Technical Note: The Estimate of about 650,000 H-1Bs as of 9/30/09

argued that the Government Accountability Office had hidden the size of the total H-1B population in its recent report, either out of political expediency or maybe laziness, and thus blurred the debate over the appropriateness of that controversial program.

The point of the earlier blog was that the total size of a population, in this case an alien worker population in the high-tech field, has an important impact on the labor market. The impact of the program on American workers and American life cannot be measured very well if the size of the population itself was not counted, or at least estimated. With those notions in mind I decided to do an estimate of my own, and came up with one in the neighborhood of 650,000. This is how I did it. (Readers allergic to statistics should stop here.)

Program Background. Workers in the H-1B program are tied to their employers; and the employers seek permission from the federal government to bring in workers in this status from overseas, or to adjust workers in the U.S. from another status, usually that of an F-1 or J-1 student. Though the Labor Department and the State Department play a role in the program, the control of the numbers rests with U.S. Citizenship and Immigration Services, a component of the Department of Homeland Security.

Employers file a petition for workers, one at a time, with USCIS and there are congressionally enacted ceilings on the number of initial petitions that can be approved in a given fiscal year. Currently there is an annual 65,000 ceiling for bachelor's level workers, an additional one of 20,000 for aliens with U.S. graduate degrees (usually in science and engineering), and certain other employers may bring in H-1Bs outside the ceilings. The latter group includes government agencies, academic institutions, and organizations whose workers are employed in academic settings.

The initial visas secured through this process are usually good for three years, and are renewable, routinely, for another three years. Further, a worker who has secured an approved petition for a labor certification, and thus potentially a green card, may extend his or her H-1B status beyond the six years, even though the approved petition is unlikely to produce an immigrant visa for many years. This happens because many H-1Bs who have an approved immigrant petition are from India and other nations where there are long backlogs of such approved petitions. These six-year-plus extensions are usually issued one year at a time.

In short, if an H-1B handles these opportunities well, the alien can stay in H-1B status until such time as he or she can adjust to permanent resident status. Though labeled a temporary visa, it can have all the characteristics of a permanent one.

Meanwhile, as the GAO report explains at some length on pp. 30-36, the various government data systems dealing with the H-1B aliens do not mesh neatly, and hence the total number of people in H-1B status in the country at any one time cannot be counted precisely. (Page numbers referenced are those in the printed report itself, not the page numbers are listed by Adobe Acrobat, which lists the cover as p. 1 and thus has a different pagination system.)

While the GAO publishes some recommendations in its report, it makes no effort to estimate the total population of H-1Bs, nor does it recommend changes to the way the government handles its counting systems. Further – and making it harder to make an estimate of the size of the population – its presentation on related matters are often stated in percentages or shown in charts, without noting the specific numbers. (See p. 12, for example.)

Definition. The objective of this exercise is to estimate the number of aliens who, having obtained H-1B status, were alive and in this nation, either in that status or in another legal status, on September 30, 2009. Given the three-year duration of most petitions, by September 30, 2009, everyone in legal status would have secured an approved petition at some time in the prior three years; further, no one who secured status before October 1, 2006 would still be in legal status unless a renewal had been secured.

Estimating the size of the H1-B population on September 30, 2009

This is a three-step process.

First, we estimate the gross size of the supply of approved petitions; second, we estimate the attrition caused along the way in the admissions process; and, third we estimate the percentage of those actually admitted who stop being legal workers in the U.S., largely by leaving the nation.

Step One: Approved petitions

The gross number of H-1B approvals (both initials and extensions) filed in Fiscal Years 2007, 2008, and 2009, is printed in an annual USCIS publication, "Characteristics of H-1B Specialty Occupation Workers: Fiscal Year 2009 Annual Report", p ii, and in its FY 2008 predecessor. This provides us with the basic supply side numbers. (All data in this document are for fiscal years.)

That total is supplemented by an annual population of 12,000, according to a (probably conservative) USCIS estimate, of pre-H-1B workers, drawn from recent F-1 college graduates. The Bush administration, through a process described in an earlier blog of mine, made it possible for potential H-1B workers, without those visas, to works as if they were H-1B workers by extending their legal authorization to work derived from their earlier F-1 status. Since the population was created in April 2008, I calculate that after the passage of a year and a half, it would total 18,000 by the magic date, September 30, 2009. (This is a lot of explanation for 18,000 workers but these are complex processes.)

Here are the numbers for the first step:

The 789,967 number above is not a count of people; it is an estimate of the total number of approvals issued by USCIS. There are three types of attrition that occur at this stage of the process: some aliens are the subject of two or more approved petitions, meaning that there are more approvals than aliens (something the government's statistical systems could handle, but do not); some applications approved by both USCIS and the State Department are not, in fact, used; and some H-1B petitions that have been accepted by USCIS are not approved by the State Department overseas (something which is recorded). In the latter instance, an adjudicator may note, for instance, that the alien who has secured an approved petition does not, in fact, have the academic credentials claimed; hence an approved petition, but no visa for the alien.

It is necessary to estimate the first and second variables; the State Department's final refusal rate is calculated from this data set and its successors.

The 750,000 number above is the estimated number of aliens who secured H-1B status during the FYs 07-09. Some of them departed during that three-year period through returns home, a major factor, and through two other, statistically less significant ones: death in the U.S. and emigration to a third nation (such as Canada). While good age-specific death rate data are available for the population generally, the other two variables must be guesstimated.

H-1Bs are mostly under 35, we are told, so I used the average annual death rate of 35-year-old Americans of both sexes and all races for that age group. That is 111/100,000; since the H-1Bs are above average in income and in education, and rarely work in dangerous industries such as construction, agriculture, and mining, their death rates are probably below the average but I used the 111 anyway, and for the three years collectively I got a rounded figure of 2,500.

At least one Canadian province, Alberta, has or had an arrangement to speed the admission of people with current H-1B approvals, and there was some use of this provision. Someone able to secure an H-1B in the U.S. could probably get a temporary work permit in other nations as well. I used a rough estimate of 1,000 aliens leaving H-1B status in this way. As to returns home, I estimated that 10 percent of the population of interest took this route.

An additional element is the fact that a minority of the extensions are not for three years, but for a single year, as noted earlier. The numbers are not known. The factor does not apply to either the new arrivals, or those getting their first extension (of three years), only to the minority who have been here for six years or more. I have made an allowance of 20,000 for this factor.

In addition to the arrival of H-1Bs, and their (much more limited) departures, there is a third process which needs to be mentioned, if not measured. Some H-1Bs every year adjust from H-1B to green card status, and a smaller number adjusts from H-1B to other temporary worker statuses. (An H-1B moving from the payroll of Westinghouse to that of the World Bank would become a G-1, for example). We made no attempt to measure this population because the members of it remained legal workers in the U.S., neither adding to, nor subtracting from, the population of interest.

The estimate above is just that, a combination of hard numbers, fairly hard numbers, and estimates. It is a shame that the GAO, with its remarkable intellectual resources, did not conduct a similar exercise. I would welcome alternative estimates at

In the meantime, it is appropriate to think, and talk of the total population of 650,000 or so H-1B workers and not to dwell on the considerably less meaningful figure of 65,000 that is usually used by mass migration and industry advocates.

# # #

Anyone wanting to download, and print, the GAO report cited above should be aware that I, at least, ran into a bug when I tried to do so. A graphic – on p. 54 of the report as printed and p. 62 in Adobe – thwarted me from printing the whole document. I needed to print pp. 1-61, and then, separately, pp.63-118. You can probably do without the material on p. 62, unless you have a strong interest in the geographic distribution of H-1B program complaints. DSN