Key Points:

• Under the Department of Labor’s 2026 proposed NPRM’s experience benchmarking alternative, 56 percent of H-1B registrations would fall below the new Level I floor and receive no lottery entries, up from 21 percent under the NPRM primary rule.

• Combined with the wage-level-weighted lottery, experience benchmarking would raise mean compensation by $27,686 (+24.7%) over the prior random lottery, $7,076 above the NPRM primary rule’s $20,610 gain.

• Average age falls 2.3 years and the prior F-1 share rises 9.1 percentage points under experience benchmarking, the opposite direction from the NPRM primary rule, because the benchmark rewards wages that are high relative to a worker’s credentials rather than high in absolute terms.

• About 12 percent of the compensation gain could be offset by strategic occupational reclassification, compared with 18.9 percent under the NPRM primary rule, reducing the expected increase from 24.7 percent to 21.6 percent.

Background

Since March 2026, DHS allocates the 85,000 cap-subject H-1B visas through a wage-level-weighted lottery, assigning 1 to 4 lottery entries based on each registration’s prevailing wage level. On March 27, 2026, DOL published a Notice of Proposed Rulemaking (hereafter, the 2026 NPRM) raising the four prevailing wage percentiles from approximately 17th/34th/50th/67th to 34th/52nd/70th/88th; our previous brief analyzes that primary percentile-increase rule (hereafter, the NPRM primary rule). The NPRM also invites comment on three alternatives.

Under the experience benchmarking alternative, DOL would depart from the uniform-percentile approach and set each worker’s Level I wage to the estimated median earnings of native-born U.S. workers in the same occupation and geographic area with comparable education and experience. Levels II, III, and IV would apply the same education-experience wage premium to the OEWS-derived 62nd, 75th, and 90th percentiles for that occupation and area. The wage premium would be estimated from a series of Mincer earnings equations fit on American Community Survey (ACS) microdata for native-born U.S. workers. We simulate the H-1B lottery outcomes under experience benchmarking using a synthetic FY 2024 registration pool built from data obtained by Bloomberg.¹

How Experience Benchmarking Would Change the Wage-Level Distribution of H-1B Registrations

Under experience benchmarking, 56 percent of H-1B registrations would fall below the new Level I threshold, compared with 20.9 percent under the NPRM primary rule. The wage-level-weighted lottery excludes these “Too Low” registrations entirely, so their employers would have to raise offered wages or withdraw.² Figure 1 shows how current OFLC wage level assignments would shift under the experience-benchmarked scale.

We estimate that 87.4 percent of current Level I registrations would fall below the new Level I floor; for current Level II registrations, 66.8 percent drop out. Even at current Level IV, 5.1 percent would land below the new Level I because employers at the very bottom of the current Level IV wage band still fall short of the education-experience benchmark. At the top of the schedule, only 5.3 percent of the pool reaches experience-benchmarked Level IV, compared with 17.4 percent at current Level IV and 7 percent at the NPRM-primary Level IV.

Who Does Not Qualify Under Experience Benchmarking?

Among the 56 percent of registrations that would land below Level I under experience benchmarking, the demographic profile differs sharply from those above the threshold. Table 1 reports selected characteristics for each group.

Registrations that do not qualify are systematically older (average age 34.2 vs. 29.8 for those above Level I), less educated (24.5 percent hold a master’s degree or higher, versus 42.0 percent above Level I), more concentrated in Computer and Mathematical occupations (78.9 vs. 68.5 percent), and disproportionately from Indian nationals (79.3 vs. 58.1 percent). Only 20.5 percent are prior F-1 student visa holders, less than half the share above Level I (46.7 percent). These are mid-career bachelor’s-degree workers in tech whose offered wages are high in the broader occupation but below the experience-benchmarked median for their education and experience. We estimate that the median non-qualifying registration would have to raise its compensation by about 26 percent to reach the minimum lottery-qualification threshold under experience benchmarking.

Compensation Effects

Table 2 compares the compensation distribution of selected registrants under four scenarios: (1) the pre-2026 random lottery, (2) the wage-level-weighted lottery with current OFLC prevailing wages, (3) the wage-level-weighted lottery under the NPRM’s primary percentile rule, and (4) the wage-level-weighted lottery under experience benchmarking. In each weighted scenario, registrations classified as “Too Low” are dropped and the remaining pool is resampled back to the original size. Two assumptions underlie this construction. First, total registration demand is unchanged across rules: the pool refills to its original size after dropouts. Second, the demographic composition of qualifying registrants is held fixed: the resampled pool inherits the composition of FY 2024 registrations that would clear the new Level I. The 85,000-winner cap remains the same.

Experience benchmarking adds $7,076 (+5.3%) to mean compensation above the NPRM primary rule and $10,074 to the median (+8.2%). Combined with the wage-level-weighted lottery, the total gain over the random lottery would reach $27,686 (+24.7%) at the mean and $31,835 (+31.3%) at the median.

Wage-Level Distribution of Selected Registrants

Figure 2 shows the wage level composition of selected registrants in each of the three weighted-lottery scenarios, each scored using its own prevailing-wage thresholds.

Under the NPRM primary rule, the qualifying pool concentrates at Level I (37.3 percent of selections) because most registrations that previously cleared current Level II or Level III are now scored at Level I under the higher 34th/52nd/70th/88th thresholds. Under experience benchmarking, the distribution is more balanced: Level III holds the largest share at 33.9 percent, and Levels I, II, and IV each fall in a tighter 20–24 percent band. Both rules lift selected wages across the full distribution: the 25th, 50th, and 75th percentiles each rise more under experience benchmarking than under the NPRM primary rule.

Who Gets Selected?

While both experience benchmarking and the NPRM primary rule raise average compensation, they push several demographic margins in opposite directions. Table 3 reports the headline numbers.

Under the NPRM primary rule, selected workers are slightly older (32.3 vs. 31.7 years) and slightly less likely to be women or recent F-1 graduates. Under experience benchmarking these margins reverse: selected workers are 2.3 years younger, the share aged 35 and under rises by 13.2 pp, the share of women rises 2.4 pp, and the F-1 share jumps 9.1 pp.

The mechanism follows from the benchmark. Under experience benchmarking, Level I is the median wage of native-born U.S. workers with the same education and experience in the same occupation and area. A young bachelor’s-degree worker with two years of experience faces a lower Level I threshold than an older master’s-degree worker in the same occupation, because the benchmark for the former is lower. A highly paid new graduate therefore clears Level I more easily than under the NPRM primary rule, where all registrations face the same 34th-percentile floor regardless of credentials. Under the NPRM primary rule, selected workers tend to have high absolute wages, which favors senior workers. Under experience benchmarking, selected workers have wages that are high relative to their credentials.

Experience benchmarking produces a smaller shift toward higher-educated winners than the NPRM primary rule. Doctorates rise only 0.8 pp under experience benchmarking, compared with 3.3 pp under the NPRM primary rule. The mechanism is relative benchmarking: experience benchmarking sets higher dollar thresholds for PhDs, so a PhD earning the median PhD wage lands at Level I and receives one lottery entry, the same as a bachelor’s-degree worker earning the median bachelor’s-degree wage. Under the NPRM primary rule, by contrast, the PhD’s high absolute wage is much more likely to clear the 88th percentile and win four entries.

India’s selection share falls 11.8 pp under experience benchmarking, more than twice the 5.2 pp drop under the NPRM primary rule. Computer and Mathematical occupations account for 74 percent of FY 2024 H-1B registrations, and the Indian H-1B population in that group is concentrated in mid-career bachelor’s- or master’s-degree profiles. Many of these registrations offer wages that are high in the broader occupation but near or below the median for their education and experience group. The reallocated slots flow to China (+3.7 pp), Canada (+3.5 pp), and the Other grouping of countries outside the top ten (+4.0 pp), whose H-1B mix contains more PhDs and more workers at the high end of their occupation’s wage distribution.

Computer and Mathematical occupations lose 3.8 pp of selection share under experience benchmarking, slightly more than under the NPRM primary rule (−2.8 pp) and more than double the loss under current prevailing wages.

Strategic Reclassification Under Experience Benchmarking

Employers can shift a job’s occupational classification into a closely related O*NET occupation to gain lottery weight. As in our earlier brief on reclassification, we simulate this by allowing each registration to reclassify into one of its five nearest-neighbor O*NET occupations. Under experience benchmarking, 66.9 percent of registrations could achieve a higher wage level by doing so, about 5 pp more than the 61.3 percent share under the NPRM primary rule and the 60.7 percent share under current thresholds. As the number of related occupations considered grows to 20, the share under experience benchmarking rises to 87.2 percent, just shy of the NPRM primary rule’s 89.6 percent and well above the 81.0 percent under current thresholds.

Reclassification offsets 12.3 percent (first-alt) and 12.4 percent (max-gain) of the rule’s mean compensation gain, well below the 18.9 percent offset under the NPRM primary rule.

Two channels drive the smaller offset. First, the base compensation gain is larger under experience benchmarking — $27,686 over the random lottery versus $20,610 under the NPRM primary rule — so the same dollar leakage represents a smaller share of the gain. Second, experience benchmarking compresses the bottom of the wage ladder relative to the NPRM primary rule. Within FY 2024 approved petitions, the average Level I to Level II step is about $13,000 under experience benchmarking versus $24,000 under the NPRM primary rule, and the average Level I to Level III gap is $29,000 versus $47,000. At the top of the ladder the gap widens — Level III to Level IV averages $33,000 under experience benchmarking vs. $24,000 under the NPRM primary rule — but reclassification activity remains heaviest at the bottom (see Appendix Table A1), where the smaller step under experience benchmarking makes each reclassification dilute mean compensation by less.

Reclassification dilutes the selected pool’s mean wage only when extra entries tip a registrant from losing to winning, and tighter spacing means each such marginal reclassifier brings a wage closer to the typical winner’s. Reclassification activity is itself slightly higher under experience benchmarking, with both the share of registrations that can jump and the average jump somewhat larger; that would raise the offset on its own, but the two channels above outweigh it. The residual compensation lift after potential reclassification is therefore much larger under experience benchmarking: +$24,293 (+21.6%) vs. +$16,710 (+14.9%) under the NPRM primary rule.

Appendix

Estimation Methodology

Experience benchmarking sets each prevailing wage level to reflect a worker’s education and experience,³ not just the wages of all workers in their occupation and area. To produce the schedule we combine two data sources: BLS Occupational Employment and Wage Statistics (OEWS) percentile wages by 6-digit SOC and metropolitan area, and the American Community Survey (ACS) Public Use Microdata Sample (PUMS) for native-born U.S. workers from the 2018–2022 5-year sample.⁴ Restricting the ACS to native-born workers ensures the benchmark for alien workers reflects the pay of similarly qualified U.S. workers, the standard the policy is designed to apply.

The schedule centers on a single multiplicative ratio that aligns OEWS percentiles with the education-experience profile of the underlying workforce. We estimate this ratio from a Mincer earnings regression of log annual wages on a quartic in potential experience and indicators for eight education levels (less than high school, high school, some college, associate’s, bachelor’s, master’s, professional, and doctoral), with metropolitan-area fixed effects:

$$\log W_i = f(\text{PotExperience}_i) + \theta_{\text{Education}(i)} + \theta_{\text{MSA}(i)} + \varepsilon_i$$

We apply a bias correction for exponentiating predicted log wages. The four prevailing wage levels apply the ratio uniformly to OEWS $p_{50}$ (Level I), $p_{62}$ (Level II, interpolated between $p_{50}$ and $p_{75}$), $p_{75}$ (Level III), and $p_{90}$ (Level IV).

The bias correction is Duan’s smearing factor, a standard retransformation for log-linear predictions. For SOC × area cells without complete OEWS percentile data, we recover the underlying wage distribution by inverting published OFLC Level I and Level IV under a log-normal assumption (detailed in the appendix of our earlier brief) and apply the same ratio; the two methods together cover 99 percent of the 451,984 OFLC cells. For a very small share of experience-education cells where the Mincer-implied Level I falls below the federal minimum wage of $7.25 per hour, the schedule is raised to that wage. The INA-required Level I ≤ II ≤ III ≤ IV monotonicity holds by construction, since a single ratio applied to a monotone OEWS percentile base preserves the ordering. We additionally impose monotonicity along the education ladder (up to master’s) and rule out wage profiles that decline with potential experience. These corrections go beyond a standard Mincer regression and bind only in edge cases.

Under experience benchmarking, all four wage levels become curves that rise with potential experience as the conditional median climbs above the percentile-based floors. Figure A1 traces this for a bachelor’s-degree Computer Systems Analyst (SOC 15-1211) in the Philadelphia-Camden-Wilmington MSA. The current OFLC Level I for this cell is $69,971 and the NPRM primary rule would raise it to $85,106, both flat across worker experience.⁵ The experience-benchmarked Level I starts at $75,150 at zero potential experience, between the two flat thresholds, and rises above both within a few years; Levels II, III, and IV share the same shape, fanning out as the same ratio is applied to higher OEWS percentiles.

Wage-Level Transitions Under Reclassification

Reclassification activity under experience benchmarking concentrates higher up the wage ladder than under the NPRM primary rule. The largest single flow is Level III to Level IV at 17.3 percent of the pool, compared with 7.2 percent under the NPRM primary rule, and Level I to Level II is much smaller at 14.7 percent versus 30.2 percent.

This analysis was produced by PWBM staff under the direction of Alex Arnon.

Footnotes

1. H-1B lottery registration data for FY 2024, obtained by Bloomberg through Freedom of Information Act requests; see the public release. ↩

2. We treat registrations whose offered wage falls within 5 percent of the new Level I threshold as Level I, assuming their employers would slightly raise the offer to clear the new floor. ↩

3. As is standard in labor economics, we use Mincer’s potential-experience proxy (age minus years of schooling minus six) since the ACS does not measure work experience directly. ↩

4. We obtain the ACS data through IPUMS USA: IPUMS USA, University of Minnesota, www.ipums.org. ↩

5. OFLC values are from wage year 2022–23; the underlying OEWS percentile wages are BLS’s May 2021 publication. Compensation values in Tables 2 and 7 are adjusted to FY 2025 using CPI-U; Figure A1 wage levels are shown in OFLC wage year 2022-23 nominal dollars to maintain comparability with the OFLC and NPRM primary reference lines. ↩