Books to Read: James E. Hartley - The Representative Agent in Macroeconomics

You can find amazon reference here. Below a review produced by David Colander. This is book is a must read in my opinion even if it is quite onesided and essentially devastating in its critique of the use of representative agent models.

E Macroeconomics and Monetary Economics The Representative Agent in Macroeconomics. By James E. Hartley. Frontiers in Political Econ- omy, vol. 10. London and New York: Rout- ledge, 1997. Pp. x, 229. $69.95. ISBN 0-415- 14669-0. JEL 98-0076

James Hartley's book is interesting, worth a read by those economists who are unfamiliar with the ongoing debates with macro. His in- stincts are good; representative agent models are not especially enlightening and contain logical problems when applied to real-world issues. If those are not your instincts take a look at the book to question them; if those are your instincts take a look at the book to feel better about your beliefs. The book is less relevant to those who are involved in macro and are familiar with, say, Andrew Kir- man's work, with some of the recent work in post-Walrasian macroeconomics, or with the recent work in complexity. Although I liked the book, I was a bit con- fused by what the book is and what it isn't. Hartley explicitly states that "this book is not one of the ever-increasing number of 'defini- tive critiques of neoclassical economics' " but is instead "a solidly neoclassical" book.

He further states that the "the criticisms of cur- rent practice and the suggestions for future practice are criticisms and suggestions from within the family." I haven't seen an ever- increasing number of books that claim to be "the definitive critiques of neoclassical eco- nomics." In fact I seldom hear the term neo- classical economics any more-except from heterodox economists-and I'm not quite sure what it includes. From reading Hartley's book I would say that it is a strong criticism of practices of New Classical economics, and of those parts of New Keynesian economics that use representative agent models. I fully agree it is not definitive criticism, but a cri- tique it is.

The book consists of five parts and 14 chapters. Part I (chapters 1-3) provides an introduction in which he discusses the history of representative agent and what he sees as the arguments for its use. Parts II, III, and IV (chapters 4-13) provide the core of the book; they develop the three central reasons that he believes underlie the use of representative agent models. Part V (chapter 14) is a four- page chapter that discusses what will replace the representative agent model. While I liked the general thrust of the book, I had problems with various aspects of it. For me, perhaps because I am an historian of thought, the history was not illuminating. One problem was the jump between Mar- shall's concept of the representative firm to the representative agent of new classical models, as if they were the same thing. This does not fit my understanding of Marshall. Marshall used the representative firm as a method of talking about a dynamic equilib- rium of an industry. It was a contextual and limited use of the concept; nowhere does Marshall try to formalize his concept into a deductive tool independent of its context. It is true there were serious problems with Mar- shall's concept of representative agent, but the debate about the representative firm's usefulness is quite irrelevant to the debate about representative agent.

Another problem I have as an historian of thought is that there is little discussion of the controversy that surrounded the introduction of the representative agent model, and of how its introduction was fought by advocates of the Keynesian-monetarist tradition. Hartley ignores old Keynesians, a la Tobin; post- Keynesians, a la Davidson; and coordination macroeconomists, a la Clower, and even mainstream Keynesians, such as Solow, who have always opposed the concept. There is a story to be told about the introduction of the representative agent model; it is a story of intrigue, incentives in the profession, and some false paths created by neo-Keynesians claiming too much for their models. This book is not that story. The book is on stronger grounds when talk- ing about the more recent use of repre- sentative agent. This starts in chapter 3, where Hartley discusses his view of the New
Book Reviews 2173 Classicals' and the New Keynesians' justifica- tion for representative agent models.

He in- cludes three reasons: (1) their providing a means of avoiding the Lucas' critique, (2) their aid in the construction of Walrasian models, and (3) their providing a microfoun- dation of macro. Clearly these three justifications played a role in the introduction of these models, but I would not regard them as definitive. By that I mean that if someone pointed out to a repre- sentative New Classical that these three justi- fications are undermined, which I think this book convincingly does, he or she is not going to say: "I'm wrong; I will go out and give up my representative agent models." Instead, he or she will say "What else is new?" I say this because the main thrust of the book's argu- ments was well-known before, and nothing changed.

In my view, the justifications of the use of representative agent models are more complicated, and more tied to the sociology, institutional structure, and pedagogy of the profession, than to the logic of the models. Hartley does a nice job of undermining the three reasons he gives for using repre- sentative agent models, and herein lies the strength of the book. I would have empha- sized the interdependence problem much more than does Hartley, however. Hartley jumps between what might be called aggrega- tion bias problems-the problems of a single concept capturing a multiple -dimension real- ity that cause the aggregation concept to be biased-and the deeper methodological ag- gregation problem that undermines the entire attempt of looking at complex systems through a lens of microfoundations. He doesn't forcefully point out that in truly com- plex systems microfoundations, independent of their macrofoundations, are meaningless. Ultimately, it is the development of an al- ternative that will lead to the end of the rep- resentative agent model. Here the book's dis- cussion is simply too short. It was the belief that only an alternative would lead to change that has led my work to post-Walrasian macro- economics, complexity, and complex systems theory. Herein lies a different tradition, which I believe can provide a cogent alterna- tive, and which can redirect economics from deductive models to finding patterns in data. It would have been nice if Hartley had dis- cussed such issues.