Pay transparency: three good reasons to wait, and three better reasons not to

A pay structure you cannot explain is not a structure. It is a history of negotiations.

The EU Pay Transparency Directive gave member states until 7 June 2026 to turn it into national law. Belgium did not make it; on current signals, neither did the Netherlands. The government requested a six-month extension and sent the Commission a list of interpretation questions. On 6 August the Commission published an FAQ that, as the VBO FEB noted drily, answers the general questions, ignores the Belgian ones, and binds nobody.

So the obligations are known in substance, the Belgian text is not, and the first reports are due on 7 June 2027 for employers with 150 or more staff, covering a calendar year that is three-quarters gone. That produces a predictable conversation in any management team. The three reasons to wait are good reasons. They are also, on inspection, three reasons to start.

What the directive requires

Four things matter for a mid-sized employer. Candidates must be told the pay or pay range before the interview, and you may no longer ask what they currently earn. Any employee may ask, in writing, for the average pay by sex of colleagues doing equal work, and you must answer within two months; pay secrecy clauses are void and your pay-setting criteria must be objective, gender-neutral and accessible. Employers with 250 or more staff report their gender pay gap annually from June 2027, 150 to 249 every three years from the same date, 100 to 149 from 2031, with bonuses and benefits inside the perimeter and results broken down by category of workers. And where any category shows an unexplained gap of five per cent or more that is not fixed within six months, you conduct a joint pay assessment with the unions in the room, while the burden of proof in any discrimination claim shifts to you.

Read together, the object of the directive becomes visible. It is not the size of the gap. It is whether your pay structure can be described, and whether the description holds when a colleague, a union or a court asks for it.

Three good reasons to wait

The Belgian law does not exist. The strongest argument. Belgium layers its own architecture on every directive; a 2012 law already requires companies over fifty employees to file a biennial pay analysis, and nobody knows how the two will be reconciled, what the penalties are, or how "equal value" will be read in a country with more than a hundred sectoral pay scales. Designing compliance before the parameters are known looks like poor use of scarce management time.

Belgium does not have a pay gap problem. Also largely true. On Eurostat's measure Belgium has for years reported one of the lowest gender pay gaps in the Union, in the low single digits against an EU average around eleven to twelve per cent. Sectoral bargaining and indexed scales did that work decades ago. This, the argument runs, is a German and Dutch disease, and Belgium is being asked to pay for the cure.

The cost is certain and the benefit is not. The Commission promises more applicants and better retention. The VBO reports that not one employer it consulted mentioned those spontaneously; they mentioned burden, cost and legal risk. Job evaluation costs money, consumes management time, forces a conversation most companies have carefully avoided, and produces a number that can be used against you.

Each is a serious position. A leadership team holding all three is not negligent; it is conventional. Here convention is the expensive path, because all three answer a question the directive does not ask.

Three better reasons not to

The law you are waiting for cannot change the part that takes time. Everything Belgium can still decide sits in the reporting layer: templates, channels, penalties, definitions. Everything slow sits underneath it: grouping roles into categories of equal work, evaluating them on documented criteria, mapping every employee and pay component onto that grid, and computing the gap per category for the first time. None of it depends on a Belgian choice, and the VBO, with no interest in overstating urgency, says to start the classification now because it is "the foundation for all subsequent obligations". A company of three hundred people with twenty years of individually negotiated pay needs twelve to eighteen months to reach a defensible structure. The first report is due in eight and a half. Waiting for the law is not waiting; it is choosing to be surprised.

The directive does not measure the country. It measures the category. Belgium's national gap is an average across millions of workers. The five-per cent trigger is computed inside one employer, inside one category of equal work. A company can sit in the best-performing country in Europe and still have three categories over the line, for reasons that have nothing to do with discrimination: a technician hired in the tight 2022 market at a premium the 2016 cohort never got, a bonus scheme that rewards a function men happen to dominate, a legacy of negotiated company cars. Each is defensible if an objective criterion is written down. None is defensible if the answer is "that is what we agreed at the time". The national figure is not a defence. The only defence is a structure.

The cost is real, but it is the cost of a decision you should have taken anyway. Strip away the compliance language and the directive forces a decision about how the company sets pay: which roles exist, what each is worth relative to the others, where seniority bends the curve, how variable pay is earned. In most owner-managed companies those decisions were never taken. They accumulated, one negotiation at a time, and the accumulation is called a pay structure because nobody has looked at it whole. It is the pattern this series described for pricing in Field Notes #005: a first-order lever left to run on history because examining it is uncomfortable. Transparency does not create the bill. It brings forward one you would have paid at the next reorganisation, acquisition or exit, and lets you choose the moment.

What the evidence says you get for it

This is one of the few areas of labour regulation where the experiment has already been run and measured by people with no stake in the answer. Denmark mandated gender pay reporting in 2006, the United Kingdom in 2017, and a wave of US states has required pay ranges in job postings since 2021. Three findings are robust enough to plan on.

The gap closes from the top, at no cost to productivity. In Denmark, Bennedsen, Simintzi, Tsoutsoura and Wolfenzon found reporting firms narrowed their gap by about two percentage points, and the mechanism was not faster wage growth for women but slower growth for men at the top of each range. Productivity was unaffected; profitability, if anything, marginally improved. UK studies found the same pattern and a similar magnitude. Transparency does not hand you a bill to lift everyone to the top. It removes your ability to keep paying premiums you cannot explain, and the range compresses through normal drift over two or three years.

Hiring gets easier for firms that can show a structure. Reporting firms in both countries hired more women afterwards, and UK job seekers became measurably less likely to apply to employers with large published gaps. Where US states required a range in every posting, postings drew more applicants, more of them from outside the usual pool, and the range did the filtering work a first interview used to do. In a labour market where technical and commercial roles have been hard to fill for five years, a defensible range in every vacancy is a small, durable edge over "competitive package".

Unexplained differences cost more than explained ones. The finding that should most interest anyone who manages people comes from Breza, Kaur and Shamdasani's field experiments: when co-workers learned they were paid differently and the difference tracked something visible about performance, output and cooperation held; when it could not be tied to anything, output, attendance and willingness to work together all fell. Cullen and Perez-Truglia added the direction: learning that peers earn more reduces effort, learning that managers earn more increases it. The lesson is not that pay differences are dangerous. Inexplicable ones are. The remedy is not equal pay but explicable pay, which is exactly what a job architecture produces, and a company that also publishes how pay progresses upward turns a compliance duty into a retention tool.

One cost is equally well documented and should be said plainly. Transparency compresses pay and weakens the link between pay and performance; Mas found top earners cut by around seven per cent when California cities disclosed salaries, and Obloj and Zenger found less inequity but also less pay-for-performance in US universities. For a manager who relies on discretionary rewards to keep the best people, that is the real trade-off. The answer is not to resist transparency but to make performance measurable enough that paying for it survives daylight. If you cannot write down why your best salesperson earns forty per cent more than the median one, transparency will not stop you paying it. It will stop you paying it quietly, which forces a better conversation.

What to do Monday morning

Three actions, none requiring the Belgian law.

Ask HR for one page: headcount per legal entity against the 100, 150 and 250 thresholds, per country, with the first reporting date each faces. If that page does not exist by Friday, you have your answer about readiness.

Pull the last analysis report filed under the 2012 Belgian law, if you have one, and read it as if you were about to buy the company. It is the closest thing you have to a dress rehearsal, and in most companies it went from payroll to the works council without passing the people who set pay.

Take three roles you hire for regularly and write one sentence each on why the highest- and lowest-paid incumbent earn what they earn. If the sentence contains the word "negotiated", you have found your first category of concern without a consultant.

Then set one date, in the first half of 2027 at the latest, on which the company computes its own gap per category, internally, before anyone else does. A leadership team that first sees its five-per cent categories in a filed report has lost the initiative. One that sees them nine months earlier has a budget to fix them and a story to tell.

The Belgian government bought itself six months. It did not buy them for you.

Photo: Unsplash