A Feasible and Objective Concept of Optimality: the quadratic loss function and U.S. monetary policy in the 1960s∗

The introduction of linear-quadratic methods in monetary economics in the 1960s tinged the intense debate about the optimal monetary policy instrument. These methods were widely used outside monetary economics because they delivered easy solutions to complex stochastic models. This same reason explains the success of quadratic loss functions according to the conventional wisdom among monetary economists. In this traditional narrative, Henri Theil and Herbert Simon are often cited by their proofs that models with quadratic objective functions have the certainty equivalence property. This attribute made the solution of these models feasible for the computers available at that time. This paper shows how the use of a quadratic loss function to characterize the behavior of central banks inaugurated an objective or uniform way of talking about optimality. In this respect, the discourse on optimal monetary policy stabilized. Moreover, a richer account of the quadratic approach to monetary policy debate emerges by analyzing how quadratic loss functions were used in operations research and management problems by groups of scientists that included economists like Modigliani and Simon. I argue that feasibility is only one important factor that explains the wide popularity of quadratic functions in monetary economics.