There is a debatable but serious intellectual case for tariffs on migrant workers. The Nobel laureate economist Gary Becker argued decades ago that charging a tax on immigrant workers could allocate scarce visas to the workers and firms that value them most, and generate public revenue—without the distortions of rationing by quotas.
Last September, the United States moved from theory to practice with a presidential proclamation obliging employers to pay $100,000 for each new H-1B worker entering the country—the largest-ever tariff on international trade in services. The H-1B visa is used by employers to hire highly skilled professionals in the technology sector, and its issuance is linked by extensive research to higher productivity and economic growth rates.
But the only public economic analysis justifying the tax rests on several analytical choices that are incorrect and highly idiosyncratic. These problems go beyond differences of interpretation. They undermine the entire economic justification for this migration tariff.
A new tax on the services trade
What is the policy? This is a new tax (or tariff). The proclamation calls the $100,000 a compulsory “payment”—a pirouette to avoid calling it a tax, because the statutory authority invoked gives no power to tax migrants. But a compulsory $100,000 charge whose explicit purpose is to capture employer “payroll savings” for the US Treasury is a tax by any standard definition in law or economics. The Supreme Court has held that compulsory payments to the government (including for immigration) are taxes unless tied directly to and earmarked for the costs of regulation. The H-1B tax makes no such link.
Why charge $100,000? A single public study provides the quantitative case for that amount. It was recently released by former White House economist George Borjas, who served as the US administration’s senior in-house immigration economics advisor while the tax was designed and enacted. That study estimates that H-1B workers earn 16 percent less than comparable US natives. Over a six-year visa term, this gap would imply roughly $100,000 in employer savings—the exact amount of the tax. Thus the study’s author recommends a tax calibrated to “maximize government revenue” by extracting those savings.
The empirical linchpin of the argument is the alleged 16 percent wage gap. The case for the $100,000 magnitude rests crucially on the size of the estimated wage gap and whether that gap exists at all. There is cause for concern. Some of the methods used to estimate the H-1B wage gap have been cogently criticized by analysts at the Economic Innovation Group, partially overlapping the problems I uncovered.
A stack of erroneous methodological choices determines the result
In my own new analysis at the Center for Research and Analysis of Migration, the leading global network of migration economists, I calculate that H-1B workers earn a small premium relative to comparable natives, of about 6 percent. Not a 16 percent penalty. A 6 percent premium, on average, with large variation around that average. My calculation draws from the same records released under the Freedom of Information Act (FOIA) on individual H-1B workers, and the same American Community Survey data on US natives. This is not a minor discrepancy. It implies that the economic case for the $100,000 tax may be built on a flawed foundation, not just imprecise but directionally incorrect.
I trace this large discrepancy to a series of errors, omissions, and unorthodox choices in the analysis used to justify the H-1B tax.
- Undisclosed filling-in of missing data: The government database does not record a level of education for about a third of the H-1B workers in key years. The analysis justifying the tax simply filled in all those blanks by assuming their highest degree is bachelor’s—without disclosing this assumption. This hidden assumption creates unknown bias, because the reason those values are missing could be correlated with H-1B workers’ wages by education level. Even worse, the bachelor’s degree assigned for 39 percent of those workers as their assumed highest degree is impossible, because they entered through a visa lottery reserved for people with a master’s degree or above. Using years with uncorrupted data, and no erroneously assumed values, reduces the wage gap.
- Comparing different definitions of wages: The analysis justifying the tax takes no notice of a substantial difference in what counts as “wages” for H-1B workers and natives. There is a fundamental accounting issue: The H-1B records report base salary only, while the census captures total wage income for natives—including bonuses, equity compensation, and second-job earnings that amount to more than 7 percent of base pay at major tech employers. Comparing H-1B workers’ base salary to natives’ total compensation guarantees that H-1Bs look underpaid. Comparing apples to apples reduces the wage gap.
- Mismatched years: The analysis justifying the tax compares H-1B workers from four different years (2021–24) to natives in just one selected year (2023). There is simply no reason to do this, and no reason is given. The straightforward approach—comparing workers within the same years—reduces the gap further.
- Unique, unorthodox definition of local labor markets: The analysis justifying the H-1B tax exclusively compares H-1B workers whose offices are in each neighborhood to US native workers whose residences are in that same narrow neighborhood (Public Use Microdata Area or PUMA), often just 1–3 miles across in the relevant cities. That artificially makes H-1B wages lower. For example, it compares H-1B workers who work in the upscale Chelsea neighborhood of Manhattan, but commute there from cheaper neighborhoods, exclusively to natives who can afford to live in Chelsea. The only proper comparison is between H-1B workers and natives across the well-integrated New York metro area labor market, which is how all mainstream peer-reviewed research in labor economics defines local labor markets. No serious research would consider the 100+ neighborhoods (PUMAs) of the New York City metro area to be separate labor markets. Moreover, US law requires H-1B wages to be compared to workers in the broader commuting zone or metro area, not the narrow neighborhood of the worksite. Using the standard definition of local labor markets shrinks the wage gap even further.
- Failing to account for differences in tenure: The study justifying the tax compares native workers, who typically have several years of tenure at their employer, to new employee H-1B workers who by definition are recent arrivals at their firm. But economic research has conclusively shown, for decades, that workers’ wages rise with their tenure at a firm, even controlling for age and overall work experience. Comparing H-1B workers with low tenure to natives with higher tenure, by itself, artificially creates a large wage gap. Correcting this omission further reduces the wage gap.
Correcting for all of these errors, omissions, and unconventional choices accounts for the entire gap between the finding of a 16 percent H-1B wage penalty, on which the study justifying the tax rests, and reveals a small wage premium for H-1B workers.
The flaws in this analysis undermine the economic rationale for the tax
The empirical problems alone warrant revisiting the H-1B tax in general and specifically the $100,000 magnitude. The study justifying the tax draws the $100,000 magnitude directly from the purported 16 percent wage penalty for H-1B workers. The straightforward corrections I report, grounded in decades of peer-reviewed and widely-accepted research in labor economics, imply that there is no wage penalty on average for H-1B workers relative to comparable natives. By the study’s own reasoning then, for this reason alone, the optimal tax proclaimed by the White House should have been zero.
But as I discuss in detail in my paper, the problems run deeper. Even a policymaker convinced that a wage gap exists has reason to prefer other instruments. These include wage floors better calibrated to occupation and location, or portability of visa status across employers to reduce workers’ dependence on a single firm. These are superior for the purposes of raising economic welfare, including natives’ welfare specifically, compared to a blunt per-head levy that deepens the very distortions it purports to address.
The economic case for some degree of tariff on migrant workers, and on services imports in general, deserves to be taken seriously and debated. That requires sound economic analysis. But the specific empirical foundation for a $100,000 tax does not survive close examination, and the choice of instrument raises separate concerns. The evidence calls for revisiting both the magnitude and the fundamental design of this unprecedented policy experiment.
Data Disclosure
This publication does not include a replication package.
