Effects of the New $100,000 Fee and Wage-Weighted Lottery on the H-1B Visa

Key Points

• A September 2025 Presidential Proclamation requires a $100,000 fee for H-1B petitions filed for workers abroad, an estimated 60 percent of registrations. H-1B workers subject to the fee are less likely to hold a master’s degree or higher (11.5 versus 63.8 percent of fee-exempt workers), more likely to be Indian nationals (82.2 versus 51.2 percent), and about five years older on average (34.2 versus 29.4).

• We estimate that employers would be willing to pay the $100,000 for only 30 to 56 percent of fee-subject registrations, depending on how much of the payment they recover through lower payroll. The fee would screen out the rest before they reach the new wage-level-weighted lottery, cutting the fee-subject share of the lottery pool from about 60 percent to roughly 31 to 46 percent.

• Together, the $100,000 fee and the wage-level-weighted lottery would raise the average pay of selected registrants above the pre-2026 random lottery (the selection design used through the March 2025 lottery, before the new wage weighting) by $7,551 to $18,799, or 6.7 to 16.7 percent, under current prevailing wages; under the higher 2026 NPRM prevailing wages the gain would be larger still, $16,542 to $28,686 (14.7 to 25.5 percent). Most of the increase comes from the wage weighting, not the fee, which on its own changes pay by −$3,606 to +$7,642, depending on how much of the payment employers recover through lower payroll.

• Relative to the pre-2026 random lottery, the wage-level-weighting and the $100,000 fee would reduce India’s selection share from 66.8 percent to between 55.7 and 63.5 percent, increase the doctorate share from 5.7 percent to as much as 9.5 percent, lower the share of selected workers with no prior U.S. visa from 44.5 percent to between 26.1 and 33.9 percent, and raise the former F-1-student share from 43.3 percent to between 48.5 and 50.8 percent.

Background

Starting with the FY 2027 visa lottery, the 85,000 cap-subject H-1B visas are allocated through a wage-level-weighted lottery, assigning 1 to 4 lottery entries based on each registration’s prevailing wage level. In previous work we have analyzed that design, as well as the impact of raising prevailing wages as proposed in the Department of Labor’s 2026 notice of proposed rulemaking (NPRM).

A second new policy operates alongside the new lottery design: a September 19, 2025 Presidential Proclamation requires a $100,000 fee for certain H-1B petitions filed on or after September 21, 2025.¹ The payment applies to petitions for workers who are outside the United States, including petitions that request consular processing. It does not apply to petitions that extend the status of, or change the status of, a worker already in the United States in valid status, and it does not apply to anyone holding a currently valid H-1B visa. We label the petitions to which the payment applies “fee-subject” and the remainder “fee-exempt.”

We simulate the combined effects of the fee and the wage-level-weighted lottery using a synthetic FY 2024 registration pool built from data obtained by Bloomberg.² An employer subject to the fee registers a worker only when the hire is worth the $100,000 to the firm, so anticipation of the payment screens out some fee-subject registrations before the lottery. How much of the payment employers can recover through lower payroll is uncertain, so we report our results across two illustrative scenarios rather than a single estimate. Because we build on the FY 2024 pool, our estimates describe how the two policies reshape the composition of selections rather than forecasting future application volumes. We present the details of this analysis in the appendix.

Who is subject to the $100,000 fee

The fee is not a uniform charge on H-1B hiring. It applies to the 60 percent of registrations that we estimate are filed for workers abroad.³ The workers the payment reaches differ from those it exempts on every characteristic we observe. Table 1 reports the two groups side by side.

Fee-subject registrations are concentrated among older, bachelor’s-degree workers in technology. They are about five years older on average (34.2 versus 29.4), hold a master’s degree or higher far less often (11.5 versus 63.8 percent), and are more concentrated in Computer and Mathematical occupations (83.3 versus 60.5 percent). Most have no prior U.S. visa, 89 percent, whereas three-quarters of fee-exempt workers are former F-1 students already in the country. Fee-subject workers are also offered less, $105,938 on average versus $117,040 for the exempt group. The split tracks how the fee is defined: workers extending or changing status from inside the United States, who typically entered on F-1 student visas and completed U.S. degrees, are exempt, while workers brought in from abroad, who are more often mid-career bachelor’s-degree workers, are subject to the payment.

Because exposure to the fee tracks these characteristics, it is highly uneven across groups. Figure 1 shows the share of registrations subject to the payment within several cuts of the pool.

Essentially all registrations with no prior U.S. visa are subject to the fee, 99.5 percent, as are 71 percent of Indian nationals, 78.9 percent of workers with no more than a bachelor’s degree, and 82.4 percent of workers aged 35 and older. Exposure is far lower for former F-1 students (6.5 percent), Chinese nationals (16 percent), and workers with a master’s degree or higher (21.5 percent). The payment’s exposure therefore concentrates on the same broad group that the wage-level-weighted lottery already disadvantages in selection and that the proposed higher 2026 prevailing wages would disadvantage further: lower-paid, bachelor’s-degree workers in computer occupations.

How many fee-subject registrations survive the payment

A registration subject to the payment enters the lottery only if its employer is willing to pay. The fee applies to 60.5 percent of all registrations, and how many of those clear that bar depends on the payroll savings employers expect from hiring an H-1B worker rather than a comparable U.S. worker. Following Borjas (2026), we treat the firm as paying only when the hire’s payroll savings, summed over a six-year term and discounted at 3 percent, outweigh the $100,000 payment and an unobserved hire-specific cost or benefit that offered wages do not capture.⁴ Because the size of those savings is uncertain, we report two variants. The parity variant sets payroll savings to zero, so survival turns on whether the firm values the particular hire, beyond the offered wage, by more than the payment. It makes no assumption that H-1B workers are paid below comparable U.S. workers and is the more selective of the two. The payroll savings variant lets each firm’s predicted savings offset the payment and is the less selective. Table 2 reports, under each variant, the share of fee-subject registrations that survive the payment, the resulting surviving share of all registrations, and the share of all registrations that reach the lottery.

Under parity, 29.8 percent of fee-subject registrations survive the payment. Under payroll savings, 56 percent do. Surviving fee-subject registrations therefore fall to 18 percent of all registrations under parity and 33.9 percent under payroll savings, down from the 60.5 percent subject to the payment, as the fee screens them out before they reach the lottery. Together with the fee-exempt registrations, which are not subject to the payment, 57.6 percent of all registrations reach the lottery under parity and 73.4 percent under payroll savings. Of the registrations reaching the lottery, fee-subject registrations make up 31.3 percent under parity and 46.1 percent under payroll savings. Because the payment reaches only fee-subject registrations and leaves the remaining 39.5 percent of registrations untouched, it changes the total number of registrations reaching the lottery far less than it changes the fee-subject subset. This compositional shift, before any wage weighting, is the channel through which the fee reshapes selections.

The combined effect on selected registrants’ compensation

The combination of the two policy changes would raise selected registrants’ pay across the whole distribution, by an amount that depends on the prevailing-wage scale and the payroll savings variant.⁵ Table 3 reports the compensation of selected registrants under the random lottery and under the wage-level-weighted lottery applied to the post-fee eligible pool (fee-exempt plus surviving fee-subject registrations), scored on both the current OFLC prevailing wages (current law) and the 2026 NPRM prevailing wages. For each scenario we show the weighted lottery without the payment, the parity variant, and the payroll savings variant.

Under current prevailing wages, mean selected pay rises from $112,282 under the random lottery to between $119,832 and $131,081, a gain of $7,551 to $18,799. Under the 2026 NPRM prevailing wages the gains would grow to $16,542 to $28,686, reaching a mean of $140,967 under parity. The gains are larger under parity than under payroll savings because the parity variant screens out more fee-subject registrations, and the lower-paid ones drop out, leaving a pool concentrated at the top wage levels and so higher-paid.

Table 4 collects these combined effects on the mean, the gain over the random lottery under each scenario.

What the fee adds on top of the weighted lottery

Most of the increase in selected registrants’ pay comes from the wage weighting, not the fee. The no-payment rows in Table 3 are the weighted-lottery results from our prior brief, which used the same registration pool and wage-level definitions. Comparing the payment scenarios with those rows isolates what the fee adds now that the weighting is in place.⁶ Table 5 reports that marginal effect.

The payment’s own contribution is smaller than the wage weighting’s, and its sign depends on which fee-subject registrations each variant removes and which it retains. Under parity it adds $7,642 to $8,078 to the mean, or 6.2 percent under current prevailing wages, because the registrations it screens out are lower-paid on average than those that remain, leaving a higher-paid pool.

Under payroll savings the fee instead lowers mean selected pay, by $3,606 to $4,066, or 2.9 percent under current prevailing wages. Two offsetting forces produce that decline. The payment retains the registrations paid furthest below their comparable U.S. wage, which also tend to be offered less in absolute terms, and those retained workers pull the selected mean down. At the same time it pushes out other low-paid fee-subject registrations whose payroll savings cannot clear the $100,000, and their exit pulls the mean back up. The weighting gives the retained low-paid registrations fewer lottery entries, muting the downward pull, but in our simulations the two forces do not fully cancel, leaving the net reduction in Table 5.

Figure 2 traces this split on both prevailing-wage scales, from the random-lottery reference through the wage-weighting step to the payment, which tilts mean selected pay up under parity or down under payroll savings.

In both variants the weighting accounts for most of the rise over the random lottery, so the fee redistributes who is eligible far more than it raises average pay.

The combined effect on the wage-level distribution

The wage weighting would move selected registrants up the wage-level distribution, further under the parity variant than under payroll savings. Figure 3 shows the wage-level composition of selected registrants under the random lottery and under each weighted-with-payment scenario, each scored on its own prevailing-wage scale.

Under the random lottery, selected registrants are spread across the current OFLC levels, with 18 percent at Level IV and 28 percent at Level I. The weighted lottery scored on current prevailing wages lifts the Level IV share to 36 percent under parity and 25.5 percent under payroll savings, while cutting the Level I share to roughly 12 to 16 percent. Scored on the higher 2026 NPRM thresholds, the Level IV share of selected registrants would rise from 7.2 percent under the random lottery to 23.1 percent under parity and 15.3 percent under payroll savings. The gap between the two variants at each level reflects the same mechanism as the compensation results: under parity, fewer fee-subject registrations clear the payment, concentrating selections at the higher wage levels.

Who gets selected under the combined policy

The combined policy would change the composition of selected workers along every margin, and the two variants again give a lower and higher figure for most of those shifts. Table 6 reports the change in selection shares relative to the random lottery.

India’s selection share falls the most, from 66.8 percent under the random lottery to 57.7 percent under the current-PW weighting with the parity variant. Under the payroll savings variant, the drop is only 3.3 points, to 63.5 percent, because more lower-paid Indian registrations survive the fee. The higher 2026 NPRM thresholds would deepen the declines, to 11.1 points under parity (55.7 percent) and 6 points under payroll savings. China and Canada gain under every scenario, together accounting for about half of India’s decline under parity. The doctorate share of selections would rise from 5.7 percent to as much as 9.5 percent, and Computer and Mathematical occupations, which make up 71.3 percent of random selections, would lose up to 5.7 points, chiefly to business, finance, and engineering.

The largest shifts are in prior visa status, shown in Figure 4. Workers with no prior U.S. visa, who are almost entirely the fee-subject group brought in from abroad, fall from 44.5 percent of random selections to between 26.1 and 33.9 percent across the four weighted scenarios. Former F-1 students, who are largely already in the country and exempt from the payment, rise from 43.3 percent of random selections to between 48.5 and 50.8 percent. L-1 transferees, TN professionals, and H-4 dependents gain as well.

The two variants diverge on sex, while the direction of the age shift depends on the wage scale. Under parity, fewer of the older, more male fee-subject workers clear the payment, so women make up a larger share of selected workers, rising from 33.8 percent under the random lottery to 35.7 percent under the current-PW weighting. Under payroll savings, more of those workers survive, and the share of women instead falls, to 31.2 percent under the same weighting. Under the current-PW weighting, parity makes selected workers slightly younger, from 31.7 to 31.5, while the higher NPRM thresholds, by concentrating selections at the top wage level, would make them slightly older, reaching 32.2 under payroll savings. The net direction therefore depends on how much of the payment employers recover and on which prevailing-wage scale applies.

Appendix

Identifying fee-subject registrations

The Proclamation applies the $100,000 payment to petitions for beneficiaries outside the United States, including petitions requesting consular processing, and not to petitions that extend or change the status of a worker already in valid status inside the United States. We flag fee-subject cases using the Form I-129 requested action recorded on FY 2024 approved petitions, treating petitions that request consular notification as subject to the payment. Registrations themselves do not record the requested action, so we work with a synthetic registration pool that resamples the approved petitions up to the FY 2024 registration count and to the share of registrations holding an advanced degree reported by USCIS. Consular cases make up about 50 percent of approved petitions but about 60 percent of this pool, because the registration pool holds a larger share of bachelor’s-degree workers than the approved pool, and consular processing is concentrated among them. The reconstruction assumes that, within each education group, the consular share of registrations matches the share of approved petitions. The data cannot directly confirm this, so the fee-subject share is an approximation of the Proclamation’s legal test.

Table A1 reports the same characteristics for the FY 2024 approved petitions, the raw data the reconstruction draws on. The approved petitions hold advanced degrees far more often than registrations do, so the fee-subject share among them, about 50 percent, is lower than the 60 percent in the reconstructed pool of Table 1.

The fee-survival calculation

We model an employer’s decision to pay the $100,000 as a comparison of the payment with the payroll savings from hiring an H-1B worker rather than a comparable U.S. worker, applying the theoretical framework of Borjas (2026). We estimate each worker’s comparable U.S. wage from a regression on U.S. workers in the 2024 American Community Survey that controls for education, age, sex, occupation, industry, and location, and take payroll savings to be the gap between that predicted wage and the worker’s offered wage. We subtract a small seniority adjustment, because the comparable U.S. workers are more experienced on average than newly hired H-1B workers, and we trim the gap to within plus or minus 50 percent of the offered wage to limit the influence of extreme values. The firm files when those savings, accumulated over a six-year H-1B term and discounted at 3 percent, exceed the $100,000 payment plus an unobserved draw from a normal distribution, a hire-specific cost or benefit that offered wages do not capture. Centered at zero, the draw is as likely to be a benefit as a cost. Because each worker was in fact hired before the payment existed, this term could not have exceeded the payroll savings, so we draw it subject to that bound. The payment is a fixed nominal amount and does not vary with the worker’s wage.

Because the regression proxies experience with age, it applies the U.S. return to experience to a worker’s full career, including any years worked abroad, whose value in the U.S. labor market these data cannot separately identify. This is one of several unmodeled margins that could move the estimated payroll savings in either direction.

The two variants differ only in the payroll savings the firm is assumed to expect. Under parity that expectation is zero, so a registration survives only when the firm values the particular hire, beyond the offered wage, by more than the $100,000 payment. This imposes no view on whether H-1B workers are underpaid relative to comparable U.S. workers, and it is the more selective case. Under the payroll savings variant the firm’s predicted savings enter the comparison, and more registrations clear the payment. Survival is a probability averaged over 1,000 draws, not a determinate outcome for any single registration. See Borjas (2026) for the full derivation of the survival condition.

The three lottery scenarios

The random lottery is applied to all registrations, with no payment screen and no wage weighting, and represents the pre-2026 status quo. The two weighted lotteries are applied to the post-fee eligible pool, which is the fee-exempt registrations plus the fee-subject registrations that survive the payment. The current-PW weighting assigns 1 to 4 lottery entries by OFLC prevailing wage level. The 2026 NPRM weighting assigns entries by the proposed 34th/52nd/70th/88th-percentile levels and, consistent with the proposed rule, drops registrations whose offered wage falls below 95 percent of the NPRM Level I wage before weighting. In every scenario the lottery selects 89,070 registrations, the count carried from our prior brief. It exceeds the 85,000 cap because some selected registrations do not result in a filed petition, and it is calibrated to a registration pool de-duplicated for multiple and ineligible filings rather than to the larger first-round total that USCIS selects to fill the cap.

Legal status

The payment’s legal status remains unsettled. On June 8, 2026, the U.S. District Court for the District of Massachusetts declared the agency policy implementing the $100,000 payment unlawful and vacated it in full, holding among other grounds that the payment is in substance a tax that Congress never authorized the executive to impose. Four days later, the court denied a stay pending appeal but temporarily stayed the vacatur while the government sought relief from the First Circuit. The government filed its stay motion on June 18. The vacatur remains stayed, and the payment remains in effect, while the First Circuit considers that motion. In a separate case, the U.S. District Court for the District of Columbia upheld the Proclamation on December 23, 2025, and the D.C. Circuit heard argument on March 9, 2026 in an appeal that remains pending. Separately from the litigation, the Proclamation’s entry restriction is scheduled to expire on September 21, 2026, absent extension. We analyze the payment’s effect on the lottery as if it remains in force.⁷

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

1. Presidential Proclamation, Restriction on Entry of Certain Nonimmigrant Workers (Sept. 19, 2025), 90 Fed. Reg. 46027 (Sept. 24, 2025). Implementation guidance is on the USCIS website. The payment applies to petitions filed on or after 12:01 a.m. eastern on September 21, 2025, and the restriction expires twelve months after its effective date absent extension. The payment’s legal status is unsettled and is the subject of ongoing litigation; see “Legal status” in the appendix. ↩

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

3. Because registrations do not record whether the worker is abroad, we build a synthetic registration pool by resampling approved petitions, which do, to match the registration count and the share holding an advanced degree. Workers abroad are about 50 percent of approved petitions but 60 percent of the synthetic pool, because such filings are concentrated among bachelor’s-degree workers, who are a larger share of registrations than of approved petitions. The appendix details the full reconstruction. ↩

4. Our fee-survival calculation applies the Borjas (2026) framework with our own wage and cost estimates, similar in approach to the experience-benchmarking regression in our prior brief. Borjas (2026) emphasizes that a fee of this kind does little to change aggregate H-1B demand; our larger screening figures describe only the fee-subject subset and, under the parity variant, assume no payroll savings, the more selective case. The framework has also been the subject of public debate. See the appendix for the full calculation. ↩

5. Compensation is adjusted to FY 2025 dollars using the CPI-U, consistent with our prior briefs. Figures average 1,000 simulated lotteries of 89,070 selected registrations each. ↩

6. The payment and no-payment rows come from two separate weighted-lottery simulations on the same synthetic registration pool, the current run on the post-fee eligible pool and the prior brief’s run on all registrations with no payment screen. They are therefore comparable up to simulation noise, and the marginal effect is the difference between the two at each prevailing-wage scale. ↩

7. California v. Mullin, No. 1:25-cv-13829 (D. Mass.), declared the agency policy implementing the Proclamation unlawful and vacated it in its entirety on June 8, 2026. On June 12, the district court denied a stay pending appeal but granted an administrative stay pending a First Circuit decision on the government’s anticipated stay motion; the government filed that motion on June 18 in No. 26-1699 (1st Cir.), where it remains pending. USCIS’s H-1B guidance, last updated July 17, 2026, continued to require the $100,000 payment. In Chamber of Commerce v. DHS, No. 1:25-cv-03675 (D.D.C. Dec. 23, 2025), the district court upheld the Proclamation; the appeal, No. 25-5473 (D.C. Cir.), was argued on March 9, 2026. Litigation status is as of July 21, 2026. ↩