This is nothing other than the problem of the stupid customer. If the work measured by the annual reports is not up to snuff, then the grants should be curtailed. I've done grant-based research work, and have produced degree theses as documented output. A good part of the research work and funding goes to the support of graduate students as research assistants. If the work is not being evaluated, then the customer is stupidly throwing money out the window with no discernible benefit.