Public\u2013public<\/strong> (funded and owned by the government)<\/li>\n<\/ul>\nWe then link these time series to macroeconomic indicators such as total factor productivity (TFP), GDP, R&D expenditure, and investment. By controlling for the business cycle and showing that our innovation measures are not predicted by fiscal and monetary shocks, we isolate the medium-term co-movements between innovation activity and aggregate outcomes.<\/p>\n
The results are striking. Figure 1 displays the macroeconomic effects of innovation, showing that public-private patents (first column) are characterised by the largest medium-term impact on both output (first row) and productivity (second row), doubling those of private sector innovations over the same horizons (second column). Government-funded but privately owned patents are also associated with sustained increases in private R&D spending, business investment, and real wages. Though public\u2013private patents<\/em> make up just 2% of the total<\/strong>, they account for about one-fifth of medium-term GDP and TFP fluctuations<\/strong> (Figure 2). Innovation through public\u2013public patents<\/em>, not shown here, has muted average effects \u2014 though it disproportionately features the most disruptive breakthroughs in health and biotechnology. These results hold across ten different econometric approaches, consistently pointing to government-funded research as a key driver of medium-term productivity gains.<\/p>\nFigure 1 <\/strong>Macroeconomic effects of innovation<\/p>\n<\/span>Notes<\/em>: The figure displays the dynamic effects of innovation shocks in each category of patents (public-private, private-private, public-public; by column) on (log) real per-capita private GDP and (log) TFP. The estimates are done by means of local projections, where the size of the shock is normalised such as to increase total patents by 1% on impact. Shaded areas report 68% and 90% confidence intervals. Sample: 1950:Q1-2015:Q4.<\/span><\/h6>\nFigure 2<\/strong> Contribution of innovation to GDP and TFP<\/p>\n<\/span>Note:<\/em>\u00a0The figure displays the forecast error variance contribution of innovation shocks in two categories of patents (public-private and private-private) on (log) real per-capita GDP expenditure and (log) TFP (by row). Shaded areas report 68% and 90% confidence intervals.<\/span><\/h6>\nWhy are government-funded patents so impactful? Because they tend to be more basic<\/em> than privately funded ones<\/strong>. Basic research expands the knowledge frontier, generating broadly applicable insights that create powerful spillovers and amplify innovation across the economy. Figure 3 shows that government-funded patents account for a large share of the \u2018most basic\u2019 ones, according to measures of basicness based on patents\u2019 similarity to subsequent ones (\u2018importance\u2019) and on the number of citations to scientific papers (\u2018reliance on science\u2019). Even though they are far fewer, government-funded patents tend to be much more basic than private ones.<\/p>\nFigure 3<\/strong> Basicness of publicly funded patents<\/p>\n<\/span>The multiplier of public\u2013private innovation<\/span><\/h2>\nThe macroeconomic patterns we uncover echo Vannevar Bush\u2019s post-war vision: the government should fund basic research that private investors find too risky or with too large upfront costs, while universities and firms translate discoveries into applications. This collaboration between public funding and private ownership appears to generate powerful spillovers.<\/p>\n
A back-of-the-envelope calculation suggests that every dollar of public R&D funding generates between eight and fourteen dollars in cumulative GDP<\/strong>, more than twice the return on privately financed innovation. These multipliers align with recent evidence that public R&D crowds in \u2014 rather than crowds out \u2014 private innovation (see Bloom et al. 2019, Myers and Lanahan 2022, Bergeaud et al. 2025).<\/p>\n<\/span>NIH, NSF, and the geography of transformative science<\/span><\/h2>\nDrilling down by federal agency, the NIH and NSF<\/strong> emerge as the most productive funders of innovation. Patents supported by these two agencies are most strongly linked to medium-term gains in productivity, output, and business-sector R&D. By contrast, patents financed by the Department of Defense or NASA exhibit smaller macroeconomic correlations.<\/p>\nThe sectoral composition of innovations sheds further light on why. In health and biotechnology \u2014 fields dominated by NIH funding \u2014 publicly financed patents outperform private ones by a wide margin. This reflects both the \u2018public good\u2019 nature of biomedical research and the long horizon of medical discovery, which private firms may find hard to justify to shareholders. Similar patterns arise in basic energy research, where the Department of Energy\u2019s publicly funded projects have generated significant spillovers into private R&D, consistent with evidence in Myers and Lanahan (2022).<\/p>\n
<\/span>Who delivers the biggest impact?<\/span><\/h2>\nBeyond federal agencies, the data reveal systematic differences across types of innovators. Research institutes and universities<\/strong> \u2014 rather than for-profit companies \u2014 produce the most growth-enhancing patents when supported by government funds. When universities rely solely on private funding, however, their macroeconomic impact is roughly half as large.<\/p>\nAmong for-profit entities, start-ups and venture-capital-backed firms<\/strong> stand out as more productive engines of innovation than established corporations \u2014 but only when they receive public support. Without government funding, their productivity advantage vanishes. These findings suggest that public financing does not displace entrepreneurial activity; it amplifies it.<\/p>\nSuch complementarities between public and private sectors echo the arguments made by Aghion (2023) and Mitra et al. (2024) on the design of innovation policy in Europe: when the state takes early-stage risks, the private sector is more likely to scale up transformative technologies. America\u2019s post-war experience offers a textbook example of this dynamic.<\/p>\n