Political Guyentist
← Back to issue

U.S. public R&D may generate as much productivity abroad as at home

Political Guyentist · July 27, 20261 min read

NBER

A new estimate across 69 economies finds the largest spillovers from nondefense research, especially in non-OECD countries.
Currently showing One U.S. public R&D shock, a 12-year global payoff
One U.S. public R&D shock, a 12-year global payoff
Roughly half the payoff from U.S. public research crosses the border The domestic social rate of return to U.S. public nondefense research and development is estimated at 140 to 210 percent by Fieldhouse and Mertens. De Souza and coauthors apply their estimated international spillover wedge to derive a conservative worldwide lower bound of 250 to 460 percent or more. These are derived point estimates, not confidence intervals, and sampling uncertainty is not shown. Three bars show the exact shares of the estimated global productivity gain received by the United States and 69 foreign economies under the paper's GDP weighting windows: 56.6 versus 43.4 percent using 1977 to 1979 weights, 52.3 versus 47.7 percent using 1994 to 1996 weights, and 45.0 versus 55.0 percent using 2017 to 2019 weights. The study sample is 1980 to 2019; gains are discounted over 15 years at 7 percent. RESEARCH, EXPLAINED Roughly half the payoff crosses the border The paper’s conservative model puts worldwide benefits from U.S. public nondefense R&D at about twice the benefits measured inside the United States. Estimated social rate of return Ranges of derived point estimates; zero-based scale 0% 100% 200% 300% 400% 500% Inside U.S. 140–210% Worldwide 250–460%+ Domestic: Fieldhouse & Mertens (2026a). Worldwide: De Souza et al.’s lower-bound spillover application. Sampling uncertainty is not shown. The worldwide range is a lower bound, not a confidence interval. Who receives the estimated global productivity gain? Same 15-year discounted estimate, using GDP weights from three points in the sample Start weights, 1977–79 U.S. 56.6% Abroad 43.4% Middle weights, 1994–96 U.S. 52.3% Abroad 47.7% End weights, 2017–19 U.S. 45.0% Abroad 55.0% “Abroad” covers the 69 foreign economies in the study. Later weights give non-OECD economies more weight. ESTIMATES, NOT OBSERVED OUTCOMES. Sample: 1980–2019; 15-year horizon; 7% discount rate. Source: De Souza, Fieldhouse, Mertens, Nath & Ramey, NBER Working Paper 35517, Table 1 & Section 7 (July 2026), doi:10.3386/w35517. Domestic-return source: Fieldhouse & Mertens (2026a). RESEARCH, EXPLAINED Roughly half the payoff crosses the border For public nondefense R&D, worldwide benefits are about twice those measured inside the United States. Estimated social rate of return Derived point estimates; zero-based scale 0% 100% 200% 300% 400% 500% Inside U.S. 140–210% Worldwide 250–460%+ U.S.: Fieldhouse & Mertens (2026a). Global: De Souza et al.’s lower-bound application. Sampling uncertainty not shown. Lower bound, not a CI. Who receives the global gain? Start, 1977–79 U.S. 56.6% Abroad 43.4% Middle, 1994–96 U.S. 52.3% Abroad 47.7% End, 2017–19 U.S. 45.0% Abroad 55.0% 69 economies abroad · 1980–2019 · 15 yr · 7%. Source: NBER WP 35517 · Table 1 & Sec. 7. July 2026 · doi:10.3386/w35517

What that means

The authors’ back-of-the-envelope calculation says global returns to U.S. public nondefense R&D are about twice the domestic returns—so the United States captures roughly half of the total productivity benefit. They point to open scientific knowledge, trade in capital goods and technological leapfrogging as likely channels. This is a working paper, not a peer-reviewed result, and the 50–50 split is an estimate built on the paper’s causal model rather than a direct accounting of gains.