Census Data Puts Tech Work on Consulting Firm Payrolls

Census W-2 files show majority-foreign-born professional-services firms added 378,120 jobs from 2006 to 2022. Small shops in the same sector with few foreign-born staff lost headcount over that stretch.

The people doing the work often sit in a client’s office. The tax form names a consulting firm. That is the layer the new federal files can actually see.

The Census Count Follows the W-2, Not the Badge

Martha Stinson and Sean Wang at the U.S. Census Bureau built the Business Dynamics Statistics of Human Capital as an experimental product, released with an April 2025 working paper. They linked IRS W-2s to the Longitudinal Business Database and to demographic files from Social Security applications, the decennial census, and the American Community Survey. The match covers 98 percent of the employment the core Business Dynamics Statistics already track.

Each firm is sorted by the share of its own workers in a group, not by who walks the client campus. The nativity bins are 0-10 percent, 10-25 percent, 25-50 percent, 50-75 percent, 75-90 percent, and 90-100 percent foreign-born. Those BDS-HC experimental nativity tables then report the usual business-dynamics measures: jobs, openings, closings, births, and deaths.

Stinson and Wang are blunt about what the product is for. It maps sorting. It does not prove that a visa caused a hire.

We find substantial changes between 2006 and 2022 in the distribution of employers along several dimensions, primarily driven by changing workforce compositions within continuing firms rather than the reallocation of employment between firms.

Martha Stinson and Sean Wang, U.S. Census Bureau, CES 25-20

That economy-wide finding is mixing inside firms that already existed. The professional-services files also show something else: a bulge of small companies whose payrolls are almost entirely foreign-born. Both can be true at once, because the W-2 employer and the client are often different companies.

WHAT THE FILES CAN SAY

• The unit: A job is a W-2 paid by a U.S. employer, including a staffing shop that places people at someone else’s site.
• The people: Foreign-born here means anyone born outside the United States, including naturalized citizens and permanent residents as well as temporary visa holders.
• The limit: The tables carry no visa flag, so they cannot count H-1B, L-1, or OPT workers as a separate group.

WHAT IS UNCONFIRMED

• Visa share: How many of the foreign-born workers in a 90 percent shop are citizens, green-card holders, or temporary visa holders.
• Client identity: Which household-name tech firm, if any, is buying the contract behind a given W-2.

Across the whole employer economy, jobs at firms with less than 10 percent foreign-born staff peaked as high as 59.1 million in 2006 and stood at 53.9 million in 2022. Total tabulated employment in 2022 was 133.7 million. The native-heavy workplace did not vanish. It stopped being the default.

Professional Services Took the Bulk of the Shift

The sharpest move is in Professional, Scientific, and Technical Services, the NAICS sector that includes software, IT consulting, and technical staffing. It also includes law, accounting, and advertising, so the sector is wider than “tech.” It is still the sector that feeds corporate engineering benches.

Firms in that sector where foreign-born workers were at least half the staff grew from 75,157 companies in 2006 to 125,325 in 2022. Combined jobs at those firms rose from 419,874 to 797,994. That is 378,120 jobs, nearly 90 percent growth. Economy-wide, companies where 75 percent or more of staff were foreign-born grew from 459,254 to 707,442, and their jobs rose from 2,304,412 to 3,366,786, a gain of 1,062,374.

FOREIGN-BORN FIRM COUNTS, 2006 AND 2022

Mid-size professional-services employers, 20 to 499 workers, moved the same way. Jobs at firms that were 10-25 percent foreign-born rose from 754,416 to 1,167,368. The 25-50 percent band went from 325,278 to 471,328. The 50-75 percent band went from 78,030 to 117,778. Once a shop is past the startup stage, the foreign-born share is still rising.

Tiny Firms With 90% Foreign-Born Payrolls

The densest pipeline is small. Among professional-services employers with 1 to 19 workers, companies where more than 90 percent of staff were foreign-born rose from 40,743 to 75,698. Jobs inside those companies climbed from 89,156 to 148,162.

The shops many people still picture as the domestic base barely moved. Small professional-services companies with less than 10 percent foreign-born staff went from 437,379 to 453,797. Jobs there fell from 1.6 million to 1.52 million. Headcount left the native-heavy small firm and showed up at the 90 percent shop.

Stinson and Wang’s startup cut of the same files fits that picture. Thirteen percent of employment at small young firms sits in companies that are 75 percent or more foreign-born, against 7 percent at small older firms. Only 65 percent of startup jobs are at firms with less than 25 percent foreign-born staff, against 75 percent at older firms. Some of these companies were built around the model. They did not drift into it.

Labor filings at large campuses keep showing the same vendors on the ground, even when the client’s own H-1B list looks modest. The Census product cannot name those clients. It can show that the W-2 employer of record grew in a band of tiny, heavily foreign-born professional-services firms that match how IT staffing actually bills.

Large Employers Added More Than 573,000 Mixed Jobs

The 500-plus layer in professional services did not become a 90 percent foreign-born workplace. It became mixed at a scale that barely existed in 2006.

Jobs at those large firms where 25-50 percent of workers were foreign-born rose from 278,058 to 851,557, a gain of more than 573,000, about 206 percent. Jobs in the 10-25 percent band rose from 1.8 million to 2.55 million. Large companies rarely land in the 90 percent bin. They buy project labor from firms that do.

That split is why two honest pictures of “U.S. tech” can disagree. A big software maker’s own W-2 file can stay majority native-born, or only moderately foreign-born, while the people in the building include contractors whose W-2s belong to a consultancy. Diversity reports and visa tallies that start from the client’s payroll miss the second stack. BDS-HC starts from the stack that pays the tax form.

Supporters of global hiring say the mix fills skill gaps and keeps product work in the United States. The files cannot test that claim. They can show that the domestic small-firm side of the same sector did not grow with it.

The 15% Rule Already Named the Pipeline

Federal labor rules have long had a name for an employer whose U.S. staff is thick with H-1B workers. Under the 15 percent H-1B-dependent employer test, a firm with 51 or more full-time equivalent staff in the United States crosses the line when H-1B workers are at least 15 percent of that count. Smaller shops hit the line sooner.

H-1B-DEPENDENT THRESHOLDS

• 25 or fewer FTE: The employer is dependent if it has more than seven H-1B workers.
• 26 to 50 FTE: The line is more than 12 H-1B workers.
• 51 or more FTE: The line is H-1B staff equal to at least 15 percent of U.S. full-time equivalents.

Dependent employers face extra attestations when they file a certified Labor Condition Application from DOL, unless they use the LCA only for exempt workers. The Census nativity bins are not an H-1B count. They still line up with the industry that lives on that paperwork: IT consultancies that recruit abroad, sponsor visas, and deploy people to client sites while keeping the W-2.

Ramachandran Sundararajan, chief people officer at HCLTech, has said about 80 percent of that company’s U.S. people are locals and that its H-1B flow runs between 500 and 1,000 a year. Large Indian IT firms have spent years saying they can hire onshore or move work offshore if visas tighten. The Census files end in 2022, before the later fee fight. They describe the model as it stood when those options were cheap.

A Jury in Los Angeles and a $100,000 Fee

On October 4, 2024, a federal jury in Palmer v. Cognizant Technology Solutions Corp., 17-cv-06848, in the Central District of California, found a pattern of intentional discrimination against non-South Asian employees on race and non-Indian employees on national origin when workers were cut from the company’s unassigned “bench.” The jury also found that punitive damages were available. The 2017 complaint had alleged that at least 75 percent of Cognizant’s U.S. workforce was South Asian, against about 12 percent of the U.S. IT industry; that share was the plaintiffs’ claim, not a Census finding. Cognizant denied the charges and said it would appeal.

Cognizant does not tolerate discrimination and takes such claims seriously.

Cognizant spokesperson, after the October 2024 verdict

The Census tables do not adjudicate that case. They do show why a bench-and-deploy consultancy can look like an outlier in court and like a common W-2 pattern in the professional-services files.

Policy then hit the same firms’ new inflows. After a September 2025 proclamation that added a $100,000 payment on certain consular H-1B petitions, USCIS’s technical appendix to a later fee rule compared cap registrations. Infosys Limited went from 8,886 registrations in the FY2026 cap season to 759 in FY2027. Tata Consultancy Services Limited went from 5,955 to 284. Cognizant Technology Solutions US Corp. went from 3,752 to a recorded drop of 3,752 in that table. HCL America Inc. went from 3,855 to 387.

The cap itself still filled. USCIS said it had received enough petitions to reach the 65,000 regular cap and the 20,000 advanced-degree exemption for FY2027. What moved was who filed. Direct employers with lower consular shares cut less. The staffing layer cut a lot, which is what you would expect if the Census pipeline is real and if the substitute for a new visa is offshore delivery rather than a local W-2.

THE CLOCK AFTER THE 2006-2022 FILES

1. October 4, 2024: A Los Angeles federal jury finds a pattern of bench discrimination at Cognizant and keeps punitive damages in play.
2. April 2025: Stinson and Wang release CES 25-20 and the first public BDS-HC tables for 2006-2022.
3. FY2026 cap season: Eligible H-1B registrations fall to 343,981 from 470,342, after USCIS moves to beneficiary-centric selection.
4. FY2027 cap season: Infosys, TCS, Cognizant, and HCL America slash registration volume in USCIS Table A.6; the numerical cap still fills.

USCIS also published the FY 2026 H-1B registration totals that show the integrity cleanup before the fee shock. Unique employers were about 57,600, against about 52,700 a year earlier. Unique eligible beneficiaries fell to about 339,000 from about 442,000. Eligible registrations fell 26.9 percent, to 343,981 from 470,342. Average registrations per beneficiary were 1.01, against 1.06. Multiple-registration stuffing, which had produced 408,891 multi-registrant cases in FY2024, was down to 7,828 in FY2026. The lottery was no longer a volume game in the same way. The contractor layer then ran into a price shock on consular petitions.

Census Files Now Run Through 2023

The 2022 vintage is still the series behind the Stinson and Wang public statistics on firm workforce mix. In September 2026 the Bureau posted 2023 BDS-HC tables that stretch the time series to 2000 through 2023 and add extra cuts by sector and firm age and by sector and state. The product page was revised on September 16, 2026.

Those extra years will not relabel a W-2. They will show whether the 90 percent professional-services shops kept adding jobs after visas got expensive, or whether the next increment moved offshore and dropped out of the U.S. employer file altogether. If the work leaves the W-2, BDS-HC will record a smaller U.S. shop, not a native-born recovery. That is the next measurement, not a verdict.

The 2006-2022 record is already clear enough on the mechanism. Large tech payrolls got more mixed. The 90 percent layer grew in small consulting firms. Native-heavy small shops in the same sector did not. Anyone still counting “tech jobs” only on the client’s badge is reading the wrong form.

Frequently Asked Questions

What Does the Census BDS-HC Product Measure That QWI Does Not?

The Quarterly Workforce Indicators track jobs, hires, and separations from the worker’s side and can split those flows by worker traits and firm traits. BDS-HC flips the unit. It classifies the firm by the mix of its whole W-2 workforce, then reports employer outcomes such as job creation, destruction, births, and deaths for each mix. Stinson and Wang built it so users can see the economic setting a group of workers faces because of who employs them, not only the average outcome of workers with a given trait.

Does Foreign-Born in These Tables Mean H-1B Visa Holders?

No. Nativity in BDS-HC comes from Social Security, census, and survey records and groups anyone born outside the United States, including naturalized citizens and permanent residents. Temporary visa holders are in that group when they have a W-2, but they are not tagged. Overlap with H-1B-heavy industries is a separate fact from Department of Labor and USCIS filings, not a column in the Census tables.

What Makes an Employer H-1B-Dependent Under Federal Rules?

The test is a headcount ratio, not a nativity share. For firms with 51 or more U.S. full-time equivalents, H-1B workers at or above 15 percent of that FTE count trigger extra Labor Condition Application attestations, unless the LCA is used only for exempt workers. Firms with 25 or fewer FTE are dependent above seven H-1B workers, and firms with 26 to 50 FTE are dependent above 12. Independent contractors and bona fide consultants are kept out of the FTE count.

Is Professional, Scientific, and Technical Services the Same as Big Tech?

No. The sector is NAICS 54. It includes computer-systems design and IT consulting, which is why the H-1B debate lands there, and it also includes legal services, accounting, architecture, advertising, and scientific research. A jump in foreign-born share in this sector is not a headcount at Apple or Microsoft. It is a jump among the U.S. employers coded in that sector, including the consultancies those companies buy.

Are the BDS-HC Tables Official Proof That Immigration Caused Job Loss?

No. The authors state that the tables are not built for causal claims about how firms match with workers or how workforce mix changes business results. They describe sorting and the different job-creation and exit rates attached to different mixes. Disclosure avoidance also uses noise infusion, so fine cell counts should be read as experimental statistics, not as a court exhibit.