Pay transparency: what the laws actually did.
Canada's public salary disclosure cut the gender pay gap by 20 to 40%. Austria's internal pay reports did nothing measurable. US posting laws raised wages 1 to 4%. The difference is whether anyone could ask about a specific person. That is the part of the EU directive to prepare for first.
An HR director has budgeted for the gender pay gap report due in June 2027 and is quietly confident. The first thing that actually arrives is an email from one engineer asking for her pay level and the average for colleagues doing work of equal value. The research says the email is the instrument that moves pay. The report is the one that moves nothing by itself.
Four natural experiments
Baker, Halberstam, Kroft, Mas and Messacar studied Canadian provinces that published the salaries of university faculty by name. The gender pay gap fell by 20 to 40%. Gulyas, Seitz and Sinha studied Austria's 2011 law, which required firms to produce internal pay reports by sex. They found no effect on men's or women's wages and could rule out any narrowing of the gap larger than 0.4 percentage points; there was no compression within firms either. Cullen and Pakzad-Hurson, in Econometrica, studied US state laws that protect workers' right to discuss pay and found average wages fell about 2%, least where workers had little individual bargaining power anyway: once everyone can compare, employers stop paying premiums to the few who negotiate. Arnold, Quach and Taska, in a 2025 working paper, studied US laws requiring salary ranges in job postings: postings with pay rose 30 points, wages rose 1.3 to 3.6% across three datasets, and employment and posting counts did not change.
What each instrument changed
- Named salary disclosure (Canada): gender pay gap−20 to −40%
- Internal pay reports (Austria): gender pay gapupper bound 0.4 pointsno measurable change
- Right to discuss pay (US states): average wages−2%
- Salary ranges in postings (US states): wages+1.3 to +3.6%
What separates the ones that worked
The Austrian report went to a committee. The Canadian disclosure let a woman see what the man at the next desk earned. The posting laws let a candidate see the range before applying. The right to discuss let colleagues compare. In every case where pay moved, an individual could compare their own situation to a specific other. Where only an aggregate reached a few people, nothing happened.
The EU directive contains both kinds of instrument. The reporting duty, from June 2027 for employers with 250 or more workers, is the Austrian kind, with one difference: a 5% unexplained gap that is not remedied within six months triggers a joint pay assessment with workers' representatives, which gives the report teeth the Austrian one lacked. The right to information is the Canadian kind: any worker, at any employer, may ask for their pay and the average by sex for the same work or work of equal value, and must be answered within two months. That is the instrument that moves pay, and it is the one that arrives first.
What it means for the first two months
- Answer the individual request as the main event, not the report. It is the request that changes behaviour.
- Define work of equal value before the first request, with the criteria recorded, because the answer compares the person to a group and the group has to be defensible.
- Expect compression, not just lifts. The US evidence on the right to discuss is that premiums paid to a few negotiators disappear once comparison is possible.
- Reconcile HR pay with payroll before answering anything. A request answered from a stale record is a request answered wrong, in writing.
The assistant's role in this is the tedious part: finding the job codes without a grade, reconciling HR pay against payroll, and drafting the answer to a request from current records with every figure traced to its source. The judgement about equal value, and the explanation of any gap, stay with the people who will be asked about them later.
Sources
- Baker, Halberstam, Kroft, Mas and Messacar, “Pay Transparency and the Gender Gap”, American Economic Journal: Applied Economics, 2023
- Gulyas, Seitz and Sinha, “Does Pay Transparency Affect the Gender Wage Gap? Evidence from Austria”, American Economic Journal: Economic Policy, 2023
- Cullen and Pakzad-Hurson, “Equilibrium Effects of Pay Transparency”, Econometrica, 2023
- Arnold, Quach and Taska, salary-range posting laws, NBER working paper 34480, November 2025
- Directive (EU) 2023/970 on pay transparency (EUR-Lex)
More notes
- № 19Skills became infrastructure. Judgement didn't.In October 2026 the two largest HR suites moved skills and task data into the core of their platforms. Inferred skills are now a commodity. The evidence says the bottleneck was never the taxonomy; it was what anyone did with it.Perspective4 min read
- № 17“Potential” is where the promotion gap hides.In a study of 29,809 management-track employees, women were rated 8.3% lower on potential despite higher performance ratings, and were 14% less likely to be promoted. Their lower potential ratings did not predict their future performance. Every HR system has the two columns needed to check this.Research3 min read
- № 16Works councils and robots: the evidence the pilot needs.German plants with works councils kept more of their workers through robot adoption, trained more of them, and did not adopt more slowly. Finland's mandate raised productivity. For an AI rollout that starts with the works council, the numbers are now on the employer's side.Research3 min read
See PeopleHub on one of your own workflows.
Forty-five minutes, and a clear account of what PeopleHub can and cannot do for it.
Book a working session


