For the First Time, DHS Estimates the Total Size of the H-1B population

Finally a number reflecting the stock, not just the flow — but it seems low

By David North on June 27, 2020

For too many years the major number attached to the H-1B program was the ceiling on the number of new private-sector H-1B workers to be permitted in each year — 85,000. This number was published many times, and gives a totally misleading notion of the size of the program because approvals usually last for three years and many H-1B workers have one to many renewals.

Now USCIS has thrown a new, and more realistic, number into the conversation. The agency has just issued an estimate of 583,420 total H-1B workers in the country (as of September 30, 2019). It is entitled "H-1B Authorized to Work Population Estimate".

It is good that the agency has finally issued a number on the size of the H-1B population — the stock, in other words, not just the annual flow. But on the other hand, the estimate itself is either far too low or the description of how DHS got to the number is badly explained, or perhaps both. So one plus, and two minuses.

Using only DHS-released statistics we have these two dueling numbers, only partially explained in the new report:

  • Total H-1B approvals in the fiscal years 2017, 2018, and 2019: 1,085,040
  • Total H-1B workers in status at the end of FY 2019: 583,420

The larger number comes from a series of DHS reports, most recently the "Summary of Approved H-1B Petitions by Employers, Fiscal Year 2019". No one could expect that the total number of workers would be the same as the total number of approvals because of a variety of factors to be discussed below, but the 46 percent drop-off is hard to understand. And the estimate does not even mention the 1,085,040 number.

Instead, it uses what I believe is a previously undisclosed number, 725,613, as the basis for its estimate; this is defined as "approved H-1B petitions with a validity period through FY 2019 and beyond".

When we asked an agency insider about the 360,000 or so gap between the 1,085,040 and the 725,613, we were told that all or most of it could be explained by the fact — with which we do not disagree — that many approvals were for less than the usual three years. But how significant was this factor?

We are not told the average length of such approvals for all three years, but, in a footnote to a remarkably dense technical appendix, we are told that for the years 2017 and 2018, the average length of an approval was 2.3 years. Now 2.3 years is 77 percent (rounded) of three years and 77 percent of 1,085,040 is about 835,000. This suggests that the average length of approvals was much lower in 2019 (which aggravates the industry) or there is some other factor at work. The estimate would be much more believable if that whole process had been explained carefully.

The next part of the estimation's process (and this is described) takes the number 725,613 and reduces it to 619,327; the latter number is "total authorized unique beneficiaries" or human beings. This reduction, again appropriately, relates to the fact that some approvals within the three-year period were granted to the same persons more than once. Some H-1Bs changed employers, and others added an additional (presumably part-time) employer to the one that they already had. I do not have many worries about this part of the process — there is some worker movement — but I am concerned about the accuracy of this part of the estimate, which is based on Social Security numbers.

The second footnote in the estimate says "About 78 percent of the approved petitions with validity periods through the end FY 2019 (and beyond) have SSN[s] captured by the USCIS SOR [system of records]."

Really! In close to a quarter of the records — dealing with workers who often make $100,000 a year or more — there is no SSN. What an admission!

Further, it would be helpful to know the time frame for these lateral movements; it would be nice to know how much job-changing is going on in a given year, and how much of it is H-1Bs adding an additional employer. A missed opportunity.

The final part of the estimate reduces the 619,327 to 583,420 mentioned earlier. This part is not troublesome and deals with the extent to which H-1Bs switched their status to another nonimmigrant status (1,475), were denied visas by consular officials (2,100), or moved on to green card status (32,332). That a handful of them must have died is not mentioned.

My overall sense is that the estimate — it carries no authors' names — was written by a technician or several technicians who know little about the program, and the meaning of their findings, but who could muster up some really impressive equations such as: