Who tells you what a job pays, and who doesn’t.

About half of US job postings now tell you what the job pays. Which half is not random. We took 39,084 US postings from two windows this year, sorted them by whether a pay-range law was in force where the job is located on the day it was posted, and found the same answer in both windows: where a law applies, two to three postings in four state pay (64.4% in spring, 73.4% in fall); where none does, about one in three. Then we looked at Virginia, whose law took effect between our two windows, and at the employers that post on both sides of the line.

Issue 01previewed the gap on one window and said we had not yet tested what changed when a law took effect. This issue does that, as far as the data allows, and goes further: which states are high and low, whether the same employer behaves differently in different states, why remote postings are the odd group out, which kinds of work carry the widest gap, what a stated range actually looks like, and whether “competitive salary” means what people assume. No employer is named in this issue. Section 10 explains the two windows and a change to how we define the universe.

Twice as often, in both windows

The comparison is simple. A posting is in the law group if a law requiring a pay range in the posting itself was in force in its jurisdiction on the day it was posted: eleven states and the District of Columbia throughout, plus Virginia from July 1, Maine from July 29 and the city of Cleveland. Everything else with a state is the no-law group. In the fall window, August 31 to September 20, 73.4% of the 5,052 law-group postings state a pay figure and 34.5% of the 7,210 no-law postings do. In the spring window, May 11 to June 21, it was 64.4% of 10,320 against 30.5% of 13,554. The ratio is 2.13 to one in fall and 2.11 in spring; the gap in points is 38.9 and 33.9.

Both groups moved up between the windows, by nine points in the law group and four in the no-law group. Some of that is real and some is the feed: it became US-only on September 10 and a wider retirement sweep started the same day, so the fall window has a different mix of employers and roles from the spring one. That is why the rest of this issue compares groups against each other inside a window, or pools the two windows and says so, rather than reading the two totals as a trend. Across everything in the fall window, with or without a state, 51.1% of postings state pay. Pooled across both windows the law group is at 67.4% (n 15,372) and the no-law group at 31.9% (n 20,764).

Virginia switched inside our data

Virginia’s law took effect on July 1, 2026, between our two windows and with no employer-size threshold, which makes it the cleanest before-and-after we have. In the spring window, 770 Virginia postings stated pay 34.7% of the time, in line with the other no-law states at 30.2%. In the fall window, 438 Virginia postings stated pay 68.9% of the time, in line with the law states at 74.0%. Over the same two windows the other no-law states moved from 30.2% to 34.5%.

The obvious objection is that different employers might be posting in Virginia in the fall. So we checked the employers that posted in Virginia in both windows: 60 of them, with 314 spring postings and 156 fall postings between them. Their share moved the same way as the state’s. Their fall postings fall under our 300-posting floor, so we do not print their shares, but the direction is not in doubt: the same employers changed what they put in their Virginia postings.

Two limits. We cannot show the week-by-week turn, because our pay extraction was broken from June 22 to August 30, which straddles July 1; the methodology at the end has the detail. And this is a before-and-after with a comparison group, not a controlled experiment, so we describe it rather than claim a cause. Maine’s law took effect on July 29, also between the windows, but Maine has 93 spring postings and 38 fall postings in our feed, too few to publish.

By state

Pooling the two windows gives 32 states with 300 or more postings: ten with a law in force throughout, 21 without one, and Virginia, which changed between the windows and is in section 02 rather than on these charts. The two groups do not overlap. The lowest law state is Maryland at 58.0%; the highest no-law state is Oregon at 50.4%. Every one of the ten law states sits above every one of the 21 no-law states.

Minnesota leads at 73.6%, then Colorado, the District of Columbia, Illinois, Washington, California and New York between 67% and 71%, and Massachusetts at 66.3%. New Jersey (58.8%) and Maryland (58.0%) trail the rest of the law group, which runs from 66.3% to 73.6%; New Jersey’s law is the second-newest on the list and Maryland’s applies only to work done at least partly in Maryland. California alone is 5,239 postings, so its 68.0% carries much of the law-group total.

At the top of the no-law group are Oregon (50.4%), Connecticut (45.8%) and Nevada (41.4%). Connecticut’s law requires a range when a candidate asks for one and Nevada’s once a candidate has interviewed, which we do not count as posting laws, and Connecticut’s becomes a posting requirement on October 1, 2026. The big no-law states cluster between 25% and 35%: Ohio 35.0%, North Carolina 34.7%, Florida 31.7%, Georgia 31.4%, Pennsylvania 30.0%, Texas 25.3%. Texas has 3,164 postings in the pool, more than any state but California, and a quarter of them state pay. South Carolina is lowest at 20.4%.

The law does not travel with the employer

The state pattern could in principle be an employer pattern in disguise: maybe the kinds of employers that post in California and New York are the kinds that state pay everywhere. The feed lets us check, because many employers post in both kinds of jurisdiction. We took every employer with ten or more postings in law jurisdictions and ten or more outside them, across the pooled windows: 75 employers, accounting for 18.0% of law-group postings and 19.0% of no-law postings. Inside law jurisdictions their postings state pay 67.3% of the time. Outside, 35.7%.

For comparison, employers that post ten or more times outside law jurisdictions and never inside one state pay 29.4% of the time (1,903 postings), and employers that post only inside them state pay 69.0% of the time (819 postings). Taken together, the two-sided employers look much more like the local employers wherever they post than like themselves on the other side of the line: 67.3% inside against 69.0%, and 35.7% outside against 29.4%. Taken one by one, they do not all behave that way.

Employer by employer, the picture splits. Of the 75, 28 state pay at least 20 points more often inside law jurisdictions than outside, and another nine are between five and 20 points higher. Thirty-one are within five points either way, and most of those have simply made one decision for everywhere: 20 of the 75 state pay on at least 90% of their postings in both groups, and 14 on 10% or fewer in both. Seven state pay more often outside law jurisdictions than inside, by five points or more. The median employer’s gap is 4.3 points, which is the “one policy everywhere” group pulling the middle down; the average is carried by the 28. We are not naming any of them.

Remote postings and the laws

Inside law jurisdictions, hybrid postings state pay 76.0% of the time and on-site postings 69.6%. Remote postings state it 61.0% of the time, the lowest of the three. Several of the laws say a remote role that could be performed from the state is covered, so being remote is not by itself an exemption. Our reading is that employers apply a state’s rule to postings they place in that state, and treat a remote posting as belonging somewhere else, or nowhere. Postings with no state at all, which are 54.5% remote, state pay 58.5% of the time when remote.

Outside law jurisdictions the order flips. Remote postings state pay 42.0% of the time, hybrid 39.8% and on-site 30.4%. One possible reading is that remote postings reach candidates across state lines, some of whom live in law states; this issue does not test it. One more group is worth a line: postings that state no arrangement at all state pay 24.8% of the time inside law jurisdictions and 10.8% outside. Postings that say less about where the work is also say less about what it pays.

By kind of work

Twenty role families have 300 or more postings in each group. The gap is present in all twenty, and in 14 of them postings in law jurisdictions state pay at least twice as often as postings outside. The widest gap is engineering and architecture: 80.8% inside, 25.5% outside, 55 points. Banking, lending and insurance (75.1% against 26.0%), accounting, audit and tax (78.7% against 31.9%), HR and recruiting (67.7% against 23.7%) and registered nursing (69.3% against 27.6%) follow, each with a gap over 40 points.

At the other end, software engineering has the narrowest gap of the twenty families: 64.6% inside law jurisdictions and 53.8% outside, 10.8 points. It is the only one of the twenty where more than half of no-law postings state pay. Sales (72.7% against 46.2%) and legal and compliance (85.1% against 46.4%) are the next most forthcoming outside law states, and legal has the highest disclosure of any family inside them. Creative, design and media (51.2% against 24.5%) sits near the bottom on both sides; marketing and communications (58.9% against 32.7%) is among the lowest four inside law jurisdictions but mid-table outside them. The professions that disclose least where no law applies are general and location management at 19.5%, finance advisory at 20.9% and HR and recruiting at 23.7%.

What a stated range looks like

A common worry about pay-range laws is that they produce ranges so wide they mean nothing. The feed does not show that. Among the 10,354 law-jurisdiction postings that state pay, 95.3% give a true range with two different ends; the rest give a single figure. Among the 6,614 no-law postings that state pay, 90.5% give a range and 9.5% a single figure. We did not test why; one possibility is hourly wages posted as one rate. 11.6% of on-site no-law postings that state pay give a single figure, against 5.8% inside law jurisdictions.

The ranges themselves are the same shape on both sides of the line. The median range spans 26.1% of its midpoint inside law jurisdictions and 26.5% outside. A quarter of ranges are narrower than 15.4% of the midpoint in both groups, and three quarters are narrower than 40.0% in both. Wide ranges are a minority: 11.8% of law-jurisdiction ranges and 15.7% of no-law ranges span more than half their midpoint, and ranges wider than the midpoint itself are 0.5% and 0.9%. Where the pay period is known, hourly ranges are tighter than annual ones: a median 18.2% of the midpoint against 30.0% inside law jurisdictions, and 20.0% against 31.6% outside.

“Competitive salary”, and what actually travels with pay

Job seekers tend to read “competitive salary” as a sign the employer is hiding the number. Across the pooled windows, 19.7% of postings use that phrase or one like it (competitive pay or compensation, commensurate with experience, depending on experience, DOE). Of the 7,684 postings that do, 55.1% also state a pay figure, which is more than postings overall at 46.9%. Postings that state pay use the phrase more often than postings that do not: 23.1% against 16.6%. In law jurisdictions, 74.2% of postings that use the phrase state a figure too. The phrase is boilerplate that sits next to the number at least as often as it replaces it.

The better tell is silence. Outside law jurisdictions, 56.3% of postings give neither a pay figure nor any of these phrases; inside them, 27.8%. And what does travel with a stated pay figure is a benefits list. Among postings that state pay, 79.6% list benefits in law jurisdictions and 81.1% outside them. Among postings that state no pay, 40.1% and 50.8% do. When a posting lists benefits, it lists about nine or ten of them on average regardless of group.

What to do with this

• Read the state before you read the employer. A missing pay range outside a law state is the norm, not a signal. A missing pay range inside one is unusual, and worth a question, because about two thirds of postings there include it.
• Find the employer’s law-state postings. Of the 75 employers that post often on both sides of the line, 28 state pay at least 20 points more often inside law jurisdictions. The same title in California, New York, Colorado or Washington often carries the range you are not being shown.
• Do not read “competitive salary” as a refusal. Postings that say it give a figure more often than postings overall: 74.2% of them inside law jurisdictions and 43.8% outside. The posting to be wary of is the one that says nothing about pay at all and lists no benefits.
• Remote roles are the grey area.Even inside law jurisdictions, remote postings state pay less often than on-site ones. If a remote posting could be performed from a law state, the range is often required; asking which state’s rules the employer applies is a fair question.
• Watch Connecticut.Its posting requirement takes effect on October 1, 2026. Connecticut postings are at 45.8% in our pooled windows. Virginia’s share moved between our two windows after its law took effect; we will measure whether Connecticut’s does.

How we counted

The short version. The long version, with every definition, lives on one page and every report links to it: About Longbow Research.

• Two dated windows, not a snapshot. This issue counts postings by the date the employer published them: May 11 to June 21, 2026 (26,151 postings) and August 31 to September 20, 2026 (12,933). Those are the two stretches in which our structured pay field was complete. Between them it was not; the methodology block below shows the weekly numbers.
• Definitions v2. Issue 01 and Issue 02 counted the postings live in our feed on a snapshot day. The feed now retires a posting it has not seen for fourteen days, so a live set describes the last few weeks rather than a market. From this issue the universe is postings dated inside a stated window, whether or not they are still live when we publish. The change is logged on About Longbow Research.
• Law means law on the posting date. Fifteen jurisdictions required a pay range in the posting itself during at least part of our windows; the methodology lists each with its effective date. Virginia and Maine took effect between the windows, so their spring postings are no-law and their fall postings are law. On-request and after-interview laws (Connecticut, Rhode Island, Nevada, Cincinnati, Toledo) and Columbus, whose ordinance is unenforced until 2027, are no-law.
• Thresholds we cannot see.Most laws exempt small employers. The feed carries no employer-size value for these postings, so “law in force” means in force in the jurisdiction, not necessarily binding on that employer. Some no-pay postings in law states are therefore lawful.
• Clear denominators. Every share of postings stating pay is a share of all postings in the group, with no restriction to postings that carry some other field. Postings with no state (2,948; 92.5% of postings carry a state) are reported on their own and never pooled into either law group.
• The pay field is a floor. A dollar-amount pattern in the posting text adds 5.1 points our field misses, while the field finds 23.5 points the captured text lacks. Counting either, the gap is 72.9% against 36.1%. The methodology block has the detail.
• No group under 300 postings.That is why Maine, the within-employer Virginia shares, and Ohio’s city ordinances appear as counts or directions but not as percentages.
• No pay levels, no names, no causes. Only 21.5% of stated pay carries a pay period we can read, so this issue publishes disclosure rates and range shapes, not dollar figures. No employer is named. Virginia is described as a before-and-after with a comparison group, which is as far as the data goes.

Coming next. Issue 04, What Employers Actually List: the skills named in postings, by role family, and how often people skills outrank tools. Each issue goes out as an edition of The Longbow Index on LinkedIn the day it publishes.