Key Points

• The policy expands the economy through three reinforcing channels. By 2059, effective labor input rises 1.7 percent, total factor productivity rises 1.3 percent through the larger STEM workforce, and capital rises 3.9 percent as households save against a permanently higher productivity path. Together these raise output 4.0 percent above current law.

• The wage gains are broadly shared but uneven. Average labor income rises 2.9 percent by 2059, with the largest gains going to low-education workers (4.7 percent for domestic and 4.2 percent for foreign-born). Incumbent foreign-born STEM workers are the only group whose wages remain persistently below baseline, because the additional immigrants land most directly on their skill cell.

• The federal budget improves steadily. Because the additional immigrants are predominantly working-age and highly educated, revenues rise 3.8 percent above baseline by 2059 while outlays rise only 0.9 percent. The primary deficit falls 49.0 percent below current law by 2054 and 41.7 percent by 2059, and debt held by the public ends the projection 5.5 percent lower.

• Every native-born group is made better off, at every age and income level. This Pareto improvement for native-born workers rests on three standard mechanisms: domestic and foreign-born STEM workers are complements rather than close substitutes, the larger STEM workforce raises economy-wide productivity, and a working-age, high-earning inflow improves the federal budget. Lifetime gains are largest for the youngest low-education domestic households, exceeding $100,000 across most of the income distribution for the cohort born a decade after the policy takes effect.

Background

Workers in science, technology, engineering, and mathematics (STEM) occupations are central to U.S. productivity growth, and their contributions extend beyond the firms that employ them: concentrations of skilled workers generate knowledge spillovers that raise the productivity of others in the local economy. College-educated STEM workers accounted for about 12 percent of total U.S. employment and 29 percent of college-educated employment in 2024, up from 7 percent and 27 percent in 2000. Foreign-born workers play a disproportionate and growing role in these fields: 24 percent of college-educated STEM workers were foreign-born in 2024, up from 18 percent in 2000, and foreign-born STEM employment grew 156 percent over the period, far outpacing the 83 percent growth among native-born STEM workers. These shares count only workers holding at least a bachelor’s degree, a narrower population than the broader STEM workforce measures reported elsewhere, which include technical occupations that do not require a four-year degree.¹

Under current law, the principal route to permanent residency for skilled workers, the employment-based (EB) green card categories, is subject to an annual cap of roughly 140,000 visas and a per-country limit of 7 percent. Because demand from high-sending countries far exceeds supply, these rules generate multi-decade backlogs, particularly for applicants from India and China. PWBM evaluates a proposal, modeled on the America COMPETES Act of 2022 and the Keep STEM Talent Act of 2023, that would exempt immigrants holding an advanced STEM degree from both the annual EB cap and the per-country limit, effectively treating STEM green cards as an uncapped category while leaving the number of non-STEM visas unchanged.² PWBM analyzes this policy with a micro-founded overlapping-generations model in general equilibrium, in which wages, capital, productivity, and the federal budget all adjust together. The model distinguishes workers by education, STEM occupation, experience, and nativity, and embeds new PWBM estimates of how readily these worker types substitute for one another along with the link between STEM employment and economy-wide productivity.³ The policy takes effect in 2027 in the model.

Who the Policy Admits

Because the exemption removes the per-country ceiling, it first accelerates the admission of applicants from backlogged countries who have already been approved but are waiting for visa numbers. The population effect then compounds over time as successive cohorts arrive, remain, and have U.S.-born children. An earlier PWBM microsimulation analysis of the same proposal estimated that it would raise the U.S. population by about 825,000 by 2034 and by more than 2.5 million by 2044.⁴ Figure 1 shows the working-age population (ages 21 to 65) widening steadily above the baseline over the projection horizon.

The most direct effect is on the composition of the workforce. Figure 2 shows that the college-educated STEM share of the working-age population rises from 13.4 percent in 2024 to 18.3 percent by 2060 under the policy, compared with 17.6 percent in the baseline, and that the increase is driven almost entirely by additional foreign-born workers. This rising STEM share is the engine of the results that follow: it raises productivity through the STEM-innovation channel, and it increases the relative supply of STEM labor, which reshapes wages across worker types. The college-educated non-STEM and non-college populations are little changed in absolute terms; the non-college share of the workforce declines only because the total working-age population grows faster around it.

Macroeconomic Effects

The policy raises output through the labor market first. As Table 1 shows, effective labor input rises 1.7 percent above baseline by 2059, reflecting the larger working-age population. Total factor productivity rises from 0.4 percent in 2029 to 1.3 percent by 2059 as the STEM workforce expands. Capital responds with a lag, building from near zero to 3.9 percent by 2059 as households accumulate assets against the higher productivity path. Together, the expansion of labor, capital, and productivity raises output to 4.0 percent above current law by 2059, with factor accumulation playing the quantitatively largest role and the productivity channel placing the economy on a permanently higher path. On the household side, private consumption rises 3.5 percent and the average wage per efficiency unit of labor rises 2.7 percent by 2059.

The Federal Budget

The policy produces a favorable fiscal impact that strengthens over time, because the additional immigrants are predominantly working-age and highly educated. As Table 2 shows, federal revenues rise from 0.9 percent above baseline in 2034 to 3.8 percent by 2059, tracking the expanding base of labor, consumption, and capital income. Outlays rise far more slowly, from 0.1 percent to 0.9 percent over the same period, because the new immigrants draw relatively little from means-tested transfer programs. Because revenues grow faster than outlays, the primary deficit falls sharply relative to current law, reaching 49.0 percent below baseline in 2054 before moderating to 41.7 percent in 2059, and debt held by the public ends the horizon 5.5 percent below baseline.

Wages Differ Sharply Across Workers

The 2.9 percent rise in average labor income by 2059 masks substantial heterogeneity across skill and nativity groups, shown in Table 3 and Figure 3. Low-education workers gain the most, reaching 4.7 percent for domestic and 4.2 percent for foreign-born workers by 2059, because their growing relative scarcity reinforces the productivity-driven rise in their marginal product. High-education non-STEM workers see more moderate gains that strengthen over time as general equilibrium benefits diffuse through the economy.

Incumbent foreign-born STEM workers are the only group with persistently negative wage effects, because the inflow expands the supply of workers closest to them and the labor substitution effect outweighs the economy-wide productivity gain for this group. Domestic STEM workers, by contrast, gain modestly. Almost all of this variation operates through wages rather than hours worked, which change little across groups.

The substitution elasticity between domestic and foreign STEM workers is the key assumption.

PWBM estimates that native and foreign-born college-educated STEM workers substitute for one another with an elasticity of 5.63, far below the value of roughly 18 previously reported for STEM workers as a group. At 5.63 the two are considerably more complementary than prior estimates suggest, so an additional foreign STEM worker raises the productivity of a domestic one rather than simply displacing them. This single parameter is what separates a result in which domestic STEM workers gain from one in which they lose.⁵

Lifetime Welfare

The wage and productivity effects accumulate over the life cycle into lifetime welfare effects, measured as equivalent variation: the one-time payment that would leave a household indifferent between living under current law and under the policy.⁶ These values capture three channels at once: the change in each household’s wage path, the economy-wide gains from higher productivity and capital, and the fiscal improvement. The model closes the long-run federal budget through the consumption tax, so the lower debt path implies a lower required consumption tax rate, which itself adds to lifetime welfare.⁷

The distributional pattern mirrors the wage results but is sharper, because younger cohorts are exposed to the new wage path for more of their working lives. As Table 4 illustrates for households entering at the time of the policy, low-education domestic households realize the largest gains, ranging from $90,600 to $110,000 across the income distribution. The cohort born a decade after the policy takes effect gains more still, from $96,700 to $137,800, exceeding $100,000 in four of five income bands. High-education non-STEM households gain less, and incumbent foreign-born STEM households are the main group with losses: a high-income foreign STEM household aged 10 at the time of the policy loses roughly $62,000 in equivalent-variation terms, as the lower STEM wage path is capitalized over many remaining years.

The losses fall on incumbent foreign-born STEM workers, and they are concentrated among the younger and prime-age cohorts with the most remaining exposure to the lower STEM wage path. Older foreign STEM households gain, because the productivity and capital effects outweigh a wage loss they will experience for fewer years. Every native-born group, at every age and income level, is made better off, so the policy is a Pareto improvement for native workers. A robustness exercise confirms that the estimated native-foreign wage structure is central to this result. Under perfect substitution, the aggregate productivity and capital gains are transmitted more uniformly, low-education households still gain but by much less, high-education non-STEM households gain more, and the concentrated foreign STEM losses disappear. Moving from that benchmark to the estimated imperfect-substitution structure raises average low-education equivalent variation by about $21,500 while lowering it by about $12,600 for high-education non-STEM households and about $27,500 for STEM households.

---

This brief summarizes a Penn Wharton Budget Model working paper, “The Economic and Fiscal Impact of STEM Immigration in General Equilibrium” (PWBM Working Paper W2026-1), by Alexander Arnon, Duncan Haystead, Felix Reichling, Germán Sánchez Sánchez, Kent Smetters, and Jesús Villero.

Footnotes

1. Figures are calculated from the 2000 Decennial Census 5 percent sample and the 2024 American Community Survey, accessed through IPUMS USA, using a standard crosswalk from STEM occupation codes to Census occupation codes. STEM status is restricted to workers with at least a bachelor’s degree. ↩

2. The statutory annual minimum for employment-based visas is 140,000 and often rises when unused family-sponsored visa numbers spill over into the employment-based pool. Eligible applicants would hold a master’s or doctoral degree in a STEM field from a qualified institution; the exemption extends to spouses and minor children. Neither bill was enacted. ↩

3. PWBM estimates the elasticity of total factor productivity with respect to STEM employment at 0.26, in line with the upper range of prior estimates. The model also includes Social Security, Medicare, Medicaid, ACA subsidies, SNAP, and federal debt dynamics. ↩

4. See PWBM, “Budgetary Effects of Granting Green Cards to Immigrants with Advanced STEM Degrees” (January 2024), which analyzes the same proposal using the microsimulation model alone, without the general equilibrium feedback modeled here. ↩

5. The model also embeds an elasticity of substitution within the high-education group of 4.90 and between high- and low-education workers of 1.81. These complementarities are why low-education workers gain when the high-education STEM workforce expands. ↩

6. Income percentiles are calculated separately for each demographic group and age. Households younger than 30 are categorized by their gross income in the year they turn 30. ↩

7. The required consumption tax rate turns negative after 2060, equivalent to a consumption subsidy of about 1.7 percent. ↩