Can Cold Turkey Reduce Inflation Inertia? Evidence on Disinflation and Level‐k Thinking from a Laboratory Experiment

• Published date: 01 December 2022
• Author: MARCUS GIAMATTEI
• Date: 01 December 2022
• DOI: http://doi.org/10.1111/jmcb.12904

DOI: 10.1111/jmcb.12904

MARCUS GIAMATTEI

Can Cold Turkey Reduce Ination Inertia? Evidence

on Disination and Level-k Thinking from a

Laboratory Experiment

It is widely believed that ination inertia varies with the policy pursued.

In a novel experiment, price setters determine ination rates and react to a

central bank’s indicator, which is implemented exogenously either as cold

turkey or gradual disination. In a third treatment, subjects in the role of a

central banker set the indicator endogenously, potentially reducing inertia

by signaling to be a tough central banker. I nd inertia to be structurally

stable and invariant to policies. The data can be organized by a model of

level-kthinking,which shows that cold turkey improves only a few subjects’

adjustment while leaving many behind.

JEL codes C72, C92, E31, E52, E71

Keywords: cold turkey, disination, gradualism, inationinertia, level-k

The persistence of ination is a well-documented phe-

nomenon in the macroeconomic literature. While ination inertia is widely evi-

denced, some authors argue that ination inertia can be reduced because it may vary

with the policy pursued (King 1996, Erceg and Levin 2003, Westelius 2005). This

line of causation is also central to Lucas (1976). Lucas motivates his critique using

Parts of the experiments were nanced by the Chair of Economic Theory, University of Passau. The

author declares that he has no relevant or material nancial interests that relate to the research described in

this paper.The author is especially grateful to Johann Graf Lambsdorff, who always and with all his enthu-

siasm believed in the project and supported it. The author thanks Rosemarie Nagel, Susanna Grundmann,

Manuel Schubert, Katharina Werner, Stefan Grundner, Isabelle Riviere-Bowen, and Ann-Kathrin Crede

for their support and a lot of suggestions. Thanks also go to the participants of the Behavioral Macroeco-

nomics Workshop at the University of Bamberg in June 2018, of the research seminar of the University

of Passau and of the Workshop on Theoretical and Experimental Macroeconomics at the 2015 Barcelona

GSE Summer Forum, especially to John Duffy and Frank Heinemann. And I want to thank the referees

for providing comments that helped to improve the paper.

M G is a Professorat Bard College Berlin. External fellow at the University of Passau,

Germany, and at the University of Nottingham (CeDEx) (E-mail: marcus.giamattei@uni-passau.de).

Received July 31, 2018; and accepted in revised form September 1, 2021.

Journal of Money, Credit and Banking, Vol. 54, No. 8 (December 2022)

© 2021 The Authors. Journal of Money, Credit and Banking published by Wiley Periodicals

LLC on behalf of Ohio State University.

This is an open access article under the terms of the Creative Commons Attribution License,

which permits use, distribution and reproduction in any medium, provided the original work

is properly cited.

2478 :MONEY,CREDIT AND BANKING

rational expectations where there is no place for inertia in response to anticipated

shocks. He points to the more general problem that subjects’ behavior is not policy

invariant and changes with the policy under consideration. Subjects may not pay at-

tention to minor changes in policies. But major changes would induce large costs to

those who fail to adjust and consequently increase attention (Akerlof et al. 2000),

this way overcoming inertia. Such a major policy shift could reduce the number of

inertial subjects and accelerate adjustment (Schaling and Hoeberichts 2010, Cogley,

Matthes, and Sbordone 2015). It may also provide a clear signal that could dominate

idiosyncratic noise or strategic uncertainty about others and thereby reduce inertia

(Angeletos and Lian 2016).

Whether inertia varies with policies is hard to study with observational data, be-

cause policies are also chosen endogenously in response to given inertia. Therefore,

I implemented a novel experimental design to allow for clear causal inferences. In

my experiment, I focus on the role of cognitive limitations and nonstandard decision-

making causing ination inertia. Totest my hypothesis, I take two classical policies to

ght ination. Central banks can do this by increasing the key interest rates. But dis-

ination can follow two very different approaches—cold turkey (CT) or gradualism.

Starting with Sargent (1982), CT implies the enforcement of a clear and sudden policy

change. Rational subjects immediately adjust to such a policy change and disination

will occur at no cost. Gordon (1982, 11) challenges the view that CT disination is

similar to a headache remedy that brings “an instant cure with no side effects.” He

favors implementing disination gradually to account for the idea that people may

not be fully rational such that “ination has a stubborn, self-sustaining momentum,

not susceptible to cure” (Hall 1982, 3).

In the experiment, four price setters set individual ination rates and react to an

indicator set by the central bank. In two treatments, I vary the indicator exogenously.

My design assumes that price setters’ decisions are strategic complements, and op-

timal individual levels of ination depend on the average price setter ination rates

and the indicator of the central bank. Price setters maximize their payoff by choos-

ing prices that are close to the equilibrium ination rate and at the same time not too

far away from the ination rates that others set. Price setters have full information

about the economy and the indicator and can exibly adjust their ination rates. Still,

I observe high levels of inertia. This departure from rationality can be modeled in

the form of level-kthinking. The concept of level-kthinking includes the cognitive

failure to adjust rationally as well as the (rational) response to expectations that other

price setters might fail to do so (Lambsdorff, Schubert, and Giamattei 2013). I hy-

pothesize that CT reduces inertia and improves level-kthinking. Nevertheless, inthe

experiment only, some subjects show improved level-kthinking, while a large part is

left behind and appear to be overburdened by the strong policy shift. I observe that

the overall inertia is not reduced in the CT treatment.

If policy is not implemented as an automated rule as in the rst two treatments

but central bankers can exibly decide on the indicator, inertia might, in addition, be

responsive to the central banker’s type. CT might be effective in a way that is not

captured by the rst two treatments. Price setters may be unsure about which type

MARCUS GIAMATTEI :2479

of central banker they are confronted with. A “weak” central banker avoids costly

adjustments. A “tough” central banker is willing to incur large adjustment costs and

implements CT. Therefore, CT may serve as signal of being a “tough” central banker

and to differentiate from a “weak” central banker. Once price-setters identify a tough

central banker, they might anticipate strict future policies and adjust more quickly. I

test this proposition by implementing a third endogenous treatment where the role of

the central banker and actively embodied by subjects who decide on the disination

strategy. Yet, even in this treatment, I do not observe that CT reduces inertia. Over-

all, my ndings suggest that inertia is persistent and not easily affected by policies,

contrary to the idea by Lucas (1976).

1. LITERATURE

It is well known that the standard New Keynesian model does a poor job in ex-

plaining ination inertia (Buiter and Grafe 2001, Mankiw 2001, Fuhrer 2009). Sev-

eral modications to models with rational expectations have been proposed such as

mechanical indexation (Yun1996, Christiano, Eichenbaum, and Evans 2005, Ascari

and Ropele 2012), real wage rigidities (Blanchard and Galí 2007, Ascari and Merkl

2009), staggered pricing policies (Calvo, Celasun, and Kumhof 2007), sticky infor-

mation (Mankiw and Reis 2002, Agliari et al. 2017, Branch and Evans 2017), rational

inattention (Zhang 2017), or habits (Collard, Fève, and Matheron 2007). But, as ar-

gued by Nimark (2008), these emerge as ad-hoc xes, aimed at identifying features

that might align theory with evidence rather than pushing toward a more general

theory.

Another line of literature focuses on bounded rationality to explain ination inertia

(Roberts 1997, Ball 2000, Steinsson 2003, Adam 2007, Ormeño and Molnár 2015).

Failures in expectation formation may be one type of bounded rationality. As evi-

denced by survey data, subjects only form slow-movingination expectations (Adam

and Padula 2011, Fuhrer 2017). Subjects may be adaptive learners (Orphanides and

Williams 2005, Milani 2007, Al-Eyd and Karasulu 2008, Evans and Honkapohja

2009, Kurz, Piccillo, and Wu 2013, Hachem and Wu 2017) and have to learn the

parameters of the model.

Failures in decision making are another type of bounded rationality. Subjects may

fail to optimize even if they haveall information due to cognitive constraints, missing

economic literacy (Burke and Manz 2014), limited capacities for monitoring current

conditions (Woodford2003a) or to save cognitive costs (Gabaix et al. 2006, Magnani,

Gorry, and Oprea 2016). Such individual decision failures may translate into aggre-

gate inertia. This may arise because ination decisions are strategic complements

(Ball and Romer 1991, Fehr and Tyran 2005). By introducing a few nonrational sub-

jects together with strategic complementarity, inationrates may become inertial due

to the bounded rationality of some subjects and the high-order beliefs of others. With

complementarity, rational subjects have an incentive to imitate nonrational behavior