The best-laid schemes of mice and May: A response to Zamila Bunglawala

When Theresa May first entered Number 10 she announced her desire to tackle the“burning injustices” of British societyincluding racial and ethnic disparities. In particular, she pointed to such things as the ethnic minority employment gap and the outcomes for different ethnic groups in the criminal justice system. Her flagship race policy last year was the creation of the Race Disparity Unit, which provides statistics on social and economic outcomes by ethnicity, presented on the website Ethnicity Facts and Figures.

It is reasonable – indeed necessary – to ask why racial disparities continue to persist in our society and what can be done to address them. Few decent people want to live in a world where certain groups languish through no fault of their own. However, as I will argue, the Government’s approach to the issue of racial disparity has been simplistic and liable to backfire.

The muddled thinking of those behind the Unit is revealed in a recent article in The Times by Zamila Bunglawala, who is Deputy Director of Strategy and Insight at the Race Disparity Unit in the Cabinet Office. I quote in full three clauses in her piece:

[1] The audit does not explain the causes of ethnic disparities, and to what extent it is down to bias and discrimination. [2] That said, the website provides vital information to help us explore these questions, and informs policies and programmes to change outcomes… [3] Many may say that we all have biases, but when a bias hurts a person’s feelings, results in them being unjustly treated – or, as the statistics of the Race Disparity Audit show – results in them losing out on a job or promotion, it is clear that something needs to be done”.

She initially states that the statistics presented in the audit do not evidence causation (Clause 1). This is correct. What is presented on the Ethnicity Facts and Figures website are series of correlations; patterns of distributions of various social outcomes across ethnic groups. And correlation is not causation as every undergraduate in the social sciences should be able to tell you.

She then goes on to say that despite this, correlations can help you “explore” causation (Clause 2). To be charitable, the most that analysis of correlations can do is rule out causation (i.e. if there is no correlation then there cannot be causation) but this requires rigorous regression analysis which few beyond academia and some sections of the Civil Service have expertise in. Certainly, the website makes no stab at this.

Finally, she roundly contradicts her earlier statement when she concludes that the statistics do evidence discrimination with racial bias at the root (Clause 3, which contradicts Clause 1).

Nowhere in Bunglawala’s article is she even prepared to countenance the idea that differences between ethnic groups might, at least in part, stem from cultural differences or the historical shaping of collective preferences. Instead, discrimination is presented as the one concern, the queen of explanations. This narrow focus is underlined by the article’s emphasis on the author’s own personal experience of racial bias (e.g. “I have become familiar throughout my career for being mistaken for the PA”). But while such anecdotes can illuminate the legitimate frustrations felt by some ethnic minority individuals, they do not shed much light on the general extent of racial bias in organisations. Nor what can be done about it.Nor how we will know if and when we are successful in rooting it out. These questions are neither asked nor answered.

At the time of the Unit’s launch, the mantra was “explain or change”. If a public body or business produces statistics which show an ethnic disparity, they must either explain it or “correct” it in some way. But if you genuinely cannot explain something, how on earth do you go about changing it? Would you trust a plumber who could not explain why you had no hot water to fix the problem? Complex social causes which may be behind so many statistical disparities between groups including class, geography, age, and culture – are reduced to one simplistic explanation: the prejudices and biases of those in charge.  

There is a hardening official consensus around the causes of differences between groups and how we as a society are supposed to deal with them. It is this which underlies the Government’s diktat for all companies to publish their pay according togender (soon to be extended to include ethnicity), the Lammy Review into minority representation in the criminal justice system, and ultimately, the Race Disparity Unit. There has been very little serious Parliamentary scrutiny or challenge of these policies. It is assumed across the political spectrum that publishing statistics on distributions across groups will be both a successful diagnosis of a problem as well as a cure.

But not only does this approach fail to produce a reliable diagnosis; it can actively do harmWhat if the “nudge” to action, which is intended to shame people into compliancecompels them to rush into making mistakes? What if there are unintended consequences we have not begun to think about? Indeed, recent research from the United States has argued that most conventional corporate diversity and inclusion schemes tend to backfire – those businesses that deployed such schemes actually became less ethnically varied.

The Civil Service’s own workforce strategy exemplifies some of the problems I have described. Bunglawala says that she is “one of about a 5 per cent ethnic minority [of] senior civil servants”. This number refers to the share of ethnic minority individuals who are not white, working in the Senior Civil Service (SCS) and is used in diversity monitoring.

There is another figure, on the Ethnicity Facts and Figures website of 8 per cent and that refers to those at Senior Civil Service level, composed of the SCS plus some health professionals, military personnel, and senior diplomats. The fact that it is higher is revealing as it will largely be down to successful minority medical professionals in the NHS.

The fact of career clustering by choice and based on different preferences is the one thing that the diversity and inclusion specialists have nothing to say about. Why is 5 per cent bad when 32 per cent of British-born NHS consultants are from a non-white ethnic minority (as found in a 2016 Policy Exchange report called Bittersweet Success? Glass Ceilings for Britain’s Ethnic Minorities at the Top of Business and the Professions, which I co-authored)?

Bunglawala writes that the Civil Service aims for its senior level to be 13.2 per cent ethnic minority by 2025. This is a misstatement of Civil Service policy. Actually, the aim is to have the share of new recruits to the senior level at 13.2 per cent non-white ethnic minority by 2025. The Civil Service estimates that hitting this target would increase representation to 10 per cent although they are someway off at 5.6 per cent of new recruits currently.

It is not clear why 13.2 per cent is deemed the desired level. I suspect it is because this is very roughly the share of the wider British population that is non-white ethnic minority. If so, then this is not the relevant figure when considering entry into elite professional jobsboth because minorities tend to be younger and also because roughly half of them are first generation immigrants, many of them do not speak English fluently, and so are not in the running to be taking up positions of leadership in the Civil Service.

It would be much more appropriate to create benchmarks based on the historic pipeline of qualified talent; one measure of that is the percentage of non-white ethnic minority people who left the Russell Group universities around the mid-90s which was 9 per cent.

Setting the benchmark too high is unwise because it encourages employers to scrape the barrel of the available talent pool, over-promoting people into jobs they are not qualified to do. The pressure to meet the targets is likely to influence decision-making at the highest level, but there seems to be very little scrutiny over its impact. Do senior civil servants such as Bunglawala ever reflect on the risks of getting it wrong?

Minorities also tend to cluster in different parts of the SCS. If we look at some individual departments and judging against our 9 per cent benchmark, we see good representation if not over-representation. For instance, 14 per cent of the SCS in the Ministry of Justice is from a non-white ethnic minority, as is 11 per cent at HM Treasury, and 26 per cent at Public Health England. Note that these departments map onto medicine, law, and finance – areas that ambitious minority individuals have historically excelled in. If there is a shortfall elsewhere, mostly felt in small departments and ministries, then perhaps it is the case that some parts of the Civil Service, DEFRA or the Ministry of Defence for example, do not connect with minority experience or ambition?

However, it is possible to mistake the appearance of things working with things actually working. You might think that 26 per cent is good but perhaps it ought to be 30 per cent? We have no way of knowing what to expect because we do not have accurate measures of preference and achievement which are the key variables necessary to predict what one ought to see. Without them any attempt to control the labour market is liable to distort it.

There is another fact that threatens to undermine the whole endeavour. Statistics on minority representation in places of employment are often not entirely what they claim to be. Usually, they are the share of people of known ethnicity within an organisation. Often people do not declare their ethnicity. In fact, in 2018, one quarter did not declare their ethnicity in the Civil Service. On grounds of pragmatism, it is fine to use just those of known ethnicity so long as the number of “undeclareds” does not get too large; then the statistic loses its validity. And increasingly this is what is happening. In the last 10 years, the share of undeclareds in the Civil Service has risen from 20 per cent to 25 per cent. The growth is most striking amongst those aged 20-29, from 26 per cent in 2008 to 43 per cent in 2018.

What if young people are either put off by tick-box diversity because they want to be valued by their individual accomplishments or worse scared by all this talk of discrimination so that they come to believe that disclosing their ethnicity will lead to them being discriminated against? Whatever the reason for not declaring their ethnicity, the result is that the statistics which form the basis by which the Civil Service judges its openness are slowly rotting. If such a trend is repeated in all walks of life, then this defeats the entire point of the Race Disparity Unit. (Note that non-disclosure amongst the SCS in the Cabinet Office is as high as 51 per cent.)

Moreover, what is to be done if it is shown that disparities favour ethnic minorities? The Prime Minister is currently consulting on how companies should best be compelled by Government to publish details of how they pay their staff by ethnicity, in order that “decisive action” can be taken. However, as the Ethnicity Facts and Figures website reveals, black and Chinese civil servants are paid more than white ones. Indeed, amongst senior civil servants, whites (median pay £81,000 per annum) earn less than their black (£90,000), Asian (£86,700), and Chinese (£88,400) colleagues. Pay whites more? Pay minorities less?

There is an irrational idea at the heart of thinking at the very top of government on questions of race and ethnicity, not to mention gender. It is the view that groups should be evenly spread across every walk of life and if they are not, then they should be made to be so, and it is all someone else’s fault. It also fails to include the possibility that things might be going right and that not everyone wants the same thing.

It is important to acknowledge the fact of ethnic/racial discrimination in the work place. The CV test has proven this. At the same time, we must be cautious in how we rectify things. Research has shown that voluntary measures work best in improving ethnic diversity. Rather than using ambiguous statistics to frighten people into action, it would be better to foster conversations based on respect and consent within our workplaces.

It is fair and reasonable to question the disparities that exist but it is quite another thing to automatically assume they are caused by racial bias and need to be corrected in all instances. After all, should society actively try to reduce the number of British Indian adults who are medical professionals (one in ten) in order to maximise the number entering the Civil Service? Who would gain and lose by such a crude approach? The consequence of bad diagnoses – no matter how well-intentioned – is likely to be bad solutions. As Burns wrote: “The best-laid schemes of mice and men, gang aft agley (go often astray), and leave us nought but grief and pain, for promised joy”.