The National Science Foundation collects the Higher Education Research and Development Survey (HERD) from institutions in the US that participate in significant amounts of research. This survey is the primary source of information on research expenses in U.S. colleges and universities according to the NSF, and is often used to determine trends and patterns in research on a national scale.
This survey is intended to be a census of all schools participating in substantial research in the US. Annually, NSF collects data from approximately 900 institutions, targeting schools with at least a bachelor's degree in any field and at least $150,000 annual spending in research and development. Responses are self-reported by the institutions, and these responses are checked for consistency with previous responses, with schools given the chance to resolve both consistency issues and missing responses when possible.
Schools are asked to answer a number of specific questions regarding their funding sources and expenditures. Each question provides a few attributes of possible funding, including funding sources (federal vs. non-federal, etc.), purpose (research vs equipment, etc.) and field of research that funding is used in.
The complete HERD data is available at the NSF HERD site. The data doesn't always line up exactly with the institutions as we've identified them through the Department of Education's IPEDS data, and sometimes the data for several institutions is combined, which is noted by us when relevant. We are using files from 1972 to 2015 downloaded on July 10,2017 and we received a HERD crosswalk to help assign IPEDS UNITIDs to the institutions on September 15, 2017 from a HERD program officer. The November 2019 HERD data for 2018 was downloaded January 7, 2020.