editor_options
markdown wrap 72

SciMap-Methods

Code to accompany the manuscript "Economic Loss due to Health Funding Cuts as Distributed Across Geospatial Units" bv Mallory J Harris, Alyssa H Sinclair, Clio Andris, and Joshua S Weitz.

The two scripts used to conduct all analyses NIH_impacts.Rmd (estimate county and district-level economic losses from IDC cap and terminated grants) and NIH_analysis.Rmd (used to generate figures and calculate all results reported in the manuscript).

Input data used in the analysis are in the data folder, outputs csvs are in the output folder, and output figures are in the figs folder. More detailed descriptions follow. Code developed for R version 4.4.1. Each script should take 1-5 minutes to run.

Data

• county_pop - county FIPS, state name, county name, and population size for each county (currently used to match county FIPS to name and state)
• fips_dictionary - lists the county FIPS code of all lat/lon pairs
• geoid_dictionary_July4 - lists the Congressional District GEOID (New119Fips) and district name (NAMELSAD) of all lat/lon pairs
• JT_congress_July30 - commuter flows at the district level. COMMUTES reports the number of people commuting from Cong_ORIGIN GEOID to Cong_DESTINATION GEOID.
• NIH_prior - grant information from NIH RePORTER for grants active for several years prior to FY2024 (used to geolocate terminated grants)
• NIH_raw - grant information from NIH RePORTER for grants active in FY2024 (used to estimate direct and indirect costs)
• nih_terminations_airtable - grant terminations compiled by Grant Witness (as of August 19, 2025)
• OD_countySum001_2016 - as above, commuter flows at the county level.
• org_names_corrected - helper file to correct capitalization and abbreviations in names for terminated grants. Generated manually, does not affect this analysis.
• repeated_orgs - helper file to merge suborganizations of a single organization in the same county and congressional district that. Generated manually, does not affect this analysis.
• Ruralurbancontinuumcodes2023 - metro/non-metro classificationof counties.
• state_abbrev - state names and their two-letter postal abbreviations
• state_and_county_fips_master - dictionary of county FIPS code, county name, and state name
• state_fips_master - dictionary of county FIPS code, county name, and state name (only used to construct state_FIPS > state name mapping)
• state_multiplier-urm - breakdown of NIH awards and economic activity generated at state level (from URM)
• us_representatives_119th_congress - representative name and party affiliation by district

Output

The main output files are NIH_impact_cong, NIH_impact_county, and NIH_impact_state, which report losses at different geographic scales. Detailed descriptions of the fields within those files follow.

Column names are constructed as follows:

[loss source] [loss type] [self-reporting?] [log scale?]

loss source:

• terminated: grants that have been terminated according to grant-watch.us
• IDC: estimated losses due to the proposed cap on indirect costs, based on FY2024 funding
• grant_funds: total grant funding for FY2024
• overlap: indirect costs associated with a terminated grant (this is an overlap between the two loss sources, avoids double-counting in estimating total losses)
• combined: losses associated with both the proposed cap on indirect costs and terminated grants (accounting for potential overlap)

loss type:

• loss: raw financial loss
• econ_loss: economic loss, calculated by applying $2.56 economic multiplier estimated by United for Medical Research (UMR)
• job_loss: job loss, estimated by applying multiplier estimated from UMR report

self-reporting

• no_self: column names containing "noself," indicate that self-reported terminations were excluded from calculations (i.e., only terminations reported by governmental sources are considered). Otherwise, self-reported terminations are included (only relevant for terminated, overlap, and combined columns).

log-scale

• log: column names with the suffix "log" are log-scaled for visualization purposes.

All three files contain the following geographic identifiers: state (full state name), state_code (state abbreviation), and state_FIPS (two-digits state FIPS code). Additional identifiers for Congressional Districts are GEOID (numeric ID for congressional districts), rep_name (name of Congressional Representative), and pol_party (political party of Congressional Representative). Additional identifiers for counties are FIPS (five-digit county FIPS code, note that leading zeroes should be present in some cases), and name (county name).

Additional output files

Intermediate outputs are also provided. They are: losses estimated under static models at the county and district level (static_loss_cong, static_loss_county); losses from terminated grants at the point (institution)-level (terminated_points); losses from the IDC cap at the grant level (NIH_clean_fips); flows of economic losses between county origin-destination pairs (county_commute_NIH); and flows of economic losses into counties, districts, and states based on commuter flows (NIH_impact_county, NIH_impact_cong, NIH_impact_state); comparisons of static versus commuter estimates for counties and districts (static_vs_commuter, static_vs_commuter-CD); and sensitivity analyses of county-level estimates using state-specific economic multipliers or varying commuter influence $\rho$ (NIH_impact_county-diffmult, NIH_impact_county-sens).

Figs

Folder contains outputs relevant to manuscript figures. Files are: institutions located within a given focal county (fig1-points, for Figure 1), economic losses in surrounding counties originating from a given focal county (fig1-vals, for Figure 1), the pdf output of Figure S2 (figs2), the pdf output of Figure 3 (fig3), a latex intermediate for Table 1, the pdf output of Figure S4 (fig-supp-rho25), the pdf output of Figure S5 (fig-supp-varyrho), and the histogram for Figure S6 (supp-state_mult_hist).