PITY Britain’s press officers. April 4th was the deadline for employers in Britain with 250 workers or more to publish details of the difference between the average pay of their male and female employees, under a new annual reporting requirement (The Economist Group recorded a median gap of 29.5%). The overall figures are eye-opening: eight out of ten employers pay men more than women. Theresa May, the prime minister, has promised to tackle this “burning injustice”.

Some take the numbers to mean that women are paid less than men for the same job. In fact the exercise bluntly compares employees’ pay without accounting for their differing roles—so Premier League football clubs have vast but meaningless pay gaps, as the men on the pitch are compared with the women on the turnstiles. This may even create perverse incentives, as firms could appear better by outsourcing low-paid jobs done by women. Yet the exercise could also lead to deeper questioning of what causes differences in pay (see article). To that extent, the mad scramble among firms—not just for data, but also for justifications—is encouraging.

Even Stevens, and Stephanies

Two problems stand out. First, the data confirm that too few women hold senior positions. Firms have got better at attracting female entrants, but they leak talent, so women become rarer higher up the ranks. Remedies to fix that are well known: generous parental leave (for men and women alike), affordable child care and flexible working hours all help ensure that motherhood need not mark the end of a woman’s career.

Second, women still face being paid less than men for doing jobs that are fundamentally similar. Many suspect that clear-cut sexism has given way to more subtle discrimination. Birmingham council was successfully sued for rewarding male-dominated work like street-sweeping more generously than female-dominated work like cleaning. However, quantifying bias with rigorous statistics is tricky, and made more so by the growth of the knowledge economy—it is harder to compare the jobs of two consultants than two factory hands. Politicians, courts and bureaucrats are ill-placed to compare jobs from afar, let alone to set pay. Better for firms to be open about their pay and aware of biases, which may be unconscious.

At the heart of unequal pay is the asymmetry between employers, who know everyone’s salary, and employees, who are entitled to know no one’s but their own. As collective wage-bargaining has been replaced by salaries that are negotiated individually, firms have had more scope to pay unequally. Transparency would help overcome this and make the labour market more efficient, as workers and employers could find the jobs or employees that suited them best.

How far should this go? In Sweden firms with more than ten employees fall under a public ombudsman, who can inspect their mandatory pay survey, carry out in-depth audits and ask for changes. In America several big firms have begun voluntary pay surveys, often under shareholder pressure.

More radically, in Finland, Norway and Sweden anyone can look up what their colleagues and neighbours earn. For many, this economists’ ideal is a step too far. But discretion could be preserved by borrowing from a new German policy that lets workers in firms with at least 200 employees ask about average pay within a peer group. This gives workers the tools to bargain with, without sacrificing anyone’s privacy.

Such measures could help iron out other discrimination. The first pay survey of Salesforce, a cloud-computing firm, in 2016 led to $3m in salary corrections that benefited men as well as women. Its next survey dealt with discrimination based on ethnicity, leading to a further $3m in corrections.

Britain’s exercise has stimulated reflection in firms that had given barriers to, and biases in, pay and promotion little thought. The goal should not be to eliminate gaps but to get firms to think about which differences in seniority and pay are justified and which are not. The fact that most have accompanied their data with analysis and explanations, and that some have set targets, is at least a start.