Reply to Coates on financial cost-benefit analysis

Glen Weyl and I have been going back and forth with John Coates on the question whether financial regulators should use cost-benefit analysis. Weyl and I defended CBA of financial regulation here and here. Coates wrote an article criticizing CBA here. Our response is now posted, as is his reply to our response. Below is our reply to his response to our response to his response to our earlier arguments.

In his (latest) response, Coates usefully narrows the focus to the crucial issue: is there any reason to think that CBA of financial regulation and CBA of other types of regulations (like safety or environmental regulations) are different? Weyl and I agreed with Coates that lots of efforts to perform CBA of financial regulations have been shoddy, but that’s just because the methodology is at an early stage. Early environmental CBAs were shoddy as well, but they have improved greatly over the years, thanks to pressure from the White House, which ultimately forced regulators to enlist economists to help them improve environmental CBA.

Weyl and I think that, on theoretical grounds, CBA should be a lot easier for financial regulations (which are, after all, all about money, with tons of data) than environmental regulations (which involve many difficult-to-monetize valuations). Coates makes the opposite argument. In his latest reply, he makes the following points.

1. Coates rejects our merger guidelines as an example where cost-benefit principles have informed market regulation. His major point seems to be that those guidelines are not themselves instructions to perform CBAs, or the result of formal CBAs that were reviewed by courts, and that they are implemented loosely rather in a rigid way. We just don’t understand the force of this argument. The guidelines are the result of economic analysis (Weyl participated in writing them), and they basically enable the government to do cost-benefit analyses of mergers by creating certain economically informed presumptions. Mergers generate the same kind of problems about cost and benefit estimation that financial regulation does. At a minimum, this suggests that these valuation problems are not insuperable.

Coates also says that academics and practitioners admire the guidelines because they prefer rules (“constrained discretion”) over ad hoc judgments. But there is no reason to think that rules in general are better than discretion. Bad rules are worse than discretion, at least if discretion is used in good faith. The guidelines are good both because they are rules (or presumptions) and they are good rules grounded in economic principles.

2. Coates also repeats his argument that (if we understand it correctly) the natural sciences play a greater role in other forms of regulation than in financial regulation. The merger guidelines example was intended to show that this assumption is false: “market regulation” under the rubric of antitrust law is social science all the way down. But let’s consider his argument from safety regulation: the rear-facing camera. Coates argues that the main issue is one of natural science and engineering, which makes it easy to determine whether a mandate is cost-benefit justified.

A simple response to this argument is just to acknowledge that some kinds of regulation are easier than others. We certainly do not deny that. We do doubt whether the camera mandate is as simple as he says. Everything depends on how people respond to the new technology, and we know that how people respond to new technology is often difficult to predict. And, of course, safety regulations raise difficult issues about valuing human life. But the broader point is that many types of regulation seem easy just because we’ve already advanced down the learning curve. That will be true for financial regulation just as for any other type of regulation.

3. Finally, Coates seems to back off from his claim that financial regulation is special, and to argue that, across all areas of regulation, we need to distinguish between areas of regulation that have what he calls “non-stationary” (which seems to mean rapidly evolving) features and those areas that do not. And so his critique may turn out to apply to regulation of drugs, for example, or regulation of any activity where technology is changing at a rapid pace. Maybe he thinks that non-stationary features dominate the financial system but not other systems, but he doesn’t show this. Weyl and I are much more optimistic, based on recent developments in academic economics, including IO and antitrust, where (to repeat) the object of regulation is an incredibly non-stationary phenomenon–the market itself.

But if Coates is right, what does this mean? He advocates regulation based on “conceptual CBA,” which as far as we can tell, means CBA based on guesses rather than reasonable estimates.  We suspect that a more plausible response to his skepticism is not regulation but deregulation. If the government cannot explain why it is imposing costly constraints on the market, then regulation will be difficult to defend politically as well as legally.