not force companies to measure those? We know there’s a link between lifelong earnings and obesity, why not track whether those who are fat are being held back?

Or rather, let’s not force companies to do any of these things. If a company wants to, and sees the chance to develop a competitive edge by doing so, then let them. Maybe though, all the intersections actually ends up boiling down to individuals.

And maybe, just maybe, it is companies that are based placed to know their own flaws and how to tackle any prejudices which hit their bottom line, better than any government ever could.