University libraries are often discussed as support services, while research grants and economic output are treated as separate measures of institutional performance. A new US analysis suggests those parts of the academic system are much more tightly connected than budget lines imply.
Researchers R. Eugene Turner and Suzanne L. Marchand examined library expenditure, university research and development funding, global university rankings, arts and humanities strength, state gross domestic product and per capita income. Their results reveal strong statistical relationships across several levels of the US research system, including a ratio of $21.34 in R&D expenditure for every $1 of library expenditure among 71 public universities.
The findings, published in Scientometrics on 24 September 2026, do not show that an extra dollar spent on a library directly creates $21.34 in research funding or $264 in state economic output. The study is observational, combines measures from different periods and examines institutions and states rather than experimentally manipulating budgets. What it does provide is a quantitative map of how research infrastructure, academic breadth and economic indicators move together.
Looking beyond the usual measures of research strength
Research performance is commonly described through publications, citations, patents, rankings and grant income. Turner and Marchand argue that this can obscure the infrastructure that makes scholarship possible. Libraries provide journal subscriptions, books, datasets, archives, specialist collections, interlibrary loans and professional expertise, yet their expenditure is rarely placed alongside research funding in analyses of university performance.
The researchers therefore asked whether library expenditure was quantitatively related to university R&D spending, whether arts and humanities strength was associated with the same research ecosystem, and whether university R&D expenditure tracked broader economic prosperity at state level.
This matters because the relationship can easily be misread in either direction. A wealthy research university may be able to afford a large library because it already attracts substantial funding. Alternatively, stronger information infrastructure may help researchers prepare proposals, collaborate and produce scholarship that improves their ability to compete for funding. Both processes may operate simultaneously, alongside institutional size, history, reputation and regional wealth.
The analysis connected several large datasets
The study drew library expenditure data from the Association of Research Libraries. To reduce the influence of single-year fluctuations, the researchers used the most recent three years available, 2020 to 2022, covering 102 university research libraries in the United States. These institutions were separated into public and private universities.
University R&D expenditure estimates came from the US National Science Foundation. For the state-level analysis, the team summed 2014 to 2020 R&D spending for 332 universities that each recorded more than $10 million in expenditure. Several jurisdictions were excluded for specific structural reasons, including the unusually large influence of Johns Hopkins University on Maryland’s totals and the distinctive concentration of non-university libraries in Washington, DC.
The researchers also compared continuous scores from major university ranking systems, including the Academic Ranking of World Universities, QS World University Rankings, Times Higher Education World University Rankings and the Aggregate Ranking of Top Universities. Arts and humanities programme scores were incorporated to examine whether strength outside science and engineering was associated with the wider research enterprise.
State per capita income data were taken from 2023 figures, while the researchers used regression analyses and principal component analysis to examine how the institutional variables clustered. The different time windows are important when interpreting the results because the analysis is not a same-year causal test.
Public and private universities showed different spending relationships
Among 71 public universities, library expenditure alone described 67% of the variance in R&D funding. The estimated slope corresponded to $21.34 in R&D expenditure for every $1 in library expenditure.
The relationship was weaker and differently scaled among 29 private universities. There, the slope was approximately $9.60 in R&D expenditure per $1 of library expenditure, and the regression described 35% of the variance in research funding. A formal comparison found that the public and private slopes differed significantly, with F = 17.1 and p < 0.0001.
That difference is useful because it cautions against treating all universities as interchangeable. Public and private institutions differ in funding structures, endowments, research portfolios, tuition models and disciplinary composition. A single national ratio would therefore hide meaningful institutional variation.
The analysis also found that universities with ranked arts and humanities programmes tended to occupy a stronger R&D position. The researchers estimated an average difference in intercepts equivalent to $234 million per year in R&D expenditure between institutions with and without ranked arts and humanities programmes. They interpret this as evidence that arts and humanities strength is compatible with, rather than a drain on, a high-performing research system.
A principal component analysis separated size from research infrastructure
One possible explanation for the library relationship is simple scale: large universities have more students and faculty, larger libraries and larger research budgets. The principal component analysis helped the authors examine that possibility.
For public universities with an arts and humanities ranking, the first principal component explained 53% of variance and the second explained 22%. Library expenditure, R&D dollars and arts and humanities scores clustered together, while measures associated with university size formed a different cluster. The near-right-angle relationship between those clusters indicated that size was largely independent of the library and research expenditure grouping within this sample.
This does not eliminate every possible confounder. Institutional prestige, historic funding advantages, local industry, faculty composition and research specialisation could still influence both library budgets and grant income. It does, however, make the simple explanation that the pattern is only a consequence of university size less convincing.
The pattern extended from universities to state economies
The strongest state-level association involved total economic output. Across the geographical units analysed, variation in university R&D expenditure described 86% of the variation in state GDP. The fitted relationship corresponded to approximately $264 in gross state economic production for each $1 of university R&D expenditure.
When population was added to R&D expenditure in a multiple regression predicting state GDP, the adjusted R² rose to 0.98. Population is naturally a major determinant of total economic output, so this result should not be interpreted as evidence that university research alone explains nearly all differences between state economies.
The population-normalised comparison was more modest. Variation in per capita university R&D expenditure described 34% of the variation in state per capita income. The estimated relationship was $59.51 in per capita income for each $1 of university R&D funding per capita. Massachusetts recorded the greatest university R&D expenditure per person, while Delaware had the highest per capita earnings among the jurisdictions considered.
These ratios are striking, but they are associations rather than literal rates of return. A state with a productive technology, healthcare or advanced manufacturing sector may simultaneously generate higher incomes, attract research-intensive universities and support larger R&D budgets. Research universities may also contribute to those industries through skilled graduates, knowledge transfer, startups and collaboration. The observational design cannot fully separate those pathways.
Why the findings matter for university budgets
The study’s practical argument is not that administrators should mechanically multiply library budgets to obtain a predicted grant return. Instead, the ratios can be viewed as institutional guardrails showing the scale at which research support and research activity have historically coexisted.
That distinction is especially important as universities face rising subscription costs, expanding data requirements, artificial intelligence tools and pressure to demonstrate direct financial returns from support functions. A library budget may look like an overhead when considered in isolation. In a system-level view, access to literature, datasets, archives and specialist information services forms part of the infrastructure through which researchers develop questions, prepare proposals and communicate results.
The arts and humanities result broadens that argument. The data do not support a simple picture in which disciplines outside heavily funded science and engineering necessarily weaken research competitiveness. Institutions with ranked arts and humanities programmes showed stronger R&D profiles, although the analysis cannot determine whether those programmes themselves generated the advantage or whether well-resourced universities are simply strong across multiple domains.
Important limitations keep the ratios in perspective
The authors explicitly describe the work as observational rather than experimental. Correlation therefore remains the central limitation. The analysis cannot establish that higher library spending causes greater R&D expenditure, or that university R&D causes the measured differences in state prosperity.
The datasets also cover different periods. Library spending was measured from 2020 to 2022, university R&D data span earlier years, and GDP and income were measured in 2023. Research investment can take years to affect publications, innovation, employment or regional income, so the appropriate lag structure is uncertain.
State boundaries are another imperfect unit of analysis. Research collaborations, graduate migration, corporate investment and knowledge spillovers routinely cross state lines. A university can therefore contribute to economic activity outside the jurisdiction in which its expenditure is recorded.
The study also focuses on large research institutions and cannot automatically be generalised to smaller universities, teaching-focused colleges or higher-education systems outside the United States. Future work could examine publication types, citation patterns, discipline-specific time lags and alternative combinations of expenditure years.
A wider view of what supports research
The most useful contribution of the analysis may be its insistence that research output is produced by an ecosystem rather than a grant office alone. Across the institutions studied, larger library budgets were strongly associated with larger research budgets, arts and humanities strength coexisted with greater R&D funding, and university research expenditure tracked regional economic indicators.
None of those findings turns correlation into causation. They do, however, challenge narrow budget discussions in which libraries, broad disciplinary portfolios and research funding are treated as unrelated competitors. The data suggest that successful research universities may be better understood as integrated systems whose information infrastructure, academic breadth and research capacity develop together.
Source Information
Study: Libraries, research and development funding, and economic prosperity: data for the holistic university
Authors: R. Eugene Turner and Suzanne L. Marchand
Journal: Scientometrics
Published: 24 September 2026
DOI: 10.1007/s11192-026-05787-8
