reduced-form inflationpersistence should be measured. A natural measure of persistence is the impulse response function implied by the inflation process, which shows how quickly the effect of a shock to inflation vanishes. Following the previous literature, Fuhrer (2010) emphasizes persistence measures based on autocorrelation, which are indeed reasonable if inflation dynamics are well described by a conventional (causal) autoregressive (AR) process, as is typically assumed. However, if this not the case, there is, in general, no one-for-one correspondence between the
Studies in Nonlinear Dynamics &
Volume 13, Issue 4 2009 Article 1
Changes in U.S. InflationPersistence
Kyu Ho Kang∗ Chang-Jin Kim†
∗Washington University in St. Louis, email@example.com
†Korea University and University of Washington, firstname.lastname@example.org
‡Washington University in St. Louis, email@example.com
Changes in U.S. InflationPersistence∗
Kyu Ho Kang, Chang-Jin Kim, and James Morley
We investigate the existence and timing of changes in U.S. inflationpersistence. To do so, we
develop an unobserved components
autocorrelation properties of a stationary inflation rate, the picture is considerably murkier.” We revisit the changing nature of inflationpersistence in the US. We add to the literature on inflationpersistence in three ways. First, we use a quantile regression approach which allows us to examine the degree of inflationpersistence at different conditional quantiles of inflation. Thus far the literature focuses on persistence evaluated at the conditional mean. This neglects the fact that inflation following shocks drawn from the tails of the shock distribution might exhibit
Author Notes: This is a completely revised version of our previous working paper entitled
“Modeling Expectations with Noncausal Autoregressions.” We thank Michael McAleer, Efrem
Castelnuovo, Martin Ellison and Antti Ripatti for useful comments. Financial support from the
Academy of Finland and the OP-Pohjola Group Research Foundation is gratefully acknowledged.
This research was completed while the second author was a Fernand Braudel Fellow in the
Economics Department of the European University Institute.
volatility, persistence and predictability of inflation underwent a
turning point around 1995. During the last decade, volatility and persistence have been increasing
and predictability has been going down.
KEYWORDS: marginal likelihood, Bayesian model comparison, auxiliary particle filter,
stochastic volatility, great moderation, inflationpersistence
Author Notes: This paper was presented at the 3rd Italian Congress of Econometrics and
Empirical Economics, University of Ancona, January 30-31, 2009, and at the EEA-ESEM meeting
in Barcelona, August 23-27, 2009. The authors
which all models generate higher root mean square errors than those in other sample periods and output gap has the greatest mean and variability among all sample periods. Our research makes two contributions. First, we address the inflationpersistence phenomenon from a mathematical perspective. To the best of our knowledge, this is the first paper to introduce memory into the hybrid Phillips curve via the Caputo fractional derivative. Our inflation forecasting methodology avoids the estimation inaccuracy of the differencing parameter and an issue that multivariate
1 Introduction The dynamics of inflation, as defined by its moments, volatility, and persistence, affects the ability of central banks to control it. This paper focuses on inflationpersistence in the US using monthly Consumer Price Index (CPI) data from 1871:1 to 2016:1. The data span the modern history of the international monetary systems, including the classical gold standard era (1870–1914), the interwar period (1915–1944), the Bretton Woods system (1945–1971), and the post Bretton Woods system (1971–present), and thus provide a unique opportunity to
or inflationpersistence. Moreover, we observe multi-stability via a Chenciner bifurcation.
∗Marji Lines, Department of Statistics, University of Udine. Frank Westerhoff, Department of
Economics, University of Bamberg.
Many macroeconomic models restrict their attention to the behavior of a
representative agent which, when it comes to the formation of expectations, relies
on a single strategy to forecast macroeconomic variables. For some time this
strategy was given by an extrapolative, adaptive or regressive prediction rule but
. (2013) for a recent overview], cointegration techniques and structural breaks. Our study complements the literature by considering the consequences of changing inflationpersistence for testing the Fisher effect as a novelty. Importantly, we propose a simple OLS regression-based test for the Fisher hypothesis which is a long standing issue in monetary economics. In direct connection, the persistence properties of inflation and nominal (and real) interest rates are analyzed. An important early contribution is Barsky (1987) , who studies the effect of changing
not have a unit root and this can be the case of the US economy. When taking his argument into account, I focus on the possibility of abrupt changes in trend inflation as in Levin and Piger (2006) rather than stochastic trend which evolves as a driftless random walk as in Cogley, Primiceri and Sargent (2010) . Changes in the group of AR coefficients, related to inflationpersistence, can be understood through shifts in structural parameters in the New Keynesian Phillips curve. Shifts in inflationpersistence can be caused by changes in firms’ price setting