Independent report

Employment, fairness at work, and enterprise

Updated 28 April 2021

Summary

This chapter looks at ethnic minority progress at work. It considers the history of, and current state of, pay and employment, and the trends in social class mobility across generations.

The employment rates for the White British and Indian ethnic groups were 77% and 76% respectively in 2019. For some others it was significantly lower at 69% for Black people, and 56% for people in the combined Pakistani and Bangladeshi ethnic group (this last figure is the result of a much lower female participation rate).^{[footnote 1]} Unemployment differences have been declining, though remain significantly higher for younger people.^{[footnote 2]}

The pay gap, meaning the difference between the median hourly earnings of all ethnic minority (not including White minority) groups and White groups, is at its lowest level since 2012 at 2.3%. Employees from the White Irish, Indian and Chinese ethnic groups on average have higher hourly earnings than the White British ethnic group.^{[footnote 3]}

Ethnic minorities have been making progress up the professional and occupational class ladder, though some more than others, and there remains under-representation at the very top. Employees from ethnic minority backgrounds are more likely than those from a White British background to say experiencing discrimination contributed to their failure in achieving their career expectations (20% versus 11%).^{[footnote 4]}

Also, using data from previous reviews led by Baroness Ruby McGregor-Smith and Sir John Parker, evidence heard from a wide range of stakeholders, and an examination of new data, the Commission identified 4 areas of focus:

1. Ethnicity pay gap, evaluating trends in pay and considering the value of ethnicity pay reporting in promoting fair outcomes, using the NHS as a case study.
2. Fairness at work, challenging existing approaches and examining alternative ways to promote fairness for ethnic minorities that leads to better outcomes and achieves inclusivity.
3. Empowering the next generation of entrepreneurs, stimulating the entrepreneurial instincts of enterprising young people from all backgrounds.
4. Artificial intelligence, considering how to identify and mitigate bias in artificial intelligence, and use it as a tool to promote fairness.

The case studies highlighted in this chapter have been identified by the Commission, during its evidence gathering phase, as positive examples of what works and are used for illustrative purposes only in the context in which they arise. The Commission fully recognises that there will be many other examples of similar good practice in the respective fields and industries, so wishes to make clear that in highlighting them they are not particularly endorsed or being given preferential treatment.

Employment and unemployment

The pandemic will have a significant effect on the British economy, causing a big reordering of employment and possibly a return to sustained high levels of unemployment. Some of the data presented below could therefore be overtaken by events. Nevertheless, the pay and employment story has been a broadly positive one in the last quarter of a century. There has been a gradual convergence on the White average in employment, pay and entry into the middle class, with some groups overtaking the White majority and others somewhat underperforming.

In the 1970s, still largely an era of full employment, there was very little disparity between ethnic groups in employment rates.^{[footnote 5]} This was at a time when racial hostility was much greater as well as more tolerated than today, underlining the point that equal outcomes do not necessarily signify lack of discrimination just as differential outcomes are not necessarily caused by discrimination. Employment rates then widened sharply in the 1980s and into the 1990s, due to deindustrialisation and recessions, but since the early 2000s have been narrowing again.^{[footnote 6]}

Figure 5: Percentage of working age people who were employed, by ethnicity over time (UK, 1972 to 2020)

Source: Evidence commissioned by the Commission, Norrie, R., Goodhart, D., using data from the General Household Survey and the Labour Force Survey analysed by Professor Yaojun Li.

The gap in employment between the White ethnic group and the Indian, Pakistani, and Bangladeshi groups has decreased over the past 20 years.^{[footnote 7]} This is mainly due to the fact that these ethnic minority groups started with notably lower employment rates in 2001. For example, the employment rate for the Bangladeshi ethnic group increased by 20.6 percentage points from 2001 to 2019, while the rate for the White ethnic group increased by 4.0 percentage points in the same period.^{[footnote 8]} Disparities in unemployment rates have been declining quite sharply since 2013.^{[footnote 9]} 8% of the Pakistani and Bangladeshi group, 8% of the Black group, and 6% of the Mixed group were unemployed in 2019, compared with 4% of the White group.^{[footnote 10]}

The pre-pandemic employment rate for the White British was 77% in 2019. The rate for people from the Indian ethnic group was just lower, at 76%, while most other minority groups had a lower employment rate: 69% for Black people and 56% for people in the combined Pakistani and Bangladeshi ethnic group. For White minorities, classified as White Other, the rate was higher than for the White British group (83%).^{[footnote 11]}

The pandemic is likely to have a mixed impact on the employment rate and financial stability of ethnic minority groups. For example, working in sectors shut down by the pandemic and being self employed is particularly prevalent among Pakistani and Bangladeshi men. This brings uncertainty of income in households which also typically have less savings to fall back on. Only 30% of people from the Bangladeshi, Black Caribbean and Black African groups live in households with enough to cover one month of income, compared to 60% of the rest of the population. On the other hand, the high representation of other ethnic minority groups in some key worker roles reduces their risk of income losses, at least in the short term.^{[footnote 12]} These are only early trends and changes to employment rates usually lag behind economic shocks. Historically, employment rates of ethnic minority groups have gone down more than overall employment rates during economic downturns in the UK. This indicates a risk to ethnic minority groups if the economic fallout from the COVID-19 crisis fails to turn into a V-shaped recovery.^{[footnote 13]}

There are substantial differences between men and women in employment rates in the combined Pakistani and Bangladeshi ethnic group: 73% compared with 39% respectively in 2019, likely due to cultural-religious reasons. Less marked differences are observed for all other ethnic groups: White British men and women 80% and 74% respectively, and Black men and women 71% and 67% respectively.^{[footnote 14]}

Age is another important factor, with employment rate differences between the 16 to 24 and 25 to 49 age groups largest in the Indian ethnic group (38% and 84%) and smallest in the White British group (58% and 86%).^{[footnote 15]} This difference is likely to be because young people from the Indian ethnic group are also among the most likely to go on to further education.^{[footnote 16]}

Disparities in unemployment rates between ethnic groups are also wider for the 16 to 24 age group. Overall unemployment in 2019 for White people was 4%, and 7% for all minority groups combined.^{[footnote 17]} For people aged 16 to 24, unemployment rates are far more pronounced: the White ethnic group has an unemployment rate of 10%, compared with 19% for ethnic minority groups.^{[footnote 18]} Black African and Bangladeshi ethnic groups see the highest rates of youth unemployment at 26% and 24% respectively.^{[footnote 19]}

Figure 6: Unemployment rate by ethnicity and age group (UK, 2017 to 2020 combined)

Unemployment rates for the 16 to 24 group are high even for those from Indian and Chinese ethnic groups who comfortably outperform the White average in education and incomes overall and generally benefit from positive stereotypes. It is noticeable that young people from the Black Caribbean ethnic group (who are more likely to come from families that have been established in the UK for longer) have a much lower unemployment rate than those from the Black African ethnic group, even though prejudice faced by both groups is likely to apply in equal measure.

Discrimination is likely to be a part of the story too, an assumption reinforced by field experiments in which responses are compared in imaginary job applications between people with majority and minority sounding names. We will look at this later in the ‘Bias at work and what to do about it’ section. ## Pay, social class and entry to top jobs

The hourly median pay gap between all minorities and the White British ethnic group has shrunk to 2.3%, its smallest level since 2012 when it was 5.1%.^{[footnote 20]} This headline figure hides some large variations: the Pakistani ethnic group earned 16% less on average than the White British group, the Bangladeshi ethnic group 15% less, and the Black African group 8% less. Meanwhile the White Irish (41%), Chinese (23%) and Indian (16%) ethnic groups earned more on average than the White British group.

In terms of gender, men earned a higher hourly median wage than women in all but 3 ethnic groups in 2019 (Arab, Bangladeshi, and Black Caribbean). Black Caribbean and Arab women also earn more on average than White British women.^{[footnote 21]}

Figure 7: Hourly median pay gap compared with the White British ethnic group, by ethnicity (England and Wales, 2019)

The ethnicity pay gap data only includes those who are employed in the UK, omitting those who are either self employed or not in work. This is significant, particularly when considering that many ethnic minority groups already have lower employment rates than the White British group.

These ‘raw’ pay numbers also do not take into account things like age, qualifications, region, whether someone was born in the UK or not, and when adjusted it can reduce the pay gap for some and increase it for others. So, for example, after adjustment, the strongly positive pay gap for the UK-born Chinese ethnic group turns into a small negative one, and for the UK-born Indian group is reduced to a smaller positive pay gap.

In other words, although people from the Chinese ethnic group are earning more on average than White British people, they are actually earning less when taking account of all of these other factors.

Figure 8: Hourly median pay gap compared with the White British ethnic group, adjusted for characteristics affecting pay, by ethnicity (England and Wales, 2019)^{[footnote 22]}

The overall convergence story on employment and pay is also reflected in a shift up the social scale. Some of this has happened almost automatically as the social structure has changed in the past 50 years, eliminating manual working class jobs and creating many more middle class and professional jobs, but some ethnic groups have moved up the social scale more than others. Over the past 50 years several ethnic groups have made exceptional progress in the UK.

According to work on social mobility by Professor Yaojun Li, ethnic minority children with parents in routine manual roles^{[footnote 23]} were much more likely to achieve upward mobility compared with their White peers. Only 5% of children from the Chinese ethnic group remain in the same routine manual positions as their parents, compared with 24% of White children.

With the exception of men from the Black Caribbean and combined Pakistani and Bangladeshi ethnic groups, most ethnic groups are now broadly level with the White ethnic group in terms of occupational class.^{[footnote 24]} Black Caribbean men arriving in the UK had a lower start in terms of class position and were well behind White men in occupational advancement or in access to the top occupational classes (including jobs such as teacher, manager, social worker and engineer) for most of the past 50 years. However, according to Li’s work there are also signs that they are catching up, with second-generation men (those born in the UK or arriving before the age of 13) being little different from White men and second-generation women surpassing White women.

Meanwhile, Pakistani and Bangladeshi men, along with Black African and Black Caribbean men, were the most vulnerable to unemployment in times of economic downturn, with the chances of getting a position in the top occupational class also declining over the decades for first generation Pakistani and Bangladeshi men. Women in the Pakistani and Bangladeshi ethnic group also tend to have persistent disadvantages relative to White women in terms of both employment status and class position. Three-quarters of the first generation and around half of the second-generation women in this group were economically inactive, although the situation has improved in the current decade.

In spite of the variations, there have been more signs of social progress than social regress over the past 50 years, with some groups, like those from the Indian and Chinese ethnic groups doing even better than those from the White ethnic group, and other groups catching up.

What about the extent to which ethnic minorities advance into the very top positions in professional, business and public life?

The executive recruitment agency Green Park’s Colour of Power project has recently found that 52 out of 1,099 of the “most powerful jobs” were held by ethnic minority individuals. That is 4.7%. It concluded that Britain’s leadership positions had failed to improve between 2017 and 2020, with only 15 additional ethnic minority-held roles since 2017.^{[footnote 25]}

When looking at the top of the FTSE 100, we see that there has been an improvement in the representation of ethnic minorities on board just in this last year. In January 2020, 52 FTSE 100 companies had ethnic representation on their boards, and in November 2020 this figure rose to 74.^{[footnote 26]} However, at the same time the number of Black people at the very top of the top FTSE 100 companies (meaning chair, chief executive or chief financial officer) recently fell from 2 to 0 (reported in February 2021).^{[footnote 27]}

Overall, some sectors of the UK are more open than others, the top of business and academia remain notably White while the public sector in general has a better track record than the private.^{[footnote 28]}

Now we take a closer look at the share of ethnic minorities in the upper echelons of 3 professions – the Civil Service, law, and medicine.

Civil Service

As of 2020, 9.1% of senior civil servants came from an ethnic minority background, which has increased from around 4% in 2010.^{[footnote 29]} There is substantial variation across government departments. In 2019, the department with the highest levels of ethnic diversity is the Department for Health and Social Care, where 12.8% of senior civil servants came from an ethnic minority background. In contrast, only 2.2% of senior civil servants at the Ministry of Defence came from an ethnic minority background.^{[footnote 30]}

Figure 9: Percentage of civil servants who were from ethnic minority groups (not including White minorities), by government department and grade

The Civil Service fast stream is the Civil Service’s talent spotting programme for its future leadership. The most recent data shows that, in 2018, over 12,000 applications were received from ethnic minority applicants (31.2% of those where ethnicity was known). Of people recommended for appointment, 15.8% were from an ethnic minority background. Both these figures have been increasing steadily since the late 1990s.^{[footnote 31]}

The Civil Service has a series of ethnic minority targets for appointments to the senior service. Its current target is to have 9.5% of appointments to the Senior Civil Service from an ethnic minority background, rising to 13.2% in 2022/25.^{[footnote 32]} It is currently falling short of these; indeed, they may be unrealistic targets given the minority share of the talent pool that entered the Civil Service at the turn of the century.

Law

As of 2019, 21% of lawyers working in law firms in England and Wales were from an ethnic minority background. 15% of lawyers are Asian (up 6 percentage points since 2014) compared with 7% of the workforce in England, Scotland and Wales in 2018. 3% of lawyers are Black (no change since 2017) and this reflects the workforce in England, Scotland and Wales in 2018 (3%). There has been no change in the percentage of lawyers from the Mixed (2%) or Other (1%) ethnic groups.^{[footnote 33]}

Ethnic minority partners tend to be clustered more in ‘sole-owner’ law firms as well as smaller firms. The largest firms (50 or more partners) have the lowest percentage of ethnic minority partners (8%). In contrast, 36% of one-partner firms have a partner from an ethnic minority background.^{[footnote 34]}

The most elite law firms, known as the ‘magic circle’ and including Freshfields, Allen & Overy, Slaughter & May, Clifford Chance and Linklaters, show some variation in their diversity levels at partner level.^{[footnote 35]} For instance, 6% of partners at Slaughter and May are of ethnic minority background compared with 10% at Linklaters as of 2018/19. These firms are mostly more diverse at associate level than the national average. Around 18% to 26% of magic circle associates are from an ethnic minority compared with 15% of associates across all law firms. The exception was Freshfields at 13%.^{[footnote 36]}

Among barristers, levels of ethnic diversity are broadly similar to solicitors. Overall, 14.6% of practicing barristers come from an ethnic minority background. At the QC level,^{[footnote 37]} 8.8% are from ethnic minority backgrounds. Finally, the ethnic diversity of pupil barristers is higher, with 22.9% of them coming from an ethnic minority background.^{[footnote 38]} Judges from ethnic minority backgrounds are still relatively rare but with this growing pipeline are expected to become more common.

Medicine

Medicine is a profession where ethnic minority groups are strongly represented. In 2020, almost half (46.1%) of doctors working in the NHS were from an ethnic minority background, with 30.2% from the Asian ethnic group, 5.2% from the Black ethnic group, 3.5% from the Mixed ethnic group and 2.6% from the Chinese ethnic group.^{[footnote 39]}

Senior doctors (including consultants and speciality doctors) were overall more likely to be White (56.2%) or Asian (31.4%) than junior doctors (including speciality registrars and Foundation year doctors) (50.5% and 29.3%), and likewise a higher percentage of junior doctors were from Black, Chinese, Mixed or Other ethnic backgrounds compared with senior doctors. However, representation varies between these grades. Overall, around 40% of consultants are from ethnic minority backgrounds, as is a similar share of foundation Year 1 doctors.^{[footnote 40]}

Figure 10: Percentage of doctors who were from each ethnic group, by grade

Ethnicity pay gap reporting

An ethnicity pay gap is calculated as the difference between the median hourly earnings of the reference group (White or White British) and other ethnic groups as a proportion of median hourly earnings of the reference group.^{[footnote 41]} For this reason, pay gaps, on whatever grounds, do not necessarily represent unequal pay for equal work, but can describe the structure of an organisation and how different ethnic groups are distributed across its pay bands.

Ethnicity pay gap reporting is a potentially useful tool but needs to be approached with care. Reported ethnicity pay data should be disaggregated by different ethnicities to provide the best information possible. The pay gaps, once identified, should be reviewed to gain an understanding of why they exist in different organisations.

Discussions with businesses have revealed that, like the Commission, they are aware of the pitfalls around the execution of ethnicity pay reporting, but feel that this work needs to start somewhere. We specifically consider the NHS as a case study later in this section, reviewing disparities in pay among ethnic minority healthcare staff as well as disparities in recruitment and progression.

We recognise the appetite that some employers have to act and publish their ethnicity pay gaps. A number of private and public sector employers (such as the NHS) have already voluntarily published their ethnicity pay gaps. We believe that ethnicity pay gaps should continue to be reported on a voluntary basis and that the government should provide guidance to employers who choose to do so.

It is clear that pay gap reporting as it is currently devised for gender cannot be applied to ethnicity. There are significant statistical and data issues that would arise as a result of substituting a binary protected characteristic (male or female) with a characteristic that has multiple categories.

The main statistical problem that arises with ethnicity pay reporting is the unreliability of sample sizes. If an employer with 250 employees (the threshold suggested in the 2018 BEIS consultation on ethnic pay gap reporting)^{[footnote 42]} reports a gender pay gap, on average they will be comparing 125 men with 125 women.

If they report an ethnicity pay gap as well, on average they will be comparing 225 White employees with 25 ethnic minority employees. Any findings from such a comparison will be unreliable and make it impossible to look at the workforce stratified by the 18 ONS ethnicity classifications.

If an employer is in an area with a low ethnic minority population there may not be a diverse local candidate pool for firms to employ from. The 2011 Census data shows that of the 650 constituencies in the UK, 437 are over 90% White,^{[footnote 43]} so many employers around the country simply do not have enough ethnic minorities for the recording sample to be valid.

For example, any employer in the Lake District can expect 98% of its candidate pool to be White. An employer there with 300 staff could then expect to have on average just 6 ethnic minority employees.

Any comparison between the median of 294 employees with the median of 6 employees will be meaningless and is likely to change considerably just from adding or subtracting one ethnic minority employee.

Additionally, the age distribution of ethnic minority groups can influence the ethnicity pay gap. Those from ethnic minority groups are more likely to be younger, meaning they have not yet had the opportunity to reach the peak of their careers. In order to account for this, firms would have to control for age, which makes sample sizes smaller and the reported data subject to fluctuations year on year.

Recommendation 9: Promote fairness – Investigate what causes existing ethnic pay disparities

The Commission recommends that all employers that choose to publish their ethnicity pay figures should also publish a diagnosis and action plan to lay out the reasons for and the strategy to improve any disparities. Reported ethnicity pay data should also be disaggregated by different ethnicities to provide the best information possible to facilitate change. Account should also be taken of small sample sizes in particular regions and smaller organisations.

To support employers undertaking this exercise, the Commission recommends that the Department for Business, Energy and Industrial Strategy (BEIS) is tasked with producing guidance for employers to draw on.

NHS pay, recruitment and progression

In 2018, the NHS became one of the first public sector organisations to publish breakdowns of pay for all staff by ethnic group, but the picture presented by the overall NHS data is complicated.

For instance, in 2019, the average monthly basic pay of (full-time equivalent) Asian men overall, was £3,864, greater than that of White men (£3,145) who in turn earned more than Black men (£2,646).^{[footnote 44]} Similarly, Asian women (£2,717) earnt more than White women (£2,491), who in turn earned more than Black women (£2,320).^{[footnote 45]} There is a disparity here but more information is required to understand why and where this disparity exists. The high number of Asian groups at consultant level in the NHS is likely to explain some of this difference.

Generally, there is no consistent pattern with examples of both positive and negative pay gaps, as well as examples of parity. One of the standout figures is that Black male senior managers earn just 83p for every £1 earned by their White counterparts, while Asian male senior managers earn £1.01.^{[footnote 46]} This might have something to do with the seniority levels within this category. At the ‘very senior manager’ grade, Black men earn 97p for every £1 earned by White men, while Asian men earn £1.04. The corresponding figures for women are 99p and £1.06 respectively.^{[footnote 47]}

Such a picture is not consistent with a pattern one might expect of systemic discrimination, although undoubtedly, there will be cases of discrimination and bias in what is the largest employer in the country. The weight of the evidence considered tends to point towards the White group receiving marginally more pay within staff groups. An analysis of pay by ethnicity controlling for age would provide further insight into pay differences.

Case study: The Surash Pearce Report on Ethnic Pay Gap

This report^{[footnote 48]} (published in September 2020) presents a comprehensive review into ethnic pay gap and workforce development at The Newcastle upon Tyne Hospitals NHS Foundation Trust.

The report used the following data during its review:

• electronic staff records
• WRES data – 2018/19 data
• NHS National Staff Survey
• recruitment data from TRAC recruitment management system
• local and national Clinical Excellence Awards data

By combining these varied data sets the review was able to build a more in-depth understanding of differences in the pay, progression and experiences of the trusts’ ethnic minority staff.

Some of the key headline findings include:

• BME staff are more likely to pursue continuous professional development and non-mandatory training
• BME staff at Newcastle Hospitals report being more likely than White staff to be bullied or harassed by other members of staff, with less opportunity to progress in their careers, but more discrimination from managers
• White staff report experienced more abuse from service users compared with BME staff at Newcastle Hospitals
• a White applicant to a consultant job in Newcastle is more likely to be shortlisted and more likely to be successful at interview (relative likelihood 1.53 times)
• additional programmed activities (PA)^{[footnote 49]} payments: female consultants lag behind male consultants
• Clinical Excellence Award (CEA): Locally a BME consultant makes 24.5% less than a White consultant, compared with national awards which is only 5.4% less.
• more female BME consultants applied for the local CEA^{[footnote 50]} (2017/18) and were more likely to be awarded CEAs, however the value of their award was the lowest level.

Impact:

Newcastle Hospitals believe they are the first Trust in the country to publish such detailed, transparent and extensive information about this issue. Some of the information in this report makes difficult and uncomfortable reading but the Trust has reflected on it and is using the report to learn from and as a basis for improvement. Newcastle Hospitals has used the information to inspire positive dialogue and taken action to address race inequality in the workforce.

Case study: Occupational preferences research

During the Commission’s investigations of this topic, an exploratory piece of research^{[footnote 51]} was commissioned to try to understand possible reasons behind the occupational preferences of NHS staff and if there were any differences by ethnicity. The study had a small sample size (n=116), along with an uneven distribution for gender, White and ethnic minority groups, and full and part-time employees. As a result, findings should be regarded as indicative only.

However, this is an example of the kind of work that could be done by Trusts or the NHS as a whole to consider the causes of any career progression gap between different groups of employees:

The study explored factors influencing behaviours in the attainment of managerial and leadership roles across ethnic groups – that is, White and ethnic minority populations in the NHS. As the authors stated “the objective of this study was to offer preliminary findings on whether there is a significant difference in behaviours of ethnic groups that explains variance in attaining managerial and leadership roles”.

The study found that a much higher proportion of respondents from Black and Asian groups perceive becoming a manager or leader to be a risky career choice. IT also found that a higher number of Black respondents disagree that their manager and others more senior provide them with feedback to become a manager, in comparison with White respondents. This merits better understanding as there are simple HR activities which can address these perceptions.

The research also suggested across all ethnic groups there was a link between the level of educational attainment of respondents’ parents and the respondents’ career progression. Therefore suggesting that socio-economic background is a factor meriting further investigation where racial disparities arise.

As noted previously, the pay gaps, once identified, should be reviewed to gain an understanding of why they exist in different organisations. There are several ways in which this could be done. An example of this is demonstrated through the case study above of an exploratory piece of research conducted of NHS staff.

Recommendation 10: Promote fairness – Improve understanding of the ethnicity pay gap in NHS England

The Commission recommends that NHS England as a whole should commission a strategic review of the causes of disparate pay and, where discrimination is pinpointed, spell out the measures that might meaningfully address it. Such a review would shine a light on the barriers to in-work progression and how to overcome them – for example, in promotion, are foreign qualifications equally validated yet ‘informally’ seen as ‘inferior’. It would ask how the NHS performs on pay gaps compared with international comparators and if other metrics than pay gaps reveal barriers better.

Bias at work and what to do about it

Bias and discrimination at work are usually hard to pin down, especially when they are not only a social taboo, but also illegal. The field experiments using job applications tests, mentioned below, provide conclusive evidence that bias, at least in hiring, does exist. They also highlight the need for more scrutiny of the interventions that do make a difference.

Job application field experiments have been carried out in the UK since the late 1960s. The most recent was conducted by Dr Valentina Di Stasio and Professor Anthony Heath as part of a wider EU-funded project and found a call-back ratio of 1.6:1.^{[footnote 52]} In other words, to receive a call-back, people with ethnic minority names had to write 1.6 letters for every 1 written by someone with a majority name. A less recent study focused on UK cities was conducted by NatCen on behalf of the Department for Work and Pensions. This study found a similar call-back ratio of 1.7:1, with ratios varying between ethnic minority groups from 1.5:1 to 1.9:1.^{[footnote 53]}

Figure 11: Response rates to job enquiries (‘call-back ratios’) compared with the White ethnic group, by ethnicity (DWP, 2009)

There are important caveats. We know that discrimination occurs, but these experiments cannot be relied upon to provide clarity on the extent that it happens in every day life.

While these application tests show discrimination against names that are recognised as not being traditionally British, it is unclear if this effect is about race, class or perceived foreign culture. More granular studies should be conducted which manipulate first names as well as surnames, class as well as ethnicity, and which include greater and lesser levels of CV qualifications, as all have been found to affect the results.^{[footnote 54]} They should also attempt to assess the ethnic makeup of those making hiring decisions, or at least the ethnic composition of the hiring organisation or unit, to test for variations in bias among assessors.

Field experiments are, of course, not the only possible evidence for discrimination. Varying promotion rates may also signify discrimination, with more ethnic minoirty employees overall saying that experiencing discrimination contributed to them failing to achieve thier career expectations than those from a White British background (20% compared to 11%). Black employees had a higher rate of experiencing discrimination at 29%,^{[footnote 55]} but this was still not the experience of the majority.

Subjective factors may also affect perceptions of discrimination. Human beings tend to discriminate, even when unintended. We are all susceptible to differentiating between in-groups and out-groups and will be prone to favour those we perceive as belonging.

The Commission recognises that some organisations have already adopted initiatives that produce fairer recruitment, for example through name-blind applications and diverse interview panels. For example, Baker McKenzie, the international law firm, uses a combined approach of blind screening and the Contextual Recruitment System (CRS) developed by Rare Recruitment (see case study below).^{[footnote 56]}

Evidence submitted from the Employers Lawyers Association (ELA) to the Commission’s call for evidence further highlighted that many firms have adopted Rare Recruitment’s Contextual Recruitment System in recruiting candidates from diverse backgrounds. This initiative was considered important in helping the legal profession in particular to reduce the evidenced racial inequalities that exist when recruiting young people.^{[footnote 57]}

Case study: Rare Recruitment’s Contextual Recruitment System

Research shows high-achieving people from disadvantaged backgrounds outperform their peers in elite jobs. Rare’s pioneering Contextual Recruitment System (CRS) provides real time contextual information that allows organisations to identify exceptional candidates they might otherwise miss.

The system uses big data to help organisations identify candidates with the greatest potential. It measures candidates’ achievements, including academic performance at university, against uniquely comprehensive datasets and classification systems developed by Rare over a decade.

The system works by combining publicly available information with candidates’ application responses. Rare uses information from their bespoke databases. For example, in the UK, the first database contains the exam results of more than 4,000 secondary schools and sixth-form colleges nationally, while the second contains 2.5 million residential postcodes.

The system delivers 2 outputs to employers: flags to measure disadvantage and Performance Index (PI) to measure outperformance against students at the same school. Rare reports that firms adopting their CRS hire 61% more people from disadvantaged backgrounds.

Unconscious bias is a form of ‘fast thinking’ – those quick decisions we make without realising and are no doubt connected to particular values or world views that are hard-wired into our minds. Such bias is commonplace because it is reflexive and automatic. It is also laced with preferences and prejudices based on our upbringing and family and social backgrounds.

All people, not just White people, are subject to these biases but it matters more if you are 84% of the population and tend to dominate the top positions. There is a perception that people at the top tend to have affinity bias, appointing people in their own image. Polling by Harvey Nash’s Engage network of senior executive and board leaders from ethnic minority backgrounds found that 63% believed that the unconscious biases of boards and CEOs is a factor in explaining why there is so little diversity at board level.^{[footnote 58]}

Indeed, FTSE 100 CEOs do have a certain profile. They are nearly all White^{[footnote 59]}, male, university graduates and one-quarter have MBAs. 18% have studied at Oxford or Cambridge, and almost half of them have been promoted from inside the same company (46% in 2019).^{[footnote 60]} If ambitious minority people do not fit the mental image of what leadership looks like, they fear they will be overlooked. For those who come from a different background or have a different set of experiences, the implicit culture or feel of professional life can seem strange, especially for those from a minority or working-class background.

Research by the Equality and Human Rights Commission found that for FTSE 350 leadership role descriptions provided by 25 companies, 28% explicitly specified some form of cultural fit or ‘chemistry’.^{[footnote 61]} Youth Futures Foundation and research from NatCen Social Research illustrate how these requirements can further disadvantage those from underrepresented backgrounds:

Such ‘information gaps’ can grow even larger if young people from more affluent backgrounds have a better understanding of the demands of a job application process as well as a greater appreciation of the relevant cultural norms (which can be particularly beneficial in interviews and assessment centres).

(Youth Futures Foundation, call for evidence)

Young people across ethnic groups were aware that they may lack the cultural capital that high skilled employers may value. In turn, they suspected that prospective employers may make (conscious or unconscious) judgements based on the way graduates from working-class backgrounds dress and present themselves, as well as their vocabulary, accent and ability to articulate themselves in an interview.

(NatCen Social Research^{[footnote 62]})

Many companies have been prompted into intense soul-searching with regard to race, prompted by the Black Lives Matter movement last year. They have adopted various diversity and inclusion indexes, tick-box exercises and charters, such as unconscious bias training. The result, however, seems to be a focus on process rather than outcomes: this training scheme, that equality initiative, a newly designed culturally neutral form.

Most researchers remain sceptical about the impact of unconscious bias training, quotas and diversity specialists. Research by Kalev and Dobbin, published in the Harvard Business Review, found that mandatory diversity and inclusion measures have not always been successful.^{[footnote 63]} Many of the products of the diversity and inclusion industry are not standardised or certified by an official body, or independently assessed to show any level of efficacy, and some of the most popular products, such as unconscious bias training, have been shown to have unclear or mixed impact. Indeed, research by Her Majesty’s Inspectorate of Probation found that sometimes, probation staff feel reluctant to speak freely for fear of their words coming across as racist, limiting opportunity for further discussions.^{[footnote 64]} The Equality and Human Rights Commission found that evidence for unconscious bias training’s ability effectively to change behaviour is limited and “there is potential for back-firing effects when UBT participants are exposed to information that suggests stereotypes and biases are unchangeable.”^{[footnote 65]}

While the Commission recognises the place of such practices in the journey to promote diverse and inclusive work environments, it maintains that diversity and eliminating disparities requires impactful organisational redesign and training that leads to truly inclusive environments. By inclusive, this means an environment where anyone feels comfortable to be themselves and confident that they have the same chance of succeeding as anyone else with the same qualifications and experience. Diversity training and policies that treat people differently according to ethnicity does not work.

We all have a responsibility to reduce bias and there must be some tolerance for human fallibility. As Musa al-Gharbi makes clear in his article ‘Diversity is Important. Diversity-Related Training is Terrible’, diversity-related training should be a process of mutual exploration in which bias, discrimination, nepotism and motivated reasoning should be seen as ‘general cognitive tendencies’ which people of all backgrounds and ethnicities are susceptible to.^{[footnote 66]} This indicates ‘nudge’-style procedures (such as name-blind CVs, transparent performance metrics, family-friendly policies, proactive mentoring and networking procedures) are more useful than methods that overtly discriminate against some groups, for example quotas.

The Commission recognises that employers will want to do something to show a commitment to fairness. However, finding one ethnic minority face for a board, for example, is not a substitute for a proper fairness strategy. It should not be assumed that someone’s ethnic background will change a board’s culture, nor that when people from different backgrounds come together they will be more creative. Greater emphasis should be placed on diversity of thought and perspective around a board table which is not associated with anyone’s race or ethnicity. Widening the pool of places where positions are advertised or people recruited could attract more diverse candidates in every sense of the word without discriminating against any group.

Organisations can be (re)designed to change behaviour, and therefore outcomes. Sometimes the changes are so simple that they seem trite, for example, using images of successful colleagues with an ethnic minority background on walls. But these ‘nudge’ measures and a better understanding of how selection for top jobs operate have been proven to produce among the best results and beneficial outcomes, without alienating people.

One example of increasing representation through the use of nudge measures such as mentoring is Liverpool City Council’s initiative ‘Step Forward - Step Up’. The programme, due to be launched in September 2021, will be delivered by a consortium of key city employers led by the Lord Mayor of Liverpool. Through a series of development modules (such as leadership development programme, mentoring and action learning sets), it will seek to develop and create a diverse talent pipeline in Liverpool. Participating employers support the programme by sponsoring participants’ fees, promoting career opportunities, contributing to modules, providing mentoring and committing to embedding diversity at all levels, among other things. This programme was developed in partnership with the successful model created by Bristol City Council’s ‘Stepping Up’ initiative,^{[footnote 67]} a locally driven public-private collaboration to advance representation.

From the evidence it received, the Commission identified a promising inclusion tool by Zyna Search developed by In Diverse Company that provides an alternate way for businesses to measure inclusion based on behavioural psychology and research. The bespoke model, called the Cultural Inclusion Maturity Model (CIMM)^{[footnote 68]}, is focused on driving consistent and meaningful behaviour changes in a measurable way, and tailored to meet respective business needs. In Diverse Company’s work with private and public sector firms has revealed that, despite an increased ethnic diversity at the board level, other employees in the organisation still face marginalisation.

Shifting cultures takes a more sustained effort than increasing representation. Both however, are critically important, and are likely to feed into each other. The Commission has seen (preliminary) evidence of the CIMM’s novel approach yielding positive results, for example within a leading pharmaceutical company and also a recruitment company that In Diverse Company have partnered with.

Case study: Cultural Inclusion Maturity Model

The Culture Inclusion Maturity Model was developed with a mission to create a dynamic system of inclusion measurement that enables organisations to recognise, track and monitor progress in establishing an inclusive workplace on a global scale. The constructs of the model are underpinned by psychological and business research.

The unique methodology measures inclusion by assessing the behaviours of leaders, teams and employees and auditing the organisation’s capability (defined by policies, processes, and procedures). The model not only provides a specific measure for an organisation which can be compared with others, it also clearly defines where the measure is in terms of its maturity level and as such enables organisations to see how they can make improvements.

At the end of the process, organisations receive:

• an inclusion accreditation – either bronze, silver, or gold
• a comprehensive breakdown of findings – highlighting areas of the organisation that are more or less inclusive than others
• a list of recommended actions – tailored to the organisation, specifically designed to increase overall inclusion and employee well-being

The model has been tested for validity and reliability in the UK, Europe, Asia and the US, and across a range of sectors to ensure the measure is robust. Progress is measured by running the CIMM on a regular basis (every 12 to 18 months) with key outcomes for organisation including, improved:

• well-being – inclusive work cultures provide a safer place for employees to perform at their best
• transparency – inclusive cultures drive transparency and reduce instances of corporate wrongdoing and corruption, build trust between employees and managers, and create reputable brands
• business performance – inclusive cultures drive better decisions, productivity, innovation, creativity leading to better business performance
• diversity – inclusive cultures, ones where all employees feel they belong and can also be open about their differences, are ones that not only attract diverse talent but also they get the most from diversity of perspective

Impact: there is early evidence to show additional organisational benefits, such as a reduced number of grievance and disciplinary actions, reduction in sickness levels, and increases in staff engagement and customer satisfaction.

The Commission also recognises the important role of public sector procurement and supporting in-work progression. It welcomes the publication of the government’s ‘Transforming public procurement’ green paper^{[footnote 69]} and supports the work of the In-Work Progression Commission led by Baroness McGregor-Smith.

Recommendation 8: Promote fairness – Advance fairness in the workplace

a) The Commission calls on organisations to now move away from funding unconscious bias training. The existing training should be replaced with new interventions that when implemented, can be measured or evaluated for their efficacy, such as:

• mandated sponsorship groups to ensure wider exposure of ethnic minority individuals to their peers, managers and other decision makers
• training and routine skills support for all employees in their professional and personal lives (for example on collaboration, confidence, communication, and presentation skills), which could disproportionately benefit more disadvantaged groups

b) The Commission also calls on the government to work with a panel of academics and practitioners to develop resources and evidence-based approaches of what does work to advance fairness in the workplace. The landscape of diversity training is highly mixed, and the government can play a role in guiding organisations to high quality materials and resources.

These resources should include guidance for employers, and be piloted in the Civil Service to replace the use of unconscious bias training.

NHS staff experience

The NHS is a success story with significant over-representation of ethnic minorities in high status professional roles, but also a less happy story, with a consistently negative experience reported by many of its ethnic minority staff at lower levels.

The Commission heard from frontline staff who felt that there were still informal progression or recruitment practices that could be slowing the reduction of disparities. The Commission heard that “For exec and non-exec roles, having transparency in the application process is vital. We have examples of conversations and pats on the back to get appointments to the board.” This was raised as a particular issue when moving into management: “At junior stage I didn’t feel discrimination but getting forward into other roles it becomes difficult. It goes back to these networks that the average Black person doesn’t have”.

These processes are challenging to nail down and then to police, given their informal nature, but more broadly, the NHS recognises that there are problematic disparities which impact the experience, wellbeing and health outcomes of staff themselves, as well as their career progression, and the performance of the Trusts in which they work.^{[footnote 70]}

Ethnic minority staff report worse experiences, when compared with White staff. The Commission heard examples of the kinds of negative experiences faced by some on the frontline; “the issues came when I was actually qualified. I would go on to wards and not even be recognised [as a nurse]– people would even ring agencies to confirm if I was an actual nurse”.

The NHS’s Work Race Equality Standard (WRES) is a series of indicators chosen to measure equality within the NHS. Evidence from WRES shows ethnic minority NHS staff are more likely to report personally experiencing discrimination at work from a colleague (15.3% of people who were not White, compared with 6.4% of White workers).^{[footnote 71]} They are also more likely to report harassment, bullying or abuse (29% compared with 24.2%), and less likely to believe there were equal opportunities for career progression or promotion (69.9% compared with 86.3%).^{[footnote 72]}

In terms of actual opportunities, WRES shows White applicants were 1.46 times more likely to be appointed from shortlisting across all posts, while ‘BME’ staff were more likely to be formally disciplined by a factor of 1.22.^{[footnote 73]}

It should be noted that the surveys from which this data is taken have a low response rate for ethnic minorities (16.7%).^{[footnote 74]} WRES indicators are limited to White staff versus ‘BME’ staff, and so do not tell the full story as to where the actual disparities lie.

The Commission accepts the work of WRES as important to monitor fairness within the NHS but suggests that further detail is required. This includes disaggregating within the ‘BME’ category, as well as looking at opportunities in more detail, most notably in terms of disparity within staff groups.

In an effort to establish how NHS organisations can address workforce inequalities, the King’s Fund looked at case study examples of where interventions such as establishing staff networks, ensuring safe routes for raising concerns, and enabling staff development and career progression were being implemented.^{[footnote 75]} It was found that these were broadly perceived to be beneficial, but also recognised that there is no single ‘one size fits all’ approach that can work.

The Care Quality Commission (CQC) takes a range of diversity indicators, including WRES scores and recruitment data, into account when running its inspections, as well as engagement with bodies or groups within healthcare providers to gain qualitative understanding of such issues. However, the Commission has heard about a lack of trust in the CQC process among frontline staff. As one paramedic stated, “the CQC has made it clear they ignore race issues”.

The Commission heard feedback that more needs to be done by the CQC to ensure disparities are better understood and considered in inspections. The results from this have the potential to be one of the strongest tools for encouraging change and progress for one of the largest employers in the UK.

Recommendation 2: Build trust – Review the Care Quality Commission’s (CQC) inspection process

The Commission recommends that the Department of Health and Social Care (DHSC) commission a review into the CQC’s approach to scoring employee diversity and inclusion in their inspections.

The Commission recommends that this review is chaired by an expert with close knowledge of the health care system and CQC internal processes, ideally a former inspector or inspector of an alternative inspection body. The review team should work closely with the NHS Workforce Race Equality Standard team and the disciplinary bodies of the medical professionals to ensure that the views of these bodies feed into this work.

Enterprise

The UK continues to rank as one of the world’s most entrepreneurial countries.^{[footnote 76]} Enterprise is a crucial engine of economic growth and the entrepreneurial path should be accessible to all. It has some of the lowest barriers to entry, yet potential entrepreneur’s experiences are influenced by their ethnicity, gender, age and geographic location.

There is clearly an appetite for those from ethnic minority backgrounds to pursue enterprise, and benefits to the rest of society when they do. In 2018 around 250,000 firms in the UK were ethnic minority-led, and generated around £25 billion of growth value added (GVA) – the equivalent of the city of Manchester.^{[footnote 77]} Additionally, ethnic minority owned firms were more likely to export their goods and services than White owned firms in all regions across the UK, bringing vital income to the UK economy.^{[footnote 78]}

The Commission believes in the power of agency. The need to stimulate the entrepreneurial instincts of the country’s people is greater than ever. Whilst this is the case, we know that more support for ethnic minority entrepreneurs and those from lower socio-economic backgrounds is needed. The data shows that:

• Those from Asian and Other ethnic minority backgrounds are less able to get access to finance to start working on their idea. 49% of Asian and other ethnic minorities previously aspiring entrepreneurs cited ‘difficulties getting finance’ as a reason they stopped working on their idea, compared with 25% of White British people.^{[footnote 79]} We explore some of the reasons as to why later.
• Despite those from Black backgrounds having similar success rates to their White counterparts in starting a business, they also see worse outcomes.^{[footnote 80]} This is thought to be a result of their location and lower household income levels (40% of ethnic minority people in the UK live in London,^{[footnote 81]} which faces higher costs as well as higher competition).
• In terms of performance, Black business owners report a median turnover of £25,000 – this is £10,000 less than White business owners and £15,000 less than Asian and Other ethnic minority business owners. In 2019, 72% of Black and 62% of Asian and Other ethnic minority business owners report making a profit, compared with 84% of White business owners.^{[footnote 82]}
• For female business owners, over one third of those from Black, Asian and Other ethnic minority backgrounds reported making no profit in 2019, compared with 15% of White female business owners.^{[footnote 83]} Overall, median turnover for female entrepreneurs’ businesses was a third of male entrepreneurs’ and productivity was less than half.
• Those who are afforded more and better opportunities to build social capital (for example, through their jobs) have higher success in comparison to those with less opportunity (for example, those previously looking after the home). However, there is an under representation of Black people in managerial and senior roles, thereby reducing their opportunities to develop relevant social capital.^{[footnote 84]}
• Ethnic minority entrepreneurs also experience poorer outcomes in accessing venture capital than their White counterparts – Black entrepreneurs were cited as experiencing the poorest outcomes of all (with only 38 Black entrepreneurs, in a sample of 3,784, receiving venture capital funding in the past 10 years).^{[footnote 85]}
• 40% of the total capital invested at seed stage between 2009 and 2019 went to founding teams with at least one member from a top ranking university.^{[footnote 86]} This indicates that those who receive higher education from such institutions disproportionately benefit from receiving starting capital towards their venture.

The Commission found there to be a fundamental barrier in communication between ethnic minority groups and high street banks. Many aren’t engaging and feel discouraged in seeking finance due to a fear of rejection. There is evidence to show that business owners from ethnic minority backgrounds are disproportionately declined for lending^{[footnote 87]}, and while there is no evidence of racial discrimination by banks^{[footnote 88]}, this has resulted in perceptions of a systemic disadvantage underpinning discouragement.^{[footnote 89]}

When given evidence to the Commission, Jamaican-born entrepreneur Levi Roots, recalled his own experience as a young entrepreneur in Brixton during the 1980s. He was repeatedly declined for investment and felt dissatisfied with the advice he received from his local bank. Roots believed that the support and opportunities available to him as an aspiring entrepreneur were limited in comparison to those in the more affluent area of Herne Hill nearby.

While some have found financing to support their business ideas through avenues such as the government-backed Start Up Loans scheme^{[footnote 90]}, many others are reluctant to borrow from banks, preferring to finance their business ideas through considered safer means such as personal savings, or support from friends and family. However the limited personal resources of those from lower socio-economic backgrounds consequently hinders their development of their ideas. There is work here for high street banks and lending institutions in making themselves more accessible to these community groups who currently do not consider lending to be a viable access route for them.

The Commission considered ways in which to better connect the industry with disproportionately affected groups from ethnic minority and lower socio-economic backgrounds and widen access for them. Banks are already asking this question and conducting research, but we have seen few action based solutions. More than ever, the younger generation have the desire to be their own boss, but of all age groups, are unlikely to act on this ambition.^{[footnote 91]} They are alive to the latest technologies, trends and ideas, but are unfortunately, among those hardest hit by the economic downturn of the COVID-19 pandemic.^{[footnote 92]} For many, starting a business poses too big a risk when compared to the stability of more traditional forms of employment.

The Commission heard from many young people who shared their ambitions and voiced their frustrations regarding perceived barriers. Many among them felt strongly that there was a need to mitigate against the regional inequalities of labour market opportunities available to them. It is for this reason that the Commission chose to focus on providing increased opportunities to young people. We recognise the benefit of having targeted interventions to support them, particularly those facing worse outcomes as a result of demographic factors. We believe enterprise is a crucial way to raise individual agency and contribute to the country’s economic recovery.

Universities have been known to be regional engines of innovation. They contain a copious amount of untapped potential and present the ideal infrastructure and resources to generate ideas, activities and meet those like minded. Yet relatively few university graduates go on to become entrepreneurs.^{[footnote 93]} The Commission felt that there was much fertile ground in better utilising the role they can play in nurturing entrepreneurial spirit. As it stands, opportunities across universities are being missed, with a disproportionate amount of capital at seed stage invested in those from the UK’s elite^{[footnote 94]} and tier 1^{[footnote 95]} educational institutions.^{[footnote 96]}

Yet, several universities across the UK have been recognised for the support they provide to aspiring entrepreneurs.^{[footnote 97]} These institutions reside in areas with large ethnic minority populations, serve their local populations and support their local economies, yet face limitations from the low endowments provided to them, in comparison to those more research-focused and internationally-facing.^{[footnote 98]} Generating investment into them could further aid in their ability to provide talented students with early opportunities to build companies, and grow talent within their regions. Financial institutions are in a position to offer the required resource, and should be encouraged to do so.

During our evidence session with HSBC UK, we found there to be much goodwill and a desire to support under-represented communities bridge the gap of enterprise ambition with reality. There was an awareness of the disparities these groups faced and agreement that institutions who were committed, and working hard to support underrepresented groups through their journey, should benefit from additional support in doing so. After all, entrepreneurial success requires rounded support – this means financial investment accompanied by business support services such as mentoring, capacity building and network access.^{[footnote 99]}

Together with HSBC UK, the Commission developed a proposal which aimed to encourage the entrepreneurial ambitions and innovation of the next generation through the spirit of competition. Competition is a key engine of innovation and we have seen cases of this time and again – take Dragon’s Den for example. There are many individuals across the UK from all backgrounds who, with some financial investment and support, have the opportunity to become a household name like Levi Roots.

Since its inception in 2014, HSBC UK’s SME lending fund has supplied over £75 billion to boost British business growth.^{[footnote 100]} As a part of their renewed commitment in 2021 to lend a further £15 billion to SMEs through this fund, the Commission felt it was important that young people were given the opportunity to contribute to the country’s economic recovery through this fund.

Recommendation 17: Create agency – Encourage innovation

The Commission recommends that HSBC UK works in collaboration with universities across the UK to pilot a competitive enterprise programme that will target aspiring entrepreneurs from underrepresented and low-income backgrounds.

The universities that would be considered to take part in the pilot will be those who would benefit most from an increased endowment that would: 1) bolster their offer of support to aspiring entrepreneurs; and 2) further enable them to nurture entrepreneurial talent.

The programme will support participants in the development of their proposals through the provision of advice, mentorship and access to networks, and provide financial backing towards the winning entrant’s enterprise. The Commission envisages that participants of the programme will form an alumni community that will act as an additional source of support.

This should act as a model for other banks and financial institutions to emulate in collaboration with universities as a way to nurture talent, encourage innovation, and offer support to aspiring entrepreneurs from underrepresented and low-income backgrounds across the UK.

Entrepreneur Levi Roots was supportive of this recommendation, saying:

There is no doubt that disparities, especially towards Black and ethnic minority entrepreneurs in employment and higher education have become that slippery, greasy pole of opportunities on which our aspiring young are now attempting to navigate. The ability to adapt and be resilient through such interventions is key to speeding up and supporting their ascension towards these aspirational goals.

Artificial intelligence

Artificial intelligence (AI) systems raise complex questions of fairness. AI can be hugely beneficial – for example through speed, transparency, and better decisions. However, it brings novel ways for bias to be introduced. Because these systems can be replicated at a huge scale, the Commission knows that understanding and accountability in these systems are critical. From hiring, to loan administration, policing, insurance pricing, and more. These systems can improve or unfairly worsen the lives of the people they touch.

To steward automated systems responsibly, we must acknowledge their novelty and complexity. First, we need to define fairness mathematically so we can evaluate these systems. There are many^{[footnote 101]} possible definitions which fit different contexts. Our existing legal structures, such as the Equality Act, may need to be re-evaluated for the algorithmic age.

Second, bias can enter the system in various ways. First, through data – for example, if a loan decision system receives data for a particular ethnic group that only contains default borrowers, the system may learn to use ethnicity data to unfairly deny loans to that group. Second, through the model; for example, unfair rules may be unknowingly hardcoded into the system (in practice, this is less common). Last, through decisions; for example, a system may give a fair output which humans may be more confident about than is deserved.

The rate of change in these systems means specific remedies are premature. Organisations and decision-makers should use tools that detect and mitigate bias before, during, and after a system deployment. For example, Algorithmic Impact Assessments^{[footnote 102]} are questionnaires that help raise fairness risks before a system is used. There are also now myriad technical tools that can be used to inspect data, models, and outputs for bias.^{[footnote 103]} The government and organisations need to stay up to date with the rapidly advancing field of fairness, accountability, and transparency^{[footnote 104]} in automated decision systems.

Last, before dismissing any system, it should be compared with the alternative. An automated system may be imperfect, but a human system may be worse.

Case Study: Addressing bias in hiring to let talent flourish

How could we ensure automated hiring systems hire for talent and talent only? First, we must have 2 hard conversations. What do we mean by talent? And what does it mean to be fair to it?

Human bias is the greatest risk at this stage. Assuming definitions of talent and fairness (a big assumption), we might intervene in various ways. A hiring system trained on the data of a company’s historical employees may learn biased relationships if the company has disproportionately hired one ethnicity, therefore, the hiring analyst should think about the data collected. One could modify the model, for example, by penalising it during the “learning” phase if ethnicity and predicted outcome are statistically dependent. Or, one could modify the outputs.

For example, if the model predicts different hiring outcomes for the same candidate if only the value of the “ethnicity” variable changes, the output is discarded and not used.

These are only examples and the reality is much more complex. However, it shows that there may be ways to trust automated systems and remove human biases to allow talent, not other characteristics, to be rewarded.

The Commission acknowledges that the government is currently working to advance this field in order to accelerate AI adoption and fair outcomes. We welcome the Centre for Data Ethics and Innovation’s (CDEI) ongoing work in this space and supports the recommendations from its review into bias in algorithmic decision-making.^{[footnote 105]} If acted on, the recommendations will help change the behaviour of all organisations who make potentially life-changing decisions on the basis of data and algorithms.

Recommendation 3: Build trust – Improve the transparency and use of artificial intelligence

The Commission supports the recommendations of the Centre for Data Ethics and Innovation (CDEI) and calls on the government to:

• place a mandatory transparency obligation on all public sector organisations applying algorithms that have an impact on significant decisions affecting individuals
• ask the Equality and Human Rights Commission to issue guidance that clarifies how to apply the Equality Act to algorithmic decision-making, which should include guidance on the collection of data to measure bias, and the lawfulness of bias mitigation techniques

Footnotes:

1. Ethnicity Facts and Figures, (2021) Employment by ethnicity. Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/employment/employment/latest#by-ethnicity ↩

2. Ethnicity Facts and Figures, (2021) Unemployment by ethnicity. Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/unemployment-and-economic-inactivity/unemployment/latest ↩

3. ONS (2020) Ethnicity Pay Gaps. Release date 12 October 2020. Available at: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/articles/ethnicitypaygapsingreatbritain/2019 ↩

4. CIPD, (2017), ‘Addressing the barriers to BAME employee career progression to the top’. Survey of 1290 UK employees. Available at: https://www.cipd.co.uk/Images/addressing-the-barriers-to-BAME-employee-career-progression-to-the-top_tcm18-33336.pdf ↩

5. Evidence commissioned by the Commission on Race and Ethnic Disparities, received on 26 October 2020. Norrie, R., et al, D., (2020) ‘An overview of ethnic disparity in the labour market and measures to address it.’ ↩

6. ibid. ↩

7. ONS (2020), A09: Labour market status by ethnic group. Source: Labour Force Survey. Differences calculated from April to June 2001 to April to June 2019. A020). Available at: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/datasets/labourmarketstatusbyethnicgroupa09 ↩

8. ONS (2020), A09: Labour market status by ethnic group. Source: Labour Force Survey. Differences calculated from April to June 2001 to April to June 2019. Available at: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/datasets/labourmarketstatusbyethnicgroupa09 LFS data is available for 2020, but these have not been used due to the impact the COVID-19 pandemic may have on employment rates. The rate for the Bangladeshi group increased by 22.8 percentage points. from 2001 to 2020, compared to an increase of 3.2 percentage points for the White ethnic group. ↩

9. Evidence commissioned by the Commission on Race and Ethnic Disparities, received on 26 October 2020. Norrie, R., Goodhart, D., (2020) ‘An overview of ethnic disparity in the labour market and measures to address it’. ↩

10. Ethnicity Facts and Figures, (2021) Unemployment by ethnicity. Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/unemployment-and-economic-inactivity/unemployment/latest#by-ethnicity ↩

11. ibid. ↩

12. Platt, L., Warwick, R. (2020), ‘Are some ethnic groups more vulnerable to COVID-19 than others?’, Institute for Fiscal Studies. Available at: https://www.ifs.org.uk/inequality/wp-content/uploads/2020/04/Are-some-ethnic-groups-more-vulnerable-to-COVID-19-than-others-V2-IFS-Briefing-Note.pdf ↩

13. Froy, F. and L. Pyne (2011), ‘Ensuring Labour Market Success for Ethnic Minority and Immigrant Youth’, OECD Local Economic and Employment Development (LEED) Working Papers, 2011/09, OECD Publishing.’Available at: http://dx.doi.org/10.1787/5kg8g2l0547b-en ↩

14. Ethnicity Facts and Figures, (2021) Employment by ethnicity and gender. Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/employment/employment/latest#by-ethnicity ↩

15. Ethnicity Facts and Figures, (2021) Employment by ethnicity and age. Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/employment/employment/latest#by-ethnicity-and-age ↩

16. Department for Education (2020) Permanent data table: ‘Free School Meals, Gender and Ethnic Group’ from ‘Widening participation in higher education’, Data covers pupils from English state-funded schools and special schools who have progressed to HE in UK Higher Education Providers (including Alternative Providers) and English Further Education Colleges in the 2018/19 academic year. Available at: https://explore-education-statistics.service.gov.uk/data-tables/permalink/77f3aabf-1e21-4c2f-bb58-b5671c695307 ↩

17. Ethnicity Facts and Figures, (2021) Unemployment by ethnicity. Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/unemployment-and-economic-inactivity/unemployment/latest#by-ethnicity ↩

18. Ethnicity Facts and Figures, (2021) Unemployment by ethnicity and age (White and Other than White). Covers England, Scotland and Wales in the year 2019. Source: Annual population survey. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/unemployment-and-economic-inactivity/unemployment/latest ↩

19. Evidence commissioned by the Commission on Race and Ethnic Disparities, received on 26 October 2020. Norrie, R., Goodhart, D., (2020) ‘An overview of ethnic disparity in the labour market and measures to address it’. Norrie and Goodhart use a 4 wave pooled version of the Labour Force Survey to attain these unemployment rate estimates. ↩

20. ONS (2020) Ethnicity Pay Gaps. Release date 12 October 2020. Available at: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/articles/ethnicitypaygapsingreatbritain/2019 ↩

21. The median hourly pay in England and Wales for White British women was £11.21 in 2019, compared with £12.09 for Black Caribbean women and £12.49 for Arab women. Source: ONS Ethnicity Pay Gap 2020 ↩

22. Pay determining characteristics used by the ONS here are the following: ethnicity; country of birth; occupation; highest qualification level; age; sex; marital status; working pattern; disability status; working in the public or private sector; geography; whether they have children or not. ↩

23. Examples of routine manual jobs include retail assistants, cleaners, van drivers and waiters. ↩

24. Evidence commissioned by the Commission on Race and Ethnic Disparities, received on 25 November 2020. Li, Y.,(2020) Social Progress: The social mobility of ethnic minorities in Britain in the last fifty years (1972-2019) ↩

25. Green Park (2020), The Colour of Power: Report Insights. Published 27th July 2020. Accessed 20th February 2021. Available at: https://www.green-park.co.uk/insights/the-colour-of-power/s191468/ ↩

26. EY, (2021), ‘Significant progress on improving ethnic diversity of FTSE 100 boards reveals new data from the Parker Review’, Press release. Accessed 16 March 2021. Available at: https://www.ey.com/en_uk/news/2021/03/significant-process-on-improving-ethnic-diversity-of-ftse-boards-reveals-new-data ↩

27. Green Park (2021), GREEN PARK BUSINESS LEADERS INDEX

    BRITAIN’S TOP FIRMS FAILING BLACK LEADERS Published 3rd February 20201. Accessed 20th February 2021. Available at: https://www.green-park.co.uk/insights/green-park-business-leaders-index-britain-s-top-firms-failing-black-leaders/s228945/

    ↩

28. According to the Race in the workplace: The McGregor-Smith review, when subjected to CV testing private sector employers showed a discrimination rate of 35% compared with 4% for the public sector. ↩

29. Civil Service statistics: 2020, Published August 2020. Available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/940284/Statistical_bulletin_Civil_Service_Statistics_2020_V2.pdf ↩

30. Civil Service Diversity and Inclusion Dashboard. SCS representation by department and ethnicity. Data for 2019. Available at: https://public.tableau.com/profile/cabinet.office.diversity.and.inclusion#!/vizhome/CivilServiceDiversityandInclusiondashboard/Introduction ↩

31. Civil Service HR (2018) Civil Service Fast Stream: Annual Report 2017 and 2018, Table 24. Available at: [https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/767789/Civil_Service_Fast_Stream_Annual_Report_2017-_2018.pdf](https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/767789/Civil_Service_Fast_Stream_Annual_Report_2017-_2018.pdf) ↩

32. Civil Service Diversity and Inclusion Dashboard. Updated 11th September 2019. Available at: https://www.gov.uk/government/publications/civil-service-diversity-inclusion-dashboard/civil-service-diversity-and-inclusion-dashboard ↩

33. Solicitors Regulation Authority, (2020), ‘How diverse is the legal profession?’. Available at: https://www.sra.org.uk/sra/equality-diversity/key-findings/diverse-legal-profession/ ↩

34. ibid. ↩

35. Chambers Student, (2019), ‘Law firm diversity 2018/19’. Available at: https://www.chambersstudent.co.uk/where-to-start/newsletter/law-firm-diversity-201819 ↩

36. ibid. ↩

37. Queen’s Counsel (QC) are barristers or solicitor advocates who have been recognised for excellence in advocacy. They’re often seen as leaders in their area of law and generally take on more complex cases that require a higher level of legal expertise. (Definition from ‘The Law Society’) ↩

38. Bar Standards Board, (2021), ‘Diversity at the Bar’. Available at: https://www.barstandardsboard.org.uk/uploads/assets/88edd1b1-0edc-4635-9a3dc9497db06972/BSB-Report-on-Diversity-at-the-Bar-2020.pdf ↩

39. Ethnicity facts and figures (2021) NHS workforce by ethnicity and grade (medical staff). Source: NHS Workforce Statistics 2020. Covers England. Data measures those working in trusts and CCGs, and does not include staff working in primary care, like GPs and dentists in dental practices staff working in NHS support organisations, and central bodies like commissioning support units and NHS Digital, or agency staff and contractors. Available at: https://www.ethnicity-facts-figures.service.gov.uk/workforce-and-business/workforce-diversity/nhs-workforce/latest ↩

40. Ethnicity facts and figures (2021) NHS workforce by ethnicity and grade (medical staff). Source: NHS Workforce Statistics 2020. Covers England. Data measures those working in trusts and CCGs, and does not include staff working in primary care, like GPs and dentists in dental practices, staff working in NHS support organisations, and central bodies like commissioning support units and NHS Digital, or agency staff and contractors. Available at: https://www.ethnicity-facts-figures.service.gov.uk/workforce-and-business/workforce-diversity/nhs-workforce/latest ↩

41. ONS (2020) Ethnicity Pay Gaps. Release date 12 October 2020. Available at: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/articles/ethnicitypaygapsingreatbritain/2019 ↩

42. BEIS, (2018), ‘Closed consultation: Ethnicity pay reporting’. Available at: https://www.gov.uk/government/consultations/ethnicity-pay-reporting ↩

43. UK Parliament House of Commons Library (2020) Constituency data: ethnicity. Published 16th November 2020. Internal calculation. Data available at: https://data.parliament.uk/resources/constituencystatistics/Ethnic-group.xlsx ↩

44. NHS Digital, (2019), ‘NHS Workforce Statistics - March 2019 (Including supplementary analysis on pay by ethnicity)’. Available at: https://files.digital.nhs.uk/E1/0FC8B8/Ethnicity%20pay%20gap%20-%20FTE%20Basic%20comparison%20tool%20-%20by%20staff%20group.xlsm ↩

45. ibid. ↩

46. ibid. ↩

47. ibid. ↩

48. Surash, S. & Pearce K. (2019), ‘The Surash Pearce report A comprehensive review into ethnic pay-gap and workforce development at The Newcastle upon Tyne Hospitals NHS Foundation Trust’. Available at: https://www.damejackiesblog.co.uk/wp-content/uploads/2020/09/Surash-Pearce-Report-2019_low_res.pdf ↩

49. A typical full-time consultant’s job plan will have 10 programmed activities. Additional programme activity (PA) can be paid to those undertaking additional clinical/service commitments. Please see full report for details ↩

50. Clinical excellence awards (CEA) are awarded locally and nationally for clinical excellence. Please see full report for details ↩

51. Occupational preferences research: This study was conducted by MindGym ↩

52. Di Stasio, V., and Heath, A., (2019), ‘Are employers in Britain discriminating against ethnic minorities?’, GEMM Project, available at: http://csi.nuff.ox.ac.uk/wp-content/uploads/2019/01/Are-employers-in-Britain-discriminating-against-ethnic-minorities_final.pdf ↩

53. Wood, M., Hales, J., Purdon, S., Sejersen, T., and Hayllar, O., (2009), ‘A test for racial discrimination in recruitment practice in British cities’, available at: https://natcen.ac.uk/media/20541/test-for-racial-discrimination.pdf ↩

54. Darolia, R., et al., (2016), ‘Race and gender effects on employer interest in job applicants: new evidence from a resume field experiment.’ Applied Economics Letters 23(12): 853-856; Neumark, D. and J. Rich (2019). ‘Do field experiments on labor and housing markets overstate discrimination? A re-examination of the evidence.’ ILR Review 72(1): 223-252. ↩

55. CIPD, (2017), ‘Addressing the barriers to BAME employee career progression to the top’. Survey of 1290 UK employees, available at: https://www.cipd.co.uk/Images/addressing-the-barriers-to-BAME-employee-career-progression-to-the-top_tcm18-33336.pdf ↩

56. Confederation of British Industry, (2020),Bridge the gap: Actions your business can take to close the ethnicity pay gap, available at: https://www.cbi.org.uk/media/4931/12567_epg_guide.pdf ↩

57. Evidence submitted to the Commission on Race and Ethnic Disparities call for evidence, Employment Lawyers Association (ELA) ↩

58. ‘The Ethnicity Gap’, published by inclusion 360. Available at: https://www.inclusion360.co.uk/our-advice/the-ethnicity-gap ↩

59. Parker Review (2020) ‘Ethnic Diversity Enriching Business Leadership: an update report from The Parker Review’. Available at: https://assets.ey.com/content/dam/ey-sites/ey-com/en_uk/news/2020/02/ey-parker-review-2020-report-final.pdf ↩

60. Robert Half, (2019), ‘The Robert Half FTSE 100 CEO Tracker’. Available at: https://www.roberthalf.co.uk/reports-guides/ftse-100-ceo-tracker. Accessed 24th March 2021 ↩

61. Equalities and Human Rights Commision (2016) ‘Inquiry into fairness, transparency and diversity in FTSE350 board appointments.’ Available at: https://www.equalityhumanrights.com/en/publication-download/inquiry-fairness-transparency-and-diversity-ftse-350-board-appointments ↩

62. Iyer,P., Shields, J., Elliott, C., and Gill, V., (2021), ‘Understanding young people’s experiences of employment Qualitative consultations with working-class Black, South Asian and White young people in England FINAL REPORT’, NatCen Social Research, Prepared for the Commission for Race and Ethnic Disparities. ↩

63. Dobbin, F., and Kalev, A., (2016), ‘Why Diversity Programs Fail And what works better’, Harvard Business Review. Available at: https://hbr.org/2016/07/why-diversity-programs-fail ↩

64. Her Majesty’s Inspectorate of Probation, (2021), ‘Experiences of Black, Asian and minority ethic service users on probation: A report summarising service user perspectives’. Available at: https://www.justiceinspectorates.gov.uk/hmiprobation/inspections/race-equality-in-probation/ ↩

65. Atewologun. D, Cornish.T, et al, (2018), Unconscious bias training: An assessment of the evidence for effectiveness, Equality and Human Rights Commission, Available at: https://www.ucd.ie/equality/t4media/ub_an_assessment_of_evidence_for_effectiveness.pdf ↩

66. ]Al-Gharbi. M., (2020), ’Diversity is Important. Diversity-Related Training is Terrible’. Available at: https://musaalgharbi.com/2020/09/16/diversity-important-related-training-terrible/3/ ↩

67. Bristol.gov.uk, Stepping Up programme. Available at: https://www.bristol.gov.uk/mayor/stepping-up-programme ↩

68. In Diverse Company, Cultural Inclusion Maturity Model. Available at: https://indiversecompany.com/inclusion-maturity-model/ ↩

69. Cabinet Office, (2020), Green Paper: Transforming public procurement. Available at: https://www.gov.uk/government/consultations/green-paper-transforming-public-procurement ↩

70. NHS England, (2020), ‘We Are The NHS. People Plan 2020/21 - Action for all of us’ Available at: https://www.england.nhs.uk/wp-content/uploads/2020/07/We-Are-The-NHS-Action-For-All-Of-Us-FINAL-March-21.pdf ↩

71. NHS, (2020), ‘NHS workforce race and equality standard: 2019 data analysis report for NHS trusts’. Available at: https://www.england.nhs.uk/wp-content/uploads/2020/01/wres-2019-data-report.pdf ↩

72. ibid. ↩

73. ibid. ↩

74. The report does not offer a response rate for White staff ↩

75. The Kings Fund, (2020), Workforce race inequalities and inclusion in NHS providers. Available at: https://www.kingsfund.org.uk/publications/workforce-race-inequalities-inclusion-nhs ↩

76. The Global Entrepreneurship and Development Institute, (2019), ‘Global Entrepreneurship Index’. Available at: https://thegedi.org/global-entrepreneurship-and-development-index/ ↩

77. Federation of Small Businesses, (2020), ‘Unlocking opportunity: The value of ethnic minority firms to UK economic activity and enterprise’. Available at: https://www.fsb.org.uk/resource-report/unlock.html ↩

78. ibid. ↩

79. British Business Bank, (2020), ‘Alone together: Entrepreneurship and diversity in the UK’. Available at: https://www.british-business-bank.co.uk/research-alone-together/ ↩

80. ibid. ↩

81. Office for National Statistics, (2013), ‘2011 Census: Key Statistics and Quick Statistics for local authorities in the United Kingdom - Part 1, Table KS201UK’, accessed 15 February 2021. Available at: https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationestimates/datasets/2011censuskeystatisticsandquickstatisticsforlocalauthoritiesintheunitedkingdompart1 ↩

82. British Business Bank, (2020), ‘Alone together: Entrepreneurship and diversity in the UK’. Available at: https://www.british-business-bank.co.uk/research-alone-together/ ↩

83. ibid. ↩

84. Ethnicity facts and figures, (2020), ‘Employment by occupation’. Available at: https://www.ethnicity-facts-figures.service.gov.uk/work-pay-and-benefits/employment/employment-by-occupation/latest#by-ethnicity-and-type-of-occupation ↩

85. Research conducted by Extend Venture analysed data on venture capital investments into companies founded and funded between 2009 and 2019 to provide insights on the landscape of UK venture capital funding across ethnic groups and other demographic factors within the last 10 years. ↩

86. Top ranking universities were classed as Oxford, Cambridge, Harvard, Stanford and their respective business schools. ↩

87. Carter S., Mwaura S., Ram M., Trehan K., Jones T., (2015) ‘Barriers to ethnic minority and women’s enterprise: Existing evidence, policy tensions and unsettled questions’ International Small Business Journal. 2015;33(1):49-69. Available at: https://journals.sagepub.com/doi/full/10.1177/0266242614556823 ↩

88. Research shows this is predominantly due to businesses’ credit and financial characteristics being at odds with the market standard approaches that are used to make credit decisions, when applied equally to all firms. ↩

89. British Business Bank, (2020), ‘Alone together: Entrepreneurship and diversity in the UK’. Available at: https://www.british-business-bank.co.uk/research-alone-together/ ↩

90. In 2019, 20% of start-up loans were awarded to people who were ethnic minorities (not including White minorities) according to British Business Bank data. ↩

91. Centre for Entrepreneurs, (2017), ‘Putting the uni in unicorn: The role of universities in supporting high-growth graduate startups’. Available at: https://centreforentrepreneurs.org/wp-content/uploads/2017/08/CFE-University-Entrepreneurs-Report-WEB.pdf ↩

92. Dias, Joyce, et al., (2020), ‘COVID-19 and the career prospects of young people, Institute for Fiscal Studies’. Available at: https://www.ifs.org.uk/publications/14914 ↩

93. Centre for Entrepreneurs, (2017), ‘Putting the uni in unicorn: The role of universities in supporting high-growth graduate startups’. Available at: https://centreforentrepreneurs.org/wp-content/uploads/2017/08/CFE-University-Entrepreneurs-Report-WEB.pdf ↩

94. The elite universities category represents Oxford and Cambridge ↩

95. Tier 1 represents the next 10 universities in the UK and the next 10 business schools ↩

96. Extend Venture, (2020), ‘Diversity beyond gender’. Available at: https://www.extend.vc/reports ↩

97. Aston University, Coventry University and Manchester Metropolitan University are among the 33 business schools across the UK that have been awarded the Small Business Charter in recognition of their ongoing work to support small businesses, local economies and student entrepreneurship. ↩

98. British Business Bank, (2020), ‘Alone, together: Entrepreneurship and diversity in the UK’. Available at: https://www.british-business-bank.co.uk/research-alone-together/ ↩

99. ibid. ↩

100. HSBC UK, (2021), ‘HSBC UK launches £15BN SME Fund to support British businesses’. Available at: https://www.about.hsbc.co.uk/news-and-media/hsbc-uk-launches-15bn-sme-fund-to-support-british-businesses-to-grow-again ↩

101. Mehrabi.N, Morsatter. F, et al., (2019), ‘A survey on bias and fairness in machine learning’. Available at: arXiv preprint arXiv:1908.09635 ↩

102. Reisman. D, Schultz. J, et al., (2018), ‘Algorithmic impact assessments: A practical framework for public agency accountability’. Available at: https://ainowinstitute.org/aiareport2018.pdf ↩

103. Mehrabi.N, Morsatter. F, et al., (2019), ‘A survey on bias and fairness in machine learning’ Available at: arXiv preprint arXiv:1908.09635 ↩

104. ACM Conference on Fairness, Accountability, and Transparency (ACM FAccT) is the premier academic conference on Fairness, Accountability, and Transparency. ↩

105. Centre for Data, Ethics and Innovation, (2020), ‘Review into bias in algorithmic decision-making’. Available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/957259/Review_into_bias_in_algorithmic_decision-making.pdf ↩