A regulatory wrinkle from rational expectations

The rational expectations hypothesis can be understood in various ways. One is as an equilibrium condition in a model--the model is in equilibrium when expectations of agents in the model align with the predictions of the model (though that does not mean it is a stable equilibrium). Another is that expectations of agents within the model should not be set differently from the predictions of the model without justification. That is, rational expectations is the default available-information-is-used hypothesis.*

One of Arnold Kling's recurring points in his useful and enlightening analysis of the 2008 Financial Crisis, Not What They Had in Mind, is that the regulators were not systematically cleverer than the banks and other financial institutions. That, on the contrary, both the regulators and the regulated tended to share beliefs about risk, prospects, etc. Of course, why would we expect them to do otherwise? They had access to much the same information and were using essentially the same analytical tools--given that analytical tools are part of information. (Indeed, to the extent there were differences in access to information, market participants would be likely to be better informed than regulators.)

The convergence in information and expectations was (and can be expected to be) of a somewhat interactive nature, as much of the information was created as a response to the regulations. I don't mean "made up" (though there was some of that; but where there is wealth to be had, fraud is always a potential issue). I mean that information was created both to conform to the regulations and as a result of the regulations. 

Discretionary convergences

So, does not discretionary regulation have a rational expectations problem? On what basis do we expect the expectations of the regulators to be usefully systematically different from the expectations of the regulated about market conditions? If we have no basis to expect them to be usefully systematically different about (for example) risks--and surely we do not--why do we have discretionary regulation?

One response might be: but regulators and regulated have different incentives. Well, yes; but why does that make a difference if they still have converging expectations? And creating a difference by having the regulators be significantly less informed about market conditions than participants hardly seems a desirable way to discourage expectations convergence. (Having them be systematically better informed is not a plausible situation if markets are even weakly information efficient.)

The US political system is particularly prone to creating discretionary regulations, because delegating regulatory activity to specific agencies allows more "responsive" regulation (i.e. it does not have to fight its way through Congress), because it creates someone for Congress to blame (they can distance themselves from negative political fall out) and because it permits more use of "expert" knowledge to "fill in the blanks" of legislation that is more likely to pass if it has useful ambiguity, leaving areas open to later (lobbying and) decision. 

If we at least pretend that the point of regulation is good public policy (rather than creating politically useful externalities), given this converging expectations difficulty, what is required for socially-beneficial regulation is for differing incentives to usefully create different reactions even though regulator and regulated are likely to have the same expectations about, for example, risks.

What makes this even more problematic is that, as Kling points out, financial regulatory regimes in particular are not stable:

It turns out that financial regulation is not like a math problem, which can be solved once and stays solved. Instead, financial regulation is like a chess game, in which moves and counter-moves proceed continually, eventually changing the board in ways that players have not anticipated.

A great deal of financial innovation is aimed at what Kling calls "regulatory arbitrage"--getting around regulatory constraints. So, any given regulatory regime is inherently prone to becoming increasingly detached from the actual structure of financial markets even as those structures will be significantly affected by said regulatory regime. In such circumstances, as Kling notes, regulations become ways, not so much of stopping the last crisis, as helping to create the next one.

And expecting regulators to react usefully to the changes when they will have the much same expectations as those they are regulating seems, to put it mildly, a big ask. 

So, there seems to be a rational expectations problem with discretionary regulation, particularly in financial markets.

Broad bargains

If you are really going to get around the converging expectations problem, then the incentives difference has to be maximised without enlarging the gap between market participants and regulators about market conditions--i.e. not increasing the degree to which regulators are less informed than market participants. Such can be done by making sure that--to continue with the case of banking and financial markets--the "game of bank bargains" is played in as broad a bargaining process as possible. That is, minimise the likelihood of the interests of significant groups either not being considered or being discounted. Which is another mark against discretionary regulation, because that has a fundamental tendency towards being framed by those most involved; hence the whole regulatory capture problem.

So, is your financial regulation bargain broadly or narrowly based? If it is the latter, then converging expectations (and incentives) between regulator and regulated are not likely to result in a socially-beneficial regulatory regime; and the more said regulatory bargain relies on discretionary regulation, the more that is so.

A point, by the way, that applies to bureaucratic approval processes generally. Especially when we realise that regulatory capture is not solely a feature of regulated firms, but can apply to well-organised/well-connected interests generally. Land use regulations are classic examples of that.

Broad-based bargaining producing general rules which are transparent would seem the way to go. Any system which ultimately relies on the regulators having different expectations about market circumstances than the regulated has a problem. Given that the more informed the regulators are, the more convergence in expectations is to be expected--while generating different expectations by regulators being less informed is hardly desirable--rather more scepticism about regulatory structures (especially discretionary regulatory structures) than seems generally evident seems sensible. If the point is good public policy.

Arnold Kling's suggestion of trying to have a financial system which is easier to fix rather than harder to break also seems to be worth considering.


* Economising on information and cognitive effort would presumably be what you would base any divergence in agent  expectations from model predictions on.

[Cross-posted at Skepticlawyer.]