Rational Expectations Hypothesis, DSGE models and the Behavioural paradigm
_ This blogpost is an extract of a future publication on the role of expectations _
Throughout the development of economic theory, the importance of expectations has grown, significantly influencing both research and economic policies (1). Although it is now almost impossible to find an economic discussion that does not consider this concept, economists' attention to how expectations are actually formed and how they interact with other economic variables has followed a more discontinuous path, with cycles of highs and lows (2).
Here, we focus on briefly illustrating the paradigm that has most influenced the development of economic models, namely the one based on the rational expectations hypothesis (3), as well as some of its developments.
The Rational Expectations Hypothesis
The concept of expectation in economics has been reshaped over the past fifty years in response to the famous Lucas critique (4). Lucas criticized the prevailing paradigm, which assumed that expectations were invariant to economic policies, arguing, quite rightly, that expectations would be impacted and therefore altered by these policies. However, his critique went much further, endorsing the ineffectiveness of policies (5), supported by the powerful construct of the rational expectations hypothesis (6). This hypothesis presents a situation where the agents' subjective average probabilities about the distribution of relevant variables coincide with the conditioned probabilities in the "true" model of the economy, making "optimal" use of all available information and deviating from perfect prediction only due to some random noise. This is why policies are ineffective: if agents have rational expectations, policies may introduce more noise but cannot, on average, alter the economy's trajectory.
This formulation is based on the assumption that actors use all available, scarce, and costly information in the most efficient way possible, and thus create expectations that:
i) are endogenous to the model (7);
ii) coincide with the expected value of a probability distribution (8);
iii) at time t+1 are conditioned by the information at time t.
Therefore, rational expectations (9) represent the most accurate predictions that agents can make based on the information available to them. This suggests that, in aggregate, economic agents do not make errors in their predictions, even though they may make mistakes individually. In other words, expectations are accurate on average. The forecasting error will therefore be a random variable with a zero mean and no serial autocorrelation.
This theoretical framework has become the paradigm generally used to make economic policy choices through Dynamic Stochastic General Equilibrium (DSGE) models (10) – developed in the seminal works of Kydland & Prescott (11) and Long & Plosser (12) – which today represent the macroeconomic standard and are used by most major international institutions. These models are therefore based on the concept of equilibrium, which, in this specific case, is defined as follows: possible expectations for tomorrow are calculated assuming that there are no significant changes to the known possibilities today (probability distributions are invariant over time). Consequently, this implies the stationarity of economic variables.
Limitations of the Rational Expectations Hypothesis and DSGE Models in the Complexity of the Economic System
Now, although in certain stable and conventional contexts these models can guide the understanding of dynamics, it is not difficult to notice how their blind application to the complexity of the economic system proves to be detrimental. In fact, although DSGE models do include non-linear functions, such non-linearity is not applied in this specific sense, meaning that it does not generate emergence, typical of open and complex systems (like the economy).
The recurring financial system crises are partly manifestations of the failure to model such dynamics more complexly, so every crisis catches us unprepared, repeatedly renewing a coercion that cannot be contained.
The following graph (13) clearly shows the inconsistency of the rational expectations hypothesis through the example of unemployment in the United Kingdom from 1860 to 2011, highlighting significant events associated with it.
We see that the path is not stationary at all, but there are unexpected changes both of an extrinsic nature (such as the two world wars) but also, like the 2007/2008 crisis, of an intrinsic nature, i.e., inherent to the functioning of the economy. The causes here must be sought within the system, not outside. This simple example illustrates, once again, how the assumptions underlying DSGE models are insufficient.
Some Critiques and Responses to the Rational Expectations Hypothesis: The Behavioral Approach
Although the rational expectations hypothesis has represented and still represents one of the pillars in the construction of economic models (we mentioned DSGE models) and economic policies, numerous critiques have been made, especially regarding the assumption of perfect information for agents and the consequent predictability of economic variables. In this section, we briefly present some of the most famous critiques and responses, focusing on the analysis of the economic agent.
Simon (14) replaced the concept of "maximization" with that of "satisficing" and coined the term and concept of bounded rationality, proposing it as an alternative foundation for economic modeling.
Sargent (15) demonstrated how these assumptions are highly unrealistic: in reality, information is costly, and individuals’ processing abilities are limited.
Kahneman and Tversky (16) experimentally demonstrated the failure to adhere to the principle of economic rationality in risky situations, constructing their "prospect theory."
In general, these critiques of economic rationality attempt to integrate psychology into economic models. We agree with Dow, who asserts that this attempt has been partly limited and limiting.
Indeed, in the behavioral strand that developed from the critique of the rational expectations hypothesis, there is a clear separation between what Dow defines as "cognition" and what he defines as "emotion." The conventional framework was maintained, with some adaptation to accommodate experimental evidence, but without changing the perspective. Kahneman himself pointed out that behavioral economics maintained the basic structure of standard economic theory, limiting itself to introducing cognitive obstacles to explain deviations from rational behavior.
What emerges is the conflict between so-called normative rationality (what should be) and instrumental rationality (practical). In the behavioral paradigm, the unexplained is often classified as irrational, rather than revisiting the theory to account for these deviations in a broader conception of rationality. In this framework, psychology focuses on individual preferences and decision-making processes, standardized for analysis and simplified to fit models with "representative" agents, where deviations are considered irrational, essentially viewed as a "random shock."
As we have seen, psychology in this approach reinforces the dualism between rational behavior and irrational behavior.
In general, both in the paradigm of the rational expectations hypothesis and in the behavioral one, it is easy to notice how economic actors are not truly configured as agents. Instead, they are operationalized, becoming homogeneous (or representative) re-agents to given conditions. Thus, expectations are essentially a derivative of what has already happened.
In the end, everything remains the same, until the next crisis, which is itself operationalized through a random shock.
Therefore, there is no real room for purposeful action. The goal is defined a priori by the model, and the re-agents are designed to achieve it.
Now, what needs to be done instead to build a theory of expectations that truly incorporates psychology is to overcome this duality. One of the economists who has most successfully considered economic behavior within a complex theoretical framework, thus formulating the role of expectations based on a different conception of economic rationality that truly takes psychology into account, was Keynes. For Keynes, it is not even necessary to define behavior as "rational" or "irrational"; what matters are the reasons that lead to economic action, which develops within an open system characterized by uncertainty, alongside his logical view of probability.
Thus, he introduced a concept as widely cited as it is misunderstood: that of "animal spirits" to describe the impulse for action despite uncertainty.
In the next blogpost, we will delve deeper into Keynes' thoughts on expectations.
_ This blogpost is an extract of a future publication on the role of expectations _
Notes
(1) For an overview of the development of the treatment of expectations, see Arena et al. (2021). Expectations in Past and Modern Economic Theory. Foreword.
(2) See on this point Visco, I., & Zevi, G. (2020). Bounded Rationality and Expectations in Economics. We agree with Visco and Zevi in asserting that at the extreme of the "high cycle," we find Keynesian theory, which acknowledges the significant role of expectations in the functioning of the economy. This role is also linked to fundamental concepts in the theoretical framework of the British economist, such as the differentiation between risk and uncertainty. We will delve deeper into this in the continuation of the article.
(3) We emphasize that this theoretical construct is not synonymous with the principle of economic rationality, which has a longer history and broader meaning and has always played a fundamental role in the development of economic theory (from Smith, Jevons, Stuart Mill, Pareto, Robbins, Keynes, Simon, to contemporary formulations). Its importance lies in the fact that it represents the fundamental behavioral premise, as well as one of the epistemological foundations of economic theories. See on this Zouboulakis, M.S. (2014). The Varieties of Economic Rationality. In the history of economic thought, this principle has evolved, been modified, rejected, revisited, reformulated, and at different times grounded on various bases: psychological, social, and logical-computational. For an in-depth review of the literature and a historical-philosophical analysis, see also Mele, A. R., & Rawling, P. (Eds). (2004). The Oxford Handbook of Rationality.
(4) Lucas, R. E., Jr. (1976). Econometric policy evaluation: A critique.
(5) In addition to Lucas, prominent proponents of this hypothesis (key figures in neoclassical theory) include Sargent and Wallace. See Sargent, T. J. & Wallace, N. (1981), Some Unpleasant Monetarist Arithmetic. According to this perspective, any economic policy intervention that is publicly announced and thus becomes known to market participants has no impact on real variables. This is because agents, being rational, anticipate the future effects of such policies and adjust their behavior accordingly. In other words, they act in a way that ensures their predictions are actually realized, a phenomenon known as "self-fulfilling prophecies." In this framework, therefore, only random events can lead to changes in economic variables relative to their natural state.
(6) This construct was originally developed by Muth—see Muth, J. F. (1961), Rational Expectations and the Theory of Price Movements. Although Muth's contribution was initially intended for certain specific and rather narrow circumstances, Lucas extended it, adopting it as a necessary condition of consistency in macroeconomic models.
(7) This means that the formation of expectations is an endogenous process within the economic system, that is, internal and influenced by the dynamics of the system itself.
(8) That is, the conditional expected value.
(9) In note 3, we observed how the concept of "rationality" differs from the construct of the "rational expectations hypothesis" as presented thus far. In this latter case, in agreement with Dow, "rationality" is nothing more than the application of deductive calculative logic (i.e., formal logic) to a set of given premises (optimization of behavior with respect to a set of preferences, endowments, and technologies). This, therefore, requires agents to make calculative choices among all possible options (including contingent ones) and for these choices to be consistent. As long as deductive logic is correctly applied and contradictions are avoided, the system itself will be internally consistent. See Dow, S. (2013). Keynes on Knowledge, Expectations and Rationality, p. 119.
(10) These models, while preserving the concept of equilibrium, integrate microeconomic aspects into the modeling and, most notably, incorporate the possibility of random shocks to justify the presence of economic fluctuations. We will not delve here into a detailed critique of such models but will limit ourselves to highlighting their Achilles' heel: introducing a random shock within the model explains nothing about the phenomenon itself but merely serves to justify a forecasting error. Therefore, it does not fundamentally change our understanding of the phenomenon. A clear and comprehensive critique can be found, among others, in Gallegati, M. (2022). Il mercato rende liberi e altre bugie del neoliberismo.
(11) Kydland, F. E., & Prescott, E. C. (1982). Time to Build and Aggregate Fluctuations.
(12) Long, J. B., & Plosser, C. I. (1983). Real Business Cycles.
(13) The chart is taken from a 2014 article on VoxEU. See Mizon, G. & Hendry, D. (2014, June 18), Why DSGEs crash during crises.
(14) Simon, H. A. (1955). A Behavioral Model of Rational Choice.
It is interesting to read on p. 99 of the cited paper Simon's description of his project (italics mine): "to replace the global rationality of economic man with a kind of rational behavior that is compatible with the access to information and the computational capacities that are actually possessed by organisms, including man, in the kinds of environments in which such organisms exist."
Simon proposes a new conception of rationality that takes into account the capabilities possessed by organisms, including humans, within an environment. As we will see, this theoretical project is entirely aligned with the argument we will propose and with the concept of naturalization that we have outlined. Nevertheless, we do not believe Simon succeeded in effectively advancing this project.
(15) Sargent, T. J. (1993). Bounded rationality in macroeconomics.
(16) Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk.
(17) Dow, S. C. (2011). Cognition, market sentiment and financial instability.
(18) For a definition of the terms "cognition" and "emotion," I refer to the cited article by Dow. It will be interesting to see how this terminological dualism is resolved in the concept of drive.
(19) Kahneman, D. (2003). Maps of bounded rationality: psychology for behavioral economics, p. 1469.
(20) As Dow points out, this theoretical drift has also been fueled by a misinterpretation of neurophysiological evidence. Since the areas of the brain responsible for reasoning and emotion are distinct, an analytical separation was made in the development of economic theories.
(21) This approach has its roots in the Scottish Enlightenment: Adam Smith himself embraced a much more nuanced conception of rationality that certainly did not endorse this dualism. On the contrary, he saw sentiment as a driver of knowledge. His views are well articulated in The Theory of Moral Sentiments. See Smith, A. (1759). The Theory of Moral Sentiments.
(22) Akerlof and Shiller themselves use the term to encompass a wide range of "irrational" behaviors; their approach remains closer to the behavioral paradigm, perpetuating the dualism and failing to fully grasp the Keynesian interpretation. See Akerlof, G.A. & Shiller, R.J. (2009). Animal Spirits: How Human Psychology Drives the Economy, and Why It Matters for Global Capitalism.
Bibliography
Akerlof, G.A: & Shiller, R.J. (2009), Animal Spirits. How Human Psychology Drives the Economy, and Why It Matters for Global Capitalism. Princeton University Press.
Arena, R., Dal-Pont Legrand, M., & Guesnerie, R. (2021). Expectations in Past and Modern Economic Theory. Foreword. Revue d’économie Politique, Vol. 131(3), 323–332. https://doi.org/10.3917/redp.313.0005
Dow, S. C. (2011). Cognition, market sentiment and financial instability. Cambridge Journal of Economics, 35(2), 233–249. http://www.jstor.org/stable/24231970
Kahneman, D. (2003). Maps of Bounded Rationality: Psychology for Behavioral Economics. The American Economic Review, 93(5), 1449–1475. http://www.jstor.org/stable/3132137
Mele, A. R., & Rawling, P. (Eds.). (2004). The Oxford Handbook of Rationality. Oxford University Press. https://doi.org/10.1093/oxfordhb/9780195145397.001.0001
Mizon, G., & Hendry, D. (2014, June 18). Why DSGEs crash during crises. VoxEU. https://cepr.org/voxeu/columns/why-dsges-crash-during-crises
Long, J. B., & Plosser, C. I. (1983). Real Business Cycles. Journal of Political Economy, 91(1), 39–69. http://www.jstor.org/stable/1840430
Sargent, T. J. (1993). Bounded rationality in macroeconomics. Clarendon Press. http://www.gbv.de/dms/hbz/toc/ht005051088.pdf
Simon, H. A. (1955). A Behavioral Model of Rational Choice. The Quarterly Journal of Economics, 69(1), 99–118. https://doi.org/10.2307/1884852
Smith, A. (1759) The Theory of Moral Sentiments, D.D. Raphael and A.L. Macfie (eds.), Oxford: Oxford University Press, 1976.
Visco, I., & Zevi, G. (2020). Bounded rationality and expectations in economics. Banca d’Italia. https://doi.org/10.32057/0.QEF.2020.575
Zouboulakis, M. S. (2014). The Varieties of Economic Rationality. Routledge.
Comments