Does it really matter to be rational?


In the April 2017 issue of the American Economic Review we find a fascinating paper by Hassan and Mertens “The Social Cost of Near-Rational Investment”. I think it deserves close attention for our work of clarifying methodological issues in Behavioural Finance. The paper uses a standard Real Business Cycle and Rational Expectations Model to gauge the economic effects of ‘near-rational behaviour’, both in theoretical terms and presenting an empirical estimation. I think it raises important philosophical questions, although the authors do not have these in mind.

The central argument is straightforward and does not need much maths. If agents are rational, it does not pay for them to be perfectly rational, because there are small mental costs of that, such as effort. These costs can be close to negligible. The authors even present an estimation of these costs, based on empirical data on forecasting errors collected by the Association of Professional Forecasters. The costs may be around 0.13 percent of permanent consumption. But near-rational behaviour causes high social costs (estimated about 5.31 percent of permanent consumption) because markets operate as an aggregation mechanism that mediates externalities among agents. Small errors feed into the market process and bias asset prices. Agents take those prices as information for updating their information. The externalities accumulate, so that the information processing function of markets is damaged. A particularly important result of the paper is that this effect becomes the stronger, the more dispersed private information is, hence the more noise it carries, because in this case markets would be most valuable in correcting for the errors in private information. In this case, the individual returns of perfectly rational behaviour (expending effort for correcting the private information) are also the lowest, thus further driving the individual behaviour that creates huge social costs.

I think that the paper reinforces a point that I submitted in an earlier blog-post, namely that rationality is a public good. That creates a paradox, because it is rational not to be perfectly rational. This is an argument that fits into the line of the ‘rational ignorance’ literature in Public Choice. Further, we learn that markets only can work efficiently if agents are fully rational in the sense that they do not take the costs of rationality into consideration. If they do, they create externalities. In this sense, full and perfect rationality cannot be equated with ‘individual rationality’, but is a form of social rationality: Everyone a social planner in terms of welfare theory!

The paper can be evaluated in different ways in terms of our general philosophical concerns. On the one hand, it shows that the established models of rationality and markets always contain the seeds of paradox which are carried along with the very notion of rationality, if it includes reflexivity (such as rational ignorance). On the other hand, if near-rationality already produces high social costs, what about the much more serious failures of rationality that are the topic of standard Behavioural Finance? Wouldn’t we conclude that markets are a disaster, under these circumstances?

Mertens has another paper where he discusses the policy implications of these arguments. His work is macroeconomics, and so the focus is on government action in stabilizing the business cycle, such as consumption smoothing and, most importantly, interest rate policies that take asset price gaps in the previous period into consideration, with the effect that agents would pay less attention to current asset prices, but also to future government policies. The effect is that they would rely less on the market in forecasting economic activity, but on their private information, thus reducing the externalities which are mediated via the market.

I suggest that we can also adopt an evolutionary perspective: Perhaps societies evolve institutional, normative and behavioural mechanisms by which the interaction between markets and agents is improved in establishing specific behavioural information structures. Only considering government policies seems far too narrow here, and raises the standard argument about government failure. The real question is whether both Behavioural Finance, sticking to the norm of rationality in judging behavioural ‘failures’, and the Rational Expectations models present a distorted picture of real-world markets. This is the point where narratives may come into play, just to name one aspect raised in my previous blogs. Narratives are external to the market mechanisms in the narrow sense, as they direct attention to factors which are external to the price-quantity dynamics, such as stories about technological prowess of entrepreneurs, comparisons across different historical episods, and so forth. The narrative would create more leeway between private expectations and current market dynamics. In other words, expectations would become less ‘rational’, in the formal sense, but more reasonable, as the narratives trigger reflective thought, highlighting private information, and resulting in apparently ‘deviant’ behaviour, given the current market situation. These are the ‘animal spirits’, indeed. They turn agents less rational, but more reasonable, thus undergirding better functioning of markets.

Cited papers:

Hassan, Tarek A., and Thomas M. Mertens. “The Social Cost of Near-Rational Investment.” American Economic Review 107, no. 4 (April 2017): 1059–1103. doi:10.1257/aer.20110433.

Mertens, Thomas M., Volatile Stock Markets: Equilibrium Computation and Policy Analysis (November 21, 2011). Available at SSRN.


Funded by: