You are on a screening call, or an application form has a box marked current salary with an asterisk beside it, and the question is in front of you. What are you on at the moment?
Everyone has advice about this, and most of it is about tone. Deflect gracefully. Turn it into a question about their range. None of that is wrong, but it skips the two things that decide what the question costs you: whether the employer is allowed to ask where you are, and what happens to the offer, and to your odds of getting one, when you answer or do not. Both have been measured.
The question comes before the offer, and that is the point
Moshe Barach and John Horton, whose experiment is below, ran a representative US survey alongside it. Of 391 respondents, 29.4 per cent said the employer at the last job they interviewed for had asked about their pay history. Of those who were asked, 82.6 per cent were asked before an offer was made.1
The timing is the tell. A firm that asks only after making an offer, the authors write, "can only be using it for bargaining purposes"; a firm that asks before "can use it during the screening process and during the bargaining process".1 So the number does two jobs. It helps decide whether you are worth pursuing, and it helps decide what to open with. Whatever you do with the question, do it knowing that.
Where it is now illegal to ask, and where it is not
United States. There is no federal rule for private employers. Federal agencies have one for their own hiring: under an Office of Personnel Management regulation they had to comply with by 1 October 2024, agencies hiring into the General Schedule and the other pay systems the rule names may no longer set pay on a candidate's non-federal salary history for a first civilian appointment or a reappointment after a break in service, and the rule adds that this holds "regardless of how they learn that information", including when the candidate volunteers it.18 For everyone else, HR Dive's running list of salary-history bans, last updated 28 April 2026, counts 22 statewide bans, though several cover only public employers and the details differ state by state.11 California's is typical of the stronger ones. Labor Code section 432.3 says an employer "shall not, orally or in writing, personally or through an agent, seek salary history information, including compensation and benefits, about an applicant for employment", and shall not rely on it in deciding whether to offer a job or what to pay.9 New York's Labor Law section 194-a adds that an employer may not refuse to interview, hire or promote you, or otherwise retaliate, because you declined to provide it.10
Two clauses in the California statute matter for what you say. Nothing stops you from "voluntarily and without prompting disclosing salary history information", and if you do, nothing stops the employer "considering or relying on that voluntarily disclosed salary history information in determining the salary".9 The ban is on the question, not the number. Mention your salary unasked in California and you have handed back the thing the law took away. Other ban states draw the line elsewhere. Illinois lets you volunteer the number but forbids the employer from considering or relying on it, in the state Department of Labor's words, "as a factor in determining whether to offer employment or compensation".17 And the rule that governs you is the one for the job and the employer, not for your address. It can be a city ordinance rather than a state law, it can turn on whether the employer is public or private, and for a remote role it is not necessarily the state you live in, so check the state and local rules for the specific role before you decide what to say; HR Dive's list covers localities as well as states.11 The bans also generally leave employers free to ask what you are looking for, which HR Dive notes in several of its state summaries.11
European Union. Directive (EU) 2023/970 contains a one-sentence ban on the question, Article 5(2), which we quoted in full in the article on when an employer has to tell you the pay. The European Commission's explainer of 5 June 2026 puts it plainly: employers will have to inform job seekers of the starting salary or pay range and will "no longer be allowed to ask them about their pay history".12 The deadline for writing that into national law was 7 June 2026. That is not the same as the rule being in force where you live. A directive reaches a private employer through national legislation, and as of 1 June 2026 one transposition tracker found that no member state had brought full nationwide transposition into force, with twelve having published no draft at all.13 Check your own country's status before you rely on it. The ban is arriving country by country.
United Kingdom. No ban, and none proposed. The government's July 2026 consultation on equal pay proposes a statutory duty to publish pay information in job adverts; the consultation document does not mention salary history questions anywhere.14 Lewis Silkin, reading the same document, notes the omission as the clearest difference from the EU approach.15 What exists instead is the Fawcett Society's #EndSalaryHistory pledge, under which employers commit "to not solicit current salary information from prospective employees in any manner including application forms, job interviews and portals".16 Voluntary, and binding on nobody who has not signed it. In the UK the question is lawful, and you should expect it.
What employers do when they cannot see your salary
The reason these laws exist is a claim about behaviour: that employers use your last salary to set your next one. One experiment tested it directly, by taking the number away.
In late 2014, an online labour market for hourly freelance work randomly assigned every employer who posted a job during a fortnight into two groups: 2,974 treated employers, who could no longer see applicants' past hourly wages on the platform, and 2,948 controls, who could. Nobody was told. The interface simply changed.1
Treated employers looked at about 7 per cent more applicants and asked at least one question of 13 per cent more of them. They were not put off hiring: the fill rate went up about three percentage points from a base of 40 per cent. And the people they hired were different. Workers hired by treated employers had wage bids nearly 11 per cent lower, and past average wages about 13 per cent lower, than those hired by controls. The authors call it "bargain hunting". Without the history, cheaper-looking candidates looked like better deals.1
The bargaining result is the one for you. In the roughly 11 per cent of hires where the two sides haggled at all, workers facing treated employers kept 0.96 of their original bid, against 0.88 for controls, which the paper reports as 9 per cent more of their initial bid.1 When the employer could not see what you used to earn, you conceded less. And when the experiment ended and the number came back, employers went back to screening as they had before.1
The limits are the setting. A freelance platform, hourly contracts, workers bidding first, applicants unaware their history was hidden; the paper says all of this itself.1 It was published in the Journal of Labor Economics in 2021.2
Observational evidence from the US bans points the same way. James Bessen, Erich Denk and Chen Meng, comparing states before and after their bans, found that private employers posted wages more often and "increased pay for job changers, particularly for women (6.2%) and non-whites (5.8%)".6 Benjamin Hansen and Drew McNichols, in a working paper on the early effects of California's ban, found the gender earnings ratio rose by about 1 per cent in states with bans, "principally driven by those who have recently switched jobs".7 Neither is an experiment, and both estimate what bans did to whole labour markets, not what one applicant's silence does. For that there is a second experiment.
What answering costs, and what silence costs
Amanda Agan, Bo Cowgill and Laura Gee hired 256 real US recruiters at their advertised hourly rates and had them screen 2,048 applications for a software engineering job with a stated pay band of $70,000 to $120,000. The recruiters were not told that the job and the candidates were fictitious. The researchers randomised whether the application form asked for salary history, whether each candidate disclosed or stayed silent, and, for those who disclosed, how much.4 The paper is in the American Economic Journal: Applied Economics, July 2025; the figures here are from the revised working paper, which is open to read.3
The headline is that silence has a price. Recruiters read a candidate who did not disclose as having lower quality and weaker competing offers, and the salary offers made to candidates who disclosed were about $7,300, or 7.5 per cent, higher than those made to silent ones.4
But that is the average silent candidate against the average discloser, and the paper takes it apart. Recruiters treated silence as if the hidden salary sat at about the 25th percentile for someone with that CV. The authors' own conclusion: "Workers below this percentile are better off silent."4 If what you earn now is low for the job you are applying for, refusing is cheaper than answering.
If it is high, the number cuts the other way. Each extra dollar disclosed raised the offer by $0.68, and raised recruiters' estimate of the candidate's competing offers by $0.77, more than it raised their estimate of what the candidate was worth. Recruiters read the number less as a measure of talent and more as a price tag.4 And price tags can be too high. Above $70,000, every extra $10,000 disclosed cut the probability of being recommended for a callback by four percentage points. Disclosing a high salary got candidates better offers when they were chosen, and got them chosen less often.4
Gender changed the size of the penalty. Silent women were punished less than silent men: a silent man's offer was reduced by $6,900 more than a silent woman's, with the flip side that women gained less from disclosing.4
Two limits worth carrying. This is one occupation, in the US, judged on paper, with no conversation. The authors flag the last of these themselves: in an interview, "a candidate whose previous salary is 'too high' could clarify their expectations and potentially avoid rejection".4 A number in a form field has no such chance.
There is also evidence that bans do not end the number, only the question. Surveying American workers in 2019 and 2021, Cowgill, Agan and Gee found that 28 per cent volunteered their salary history "even when a ban prevents employers from asking", that a further 47 per cent would if enough other candidates did, and that men were more likely than women to volunteer it.5 Where the question is banned, the people still answering are disproportionately the ones with a number worth telling.
The number that actually sets the offer
If not your history, then what? Nina Roussille's study of Hired.com, a recruitment platform for full-time engineering jobs, is the cleanest answer. Candidates there must post an "ask" salary. Firms then bid, seeing only the CV and the ask. After adjusting for CV characteristics, women asked 2.9 per cent less than men, were bid 2.2 per cent less, and ended 1.4 per cent lower, and controlling for the ask "explains the entirety of the residual gender gaps in bid and final salaries".8
Then the platform changed one thing. In mid-2018 the empty ask box was pre-filled with the median bid for similar candidates. The ask gap, the bid gap and the final-offer gap all went to zero, and women "did not receive fewer bids or final offers than men did due to the change".8 The paper is in the Quarterly Journal of Economics, August 2024.
What that says, for anyone deciding what to tell a recruiter, is that the offer follows the number you put in front of the employer for this job. Your history matters exactly as far as it becomes that number. The market rate for the role, which the employer may already be obliged to show you, is a better one to hand over, and whether the offer can move at all is a separate question from what you say now.
What to say
Where the question is banned, do not answer it and do not volunteer. Where a ban covers the role, a form field for current salary is a field you leave blank or mark not applicable. In New York the employer may not penalise you for that.10 Say nothing about your pay unprompted. In California a volunteered number can be used to set your salary where an asked-for one cannot,9 and even in a state such as Illinois, which forbids relying on it,17 volunteering tells the employer something you did not have to.
Where it is lawful and your pay is low for the job, decline, and give a range instead. The experiment prices declining at about 7.5 per cent against the average discloser, and prices disclosing a low number below that.4 "I'd rather not anchor to my current role; for this one I'm looking at X to Y" is the whole script, with X and Y drawn from the market, not from your payslip.
Where it is lawful and your pay is high, decide which you want more. The number raises the offer and lowers the chance of getting one.4 On a form, the loss falls on the callback. In a conversation you can attach the expectation to it, which the authors say is the mitigating step their experiment could not capture.4 So do that: the current figure, then in the same breath what you are looking for.
Do not invent a number. None of this rests on a lie, and a payslip request at offer stage can end a process that a range would have survived.
Practise the moment, because it is short. The question comes early and is over in a sentence, and the sentence is easier to say on the day if you have said it out loud before. If the rest of the screening call could use a rehearsal too, JobCraftly's interview practice runs one against the job description you are actually applying for and gives you specific feedback on your answers.
The employer asks early because early is when your old number is worth most to them. It is worth least to you at the same moment. Decide before the call which number you are going to give, and make it the one for the job in front of you.
