New Evidence on the Ability of Asset Prices and Real Economic Activity Forecast Errors to Predict Inflation Forecast Errors
Date | 01 August 2017 |
Published date | 01 August 2017 |
Author | Nicholas Apergis |
DOI | http://doi.org/10.1002/for.2453 |
New Evidence on the Ability of Asset Prices and Real Economic
Activity Forecast Errors to Predict Inflation Forecast Errors
NICHOLAS APERGIS
1
*
1
Department of Banking and Financial Management, University of Piraeus, Greece
ABSTRACT
This paper investigates the impact of both asset and macroeconomic forecast errors on inflation forecast errors in the
USA by making use of a two-regime model. The findings document a significant contribution of both types of forecast
errors to the explanation of inflation forecast errors, with the pass-through being stronger when these errors move
within the high-volatility regime. Copyright © 2016 John Wiley & Sons, Ltd.
key words inflation forecast errors; asset and macroeconomic forecast errors; two-regime model; USA
INTRODUCTION
Reliable forecasts of inflation are substantially important for policymakers who implement both monetary and fiscal
policies, for investors who hedge the risk of their investments, as well as for firms that attempt to effectively invest
and set prices (Lucas, 1972; Sargent & Wallace, 1975). More specifically, the understanding of inflation forecast er-
rors is called for to investigate whether there is room for improvement in inflation forecasting. The primary research
goal of this paper is to explore the role of both asset prices and real economic activity forecast errors with relevance to
predicting US inflation forecast errors. The empirical analysis makes direct use of inflation forecast errors coming
from the Professional Forecasters Survey (PFS). Thomas (1999) provides supportive evidence that relevant surveys
outperform simple time series benchmarks modeling for forecasting inflation. In addition, it is assumed that the link
between the two groups of forecast errors is regime-dependent, with the pass-through being time-varying. The pri-
mary novelty of the study is that it is considered the first, to the best of the author’s knowledge, to explore the asso-
ciation between PFS inflation forecast errors and those coming from both asset prices and real economic activity. It is
also worth noting here that the empirical analysis is pursued through the link between inflation and asset errors and
not through the direct link that associates the values of the underlying variables. The primary reason is twofold: first,
the analysis considers the deviations (errors) between actual values and professional forecasting values; second, the
errors match the shocks inflation receives. The literature considers both the microeconomic and macroeconomic con-
sequences of the link between errors than actual values. In particular, at the microeconomic level, inflation forecast
errors stem from the inefficiency of decisions made by agents whose perception of future relative prices is not correct
(Lucas, 1972; Sargent & Wallace, 1975), while on a macroeconomic level the presence of such errors stems from the
deviations of actual monetary policy decisions from a forward monetary (Taylor) policy rule recommended by
Clarida et al. (1998, 2000).
The analysis is related to the strand of the literature that forecasts inflation from asset prices. Jorion and Mishkin
(1991) find that inflation forecasts can be highly consistent only if they are derived from short-term interest rates in
the case of 10 OECD countries. Goodhart and Hofmann (2000) document that stock returns do not exhibit any pre-
dictive content for inflation in the case of 12 developed economies, while Stock and Watson (2003) argue that asset
prices, as forward-looking assets, can accurately predict inflation movements.
In a different strand, the literature has extensively explored the role of exchange rates as a potential channel
through which inflation can be predicted, especially in open economies. In the USA, exchange rates have long en-
tered conventional Phillips curves (Goodhart & Hofmann, 2000), while they provide supportive evidence that hous-
ing prices is an important driver characterized by significant predictive content for inflation rates.
In the strand of the literature that provides evidence for a strong link between inflation forecast errors and real eco-
nomic activity forecast errors, the standard approach has been the Phillips curve (PC) model, which assumes the pres-
ence of a trade-off between unexpected inflation and certain indicators of real economic activity, such as
unemployment rates. Stock and Watson (1999) illustrate that such models have better forecasting performance using
leading indicators of economic activity, while Atkenson and Ohanian (2001) report that these models cannot provide
stronger, more reliable forecasts than those provided by the naive model. Such conflicting findings on inflation
*Correspondence to: Nicholas Apergis, Department of Banking and Financial Management, University of Piraeus, 80 Karaoli & Dimitriou St,
Piraeus, 18534 Greece.
E-mail: napergis@unipi.gr
Journal of Forecasting,J. Forecast. 36, 557–565 (2017)
Published online 16 November 2016 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/for.2453
Copyright © 2016 John Wiley & Sons, Ltd.
To continue reading
Request your trial