Salary Transparency Laws and Job Market Data Quality

Salary Was the Missing Field

For decades, compensation was the most commercially valuable field in a job posting and the least likely to be populated. Employers treated salary as something to negotiate, not advertise. For anyone building labor market analytics, this meant the single highest-demand data point was absent from the vast majority of records.

Regulation is changing that, and the data tells the story with useful specificity.

Canaria tracks salary disclosure across 865 million job observations spanning 2022 to the present, drawn from Indeed, LinkedIn, 200,000+ employer ATS career portals, and dozens of additional sources. Each record carries 82 structured fields. Over that period, a wave of state and federal transparency laws has reshaped the completeness profile of job market data. The shift is uneven, sometimes counterintuitive, and nowhere near finished.

The Legislative Timeline

Colorado's Equal Pay for Equal Work Act took effect January 1, 2021, making it the first state to require compensation ranges in job postings. The initial reaction was telling: some employers excluded Colorado applicants rather than disclose salaries.

New York City followed in November 2022. California and Washington State enacted laws effective January 1, 2023, covering roughly 25% of the US labor market. Hawaii joined in 2024. Illinois and Minnesota brought requirements into force in 2025.

On the international front, the EU Pay Transparency Directive began taking shape in 2025, signaling that salary disclosure is becoming a global expectation. Pending US legislation in New Jersey, Massachusetts, and Vermont would extend coverage to the majority of the American workforce.

Year-Over-Year Disclosure Rates

Here is salary disclosure across our dataset, measured as the percentage of records that include any stated salary information:

The 2022 figure of 30.9% is the highest overall rate, but not because transparency laws were more effective that year. It reflects source composition: the mix of platforms captured in 2022 skewed toward sources with higher disclosure rates. As ingestion expanded to include more international coverage and additional ATS feeds, the denominator grew faster than the salary-populated numerator. Any time-series analysis must account for source mix changes. Our methodology details how we handle source normalization across years.

The meaningful trend is the recovery from 2023 to 2025. Coverage climbed from 19.9% to 23.0%, with an absolute increase of 23.3 million salary-populated records between 2024 and 2025. That 2025 figure of 62.7 million is the largest absolute count in our dataset.

Transparency Law States vs. the Rest

State-level data reveals whether legislation actually moves the needle:

Colorado leads at 24.6%, with the major 2023 law states clustering around 23.4-23.9%. States without transparency laws trail, but not by as much as you might expect. The gap between Colorado (24.6%) and Texas (21.4%) is 3.2 percentage points. At scale, that matters: applied to Texas's 47.7 million records, a Colorado-level rate would yield 1.5 million additional salary data points. But market forces, particularly platform defaults and employer competition for candidates, are driving substantial disclosure even without mandates.

Newer law states tell a different story. Illinois (18.4%) and Minnesota (20.2%) sit below the non-law state average, likely reflecting the lag between legislation and steady-state compliance. Colorado had years to reach 24.6%. Illinois is months in. The trajectory matters more than the snapshot.

Platform Matters More Than Geography

One of the less intuitive findings is that the source platform often matters more than state law. Disclosure rates by platform:

• SimplyHired: 78.2% of records include salary
• Indeed: 49.7%
• Jobs2Careers: 42.3%
• CareerBuilder: 3.6%
• LinkedIn and most ATS feeds: approximately 0%

SimplyHired's 78.2% rate reflects product design, not employer generosity. The platform estimates and displays salary ranges even when employers do not provide them. Indeed follows a similar but less aggressive approach. LinkedIn almost never includes salary in structured fields. Salary on LinkedIn appears in description text, requiring NLP extraction rather than simple field parsing.

This means salary coverage in any state is partly a function of which platforms dominate that state's posting ecosystem. A state where Indeed is the primary channel will show higher coverage than one where LinkedIn dominates, regardless of legislation. You can compare raw vs enriched records to see how source composition affects field availability across our datasets.

What This Means for Compensation Analytics

The uneven distribution creates specific challenges:

Selection bias is the core problem. When only 23% of records include salary, those records skew toward certain industries (healthcare, government, retail), seniority levels (entry and mid-level), and geographies. Treating the disclosed subset as representative produces biased estimates. Understanding which fields are populated and why requires familiarity with the underlying schema.

Source blending amplifies bias. Combining Indeed (49.7% coverage) with LinkedIn (near 0%) means salary-populated records come almost entirely from Indeed. Your "average salary for software engineers in SF" is actually the Indeed-specific average. Checking provider-level breakdowns helps isolate this effect.

Year-over-year comparisons require controlling for source mix. Our coverage went from 30.9% to 19.9% between 2022 and 2023 due to source composition changes, not declining transparency. Any time-series analysis must account for this.

Filling the Gap with Prediction

Stated salary coverage will continue climbing as more states and countries adopt transparency mandates. But even optimistic projections leave the majority of records without employer-provided compensation. We address this with a salary prediction model trained on over 50 million observations, achieving MAPE under 15%. It requires state, zip code, and SOC code as inputs. When prerequisites are available, it covers records lacking stated compensation. When prerequisites are missing, it returns -1 rather than an unreliable estimate.

The combination of stated and predicted salary transforms a 23% coverage field into majority coverage for records with valid geographic and occupational inputs. For historical analysis, where pre-2022 stated rates are even lower, this makes longitudinal compensation research feasible. See the glossary for definitions of key terms like MAPE, SOC codes, and annualization.

Where This Is Heading

The regulatory trajectory points toward near-universal salary disclosure within five to ten years. The EU directive will eventually cover 450 million workers. Transparency-law states will push toward 30-40% as compliance matures. Non-law states will follow more slowly as employer competition and platform defaults continue raising the baseline.

For data buyers, the practical question is whether to wait for stated coverage to improve or to work with prediction-augmented data now. The answer depends on your tolerance for model uncertainty versus selection bias. Both have measurable costs, and the data to evaluate that tradeoff is available.

To see salary coverage for your target occupations and geographies, request a sample.