What does a research grant actually pay for?
Considerably more than the research itself — and the largest single component is frequently people, with a substantial additional amount going to the institution rather than to the project.
The direct costs:
Salaries, usually the dominant item — researchers, technicians, and a proportion of the principal investigator's time.
Equipment and consumables.
Fieldwork, travel and participant costs.
Access charges for facilities, computing and data.
Publication costs, including open access charges.
The indirect costs — the part outsiders find surprising. Universities charge overheads on top of direct costs, covering the infrastructure that makes research possible: buildings, heating, libraries, IT, finance and HR functions, technical support, and research administration.
These are genuinely real costs. A laboratory occupies space that must be built, maintained, heated and insured, and none of that appears in a project budget as a direct item. Overhead rates can add a substantial percentage to the direct costs, and vary considerably by institution and funder.
Full economic costing is the UK framework for calculating what research actually costs. Crucially, most funders do not pay 100% of it — a common arrangement pays around 80%, meaning the institution subsidises the remainder from other income. This has significant consequences: winning grants can cost a university money, and the shortfall is cross-subsidised, historically in substantial part from international student fees.
What grants generally do not cover: the time spent writing unsuccessful applications, which is enormous in aggregate given typical success rates; permanent staff costs beyond the funded proportion; and work outside the specified scope.
How funding is awarded: peer review of applications by panels and external reviewers, assessing significance, approach, feasibility, value for money and — increasingly — impact beyond academia, data management and open access plans.
Success rates are low, frequently well under a third and sometimes far lower, which means most of the effort invested in applications produces nothing.
The consequences of the model: researchers spend a large share of their time seeking funding rather than doing research, and the system favours those with the track record to be credible — which compounds existing advantage.