Wednesday, 29 July 2009

Buzz about behavioural finance

Lots of behavioural finance conversations going on on the blogs today and yesterday.
  1. Chris Dillow of Stumbling and Mumbing replies to my proposal for governments to take into account cognitive bias while regulating.
  2. Simon Johnson of Baseline Scenario responds to a debate between Richard Thaler and Richard Posner about financial regulation.
  3. Alex Tabarrok from Marginal Revolution highlights the difficulty of fighting asset bubbles, even if you have overcome the challenge of identifying them.
  4. Kenneth Arrow (via Conor Clarke of The Atlantic) argues that behavioural economics doesn't predict anything.
  5. Update: A friend points out this letter in the FT from John Maule calling for behavioural approaches to be used more in regulation and investment decisions.
I'd love to have time to engage in depth with all of these debates, but let me start with a couple of key points.


Commenters on both Chris's and Simon's posts use a familiar argument to dissent from the idea of behavioural regulation. Surely, this line goes, regulators are just as irrational* as consumers - perhaps more so, since they don't have their own money at stake.

In fact, this argument is wrong in two respects. First, regulators are not as irrational as consumers, for the following key reasons:
  • They are better trained. Just as doctors know more about health than patients, regulators know more about consumer finance than the vast majority of consumers. Experimental results (see Alex's posting for some details) show that more experience and better information makes people more rational.
  • They have more time to analyse the issues. Many behavioural phenomena arise because people make decisions in the moment, when careful reflection would identify an alternative course of action.
Second, regulators are not trying to second-guess every action of consumers. They should only act in cases where consumers clearly act against their own interests.

There is a valid philosophical point about what a consumer's interest really is. It may be that the customer's genuine preference in the moment is to sign the mortgage with a low introductory interest rate, even though the rate will reset to an excessive level in six months. You could make a case relying on high discount rates or rational ignorance which justifies why it isn't in the consumer's interest to inform themselves - or care - about impoverishing themselves next year.

Just as, to use Thaler's example, you could make a case that hard-up parents might be better off buying a cheap crib which endangers their baby's life, in order to save £50 which they would instead spend on feeding their child.

But most people would agree that there are some cases in which you can clearly see a consumer's best interest being violated.

These are the cases where regulators will act - when sufficient evidence is available - not in situations where the regulator might just have a different opinion to the consumer about the best choice to make.

The broad case is that, given certain assumptions, classical economic rationality is actually the best way for us to act. We demonstrably depart from it in some situations, and within carefully constrained limits, there is a case for pushing us back to it.

Chris is not arguing against my point directly, but makes some entirely valid public choice arguments for being careful in the scope of this regulation. Of course these examples don't prove that governments should never do anything at all. Equally, they don't prove that regulators should never try to combat cognitive biases. Of course it is more fun to try to pick positions that are completely opposed to each other, but the reality is that it's all about drawing boundaries at the best achievable position on a spectrum.


Separately, Arrow's point (ha ha) about behavioural economics. This is a valid criticism of some behavioural economics research. Lots of it aims to describe departures from classic rationality but doesn't help us to build better models. If these papers do have predictive value, it's mainly at the level of the individual and doesn't shed light on group or macroeconomic outcomes.

However, some behavioural work - and I include my own research - is specifically intended to build predictive models. Arrow's own work with Debreu is my inspiration here, so he has every right to set the challenge - which is to create an alternative to the classical general equilibrium theory, taking into account better behavioural models.

Rational choice may be the Enlightenment ideal for our thoughts and behaviour, just as costless transactions, perfect information, Coasean incentives and Pareto optimality are an ideal model for a welfare-maximising society. But until we can change our biology, or supplement it with electronics - and really, would we want to? - then we had better keep looking for improved models that describe and predict our real behaviour. I think that's what Arrow means, and if so, I wholeheartedly agree.


* I must get around to a better definition of rationality, because it's a fundamental term in this kind of debate and the argument can end up being about nothing but the meaning of the word. In this post, I broadly use rationality to mean the discounted-lifetime-utility-optimising behaviour of classically modelled economic agents, and irrationality to describe departures from that model. Tyler Cowen has an excellent essay about this.

No comments: