Skip to content
Commuters crossing the main concourse of Grand Central Terminal, New York
AnalysisEconomics

Verifiable Floors, Cheap Ceilings: How Much Do New York’s Posted Pay Ranges Tell Job Seekers?

Analysis | 15 November 2026 | 10 min read

By Lan Jiang

Commuters crossing the main concourse of Grand Central Terminal, New York. Photograph: Zoshua Colah / Unsplash.

New York employers must now post the pay they would offer. The floor of a range can be checked by the person hired; the ceiling is far harder to check, and that decides where the information lies.

The question

Do the salary ranges in New York job postings provide meaningful information?

As approved by the editor before writing began.

Introduction

Since November 2022, employers advertising jobs in New York City have had to state the minimum and maximum pay they “in good faith” (honestly) believe they would offer; New York State followed in September 2023. Disclosure in New York State postings on Indeed rose from 31% to 61% in the year to August 2023 (Indeed Hiring Lab, 2023). Yet ranges such as $50,000–$180,000, cited in New York City enforcement complaints, raise the question: does a posted range inform job seekers, or is it wide enough to say nothing?

This essay argues that the answer follows a predictable pattern. The lower bound of a range, its floor, can be checked by the person hired, whereas the upper bound, its ceiling, is far harder to check. Ceilings should therefore rarely be exceeded, making ranges look accurate while saying little about what most hires are paid. Information should concentrate in the floor, and ranges should say least where one posting covers several levels. A second kind of information, where a range sits relative to the occupation’s market pay, matters as much as its width.

The Good-Faith Standard

New York City Local Law 32 of 2022, as amended by Local Law 59, requires employers with four or more employees to state “the range the employer in good faith believes at the time of the posting it would pay”. New York Labor Law § 194-b, in force since 17 September 2023, uses similar wording. Neither law sets a maximum width.

Regulations the State Department of Labor proposed in September 2023 would go further. Ranges would cover base pay only, not bonuses, a posting spanning several levels or locations would need a range for each, and a range made “deliberately lower or higher” than the employer would pay, or an “overly broad range of compensation without further explanation”, would lack good faith (Mandelman and Popper, 2023; Morgan Lewis, 2023). Comments closed in November 2023, but as of October 2026 the Department lists no adopted regulation and refers only to “upcoming regulatory activity” (New York State Department of Labor, no date). The proposal shows how the State views bad faith but is not binding.

Enforcement relies on complaints. In New York City, a first violation corrected within 30 days carries no penalty, and only current employees may sue; the state law allows no private lawsuits (New York City Commission on Human Rights, no date). By January 2024 the Commission had filed three complaints questioning good faith: two against ranges whose maximum was up to three or four times the minimum, and one against a range of $150,000–$225,000 without giving reasons. All remained open without reported penalties (Herbert Smith Freehills Kramer, 2024). The third case suggests the standard is not a ratio threshold alone.

A Model of Range Setting

The wage paid on a posting depends on the hire’s experience, level and bargaining position, which the employer does not know when posting. Let [wL, wH] be the wages the employer would genuinely pay across plausible hires, and [a, b] the posted range. Then

b − a = (wH − wL) + (b − wH) + (wL − a)

that is, genuine differences between hires (heterogeneity), plus ceiling inflation, plus floor deflation.

The bounds face different checks. A floor set above what the employer pays is noticed by the person hired, whose offer falls below the advertised minimum, whether or not anyone complains. A floor set too low is legal but costly, because applicants use the minimum to decide whether to apply. Both forces push a towards wL, the discipline that the theory of disclosure predicts for checkable claims (Milgrom, 1981).

A ceiling above what the employer would pay is far harder to check. No applicant observes it, and proving it requires internal pay scales, current employees’ pay or evidence from litigation. Its expected legal cost is low, though not zero, while a higher ceiling may attract applicants. Market costs restrain it: very wide ranges reduce applicants’ interest (Kuhn, 2024), and a high ceiling may raise what candidates ask for. The ceiling is thus closer to cheap talk: it costs the same whether true or not, so it conveys only coarse information (Crawford and Sobel, 1982). This matters most where pay is bargained: about a third of US workers knew their pay precisely when they first met the employer, and another third bargained (Hall and Krueger, 2012, p. 56), in a survey predating the laws.

Because the floor is pushed towards actual pay and the ceiling is not, the model yields five predictions, compared in Table 1 with a model of honest heterogeneity only (a = wL, b = wH).

Table 1. Predictions of the verifiability model (checkable floor, hard-to-check ceiling) and an honest-heterogeneity model
PredictionThis modelHonest heterogeneityTells the models apart?
P1. Actual pay lies mostly in the lower half of rangesYesYes, if most hires are paid near the bottomNo
P2. Mandatory disclosure raises maximums more than minimumsYesNoYes
P3. Ranges are widest, relative to the market, where postings span levels and pay is bargainedYesYesNo
P4. Enforcement against breadth lowers ceilings more than it raises floorsYesNoYes
P5. Pay exceeds the ceiling less often than it falls below the floorYesNoYes

Measuring Information

The usual width measure, the ratio of maximum to minimum, cannot answer the question alone: the same ratio means different things in different occupations. A better benchmark is what a job seeker expects before seeing the posting, called the prior. I take the occupation’s 10th–90th percentile range (P10–P90, the middle 80% of its wages) in the Bureau of Labor Statistics’ Occupational Employment and Wage Statistics (OEWS) as this prior and compare widths on a log scale, that is, in proportional rather than dollar terms:

I = log2 ln(P90/P10)ln(b/a)

I = 0 means the range is as wide as the middle 80% of the occupation’s wages; each unit above zero, called a bit, halves that spread; I < 0 means the posting is wider than the occupation itself, the precise sense in which a range “says nothing”. The threshold varies: for New York registered nurses a ratio of 1.83 already says nothing, whereas for lawyers it is 4.16 (Appendix, Table A1).

I measures precision only, against a prior with its own limits (Appendix). A narrow range far from the occupation’s pay may reflect an unusual employer or misstated pay, so I also report location: the share of a range inside the OEWS P10–P90 band. Both measures trust the two bounds equally; the next section tests whether they should.

Evidence

Width. By ratio, most New York ranges are narrow. Of more than 60,000 postings the City Council analysed from late 2023, 97% had a maximum below twice the minimum; the widest median ratios were 1.86 for stock exchanges and 1.65 for insurance carriers (New York City Council Data Team, no date). Since six of the seven thresholds in Table A1 exceed 2, few postings likely fall below I = 0, but ratios are only indirect evidence on P3.

Precision and location. I compared 17 New York City postings with stated ranges (13 for software developers, 4 for registered nurses), collected in October 2026 (ZeroG Talent, 2026; New York University, 2026; Corvi, 2026), with May 2025 OEWS percentiles for the New York metropolitan area (Bureau of Labor Statistics, 2026).

Nursing ranges added 0.9 to 1.6 bits. Only one software range, $150,000–$350,000 (ratio 2.33), was wider than the occupation’s spread; three more, with ratios of 2.00 to 2.11, added under 0.2 bits. One of these, $175,000–$369,000, covered mid-level, senior and staff engineers in one posting, the practice the State’s proposal would prohibit. The larger difference is location: on average 78% of a nursing range lay inside the OEWS band, compared with 39% of a software range, and two software ranges lay entirely above it. For these employers a narrow range mainly signals pay above the occupation’s market. The sample is small and drawn from job-listing websites that over-represent AI start-ups; it illustrates the measures rather than estimating prevalence.

Floors and ceilings. Glassdoor matched 147,568 job listings from January to July 2024 with employee-reported salaries (Liu, 2024). In 67% of listings reported pay fell within the range, in 22% below the minimum and in 11% above the maximum; more than 60% of within-range salaries fell below the midpoint. Floor breaches were twice as common as ceiling breaches, as P5 predicts, and the bunching below the midpoint matches P1. This separates accuracy from information: a ceiling that is rarely exceeded is accurate, but if most pay sits in the lower half it says little about what a hire will receive. A floor close to actual pay is breached more often precisely because it is close.

Glassdoor notes two caveats that apply to all these figures, not only those that fit the model: salaries date back to 2023 while listings date from 2024, and one title can cover several levels. Each pushes measured pay down relative to the range and could produce P1 and P5 without inflated ceilings. Glassdoor also finds that accuracy in the five states with good-faith laws did not differ greatly from voluntary postings elsewhere. This fits the claim that applicants, not enforcement, discipline the floor, and implies the laws add little to accuracy, although employers that post voluntarily may differ from those required to post.

Ceilings over time. In Colorado, Feng (2023) estimates that mandatory ranges raised posted minimums by 1.3% and maximums by 8.1%, as P2 predicts, while Arnold, Quach and Taska (2023), with different data and comparison groups, find no widening; the disagreement is unresolved. P4 has not been tested, and a clean test of P5 needs new hires’ pay matched to the posting they answered.

Objections and Limits

Wide ranges may be honest. As Table 1 shows, honest heterogeneity also predicts P1 and P3; only P2, P4 and P5 separate the models, and the evidence on them is mixed or indirect. Honesty and information also differ: a range spanning three levels can be set in good faith yet tell an individual applicant little.

Any range is better than none. This is largely true. Without a range, a job seeker relies on the occupational prior, so any range with I > 0 adds precision, as 16 of my 17 ranges did. The question here is how that information is distributed, not whether the laws added any.

Wide ranges may help some applicants. In a study of Cornell undergraduates responding to hypothetical job postings, Lee, Park and Chang (2026, p. 7) find that participants who chose a wide range made counter-offers about $9,934 higher, and that women were more than twice as likely as men to prefer narrow ranges. A ceiling that forecasts pay poorly can still set a reference point for negotiation, and its value differs across applicants.

The prior matters. A job seeker targeting AI start-ups holds a narrower, higher prior than OEWS, against which these ranges look less precise but better located.

Policy Implications and Conclusion

A width standard, such as New York’s proposed rules on unexplained broad ranges and multi-level postings, removes avoidable heterogeneity, while a reference standard, which ties the range to documented pay, gives the ceiling something checkable; the two work best together. Delaware’s pay-transparency law, effective 26 September 2027 for employers with more than 25 employees, combines both: a range must be based on a pay scale, a previously set range, current employees’ actual pay or the budgeted amount for the position, and its breadth is relevant to good faith (Delaware General Assembly, 2025; Durham and Dantzler, 2025). A reference standard is only as strong as its least verifiable basis, and a budget is an internal figure no applicant sees. Even so, it targets the cause identified here, whereas a fixed cap on width would also penalise honestly wide ranges. Requiring explanations is a low-cost addition: Kuhn (2024) finds that a very wide range attributed to qualifications reduced interest in the employer, while a location-based explanation tended to improve impressions.

New York’s posted ranges inform job seekers, but unevenly. Most are narrow by ratio; compared with what a job seeker already knows, nearly all add some precision, least where a posting covers several levels, and where a range sits matters as much as its width. The model locates the information in the floor, the bound applicants can check; the evidence is consistent with this but not decisive. For job seekers, the practical rule is to use the minimum to decide whether to apply and to treat the maximum as room to negotiate, not as a forecast. For regulators, it is to make the maximum checkable rather than merely limit its distance from the minimum. H-1B visa filings, which record the wage offered by employer, job and work site, would allow a direct test against posted ranges.

Appendix: Data and Measurement Notes

  • Occupations. Software developers (Standard Occupational Classification, SOC, 15-1252) and registered nurses (SOC 29-1141); postings titled “software engineer” are matched to 15-1252. Hourly ranges are converted at 2,080 hours a year.
  • Prior. OEWS May 2025 10th and 90th percentile annual wages for the New York-Newark-Jersey City, NY-NJ metropolitan area, which extends beyond New York City. They were read from Wagedex, which republishes BLS data, because the BLS site could not be reached when the pilot was compiled.
  • Pay definitions. Posted ranges cover base pay under the State’s proposed rules. OEWS wages include commissions, production bonuses and tips but exclude overtime premiums, non-production bonuses, stock and benefits. Where incentive pay is large, the OEWS spread is wider than the base-pay spread, which biases I upward; stock is excluded from both.
  • Estimation and top-coding. OEWS estimates percentiles from wage intervals. In recent releases BLS reported wages of $239,200 a year or more only as a single capped (“top-coded”) value; the lawyer and marketing-manager 90th percentiles in Table A1 exceed that level and should be checked against BLS tables.
  • Location. Share of a range inside the band = max(0, min(b, P90) − max(a, P10)) / (b − a).
Table A1. Ratio of maximum to minimum at which I = 0, New York metropolitan area, OEWS May 2025
Occupation (SOC)P10P90P90/P10
Registered nurses (29-1141)$87,180$159,2401.83
Customer service representatives (43-4051)$36,930$77,2902.09
Software developers (15-1252)$101,690$224,5902.21
Accountants and auditors (13-2011)$71,850$174,7502.43
Marketing managers (11-2021)$116,040$324,9902.80
Financial and investment analysts (13-2051)$79,340$225,3902.84
Lawyers (23-1011)$94,070$391,6204.16

About this piece

Correspondent

Lan Jiang

References

  1. 1.Arnold, D., Quach, S. and Taska, B. (2023) The Impact of Pay Transparency in Job Postings on the Labor Market. Working paper, 15 March. wpcarey.asu.edu/sites/default/files/2023-08/Arnold.pdf
  2. 2.Bureau of Labor Statistics (2026) Occupational Employment and Wage Statistics, May 2025: New York-Newark-Jersey City, NY-NJ. Washington, DC: U.S. Department of Labor. Data accessed via Wagedex. wagedex.com/metros/new-york-newark-jersey-city-ny-nj
  3. 3.Corvi (2026) Registered Nurse Jobs in New York City. corvi.careers/jobs/new-york-city/registered-nurse-jobs/
  4. 4.Crawford, V.P. and Sobel, J. (1982) ‘Strategic Information Transmission’, Econometrica, 50(6), pp. 1431–1451. doi.org/10.2307/1913390
  5. 5.Delaware General Assembly (2025) House Substitute 2 for House Bill 105, 153rd General Assembly. Dover, DE: Delaware General Assembly.
  6. 6.Durham, C.D. and Dantzler, N. (2025) Delaware Enacts New Pay Transparency Law To Go Into Effect September 2027. Duane Morris LLP. www.mondaq.com/unitedstates/employee-rights-labour-relations/1698542/delaware-enacts-new-pay-transparency-law-to-go-into-effect-september-2027
  7. 7.Feng, K. (2023) The Effects of Pay Transparency on Posted and Realized Wages. Preliminary draft, 16 September. East Lansing, MI: Michigan State University. conference.iza.org/DATA_2023/feng_k34013.pdf
  8. 8.Hall, R.E. and Krueger, A.B. (2012) ‘Evidence on the Incidence of Wage Posting, Wage Bargaining, and On-the-Job Search’, American Economic Journal: Macroeconomics, 4(4), pp. 56–67. doi.org/10.1257/mac.4.4.56
  9. 9.Herbert Smith Freehills Kramer (2024) Complying with Pay Transparency Obligations: What New York City’s Recent Enforcement Activity Signals for Employers in 2024. www.hsfkramer.com/insights/2024-03/complying-with-pay-transparency-obligations-what-new-york-citys-recent-enforcement-activity-signals-for-employers-in-2024
  10. 10.Indeed Hiring Lab (2023) Pay Transparency in a Majority of US Job Postings. www.hiringlab.org/2023/09/14/pay-transparency-majority-of-us-job-postings/
  11. 11.Kuhn, K.M. (2024) ‘Ambiguity in Advertised Compensation: Recruiting Implications of Nominal Compliance With Pay Transparency Legislation’, Journal of Applied Psychology, 109(4), pp. 599–609. doi.org/10.1037/apl0001165
  12. 12.Lee, A.J., Park, T.-Y. and Chang, S. (2026) ‘The Implications of Pay Range Transparency on Job Application Preferences and Negotiations’, Journal of Applied Psychology. Advance online publication. doi.org/10.1037/apl0001360
  13. 13.Liu, L. (2024) Navigating Pay Transparency: A Closer Look at Employer Pay Ranges. Glassdoor. www.glassdoor.com/blog/pay-range-accuracy/
  14. 14.Mandelman, M.A. and Popper, N.G. (2023) New York State Promulgates Regulations for Its New Pay Range Disclosure Law. Workforce Bulletin, Epstein Becker & Green. www.ebglaw.com/workforce-bulletin/new-york-state-promulgates-regulations-for-its-new-pay-range-disclosure-law
  15. 15.Milgrom, P.R. (1981) ‘Good News and Bad News: Representation Theorems and Applications’, The Bell Journal of Economics, 12(2), pp. 380–391.
  16. 16.Morgan Lewis (2023) New York State DOL Issues Proposed Regulations, Guidance on New York Salary Disclosure Law. LawFlash. www.morganlewis.com/pubs/2023/10/new-york-state-dol-issues-proposed-regulations-guidance-on-new-york-salary-disclosure-law
  17. 17.New York City Commission on Human Rights (no date) Pay Transparency. www.nyc.gov/site/cchr/media/pay-transparency.page
  18. 18.New York City Council Data Team (no date) Salary Transparency. council.nyc.gov/data/salary-transparency
  19. 19.New York City Local Law 32 of 2022, as amended by Local Law 59 of 2022 (New York City Administrative Code § 8-107(32)).
  20. 20.New York Labor Law § 194-b (2023).
  21. 21.New York State Department of Labor (no date) Pay Transparency. dol.ny.gov/pay-transparency
  22. 22.New York University (2026) Registered Nurse [job posting]. uscareers-nyu.icims.com/jobs/15886/registered-nurse/job
  23. 23.ZeroG Talent (2026) Software Engineering Jobs in New York, NY. zerogtalent.com/software-engineering-jobs-in-new-york-ny

Cite this piece

Lan Jiang. (2026, November 15). Verifiable Floors, Cheap Ceilings: How Much Do New York’s Posted Pay Ranges Tell Job Seekers?. The YEL Journal. https://yel-journal.lovable.app/article/verifiable-floors-cheap-ceilings-new-york-pay-ranges

Corrections

No corrections have been made to this piece. We correct errors openly, with a dated note here. Corrections policy

The readers' list

New pieces from the Journal, by email. Unsubscribe any time.

Read next