High-Skilled Immigration
September 4th 2026

Explaining the Proposed $103k H-1B Fee

A Q&A on the legal basis and likely effects of USCIS’s proposal
September 4th 2026

USCIS’s new NPRM, “Fee for Certain H-1B Petitions,” proposes a new fee of $103,265 on all cap-subject H-1B petitions. The fee is intended to recover all the costs for the legal immigration system and would reimburse services unrelated to the adjudication of H-1Bs. DHS says 64% of the revenue will go to other agencies.

This is the second attempt in a year to put a six-figure price on an H-1B, extending the fee to cover international graduates of US schools seeking to stay after they graduate. Unlike the $100,000 payment required by Presidential Proclamation 10973 (currently vacated by the courts), this fee applies to people already in the United States, including those on student visas.

Given that demand for H-1Bs far exceeds the 85,000 cap, a higher fee on cap-subject H-1Bs could help prioritize top talent for whom employers are willing to pay. But multiple analyses suggest that the $103k fee is too high and will hinder recruitment and retention of top talent.    

This Q&A reviews the proposal’s likely effects and legal basis, arriving at two key findings:

  • A $103k fee would suppress H-1B demand, undermining the American economy. Five analyses using three methods all find that the fee would cut H-1B demand by 63–91%; four of the five predict that the 85,000 cap would go undersubscribed. USCIS assumes filings will not fall at all and expresses no uncertainty. Its own analysis of demand sensitivity in the Technical Appendix is riddled with errors, rendering the results completely meaningless. 
  • The fee relies on mechanisms prohibited by fiscal law. USCIS’s fee authority through the Immigration Examinations Fee Account (IEFA) extends to recovering the costs of adjudication and naturalization services. The proposed fee exceeds that authority, aiming to recover all costs related to administering and enforcing the legal immigration system across numerous agencies.1 Using H-1B fees to fund work by another agency that receives appropriations augments that agency’s budget from a source Congress never authorized, violating fiscal law.

Public comments on the NPRM will be accepted by DHS through September 24, 2026.

What is the proposed H-1B fee?

What is the new H-1B fee that USCIS is proposing?

USCIS’s new NPRM, “Fee for Certain H-1B Petitions” proposes an additional fee of $103,265, required of every H-1B cap-subject petition, claiming authority under sections 286(m)-(n) of the Immigration and Nationality Act.2 This would be on top of currently required fees, which range from $2,225-7,595 depending on the employer. For most cap-subject H-1B petitions of employers with more than 25 full-time employees, required filing fees currently total $3,595.

The new fee would apply to all 85,000 initial H-1B petitions filed for workers selected under either the annual regular (65,000) H-1B cap or the advanced-degree exception (20,000) to the annual H-1B cap.

How does this differ from last year’s proposed $100,000 payment? 

Presidential Proclamation 10973 (Restriction on Entry of Certain Nonimmigrant Workers)

requires a $100,000 payment as a condition of eligibility for certain H-1B petitions filed from September 21, 2025 to September 21, 2026. This payment is not currently being collected as it was vacated by the US District Court in June 2026. 

Unlike the Presidential Proclamation payment,

  • the proposed fee does not apply to cap-exempt organizations; and
  • the proposed fee does apply to all 85,000 initial cap-subject petitions, not only those H-1Bs first entering the country to obtain status. 

The Presidential Proclamation payment only applied to entries, sparing those seeking changes of status from within the United States, most commonly international students. The proposed fee does not spare them. 

What is the stated goal of the proposed fee? Where would the money go?

Per the NPRM, the goal of DHS’s new $103,265 fee is to recover $8.8 billion in costs associated with the administration of the lawful immigration system.

According to the NPRM (p. 54832), about 64% of the costs being captured in the new $103,265 fee are incurred outside of USCIS. The costs identified are divided between USCIS, Customs and Border Protection (CBP), Immigration Customs and Enforcement (ICE), the Department of Justice’s Executive Office of Immigration Review (EOIR), the Department of State (DOS), and the Department of Labor (DOL).

DHS also asserts that the fee will have the “indirect benefit of better protecting the wages and job opportunities of U.S. workers.” 

How does the proposed fee compare to historical USCIS fees? 

As detailed later in this piece, the proposed fee differs from historical fees in at least six distinct ways, including: 

  1. Magnitude. USCIS is proposing raising the required H-1B fees by more than an order of magnitude across all mandatory fees charged to H-1B employers (for registration, base filing, training, fraud, and asylum support). For a typical employer with more than 25 employees, the fee will increase from $3,595 to $106,860. This new total is ~175 times USCIS’s own modeled cost of adjudicating an H-1B petition.3 The $8.78 billion target exceeds the $7 billion the entire Immigration Examinations Fee Account (IEFA)  collected in FY 2025 across every immigration benefit, so one fee on one form for one sub-population would raise more than USCIS’s entire current fee schedule combined.
  2. Funding law enforcement components at DHS. The rule would route $1.13 billion a year to other DHS components, namely ICE and CBP, which have never been funded out of the IEFA because they do not provide adjudications services, the purpose of the account. DHS notes in the NPRM that petitioners will be funding “non-USCIS administered programs to which they have no connection or from which they receive no direct benefit.”
  3. Funding agencies outside DHS. More than half of the DHS’s expected revenue from the fee (53%, or $4.65 billion) would leave DHS entirely, going to EOIR at DOJ, DOL, and DOS. DHS acknowledges that DOL and DOS “have not independently used the authority in section 286(m)” to charge for the listed costs, so the proposed rule has USCIS collecting funds on their behalf under a statute they have never invoked themselves. The reimbursement agreements that would move the money don’t exist yet — DHS says their terms “would be determined before a final rule,” meaning the public can only comment on the total, and not the mechanism.
  4. Methodology. The new fee departs from Activity-Based Costing, a methodology which even USCIS concedes “reflect[s] best practices” and has been the standard for determining immigration benefit request fees for both USCIS and its predecessor agency Immigration and Naturalization Service (INS) since 1998.4 Instead, the new model assembles a $8.78 billion revenue target based on a review of budget costs. USCIS also departs from long-standing practice by failing to provide cost estimates under the Federal Accounting Standards Advisory Board’s three-tier cost hierarchy (SFFAS No.4).
  5. Augmentation of appropriated funds. The rule would route billions of dollars a year to federal agencies Congress funds through appropriations. DHS’s preamble concedes it has never before set a fee to fund “costs that have been generally funded by a Congressional appropriation.” 
  6. Funding for activities that are not for adjudication and naturalization services. Sections 286(m) and (n) of the INA only authorize fees for “providing immigration adjudication and naturalization services” and reimbursement of appropriations for “expenses in providing immigration adjudication and naturalization services.” However, DHS notes that it “is not setting this fee to be associated with or required for providing a specific service,” and billions of the proposed allocation would fund activities that cannot be classified as adjudication services.

What fees are currently required to file an initial, cap-subject H-1B petition?

To lodge an H-1B petition with USCIS, sponsoring employers are required to submit payment for some combination of the following fees, depending upon case- and petitioner-specific eligibility requirements:

How would the fee affect the H-1B program? What would that mean for America?

Is USCIS trying to deter H-1B filing?

That would cut against DHS’s own stated objective to not let the volume of cap-subject H-1Bs decline. 

According to DHS, the purpose of the rule is to recover revenue sufficient “to recover the lawful immigration-related costs attributed to DHS and other federal agencies” through a fee on initial, cap-subject H-1B petitions. The Department notes that “the success of the USCIS fee model and this rulemaking in generating the necessary revenue depends on the filing volumes of cap-subject H-1B petitions not falling short of those projected herein.”

The NPRM assumes that the cap-subject portion of the H-1B program would remain fully subscribed despite the $103k fee. If the Department does not meet the 85,000 visa cap annually as they expect, the fee would raise just a fraction of the revenue DHS says it needs to cover the costs of the immigration system. Further, if the fee exceeds the revenue-maximizing level, DHS may only be able to recover its estimated costs by instituting a lower (and potentially much lower) fee on cap-subject H-1B petitions.

Would the fee deter H-1B filing?

Despite USCIS counting on no change in H-1Bs with the introduction of a fee, most evidence suggests that approved H-1Bs would decline and that a lower fee would allow USCIS to raise more revenue and recover a greater share of costs.

Five methods using three different techniques (wage gap methodologies, a historical case study, and survey evidence) unanimously conclude that a $100,000 fee would lead to a sharp decline in H-1B demand; four of them find that the fee would also lead to a decline in H-1Bs.

  1. Borjas (2026): The NPRM’s assumption of no change in filing volume rests on a 2026 working paper from economist George Borjas. Finding a -16% wage gap between Americans and similar H-1Bs, Borjas uses it to simulate employer demand and identify a revenue-maximizing H-1B fee. The latest revision of his paper concludes that a fee between $95,000–$152,000 would maximize the revenue generated from fees on H-1B petitions (although the NPRM cites Borjas’s earlier finding of $100,000–$200,000). The latest revision of Borjas’s paper contradicts the NPRM’s confident claim that Borjas’s “simulation exercise showed that the demand for H-1B visas will not go below 85,000 for fees above $100,000,” since his revised range for the revenue maximizing fee is consistent with a reduction in filing. 

    Nevertheless, at $100,000, under Borjas’s preferred specification, demand for H-1B registration falls 63%. Given recent demand, that translates to roughly 89,000 petitions, which would narrowly avoid a reduction in cap-subject petitions.5 
  2. Ozimek and He (2026): Borjas’s paper drew technical criticism from several economists who found smaller wage gaps between foreign workers on H-1B visas and comparable native-born workers. Jiaxin He and Adam Ozimek reported a -5% wage gap after correcting for errors they found in Borjas’s paper. Using Borjas’s simulation method with Ozimek and He’s calculated gap, we find that a $100,000 fee would cut approved cap-subject H-1Bs by 13%.
  3. Clemens (2026): Economist Michael Clemens also disputed Borjas’s paper. After multiple corrections of Borjas’s analysis, Clemens finds a +6% wage premium, concluding that the typical H-1B worker earns 6% more than comparable native workers doing similar jobs. Using Borjas’s simulation method with this estimated premium, a $100,000 fee cuts H-1B registration by 78%, reducing approved cap-subject H-1Bs by 38%.

    Both the 13% decline found by Ozimek and He and the 38% decline found by Clemens are conservative, because these estimates borrow one further parameter from Borjas that softens the resulting decline in petitions.6
  4. Proclamation Payment case study: The September 2025 proclamation imposing a $100,000 payment on new H-1B petitions for workers outside the United States, which remained in effect for eight months before a federal court vacated it in June 2026, provides the only real-world test of employer behavior in response to a $100,000 price. The NPRM documents the effects of the proclamation in USCIS’s Technical Appendix (Table A.7). DHS’s Fee Elasticity Analysis reports that initial consular H-1B receipts for workers outside the country fell 91.2%, with FY 2027 registrations at the largest consular-reliant employers down 80–100%. Taking our estimate of committed demand as 243,000, a 91% decline in demand means a $100,000 fee cuts H-1B petitions by 75%. DHS dismisses this decline’s relevance to how the current proposal would play out, arguing it was an unrepresentative “short term shock.” However, the same data undermines this claim, showing that the year-over-year decline in petition volume grew each month after the imposition of the fee, from 81% in the first full month up to over 97% in the last month that the fee was in effect. 
  5. IFP survey: In July 2026, IFP conducted a survey of organizations that regularly sponsor H-1B workers, including many of the largest H-1B filers, on how a universal $100,000 payment would change their filing volume. The companies’ median expected reduction was 77.5%, which would translate into a 36% reduction in cap-subject H-1B visa numbers utilized.

Each of these analyses projects a large decline in H-1B demand, ranging from 63% to 91%. The potential decline in actual H-1B petitions filed and approved is uncertain because H-1Bs will continue to be fully subscribed unless demand falls by more than 65%, the break-even point assuming excess demand resembles FY 2026’s pattern. In FY 2026, registrations were 343,981, with a 71% selection-to-issuance rate (about 120,000 selections were necessary to generate a yield of 85,000 cap-subject petition approvals). All but one of these analyses suggests a decline in demand above the 65% break-even point, and hence a decline in petitions. If true, the fee will raise much less revenue than USCIS expects. 

In short, DHS’s revenue estimates are highly sensitive to a highly uncertain estimate of how H-1B filing would be affected by the new fee, and the NPRM ignores this uncertainty by citing an outlier. 

The fall in demand would be even higher than the 63–91% decline noted above if DOL finalizes its rule revising prevailing wage requirements, which the NPRM concedes “may affect the overall demand for employment-based immigration benefits.” The NPRM’s confident assertion that the NPRM “accounts for the impact of revised prevailing wage requirements on employer behavior” is not justified by any information provided in the NPRM.   

DHS has not accounted for any of this uncertainty in its NPRM. OMB Circular A-4 directs agencies to characterize uncertainty in projected benefits and costs of new rules and, at minimum, to provide a numerical sensitivity analysis showing how results vary with plausible changes in key inputs. The Department published no such sensitivity analysis in this NPRM, despite relying on an estimate at the extreme end of the plausible range. 

Why does USCIS claim that the fee won’t deter H-1B filing? What’s wrong with their analysis?

USCIS’s assumption that the volume of H-1B petitions will not change is based entirely on Borjas’s simulation method from an outdated version of his paper. Nevertheless, USCIS’s regulatory impact analysis examines the elasticity of demand for H–1B receipts to price as a sanity check, concluding that employer demand for H-1Bs is inelastic. Unfortunately, this analysis is deeply flawed, both theoretically and methodologically, completely invalidating its results. 

First, the analysis is theoretically unsound: an elasticity estimated from a small fee increase when the total fees are low is not informative about the impact of a fee hike two orders of magnitude larger. The effect of a $920 fee increase on receipts may be informative about the local elasticity, but is not informative about the elasticity at a $103,265 fee increase. Marshall’s Second Law of Demand states that elasticity should increase with price, so even if demand is inelastic to a $920 increase, that does not indicate demand is inelastic to a $103,265 increase. 

Second, the analysis is methodologically unsound, with multiple technical errors that render the result meaningless:

  1. USCIS uses the wrong price variable, completely invalidating their results. The quantity of interest is the price elasticity of receipts to fee increases, using the per-H-1B fee for an employer in a given year. But instead of using the per-visa fee, USCIS uses the sum total of all H-1B fees paid by that employer that year (total number of H-1Bs from an employer multiplied by the average fee paid per H-1B).7 This means that the count of receipts sits on both sides of the regression, which leaves them estimating simply how much revenue increases with volume. This error completely invalidates the results — what the model identifies is close to an accounting identity. A company that files 100 petitions pays approximately 100 times the fees of a company that files one, no matter the elasticity. This procedure will mechanically always produce a coefficient close to 1, regardless of whether employers are highly sensitive to price or entirely insensitive to it. Sure enough, for each of the five years, the authors estimate values of between 0.94 and 0.97, but USCIS incorrectly identifies it as the elasticity of demand. 
  2. USCIS reports values inconsistent with the model they describe. USCIS also estimates a “long-run” elasticity that uses data from FY2021-2025. This allows them to add time and petitioner fixed effects to account for variation across different years and different employers. However, the reported AIC and BIC values in table A.5 are inconsistent with the addition of these fixed effects, which means the model reported in Table A.5 is not the model described in Equation 2.8 This makes their results entirely unreliable.
  3. USCIS misreads its coefficient. The resulting “long-run” elasticity estimate USCIS reports, inclusive of the errors described above, is 0.62.9 The authors misinterpret this value and write “as this value is less than 1, we can conclude that fee elasticity of demand for Form I-129 H-1B petitions is inelastic.” However, an inelastic response should actually correspond to an elasticity between -1 and 0. Positive elasticities would mean that higher visa fees make employers more interested in filing visas, not less. This is nonsensical.
  4. USCIS presents an incorrect specification for its negative binomial regression. This regression takes employers as the unit of interest and estimates a model where the expected number of receipts is a function of visa fees and firm characteristics. USCIS presents a regression specification that includes the observed count of receipts rather than the expected count, and includes an additive error term. Both of these are incorrect for a negative binomial model.10

What would the fee mean for the US economy?

The NPRM’s proposed fee would represent a massive tax on one of America’s largest and most important high-skilled visa programs. The H-1B program is the largest channel through which international students graduating from American universities remain in the United States after leaving F-1/OPT status. H-1B visa holders also account for a large majority of employment-based green card recipients. 

Since FY 1997, employers have consistently met the H-1B program caps before the end of the fiscal year (except for FY 2001-2003, when Congress temporarily tripled the cap). If the proposed fee suppresses demand more than DHS expects — which would be consistent with every piece of evidence publicly available besides the Borjas paper — DHS will leave the 85,000 annual cap undersubscribed. If the decline in demand mirrors how employers responded to the $100,000 fee imposed by the September 2025 proclamation, the proposed fee will effectively destroy a key segment of the international STEM talent pipeline. 

Americans would ultimately bear the costs of a much smaller H-1B program. Seminal research from economists Giovanni Peri, Kevin Shih, and Chad Sparber finds that the inflow of STEM workers through the H-1B program can explain 30–50% of all aggregate productivity growth in the US between 1990 and 2010. Sustained cuts to this key slice of the high-skilled STEM talent pipeline would lead to a smaller American economy, both in aggregate and on a per-capita basis. One National Academies paper finds that a one-third decline in foreign graduates from American universities would shrink the American economy by between $220 billion and $439 billion smaller after ten years, reducing average household income by between $1,820 and $3,645. Large reductions in inflows to the H-1B program could impose costs of similar magnitude on the US economy (or larger, if in conjunction with other fees like the forthcoming $100,000 fee on Optional Practical Training).

Further, the proposed H-1B fee could reduce the program’s contributions to the Treasury through fees and taxes if DHS’s aggressive elasticity assumption is wrong. 

Some back-of-the-envelope math can estimate how much H-1B visa recipients will pay in federal taxes over the next decade under different assumptions.

As a baseline, we assume the average new H-1B recipient earns a salary of $121,863, an estimate derived from a Penn Wharton Budget Model simulation of DHS’s weighted lottery. Assuming a 76% long-term retention rate for visa recipients11 and 2% annual real wage growth, workers coming to the United States over the next ten years would pay approximately $133 billion in federal income tax and employee-side payroll taxes over the next decade.12 

Now consider a scenario in which approved cap-subject H-1B visas decline by 75%, in line with the 91% decline among petitions subject to the $100,000 fee starting in September 2025. How much H-1B composition would change is uncertain, but we can identify a range for the plausible fiscal effects. Suppose the fee makes each cohort of new H-1Bs more high-skilled and high-earning. We can estimate this effect on revenues by using Penn Wharton Budget Model’s modeled average salary (of $167,000) if DHS were to adopt a wage ranking to select H-1B petitions. In this scenario, the federal government would raise just $74 billion between the H-1B fee and tax payments. That means the fee will cost the federal government $59 billion. Even if the average salary of remaining new H-1B recipients would rise to $250,000 — far above the Penn Wharton Budget Model’s estimate of 75th percentile earnings under a wage ranking system — the federal government would still lose $28 billion in net revenues.13

Given the range of estimates of how demand would respond to such a large fee and uncertainty about how much H-1B salaries will change, it is highly unlikely that the government will break even from the fee, and quite possibly (if not probably) will lose money on net. 

What could USCIS do to recover more of the listed costs?

Given high uncertainty about how demand will respond to the fee, the Department’s proposal risks deterring H-1B filing. As we discuss in the next section, we do not believe that USCIS has the authority to recover the costs identified at other agencies. But, if DHS insists on claiming the authority in the NPRM, it could address its objectives in a lower-cost way that is sensitive to real uncertainty about what the revenue-maximizing fee is.

Instead of guessing at the fee that clears the market, DHS has not considered any alternative in which it could discover it, using the registration system it already runs. Instead of the current proposal, USCIS could open the cap registration period with the fee set at $103,265 and require a binding commitment to pay it. If registrations fall short of that number, USCIS could announce a lower fee and reopen registration for a set period, repeating the process until it secures enough commitments to reach the cap. Registrations from earlier rounds would automatically carry over to the new price, so every round could only add employers. Every selected petitioner would then pay the same final price.14 A descending auction therefore collects the most revenue that USCIS can collect while still meeting the cap and does not require predicting elasticity that DHS cannot possibly know with confidence. If USCIS is right about employers’ willingness to pay the $103k fee, then this mechanism costs the agency nothing; if USCIS is wrong, this approach would allow the agency to generate revenue as close to its cost estimates as possible.

We believe Congress would need to change the law to authorize either the NPRM or this alternative.

Is the proposed fee legal?

What authorities does USCIS have to set fees?

USCIS’s authorities for setting fees, which it claims authorize the NPRM proposal, come from Sections 286(m) and 286(n) of the Immigration and Nationality Act (INA).

Section 286(m) states:

286(m) Immigration Examinations Fee Account
Notwithstanding any other provisions of law, all adjudication fees as are designated by the Attorney General in regulations shall be deposited as offsetting receipts into a separate account entitled “Immigration Examinations Fee Account” in the Treasury of the United States, whether collected directly by the Attorney General or through clerks of courts: Provided, however, That all fees received by the Attorney General from applicants residing in the Virgin Islands of the United States, and in Guam, under this subsection shall be paid over to the treasury of the Virgin Islands and to the treasury of Guam:  Provided further, That fees for providing adjudication and naturalization services may be set at a level that will ensure recovery of the full costs of providing all such services, including the costs of similar services provided without charge to asylum applicants or other immigrants. Such fees may also be set at a level that will recover any additional costs associated with the administration of the fees collected.

Section 286(n) states:

286(n) Reimbursement of administrative expenses; transfer of deposits to General Fund of United States Treasury
All deposits into the “Immigration Examinations Fee Account” shall remain available until expended to the Attorney General to reimburse any appropriation the amount paid out of such appropriation for expenses in providing immigration adjudication and naturalization services and the collection, safeguarding and accounting for fees deposited in and funds reimbursed from the “Immigration Examinations Fee Account”.

These provisions establish the authority to collect fees (m) and expend the collected monies (n), both limited to the purpose of providing immigration adjudication and naturalization services.

Do the costs identified in the NPRM count as “services” that can justify fees?

No, the costs DHS states will be offset and reimbursed by the new fee go beyond adjudication and naturalization services. In fact, DHS seemingly concedes that it is ignoring the strictures and text of INA 286(m)-(n), which are explicitly limited to adjudication and naturalization services, by saying they intend to recover all the costs for “the full range of Federal activities and resources required to support the lawful immigration system” (NPRM at 54826).  The NPRM also states unequivocally that “DHS is not setting this fee to be associated with or required for providing a specific service” (NPRM at 54829) and that one objective of the new fee is to “generate additional revenue.” (NPRM at 54832)

Notably, both INA 286(m) and 286(n) speak in terms of “adjudication and naturalization services,” clarifying what Congress authorized. These are the services provided to individual petitioners or applicants seeking immigration benefits, like the H-1B employers filing petitions seeking the right to hire an H-1B professional. Those costs can include overhead, but don’t extend to the entire administration of the lawful immigration system. Notably, INA 286 distinguishes between “adjudication and naturalization services” ((286(m)-(n)), “inspection services” (286(g)-(i)), and “detention and removal services” (286(h)(2)(A)(v)). The NPRM lists the following agency activities for reimbursement under 286(m)-(n), which do not qualify as adjudication or naturalization services:

  • Within USCIS:
    • USCIS Voter Verification System, $75.3 million. This program is analogous to the E-Verify (and electronic employment verification program) and the Systematic Alien Verification for Entitlements program (SAVE), two small USCIS immigration status verification programs. Since these verification programs are not related to providing adjudications or naturalization services to immigrants, neither is funded through the adjudication and naturalization fee schedule — E-Verify is funded by annual discretionary appropriations, and SAVE through cost recovery from federal and state agencies requesting verifications. (NPRM at 54827)
  • Other DHS components:
    • CBP border inspection infrastructure, $76.2 million: This category includes biometric matching “for travelers entering and exiting the country through the Traveler Verification Service,” integrated into “the traveler inspection process.” (NPRM at 54829 and Supporting Document at p. 6, Table 10) These are “inspection services,” which 286 explicitly distinguishes from adjudication and naturalization services.
    • ICE vetting and screening, $1.05 billion: Of the $1 billion, $900 million is for Enforcement and Removal Operations and Homeland Security Investigations for vetting of aliens in removal proceedings, which qualify as detention and removal services under 286. The Student and Exchange Visitor Program (SEVP), which  is funded by its own fees and account provided by Congress, would get $50 million. The final $100 million is for ICE “vetting of applicants for admission.” (NPRM at 54829, Table 5) If this $100 million relates to adjudication and naturalization services, the NPRM does not give enough information to suggest that is the case. This $100 million appears to be for activities ICE undertakes for CBP, since the supporting documentation in the regulatory docket indicates this ICE activity is for aliens seeking entry to the US.
  • Outside DHS: 
    • EOIR removal adjudication and court operations, $2.96 billion: This category includes EOIR personnel, court space and guard costs, travel, utilities, and IT ($2.64 billion) and the National Qualified Representation Program, interpretation, transcription, litigation support, and FOIA costs ($318.8 million). (NPRM at 54830, Table 7, and Supporting Document at p. 6-7, Table 11) Under INA 240, removal proceedings are adversarial enforcement adjudications in immigration courts in which DHS appears as the charging party, which are not immigration adjudication and naturalization services.
    • DOL wage and hour enforcement, $350 million: DHS describes “investigations, housing inspections, wage and hour compliance, and case development.” (NPRM at 54831, and Supporting Document at p. 9, Table 13) Wage and hour activities that DHS has no authority to undertake as part of immigration adjudications or naturalizations and cannot be considered for USCIS fee reimbursement.
    • DOL litigation and legal support, $350 million: DOL’s Office of the Solicitor is identified as eligible for USCIS fee reimbursement for “legal support and litigation services for the OFLC and WHD enforcement, including regulatory drafting, work before the Office of Administrative Law Judges and Administrative Review Board, federal court defense, and general legal support at both national and regional offices.” (NPRM at 54831 and Supporting Document at p. 9, Table 13) Legal services on behalf of the US Department of Labor are not adjudication and naturalization services.
    • DOS refugee logistics, $92 million: The listed costs include activities which Congress has directed and funded the State Department to cover through DOS’s Bureau of Population, Refugees and Migration, Resettlement Support Center, and participation in the International Organization for Migration and cannot be included in DHS’s adjudication and naturalization services. (NPRM at 54830 and Supporting Document at p. 8, Table 12)
    • DOS visa issuance support, $372 million: This category includes the State Department’s Visa Services Directorate, with consular sections at 270 posts. (NPRM at 54830, Table 8 and Supporting Document at p. 8, Table 12) Consular visa issuance is a State Department function under INA 221, funded through DOS’s own fee authorities and not by DHS user fees for DHS adjudication and naturalization services.

Congress has funded several of the same functions that the NPRM intends to reimburse by USCIS user fees through separate mechanisms elsewhere in INA 286, which confirms it did not regard them as within “adjudication and naturalization services”: inspection reimbursements are handled in INA 286(i), DOS and DOL anti-fraud work is funded from the Fraud Prevention and Detection Account under INA 286(v), and DOL’s processing of labor condition applications and permanent labor certifications is funded in part from the H-1B Nonimmigrant Petitioner Account under INA 286(s)(6). Section 286 is a set of segregated fee accounts, each tied to a named category of service. Section 286(c) states the default: “all moneys received in payment of fees and administrative fines and penalties under this subchapter shall be covered into the Treasury as miscellaneous receipts,” except “as otherwise provided” in the subchapter. 

The categories detailed in INA 286 — “inspection and preinspection services,” “detention and removal services,” and “adjudication and naturalization services” — are exclusive and indicate separate functions. These categories track the programs of the Immigration and Naturalization Service at the time that Congress created IEFA : 

  • The Examinations Associate Commissioner oversaw the Inspections Program and the Adjudications and Nationality Program. 
  • The Enforcement Associate Commissioner oversaw the Detention and Deportation Program.

The 1990 Appropriations Committee understood that adjudication fees would fund “the entire cost of operating the Adjudications and Naturalization program,” (NPRM at 54826 n.73) which is quite distinct from covering services run by Inspections or the detention and removal services run by Enforcement. 

With the Homeland Security Act, the Adjudications and Nationality Program became USCIS, Inspections became part of CBP, and Enforcement became ICE.15 

Following the logic of this restructuring, “adjudication and naturalization services” is limited to services that are now a part of USCIS, and not functions under another part of DHS or other agencies touching immigration. Since 286 restricts fee collection for adjudication and naturalization services, most of the costs listed in the NPRM are not eligible for cost recovery.

Does USCIS have authority to set fees to recover costs for other agencies that have been funded by Congress?

Three background rules of federal fiscal law limit how DHS may set and spend a fee:

  1. An agency may not augment its appropriation from outside sources.
  2. Funds may not be transferred between accounts.
  3. When one agency does pay another for work, the Economy Act fixes the terms on which it may do so. 

None of these rules is absolute, but each is overcome only by an express statutory exception, not by silence or the absence of an express exclusion. DHS’s position is that INA 286(m) and (n) supply that exception. We describe the three presumptions first, then explore DHS’s case that Congress provided an exception to these rules in INA 286(m)-(n).

1. The prohibition on augmentation

In general, authorization and appropriations bills passed by Congress limit the ability of an agency to obligate or expend funds. The legal premise is that unless explicitly stated otherwise, only Congress funds an agency’s activities. An “augmentation” effectively increases the amount available in an agency’s appropriation; unless an explicit exception is made, there is a default prohibition on augmentation. This “no augmentation” principle therefore runs against DHS claiming it can recover costs for other agencies funded by Congress, and that principle could only be overcome by express inclusion of authority rather than by silence or the absence of an express exclusion.

The GAO Red Book reflects the Principles of Federal Appropriations Law for the federal government and states the underlying augmentation principle in general terms:

“As a general proposition, an agency may not augment its appropriations from outside sources without specific statutory authority. When Congress makes an appropriation, it also is establishing an authorized program level. In other words, it is telling the agency that it cannot operate beyond the level that it can finance under its appropriation. To permit an agency to operate beyond this level with funds derived from some other source without specific congressional sanction would amount to a usurpation of the congressional prerogative.” 2 GAO, Principles of Federal Appropriations Law 6-162 (3d ed. 2006), GAO-06-382SP. (Emphasis added)

“Likewise, treating a transaction which should be reimbursed as nonreimbursable may result in an improper augmentation. For example, an agency receives appropriations to do its own work, not that of another agency. Accordingly, as a general proposition, interdepartmental loans of personnel on a nonreimbursable basis improperly augment the appropriations of the receiving agency. Such nonreimbursable loans also constitute a misuse of the detailing agency’s appropriation.” 2 GAO, Principles of Federal Appropriations Law 6-235 (3d ed. 2006), GAO-06-382SP. (Emphasis added)

Two statutes codify the prohibition on augmentation: 

  • The Purpose Statute, 31 U.S.C. 1301(a), provides that appropriations “shall be applied only to the objects for which the appropriations were made except as otherwise provided by law.” An agency that receives an appropriation to do its own work may not spend it doing another agency’s, and an account created for one purpose — here, the IEFA, established to fund adjudication and naturalization services — may not be spent for a different purpose. 
  • The Miscellaneous Receipts Act, 31 U.S.C. 3302(b) states that an official “receiving money for the Government from any source shall deposit the money in the Treasury as soon as practicable without deduction for any charge or claim.”

Together, these laws codify in statute the no augmentation doctrine and place limits on how reimbursements work. They do not authorize the interpretation reflected in the NPRM.

While USCIS’s adjudication and naturalization services do not today receive annual congressional appropriations, ICE, CBP, EOIR, DOS, and DOL do. The only way to overcome the no augmentation doctrine for these agencies is an explicit exception provided by Congress.

2. The prohibition on transfers

Closely related to the prohibition on augmentation is the statutory prohibition on transfers between appropriation accounts or other working funds at Treasury. 31 U.S.C. 1532 provides that “[a]n amount available under law may be withdrawn from one appropriation account and credited to another or to a working fund only when authorized by law,” and that an amount so transferred remains subject to the purposes and limitations of the appropriation it came from unless a statute specifically provides otherwise. Moving money from IEFA’s balances into the appropriation accounts of ICE, CBP, EOIR, DOS, and DOL is a transfer, and is unlawful unless 286(n) explicitly authorizes it. 

3. Limitations on reimbursement under the Economy Act

The Economy Act, 31 U.S.C. 1535, is a New Deal-era statute providing a general authority for interagency reimbursement. Every element of the Economy Act runs in the wrong direction compared to what DHS proposes in its NPRM. The Economy Act contemplates an order placed by an agency that needs goods or services to complete work the ordering agency is obligated to complete by its own authorizing statute (1535(a)). But USCIS is looking to collect monies to cover work by ICE, CBP, EOIR, DOS, and DOL that those agencies are separately required to undertake by statute and which USCIS has no authority to complete under the INA. The Economy Act also requires adjustment on the basis of the actual cost of goods or services provided (1535(b)), which the fixed percentage share of a revenue pool (NPRM at p. 54832) is not.

Under 1535(a), an agency may place an order with another agency only if funds are available to the ordering agency, the head of the ordering agency decides the order is in the best interest of the government, the servicing agency is able to provide the goods or services, and the goods or services cannot be provided as conveniently or cheaply by contract with a commercial enterprise. Under 1535(b), payment is for the estimated or actual cost of the work, with adjustment “on the basis of the actual cost of goods or services provided.” Under 1535(d), an amount obligated for an order is deobligated to the extent the servicing agency has not performed the service by the time the ordering appropriation expires.16 

The case for an exception to the limitations on augmentation, transfers, and reimbursements

To justify the NPRM, DHS must identify express language from Congress that overcomes all three presumptions: language that authorizes augmenting the appropriations of ICE, CBP, EOIR, DOS, and DOL; authorizes withdrawing IEFA balances and crediting them to those agencies’ accounts; and authorizes doing so on terms the Economy Act would not permit. 

Despite the principles underlying the federal government’s fiscal operations, DHS’s fiscal law position reflected in the NPRM is that INA 286(n) establishes the IEFA as an “authorized augmentation account.” DHS assumes that it is authorized to use the fees deposited in the IEFA to reimburse any appropriation for the expenses incurred throughout the federal government to provide immigration adjudication and naturalization services, even accepting that transfers between appropriations are generally prohibited absent direct statutory authority. This is the same position explained in a 2019 NPRM in which DHS proposed an adjustment to the USCIS fee schedule that would have allowed fee-sharing within DHS, reimbursing ICE for various activities with USCIS fees (84 Fed. Reg. 62280 at 62287 (November 14, 2019)). While the 2019 NPRM led to publication of a final rule in 2020, it was enjoined in federal court and never went into effect. The underlying legal issues were never decided and the case was decided on grounds that DHS was found to have an Acting Secretary who was not properly appointed.

We can examine possible candidates for language authorizing an exception for the prohibition on augmentation, the prohibition on transfers, and the limitation on reimbursements. None does the work DHS needs.

  • INA 286(n)’s reference to reimbursement for “any appropriation”: INA 286(n) permits DHS to “reimburse any appropriation,” but only in “the amount paid out of such appropriation for expenses in providing immigration adjudication and naturalization services.” Does reference to “any appropriation” imply some authority to offset appropriated funding in other agencies?

    Possibly, but only when the expense supports actual immigration adjudication and naturalization services, which the costs described in the NPRM are not. As discussed above (in “Do the costs identified in the NPRM count as “services”), USCIS goes far beyond adjudication and naturalization services to attempt to recover “all of the costs for the lawful immigration program.” 

    Historically, the reference to “any appropriation” was applicable when INS lived in the Justice Department, and was responsible for detention and removal services (now under ICE) and inspection services (now under CBP), in addition to adjudication and naturalization services (now under USCIS). At the time, Congress gave appropriations to INS for all its operations, some of which could be spent on adjudications services, which then needed to be reimbursed. USCIS is proposing something quite different: raising money to send elsewhere for other purposes.

    Today, the 286(n) reference to “any appropriation” only refers to appropriations made to USCIS or appropriations made to another agency if that agency performing work on USCIS’s behalf so that USCIS can satisfy its statutory obligation to perform adjudication and naturalization services. Reimbursing another agency for its own functions is a textbook augmentation and prohibited. 
  • INA 286(m)’s reference to “offsetting receipts”: While USCIS does not presently get regular annual appropriations for adjudications and naturalization services (there can be special appropriations, as there were in the past for backlog reduction, for example), the INA 286(m) reference to “offsetting receipts” does not imply some authority to offset appropriated funding in other agencies. “Offsetting receipts” is a technical  accounting term that means that USCIS collects fees from applicants and petitioners, and those fee collections are deposited into the Treasury account for the Immigration Examinations Fee Account (IEFA). The collections are classified as offsetting receipts, meaning the government treats the money collected as receipts that offset federal spending rather than as ordinary tax revenue. The reference is intended for federal user fee accounts and does not suggest any authorities to offset appropriated funds for and authorized activities by other agencies.
  • INA 286(m)’s reference to “similar services provided without charge”: DHS may read the 286(m) proviso’s “without charge” clause as a license to charge H-1B petitioners for whatever else the government does that is not funded by fees (“without charge”), including the roughly $3 billion the NPRM would send to EOIR for asylum and cancellation adjudications and 8,400 new positions. The clause does not reach that far. It permits recovery of the costs of “similar services” — services of the same kind the fee-payer is charged for — that the fee-collecting agency provides to some applicants for free, which is exactly how USCIS has applied it for 35 years. Under its Low Volume Reallocation practice, USCIS reallocates the modeled cost of no-fee and below-cost workloads within the IEFA — Form I-589, Form I-590, credible and reasonable fear screening, certain T and U applications — onto fee-paying benefit requests adjudicated by the same agency. The clause is a cross-subsidy provision within the adjudication and naturalization program. It is not a reimbursement provision that converts another agency’s appropriated function into a “similar service” merely because it is related to immigration and is not currently funded by fees. Nor does it reach services that are provided with charge; see the discussion of programs whose users already pay above.

Some activities outside USCIS could conceivably be qualified as adjudication and naturalization services (for example, if USCIS contracted with ICE to do site visits to support adjudications or to help inform fraud indicators), but USCIS has not demonstrated that the costs it aims to recover are as narrow as these. Instead, USCIS rests on a much broader assertion that it can recover “all of the costs for the lawful immigration program” that is not found anywhere in the governing statutes. 

Under 286, when Congress wants revenue to cross a category or agency line, it says so: 

  • In 1993, Congress allowed the inspection account to reimburse “detention and removal services” and “removal and asylum proceedings,” but only for aliens arriving at air and sea ports. (INA 286(h)(2)(A)(v)–(vi)) 
  • In 1992, Congress let the Attorney General move up to $1 million from the inspection account to the detention fund, for one fiscal year, to be repaid by a date certain. (INA 286(r)(6)) 
  • In 1998, Congress directed 5% of the H-1B training fee to the Department of Labor for processing labor condition applications and, from 2000, labor certifications. (INA 286(s)(6))
  • In 2004, Congress split the fraud fee into thirds among State, DHS, and Labor and told each what it could spend its third on. (INA 286(v)(2)) 
  • In 2016, Congress allowed CBP to keep inspection reimbursements in its own appropriation, but “only to the extent provided in appropriations Acts.” (INA 286(i)) 

Every one of these is an explicit exception to the limitations imposed by fiscal law.

Section 286(n) has none of these features. It names one category, “immigration adjudication and naturalization services,” and can reimburse the amount paid out of an appropriation for that category. On DHS’s reading, IEFA is the one account in section 286 whose revenue may fund any category, in any agency, in any amount, by regulation. The reading cannot be squared with the rest of the section, which makes no mention of this exceptional, economically and politically significant claimed authority.

Are the NPRM’s anticipated costs for other agencies consistent with other public information? 

The proposed fee rule offers some cost estimates for agencies that far exceed funding needs reflected in the President’s budget requests. Notably, the development of the fee rule likely occurred simultaneously with the development and submission of the President’s budget request to Congress, which the White House Office of Management and Budget (OMB) released on April 3, 2026.

The discrepancies between the budget and NPRM cost estimates raise two concerns. 

First, they call into question the reliability of the cost estimates in the NPRM and underscore the need for rigorous, transparent accounting that the public can evaluate. 

Second, they reveal the scale of the potential (and prohibited) augmentation of appropriated funds. Congress may decline to fund these agencies and programs at even the levels described in the President’s budget request, and the fee proposes to provide even more funding than the executive branch asked Congress for. 

How much of the costs are actually related to the H-1B program?

Very little, by the agencies’ own statistics. Even if USCIS could reimburse for the services listed, the disproportionality of the share of other agency functions that support H-1Bs would bring the volume of reimbursed costs into question.

Some of the agencies that run these programs publish their workloads by visa class, form type, or program, and in every block where such a measure exists, H-1B cases are less than 7% of the workload, and generally less than 2%.

This is not to say that USCIS fee-payers can only ever be charged for their own cases. 

Cross-subsidization is lawful under 286(m) even though it may not be a standard expectation in federal agency fee-setting generally (as discussed below in “Does USCIS have authority to set fees to recover costs for other agencies that have been funded by Congress?”), and USCIS has practiced it in every fee rule since the IEFA was created. Section 286(m) lets the agency recover “the full costs of providing all such services, including the costs of similar services provided without charge to asylum applicants or other immigrants.” But “similar services” and “provided without charge” impose real limits. In particular, USCIS uses three distinct mechanisms for cross-subsidization (88 FR 402, 418–419 (Jan. 4, 2023) (Section III.D, “Full Cost Recovery”)): 

  1. Deliberate below-cost pricing: When USCIS wants to deliberately charge less than cost to encourage a behavior (such as charging below the cost of naturalization to encourage naturalizations and integration), the shortfall is diffused across the schedule. Implicitly, by concentrating diffuse costs for the lawful immigration system on H-1Bs, this rule would turn all fees for other fee-paying categories into below-cost fees, but the agency does not justify this decision or consider alternatives.
  2. Same-benefit self-funding: Fee waivers and exemptions within a form are recovered by dividing that form’s cost by fee-paying volume only.
  3. Zero-fee workloads: Zero-fee workloads include asylum applications, refugee registration, refugee/asylee relative petitions, credible and reasonable fear screening, U and T nonimmigrant status, qualifying family members of U-1 status, DNA collection, overseas verifications, automatic certificate of citizenship, determination on a child for adoption, abandonment of LPR status, refugee waivers, and posthumous citizenship applications. These are recovered under §1356(m)’s “without charge” clause, either diffusely or, since 2024 for asylum, via the targeted Asylum Program Fee.

However, the proposed fee and the extent of cross-subsidization fits none of these three mechanisms.

Does the proposed fee qualify as a tax?

Only Congress can impose taxes as general revenue raising. There is extensive case law from the federal courts on the general authority of federal agencies to charge fees to the public, which is distinguished from general revenue raising. The governing canon from the Supreme Court is Skinner v Mid-America Pipeline Co., 490 U.S. 212, 223–24 (1989), which explains that National Cable Television Ass’n v United States, 415 U.S. 336 (1974), and Fed. Power Comm’n v New England Power Co., 415 U.S. 345 (1974), stand for the proposition that

“Congress must indicate clearly its intention to delegate to the Executive the discretionary authority to recover administrative costs not inuring directly to the benefit of regulated parties by imposing additional financial burdens, whether characterized as “fees” or “taxes,” on those parties.”

Skinner also records why the Court reads such delegations narrowly: a charge that does not “besto[w] a benefit on the [regulated party], not shared by other members of society” is better understood as a tax, and the Supreme Court construed 31 USC 9701 (the general federal fee-setting authority, initially enacted as part of the Independent Offices Appropriation Act) “narrowly to avoid constitutional problems,” particularly related to the Taxing Clause.17 

The Supreme Court applied the same method in Learning Resources, Inc. v. Trump, 607 U.S. 229 (2026), holding that even broad statutory language for discretionary action did not carry a revenue power. In July 2026, the First Circuit applied Skinner and Learning Resources together to a challenge to the $100,000 H-1B Proclamation payment requirement, with an order denying stay pending appeal.18  

Applying Skinner to INA § 286(m) yields a narrow reading. This statute authorizes fees “for providing adjudication and naturalization services” set “at a level that will ensure recovery of the full costs of providing all such services, including the costs of similar services provided without charge to asylum applicants or other immigrants.” Congress therefore authorized only one cross-subsidy: paying applicants may cover the cost of services USCIS provides to others for free or at a reduced rate. If it wants to avoid implying that clause is surplusage, USCIS cannot interpret INA § 286(m) as already authorizing recovery of any immigration-related cost anywhere in the government.

Unlike the default fee-setting under 31 USC 9701, USCIS is permitted to charge identifiable service recipients for the services furnished to others, such as those unable to pay. But the other principles of 31 USC 9701 still hold: federal agencies cannot raise revenue through fees paid by the public. Legitimate fees must be set with detailed explanation of the costs being recovered. 

Could the fee double charge for services that other people, or Congress, already pay for?

The NPRM poses two double charging risks that it does little to evaluate. 

First, the fee introduces the risk of charging two different people for the same service. Many of the non-USCIS activities the fee would fund already have their own cost-based fee paid by the person who receives the service. For example, the Student and Exchange Visitor Program is funded by the SEVIS fee that students and exchange visitors pay, which ICE’s own FY 2027 budget request says recovered 137.8% of the program’s costs on average from FY 2021 through FY 2025; the NPRM nonetheless assigns $50 million in a SEVP “funding deficit” to H-1B petitioners. If that deficit returns to its average surplus, then SEVP could be double charging –  once from students and once from IEFA. Details of the reimbursement agreements could prevent this, but they are not available in the NPRM to evaluate or receive public feedback to prevent double-charging. 

Second, an agency may collect from USCIS and from Congress for the same work. The NPRM’s cost figures are projections of what agencies plan to spend in FY 2026 and FY 2027, not amounts Congress has declined to appropriate. The State Department’s $321 million in the fee rule is almost exactly its $322 million FY 2027 budget request; if Congress funds the request, the Visa Services Directorate will be paid twice.

The rule’s only safeguard against double charging is a single sentence, limited to CBP costs, promising that DHS “will align cost assignments” to avoid duplicative recovery. USCIS cannot control what Congress appropriates to State, ICE, or EOIR, and the rule does not require any recipient agency to reduce its appropriations request by the amount it would receive through reimbursement.

More fundamentally,  under INA 286(n), reimbursement is supposed to restore an appropriation for money already spent on adjudication services. Instead, the NPRM envisions a forward allocation of projected revenue to agencies that are also funded through the regular appropriations process, which is the definition of an augmentation and prohibited under fiscal law, as explained above.

USCIS may believe double-charging is not a concern because it suggests the costs USCIS aims to recover are presently unfunded costs at those agencies. But the NPRM does not provide enough evidence or methodologically rigorous cost assessment to evaluate whether USCIS is correct that there will not be double-charging. In the next question, we outline these methodological deficiencies. 

Does the NPRM meet the methodological standards under the government’s fee-setting authorities?

The governing statute doesn’t permit cost recovery in USCIS fees for non-USCIS administered programs without a detailed cost analysis of the “adjudication and naturalization services” provided by the non-USCIS program. For example, legacy INS and USCIS reimbursement of the FBI for name checks to verify criminal records of noncitizens seeking an immigration or naturalization benefit can be shown as a cost tied to a government service directly covered by the IEFA. But, for another example, neither legacy INS nor USCIS have previously sought to impose on USCIS petitioners or applicants the cost of DOL providing labor certification or associated wage and hour activities, even though that work has been undertaken at DOL since IEFA was established. That work is related to investigating individual employers or applications that individual employers sponsoring noncitizens are required to file before DOL, which USCIS is neither authorized to complete nor tasked with performing under INA as part of its adjudication and naturalization services.

Court cases offer the basic framework for how a cost must be calculated to determine a federal agency’s fee. The cases show that agencies are permitted to use reasonable averages, proxies, and indirect-cost allocations, but must explain what activities and expenses make up the fee, why those costs are attributable to the service provided, and why the chosen method of apportionment reasonably reflects differences in governmental cost. 

In Electronic Industries Ass’n v FCC, 544 F.2d 1109 (D.C. Cir. 1976), the court identified two key elements among others: the agency “must justify the assessment of a fee by a clear statement of the particular service or benefit which it is expected to reimburse” and “must calculate the cost basis for each fee assessed.”19 In Central & Southern Motor Freight Tariff Ass’n v United States, 777 F.2d 722 (D.C. Cir. 1985), the court relied on OMB Circular A-25 to explain the obligations of federal agencies in fee-setting.20 More specifically, OMB Circular A-25 requires biennial reviews of user fees and detailed cost basis information.21 

USCIS itself confirms its commitment to accurate, detailed studies of cost recovery and compliance with Circular A-25 on its budget website, stating:

“In accordance with the Chief Financial Officers Act of 1990 (CFO Act), 31 U.S.C. 901-03, and Office of Management and Budget (OMB) Circular A-25, USCIS biennially reviews the non-statutory fees deposited into the IEFA. If necessary, DHS proposes fee adjustments to ensure full cost recovery. The remaining budget authority comes from three other mandatory fee accounts and appropriated funding for the E-Verify program.” 

DHS is subject to the CFO Act, a 1990 law that currently requires major executive branch departments and various significant agencies to prepare annual financial statements and have them audited. As a CFO Act agency, DHS requires each of its component agencies (and some significant directorates and centers) to provide auditable financials as well.

The authority to charge a fee is distinct from whether its amount is permissible, and the latter requires evidence of a rational and detailed cost recovery analysis from the federal agency setting the fee level.22 In Engine Manufacturers Ass’n v EPA, the DC Circuit remanded where important staffing allocations, percentages, and methodological choices were not adequately explained, requiring an account “in intelligible if not plain English” of how costs were selected and allocated.23 As the court described, “in this case the EPA cost analysis contains page after page of impressive looking but utterly useless tables that appear to have been prepared for internal agency use.24

To comply with methodological requirements, legacy INS and USCIS have historically provided comprehensive explanation and justification for changes to the fee schedule, from the very first INS fee rule, through the following creation of IEFA (1994 fee rule) and other intervening rules, to the most recent USCIS fee schedule (2024 fee rule).

Until this NPRM, DHS calculated USCIS fees using Activity-Based Costing (ABC), an accounting methodology which allows the agency to assign costs to each benefit it adjudicates. It also allows assigning costs to the individual inputs, or “activities,” associated with each benefit. DHS has cited ABC as the basis for every other USCIS fee rule except this one, starting with the initial 1998 rule, and for all subsequent updates. (2007, 2010, 2016, 2020 (enjoined), 2024

In each of those rules, the agency offered fee-review supporting documentation for each rule, including a two stage process: 

  1. Stage One – Resources to Activities: Agency-wide USCIS staffing and operating costs are assigned to a defined list of activities performed across the agency. Examples of these activities include: adjudicating cases, background and biometrics checks, records management, customer service, information technology, management and oversight, etc. 
  2. Stage Two – Activities to Benefit Types: The total cost of each activity is divided among the benefit types that draw upon it. To do so, the agency tries to determine the approximate share of an activity’s resources that are allocated to the benefit.

Despite having both completed its FY2026/2027 biennial review and used that review to complete a thorough ABC accounting exercise in support of a separate 2026 fee rule for naturalization, USCIS did not do so in support of the H-1B fee NPRM and states in that rule that it is “not required to determine USCIS immigration benefit request fees using ABC methodology…as long as the methodology used to set a fee or fees is rational and authorized by law.25  

Under ABC accounting, which the NPRM departs from, costs are supposed to be assigned to specific activities, as in the following 2024 assignment:

  • Adjudication Labor to Make Determination ($305)
  • Management & Oversight ($77)
  • Fraud Detection & Prevention ($22)
  • Records Management ($24)
  • Inform the Public ($32)
  • Intake ($6)
  • Background/Security Check(IBIS -> TECS) ($17)
  • Direct Costs ($8)
  • Perform Biometric Services ($0)
  • Issue Document ($0)

In addition to using ABC accounting in all previous rules, DHS has also relied on the Statement of Federal Financial Accounting Standards No. 4 (SFFAS No.4) to demonstrate the reliability of its USCIS cost estimates. SFFAS No.4, set by the Federal Accounting Standards Advisory Board, is the government-wide standard, for assigning costs to the specific activities that generate them. It creates a three-tier hierarchy of cost determination, ranked from most to least precise:

  1. Direct Tracing (¶¶ 126–128): Costs assigned because they can be directly and specifically measured against the cost object. 
  2. Cause and Effect (¶¶ 129–132): Costs assigned using a measurable driver that has a demonstrable causal relationship to the cost of the cost object, even if it isn’t a direct 1:1 measurement. 
  3. Reasonable Allocation (¶¶ 133-137): Used when neither Direct Tracing nor Cause and Effect assignment is available or economically feasible. The agency picks some consistent, defensible basis (headcount share, volume share, budget share, etc.) to spread a cost across cost objects, even though there’s no demonstrated causal link between the specific cost and the specific benefit type. 

Since USCIS has not classified its costs into this hierarchy in the NPRM, the public cannot reliably assess them.

  1. Throughout these Questions and Answers, the term “agency” is used to refer to either a cabinet department or a federal agency that is a component part of a cabinet department.

  2. 8 U.S.C. § 1356(m)–(n), first enacted in 1988 and amended in 1989 and 1991. Pub. L. No. 100-459 (Oct. 1, 1988), § 209; Pub. L. No. 101-162 (Nov. 21, 1989), Title II, 103 Stat. 988, 1000; Pub. L. No. 102-232 (Dec. 12, 1991), § 309.

  3. Based upon “I-129 H-1B - Named Beneficiaries” and “H-1B Registration Fee” costs referenced in the 2024 fee rule’s Immigration Examinations Fee Account supporting documentation.

  4. As explained by GAO in a 1998 report to Congress on INS User Fee Revisions, Activity-Based Costing (ABC) was adopted in July 1995 by the OMB Director, the Department of the Treasury, and the Comptroller General of the United States (GAO Director) as part of the federal government’s Statements of Federal Financial Accounting Standards (SFFAS), which establish standards on how to determine the costs of government services. ABC's major processes are to (1) identify the activities performed to produce outputs, (2) assign or map resources required to carry out the activities, (3) identify the outputs for which the activities are performed, and (4) assign activity costs to the outputs. On August 14, 1998, after an ABC study, the legacy Immigration and Naturalization Service issued a new fee schedule regulation that selected ABC as the means to determine fees for immigration benefit services.

  5. Using raw eligible registrations to determine demand yields an overestimate since USCIS has always had to select more than 85,000 registrants to approve 85,000 petitions. To calculate excess demand for our simulations, we deflate FY2026's 343,981 eligible registrations by USCIS's own selection-to-issuance conversion rate (85,000 of 120,141 selected in FY2026, or 71%), which yields committed demand of roughly 243,000.

  6. In addition to Borjas’s approach of estimating demand, we borrow Borjas’s parameter for the dispersion of payroll savings across individual matches, which generates the right tail of employers still willing to pay $100,000. Since the data errors his critics identified inflate measured dispersion, and not just the average gap, that borrowed spread is likely too wide, making these projected declines conservative.

  7. The authors are not fully clear on how they construct their data, but quotes like “the fact that the total fees paid by a petitioner depend on the number of I-129 receipts submitted” (pg. 9) and “the mechanical relationship between filing volume and total fees” (pg. 12) only make sense if “total fees” refers to the total fees paid for all visas received, not total fees per visa.

  8. AIC and BIC are two different values that capture how well a model fits the data. Specifically, AIC = −2 ln L + 2k and BIC = −2 ln L + k ln n, where n is the number of data points, k is the number of parameters, and L is the log-likelihood. This can be rearranged to give k = (BIC − AIC)/(ln n − 2), and plugging in the values from Table A.5 gives k = 27. If the authors correctly estimated Equation 2, they would have an individual parameter for each entity, which would correspond to k ≈ 107,000.

  9. Unlike the previous estimates, this estimate is not close to 1. Entity fixed effects and imputed values for total fees break the mathematical relationship that forced b1 to be close to 1 in the previous regression. However, neither of these changes fixes the data problems that make this estimate uninterpretable.

  10. USCIS first estimates individual “short-term elasticities” for each fiscal year from 2021 to 2025 using the following regression specification:

        log(CountofReceiptsit) = Intercept + b1 log(TotalFeesit) + b2 FirmCharacteristicsit + ErrorTermit

    This is incorrect. In a negative binomial model the log-linear form applies to the expected count rather than the observed count, and there is no additive error term. The model is properly written as:

        CountofReceiptsit ~ NegBin(μit, α)
        log(μit) = Intercept + b1 log(TotalFeesit) + b2 FirmCharacteristicsit

  11. This retention rate is derived from a National Academies paper, which provides transition rates from H-1B visas to EB-1, EB-2, and EB-3 green cards, as well as H-1B extensions.

  12. For the sake of this exercise, we assume visa recipients are single filers who take the standard deduction. Other research from the Economic Innovation Group has analyzed the fiscal impact of the H-1B visa program at the household level, finding the typical household headed by an H-1B visa holder contributes $30,049 more in taxes than it uses in government services. They find that if DHS adopts a compensation-based ranking for selecting H-1B petitions, that net fiscal contribution would rise to $48,697.

  13. The Penn Wharton Budget Model projects that the 75th percentile of cap-subject H-1B petitions under a compensation-ranking selection would be between $175,000 and $200,000.

  14. This type of mechanism is also known as a “Dutch auction.”

  15. After IEFA was created, there were essentially three phases of implementation. Initially, there was no change in the fee levels set and collected by INS, but fees were used to reimburse appropriated funding. With the first INS fee rule in 1994, following establishment of IEFA, there were big increases in fees so that the adjudication, information, and records services related to processing immigration benefit applications and petitions were sufficient to reimburse appropriations received from Congress at the start of each fiscal year for such services, but other costs like IT, HR, and overhead were not recovered. With the first big USCIS fee rule in 2007, after the Homeland Security Act went into effect, fees increased to cover the full costs to run USCIS, including IT, HR, and overhead. This was a significant fee adjustment not only because it updated and recovered the full cost for all immigration adjudication and naturalization services, but because it also covered all overhead and support functions. Further, this fee rule allowed USCIS to transition from a Reimbursable fee account to a Direct account where the agency was no longer receiving appropriations to reimburse upon fee collections. Instead, USCIS had to ensure adequate fund balances existed to support operations, as it could no longer spend in advance of fee collections to fund the day-to-day activities.

  16. For more information about the Economy Act, see Principles of Federal Appropriations Law, vol. 2, at 6-202 (3d ed. 2006), GAO-06-382SP.

  17. Skinner at 223 (quoting Nat’l Cable, 415 U.S. at 342).

  18. California v. Mullin, No. 26-1699 (1st Cir. July 24, 2026).

  19. Electronic Industries, 554 F.2d at 1117.

  20. See also, Fed. Power, 415 U.S. at 349.

  21. See, e.g., Montrois v. United States, 916 F.3d 1056, 1062–63 (D.C. Cir. 2019).

  22. 20 F.3d 1177, 1182–83 (D.C. Cir. 1994).

  23. Id. at 1181.

  24. See NPRM at pp. 46–47.