Tuesday :: Mar 3, 2009

The Irrational Expectations Hypothesis

by eriposte

In response to my post yesterday criticizing "economist" William Poole, Mary linked to her post at Pacific Views late last year that summarized some of the discussion at a conference held at Columbia University in October 2008, titled "Can We Save the World Economy?" with featured guests Nouriel Roubini, George Soros and Jeffrey Sachs. I must have missed this back then but I have to thank Mary for the link. I'm including below the entire video of the roughly 90 min or so of discussion which I found both interesting and engaging, in which all three experts thoroughly challenged some of the pernicious "conventional wisdom" parading around as economics expertise in today's world.

Can We Save the World Economy? from Earth Institute on Vimeo.

In keeping with the thrust of my last two posts, let me focus on the portion where Sachs discusses the economics profession, roughly starting at the 58:00 minute mark in the video. I'm going to paraphrase what he says (not a perfect rendering), with the emphasis being mine (throughout this post):

Since the early 1980s...the way we train people to think...in mainstream economics and mainstream politics...has left them almost unable anymore to distinguish the surface from the underlying reality. Not only was it the age of Reagan and the beginning of market fundamentalism that came in the early 1980s, and the Rational Expectations revolution and all the rest, but a fundamental break in how we actually train our students to think....Because the new kind of economic modeling for 25 years, the one that won all the Nobel prizes, said - you don't have to understand the deep picture - you have to look at the prices you see on the surface and infer the deep picture from that. In other words, the surface tells you the depth because the surface and the depth couldn't be two different things. You don't really have to know underlying mechanisms in the economy because the prices reflect the underlying mechanisms. In fact, you fit the deep model, if there is one, by looking at the surface and then inferring what the deep model must be. That's literally how students are trained. So they're not trained to be skeptical. They're trained to fill in parameters of a system which assumes from the start that what you see on the surface is what you really have. Whereas what fundamental, proper skepticism ...and also what the logic of economic bubbles shows ... and what the logic of biophysical bubbles show ...is what you see on the surface can completely hide and obscure what's happening below. And it is our job as scientists and responsible citizens to understand the deep points and therefore help to rectify the discrepancies and keep ourselves away from trouble.

I will get to the Rational Expectations Hypothesis (REH) in a moment, but let me say first that Sachs' comments resonate with me not just because of the morass we find ourselves in today but because there has been a general degradation in the quality of research within parts of the academic profession over the past say 15-20 years or so, for various intrinsic and extrinsic reasons. I find this to be true even in the "hard" sciences. In one of his books - I forget now whether it was in QED: The Strange Theory of Light and Matter or in The Character of Physical Law - the famous Nobel Prize winning physicist Richard Feynman, who was never afraid to call his colleagues to the mat when needed, remarked that a key aspect of the era of quantum mechanics that allowed a whole range of new findings and discoveries about nature was the importance given to skepticism and the focus on identifying and understanding what aspects of conventional scientific wisdom were incorrect. If I recall correctly, he wrote that this was a change from a position where established scientific dogma went unchallenged for a while (see James Gleick's comments on Feynman and the notion that the alternative to doubt is authority). I am pretty sure that Richard Feynman, if he were alive today, would have had very strong words for the economics profession, where doubt was in fact replaced with authority on a large scale. Here is a very short video clip of a portion of a Feynman lecture that I think many people in academia and research should watch, especially students of economics - it is a clip that goes well with Sachs' comments above and is a good lead-in to the rest of this post.

Sachs' mention of the Rational Expectations Hypothesis (REH) is interesting because I sometimes like to think of it as the Irrational Expectations (of Economists) Hypothesis - somewhat in keeping with the utterly ludicrous comments by William Poole on the alleged non-existence of bubbles. Which brings me to the paper - "Financial Markets and the State: Price Swings, Risk, and the Scope of Regulation" by Roman Frydman and Michael Goldberg of Columbia's Center of Capitalism and Society (via Mark Thoma at Economist's View).). This is another paper worth reading and I'm going to feature a few extracts here. Let's start with Frydman's email to Thoma:

In connection with your post of Paul’s text on the failure of economics, Ned Phelps argues this is not about Neo-Keynesians or new Classicals, it is about economics that ignores the key feature of modern economies: the way they unfold over time cannot be represented by fully predetermined models (those modeling the economy with fixed rules implying a single probability distribution for the past and the future).

In fact the problem is the Rational Expectations Hypothesis, which MIT and Chicago happily share. As we argue in "Financial Markets and the State: Price Swings, Risk, and the Scope of Regulation" (and a number of recent academic papers) the mechanical nature of “scientific” models, which economists construct, prevents us from understanding the swings and design of regulatory policies. It is also dangerous, because it may lead to regulation that throws out the baby with the bath water. I think it is time that we leave “the dream of mechanical markets”, something truly difficult for both Paul and Bob Lucas.

In their paper, Frydman and Goldberg discuss the serious limitations of REH when applied to complex economic problems:

Asset prices and risk premiums reflect market participants’ decisions to buy and sell, which depend crucially on individuals’ forecasts of future market outcomes. To model fluctuations in asset prices and risk, an economist must account for market participants’ strategies – the factors that they consider relevant and how they use them to forecast market outcomes. Because modern economies are defined by their capacity to innovate, individuals cannot afford to stick to one strategy endlessly to forecast the future. This feature of decision-making in real-world markets makes the task of modeling fluctuations an extremely daunting challenge. Remarkably, the vast majority of economists presume the opposite: rational individuals forecast according to a fixed rule.11 This bit of magical thinking relies on two premises, both of which underpin the so-called “Rational Expectations Hypothesis” (REH). The first premise holds that a rational individual bases her forecasts of future values of fundamental variables and prices solely on the same model that the economist himself writes down. What has made such forecasts seem “rational” to a vast majority of economists is the second, equally strong presumption: save for a random deviation that averages to zero, an economist’s model exactly characterizes how fundamentals unfold over time and how these variables translate into the “true” value of every asset. According to REH, then, all rational individuals forecast on the basis of the same understanding of the relevant fundamental factors and how they affect outcomes.

In an REH world, market participants’ forecasts of future values of fundamentals and prices are not only correct on average; their decisions to buy and sell also result in a market price that is exactly right, fluctuating randomly around its long-run fundamental equilibrium. In setting prices, the market correctly takes into account any change in the “true” fundamental value, as long as information about such changes is transparent and widely available.

Of course, market participants may not have sufficient access to relevant information for various reasons, such as the presence of highly complex derivative instruments, loose disclosure requirements and accounting standards, or fraudulent practices. Inadequate competition and misaligned incentives may also be a problem. In such situations, regulation should be introduced to eliminate the informational asymmetry and other market imperfections. In other words, if one assumes that REH models actually provide an adequate account of how asset markets function, fixing these problems is all that would be needed for them to price assets, on average, at their “true” fundamental values.

Assuring these ideal conditions, however, would not eliminate long swings in asset prices. [...]

They point out that the REH has been improved using "Rational Bubble" models:

Although standard REH models cannot account for swings, they can be modified to generate bubble movements. “Rational” bubble models retain REH as a characterization of market participants’ forecasting strategy during a bubble. In these models, a bubble is formed if, for some reason unrelated to fundamentals, market participants somehow all decide to base their forecasts on some extraneous factor, and this leads them to believe that the price will increase exponentially at some predetermined rate. By design, REH presumes that this belief is self-fulfilling: the asset price would also move away from its “true” fundamental value at the same predetermined rate.

Rational bubble models recognize that although asset-price swings away from benchmark levels may be wide, they do not last forever: eventually asset prices start moving back toward benchmark values. As with a price swing away from the benchmark, bubble models attempt to capture this reversal with a mechanical rule: in forming their REH forecasts, market participants assume that the bubble will eventually burst, and all place the same exact probability on this outcome. When the bubble does burst, the market is assumed to return instantly to its “true” fundamental value.45


But appealing to manias to explain long swings in asset prices suggests that these movements are an aberration from otherwise “normal” times, during which the market sets asset prices at their “true” fundamental values. In fact, long swings in asset prices are the norm, not the exception


Over the last two decades, economists have uncovered massive evidence that casts serious doubt on REH-based accounts of long swings: REH is grossly inconsistent with how individuals form forecasts in real-world markets. There is also much evidence of the empirical failure of other components of the standard notion of rationality, such as its specifications of preferences.48 This evidence has led to the emergence of behavioral economics, which makes use of empirical observation to motivate its assumptions about individual decision-making. 

However, even behavioral economics tends to use REH as a starting point (the authors use the term "Irrational Bubbles" to discuss this concept):

However, despite their emphasis on empirical realism, they have continued to retain REH’s empirically and epistemologically flawed conception that markets discover the “true” fundamental value of every asset, and that an economist can exactly specify these values with his model. Behavioral models portray bubbles as swings away from these “true” values.49


In motivating their approach, behavioral economists usually appeal to empirical observations that participants in financial markets use trend-following rules. Because behavioral economists retain REH as a cornerstone of economic rationality, they interpret forecasting following chartist and other non-REH strategies as evidence of participants’ “irrationality.”50 As with “rationality,” they specify “irrationality” with mechanical rules that are supposed to capture how an individual’s forecasting strategy will be revised between now and a distant future. 


Like their REH counterparts, “irrational” bubbles are unrelated to fundamentals and serve no useful social function. If policy officials could eliminate them, the market would return to setting asset prices at their “true” fundamental values, and the allocation of capital would thereby be unambiguously improved. The policy guidance that emerges from this approach is thus the same: leave markets unimpeded, except when a bubble forms, at which point act decisively to cut it off as early as possible. 

But, unlike REH models, behavioral models suggest that such a policy is relatively straightforward. Instead of the formidable task of fighting crowd psychology and manias, all policy officials need to do is start a short-term price trend back toward the “true” fundamental value. According to behavioral models, this would lead both chartists and fundamentalists to respond mechanically to the new trend and bid the price back to this value. But this implication is contradicted by experience and research.52

The paper is worth reading to understand some of the pitfalls of modern economic theory and is a good companion paper to "The Financial Crisis and the Systemic Failure of Academic Economics" by David Colander, Armin Haas, Katarina Juselius, Thomas Lux, Hans Foellmer, Michael Goldberg, Alan Kirman and Brigitte Sloth.

P.S. On the Finance side, read this post by Yves at Naked Capitalism: "In Praise of More Primitive Finance".

eriposte :: 7:25 AM :: Comments (9) :: Digg It!