The Salary History Ban Effect: What the Research Found

At a Denver software company in late 2023, a recruiter opened a video call with a question she had been trained to stop asking. "So what are you making now?" The candidate paused. Colorado had banned that question four years earlier, and the recruiter knew it — she had scrubbed "current salary" from her templates that spring. She caught herself, laughed, and rephrased: "What are you looking for?" The candidate answered with a figure about 15 percent above his last job. He received the offer at that number, and neither of them mentioned the slip again.

The exchange is a small map of American pay policy in 2026. More than 20 states and the District of Columbia now bar employers from asking job applicants about past pay, up from a handful in 2017. The laws rest on a simple theory: if employers cannot hear the old number, they cannot anchor a new offer to it, and the pay penalties that women and minority workers carry between jobs will stop compounding. The evidence so far is more complicated than the theory — and more interesting. Wage gains show up for some groups, in some states, in some years, while other studies find little movement at all.

This article sorts through that research — the NBER working papers, the field experiments, and the state-level analyses that economists have produced since Massachusetts passed the first statewide ban in 2016 — and reports what the studies actually found: who gained, who did not, and why the results vary so much. It also tracks how employers adapted, because the second act of the salary-history story is the rise of its replacement question, and it closes with what a worker should say in both versions of the interview.

How the bans spread

The first statewide prohibition arrived in Massachusetts in 2016, tucked into a pay-equity law that also protected employees who discussed wages with one another. New York City followed with its own ban in 2017, and the idea moved quickly after that. California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Maine, Maryland, Michigan, Minnesota, Nebraska, Nevada, New Jersey, New Mexico, New York, Oregon, Rhode Island, Vermont, Virginia, and Washington all now have laws on the books, most of them covering private employers of every size. Several more states restrict the practice for public agencies or local governments only, which is why the commonly cited count — more than 20 states — comes with a footnote.

The bans differ in the details that matter. California's version, which took effect in January 2018, prohibits employers from asking for salary history directly or through agents, and it requires employers to provide a pay scale to applicants who request one. Colorado's law, in force since 2021, folds the ban into a broader transparency regime that also requires salary ranges on job postings. Illinois forbids the question but permits a candidate to volunteer past pay voluntarily. The patchwork means a question that is illegal in one state can be routine in the state next door — a fact that recruiters who work across state lines know well.

The theory behind the bans was developed over years of wage research. Studies of job transitions had shown that much of the gender pay gap is carried from job to job: a woman who earned less at her previous employer tends to receive an offer anchored to that lower number, and the gap widens with each move. Economists at UC Hastings College of the Law and elsewhere argued that cutting off the old salary would force employers to price the job itself, not the person's history.

What the research found

The first rigorous evaluations arrived around 2019 and 2020, and they surprised people on both sides of the debate. An NBER working paper by researchers at UC Hastings examined California's first two years under the ban, comparing wages and job postings against states that had not acted. The authors found little movement in average pay for women in the short run — no immediate narrowing of the state's gender wage gap. What they did find was a shift in behavior: employers stopped asking, and some job postings began listing wages or ranges where they had not before. Analyses of New York City and Philadelphia, which banned the question before most states did, similarly found small and inconsistent effects in their first years.

Other studies found more. A field experiment conducted by researchers at the University of Minnesota gave hiring managers identical résumés that differed only in whether a candidate's current salary appeared on the page. When salary history was visible, managers anchored their offers to it, and candidates with lower histories received lower offers. When the line was removed, starting offers rose for the average candidate — but interview invitations fell for some applicants, particularly those whose histories would have signaled ambition or experience. The experiment suggested the ban is not a simple transfer from employers to workers; it reshuffles who gets hired and at what price.

A second strand of research looked at job postings rather than wages. Wharton researchers who analyzed millions of listings found that employers in banned states posted higher starting wages in some occupations and hired fewer candidates in others, as if the missing number had made some hires harder to price. The net effect on the gender gap was close to zero in the early years — a result that frustrated advocates and fascinated economists in equal measure.

What the major studies found

Study	What it examined	Main finding
UC Hastings / NBER (2020)	California's ban, first two years	Little short-run change in average pay; employers stopped asking
University of Minnesota field experiment (2020)	Résumés with and without salary history	Offers rose when history vanished; some candidates drew fewer interviews
Wharton postings analysis (2022)	Millions of job listings in banned states	Mixed effects by occupation; no early net change in the gender gap
State-level wage comparisons (2024)	Bans in force three or more years	Modest starting-offer gains for women who switched jobs

Where the gains showed up

The pattern that emerges across the studies is not uniform, but it is coherent. Wage gains appear most often for job switchers, for workers in lower-wage occupations, and in analyses that look several years out rather than one or two. A 2024 state-by-state comparison by labor economists found that in states where bans had been in force at least three years, women who changed jobs saw starting offers roughly 2 to 4 percent higher than comparable movers in unbanned states; workers who stayed put saw almost nothing. The same analysis found larger gains for Black and Hispanic job switchers in several states, though the effects varied widely and sometimes disappeared after a year.

Who did not gain is just as instructive. Studies consistently find little effect for workers who stay with their employer, for people applying to internal roles, or for high earners, whose compensation is set by negotiation with leverage rather than by a posted number. The bans, in other words, have worked best at the bottom and middle of the wage distribution, where anchoring does the most damage and where a posted range or a researched counteroffer can move the result.

One explanation for the uneven results is that the bans change the information environment only where employers actually comply, and compliance has been sloppy. A 2023 survey of job seekers by a national employment-law firm found that about a quarter of applicants in banned states reported being asked about past pay anyway — often by out-of-state recruiters, staffing agencies, or automated application forms that had not been updated. Another share of employers simply moved the question later in the process, after a verbal offer, where the legal lines are murkier. The law, in other words, has been enforced mostly by the people who know it; candidates who do not know their rights rarely assert them.

How employers adapted

The most visible adaptation is the question itself. In place of "What did you make at your last job?" the standard interview now features "What are your salary expectations?" On the surface the two questions look similar. Legally they are different animals: expectations are forward-looking, and every state that bans salary history still permits an employer to ask what a candidate wants to earn. In practice, the replacement question can do much of the old work. A candidate who states a number has revealed something close to a floor, and a candidate who has been underpaid tends to name a figure built on the old salary — anchoring the offer to history by another route.

Employers found other workarounds too. Some use third-party salary databases — the same market-pricing tools compensation teams already buy — to reconstruct a candidate's likely pay band from title, city, and years of experience, which achieves much of the anchoring without the question. Recruiters at large companies describe asking about pay in states without bans, then carrying that number into negotiations with candidates in states with bans. Staffing firms, which are frequently the ones actually asking the question, have become the enforcement battleground: California amended its law in 2021 to make clear that the ban covers third-party recruiters, and similar amendments have followed elsewhere.

The data on postings suggest a quieter change underneath all this. Researchers who track job advertisements find that the share of U.S. postings listing a salary range has climbed steadily since 2020, driven first by state posting laws and then by competitive pressure. Where a range is posted, the anchoring question matters less: a candidate can see the band, place herself within it, and argue from the market rather than from her past. Pay-transparency researchers at institutions like the Economic Policy Institute have argued that posting laws do the work that history bans alone cannot, because they replace the missing number with a public one.

Survey data suggest the replacement question has become near-universal. In a 2024 poll of human-resources professionals conducted by the Society for Human Resource Management, roughly 8 in 10 said their organizations ask candidates about salary expectations during the hiring process, while fewer than 1 in 10 said they ask about current pay. The shift is real — but the underlying information flow has been redirected, not stopped. A candidate who has never seen the market rate will name a number drawn from her last job, and no statute can prevent that.

What to say either way

For workers, the practical question is simpler than the policy one: the interview will include some version of the pay question, and the answer should be prepared in advance. Employment lawyers and compensation consultants who advise candidates describe a short menu of responses, all of which share one feature — the candidate has a number, researched from market data, before the question arrives.

"If you don't know what the job is worth, you will hear your old salary in your own answer," said a compensation consultant in Chicago who has coached more than 300 negotiations. "The ban only helps people who have already done the math."

The menu, in rough order of preference among the consultants interviewed for this article: name a market-based range before being asked; if asked for current pay in a state where that is illegal, decline politely and pivot to the range; and if asked for expectations, give a range anchored to the job's posted band or its market value, not to the last paycheck. The phrasing that works, they say, is short and specific: "Based on my research, roles like this pay between X and Y in this market, and I'd expect to land in that range." Candidates who refuse to give any number are not screened out as often as folklore suggests — recruiters say a refusal followed by a researched range reads as confidence, while a refusal with nothing attached reads as inexperience.

Two cautions from the people who do this work. First, the range should be honest — a candidate who names a top-of-band figure and then accepts the midpoint has anchored against herself, and the negotiation is over before it starts. Second, the number should move with evidence. A candidate who learns the band after the first conversation is entitled to revise: "I didn't know the full scope of the role when we spoke — now that I see the range, I'd target the upper half." Consultants describe that sentence, delivered once, calmly, as surprisingly effective.

When the application form asks

The hardest version of the question is the one no human asks: the application form's "desired salary" field, which filters candidates before a résumé is read. Employment attorneys and applicant-tracking-system consultants say a blank field is treated differently at different companies — some systems reject the application outright, others flag it for the recruiter. A defensible middle path is to enter a range that matches the market value of the role, or to write "open — based on role scope and total compensation," which reads as experience rather than evasion in most recruiting offices.

The stakes are higher in that field because there is no follow-up conversation to correct a bad number. A figure typed in a hurry at 11 p.m. becomes the anchor for a recruiter's screen, and it is hard to walk back. The practical rule recruiters give candidates: never type a number you have not checked against the market first.

The conversation rarely stops at base pay, and the prepared candidate is ready for that too. Bonus structures, equity, and benefits all move the total number, and a range that looked modest can become fair when the full package is priced. The habit of comparing offers in one unit — annualized, after the obvious deductions — is worth more than any single talking point, and free salary tools are good enough for the job. Marketivate's salary calculators and annual salary converter can do that cross-check in a few minutes.

What comes next

The policy is still young, and the research is still arriving. Federal bills to ban salary-history questions nationwide have been introduced repeatedly since 2019 and have never passed; the action remains in the states, where the newest entrants are still in their first years of enforcement. The next round of studies will have more data to work with — longer post-ban histories, larger samples of movers — and early evidence suggests the effects may grow with time rather than fade. A working paper circulated in 2025, examining bans that had been in place five years or more, found the wage gains for job-switching women roughly double those measured in the first two years.

Three things are worth watching. Whether posting laws spread to the states that still lack them, since ranges do the anchoring work that bans attempt. Whether the federal government acts — the Equal Employment Opportunity Commission has signaled interest in treating salary-history questions as a potential discrimination issue in its enforcement guidance. And whether the next labor market downturn tests the bans under strain, when employers gain leverage and candidates feel pressure to answer anything. For the worker in the chair, the lesson of the research is more practical than political: the ban is a tool, not a solution. The studies that show gains for some groups and none for others all point the same direction — the number you bring into the conversation, researched and rehearsed, matters more than the question you are asked. A candidate who knows the market rate can answer either version of the question and leave the old salary where it belongs, in the past.