The ripple effects of funding on researchers and output

Using unique, new, matched UMETRICS data on people employed on research projects and Author-ity data on biomedical publications, this paper shows that National Institutes of Health funding stimulates research by supporting the teams that conduct it. While faculty—both principal investigators (PIs) and other faculty—and their productivity are heavily affected by funding, so are trainees and staff. The largest effects of funding on research output are ripple effects on publications that do not include PIs. While funders focus on research output from projects, they would be well advised to consider how funding ripples through the wide range of people, including trainees and staff, employed on projects.

Note: The table reports the number and share of unique people by occupation before aggregating them into broader categories. In our analysis, we combine Research and Research Facilitation into "Research Staff" and Clinical, Instructional, Other (who manual inspection indicates are overwhelmingly staff), and Technical into "Other Staff". The numbers reflect the distribution based on the initial occupation reported in UMETRICS for each person. Note: The table reports distribution of papers by occupation and timing of publication relative to the employment in a certain lab. Particularly, we report the number of papers that were published before a researcher appears paid by a certain lab for the first time in our sample, after the last transaction in our records, and between the two. The unit of observation is a lab-or PI-year.

A. Spending
Note: The figure shows the breakdown of spending for individual NIH Institutes and Centers (ICs).

B. Occupation
Note: The figure shows the breakdown of employment by occupation category for individual NIH Institutes and Centers (ICs). The numbers at the right show the total number of employees per award for each IC. Employment is measured as full-time equivalent workers and is estimated by prorating people by the share of their time charged to each project, their full-time / part-time status, and the number of days worked during the year.

Figure S.2.A. Ordinary Least Squares (OLS) versus Fixed Effects (FE) Estimates
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). It reports estimates for all unique publications using ordinary least squares (OLS) and lab or PI fixed effects (FE) models. The FE models account for all time-invariant differences across labs. Both estimates control for calendar year fixed effects and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.B. Publications controlling for lab size
Note: The figure plots the relationship between funding (measured in 2018 BRDPI dollars) to a lab and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). This figure plots estimates with controls for the number of employees in the lab (red series) and without this control (turquoise series). Both series control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.C. Research and clinical publications
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). These series restrict the outcome variable to clinical publications (red series), defined as clinical trials, systematic reviews, and clinical guidelines, and research publications (turquoise series) as defined by the iCite database from the National Institutes of Health, Office of Portfolio Analysis. Both sets of estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.D. Mean and PI-max RCR of publications
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and the quality measures of publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). The quality measure is the relative citation ratio (RCR): the article's mean citations per year divided by the median number of citations received by papers in the same research area published at the same time. We plot effects on the mean RCR of all papers published by the lab (red series) and the maximum RCR of those papers (turquoise series). Both estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.E. Publications by Type of Award.
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). This figure restricts the measure of funding to money received through the R01 award mechanism (red series) and through research grants (all R-class mechanisms including R01, plus K99 and U01, turquoise series). Both estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.F. Institutions by Number of Doctoral Degrees Awarded
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). To perform these regressions, we split our sample into institutions in the bottom 75th percentile in terms of average annual number of doctoral degrees awarded (red series), and in the top 25th percentile (turquoise series). Both sets of estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.G. Institutions by Enrollment
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). To perform these regressions, we split our sample into institutions in the bottom 75th percentile in terms of average annual enrollment (red series), and in the top 25th percentile (turquoise series). Both sets of estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.H. Institutions by Federal Funding
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). To perform these regressions, we split our sample into institutions in the bottom 75th percentile in terms of average annual federal funding (red series), and in the top 25th percentile (turquoise series). Both sets of estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.2.I. Career Age
Note: The figure plots the relationship between funding to a lab (measured in 2018 BRDPI dollars) and unique publications authored by the people in that lab, from publications 3 years before funding is received (-3) to 3 years after funding was received (+3). To perform these regressions we split our sample into PIs with an average career age below 10 years (red series), and above 10 years (turquoise series). Both sets of estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are shown in all panels. Standard errors are clustered at the lab level.

Figure S.3. Funding Effects by NIH IC
Note: The figure summarizes the relationship between funding to a lab (measured in 2018 BRDPI dollars) and the number of unique publications authored by the people in that lab over the 3 years following funding, separately for each NIH Institute and Center (IC) for which we have sufficient data. Each plotted coefficient is the sum of the coefficients on funding received one, two, and three years prior to a focal publication year. All estimates control for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. Ninety-five percent confidence intervals are calculated as described in Appendix B. Standard errors are clustered at the lab level.

Figure S.4. Ethnic Composition of Projects
Note: The top panel shows how (direct) spending is related to the employment measured in full time equivalent workers for employees of different ethnicities for a lab of the mean size of $362,198 in 2018 BRDPI dollars (baseline) and a lab with $100,000 in additional spending. Full time equivalent workers are estimated by prorating people by the share of their time charged to each project, their full-time / part-time status, and the number of days worked during the year. The bottom portion shows the percentage change in full time equivalent employment of each type of employee from a $100k increase in funding (along with 95% confidence intervals). Standard errors are clustered at the lab level. These are estimated from a regression of employment of each type of employee on total funding for that lab in each year controlling for lab or PI fixed effects, calendar year fixed effects, and for the number of years since a PI first received NIH funding. We divide the estimate on funding by the mean spending in each category across all lab-years. "Asian, Other" includes Indonesian, Japanese, Korean, Thai, and Vietnamese; "Western" includes Baltic, Dutch, English, French, German, Greek, Hungarian, Israeli, Italian, Nordic, Romanian, Slav, and Turkish; "All Other" includes African, Caribbean, and Unknown.