Explaining Disagreement on Interest Rates in a Taylor‐Rule Setting

• Published date: 01 October 2017
• Author: Michael J. Lamla,Lena Dräger
• Date: 01 October 2017
• DOI: http://doi.org/10.1111/sjoe.12217

©The editors of The Scandinavian Journal of Economics 2016.

Scand. J. of Economics 119(4), 987–1009, 2017

DOI: 10.1111/sjoe.12217

Explaining Disagreement on Interest Rates

in a Taylor-Rule Setting*

Lena Dr¨ager

University of Hamburg, DE-20146 Hamburg, Germany

lena.draeger@wiso.uni-hamburg.de

Michael J. Lamla†

University of Essex, Colchester CO4 3SQ, UK

mlamla@essex.ac.uk

Abstract

Most studies on disagreement focus on one specific variable, thereby neglecting the

fact that disagreement can be co-moving with disagreement on other variables. In this

paper, we explore to what extent disagreement regarding the interest rate is driven by

disagreement on inflation and unemployment. This relationship can be motivated by the

theoretical concept of the Taylor r ule. Using survey microdata for both professional

forecasters and consumers, we provide evidence that disagreement on the interest rate

is mainly driven by disagreement on inflation. We further show that disagreement is

significantly influenced by central bank transparency, as well as news on money and

credit conditions.

Keywords: Inflation expectations; interest-rate expectations; survey microdata; Taylor rule;

unemployment expectations

JEL classification:C33; D84; E31; E58

I. Introduction

Survey data of inflation forecasts from both households and professional

forecasters frequently show considerable disagreement, with time variation

in relation to macroeconomic conditions (Mankiw et al., 2004; Lahiri

and Sheng, 2008). This has important economic implications. If agents’

forecasts are dispersed, the underlying hypothesis of rational expectations

with full information is violated. This violation might be driven by agents

forming their expectations on the basis of different information sets, or by

interpreting signals differently. This can lead to non-converging beliefs in a

Bayesian learning model (Acemoglu et al., 2007), or to staggered beliefs in

†Also affiliated with ETH Zurich, KOF Swiss Economic Institute, Switzerland.

*We thank an anonymous referee for constructive comments. Furthermore, we thank Oli

Coibion, Ulrich Fritsche, Xuguang (Simon) Sheng, and the participants of the Workshop on

Uncertainty and Probabilistic Forecasting during the Financial and Economic Crisis at the

University of Heidelberg for their comments and suggestions. All remaining errors are ours.

988 Explaining disagreement on interest rates in a Taylor-rule setting

sticky information or rational inattention models (Woodford, 2001; Mankiw

and Reis, 2002; Sims, 2003), thus causing inertia in price dynamics.

Furthermore, if disagreement arises because signals from monetary policy

changes are misinterpreted, then excess volatility in asset markets might

arise, which highlights the need for a clear communication of policy

(Sims, 2009).

The sources of disagreement in survey expectations have, to our

knowledge, so far been analyzed almost exclusively for inflation

expectations, and mostly for disagreement among professional forecasters.

Furthermore, existing studies focus mainly on one measure of disagreement,

and they neglect the fact that disagreement on one economic variable might

be strongly related to the disagreement on a different economic variable. To

the best of our knowledge, only Banternghansa and McCracken (2009) and

Dovern (2014) investigate properties of multivariate disagreement based

on data from professional forecasters.

This paper contributes to the literature by exploring the potential

co-movement of disagreement measures in a more structural way. Under

the well-known theoretical concept of the Taylor rule, which is the most

prominent monetary policy rule used to describe and forecast interest-

rate movements of central banks, any disagreement regarding inflation

and/or economic growth should feed into the disagreement regarding the

expected interest rate. Consequently, in this paper we evaluate if – and

to what extent – disagreement on nominal interest rates can be explained

by disagreement on inflation and unemployment.

We employ a clear identification using cohor t-specific cross-sections

for our measures of disagreement, making use of the rotating panel

structure of the University of Michigan Surveys of Consumers, where

consumers are interviewed at least twice within six months. This allows

us to identify and track the cohorts of consumers who enter the rotating

panel together. We then construct an overlapping micropanel dataset, with

two disagreement observations per cohort. As a natural benchmark, we

also estimate disagreement in a Taylor-rule setting for the quar terly cross-

section of forecasters in the Survey of Professional Forecasters (SPF) in

the US.

In our estimations, we additionally account for a broad set of other

variables. We highlight the role of economic news observed by consumers,

and the role of changes in the communication policy of the Federal

Reserve System (hereafter the Fed). Furthermore, we control for specific

subsamples (e.g., for the initial period of high and volatile inflation in

the 1980s to mid-1990s), for the low inflation period from the mid-1990s

to mid-2000s as well as for the recent zero lower bound (ZLB) period.

Overall, we find evidence that disagreement regarding the future interest

rate is dominated by disagreement on future inflation. By contrast,

©The editors of The Scandinavian Journal of Economics 2016.