Not that I'm bitter or anything, but...
Richard Gordon (Department of Radiology, University of Manitoba) and Bryan J. Poulin (Faculty of Business Administration, Lakehead University)
Using Natural Science and Engineering Research Council Canada (NSERC) statistics, we show that the $40,000 (Canadian) cost of preparation for a grant application and rejection by peer review in 2007 exceeded that of giving every qualified investigator a direct baseline discovery grant of $30,000 (average grant). This means the Canadian Federal Government could institute direct grants for 100% of qualified applicants for the same money. We anticipate that the net result would be more and better research since more research would be conducted at the critical idea or discovery stage. Control of quality is assured through university hiring, promotion and tenure proceedings, journal reviews of submitted work, and the patent process, whose collective scrutiny far exceeds that of grant peer review. The greater efficiency in use of grant funds and increased innovation with baseline funding would provide a means of achieving the goals of the recent Canadian Value for Money and Accountability Review. We suggest that developing countries could leapfrog ahead by adopting from the start science grant systems that encourage innovation.
Update: I sent the link and abstract to the faculty in my department. Four have replied so far: 1 said, "That's crazy!", another said, "I think they've cooked their numbers," and two (both very senior) said they were very skeptical. One went so far as to say:
It would be more convincing if the second author was not from Lakehead which would do very well from getting rid of refereeing!
When asked, though, both of those senior faculty admitted that they hadn't actually read the paper...
Second update: the (junior) prof who thought they had cooked their numbers sent the detailed analysis included below, which I think is more useful (and fairer) than dismissing the paper unread because of where the authors are from...
Look at table 3. The only lines which make a difference there are the time spent by the faculty members & students writing the grants, and the administrative cost of NSERC. For the former, they provide no support for the number of hours PIs or trainees spend on writing the grant -- I think the number 120 should be much lower. I spent ~3 weeks writing my NSERC, but this was while doing several other things in parallel, so the number is at least twice too big in my case, and it was my first experience writing a grant---I think I can do a decent job in about 40 hours now. The additional 2 weeks of work on each grant by students/postdocs I think is really unsubstantiated. Almost all PIs write the grants on their own; if they ask a student for something it is a figure that the student can also use in their paper.
Secondly accounting for administrative cost by multiplying the admin costs by discovery/all grants ratio does not take into account that DGP are 5 year grants, while other grants tend to be shorter, and hence have higher admin overhead. I am too lazy to do this, but I'd be curious to see what fraction of all grant applications the DG are---the number shouldn't be hard to get.
And finally this does not take into account that grants are very easy to recycle---people who are rejected at NSERC resubmit substantially similar proposals to other funding agencies (or NSERC next year), so I think the real costs are much, much lower. There are also advantages of writing grants---I actually find that writing a grant helps me organize my thoughts on the topic---better preparing me to write a paper on it, or helping figure out what to do next in a certain field.