Identification in Macroeconomics

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

The research described in this proposal has three parts. The first concerns the fragility of identifying assumptions that are typically made to identify structural rational expectations models. The investigator focuses on new Keynesian models of the monetary transmission mechanism but the issues he raises apply more broadly to a large class of structural linear rational expectations models. Building on his critical survey of existing approaches to identification, in the second part of the proposal, he suggests a practical alternative to standard identification methods that relies on the use of natural experiments. Finally, he uses the input from estimated models, identified by their suggested approach, to analyze the structure of optimal policy rules. Broader Impact: Recent work in monetary economics seeks to identify characteristics of monetary policy rules that are optimal in the sense of minimizing a central bank objective function. In the presence of model uncertainty, John Taylor [1999] has argued that we should seek rules that work well in a broad class of theoretical environments. A good deal of the theoretical work in this area begins by specifying a representative agent maximizing model and deriving an implied set of linear functional equations that must be obeyed by a rational expectations equilibrium. One of the key results in this literature is the "Taylor principle" which asserts that monetary policy should be active in the sense that the policy maker is advised to raise the interest rate by more than 1% in response to a 1% increase in expected inflation. The Taylor principle, and similar rules of thumb, are judged to be "good" because in a large class of new Keynesian models he lead to a unique rational expectations equilibrium. In contrast, rules that do not obey the Taylor principle may be associated with indeterminacy and the possibility of equilibria that are driven by extraneous "sunspot" fluctuations. The Taylor principle has become a widely accepted rule-of-thumb for policy makers. The investigator's proposed research implies that confidence in this principle may be overstated and that policy advice based on simple structural models may rest very heavily on non-verifiable identifying assumptions. His work implies that existing researchers should test the implications of their identifying assumptions by checking implications of these assumptions for parameter constancy. Although part of their message is a criticism of existing approaches, the investigator is able to suggest workable alternatives that can act as an informed guide to future policy actions. His suggestion is to develop models that exploit past changes in regime to uncover identification assumptions that rely on natural experiments rather than on arbitrary assumptions that are difficult to justify based on primitive economic theories. This approach is widely applicable to a range of structural models and it can potentially provide a guide to policy makers that may enable the development of more robust rules of thumb that can act as guides to the conduct of monetary and fiscal policy.