1When I started reading The Behavioral Economics of John Maynard Keynes, I could not figure out what would be the specific literature dealt with in the book. For sure, the subtitle Microfoundations for the World We Live In sets the stage for the ambition of the author to provide behavioral elements regarding economic agents and the most important decisions that bear on the final macroeconomic state of the economy, in order to convey an empirical foundation for explaining involuntary unemployment in a capitalist free-market economy. However, I had no prejudice regarding the approach that would be taken, the kind of behavioral literature that would be used or the methodological rationale that would be proposed. I just felt that Keynes’ views about the psychology of agents are complex. On the one hand, Keynes’ thought was rooted in the tradition of social psychology and casual empiricism, somewhat following Marshall’s thoughts on expectations. On the other hand, his views were derived from an in-depth reflection on the nature of knowledge and action, connecting together his thoughts on probability, uncertainty, animal spirits and beauty contests. I thus wondered how behavioral economics could help us to deal with this complexity.
2After reading the book, I think that Ronald Schettkat fails to offer a convincing behavioral reconstruction of Keynes’ General Theory, although he does succeed, here and there, in establishing bridges between Keynes’ behavioral observations and some experimental results and principles in the field of behavioral economics. My opinion is that Schettkat does not address the question about what interpretation of Keynes’ General Theory to retain and how the various behavioral assumptions in the book form an interdependent system from which certain macroeconomic properties can be inferred. Instead, he regards it as obvious that Keynes’ General Theory offers a new concept of equilibrium and that the confrontation of various behavioral elements is enough to obtain a macroeconomic system with unemployment equilibrium.
3As Schettkat puts it, the goal of the book is to show that “Keynes’ revolutionary macroeconomic consequences are based on his microeconomic foundations derived from observations of economic behavior and conditions of the world we live in, which are no less revolutionary than his macroeconomics compared” (19). To the reviewer, this sentence contains both (i) the claim that a set of behavioral findings are sufficient building blocks to obtain Keynesian features in a macroeconomic system and (ii) that the findings in behavioral economics, like Keynes’ behavioral observations, are “realistic” and not hypothetical, i.e., that they are in tune with everyday observations on markets and not with an Arrow-Debreu abstract economy. Regarding this last statement, that Keynes and behavioral economics share a common realistic flavor, Schettkat does not offer an in-depth comparison of the status of behavioral statements in Keynes and the status of behavioral statements in behavioral economics, but he compensates for this lack of scrutiny by hammering on every other page that behavioral statements à la Keynes deal with the world we live in.
4For sure, Schettkat himself recognizes that merging together Keynes and behavioral economics is not obvious. The introductory chapter aims at understanding what kind of behavioral foundations are missing in the General Theory. The kind of behavioral economics to be taken on board depends on the kind of interpretation of the General Theory that is retained. Keynes’ revolution is portrayed around the identification of macroeconomic regularities involving (or allowing for) durable underemployment, rejecting classical interpretations of equilibrium, focusing on monetary values (vs. real values) and on workers’ and investors’ behaviors. In other words, Keynes’ theory is a set of empirical regularities, and Schettkat does not claim to search for a strong connectedness uniting them. On this strange, atheoretical base, the author sets the stage for various themes in behavioral economics that deal with Keynesian behavioral assumptions for agents in a monetary capitalist economy: bounded rationality (Chapter 3, “Rational Choice: A Normative concept”), uncertainty and animal spirits (Chapter 4, “Choice under Uncertainty: Animal Spirits”), behavior through time (Chapter 5, “Expectations over Time”), social interactions and the evolution of individual preferences (Chapter 6, “Socially Embedded Individuals”). Each chapter is conceived as both a contribution to enriching the microfoundations of General Theory and an argument against neoclassical theory. Actually, about half of the content of the book is devoted to criticizing the unrealistic nature of neoclassical theory, and one might wonder what the real goal of the book is.
5The findings and theoretical elaborations of behavioral economics most discussed in the book are linked to prospect theory, notably to reference points and biases and heuristics, which Schettkat uses amply to scaffold his views about time dynamics and non-ergodic representations of economic history. He also makes wide use of evolutionary interpretations of human behaviors (Gigerenzer and his co-authors’ work) and of the body of literature associated with animal spirits and some classics of behavioral macroeconomics (Akerlof, Shiller, De Grauwe and co-authors). However, the author ignores some literature in behavioral macroeconomics specifically addressing Keynes’ economics: e.g., “Behavioral Economics and the Economics of Keynes” (Pech and Milan, 2007) and “Can Post Keynesians make Better use of Behavioral Economics?” (Jefferson and King, 2010; see also Koutsobinas, 2014; d’Orlando and Sanfilippo, 2010; Bourgeois-Gironde and Guille, 2011). Likewise, the field of behavioral finance is ignored, hence likely connections with Keynes’ thought on speculation are not addressed (e.g., Westerhoff and Dieci, 2006). Given the importance of financial and banking sectors in all trends of macroeconomic modeling since 2008, this missing element is at odds with both Keynes’ views and contemporary transformations of macroeconomics. The rest of the references are mainly to classical papers in microeconomics and macroeconomics of various trends (post-Keynesian, neoclassical, new classical, new Keynesian), with an impetus on empirical analysis and theoretical modeling of the labor market.
6Strictly speaking, the book contains six chapters, coming after a first introductory chapter and followed by a conclusion (Chapter 8, “The Economy We Live In”) where the author repeats once more the gist of his message:
Keynes’ macroeconomic conclusions are based on observed microeconomic actions of humans as entrepreneurs, workers, consumers, and investors in a monetary economy, impressively confirmed by the rigorous research in Behavioral Economics applying several methods (observations, interviews, experiments, neurological analysis). Both Keynes and Behavioral Economics require “good economic theory” to be externally consistent—descriptive—with the abilities and actions of humans in the economic environment we live in. The methodological orientation of Behavioral Economics and Keynes is that of real science, which requires an empirical content of the behavioral assertions. This is no less than a fundamental shift in the requirement of good economic theory, a methodological revolution. (149)
7The introductory chapter sets the stage to identify the methodological commonalities between Keynes and behavioral economics and opposes them to neoclassical assumptions, both with respect to human agency and to market modeling (including money and time). Neoclassical economics, it is said, assumes stability of multiple markets, neutrality of money, calculus of probabilities, representative agents, rational expectations, perfect rationality (utility maximization), individual autonomy (no social embeddedness). In a nutshell, Keynes and behavioral economics are ruled by an inductive methodology, while neoclassical economists follow axiomatic principles. This neoclassical model, it is said, emerged in the wake of the counter-revolution against Keynesian economics following Phelps and Friedman’s papers on the natural rate of unemployment. To the author, Keynes’ microfoundations for macroeconomic regularities are manifold, though they were not fully developed. With only a few exceptions (Hansen’s 1936 review of the General Theory), commentators have reduced Keynes to macroeconomic aggregate relationships. As for behavioral economics, it essentially offers a (more) realistic account of individual behavior than neoclassical assumptions of rationality. Even prospect theory, framed as a maximization of expected utility, rests “on the overwhelming evidence that the neoclassical axioms are invalid as a description of actual human behavior” (14). Here, Schettkat discards accusations that behavioral economics is neoclassical in disguise (Sent, 2004; Berg and Gigerenzer, 2010) and aligns with the idea that other motivations operate to rationalize the findings of behavioral economics, even though some behavioral economists choose to present their findings as an enrichment of rational choice theories. So, the argument goes, behavioral economics, taken as a broad field of inquiries and experiments that depart from the fiction of an autonomous rational agent, provides the seeds for a new revolution, for a paradigmatic change à la Kuhn. Another fundamental divide between the Keynes/behavioral economics compound and neoclassical theory deals with normativity. To Schettkat, behavioral economics and Keynes’ economics are descriptive, while neoclassical theory is prescriptive, for the worst.
8Chapter 2 discusses fundamentals. It can be read as a second introduction to the book, setting the stage for the main topics of the next chapters: Is economics a logical or a real science? Can an axiomatic model be sufficient to understand the economy? Can economics benefit from a knowledge of actual economic behavior? And if it can, then which methods are valid to elicit this knowledge (observations, experiments, interviews)? The author answers these great questions by advancing trivial arguments against neoclassical economics and presenting the various sources of realistic knowledge used in behavioral economics or in Keynes (regarding market coordination or uncertainty). An important and interesting point is about the concept of equilibrium. To Schettkat, Keynes’ concept of equilibrium, like the neoclassical one, deals with a balance of contrary forces that compensate each other. The difference lies in the fact that Keynes’ equilibrium is devoid of any quality and is not a stable local attractor, while the neoclassical one is characterized as a welfare optimum and is stable through the laws of supply and demand. Here Schettkat’s discussion is fuzzy, borrowing to all kinds of criticisms addressed to neoclassical assumptions regarding rationality and market coordination. Then he mentions a result from behavioral economics—the difference between willingness to pay and to accept—which, it is assumed, contradicts the neoclassical view of markets. Eventually, Schettkat does not come back to his initial opposition between two kinds of equilibrium and does not draw any specific conclusion from it. This way of proceeding, jumping from one idea to another, is repeated throughout the book. To the reviewer, it makes the reading all the more wearisome and it eschews the important question of interpreting Keynes’ underemployment equilibrium. In the end, we have no idea of how behavioral economics results of various kinds may provide a solution to bridge micro and macro behaviors. After criticizing representative agent economics, the conclusion is just that “Keynes made clear that economic theory needs to relate to the world we live in and that useful economic theory cannot totally abstract from reality.” (47)
9Chapters 3 and 4 present criticisms of rational choice theory and choice under uncertainty (subjective expected utility), especially decision theory. It presents the basics in a pedagogic way (Bernoulli, Allais’ paradox, Kahneman and Tversky’s value function, bounded rationality, Ellsberg’ paradox). Then, it comes to Keynes’ concept of probability and confidence in his Treatise on Probability (85-92)—a theme already discussed in Chapter 2, §2.6, one of several examples of the disorder of arguments in the book. The difference between an urn containing a 50/50 share of black and white balls and another urn containing an unknown share of both balls is just that the first urn bears a higher confidence than the second. As new information is gathered regarding the second urn, the guess about the share may evolve but the confidence about the share will also increase, hence the weight of the arguments involving this knowledge. Behavioral scientists express the same thing through an assumption of preference for more certain alternatives (Kahneman, 2011), but Schettkat does not discuss the literature on ambiguity aversion. In an environment with more or less uncertainty (weights of arguments) and when decisions—such as investing—have to be made, animal spirits (i.e., the “innate urge to activity”, Keynes) will make them possible. Here, the author borrows from Akerlof and Shiller’s (2010) analysis of how animal spirit explanations rely on a process of individual emotional factors self-inferring each other through a contagious process in the face of uncertainty. Schettkat sides with the view that Keynes did indeed attribute significant importance to animal spirit arguments, as is shown both by celebrated passages in General Theory, e.g., that “a large proportion of our positive activities depend on spontaneous optimism rather than on a mathematical expectation, whether moral or hedonistic or economic” (90). This entry of emotions into the heart of decision-making echoes Damásio’s (1994) lifelong neurological approach to action and decision (see also Lerner et al., 2015).
10Chapter 5 is devoted to “Expectations over Time”. To the reviewer, it is the most interesting chapter for grasping what behavioral economics can and cannot contribute to Keynesian economics. In Keynes’ economics as in behavioral economics, time is an essential component of analysis, since the path of events and the contingencies of feedback effects imply that historical time is non-ergodic, i.e., the path of sequences experienced by individuals and the fact that experiences induce positive or negative feedbacks, continuously determine and reshape their expectations about the future. So, historical time shapes our preferences, which are fundamentally not stable. This is in complete opposition with new classical economists’ views on preferences and behaviors, as expressed by the rational expectations revolution and Lucas’ stance that economics deals essentially with a stabilized system of preferences and behavioral patterns (Lucas, 1986). In this respect, even Keynes does not allow for a full recognition of the sources of expectation revisions over time (changes in the stock of capital are neglected). The bridge between short term and long-term expectations depends on confidence and is not fully rationalizable: “It is reasonable, therefore, to be guided to a considerable degree by the facts about which we feel somewhat confident, even though they may be less decisively relevant to the issue than the other facts about which our knowledge is vague and scanty.” (Keynes, 1936, 148) Schettkat connects this view about decision under uncertainty with the availability and representativeness heuristics in behavioral economics (96). In the same vein, our tendency to base our decisions on our current situation, “in a sense disproportionally” (Keynes, 1936, 148) is confirmed experimentally as routine behaviors and “status quo bias”. But, as mentioned by the author, the comparison with behavioral economics ends when Keynes conveys some kind of collective decision rule, the “convention” and references to mass psychology. In the same chapter, the author addresses the issue of ergodicity vs. non-ergodicity of economic phenomena. After presenting an almost pedagogic example of its meaning (making sense of the sequences of independent random events vs. an ahistorical representation of average gains and losses), the author does not draw the conclusions from his presentation. He switches to another paragraph on discount rates. Here, Keynes’ views are at odds with behavioral findings: decreasing discount rates over time (Thaler, 1981), lower discount rates with increasing magnitude of stakes (Kahneman and Tversky, 1984), hyperbolic discounting (Ainslie, 1991).
11Chapter 6 deals with the fact that economic agents are “Socially Embedded Individuals”. The focus is first on wages and the author argues that behavioral findings are “strong confirmation of Keynes’ observations.” (115) The gist of the argument is that in this area, workers tend to make their decisions on the basis of a reference point, i.e., from their current situation, usually in monetary terms and not in real terms. In the case of the appraisal of wages, people will tend to refer to their relative wage and will assess wage variations in reference to evolutions in other sectors. This passage also offers a view of the author’s ecumenical blend of arguments, using reference to studies based on survey data, magnetic resonance imaging, lab experiments, and natural experiments. When it comes to Keynes’ behavioral insights regarding consumption, the fundamental psychological law comes in at the aggregate level, and is completed by ratchet effect arguments at the individual level (involving the social dimension of consumption). Schettkat does not confront these statements directly with behavioral findings. He continues—in a very disordered way—presenting famous behavioral results (endowment effects and loss aversion), then implicitly applying these notions to labor supply behavior, then returns to consumption and mistakes made by consumers in evaluating the utility of experience goods (such as cell phones).
12Chapter 7 is titled “The Resurrection and Fall of Homo Œconomicus”. Overall, it discusses the development of the neoclassical view about employment and the labor market in the 1970s in the wake of Friedman’s natural rate of unemployment assumption. As the story goes, new Keynesians were not well-equipped to buttress monetarist and new classical arguments; they failed to account for unemployment on the basis of frictional/micro-social arguments, though they had advanced various explanatory items (efficiency wage, motivation models à la Akerlof, models based on signaling or habit formation). The chapter reads like a criticism of neoclassical economics, pointing out methodological arguments of Friedman and then of new classical economists. It does not make use of any reference to behavioral economics.
13Without even trying to gather together the ideas scattered in the book, Schettkat’s conclusion (Chapter 8) seems to acknowledge that in the end, behavioral economics so far has not been able to dethrone the axiomatic-neoclassical framework. In my view, it is rather an admission of the failure of his own attempt to show us how behavioral economics could serve as a basis for a return to Keynes (or at least to Keynesian ideas). The reasons for this may lie in the shaky foundations that Schettkat offers us. The interconnection between individual behavioral features and macroeconomic representations of the economy are not discussed in depth. An institutional description of the economic system, allowing to consider the temporal/sequential account of decisions and feedback effects is sorely lacking. If ever behavioral economics and Keynesian thoughts were to become interconnected and associated into a research program, probably this should be done on the basis of more recent developments in post-Keynesian economics, using agent-based modelling. This connection is still in its infancy (Fagiolo et al., 2007; Oeffner, 2008; Chen and Gostoli, 2014; Axtell and Farmer, 2022). Integrating behavioral features in agents’ specification raises important tractability issues (Cherrier, 2023). And there is no reason why behavioral economics results should attract only post-Keynesian economists (Dosi and Roventini, 2019). This lack of a methodological backbone is acutely felt. Qualitatively, the findings of behavioral economics provide sometimes robust knowledge about agent’s behaviors and reactions to their environment. But this does not deliver any conception of the properties, stability, or robustness of any macroeconomic outcome that may result from these findings once immersed in the macroeconomic bath. The difficulties associated with aggregation, composition effects, and changes in the perception of their environment by agents still remain. The author seems partially aware of this problem, considering that some behavioral results are immune to aggregation effect or composition effects. Most often, he simply denies the problem: “Relaxing the strict neoclassical assumptions and applying descriptive behavioral microfoundations will result in different macroeconomic outcomes as Keynes demonstrated in the General Theory” (22).
Akerlof, George and Robert Shiller. 2010. Animal Spirits. How Human Psychology Drives the Economy, and Why It Matters for Global Capitalism. Princeton: Princeton University Press.