The regulation curve

by Alan Cohen

So I came home a couple days ago and my wife was frustrated with work. She is required to budget her time for different clients very precisely, and the requirements are a huge paperwork burden and actually constrain her ability to do a good job.

At the same time, I have a meeting yesterday with Alain Beaudet, the head of CIHR (Canada’s equivalent of the US National Institutes of Health) to comment on some proposed changes to Canada’s health research funding system. One of the things I suggested to him was to require a lot less information for evaluating grants.

The link between these two things is a general relationship in human systems between how precisely and hard we try to get a certain outcome by imposing rules, and how successful we are at getting it. The result (efficiency, quality, etc) is often optimized at intermediate levels of effort (oversight, amount of paperwork, pressure we apply, etc):

Of course this is a very general statement, and it becomes less true when we become smarter in how we apply the pressure and oversight. Also, the nice symmetric curve I show above may not always hold, and may depend a lot on the scales we use to measure the oversight and efficiency. Still, I think there is a general principle here, and I’m sure some economist has given this a name at some point (the Crier curve?). It’s kind of an economic equivalent of the biological principle of hormesis. A couple concrete examples will help:

1) Medical research

We want researchers to conduct effective, honest, useful, and groundbreaking research. We can’t just give everyone money regardless of their ability to conduct effective research, so we implement grant competitions as a form of oversight. Researchers and projects are ranked based on past productivity, perceived value of the proposal, etc. The general effect of this is to greatly improve the efficiency of the spending compared to a random allocation of the money, but there are two ways in which this process can exceed its peak of ability to produce good research: First, over time, various interest groups and administrators impose more and more criteria for evaluating the quality of the research. Does it conduct analysis of gender differences? Does it use aboriginal populations? Does it include a collaboration with at least 4 universities, 3 provinces, and 5 disciplines? Now, in order to get the funding, the researchers try to jump through all these hoops rather than pursuing what would have seemed to be good research absent such criteria. Those wo succeed are not the best researchers, but rather those who are best at the politics of research. The presence of well-intentioned rules destroys the creative process. At the same time, intense pressure to perform (get grants or lose your job!) creates a culture of slight dishonesty. Rather than wanting to publish in Nature and Science because it means we’ve done good research, we want to publish there because our career depends on it. We develop a slight tendency to ignore results that contradict what we want to find (and can publish easily), and we conduct research that will boost our CV, whether or not it is good at generating knowledge. So, both for pressure on researchers and criteria for grant evaluation, intermediate levels are more productive than high levels.

2) Fishing Quotas

(Caveat: I’m guessing on this one, but it seems the principle must apply.) We want to protect fish stocks and biodiversity of oceans, so we impose quotas on the catch of various species. However, in most cases there are a few people who disobey the laws (which are often hard to enforce, especially offshore where fishermen from other countries can come in). This puts fishermen who follow the laws at a disadvantage. If the quotas are too small, the disadvantage is too large, and it will be ignored or circumvented more often. Compliance is thus likely to be highest at intermediate levels. The same principle likely applies to almost any good where too much regulation will increase the black market.

From a political perspective, this principle is interesting because it means that in most cases both the extreme right and extreme left are liable to produce disastrous outcomes. Polarization of society means that there are fewer people working in the middle to create effective policies, and more a tug of war between the extremes. Even when this results in compromise, it may be a compromise that involves far too much regulation in some places and far too little in others, rather than a balanced amount throughout.