One aspect of contemporary macro micro-founded models is the use of a representative agent, whose interpretation is debated. Essentially, in these models a utility function is ascribed to a typical consumer or producer. It is not clear whether what is assumed is a single agent or a set of clones with identical preferences, but such modeling strategy helped its proponents getting away with heterogeneity, and enabled them to use a general equilibrium framework without bothering with the kind of aggregation issues emphasized by Sonnenschein, Mantel and Debreu.
James Hartley traced the use of representative agent back to Marshall’s use of a “representative firm” in his Principles and to Pigou in the early XXth century. Hartley’s work has subsequently focused on the resurrection of the representative agent by Robert Lucas in the late 1960s and 1970s, in order to provide micro foundations for his new-classical macroeconomics. The thrust of Lucas’s micro foundational program is nicely summarized by Kevin Hoover (section 5). Interestingly, Hoover contrasts Lucas’s representative agent modeling strategy with alternative ways to relate aggregates to individual behaviors, including Lawrence Klein’s attempt to preserve Keynes’ intuitions on micro foundations.
More interesting to me is another genealogy, one that allows the survey of the uses of the representative agent (RA) in microeconomics as well: growth theory à la Cass-Koopmans, or optimal taxation à la Diamond-Mirrlees. The common origins of these various models is Frank Ramsey. The Cambridge prodigy, who died at 27 in 1930, has contributed to mathematics, logic and philosophy, and his three economics papers – on probability and utility (1926), one the theory of optimal taxation (1927) and on the theory of savings and optimal growth (1928)- prefigured much of the models developed in the postwar era. His 1928 paper, in particular, made use of the calculus of variation in a dynamic intertemporal utility maximization setting to determine which amount of money should be saved to reach an optimal growth path. In both his growth and his optimal taxation papers, Ramsey used a single utility function:
“we denote by U=F(x1 …. xn) the net utility of producing and consuming (or saving) these quantities of commodities”
Whether can be interpreted as the application of a representative agent framework is open to debate.No, Marion Gaspard writes : “Ramsey does not try to justify the use of such macroeconomic functions, and a fortiori, he carefully avoids referring to any kind of aggregation procedure or representative agent‘ concept.” Pedro Duarte disagrees. He concedes that Ramsey is unclear as to whom the utility function belongs: a community of identical agents, a social planner? Loose social preferences? An average agent?
What matters in the end, Duarte argues, is that both Ramsey’s contemporaries and descendants interpreted his utility function as that of a “representative consumer” (Young’s words in 1929), “representative” (Pigou 1928) or “representative agent” (Phelps 1966; Diamond and Mirrlees 1971). What is unclear is how Ramsey’s modeling strategy travelled from 1920s Cambridge to 1960s Cambridge/MIT. His papers has sunk into oblivion for decades before being “rediscovered.” Whether there is a Cambridge-UK direct connection, or whether it was Samuelson, who had remained Ramsey’s sole promoter in the 1940s to 1960s, who introduced Diamond, Mirrlees, Atkinson, Stiglitz and Dixit to the “Ramsey rule” is yet unknown.