The UK Pay Transparency Regulations: apparent transparency without accountability?

The UK enacted its first legal measure to address gender pay inequity, the Equal Pay Act 1970, more than 50 years ago. Yet, in 2021, the gender pay gap (GPG) still stood at 15.4%. Departing from the remedial and individual approach that characterises equal pay legislation, the 2017 Gender Pay Gap Information Regulations (the Regulations) require private and voluntary sector organisations with 250+ employees to annually publish pay data broken down by gender. The long-term aspiration of the Regulations is to contribute to closing the GPG within a generation. It is also hoped that they will encourage the public disclosure of pay data and changes in workplace policies to reduce organisational GPGs (immediate aims) and improve employers ’ accountability (underlying aim). This paper considers whether the Regulations have what it takes to meet those immediate and underlying aims. Our assessment framework is built on the premise that for public disclosure to be useful and for employers to tackle the causes of the GPG, the information reported must be of sufficient quality, meaningful and relevant. The paper draws on both doctrinal analysis and empirical data reported by FTSE 100 Index companies to assess the Regulations and determine whether they hold the potential to meet those aims.


Introduction
The UK enacted its first legal measure to address gender pay inequity, the Equal Pay Act 1970 (the 1970 Act), more than 50 years ago. While the gender pay gap (GPG) fell by almost 10 percentage points in the 20 years from 1999 (26.9%) to 2019 (17.4%), 1 in 2021, it still stood at 15.4%. 2 Various UK governments have expressed the ambition of closing the GPG within a generation, 3 but without drastic changes, that aspiration seemed unlikely to be realised. The enactment of the Gender Pay Gap Information † The authors thank Miguel Flores, Mark Bell, and the two anonymous reviewers for their valuable comments on earlier drafts of this paper (the usual disclaimer applies). This project was funded by the University of Southampton, Stefan Cross Centre for Women, Equality and the Law (2018-19) and UCD, Sutherland School of Law (2020-21). 1 Regulations (the Regulations) in 2017 4 was presented as a key tool in the UK's wider strategy to close the GPG within a generation, 5 but do the Regulations really hold that potential?
Apart from some provisions establishing collective mechanisms to address pay inequalities, 6 the 1970 Act placed the burden to challenge pay differences on individual employees through equal pay claims. The 1970 Act was later amended to accord with EU law, 7 but the strong emphasis on formal equality and individual enforcement remained. 8 While women progressively increased their participation in the labour market and combined motherhood with paid employment, 9 the 1970 Act's reactive enforcement model proved insufficient to tackle the complex and interrelated causes of gender pay inequity. 10 The 2000 Equality Law Review showed that the formal and remedial approach that characterised the 1970 Act was a broader issue that permeated the fragmented and inconsistent UK legal framework. 11 In the field of gender equality, the first 'shift from a dispute-resolution model to a structural reform model' 12 came about with the introduction of the Gender Equality Duty (GED) in 2007, 13 almost at the same time as the Equality and Human Rights Commission (EHRC) was created as Britain's multi-mandate equality watchdog. 14 The GED was considered 'the biggest advance in women's equality since the 1970's' 15 because it placed a duty on all public authorities to have 'due regard to the need to eliminate unlawful discrimination on grounds of gender and to promote equality of opportunity', 16 including pay equity. Considering that the GED applied, inter alia, in the field of employment and public procurement, it had some potential to encourage the structural changes that individual litigation could not bring about. 17 A more profound paradigm change came with the adoption of the Equality Act (EqA 2010) in April 2010. The EqA 2010 not only codified and systematised UK equality law (repealing and replacing, inter alia, the 1970 Act 18 ), it also marked a shift towards a 'transformative equality' model, 19 more strongly grounded on substantive equality. 20 As part of that shift, the EqA 2010 replaced the GED 21 with a horizontal Public Sector Equality Duty (PSED) 22 and introduced in sections 78(1) and 78(2)(a) a power to issue regulations requiring employers to publish prescribed metrics  15 EOC Gender Equality Duty Code of Practice for England and Wales (Equal Opportunities Commission, 2006). 16 Emphasis added. 17 For a critique, see Fredman, above n 8, at 212-215. 18 B Hepple Equality. The Legal Framework, 2 nd edn (Oxford: Hart Publishing, 2014) ch 1. 19 Ibid, p 15. 20 On transformative equality as a form of substantive equality see S Fredman Discrimination Law, 2 nd edn (Oxford: Oxford University Press, 2011) pp 30-31. 21 The PSED also replaced the other race and disability duties: see Hepple, above n 18, p 163. 22 This was complemented by a socio-economic duty (EqA 2010, s 1) and by specific duties in regulations in England, Wales and Scotland: see A Blackham 'Positive equality duties: the future of equality and transparency?' (2021) Law in Context 98 at 105-108. about employees' pay information to reveal differences in pay between men and women. The combination of these proactive measures seemed likely to address gender inequity more effectively than prior legislation.
However, shortly after the enactment of the EqA 2010, the Coalition Government came into office and the UK suffered the worst economic crisis in decades. Politically, this moved the focus towards deregulation, thus making several innovative EqA 2010 provisions, including the PSED, 23 appear as 'red-tape' for businesses. 24 Many provisions were repealed or put on hold, 25 and the PSED was put under review. 26 Economically, the 2010 emergency budget kick started a new 'austerity era' with widespread adjustments in public services and welfare provision that had a disproportionate negative impact on women. 27 As part of these adjustments, the EHRC's budget was drastically cut, 28 which severely hampered its enforcement work. 29 Unsurprisingly, in this context, section 78 pay transparency regulations were not implemented. Instead, a voluntary reporting initiative called 'Think, Act, Report' was introduced, 30 largely relying on the business case for diversity as a way to 'nudge' 31 employers to advance equality.
The 'Think, Act, Report' initiative was part of the Coalition's emergent 'market fundamentalism' 32 approach. Aimed at organisations with at least 150 employees, it was not exclusively about reporting GPG data. Instead, it sought more broadly to promote gender pay equity. Almost 300 employers signed onto the scheme, but only 3% were known to have published gender pay information. 33 This failure was predictable given that voluntary equal pay audits based on the business case for diversity had already been introduced in the 2000s without much success. 34 After the 'Think, Act, Report' fiasco and an intense lobbying campaign, 35  Generally, pay transparency measures may be divided into two broad categories: those that introduce proactive or collective measures (eg establishing duties for employers to report, analyse and/or act to address their internal GPG) 38 and those seeking to improve individual access to pay information and facilitate remedial action (eg allowing the right to request salary information that could be used in a potential equal pay claim). 39 The UK Regulations fall within the first category because they shift the burden from the employee to the employer to uncover pay inequity: 40 it is the employer who is obliged to disclose data that may reveal the existence of a GPG.
The Regulations require private and voluntary sector organisations with 250+ employees to annually publish their employees' basic pay and bonus pay data broken down by gender. 41 The only legal obligation they introduced is reporting information on various basic pay metrics; neither analysis nor action is required. Despite failed previous attempts to rely on the business case to prompt voluntary action to address the GPG, the Government expected that, through this reporting exercise, organisations would become aware of their structural gender biases which wouldhopefullyencourage them to voluntarily take relevant action to address the problems identified. 42 Against this background, we approach the study of the Regulations through an evaluative lens: we seek to establish whether, as a legal construct, they can attain their goals. 43 The Regulations' immediate aims are to improve pay transparency through public disclosure and encourage organisational changes that may contribute to lowering the GPG. The underlying aim of such public disclosure seems to be improving employers' accountability. In addition, the Regulations' long-term aspiration is to contribute to closing the GPG in the UK within a generation. 44 Considering that it is still too early to assess the Regulations against this latter purpose, we analyse whether the Regulations are capable of achieving their immediate and underlying aims (Tables 1 and 2).
We focus mainly on compulsory reporting. 45 We evaluate the Regulations' scope and the information employers must report using an assessment framework built around two conditions inspired by Fox's theory on transparency and accountability, 46 ie the need for: (1) information quality; and (2) 'clear' transparency to generate constructive change and some degree of accountability (see Section 2 and Table 3).
The paper proceeds as follows. Section 1 explains the methodology. Section 2 discusses the purpose and aims of the Regulations and the assessment framework to evaluate the latter against their aims. Section 3 analyses the Regulations' scope and reporting requirements in light of the assessment framework to establish whether they hold the potential to meet those aims. A final section concludes.

Methodology
The legal literature discussing the UK Pay Transparency Regulations has so far mostly consisted of doctrinal legal research with either a primarily theoretical 47 or practitioner focus. 48  This paper provides an original contribution in taking an evaluative doctrinal approach that is complemented with socio-legal analysis. The latter is based on the quantitative and qualitative data submitted by in-scope employers. 50 Relying on academic and grey literature, the doctrinal part of the paper analyses the potential of the Regulations to achieve their aims. The socio-legal approach puts the Regulations into the socio-economic context in which they exist and are implemented. 51 More specifically, data reported by companies listed in the FTSE 100 Index is used to illustrate the arguments based on the doctrinal analysis.
We focus on FTSE 100 companies because they are the most likely to comply with the Regulations for various reasons. First, media exposure and large firm size are factors associated with better corporate social responsibility (CSR) disclosure, such as GPG reporting. 52 FTSE 100 companies tend to be large (250+ employees) and subject to more public scrutiny, thus making them more prepared to comply with pay transparency obligations 53 and more concerned about reputational damage due to non-compliance. 54 Secondly, there is evidence that listed companies have more (soft) regulatory 55 and financial incentives to show good performance regarding diversity and social governance issues. 56 Within FTSE 100 companies, we chose to focus on the 'worst performing companies' because there is increasing evidence that companies are concerned about media, employee, consumer and shareholder scrutiny about their environmental, social and governance (ESG) disclosures, including gender diversity. 57 Huang and Lu show that among FTSE 350 companies, the firms that reported the worst GPG data were more likely to disclose more information (including quantitative targets and actions to reduce their GPGs) to improve their reputation and gain better ESG ratings. 58 On that basis, we presume that the worst performing companies in our sample had incentives to go beyond the basic Regulations' requirements, ie to disclose better quality information, to explain their data (eg through voluntary narratives) and to take action to address their GPGs.
We collected qualitative and quantitative data of the 113 companies listed in the FTSE 100 Index for four reporting years (2017-18, 2018-19, 2019-20, 2020-21). The quantitative data were downloaded from the UK Government website where all reports must be submitted. 59 The qualitative data (ie supporting narratives that employers voluntarily publish to explain the data and potential follow up measures and/or action plans) were collected from each individual company's website. Out of 113 companies listed in the FTSE 100 Index between April 2017 and March 2021, a total of 86 were considered; the rest were excluded for having fewer than 250 employees. Quantitative metrics and qualitative data were analysed using, respectively, statistical and content analysis.  For instance, the target of 33% of women on FTSE 350 boards by 2020 (Hampton-Alexander Review. FTSE Women Leaders, November 2016, https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/ 613085/ftse-women-leaders-hampton-alexander-review.pdf (accessed 17 April 2023)) and the UK Corporate Governance Code (2018)

Legal Studies
To select the worse performing companies within our sample, we applied the statistical empirical rule of one standard deviation of the mean, which amounts to 37.1%. 60 We used that cut-off point to identify the subsample of companies whose GPG is above 37.1%, 15 in total. The subsample data is listed in Appendices 2 and 3 for reporting years 2017-18 and 2020-21 respectively. 61 Appendix 1 also shows the mean GPG of the sample and subsample companies, and provides additional data about the subsample according to size and sector.
We use the qualitative and quantitative data of the subsample to: (1) illustrate our evaluation of the data reporting requirements; and (2) check the extent to which these companies went beyond the mandatory reporting requirements to improve the information disclosed. The analysis of the subsample data is based on employers' reported data and voluntary commentary on the latter. For quantitative data, our assessment is based on descriptive statistics because we want to explore if the information disclosed is useful as reported for non-experts (without technical knowledge). 62 2. The Regulations: purpose and assessment framework The Regulations are a form of 'targeted transparency'. 63 As such, they establish disclosure obligations and have a regulatory purpose. 64 This section establishes a framework to assess the Regulations' potential to fulfil their aims.
The long-term purpose of the Regulations is contributing to closing the GPG within a generation. 65 However, the Regulations cannot be expected to be the panacea which will eliminate the GPG from the UK labour market. Indeed, they were presented as one tool among 'a range of measures to tackle the drivers of the pay gap'. 66 Regardless, closing the GPG within a generation seems more aspirational than tangible, and it would be too early and unrealistic for this work to assess if the Regulations can contribute to that ambition.
Considering the Regulations' impact assessment and explanatory memorandum, they appear to have two more immediate aims (Table 1). The primary aim seems to be 'deliver[ing] transparency'. 67 The Government hoped that the Regulations would lead to more 'openness' 68 and would raise awareness about the GPG among employers (regarding their own GPGs), 69 and among the public to create accountability and pressure for social change. 70 Accordingly, the aim of delivering transparency can be interpreted as ensuring public disclosure of internal information both inside and outside the workplace. 71 60 This figure results from calculating the sample GPG (hourly rate) mean (21.98%) and the standard deviation (15.16%) for 2017-18, then adding them and rounding to one digit (21.98 + 15.16 = 37.1). 61 The reporting obligations were suspended in 2019/20 due to Covid-19, which is why we focus on the evolution from the first to the fourth reporting year. GEO 'Employers do not have to report gender pay gaps' (24 March 2020), https://www.gov. uk/government/news/employers-do-not-have-to-report-gender-pay-gaps (accessed 17 April 2023). 62 According to Fox, this is a crucial feature of 'clear transparency': Fox, above n 46, at 667 (see also subsection 3(d)). The second aim of the Regulations seems to be 'encouraging employers to analyse the drivers behind their GPG and explore the extent to which their own workplace policies and practices may have contributed to that gap, as opposed to other factors outside of their control'. 72 Thus, the Government anticipated that the Regulations would generate accountability which would push employers to take action to address their GPG, even though taking such action is not mandated by the Regulations.
The narrative around these two immediate aims repeatedly returns to the idea of increasing employers' accountability to encourage (voluntary) action to address the GPG. Indeed, this seems to be the underlying aim of the Regulations, which is line with the 'conventional wisdom' that 'transparency generates accountability'. 73 However, 'the power of transparency is based on the "power of shame"', so its influence over 'the shameless' is limited. 74 For this reason, transparency is necessary but not sufficient to generate accountability. 75 Fox differentiates between soft and hard accountability. 'Soft accountability' can be identified with 'answerability', that is, 'the capacity to demand explanations' and 'to call those in authority to justify their decisions'. 76 However, answerability has limited effects if not supported by the capacity to sanction or compensate; it is this latter capacity that turns soft accountability into 'hard accountability'. 77 Considering the lack of civil or administrative sanctions for non-compliance with the Regulations and the EHRC's limited enforcement action so far, 78 this paper only considers whether the Regulations meet the conditions to generate soft accountability. In other words, we focus on the analysis of the scope and disclosure duties of the Regulations to assess if they can generate 'answerability' in practice (Table 2).
Building on Fox's theory on the relationship between transparency and accountability, we argue that there are at least two conditions for the Regulations to produce soft accountability: (I) information quality; and (II) clear transparency (Table 2). First, ensuring the quality of the information disclosed is crucial: if it does not meet certain minimum standards it will be of little use to key stakeholders. If that  is the case, in the absence of effective quality control mechanisms, what is in theory a 'mandatory disclosure' will be 'less mandatory in practice'. 79 Secondly, the data disclosed should 'reveal reliable information about institutional performance' and shed 'light on institutional behaviour, which permits interested parties […] to pursue strategies of constructive change'. 80 This is what Fox calls 'clear transparency'. By contrast, transparency duties that only require 'the dissemination of information that does not reveal how institutions actually behave in practice' and/or 'that is divulged only nominally, or […] turns out to be unreliable' are forms of 'opaque transparency'. 81 Obviously, clear transparency is needed to ensure that the Regulations' data disclosure duties lead to the publication of information that is meaningful and relevant to identify the causes of the GPG. Table 3 summarises the assessment criteria considered necessary to generate the conditions for information quality and clear transparency and, thus, for the Regulations to bring about soft accountability.
Let us briefly explain each of these assessment criteria. Criterion A, 'comprehensive coverage' refers to being 'inclusive of' all or many items or information. 82 In this context, comprehensive coverage refers to including as many employers and workers as possible in the scope of the Regulations. Of course, comprehensive coverage should not undermine the sustainability, or even growth, of the reporting duty over time. 83 'Comprehensive' coverage also demands a broad personal scope, requiring the inclusion of pay data from as many workers as possible in the reporting duty, and not excluding certain workers due to, eg, their employment status or working pattern.
Criterion B requires that the data submitted through the reporting duty is 'broadly comparable in quality, detail and vocabulary'. 84 This facilitates benchmarking between different organisations, which enables the public to compare the performance of different employers and potentially act on the basis of that information. 85 Comparability is also crucial within organisations to facilitate longitudinal analysis so that employers can monitor the evolution of their pay structures and the possible effects of their actions to address the GPG.
Criterion C demands that the information has integrity. 86 This requires accuracy (ie that the data are correct and authentic), as well as honesty, precision, completeness and consistency throughout the reporting process. 87 The data should be reliable and reflect the organisation's reality. Finally, Criterion D entails that the data reported is meaningful and relevant to identify the causes of the GPG. This is essential so that the data can be used to promote change (ie to tackle GPG drivers within a given organisation) but is only possible if the information is detailed enough.
These four criteria form the assessment framework by which the Regulations' immediate and underlying aims are evaluated in the following section.

Can the Regulations generate soft accountability?
To establish whether the Regulations are capable of meeting their immediate and underlying aims, we apply assessment criteria A to D to the scope of the Regulations and the reporting requirements. Both aspects are key to establishing the Regulations' potential to generate soft accountability because they directly affect the quality of the information reported and the extent to which it can help identify organisational GPGs.
The personal, material and temporal scope is explained and discussed in each subsection, where relevant for each assessment criterion. The Regulations require private (including publicly-traded) and voluntary sector organisations with 250+ 'relevant employees' at the 'snapshot date' 88 to annually publish their employees' pay data broken down by gender. 89 An overview of all measures and metrics is presented in Table 4.
In our analysis, we use 2017/18 and 2020/21 data from the subsample to illustrate the limits of the reporting requirements. 90

(a) Comprehensive coverage
Given that only in-scope employers ('relevant employers') must report pay data, comprehensive coverage requires a relatively low reporting threshold so that as many employers as possible are bound by the Regulations. 'Relevant employers' are organisations with '250 or more employees on the snapshot  92 It is also a high bar compared to the 'Think, Act, Report' initiative, which was aimed at organisations with 150+ employees. In other jurisdictions, GPG reporting thresholds are considerably lower: eg 10+ employees in Sweden, 25+ employees in Iceland, 35+ employees (with at least 10 employees of each gender per job category) in Denmark, 50+ employees in Spain and 100+ employees in Switzerland. 93 The breadth of the Regulations' coverage is also partly undermined by its personal scope. Only 'full-pay relevant employees' at the snapshot date must be considered in the data reported. The Regulations adopt the broad definition of 'employee' found in the EqA 2010, 94 which includes those employed under a contract of employment, a contract of apprenticeship or a contract personally to do work. 95 This definition is wider than that under other employment legislation, such as the Employment Rights Act 1996, which defines those under a contract personally to do work as 'workers' rather than 'employees'. 96 In theory, therefore, a broader range of people, including some atypical and gig economy workers, fall within the scope of the Regulations. 97 This is an advantage, given the increasing evidence of significant GPGs among these groups. 98 Nevertheless, the definition of 'employee' adopted by the Regulations still excludes those who are self-employed. As is well known, employers have often tried to argue that certain workers (particularly in the gig economy) are selfemployed. 99 Hence, many of the most vulnerable workers, including those with 'questionable selfemployed status', are likely to be excluded from employers' GPG reports in practice. 100 Nevertheless, the Regulations' personal scope also includes other forms of atypical work, namely part-time workers and job-sharers, who must be counted in the headcount as individual employees (not as full-time equivalents). This is commendable because '[ j]obs with higher proportions of parttime employees have lower earnings on average'. 101 Considering that women are more likely to work part-time, and that the Regulations require mean and median pay calculations to be based on the hourly rate of pay, including part-time and job-sharing employees in the headcount can give a more realistic picture of an employer's GPG, especially for organisations with large proportions of part-time and/or job-sharing employees. 102 This was the case for some of our subsample companies (ie #2, #6, #8 and #10). In the narratives that accompanied their data, these companies emphasised 91 Regulation 1(2). 92 This figure includes employees, owners and partners: see BEIS 'Business population estimates for the UK and regions 2021, Table 1', https://www.gov.uk/government/statistics/business-population-estimates-2021 (accessed 17 April 2023). 93 A Nyberg 'Pay surveys in Sweden'; HK Thorgeirsdóttir, 'Equal pay certificationhow Iceland became the first country to require proof of equal pay' 8; E González Gago and G de Cabo Serrano 'New advancements in gender pay gap in Spain' 3 (https:// ec.europa.eu/info/publications/eu-mutual-learning-programme-gender-equality-equal-pay-iceland-27-28-may-2019_en (accessed 17 April 2023)); S Rousselle-Ruffieux 'Equal pay principlenew measures to reduce the gender pay gapthe example of Switzerland' and 'Comments from Denmark' (2019)   Employees who work part-time receive their bonuses on a 'pro rata' basis, but the calculation for the gender bonus gap does not allow any adjustment to bring these bonuses back to their 'fulltime equivalent' level. While we encourage both men and women to work flexibly, the majority of those currently doing so are women. [Company #8, 2017-18] The metrics do not require the breakdown of data for part-time employees 104 so (only) two employers (#6 and #10) gave the actual proportions of women and men working part-time. Consequently, this wide personal scope does not help to explain the extent to which part-time workers contributed to an organisation's GPG.
The Regulation's scope has at least three additional features that limit their comprehensiveness. First, pay data must only be reported for 'full pay' relevant employees. This excludes employees receiving reduced pay during the relevant pay period, eg due to taking family leave (like maternity or paternity leave), sick leave or special leave. 105 This exclusion omits the effects of parenthood on earnings for all those parents (often women) who do not benefit from their employer's top up of the meagre statutory pay for maternity, paternity or adoption leave. 106 Secondly, while agency workers fall within the definition of 'employee', their pay data must only be reported by the agency through which they are hirednot by the organisation for whom they perform work (the 'hirer'). 107 Yet, agency workers are often paid less than 'employees' performing the same jobs, 108 so excluding them from the hirer's reporting duty gives an incomplete picture of the latter's pay structure and can partly mask its GPG. 109 Thirdly, partners are not considered 'relevant employees'. 110 Although they must be accounted for in the employee headcount, 111 they are excluded from pay/bonus calculations. 112 From a formalist perspective, this exclusion may seem justified becauseunlike employeesequity partners are not 'paid' a wage; instead they take part in the partnership profits. However, law and accountancy firms' partners tend to be the most senior (and thus the highest earners) within their organisations, and women are still largely underrepresented among them. 113 Thus, not reporting partners' earnings can yield misleading and unrealistic data. 114 To avoid this, the Regulations could have been designed to require typical" work tasks and working-time arrangements affect the gender wage gap among higher education graduates' (2016) 30 (5) Work, Employment and Society 802. 103 See subsection 3(d). 104 See subsection 3(d). 105 Regulation 1(2). If family leave is paid at the full rate, it should be included. However, under the UK Agency Workers Regulations (reg 5) agency workers are only entitled to the same employment conditions as employees where they have worked in the same role for the same hirer for 12 calendar weeks. Thus prior to reaching the 12-week qualifying period, agency workers may be paid less than employees performing the same job for the hirer. 109 The exclusion of agency workers is also considered in subsection 3(b). The Law Society has recommended that solicitors' firms report partner data: Gender Pay Gap Reporting: Setting the Standard for the Profession (Law Society, 2018) p 3. See also BEIS Committee, above n 42, para 46. the reporting of executive pay data. 115 In fact, the 'Big Four' accountancy firms voluntarily included partners' incomes in their 2018 reported data after criticism for not doing so. 116 Overall, despite the fairly broad concept of 'relevant employee', the Regulations have a number of features that suggest that they do not provide comprehensive coverage. The high reporting threshold, the focus on 'full pay' relevant employees only, and the exclusion of agency workers (for the hirer) and partners are particularly problematic.

(b) Comparable information
As noted in Section 2, comparability includes both comparability within organisations (internal comparability) and between organisations (external comparability). The fact that the four measures (and the more detailed metrics) apply to all in-scope employers and that they have remained the same since the adoption of the Regulations facilitates both internal and external comparability. The Gender Pay Gap Service website enables both internal and external comparisons. The website displays all pay reports submitted for different years by a given company. Further, the 'add to compare' function allows users to view and download a comparative table with the data of multiple selected companies.
These features have enormously improved transparency and comparability, as anyone can freely access the data of in-scope employers, for any reporting year. However, some aspects of the Regulations' personal and temporal scope can arguably undermine internal and external comparability.
Regarding the personal scope, 'relevant employers' must report pay metrics independently for each legal entity above the 250+ threshold within a consolidated group. 117 This increases reporting complexity for organisations with intricate operating models, as they are only required to publish GPG metrics for each in-scope legal entity, not for the entire group. 118 As a result, many large companies reporting their GPG at legal entity level do not voluntarily report consolidated group figures. This feature also makes comparisons across companies more difficult, as the figures presented depend significantly on corporate structure. An employee may, for example, reside within one payroll within a legal entity not deemed 'relevant' (and which therefore does not report GPG data), even though they are employed to fulfil a role in the larger corporate structure beyond the legal entity.
Additionally, the exclusion of agency workers from hirer pay reports allows employers to disguise their GPG data through the use of agency workers (intentionally or otherwise). For example, in the education sector, some organisations employ teachers as standard 'employees', but hire others through temporary work agencies. 119 Because agency workers are typically paid less than employees doing the same work, 120 an employer can benefit from cheaper labour while creating the impression through their GPG data that they outperform competitors in relation to pay equality. 115 For instance, Deloitte's 2019 GPG Report notes that although 'the gender pay gap methodology excludes equity partners … we have instead sought to provide one mean and one median calculation for the whole firm, including equity partner earnings'. In assessing comparability, the temporal scope of the Regulations must also be considered. There are three different relevant periods: 'the snapshot date', 'the pay period' and 'the relevant pay period'. The 'snapshot date' is 5 April of the reporting year. 121 This is the date when a 'photograph' of the organisation's employees' data is taken, and using these data, the employer must calculate the metrics required by the Regulations and report them by 4 April of the following year. For instance, the data of relevant employers on 5 April 2020 will be used to report by 4 April 2021 deadline. 122 The 'pay period' is the period that is used by the employer to pay employees (eg on a weekly or monthly basis, or other). 123 Depending on the sector and nature of the work, it may be that no regular pay period is used. In such instances, the pay period will be 'the period in respect of which the employer most frequently pays the employee one of the elements of ordinary pay'. 124 On this basis, the 'relevant pay period' is the 'pay period within which the snapshot date falls'. 125 For instance, if the employer pays employees on a monthly basis, the relevant pay period for reporting year 2021 will be April 2020 (because it includes the snapshot date, ie 5 April 2020).
This temporal scope makes it relatively simple for employers to collect the data needed to calculate the metrics. To some extent, it also facilitates external comparability between employers which may have different pay periods (eg weekly pay vs monthly pay). Nevertheless, using only the data from the period in which the snapshot date falls could allow employers to purposefully disguise the GPG, which would undermine external comparability. For instance, an employer paying employees on a monthly basis may fire low-paid female employees or hire highly-paid female workers in April 2020 to improve the data of the relevant pay period for GPG reporting purposes in 2021.
In summary, it could be argued that the Regulations are quite useful in terms of improving both external and internal comparability. Having four relatively clear measures and the comparative feature on the website is certainly helpful. However, they are far from perfect: the total exclusion of agency workers and the snapshot date system can allow for white-washing the data to the detriment of both external and internal comparability.

(c) Integrity
Our third assessment criterion requires that the information reported is correct and true, and that employers act honestly throughout the reporting process. Under the Regulations, employers are not obliged to explain how they calculated the metrics, so accuracy and integrity may be difficult to verify. Furthermore, some features of the Regulations may facilitate data window-dressing.
Some of shortcomings of the temporal and personal scope are also relevant in this context. Building on the example in the previous section, it would greatly undermine the integrity of the data if an employer were to fire low-paid female employees or hire highly-paid female workers in April 2020 to improve its GPG for the following reporting year. Similarly, as previously discussed, the Regulations exclude the pay data of partners and agency workers, which would arguably allow employers to provide an inaccurate picture of their true pay structure.
Some inaccuracies or integrity problems could also arise from the calculation of measures that must be reported. While the GPG, BPG and bonus proportions by gender (BPbG) ( Table 4) are straightforward and leave no room for manoeuvre, the delimitation of the quartile pay bands falls to the employer. To calculate this measure, employers must rank all employees from lowest-to highest-paid, and divide them evenly ('as far as possible') into four pay bands. 126 The employer then calculates the proportion of male and female employees in the lower, lower middle, upper middle and upper quartile 121 Regulation 1(2). 122 CIPD Gender Pay Gap Reporting: How to Calculate, Publish, and Communicate Your Gender Pay Gap (2020) p 3. 123 Regulation 5(1)(a). 124 Regulation 5(1)(b). 125 Regulation 5(2). 126 Regulation 13(2). pay bands. 127 This is intended to reveal whether female employees are concentrated in lower-paid roles, thereby prompting employers to identify barriers to women's career progression. 128 This flexibility 129 is problematic because to calculate the quartile metrics, employers can make choices which do not need to be justified and could lead to accidental or deliberate errors that are likely to go unnoticed, 'particularly where the distortions do not create statistically-improbable errors'. 130 While we could not identify such errors in our subsample, 131 according to statistician Nigel Marriott between 9% and 17% of the data reported contains errors 132 that could be automatically identified if the Government website had inbuilt statistical 'sanity checks'. 133 For instance, a sanity check could automatically spot inconsistencies between the median GPG and quartile metrics. 134 On this basis, the reporting system clearly allows for data that contravene accuracy and integrity standards. The Regulations' personal and temporal scope contain loopholes that let employers report data that do not reflect the real pay structure of the company, thereby undermining data integrity. Furthermore, the way in which the measures are calculated, in combination with the lack of automatic sanity tests, can lead to inaccurate data going unnoticed by the inexpert eye. These issues undermine the overall reliability of reported information, so doubts over whether a given employer calculated the information honestly and accurately may linger. In fact, Bailey et al have already found empirical evidence of misreporting in the GPG data disclosed (2017-18 to 2020-21) by in-scope employers. 135

(d) Meaningful and relevant information to identify the causes of the GPG
The final criterion in our assessment framework requires that the information reported helps identify the roots of the GPG within in-scope employers. The first key impediment to this is the Regulation's definition of 'pay'. Pay is understood as gross 'ordinary pay', including basic pay, allowances, 136 pay for piecework, pay for leave and shift premium, 137 and excluding remuneration in lieu of leave, remuneration in kind, salary sacrifice schemes and pay linked to overtime, redundancy or termination of employment. 138 This restrictive definition of 'pay' is regrettable, not only because it is narrower than the concept of 'pay' derived from EU equal pay lawapplicable in the UK prior to Brexit 139but also because there is evidence that GPGs can differ across compensation components. 140 For instance, in-kind remuneration is likely to be higher in executive jobs, 141  Iyengar, above n 48, p 111. Both statistically-improbable errors and misreporting have been detected: BEIS Committee, above n 42, pp 7-8. 131 We searched for mathematical errors for all data reported in 2017-18 using the tool created by Marriott Statistical Consulting 'Check your data is correct -Gender pay gap data chart tool v1.0 (2018)', https://www.personneltoday.com/ hr/gender-pay-gap-reporting-errors/. This term is further defined in regulation 3(3); Romney, above n 7, pp 381-382. 137 Regulation 3(1). The inclusion of shift premium is positive given that 'men are more likely to work unsocial hours' and such premiums are likely to contribute to an organisation's GPG; Iyengar, above n 48, p 86. 138 Regulation 3(2). 139 See eg Allonby, above n 120. 140 G Grund 'Gender pay gaps among highly educated professionalscompensation components do matter' (2015) 34 Labour Economics 118. 141 Substantial amounts of equity compensation or stock options are frequently included in executives' incentive packages. As non-cash compensation, they are considered in-kind remuneration. predominate, and it can have a strong impact 'on employers' ability to recruit and retain women'. 142 Furthermore, including shift-premium in the concept of 'pay' while excluding overtime seems inconsistent given that both strongly reward higher availability during care-friendly hours and thus incentivise gendered working patterns that stimulate pay gaps.
Indeed, the exclusion of overtime is the most disturbing feature of the Regulations' 'pay' definition. This exclusion is inconsistent with the EqA 2010 and other employment legislation such as the Working Time Regulations 1998, which recognise overtime within the definitions of 'pay' and 'holiday pay' respectively. 143 Further, overtime pay accounts for 1.1 billion hours of work annually and it can typically represent between 12-15% of certain employees' total earnings. 144 2.6 million UK employees report doing paid overtime, but men are more likely than women to do it (12% compared to 7%). 145 Therefore, overtime pay data could shed some light on the impact that working-time patterns and requirements for 'constant availability' have on the GPG, particularly in male-dominated occupations and organisations. 146 Additionally, employers must only submit data relating to gross pay, 147 so differences in take-home pay linked to eg tax credits, child-care vouchers (typically deducted from take-home pay) are excluded from the reporting obligations, although they may significantly affect employees' disposable income and employers' ability to recruit and retain women. 148 Furthermore, as noted earlier, only employees on 'full-pay' are included in the calculations, so reduced pay received by employees due to taking care leave or needing to take repeated sickness absences due to disability cannot be captured by the data submitted. Hence, employers reporting only the minimum required data are likely to miss pay disadvantages faced by women linked to care responsibilities (eg reduced or absence of care leave pay) and intersectional issues (eg sickness pay taken by older or disabled women).
Other aspects to consider are the measures and metrics themselves, summarised above in Table 3. The GPG shows the difference in 'hourly rates of pay' of male and female full-pay employees during the relevant pay period. 149 This is measured by reference to mean hourly rates (mean GPG) and median hourly rates (median GPG). Likewise, the BPG, which is the difference in bonus pay received by male and female employees during the relevant pay period, is measured by reference to the mean bonus pay (mean BPG) and the median bonus pay (median BPG). 150 Both the GPG and BPG can be useful because they are complementary. 151 The obligation to report both the mean and median gap is intended 'to give a more balanced overview' of an organisation's GPG. 152 While the mean gives the average considering the whole distribution, the median identifies the 'amount paid to the middle recipient', meaning it is not distorted by especially high or low salaries and bonuses paid to a small number of employees. 153 For instance, for Company #1 from our subsample, the mean GPG is 59% and the median GPG is much lower (29%), whereas for Company #12 the mean GPG is 40% and the median GPG is slightly higher (44%). While the GPG is extremely high for Company #1, for the average employee it is actually lower than in Company #12. This suggests that the relative 142 Iyengar, above n 48, p 85. Regulations 8-9. The hourly rate of pay includes ordinary pay (basic pay, allowances, pay for piecework, pay for leave, shift premium pay) and bonus pay, adjusted where necessary. 150 Regulations 10-11. 151 For a critique, see K Andrews The Gender Pay Gap Reporting Measures (Institute of Economic Affairs, 2018). 152 ACAS, above n 107, p 11. 153 Impact Assessment, above n 36, p 8. number of male employees who are better paid than female employees is comparatively higher in Company #12.
The main shortcoming of the first two measures (GPG and BPG) is lack of detail. These measures are not broken down according to working patterns or other observable employee characteristics, eg job categories, tenure, or part-time vs full-time employees. 154 Yet, the ONS reports show that working patterns 155 and occupation 156 bear substantial explanatory weight for the GPG in the UK, ie they are responsible for 23% and 9.1% of the explained part of the GPG, respectively. 157 Therefore, the GPG and BPG can only help identify very general trends and pay issues. 158 The lack of disaggregated data according to employees' working patterns or other relevant factors (like atypical work, in kind or overtime pay) makes it difficult to assess whether any of those variables could be linked to an organisation's overall GPG. Take, for example, Company #8 from the subsample. As shown in Figure 1 below, in 2017-18 it had a mean GPG of 59%. Looking at the metrics, it is difficult to reliably infer what may be causing that GPG. Obviously, the high BPG is likely to have a notable influence on the overall GPG, and the same could be said about the high concentration of women in the lower and lower middle pay quartiles, but it is impossible to identify more subtle structural issues linked, eg, to working patterns or to the high/low concentration of women in specific job categories. Longitudinal analysis is also difficult, if not impossible. For example, comparing the 2017-18 and 2020-21 data, there seem to be some mild improvements given that the mean GPG had declined to 54%. One can guess that the reduction of the proportion of women in the lower and lower middle pay quartiles could be the reason for that improvement. However, in the same reporting year there was also an increase in the median GPG and BPG and a reduction in the proportion of women in the two most highly paid quartiles. These latter variations could create the opposite effect, ie increase in mean GPG, and thus compensate the changes in the lower two quartiles. Without a more detailed breakdown of the data and commentary from the company, one cannot but guess what caused the overall reduction in the mean GPG. It is impossible to understand if the changes were the result of strategic evidence-based action from the company or a random result linked, eg, to changes in the industry operating environment.
Furthermore, the BPG metrics refer only to those employees who receive bonuses (not to all full-pay relevant employees), and only to the total bonuses paid to them. Yet, employers often pay bonuses on a pro-rata basis, so they may vary considerably according to working hours and patterns, which will not be shown in the data. 159 Whilst ACAS recommends that employers signal this in supporting statements, this is not required by the Regulations. 160 Indeed, only three out the 15 companies in our subsample (#2, #8 and #10) referred to this shortcoming and none of them provided additional BPG data, adjusted pro-rata, on the basis of working hours.
Additionally, the GPG and the BPG are not compared with sectoral benchmark data, making it difficult to judge whether the data reported align with wider sectoral trends or whether an organisation stands out as over-or underperforming compared with other employers in the same industry. This is relevant because structural factors and gender segregation in some sectors may be positively or negatively linked to the GPG and may also make it more difficult to take action to correct it at the 154 In other jurisdictions, pay transparency reports must expressly include the GPG mean and median for part-time workers (see eg the Irish Employment Equality Act 1998, s 20A(3)) or include working-patterns (see eg the Belgian Arrete royal portant execution du Code des societies et des associations, Art 5.2.1); see further Andrews, above n 151, p 2; Dromey and Rankin, above n 3, p 20. organisational level. While most companies in our subsample (81%) referred to structural factors in their sectors as a cause of the GPG, only one benchmarked its data against sectoral data (Appendix 4).
The third measure, the BPbG, compares the proportion of male and female employees receiving bonus pay in the 12-month period ending with the snapshot date. 161 This ought to reveal how much more likely it is that male employees receive bonus pay compared to female employees (or vice versa), and is intended to encourage employers to ensure fairness in bonus payment allocation. 162 Yet, once again, the BPbG oversimplifies the complexity surrounding bonus pay awards (ie hours of work and other factors are not considered) and does not account for gender segregation in some job categories which may be paid bonuses of different value. For instance, the bonus proportion measure could be high for women if they constitute the bulk of the sales force, but they would probably receive much lower bonuses compared to those received by highly ranked executives, who may predominantly be men. In reporting year 2017-18, that seems to be the case for Company #4 from our subsample. Despite having a higher BPbG for female (88%) than for male (85%) employees, it has a mean BPG of 92% (see Figure 2). This is partly explained in the company's narrative, which states that the mean BPG 'reflects the fact we have a higher proportion of men in more senior roles, where bonus payments make up a larger part of remuneration'. In contrast, their sales team has a 'high proportion of female employees', presumably represented in the high female proportions in the lower and lower-middle quartiles. Whereas in this company the sales team enjoys a 'significant bonus percentage', it is plausible that the bonuses paid to sales staff are much lower than the bonuses received by employees in senior roles (ie in the top quartile, where women account only for 16% of the employees).
The fourth measure is the quartile pay bands. As discussed above, quartiles are created by each employer to achieve an even distribution of relevant employees across four pay bands. In so doing, however, employers are only required to consider the 'hourly pay rate' of full-pay relevant employees. Consequently, the quartile distribution may be misleading because they are not based on the overall pay of employees, and bonus pay is not factored in the allocation of employees to the different bands. Thus, a highly-paid employee for whom bonuses constitute a significant portion of pay may legitimately be categorised in the lower middle quartile if their 'hourly pay rate' falls within that category. Consequently, employers are only required to report the proportion of male and female employees within each pay band. Hence, the mean and median GPG and BPG for each band are not calculated, which makes it more difficult to understand the key issues causing gender differences within each band.
This metric does not allow for comparisons within occupations and across occupational groups, which is one of the key observable variables that explains organisational GPGs in the UK. 163 Accordingly, within each pay band employees may be vertically segregated according to their gender and there may be pay differences between male and female-dominated occupations that contribute to the GPG, 164 but none of these issues will emerge from the data employers are obliged to report.
The quartiles may indicate general patterns that may be reflected upon, 165 but they are too simplistic to help reveal the causes behind those trends, which would require much more intricate data and analysis. The employers' optional narratives that may accompany the data do not solve this limitation as they rarely go into detail. Indeed, the companies in the sample analysed did not provide additional explanations about the demographics within the quartiles and limited themselves to acknowledging the obvious, ie that there was a high proportion of women in the lower-paid quartiles and a high proportion of men in the higher-paid quartiles, often justifying those percentages with reference to historical trends within the sector (Appendix 4).
On the whole, the measures that employers must report are far too simple to contribute to an in-depth understanding of an organisation's GPG. Indeed, 'reports and auditing requirements only [become] meaningful when they [add] more complex and thorough data'. 166 As this section has demonstrated, the minimalism of these measures makes it very difficult to understand the causes of organisational GPGs and devise evidence-based actions to address them. Thus, instead of providing 'clear transparency', the Regulations could be seen as an example of 'opaque transparency', which arises when the information disclosed 'turns out to be unreliable', 'does not reveal how institutions actually behave in practice' and/or involves strong investment 'to translate nominally public data into clearly transparent information'. 167 Indeed, the Regulations' metrics are so crude that without further data or commentary from employers, understanding the causes of the GPG is almost impossible, even with additional complex calculations.
In terms of information quality, the Regulations have apparently helped improve internal and external comparability (particularly through the information available on the Government website), but as discussed in subsections 3(a)-(c), loopholes in their personal and temporal scope likely undermine comparability in practice. This, along with the lack of comprehensive coverage and the questionable integrity and accuracy of the data reported, suggests that the information reported under the Regulations will be of low quality. Consequently, despite apparently high compliance levels, the Regulations can be expected to attract feeble soft accountability (or answerability), and the EHRC's rather weak powers and limited resources are unlikely to generate hard accountability (Section 2, Table 3).
The lack of soft accountability and limited hard enforcement action arguably led our subsample companies to show very timid signs of reporting efforts that went beyond the basic Regulations' requirements. For instance, only one benchmarked the GPG data against sectoral data and none disclosed additional BPG figures adjusted by working hours. Additionally, they tended to provide superficial justifications for their GPGs (like the underrepresentation of women in senior management) or based on exogenous factors (eg unsupported statements referring to historical trends in their sectors) (Appendix 4). While these are legitimate explanations for part of the GPG, which employers cannot be expected to be responsible for correcting, very few in the subsample were willing to explore causes of their GPGs beyond societal factors.

Conclusion
The adoption of proactive measures to tackle gender pay inequity is a landmark step in the UK, departing from the remedial and individual approach of equal pay litigation. Their enactment has undoubtedly increased media and public attention towards the GPG 168 and the other pay gaps. 169 The public availability of pay data has also allowed for public scrutiny and critique 170 and the high compliance rates 171 indicate employers are apparently strongly committed towards their reporting obligations. However, the Regulations impose the simplest proactive duty, and require neither analysis nor evidence-based action. Research shows that when employers' action is not mandated, the effectiveness of pay transparency legislation tends to be limited. 172 Nevertheless, reporting high quality pay information can be a first crucial step towards identifying and addressing the causes of the GPG.
For that reason, this paper has evaluated the Regulations' potential to realise its immediate and underlying aims, 173 which are inextricably linked to the quality of information reported. The Regulations' scope and reporting requirements have been analysed against an assessment framework built around four criteria: comprehensive coverage; comparable information; accuracy and integrity; and information relevant to identify the causes of the GPGall necessary to generate soft accountability.
Our inquiry suggests that the Regulations are largely incapable of fulfilling their immediate aims due to their narrow scope and simplistic reporting requirements. The Regulations have apparently accomplished the public disclosure aim. Given that very limited public information on pay was available before, the Regulations have helpfully introduced the ability to access and compare pay data by gender between 167 Fox, above n 46, at 667. Aumayr-Pintar, above n 166; Ceballos et al, above n 37. 173 Ie delivering transparency through the public disclosure of pay information and encouraging change to workplace policies to tackle the causes of the GPG (see section 2). and within companies. However, as a result of the Regulations' scope, the integrity of reported information remains uncertain. Leaving aside calculation errors, various loopholes allow management to adopt 'minimization techniques' where reporting data openly and accurately increases risks of reputational damage. 174 Most importantly, the information disclosed is not comprehensive enough due to, inter alia, the low reporting threshold, and the metrics are unlikely to be helpful in identifying the causes of organisational GPGs. In particular, the omission of a requirement to break data down by occupation and working pattern adds further challenges for employers in understanding their GPGs. Overall, the metrics do not require employers to report either high quality accurate information or detailed data, which might not allow them to take evidence-based actions to address internal GPGs.
Consequently, even if high compliance rates create the perception of effectiveness, in practice the Regulations have only generated 'apparent' (or 'opaque') transparency. Arguably, they have not triggered 'soft accountability' and a post-austerity overburdened EHRC is unlikely to be able to undertake sufficient enforcement action to increase accountability. 175 The Regulations' usefulness as a transformative policy tool to advance substantive gender pay equity is thus questionable.
These shortcomings suggest that, in line with previous Conservative Governments' ambivalence towards equality law, when the Regulations were enacted, the key concern was minimising burdens on business rather than tackling pay inequity. The Regulations allowed the Government to board the train of the fashionable transparency discourse 176 while not fully believing in it, which resulted in an opaque transparency measure.
A similar approach can be found in connected initiatives, such as mandatory equal pay auditing. 177 Both the Regulations and the equal pay auditing system appear to be an acknowledgement of activists' loud claims for firmer and more proactive measures to tackle the GPG. Yet, in-depth analysis of both initiatives reveals a strong aversion to impose thorough transparency requirements on employers 178 and some hesitance to depart from soft governance approaches.
While our doctrinal analysis of the Regulations is broadly relevant to any in-scope employer, the commentary based on the data of our subsample has some shortcomings. First, the four-year time frame of our data may be sufficient to appreciate obvious advantages and drawbacks of the Regulations, but more subtle aspects about their practical functioning and long-term impact may be difficult to identify. Secondly, the FTSE 100 Index used to create our subsample is based on stock market capitalisation of all companies listed on the London Stock Exchange irrespective of registration status. Consequently, many multi-national enterprises (MNEs) with operations largely outside the UK are part of the subsample. These MNEs only have small head offices in the UK for GPG legal entity reporting purposes, which somewhat distorts their metrics given that male employees tend to have more presence in head office senior roles. 179 Nevertheless, this paper offers significant insights into the Regulations' practical limitations, and can provide hints to government about aspects that should be reconsidered in a prospective review of the Regulations. Possibly the more beneficial and straightforward change could be lowering the reporting threshold to improve coverage while still ensuring that the reporting duty is sustainable. 180 174 Such practices are well documented in relation to pay equity legislation: see P Paterson and M Armstrong An Employer's Guide to Equal Pay (Kogan Page, 1972) p 53, cited by McColgan, above n 6, p 97. 175 For instance, concerning the 2018-19 data, the EHRC only launched one investigation to assess data accuracy and six investigations into non-compliance, see EHRC Formal Investigations Lead to 100% Compliance on Gender Pay Gap Reporting, (14 August 2019) https://www.equalityhumanrights.com/en/our-work/news/formal-investigations-lead-100-compliancegender-pay-gap-reporting; EHRC 'Final terms of reference for the investigation into T. Class Security Limited' (2019). 176 Idea expressed in general terms by Fox, above n 46, at 666. For instance, in France the introduction of the 'Professional Equality Index' was phased so SMEs of 50 to 250 employees had six additional months to prepare to submit their data 181 and a staggered approach with a final threshold of 100 workers is also envisaged in the new EU Pay Transparency Directive. 182 However, other changes affecting the scope and/or reporting measures should be carefully considered because they would come at a cost: they would undermine data comparability over time.
Our findings also constitute a valuable stepping-stone for further research on the practical functioning of the Regulations, eg the role of existing accountability mechanisms, including the role of the EHRC, trade unions, 183 civil society and/or employees in promoting compliance with the Regulations. Furthermore, it would be worth exploring the issues faced by smaller companies 184 and by public authorities, which may vary widely compared to those of FTSE companies. Finally, in the long-term, the extent to which the Regulations can contribute to closing the GPG within a generation could be considered. However, we anticipate that, given the Regulations' shortcomings, their long-term impact may be limited.