On the Econometric Testing of Rationality-Market Efficiency

Abstract

Procedures for testing the rational expectations hypothesis deserve careful study because rationality of expectations has such important implications for macroeconomic modeling and policy analysis.' Rationality of expectations generally imposes cross-equation restrictions.2 In this paper we discuss the implementation and analyze the econometric properties of a particular test of such restrictions. This test has been used in recent empirical studies of bond market behavior.3 Since the test focuses on the distinction between anticipated and unanticipated movements in variables (as in Barro (1977, 1978)), it is applicable to many macroeconomic issues. Indeed, the results of this paper are useful in further work (Abel and Mishkin (1983)) that clarifies the relations among tests of (1) rationality and market efficiency, (2) the short-run neutrality of anticipated policy and (3) Granger (1969) causality in macroeconometric models. Two important questions about the test of cross-equation restrictions arise naturally. First, under what conditions will the test lead to correct inference about the rationality of expectations? Second, what is the relation of this cross-equation test to the more common singleequation test of market efficiency frequently used in the literature? The answers to these questions are provided by the theorem in section II which states the asymptotic equivalence of the cross-equation test with the more common test of market efficiency. Also in section II we discuss identification and demonstrate that we can test for rationality of expectations even if some parameters are not identified. Finally, we present an empirical example in section III and concluding remarks in section IV.