Indecently long proposals

The business of applying for grants is swallowing up researchers' time to research, argues Peter Bowbrick. I once spent a month writing a very carefully planned, researched and targeted proposal for a one-year grant. However, so did a dozen other people, with the result that 13 months were spent applying for 12 months' research.

The business of applying for research grants is taking up more and more of researchers' time, leaving less time for research. The research selectivity exercise emphasises what research funding has been obtained, with some questionable evaluation of the "quality" of the research that is done with it, and none at all of the value of the research to the nation. As a result the advertisements for lecturers, readers and professors increasingly specify that the applicant must have a record of attracting funding.

This means that the ambitious, and the army of contract researchers whose immediate ambition is to get another contract, are all trying to outdo each other in producing more proposals and fuller and better proposals.

However, there is a fixed amount of money available. The result of the competition is more and more time spent for the same amount of money. Ambition being what it is, it is difficult to see where the limits lie. This is a classic example of a badly designed market producing an unwanted product - applications - instead of the desired product - research (or, more accurately, useful research).

The effect has been dramatic, one example being my grant application above. In another case, one grant-giver advertised widely and expensively that it was giving away Pounds 7,000. I wrote for details and was sent application form number 250. A quick calculation suggested that the payout would be Pounds 7,000 divided by 250 people (at least) divided by 14 days writing: I dropped it. About one in five applications for Economic and Social Research Council grants is funded. The ESRC is not able to give figures for the amount of time it funds but roughly guesses (optimistically, I believe) that it funds 30 months of salary per funded application.

At four to six weeks per application this means that 17 per cent to 25 per cent as much time is spent on preparing applications as in doing the research. Add to this the ESRC admin costs and the time of the referees and you find that the amount of time spent on applying is probably more than a third.

In one ESRC research programme, more than 300 people were spending two to three weeks writing outline proposals, of whom 19 per cent were allowed to submit full proposals and 8 per cent eventually got grants. Again figures are not available on the research time actually funded, but an estimated per cent as long was spent on writing proposals as on actual research. In this particular case the professor organising the research programme improved the outcome from the per cent using only ESRC money to 15 per cent by getting other money at the last moment.

There is no reason to believe that this competition selects the best research for funding - what evidence could possibly be produced? There is a lot of reason to believe that it selects average or below average projects.

The ESRC can only fund a third of the top proposed projects - those that are alpha-rated - so the choice between these is arbitrary. But, there is no reason to believe that alpha-rated projects are better than the others - the assessment of probable outcomes must be far less reliable than the notoriously erratic refereeing of journal articles.

Anything brilliantly original or non-routine is certain to annoy some people. With four referees it is likely that such a proposal will be violently attacked by one or two referees.

Referees also stand to lose money and promotion if they approve projects that compete with their own, particularly if it is likely to supplant or refute their work. Even if a small minority behave in this way, consciously or unconsciously, the effect can be significant if there are four referees. This is another failure of the system. Outside academia, in the world of the professional, the design of a market includes a guard against moral hazard. Yet here there is a major moral hazard in the key decision.

Clearly, therefore, if there were no research selection exercise, there would be perhaps 25 per cent more time available for research and the quality of the research produced would be unquestionably better. The quality of the output of a research programme should be judged on the amount of good or very good research produced. It should not be judged on the absolute amount of poor research produced, and certainly not on the proportion of the total output that is of poor quality.

An extreme solution would be to let recent graduates enter a lottery. The successful ones would be funded for their whole career, subject to some precautions against abuse. The unsuccessful ones would go into business, and not waste any time getting a PhD. And of course nobody would write research proposals. This procedure would be very similar to the system that operated in the days when Britain led the world in the social sciences.

Less extreme (and less efficient) would be a system where today's researchers, or their departments, would enter a lottery, with the winners allowed to fund their pet proposals. Of course this system would have ruled out Einstein if he had happened not to win the lottery (but what a businessman he would have made). But neither would Einstein have stood a chance in today's system: he had no history of research grants; he did not want any serious money for equipment, just a pencil and paper; he had only written one paper; he worked in a patent office not a university; he did not state his methodology; his idea was incompatible with existing theory (he obviously had not read the literature) - and he did not state what his outcomes would be.

Dr Peter Bowbrick is a consultant in the economics of markets.