History-dependent monetary policy rules in the lab

How to cite

Isabelle, Salle; Truong, Hung. 2023. "History-dependent monetary policy rules in the lab." AEA RCT Registry. AEARCTR-0011128. https://doi.org/10.1257/rct.11128-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Salle Isabelle* , Hung Truong

* Primary investigator

Registry ID
AEARCTR-0011128
DOI
https://doi.org/10.1257/rct.11128-1.0
First registered
Last updated
Status
Completed
Trial dates
2022-05-18 – 2022-11-01
Countries
Canada (BC)
JEL codes
E7; E52; E42; C92
Registry record
http://www.socialscienceregistry.org/trials/11128

Abstract

Price-level targeting (PLT) is optimal under the theory of rational expectations (RE), i.e. when expectations are in line with the monetary policy rule, but lacks empirical support. Given the hurdles to the implementation of macroeconomic field experiments, we utilize a laboratory forecasting group experiment -- where expectations come from human subjects -- to collect data on expectations, inflation and output dynamics under the traditional inflation targeting (IT) framework and PLT confronted with both deflationary and cost-push shocks. This sequence of shocks is designed to approximate the recent course of events. We add a third treatment where PLT is augmented with central bank communication about the time-dependent inflation rate that would be consistent with closing the price gap in each period. Under RE, this additional piece of information is redundant because agents are supposed to form model-consistent expectations.

Intervention

We have three treatments: one inflation-targeting monetary policy rule, one price-level targeting monetary policy rule and one price-level targeting monetary policy rule augmented with communication about the time-varying inflation target that is consistent with closing the price gap. We have six groups of six subjects each.

Experimental design

The experiment belongs to the class of learning-to-forecast experiments: a standard three-equation New Keynesian model serves as the underlying data generating process. Its expectational variables are taken directly from the subjects (aggregated using the arithmetic mean) in each period and fed into the model which returns the actual values of these variables. This process is repeated for 75 periods.
The first 40 periods simulate a deflationary environment: after some small white noise shocks, a large deflationary shock is implemented in period 10 and lasts for three periods. The second submission of 35 periods features an inflationary environment with a similar cost-push shock.
The only difference between the three treatments is the monetary policy rule and the information set of the subjects. In all treatments, the subjects know the inflation target of the central bank. In the control treatment, the central bank implements a standard inflation targeting rule. We then have two treatments related to the alternative of price-level targeting: the first one only implements the monetary rule, without any further information; in the second one, we communicate to the subjects in each period the inflation rate that would be consistent with closing the price gap.

Design details

Randomization method
Subjects are randomly assigned to groups in one of the three treatments upon entering the lab.
Randomization unit
experimental sessions
Clusters
six sessions of six subjects for each of the three treatments
Observations
108 subjects
Arms
36 subjects per treatment
Primary outcomes
distance of inflation and output gaps and average output gap and inflation expectations to their targeted values; coordination between subjects' expectations, in the long and in the short runs
Secondary outcomes
number of periods where the effective lower bound of the monetary policy rule is triggered
Sponsors
Name: Simon Fraser University, CRABE lab Location: SFU, Burnaby, BC, Canada Url: http://www.sfu.ca/crabe.html

Cited by

No citation data available for this registration yet.

Keywords