Making pay visible, what experiments show

Transparency may concern ranges announced in job adverts, gaps published at firm level, or individual pay accessible to all. Those three levels do not have the same effects.

The European directive adopted in 2023 mainly requires the first and second levels, with transposition by member states phased over several years. It also gives every employee a right to obtain information on average pay levels for their job category.

The third level exists in a few countries, where tax returns can be consulted, or in the public service of certain states. It is these situations that have supplied most of the available data.

A study published by David Card, Alexandre Mas, Enrico Moretti and Emmanuel Saez exploits the publication online of Californian public employees' salaries, randomly informing employees of the site's existence.

The random draw is what makes the protocol valuable: it allows similar people to be compared, only some of whom received the information, and therefore to attribute the observed differences to that information rather than to some trait of people curious about their rank.

People discovering they are paid below their unit's median report lower satisfaction and higher intention to leave. Those paid above report no symmetrical effect.

That asymmetry is the field's most consistent result: unfavourable information produces a marked effect, favourable information produces little or none.

It explains why net effects on reported satisfaction may be negative at organisation level, even though transparency is considered fair by those surveyed. The two findings do not contradict each other: one can approve of a principle while finding what it reveals hard to live with.

Work exploiting publication requirements, notably in Denmark and the United Kingdom, finds a reduction in pay gaps between women and men, obtained mainly through slower growth of the highest salaries.

That mechanism matters for interpretation: the reduction does not necessarily come from a rise in low pay, something publishing a ratio alone cannot distinguish. Depending on the criterion adopted, the same result can be read as progress towards equality or as levelling down.

Publishing ranges in adverts alters the outcome of individual negotiations, reducing their dispersion. Available work finds hiring salaries tightening around the announced values.

That effect works both ways and reduces the advantage of those most comfortable in negotiation, which is the main argument advanced by the scheme's supporters. Its critics argue that an employer who knows any deviation will become known has less reason to grant an exception, which may hold back some individual rises.