The Current-Salary Disclosure Call: Disclose, Deflect, or Decline

A recruiter or interviewer just asked what you currently earn, and you have to decide in the moment whether to give a number, sidestep it, or refuse. This guide is a scoring framework for that one repeated call, for candidates in any active process. It scores the situation on four axes - jurisdiction, who is asking, the stage, and whether your number helps or hurts - and lands you on one of three moves: disclose, deflect, or decline.

This is a different question from the salary you want going forward. The expectations number is something you stage and control. Your current pay is a fact that, once stated, can anchor and permanently cap your offer, and in many places the question is improper to begin with. Getting it wrong is expensive in a way that a generic "just deflect" script cannot see, because the right move flips depending on the state you are in and the size of your current number.

Why current pay is a higher-risk question than your expected number

Current pay is riskier because it is a hard baseline the employer can reuse, while your expected number is a negotiating position you set. Disclose a low current figure and you hand the recruiter an anchor that caps the offer; refuse blindly and you may pay a silence penalty on a number that would have helped you.

The evidence for the risk is a two-sided audit experiment run across hundreds of recruiters and over 2,000 applications, randomizing whether salary was asked and disclosed. Two findings drive this whole framework. First, employers make negative inferences about non-disclosing candidates and read salary history as a signal about competing options more than about worker quality. Second, disclosures by highly-paid candidates yield higher salary offers but act as a negative signal of value, producing fewer callbacks. So neither disclosing nor staying silent is free. Each is a priced signal, and the price depends on where your number sits.

The number that looks like a reason to always disclose is not one. Disclosers earned more on average because the pool of people who disclose skews toward people with strong numbers. The recruiter's default guess about a silent candidate is what makes the call.

The four axes, and what each one scores

Score the ask on four axes: jurisdiction (is the question legal and can a voluntary number be used), asker (form, external recruiter, internal recruiter, or hiring manager), stage (how early), and value (does your number beat the recruiter's default guess). The axis that most often flips the decision is value; the axis that flips the legal consequences is jurisdiction.

• Jurisdiction. Whether the ask is banned, and whether a voluntary disclosure can still be used against you. This sets the legal floor.
• Asker and channel. A written form removes your ability to redirect and stores a baseline. An external recruiter is often filling a field. An improper ask from an in-house team is a stronger signal.
• Stage. The early phone screen and the application form are the highest-leverage moments, because you state a number before you have shown any value.
• Value. Does your current number sit above or below the role's likely 25th percentile. This is the single comparison that decides whether disclosure helps.

The value axis is the whole decision

The value axis is a single comparison: is your current number above or below the recruiter's default guess for a silent candidate. In the audit, recruiters inferred that non-disclosers earned at or slightly below the 25th percentile of the role's range. That is the line. If your number is above it, disclosing pulls the offer up. If it is below it, disclosing anchors you down, and silence gets you the recruiter's more generous default.

Jurisdiction: is the question even legal, and can a voluntary number be used

Check the specific state or city before you decide anything, because the ban rules change the payoff of the identical disclosure. As of May 2024, twenty-two states and twenty-three municipalities had adopted some form of salary-history ban, the first passed in Massachusetts in August 2016. Bans typically prohibit asking about or relying on past pay but permit asking about expectations. There is no federal ban for private employers.

Two things vary and both matter. First, scope: most bans allow expectations questions, but New York restricts expected-range questions as well, so the standard "redirect to a range" move is itself constrained there. Second, waiver: what happens if you volunteer a number anyway. Illinois bars the employer from considering or relying on a voluntarily disclosed number. California and New York are the opposite - an unprompted disclosure can be factored into your offer. And "voluntary" is narrow: the employer cannot prompt you, ask a leading question, or create any pressure, or the disclosure was never voluntary.

Two more jurisdiction facts change the workflow. Liability can extend to an outside recruiter: in California an employer can be liable even when the inquiry comes from an agency. And pay-range posting is expanding, which feeds your value axis directly. Illinois requires employers with 15 or more employees to post pay scales from January 1, 2025, and Massachusetts requires employers with 25 or more workers to disclose wage ranges in postings from October 29, 2025. Re-check the current tracker for your target state, because this list moves; describe your state to a running salary-history-ban tracker rather than trusting a fixed snapshot.

The asker and the channel: form, external recruiter, internal team

Who is asking, and through what channel, decides whether you can redirect at all and how much a hostile stand will cost you. A written form removes redirection and stores a permanent baseline; an external recruiter is usually filling a field; an improper ask from an internal team is a stronger signal.

The channel is where most of the damage happens, and it happens in writing. In the researchers' survey, among workers who were asked about salary history, 45% were asked on a written application form. A form does two things a conversation does not: it strips your ability to redirect, and it records the number in the company's recruiter database, where it persists and gets reused later. That is why the highest-stakes disclosure decision is often made before you speak to a single human.

On the human side, the move shifts with the asker, though the evidence here is practitioner consensus rather than a rigorous study, so hold it loosely. An external or agency recruiter is often "just trying to fill a field," so a range ends the exchange and a standoff costs goodwill for no gain. An improper ask from an in-house team in a ban state is a stronger signal about the employer. Refolk's index shows the supply behind that channel: about 20% of US recruiter profiles sit at staffing and recruiting firms, which is why "deflect once, then range" resolves most asks without a fight.

When you want to know exactly who is on the other side of the table before a screen, you can look them up by role and market rather than guessing. Refolk turns a plain description of the person into their public professional history, so you can tell an agency recruiter filling a field from an in-house lead whose ask carries more weight.

The scoring procedure, from jurisdiction to executed move

Run these seven steps in order. The first three gather the axes, the fourth scores the call, and the last three execute and defend it. Done means you have either returned the ball without stating your real current pay, or you have disclosed a number that helps you, framed forward.

How the score resolves to a move

Combine the axes into one of three moves. The table below is the resolution most cases fall into. When axes conflict, jurisdiction constrains what is legal, and value decides whether disclosing would even help.

Deflect and escalate: the scripts and the rhythm

Deflect once with a redirect to your target range, and only give a researched range on the second push. Named practitioners converge on redirect-then-range, and the sources split on how fast to give ground, so treat the order as a lever you control.

Alison Green of Ask A Manager recommends redirecting to the range you are seeking, with language like keeping current pay confidential and stating the range you are looking for now. The documented rhythm is deflect once, then range. You can also flip the question: ask what the budgeted range for the role is before you discuss your own history. On a hard third push, Green's guidance is that you decide whether it is worth answering or potentially losing the opportunity by refusing.

First push (redirect):
"I keep that information confidential, but the range I'm looking for now is [target range]."

Turn it around:
"Before discussing my past compensation, may I ask what the budgeted range is for this role?"

Second push (give the range):
"I keep my current compensation private, but I'm happy to tell you what I'm looking for here: [target range]."

Third push (disclose or hold):
"I'd rather anchor this on the market and the role than my current number, but if it's a gate, here's the full picture: [package, anchored on target]."

For a written form, you cannot redirect, so you enter your target, not your actual pay. The dominant tactic is to put your desired range in both fields and, if asked, explain that your current salary is not relevant and that based on the market you believe your target range is reasonable. For number-only fields, placeholders like 000 or 999 signal you do not want to discuss figures yet, but they carry a risk covered below.

Silence is not neutral. It is a priced signal, and the price depends entirely on where your number sits.

How this goes wrong: failure modes and false positives

The most common way to lose here is to apply one reflex to every case. Below are the failure modes from the field, each with the check that catches it. Read this section as carefully as the scripts, because a wrong move here is often permanent.

• The "voluntary" trap. You think you volunteered freely, but the recruiter asked a leading question, so either protection was never waived, or in California and New York you waived it while assuming you were safe. Check: was there any prompt, leading question, or pressure? If so, the disclosure was not truly voluntary.
• Disclosing while below-market. You disclose to look cooperative, and warmth from the recruiter reads as safety, but your number sits below the 25th percentile and anchors you low. Check: compare your number to the posted range before you answer. Below the line, silence gets you a higher offer.
• Misreading the silence penalty. Staying silent is not free; non-disclosure lowers callbacks and offers. Check: if your number is strong, silence costs you a positive signal you should be sending.
• Form-field permanence. A number typed to pass an applicant tracking system is stored and reused later. Check: never enter actual current pay; enter your target or a placeholder.
• Wrong-state assumption. You assume a ban where there is none, or assume expectations are fair game in New York where they are also restricted. Check the specific jurisdiction, not a general impression.
• Over-refusing with an agency recruiter. Treating a routine field-fill as a hostile ask reads as difficult and costs goodwill. Check: an external recruiter is often just filling a tracking field, so a range ends it without a standoff.
• Placeholder backfire. Entering 0 or 999999 can look like you cannot follow instructions, or read as dishonest. Check: prefer a target number or a note field over an obviously fake placeholder.

Match your read against the audit numbers before you commit

Before you disclose, sanity-check your instinct against what the experiment actually measured, because recruiter warmth is a false positive for safety. The audit gives you the direction of each signal; use it to confirm that your move sends the signal you intend.

Read the table as two opposing forces. Disclosure raises the offer conditional on getting a callback, but it also signals strong competing options, which lowers callback odds for a highly-paid candidate. Non-disclosure earns a negative inference and a lower default estimate, but it protects you when your real number is weak. Your move is the point where these forces balance for your specific number, which is exactly the 25th-percentile comparison.

Verify before you answer, and keep the jurisdiction read current

Run this checklist in the seconds before you speak or type. It is the difference between a scored move and a reflex, and every item maps to an axis or a failure mode above.

Two parts of this framework go stale, so re-check them per search rather than trusting memory. The ban list moves: twenty-two states and twenty-three municipalities as of May 2024 is a floor, not a fixed count, so confirm your target state against a running tracker each time you enter a new process. And pay-range posting is expanding, which is good for you, because a posted range is the input to your value axis; Illinois postings from January 2025 and Massachusetts from late October 2025 mean more roles now hand you the 25th-percentile line directly.

When you are working several processes at once and each ask arrives in a different state and channel, the read is easy to fumble under time pressure. Refolk tailors each application and scores your fit per posting, so the current-pay decision is one you make deliberately against the specific role and its range rather than defaulting the same script into every process. Keep this framework open next to the posting, score the four axes, and let the move follow the score.