DOWNLOAD DATA

Main Points

• Allocation mechanisms compared: We consider five possible H-1B visa allocation methods:

  • The current policy rule.
  • The new DHS rule.
  • Three alternative allocation rules that have been recently debated.

• Compensation effects: The new DHS rule increases the average compensation of selected applicants by $9,554 (8.5%), from $112,309 under current policy to $121,863. Of the three alternative rules, one produces a similar increase ($122,504), while two produce much larger increases ($149,682 and $166,778).

• Wage level reallocation: The DHS rule reduces the share of selected registrations at Level I from 27% to 14% and increases the share at Level IV from 15.5% to 26%. The three alternative rules shift the share of selected registrations to Level IV by even larger amounts (between 51.6% and 67.2%).

• Education shifts: DHS rule increases doctorate holders by 1.5 percentage points; the alternative rules increase doctorate holders by over 5 percentage points.

• Nationality and demographic effects: DHS rule reduces India’s share by 2.1 percentage points to 65.5%; the decline is even larger for the other rules (as low as 49.8%). Women’s share declines from 32.5% under current policy to 31.9% (DHS rule); the other rules would lower it even more (to as low as 29.0%).

• Minimal wage effects: The new DHS rule and the other three rules produce small 5-year cumulative wage effects on U.S.-born workers (ranging from -0.04 to +0.01 percentage points), with lower productivity effects largely offsetting direct competition effects.

• Occupation and industry shifts: The DHS rule produces only modest occupational and industry reallocation, including in computer-related occupations. The alternative rules would produce much larger shifts away from computer-related occupations and industries toward business, financial, management, and consulting roles.

Introduction

We estimate the effects of alternative H-1B visa allocation mechanisms using stochastic simulations comparing each alternative to the current random lottery from 2021 through 2024. The study compares five allocation approaches, including the current random lottery, the new DHS rule, and three alternative methods that have been discussed in policy circles.

• Current Policy — Random Lottery: All eligible registrations have equal selection probability. Registrations are selected randomly until the statutory numerical cap is reached.

• New DHS Rule — Wage Level Weighted: Weighted selection process based on the beneficiary’s equivalent wage level—the highest OEWS wage level that the offered wage would equal or exceed. Registrations at wage level IV are entered into the selection pool four times, those at level III are entered into the pool three times, those at level II are entered into the pool two times, and those at level I are entered into the pool one time. The weighting process specifically considers the actual wage offered: a registration may be assigned a higher equivalent wage level for selection-weighting purposes based on prevailing wages for the corresponding SOC code and area, even if the job would otherwise be classified at a lower OEWS level for LCA purposes.

• Alternative 1 — Compensation Weighted: Selection probability is proportional to offered annual compensation, which is more holistically measured than wages, and includes value of benefits. Compensation is adjusted for regional price parity (RPP) by metropolitan area to account for geographic differences in the cost of living.

• Alternative 2 — Wage Level Ranked: Deterministic ranking by equivalent wage level. Registrations are sorted and selected from the highest wage level to lowest, beginning with wage level IV and proceeding in descending order with wage levels III, II, and I. This approach was put forward by DHS in 2021 as a final rule but was subsequently withdrawn.

• Alternative 3 — Compensation Ranked: Deterministic ranking by RPP-adjusted offered annual compensation. Selection proceeds from highest offer to the lowest until the cap is reached.

Under all scenarios, approximately 90,000 registrations are selected each year (in practice, slightly more than the statutory cap of 85,000 are selected). All scenarios preserve the current two-stage process, in which 65,000 new visas are allocated to the general category and 20,000 to the advanced degree exemption category. The Appendix provides additional details on the methodology.

Wage Levels vs. Compensation

Wage levels are set according to “prevailing wages” at different points of the wage distribution within each occupation and geographic area, based on wages measured in the Occupational Employment and Wage Statistics (OEWS). Level I (approximately 17th percentile) shown in Figure 1 corresponds to “entry-level” positions, Level II (34th percentile) to “qualified” workers, Level III (50th percentile, or median) to “experienced” workers, and Level IV (67th percentile) to “fully competent” workers. These levels are determined by comparing a position’s education, training, and experience requirements to the normal requirements for the occupational classification.1

There is considerable overlap in the actual compensation distributions of workers assigned to different wage levels. Many workers classified at Level I earn compensation comparable to or higher than workers at Level II or even Level III. This overlap indicates that assigned wage level does not map cleanly to actual compensation—workers with similar earnings may be classified at different levels depending on occupation, stated requirements of their positions, and employer compensation practices.

While wage levels are based on position requirements rather than individual qualifications or compensation, the New DHS Rule uses equivalent wage level (the highest level that offered compensation equals or exceeds) as a proxy for worker skill and value in the weighted lottery selection process. This approach means the probability of selection depends in part on the value of the worker to the prospective employer—as reflected in the offered pay—but the substantial overlap across wage levels in Figure 1 implies that this relationship is relatively weak.

Effects of the New DHS Rule and Alternatives

Wage Levels

As expected, the new DHS rule shifts selection toward wage levels III and IV registrations. Level I registrations make up 14% of selected registrations, compared to 27% under the current random lottery—a decline of 13 percentage points. Level IV registrations rise from 15.5% to almost 26%, an increase of 10.5 percentage points.

Alternative 1—a lottery in which selection probabilities are proportional to offered compensation—also shifts selection toward higher wage levels, but significantly less than the DHS rule. Under Alternative 1, Level I registrations remain higher than under the DHS rule because Level I registrations in higher-paying occupations may have compensation offers that are high relative to the overall pool despite the low wage level classification.

Ranking by compensation under Alternative 3 allocates 51.6% of new visas to level IV registrations. The likelihood of a level I registration being selected is roughly one-tenth as large as in the random lottery (2.5% vs 27.1%). Ranking by wage level under Alternative 2 allocates 67.2% of new visas to level IV registrations and most of the remainder to level III. Under Alternative 2, zero level I registrations would be selected.

Compensation

Current policy results in an average compensation of $112,309 for selected new H-1Bs (in 2025 dollars).

Both the DHS wage level weighting rule and Alternative 1 (the two weighted lotteries) raise average compensation by roughly $10,000 to around $122,000. Specifically, the DHS rule increases average compensation to $121,863 (an increase of $9,554), while Alternative 1 increases it to $122,504 (an increase of $10,195).

The compensation increase under the DHS rule is consistent across the four lottery years analyzed, though with some annual variation. In 2021, average compensation under the random lottery was $118,545 and would have been $127,684 under the DHS rule (an increase of $9,139). In 2024, the random lottery resulted in average compensation of $111,359, rising to $122,436 under the DHS rule (an increase of $11,077).

Alternative 3 boosts average compensation by $54,469 to $166,778—the largest increase among all scenarios, since by definition it chooses the highest paid workers. Alternative 2 results in a smaller increase to $149,682 (an increase of $37,373), because some higher level jobs in low-paying occupations are selected over lower level jobs that have higher absolute pay.

Median compensation is generally around $10,000 lower than the average across all scenarios.

The New DHS Rule and Alternative 1 shift the whole distribution upward and increase dispersion to a similar degree. The interquartile range (IQR) rises from $44,000 under current policy to about $52,000 under both the New DHS Rule and Alternative 1, mainly due to an increase at the upper end of the distribution.

Education Thresholds

Under current policy, the education mix of selected new H-1Bs is roughly evenly split between Bachelor’s degree or less (about 46%) and Master’s degrees (about 48%), with Doctorates around 5% and Professional degrees about 1%.

The New DHS Rule modestly shifts selection toward higher credentials: the share of Doctorate holders rises by 1.5 percentage points, mostly offset by a decline in Master’s degree recipients (down 1.32 percentage points), while Bachelor’s-and-below remains nearly unchanged (down 0.31 percentage points). Professional degree holders increase by 0.14 percentage points.

Alternative 1 has similarly modest effects but results in a smaller shift towards Doctorates (up 1.04 percentage points), a smaller decline in the Master’s share (down 0.63 percentage points), and larger decline in the lowest education levels (Bachelor’s-and-below down 0.72 percentage points). Professional degrees increase by 0.31 percentage points.

The largest changes occur under ranking-based approaches. Alternative 2 and Alternative 3 both raise the Doctorate share to roughly 11%—more than twice the share under the random lottery. Specifically, ranking by wage level increases the Doctorate share by 5.72 percentage points, while compensation ranking increases it by 6.25 percentage points.

Alternative 2 primarily reduces the share of selected beneficiaries with Master’s degrees (down 5.08 percentage points), while also reducing Bachelor’s-and-below (down 1.21 percentage points). Alternative 3 leads to the greatest decline in Bachelor’s-and-below recipients (down 4.93 percentage points), while also reducing Master’s degrees (down 2.57 percentage points).

F-1 Student Visas

Under current policy, 47.6% of selected new H-1Bs are conversions from F-1 student status.

The New DHS Rule reduces the F-1 share by 1.5 percentage points to 46.1%. Alternative 1 also lowers it but only by 0.6 percentage points to 47.1%.

Alternative 2 reduces F-1 conversion substantially to 43.8% of selected registrations—a decline of 3.8 percentage points. By contrast, Alternative 3 increases the F-1 conversion share to 48.3%—an increase of 0.7 percentage points. This large difference between the two ranking approaches reflects the different outcomes for F-1 students seeking visas for high-paying entry level jobs in occupations where compensation at all levels is high relative to most jobs.

Occupation

Under the random lottery, selection is highly concentrated in Computer and Mathematical Occupations, which account for 71.9% of selected registrations. The next largest groups—Architecture and Engineering (6.1%) and Business and Financial Operations (6.3%)—are each an order of magnitude smaller.

Wage level weighting (the New DHS Rule) produces only modest occupational reallocation. Computer and Mathematical Occupations decline slightly to 71.7% (down 0.20 percentage points). Architecture and Engineering declines by 0.50 percentage points, while Business and Financial Operations increases by 0.24 percentage points and Management Occupations increases by 0.06 percentage points.

Compensation weighting (Alternative 1) slightly increases the Computer and Mathematical share to 72.6% (up 0.69 percentage points). Management Occupations increase notably under Alternative 1 (up 0.76 percentage points), while Business and Financial Operations decreases (down 0.64 percentage points).

Ranking-based approaches generate much larger shifts. Alternative 2 reduces the Computer and Mathematical share by 5.74 percentage points to 66.2%, reallocating selection primarily toward Business and Financial Operations (up 2.81 percentage points) and Management occupations (up 0.75 percentage points). Arts, Design, Entertainment, Sports, and Media increases by 1.04 percentage points, and Life, Physical, and Social Science increases by 0.81 percentage points.

Alternative 3 also reduces Computer and Mathematical occupations by 2.82 percentage points to 69.1%, and substantially increases Management occupations (up 4.46 percentage points) since these are generally the highest paid jobs. Architecture and Engineering declines by 1.38 percentage points under Alternative 3.

Under the random lottery, 83.6% of selected registrations are for STEM occupations. Weighting approaches (both the DHS rule and Alternative 1) have minimal impact on STEM concentration—the share remains essentially unchanged.

Ranking approaches reduce STEM shares significantly: Alternative 2 reduces the STEM share by 6.3 percentage points to 77.3%, while Alternative 3 reduces it by 4.3 percentage points to 79.3%. This occurs because they shift selected registrations either toward business and financial occupations or toward management.

Industry

Under the random lottery, selection is dominated by NAICS 5415 (Computer Systems Design and Related Services), which accounts for 49.6% of all selected registrations among classified industries. The remaining top industries each account for only a few percent: Management, Scientific, and Technical Consulting Services (5.5%), Architectural, Engineering, and Related Services (4.6%), Electronic Shopping and (3.0%), and Software Publishers (2.7%).

Wage level weighting (the New DHS Rule) slightly reduces concentration in NAICS 5415 by 3.18 percentage points to 46.4%. Management, Scientific, and Technical Consulting Services increases by 0.54 percentage points. Electronic Shopping increases by 0.54 percentage points. Securities and Commodity Contracts Intermediation and Brokerage increases by 0.37 percentage points. Software Publishers increases by 0.34 percentage points. Architectural, Engineering, and Related Services decreases by 0.71 percentage points.

Compensation weighting (Alternative 1) has a similar effect to the DHS rule, reducing NAICS 5415 by 2.31 percentage points to 47.3%. The pattern of increases and decreases across other industries is similar to the DHS rule, though generally of slightly smaller magnitude.

Ranking approaches dramatically reduce concentration in NAICS 5415. Wage level ranking cuts its share to 32.0% (down 17.59 percentage points), with corresponding increases in Management, Scientific, and Technical Consulting Services (up 3.36 percentage points), Electronic Shopping and Mail-Order Houses (up 1.80 percentage points), Securities and Commodity Contracts (up 1.73 percentage points), Computer and Peripheral Equipment Manufacturing (up 1.74 percentage points), and Software Publishers (up 1.20 percentage points).

Compensation ranking reduces NAICS 5415 further to 25.7% (down 23.97 percentage points), with the largest increases in Electronic Shopping (up 4.74 percentage points), Management, Scientific, and Technical Consulting Services (up 3.17 percentage points), Computer and Peripheral Equipment Manufacturing (up 3.19 percentage points), Software Publishers (up 2.88 percentage points), and Computing Infrastructure Providers (up 2.51 percentage points).

Nationality

Under the random lottery, selection is heavily concentrated among nationals of India (67.6% of selected registrations), followed by China (about 13%), with all other countries individually accounting for only a few percent or less. The next largest source countries are Canada (1.7%), Korea (1.6%), Taiwan (1.3%), Mexico (1.3%), Philippines (1.2%), Pakistan (1.0%), and Nepal (0.9%).

The New DHS Rule modestly reduces India’s share by 2.14 percentage points to 65.5%. Canada’s share increases by 0.90 percentage points to 2.6%, China’s increases by 0.28 percentage points to approximately 13.3%, and Mexico’s increases by 0.24 percentage points.

Alternative 1 has similar effects to the DHS rule, reducing India’s share by 1.66 percentage points to 66.0%. Canada’s share increases by 0.84 percentage points and China’s increases by 0.30 percentage points.

Ranking approaches substantially reallocate selection away from India. Alternative 2 reduces India’s share to 55.2% (down 12.43 percentage points), with corresponding increases in Canada (up 4.07 percentage points), China (up 1.94 percentage points), Mexico (up 1.13 percentage points), and Brazil (up 0.61 percentage points).

Alternative 3 reduces India’s share further to 49.8% (down 17.84 percentage points)—just below half of all selected registrations. This represents the most dramatic reallocation across nationalities. Canada’s share increases by 5.79 percentage points to 7.5%, China’s increases by 5.45 percentage points to approximately 18.5%, Mexico’s increases by 0.98 percentage points, and Brazil’s increases by 0.58 percentage points.

Demographics

Under the random lottery, selected new H-1Bs average 31.3 years of age, with roughly three-quarters (75.2%) aged 35 or under, and 32.5% women. The share of F-1 student visa conversions is 47.6%.

Wage level weighting (the DHS rule) slightly increases the average age by 0.4 years to 31.7 and reduces the share aged 35 and under to 73.2% (down 2.0 percentage points). The share of women decreases to 31.9% (down 0.6 percentage points). The F-1 conversion share decreases to 46.1% (down 1.5 percentage points).

Alternative 1 has similar effects to the DHS rule, increasing age by 0.2 years to 31.5 and reducing the share aged 35 and under to 74.1% (down 1.1 percentage points). The share of women decreases to 31.5% (down 1.0 percentage point). The F-1 conversion share decreases to 47.1% (down 0.6 percentage points).

Ranking approaches produce much larger demographic shifts. Alternative 2 raises the average age by 1.5 years to 32.8 and reduces the share aged 35 and under to 67.4% (down 7.8 percentage points). The share of women decreases to 30.6% (down 1.9 percentage points). The F-1 conversion share decreases to 43.8% (down 3.8 percentage points).

Alternative 3 increases average age by 0.8 years to 32.1, and reduces the share aged 35 and under to 71.6% (down 3.6 percentage points). Alternative 3 produces the largest decline in the share of selected beneficiaries who are women to 29.0% (down 3.5 percentage points). The F-1 conversion share increases to 48.3% (up 0.7 percentage points).

Effects on Wages of U.S.-Born Workers

Wage effects are calculated using a model developed in a forthcoming Penn Wharton Budget Model working paper. The approach is similar to Caiumi and Peri (2024), which estimates wage effects from changes in immigrant worker supply accounting for substitution and complementarity relationships between different worker types. The model extends this framework to distinguish between STEM and non-STEM workers, and incorporates the effects on total factor productivity (TFP) growth resulting from changes in the STEM workforce.

The estimated wage effects of alternative H-1B allocations are quite small—generally ranging from -0.06 to +0.01 percentage points over 5 years. These effects reflect the net impact of two offsetting forces: (1) relative wage changes driven by substitution and complementarity effects between different types of workers, and (2) economy-wide productivity changes that affect all workers.

Both U.S.-born and foreign-born STEM workers would experience small relative wage gains from reduced competition in scenarios that select fewer STEM H-1Bs. However, these gains are generally offset—and in most cases reversed—by negative effects on TFP growth. Under every allocation except compensation weighting, the reduction in STEM workers selected leads to lower TFP growth, which reduces wages for all workers.

The New DHS Rule

For all U.S.-born workers, the 5-year cumulative wage effect of the DHS rule is <0.01 percentage points—essentially no effect. U.S.-born workers with no college degree experience <0.01 percentage points effect. U.S.-born college-educated workers in non-STEM occupations experience -0.01 percentage points effect. U.S.-born college-educated workers in STEM occupations experience <0.01 percentage points effect.

For all foreign-born workers, the effect is -0.01 percentage points. Foreign-born workers with no college degree experience <0.01 percentage points effect. Foreign-born college-educated workers in non-STEM occupations experience -0.01 percentage points effect. Foreign-born college-educated workers in STEM occupations experience -0.01 percentage points effect.

Alternative 1: Compensation Weighted

Compensation weighting produces small positive wage effects for most worker groups. For all U.S.-born workers, the effect is 0.01 percentage points. U.S.-born workers with no college degree experience 0.01 percentage points effect. U.S.-born college-educated workers in non-STEM and STEM occupations experience <0.01 percentage points effect.

For all foreign-born workers, the effect is <0.01 percentage points. Foreign-born workers with no college degree experience 0.01 percentage points effect. Foreign-born college-educated workers in non-STEM occupations experience <0.01 percentage points effect. Foreign-born college-educated workers in STEM occupations experience -0.01 percentage points effect.

Alternative 2: Wage Level Ranked

Wage level ranking produces small negative wage effects across most worker groups due to TFP effects dominating. For all U.S.-born workers, the effect is -0.04 percentage points. U.S.-born workers with no college degree experience -0.03 percentage points effect. U.S.-born college-educated workers in non-STEM occupations experience -0.05 percentage points effect. U.S.-born college-educated workers in STEM occupations experience -0.04 percentage points effect.

For all foreign-born workers, the effect is -0.04 percentage points. Foreign-born workers with no college degree experience -0.03 percentage points effect. Foreign-born college-educated workers in non-STEM occupations experience -0.06 percentage points effect. Foreign-born college-educated workers in STEM occupations experience -0.02 percentage points effect.

Alternative 3: Compensation Ranked

Compensation ranking produces small negative wage effects for most groups, though slightly smaller in magnitude than wage level ranking. For all U.S.-born workers, the effect is -0.01 percentage points. U.S.-born workers with no college degree experience -0.01 percentage points effect. U.S.-born college-educated workers in non-STEM and STEM occupations experience -0.01 percentage points effect.

For all foreign-born workers, the effect is -0.01 percentage points. Foreign-born workers with no college degree experience -0.01 percentage points effect. Foreign-born college-educated workers in non-STEM occupations experience -0.02 percentage points effect. Foreign-born college-educated workers in STEM occupations experience 0.01 percentage points effect—the only worker group experiencing a positive wage effect under compensation ranking.

Appendix: Methodology

We use H-1B lottery registration data sourced from USCIS and obtained by Bloomberg through Freedom of Information Act requests. Our approach builds on prior work by the Institute for Progress and the Economic Innovation Group.

All simulation results are based on 1,000 simulations of each scenario for each lottery year from 2021 to 2024. Results presented throughout this analysis are generally the average across all simulations and across all four lottery years, unless otherwise noted.

The equivalent wage level determination considers actual compensation offered compared to prevailing wages for the corresponding SOC code and geographic area. A registration may be assigned a higher equivalent wage level for selection-weighting purposes than it would be classified for LCA purposes based on this comparison.

The two-stage allocation process is preserved under all scenarios analyzed. Under this process, around 65,000 new visas are allocated to the general category and 20,000 to the advanced degree exemption category, resulting in approximately 90,000 total selected registrations annually (in practice, slightly more than the statutory cap of 85,000 are selected to account for potential withdrawals and denials).

Compensation adjustments for regional price parity (RPP) are based on Bureau of Economic Analysis data by metropolitan area. This adjustment accounts for geographic differences in the cost of living, ensuring that compensation comparisons reflect real purchasing power rather than nominal wage levels.

The wage effects model extends the framework of Caiumi and Peri (2024), which estimates wage effects from changes in immigrant worker supply accounting for substitution and complementarity relationships between different worker types. The model is detailed in a forthcoming Penn Wharton Budget Model working paper. Key features include:

• Distinction between STEM and non-STEM workers
• Incorporation of total factor productivity (TFP) effects from changes in the STEM workforce
• 5-year cumulative wage effects accounting for both direct competition effects and economy-wide productivity effects
• Worker categories by nativity (U.S.-born vs. foreign-born), education level (no college degree vs. college degree), and occupation type (STEM vs. non-STEM) among the college-educated workers

This analysis was produced by PWBM staff under the direction of Alex Arnon. Duncan Haystead contributed to the analysis. Mariko Paulson prepared the brief for the website.

Media (only): For the fastest response, email us at inquiries-pwbm@wharton.upenn.edu.
All other responses: Please use our Contact Us.
Sign up for PWBM Breaking News, Alerts and Newsletter. Unsubscribe anytime.