An analysis of recently released Department of Homeland Security data on H-1B filings suggests that about one-third of them were phony, in the sense that once the selections were granted to employers, the employers did not make use of them.
DHS announced that more than 308,000 applications were submitted for fiscal year 2022; we can project from the new data that something like 100,000 of those applications were, shall we say, exaggerations of the employers’ collective appetite for these skilled foreign workers. Since applications for the program have usually outrun the 85,000 per year ceiling set by Congress, some to many employers have been gaming the system by telling the government that they want two to four or more times as many workers as they really want in the hopes of getting the number they actually want.
In part, these numbers are the result of DHS shooting itself in the foot in the application process. In previous years, filing for an H-1B slot was labor-intensive (with lawyers often doing the work on the multi-page forms) and expensive, including fees both to the law firms and to the government. Forbes writer Stuart Anderson, who adores the program, has estimated that it costs $31,800 in legal and governmental fees to complete two rounds of H-1B applications; I think $10,000 to the government is a more reasonable estimate (for an initial application under the old rules).
Given that the older process was inconvenient to the employer community, DHS aimed at its own feet and created a new and simpler system. The first step became filing a one-page form and sending in a non-refundable check for $50. Employers responded with many more applications. Those that win slots have to fill out the needed long forms and pay the fees, but only after victory is assured.
Until recently, all one could do on this front was to speculate on the extent of duplicate filings. Recently, U.S. Citizenship and Immigration Services (USCIS) released some numbers, but not all, and thus we can speculate with a little more accuracy on the extent of the abuse of this process.
What the government has revealed, in addition to the raw number of the new applications, is the number of selections made in two lotteries for FY 2021 and three (so far) for FY 2022. The government senses — I think wrongly — that the ceilings for H-1B workers set by Congress years ago are targets for the issuance of these visas, not caps. So in the last fiscal year it ran a second lottery to cope with the over-applications and this year it has run both a second and a third lottery, for the same reason. What it has not done, to my knowledge, is to give us the number of employer acceptances of the H-1B selections, though we do know that the total number of acceptances should not be over 85,000. We have used 85,000 acceptances as our assumption in the calculations that follow, but this may be too high a number this year and thus understate the extent of abuse.
The table shows the number of initial filings, selections by the government in its recent multiple lotteries, and refusals by employers of selections won.
Roughly One-Third of the 2021 and 2022 H-1B Selections Were Not Accepted by Employers |
|||||
(1) | (2) | (3) | (4) | (5) | (6) |
Lotteries | Total Initial H-1B Filings | Filings Initially Selected | Selections Accepted; ≤ 85,000 | Empty Selections (Refusals) | Filings Not Selected |
2021 First | 106,100 | ||||
2021 Second | 18,315 | ||||
2021 Total | 274,237 | 124,415 | 85,000 | 39,415 (31.6%) |
149,822 |
2022 First | 87,500 | ||||
2022 Second | 27,717 | ||||
2022 Third | 16,753 | ||||
2022 Total | 308,613 | 131,970 | 85,000 | 46,970 (35.6%) |
176,643 |
Sources: Columns two and three: “H-1B Electronic Registration Process”, USCIS, November 19, 2021. Column four: partial estimate based on the ceiling set by Congress; column five: column three minus column four; column six: column two minus column three. The percentages in column five are CIS calculations based on columns three, four, and five. |
The key data is in column five, which shows the estimated number of much-prized H-1B approvals that have been refused by employers. The percentages shown, of 31.6 percent and 35.6 percent, are of refusals compared to the total number of selections. The average refusal rate in those two years is about one-third of the selected applications.
We assume that the percentage of empty applications among the non-selected would be about the same as among the selected, as both are lottery results. Thus, about one-third of all applications were empty ones.
The government does not give numbers for the empty filings, nor does it discuss the motives of employers for over-filing. Further, it does not mention one of the worrisome elements of this process: a rent-a-programmer firm with an overage of visas might sell individual visas to would-be migrants, which would allow the purchaser to come to the U.S. seemingly legally and then become an illegal alien.
The irony of all of this is that if DHS had not bowed to the industry and created a new system, it is likely that the program would have brought more alien workers to this country than the new system. But with DHS hiding some of the numbers, we cannot be sure.
DHS should revert to the earlier, less vulnerable selection process; that of making the employers file full applications and pay, at least temporarily, the full set of fees. This would decrease the amount of gaming of the system, show a more accurate picture of the real level of desire for H-1Bs, and give DHS an annual, multi-month, interest-free loan of some $3 billion. (The latter figure relates to 300,000 applications times $10,000 per application.)
Reviving that old system might do something even more positive: It might cause some employers to think about hiring more citizens and green card holders, workers who can be secured without lots of red tape and without extensive legal and governmental fees.
The author is grateful to Matthew Bonness for his advice.