Evolving PostWorld War II U.S. Inflation Dynamics

Abstract

Greenspan the natural-rate hypothesis, which they eventually acted upon to reduce inflation. Another mechanism was posited by Parkin (1993) and Ireland (1999), who argued that the inflation-unemployment dynamics are driven by exogenous drift in the natural rate of unemployment, for example due to demographic changes. Because the time-consistent inflation rate varies directly with the natural rate of unemployment, Parkin and Ireland attributed the drift in the inflation rate to drift in the natural rate of unemployment. The DeLong-Taylor-Sargent story makes contact with various elements in Lucas's (1976) critique. It makes the drift in inflation-unemployment dynamics a consequence of the monetary authority's evolving views about the economy. The story attributes alterations in the law of motion for inflation and unemployment to the changing behavior of the monetary authority, which emerges in turn from its changing beliefs. This story is consistent with one way that Lucas (1976) has been read, namely, as an invitation to impute observed drift in coefficients of econometric models to time-series variation in government policy functions. Sargent's (1999) version of the story focuses on how the coefficient drift over time affected the results of time-series tests of the natural-rate hypothesis. In the late 1960s, Robert Solow and James Tobin proposed a test of the natural-rate hypothesis. Using data through the late 1960s, that test rejected the natural-rate hypothesis in favor of a permanent trade-off between inflation and unemployment. Lucas (1972) and Sargent (1971) criticized that test for not properly stating the implications of the natural-rate hypothesis under rational expectations. In particular, the Solow-Tobin test was correct only if inflation exhibited a unit root. Before the 1970s, postwar U.S. inflation data did not exhibit a unit root, rendering invalid (in the opinion of Lucas and Sargent) Solow's and Tobin's interpretation of their test. However, in the 1970s, just when U.S. inflation seems to have acquired a unit root, the Solow-Tobin test began accepting the natural-rate hypothesis. Building on Sims (1988) and Chung (1990), Sargent (1999) constructs an adaptive model of the government's learning and policymaking that centers on the process by which the government learns an imperfect version of the natural-rate hypothesis, cast in terms of Solow and Tobin's representation. Parts of Sargent's adaptive story acquire credibility when it is noted how the Solow-Tobin characterization of the natural-rate hypothesis has endured, despite the criticism of Lucas and Sargent. As Hall (1999) and Taylor (1998) lament, that faulty characterization continues to be widely used. For example, see Rudebusch and Svensson (1999) for a widely Evolving Post-World War II U.S. Inflation Dynamics * 333 cited model that represents the natural-rate hypothesis in the Solow-Tobin form. Fisher and Seater (1993), King and Watson (1994, 1997), Fair (1996), Eisner (1997), and Ahmed and Rogers (1998) construct tests of long-run neutrality that are predicated on the assumption of a unit root in inflation.1 Estrella and Mishkin (1999) use the Solow-Tobin characterization to estimate the natural rate of unemployment. In the discussion following the paper by Estrella and Mishkin, John Williams confesses that the Federal Reserve Board's large-scale macroeconometric model also incorporates this characterization. Hall questions its validity for U.S. data after 1979 and sharply criticizes its continued use.