估算联邦研发投资的经济影响

Presentation at the NBER Entrepreneurship and Innovation Policyand the Economy Conference, 2026 Sheila Campbell, Jaeger Nelson, Eli Schrag, Heidi Williams, and Caleb Wroblewski How Do Changes in Federal Investment in Research andDevelopment Affect the Economy? In recent years, Congress has shown interest in making policy changes thataffect research and development (R&D), such as: §Modifications to federal funding for R&D investment (for example, the CHIPSand Science Act, Public Law 117-167) and§Modifications to tax provisions that affect the after-tax price of private R&D This presentation details CBO’s current basis of assessment for modeling theeconomic effects of changes in federal funding for nondefense R&D, with the As in our preliminary analysis published last July (CBO 2025), we considerpolicy experiments that either increase or decrease federal funding fornondefense R&D by $30 billion per year for 10 years, from 2027 to 2036. Road Map: Two Analytical Frameworks R&D Capital Stock Approach Human Capital Approach §Directly connects changes in federalfunding for R&D with changes in the R&Dcapital stock, which generate changes in §Allocates federal funding for R&D acrosslabor and capital, consistent with historicalspending patterns in which funding for §Directly connects with the standard wayR&D appears in BEA’s NIPA data §Builds on CBO’s prior work estimating theeconomic effects of changes in thenumber of other foreign nationals who §Builds on CBO’s prior work estimating theeconomic effects of changes in federal Background Federal Outlays for R&D Recipients of Federal Funding for R&D Key Factors §Timebetween funding and outlays §Changes in nonfederal R&D activity §Magnitude of productivity effects §Depreciation of R&D capital Time Between Funding and Outlays To estimate the timing of cumulative outlays, CBO used account-level rates that are based on historical patterns. Accounts shownhere are those whose outlays made up more than1 percent of the total federal outlays for nondefense R&D over the 2020–2024 period. The combined outlays for those accounts constituted 74 percent of federal outlays for nondefenseR&D over that period. To calculate the average rate, CBO weighted each account’s spending ratein a given yearby the account’s share of that 74 percent. DOE = Department of Energy; NASA = National Aeronautics and Space Administration; NIH = National Institutes of Health; NSF =National Science Foundation. Time Between Outlays and Productivity Effects Magnitude of Productivity Effects The extension of the Fieldhouse and Mertens line to 20 years is based on unpublished results shared with CBO. Depreciation of R&D Capital The research literature generally applies BEA’s standard R&D depreciation rate of §Crawfordand others (2014) tabulate depreciation rates for categories of R&D:7 percent for aerospace, 9 percent for health and energy, and 16 percent for §Weightedby shares of federal spending on R&D, those rates imply a Animportant conceptual issue is that those depreciation rates are arguablyintended to capture a private depreciation rate relevant to an owner of an R&D §The concept of depreciation relevant to CBO’s work is theeconomywidedepreciation rate. §Anapproach based on patent citations implies an economywide depreciationrate of 5 percent, which we use in this analysis. Human Capital Framework Human Capital Framework Federalfunding for R&D is allocated to labor or capital. §Wedraw on data from the National Science Foundation’s National Center for Science and Engineering Statistics and BEA’s R&D satellite account.§On average, 54 percent of federal funding for R&D pays for labor.§For institutions of higher education, we draw on the analysis of UMETRICS data in Sattari and others (2022) §We account for changes in R&D spending by nonfederal organizations, as in the R&D capital stock approach. For labor, we estimate effects through two channels: §Reallocation.Federal funding for R&D allows additional people to work as researchers, and that additionalwork (measured in researcher-years) contributes to TFP. §Training.The roughly 40 percent of federal funding for R&D allocated to institutions of higher educationpartially supports the training of additional researchers. Because of the time associated with training, those For capital, we apply CBO’s standard capital investment framework (Lasky 2018). Reallocation We start with an estimate of the economywide change in the number ofresearchers, drawing on Fieldhouse and Mertens (2026), Figure 7(e). We apply the estimates from Figure 2 in Sattari and others (2022) to allocate thatchange in the number of researchers across researchers of different education Finally, we account for education-specific effects on TFP by: §Estimatingaverage TFP effects per researcher engaged in innovation-relatedactivities on the basis of Prato (2025) and related evidence (CBO 2024); §Estimating, for different education levels, the sha