INET published a paper, written by John Kay, that deals with the relationship between economics and the world we live in. The Map Is Not the Territory: An Essay on the State of Economics spells out methodological critiques of economic theory in general, and of DSGE models and rational expectations in particular.
INET forwarded Kay's paper to a handful of economists and invited them to respond. Here we offer a perspective by Roger Guesnerie, Professor of Economic Theory and Social Organization and Chairman of the Board of the Paris School of Economics.
Guesnerie, as Michael Woodford before him, takes issue with rational expectations. To Guesnerie, rational expectations is a hypothesis, but “dominant economic discussion and practice have most often transformed the Hypothesis into an unavoidable and supposedly uncontroversial modeling axiom.” The optimism embedded in the efficient market hypothesis, Guesnerie continues, “has been one of the main sources of the recent turmoil: it has opened the floodgates to overconfidence, and weakened regulatory forces everywhere.”
A Response to John Kay
by ROGER GUESNERIE
There is little doubt that the crisis is a phenomenon that belongs to the intellectual category of “market failure”. It is also the case, as argued in the stimulating text by J. Kay, that it is associated with some kind of “economic theory failure”. John Kay makes a long list of points and an exhaustive comment is hopeless and probably counterproductive. Let me make three types of remarks. The first one (1) concerns the core assumptions of modern economic theory. The second (2) addresses John’s discussion of the “efficient market hypothesis”. My final comments (3-4) are a response to his criticism of the present scientific style of economic theory.
The present crisis has attracted legitimate methodological criticism that concerns two core elements of the methodology of the economic profession: the rationality hypothesis and the rational expectations hypothesis.
- Although the assumption that economic agents behave rationally makes sense, it has been rightly argued that real economic agents do not quite resemble homo-economicus. This point has been recurrently stressed by Paul Krugman in his New York Times columns, but is also forcefully made in the joint book of George Akerlof and Robert Shiller. In fact, the fast development of behavioural economics provides some, still early, response to this line of criticism.
- The Rational Expectations Hypothesis (from now the REH) is a conceptually different construct. The success of the hypothesis may be due to the semantic option (it is uncomfortable for an economist to feel irrational). Whatever it may be, a generation of economists has, in majority, believed that the “rational expectations hypothesis is nothing else than the extension of the rationality hypothesis to expectations”. Unfortunately, this latter assertion is plainly wrong: it is “rational” to have rational expectations only when others have rational expectations. Hence, in an economic situation involving many interacting optimizing agents, a rational expectations outcome (called a rational expectations equilibrium) appears as a kind of focal point, associated with a successful (in some sense) coordination of expectations. Such an outcome, or the underlying “equilibrium concept” is indeed a “Hypothesis”. Present dominant economic discussion and practice have most often transformed the Hypothesis into an unavoidable (and supposedly uncontroversial) modelling axiom. I agree with Kay’s rejection of this axiomatic view. Interpreting REH as an axiom has tremendous implications for the understanding of economic policies. (As Kay points out, “ricardian equivalence”, for instance, hangs on the validity of that axiom.).
- Directions for a critical assessment of the rational expectations hypothesis have been explored, particularly in the last two decades, but the urgency of deepening the reflection should now become crystal clear. In fact, I, Michael Woodford and others, are so convinced of the urgent need for a critical assessment of the REH, that we have together launched an “international network on expectational coordination”, (see Woodford’s contribution in this blog). The network gathers scholars who do not view the REH as an undebatable axiom, although their earlier work in the subject has followed different lines. Should the Hypothesis be generally dismissed as seems to be suggested in John Kay’s text? To what extent, and/or in which cases, will it remain a focal reference? The network participants certainly have different views on these questions and on others. They do not necessarily agree together but “agree to disagree”. I, personally, find it somewhat unbelievable that the REH should be equally reasonable across situations, (as required to make it a universal modelling option).
The critical assessment of the REH has a priori far reaching implications both in macroeconomics and finance. Concerning finance, John Kay discusses at length the Efficient Market Hypothesis (from now the EMH). Although I side with him to see the EMH as a flawed view of financial markets, I somewhat disagree on the reasons. His objections take too seriously the counter-objection that exceptions and anomalies to the EMH would be “too small to matter”. Such a counter-objection looks somewhat unconvincing and although this is not the place to produce a comprehensive academic discussion of the Hypothesis, let me make three points. All of them deserve, in my opinion, to be put at the forefront of intellectual debate, and all of them stress the critical role of the “rationality of expectations” for a theoretical and empirical assessment of the EMH.
- There is no consensus on the exact meaning of the EMH, and I will not propose my own definition. So far as intuitive flavour is concerned, the EMH has two sides. On the one hand, the EMH sustains the claim that markets collect well (and aggregate and then transmit) the information detained by market participants: market efficiency is informational efficiency. On the other hand, market efficiency is often taken to mean that “you cannot beat the market”. The two formulations of the EMH may be viewed are roughly equivalent... whenever expectations are “rational. However, this equivalence dramatically fails, if the REH is not taken for granted. Plausible models of expectational coordination can then describe worlds where the market works poorly, although you cannot beat it.
- The claim of market informational efficiency, the first intuitive meaning of the EMH, is supported by models which adopt the REH. However, if the REH is not taken as a modeling axiom, and if the plausibility of the RE outcome is assessed along the lines of some robustness criteria, the analysis might become somewhat devastating. For example, the more information the hypothetical Rational Expectations market is supposed to transmit, (i.e. the better it performs) the more fragile – i.e. the less plausible - is the corresponding outcome. In this context the fact that markets can transmit much information is implausible.
- On empirical grounds, attention has been rightly directed on eliciting the actual (real time) beliefs of agents. It seems to be the case that EMH fails to explain an important empirical observation. Recent analysis of surveys of financial actors’ views suggests for example, that at the top of a bubble, participants generally overwhelmingly believe that it will continue. This pattern flies in the face of REH.
So both the empirical and the theoretical cases for the EMH are much weaker than is generally argued. All one has to do to appreciate the significance of this observation is to consider how dangerously optimistic the view of the working of the financial markets implicit in EMH is. Indeed, some of my friends, who know well the financial system, think that such an optimism, fuelled by economic theory, has being one of the main source of the recent turmoil: it has opened the floodgates to overconfidence, (among market actors already prone to it), and weakened regulatory forces everywhere. But let us go on:
- Although the just sketched analysis sustains critical conclusions similar to Kay’s conclusions, it does not support his criticism of the “scientific style” of present economics. In a sense, the present scientific style has to be connected with the manifesto of the Econometric Society, produced in the first issue of Econometrica in 1933. The working model of the Econometric Society, adopted by a small minority in the profession before the Second World War, became somewhat hegemonic in the 8o’s, even if it is being transformed by recent developments. My response here is that the struggle to free our minds of the grip of the REH is too important to be left to “unscientific” economics (whatever that might be). Although there would be much to say on the recent developments of the Econometric Society working model, the division of labour – between theory, on the one hand, and empirical investigation (based on statistics and sophisticated methods) on the other hand – has been a powerful motor of understanding. On the empirical front, nobody would argue, I guess, that careful empirical investigation has to give up rigorous methods. On the theory front, although it is true that the rationality hypothesis and the rational expectations hypothesis have made modelling superficially easier, their critical assessment does not imply one has to abandon modelling. On the contrary, I do believe that rigorous reasoning, with the support of appropriate formal tools, will be essential input for improving our understanding of expectational coordination.
- The re-examination of the scope of the rationality hypothesis and a complete reassessment of the domains of validity of the REH are overdue. It has just been stressed that these tasks do not lead to dismiss the demand for theoretical modelling. This does not mean, obviously, that they are easy tasks and that they will not affect, perhaps deeply, our understanding of economic phenomena. For example, the second task, the re-assessment of the REH, will touch the roots of the philosophical determinism that has shaped economic culture, all schools of thought together. Outside a REH world, prediction is much more difficult, apart from any problem of non-stationarity (stressed in Kay’s text). The standard economic wisdom may be shaken up.
I would like to finish with some remarks about another, different, characteristic of modern economics, the increased balkanisation of its knowledge. Such a criticism, I believe, is implicit in John Kay’s contribution, and it may be appropriate to make it more explicit, even if I can touch upon the problem only briefly.
a) Frontiers in modern economic research have been multiplied, but the progress inside each front is probably less and less understood by outsiders, even from neighbouring subfields. A balkanisation of knowledge exists also in the natural sciences, but it is more of a problem in an economic (or social science) context than in physics: social action has often to rely on all dimensions of understanding. When, for instance, the question concerns systemic stability, not only of the financial system, but also of the economy as a whole, the need for synthesis of different partial viewpoints is paramount.
However, in order to combat balkanisation, a legitimate objective, the present scientific style, again, may not be the right target for attack. Returning to the Econometric Society program, one should notice that by putting emphasis on a general and neutral language, and on universal tools, it has been instrumental historically in fighting geographical and ideological balkanisation. This is not to deny that one may legitimately feel that mathematics increases the entry costs in many areas. Should we however throw out the baby (mathematics and modelling) with the bathwater? No, rigor is not a luxury; it is a basic need. Erroneous arguments in economics are not refuted by the observation of facts (contrarily to what often happens in a natural science like physics).
b) Fighting balkanisation certainly requires a better interaction between basic empirical knowledge and scholarly investigation, as well as a better dialogue between economic specialists and between economists and historians. Fighting the logic of entrenchment that is subverting modern research in economics is clearly desirable. But this takes us beyond the important questions raised by Kay’s text.
 Robert Solow argued, some time ago, that the field of economics could be defined as a field where it is “socially acceptable” to pursue your own interest.
 Although applications to finance of the “bounded rationality” viewpoint may not have yet been fully convincing (with a number of exceptions including for example Bolton, P, Scheinkman J, and Wei Xiong, (2006). Review of Economic Studies).
 As noticed by Barro (1984) American Economic Review
 In game theoretical terms, the so-called rationality of expectations is a “Nash equilibrium”, not a “Dominant Strategy equilibrium” situation.
 “A” focal point, not necessarily “the” focal point, since it needs not be unique…
 As stressed by Mike Woodford, in his comment on this blog.
 Although it is true, as argued by JK that small anomalies can generate large departures (on this issue, see also for example Branch-Evans, American Economic Journal, (2011)
 Although this is Bob Lucas’ argument, for whom I have the highest regard, personally and intellectually, I come to a different conclusion, for the theoretical and empirical reasons mentioned below.
 Behavioural finance has also generated objections of different nature
 The claim that you cannot beat the market is the (retreat?) position of some early proponents of the EMH. The retreat is safe behind the fortifications of the REH, but might become a rout if the protection turns out to be much weaker than assumed…
 Although the REH allows some critical assessment of the EMH à la Grossman S. and J. Stiglitz (1980) American Economic Review.
 See Desgranges, G., Geoffard P.Y and Guesnerie R., (2003), The Journal of the European Economic Association,
 The intuition is that the fact that agents do not discard their personal information when the market transmits a huge amount of information, although true in the REH equilibrium, is not “eductively” robust i.e. it makes the mental coordination process more and more fragile.
 The point has been made repeatedly (for example by Adam-Marcet (2010), mimeo, when considering the price-dividend ratio and in other contexts by Frydman-Goldberg (2011), Princeton University Press)
 Econometric evidence suggesting that models embodying the REH can mimic actual financial trajectories is not conclusive, when the models under scrutiny cannot reproduce actual beliefs. The evidence is then more of a tribute to the cleverness of the modeling than to the validity of the REH.
 All of the above is a criticism of EMH taken as a claim of the efficiency of the market in gathering the information, in other words as a claim of the wisdom of the markets, which many economist, and journalists, take for granted.
 Where it was accompanied by a text of explanation of Joseph Schumpeter
 Again, this point is made in a much more detailed way in Michael Woodford on this blog