Is the NAIRU more useful in forecasting inflation than the natural rate of unemployment?

Abstract

Recent studies have indicated that the terms "NAIRU" (non-accelerating inflation rate of unemployment) and "natural rate of unemployment" are not interchangeable. While NAIRU is an empirical macroeconomic relationship estimated via a Phillips curve, the natural rate is an equilibrium condition in the labor market, reflecting the market's microeconomic features. This paper evaluates comparatively the inflation-forecasting power of alternative time-varying estimates of the natural rate of unemployment relative to the NAIRU. I estimate the natural rate of unemployment in the U.S. since World War II. Three alternative methods are utilized: the Kalman filter, a structural determinants approach, and the Hodrick-Prescott filter. In the section that follows, I assess how each estimator of the natural rate compares to the others-as well as to the NAIRU derived from a Phillips curve-in forecasting inflationary changes in the United States in the second half of the twentieth century. The analysis reveals that the overall inflation-forecasting utility of the natural rate of unemployment relative to the NAIRU is not very different. Moreover, the conclusion appears to be quite robust to various estimators of the natural rate.2

Underspecification of the Empirical Return-Factor Model and a Factor Analytic Augmentation as a Solution to Factor Omission

“…A total of six sectors exhibit serial correlation at either lower or higher orders. The conclusion is that respecification with 𝑀 𝜀 𝑡 included in the macro-residual model sufficiently reduces impure serial correlation (Adams & Coe, 1990;Claar, 2006).…”

“…To test for serial correlation in the residuals, Q-statistics are estimated for first order serial correlation (Q(1)), and for the first five serial correlation coefficients (Q(5)). A better specified model will generally be free of impure residual serial correlation and any remaining significant residual serial correlation will be pure in nature, unless the functional form is incorrect (Adams & Coe, 1990;Claar, 2006). 4 Q-statistics are estimated for squared residuals for the first serial correlation coefficient, Q 2 (1), and for five orders of serial correlation, Q 2 (5) to test for non-linear dependence in the residuals.…”

Review of Kalman Filter Employment in the NAIRU Estimation

“…This model of the NAIRU evolution is generally considered appropriate for the United States [18,20,31]. On the other hand, some authors [15,20] prefer a random walk with drift which is more suitable for European countries.…”

“…Data on the unemployment rate, inflation rate (expressed by CPI or GDP deflator) and GDP in case a potential product is also estimated, and data on other explanatory variables included in a model are used as an input for the Kalman filter. Data are mostly available as quarterly time series, annual data is rarely used (by [16,18,29,33,34]).…”

Studying the Nairu and its Implications

“…For example, Stock and Watchon (1999) have concluded that inflation estimations with respect to Phillips curve are more accurate relative to others, while in Stock and Watchon (1996) signify the presence of, at least, sixty nine alternative indicators that can be used for inflation prediction and provide different information and thus inflation predictions. Further, it has been found that the use of unemployment as an indicator of inflation predictions is characterized about its impropriety, while the alternative ways of measuring it reduce the degree of uncertainty in forecasts (Gordon, 1988;Claar, 2006). The absence of any proper inflation measure is verified by the fact that in Phillips curve literature gross domestic product is widely used as an indicator of "core inflation" by excluding prices of foods and energy goods (Staiger et al, 1997a).…”