When submitting budgets for research proposals, researchers try to allocate all the advertised award money. So, e.g. for a $300 K announced award, one sees proposals ranging from $299 K to $300 K exactly. In practice, these preliminary budgets rarely turn out to be accurate, and though expenses come up that were not anticipated, sometimes expenses turn out lower. However, the investigators conducting the research for an awarded grant, have no incentive to spend less, as the money not spent would simply revert back to the funding agency. This is why we see all manner of expenses cropping up as a grant approaches closure, with extra, unnecessary travel, consulting, or equipment expenses charged to the grant. It is also why, when purchasing supplies and equipment, there is rarely an effort on the part of researchers to save money: the federal government pays, and what's not spent is lost. I've often seen extra, unneeded supplies bought because they may come in handy for some future, different project. In practice, they often go unused or become obsolete. This is a terrible waste. It is also very different from what one can see in corporate research (i.e. not through grants).
The suggestion: build incentives to spend only what is necessary by allowing the investigators (as private individuals) and/or their institution to keep a fraction of the money not spent, e.g. 1-5%. This can add up to a substantial incentive (thousands of dollars), for junior invstigators in particular. In my experience as a former academic researcher, many researchers, especially senior ones, would also choose to use these funds for other job-related useul purposes, such as teaching supplies, special research projects etc. but they should have the option of keeping it for themselves.
This way 1) the purpose of funds spent would be honestly stated 2) there would not be wasted/obsolete purchases 3) the federal government would actually save on excess funds disbursed on research projects