Should you name your salary expectation first?

The rule everybody gives

Never say a number first. Deflect. Turn the question around. Whoever speaks first loses.

It is the most confidently repeated rule in job hunting, and the best available evidence says it is not a rule at all. It is a trade, and which side of it you want depends on something nobody asks you.

The study that actually measured it

Salary History and Employer Demand: Evidence from a Two-Sided Audit by Amanda Agan, Bo Cowgill and Laura Gee, NBER working paper 29460, is unusually good for this subject because of how it was run.

The researchers hired 256 real recruiters, paid at their normal rates, and had them evaluate 2,048 applications for a software engineering role. The recruiters did not know the candidates and the job were fictitious. Both sides were randomised: whether the employer asked about salary, and whether the candidate disclosed and at what level.

The results point in two directions at once, which is why the simple rule survives.

Disclosing raised the money. Candidates who disclosed received offers about $7,300 higher, roughly 7.5 percent, than candidates who stayed silent. Each additional dollar disclosed raised the offer by about 68 cents.

Disclosing cost interviews. Among higher earners, every additional $10,000 disclosed reduced the probability of a callback by 4 percentage points.

The authors put it as the law of demand, which is the clearest way to see it. Naming a number is naming your price. Raise your price and fewer people buy.

There was also a difference by gender: men were penalised more on callbacks for disclosing, and women were penalised less for staying silent.

What happened when a real employer banned the question

The second study is rarer still, because it is a randomised controlled trial inside a real organisation.

Dropping Anchor.pdf) by Sherman, Brands and Ku, published in Management Science in 2023, ran for sixteen months at a UK educational institution across 230 actual hires. Roles were block randomised by salary band into "do not ask about salary history" and "ask".

The effect of the ban on what people were paid was £408, with p = 0.84. In plain terms: nothing. The authors conclude the ban was "about as likely to harm new hires as it was to help them." There was no significant gender difference either.

The most interesting detail is an accident. About a quarter of the roles assigned to "do not ask" involved disclosure anyway, and the candidates in those non-compliant roles ended up with the highest average salaries.

That is what selection looks like. People volunteer a number precisely when the number helps them. Which is the real answer to the question this article is about.

The law may have taken the choice away

Before deciding anything, it is worth knowing whether the question is even legal where you are applying.

In the United States, 22 states have some form of salary history ban, along with more than twenty cities and counties, according to HR Dive's tracker, last updated in April 2026. Nineteen jurisdictions cover private employers, including California, Colorado, Illinois, Massachusetts, New York and Washington. A handful of states restrict only public sector employers, and Michigan and Wisconsin have gone the other way and blocked their own cities from passing bans.

The scope varies far more than the headline count suggests. Some laws ban asking. Some only ban relying on the answer. Some explicitly allow you to volunteer it. A single national number flattens all of that, so check the state rather than the count.

Separately, a growing list of states now require employers to post a salary range on the advertisement itself, which changes the conversation entirely: Colorado, Nevada, Rhode Island, California, Washington, New York, Hawaii, Maryland, Washington D.C., Illinois, Minnesota, New Jersey, Vermont and Massachusetts, with Virginia, Maine and Connecticut arriving during 2026.

In Europe, the EU Pay Transparency Directive had a transposition deadline of 7 June 2026, and a lot of commentary described that date as the moment pay transparency arrived. It did not. By the deadline only four of twenty seven member states had fully transposed it: Italy, Slovakia, Lithuania and Malta. Around ten had published a draft and around ten had not started, with Sweden having withdrawn its draft. For most of the EU, nothing became enforceable against private employers on that date, whatever the headlines said. If this is your situation it is worth checking your own country's position rather than the directive's.

How to actually decide

The evidence does not give you a rule. It gives you a question to ask yourself, and the answer is different for different people in different months.

What is actually scarce for you right now: interviews, or money?

• If you are not getting enough interviews, silence is worth more than the money it costs you. You cannot negotiate an offer you were never made. This is most of the argument for the conventional advice, and for people in this position the conventional advice is right.
• If you have interest and you are choosing between offers, the trade runs the other way. Disclosure raised offers by about 7.5 percent in the study, and you are not short of callbacks.
• If your current number is strong for the role, the callback penalty is smaller in relative terms and the offer gain is real. The non-compliant roles in the UK trial are the evidence: people who volunteered did well, because they volunteered when it suited them.
• If your current number is low for the role you want, disclosure anchors you to it at about 68 cents on the dollar. This is the case where the blanket rule genuinely protects you.
• If the advertisement carries a range, the question is mostly moot, and your job is to place yourself inside that range on evidence rather than to avoid naming anything.

What the research will not support is the thing you have been told: that there is one correct move and it is always silence. There is a price and there is a quantity, and you are choosing between them.

If the underlying problem is that the applications are not turning into conversations at all, the two places worth looking first are what actually happens to an application after you submit it and how much of a CV gets read.

The rest of this thinking is in Ask for More.