Phillips Correlation and Price‐Change Distributions under Declining Trend Inflation

• Published date: 01 August 2023
• Author: Sohei Kaihatsu,Mitsuru Katagiri,Noriyuki Shiraki
• Date: 01 August 2023
• DOI: http://doi.org/10.1111/jmcb.12921

DOI: 10.1111/jmcb.12921

SOHEI KAIHATSU

MITSURU KATAGIRI

NORIYUKI SHIRAKI

Phillips Correlation and Price-Change Distributions

under Declining Trend Ination

The relationship between ination dynamics and micro-level pricing behav-

ior is an underexplored research area. In this article, we rst document that

deationary periods in Japan were associated with (i) variations in the dis-

tribution of price changes at the micro level, and (ii) a weakening in the

Phillips correlation at the macro level. We construct a multisector general

equilibrium model with sector-specic menu costs, a feature that is sup-

ported by our estimation using sectoral data, and show that declining trend

ination and an increasing share of services in privateconsumption together

can quantitatively explain the above two observations.

JEL codes: E31, E32, E52

Keywords: Phillips correlation, price-change distribution, menu cost,

trend ination

T     is to deepen understanding of

the relationship between rms’ price-setting behavior at the micro level and ination

dynamics at the macro level by analyzing both within a unied model. In particular,

we focus on the following two stylized facts in Japan, which have been observed

along with declining trend ination since the 1980s: (i) variations in the distribution

of price changes at the micro level, and (ii) a weakening in the Phillips correlation,

that is, the relation between ination and macroeconomic demand-supply balance.

Using quantitative analysis, we attempt to provide a consistent explanation for these

This research project started when the authors belonged to the Monetary Affairs Department at the

Bank of Japan. We would like to thank two anonymous referees and the staff of the Bank of Japan for

their helpful comments. We also would like to thank Rie Yamaoka and Yuko Ishibashi for resourceful

research assistance. The opinions expressed here, as well as any remaining errors, are those of the authors

and should not be ascribed to the Bank of Japan, the International Monetary Fund, or CyberAgent, Inc.

This work was supported by MEXT KAKENHI Grant Number 21H00704.

S K is at Strategy, Policy and Review Department, International Monetary Fund.

M Kis at Faculty of Business Administration, Hosei University (Email: mitsuru.katagiri@

hosei.ac.jp). N S is at CyberAgent, Inc.

Received December 31, 2018; and accepted in revised form December 21, 2021.

Journal of Money, Credit and Banking, Vol. 55, No. 5 (August 2023)

© 2022 The Ohio State University.

1272 :MONEY,CREDIT AND BANKING

two stylized facts to understand how the declining trend ination affects price-setting

behavior at the micro level and, in turn, how the changes in price-setting behavior

affect the ination dynamics at the macro level.

As a rst step in understanding the relationship between the micro- and macro-

price dynamics under declining trend ination, the following thought-provoking styl-

ized facts are provided by the Japanese experience during periods of high ination

(1982−1994) and low ination (1995−2012). First, at the micro level, the cross-

sectional distributions of price changes in different sectors have shown remarkably

different changes in their shapes. Specically,such distributions in the services sector

have been weighted more heavily around 0% and their dispersion has signicantly

narrowed over time, whereas those in the goods sector have only shifted leftwardand

have not shown any signicant changes in their shape. Second, at the macro level,the

Phillips correlation has signicantly weakened. In contrast to the sectoral differences

in price-setting behavior at the micro level, the Philips correlation has weakened not

only in the services sector but also in the goods sector.

We focus on the menu-cost hypothesis as a keymechanism to consistently account

for the stylized facts and examine its validityby carrying out both an empirical investi-

gation and a quantitative exerciseusing a theoretical model. We establish an empirical

model with price rigidities in the vicinity of 0%-price change to test the menu-cost

hypothesis in Japan. Specically, we employ a panel data analysis using a limited

dependent variable model with two-sided thresholds, as in Honoré, Kaufmann, and

Lein (2012), and estimate parameters using individual item data from Japan’s cost

price index (CPI). In line with other empirical studies of the United States or the euro

area, our empirical study identies the existence of statistically signicant menu costs

and indicates higher menu costs in the services sector than in the goods sector. Next,

we construct a multisector dynamic general equilibrium model with heterogeneous

rms, where the services sector bears higher menu costs than does the goods sector.

A quantitative analysis using comparative statics shows that declining trend ination

and a rising share of services in consumption, both of which have been observed in

Japan, together can quantitatively explain the stylized facts at the micro and macro

levels in Japan within a unied model.

Intuitively, the decline in trend ination weakens the Phillips correlation because

rms are reluctant to pay menu costs to make price changes, thus, reducing price

responsiveness to macroeconomic uctuations. Because menu costs are higher in the

services sector, the rise in the share of services in private consumption strengthens

this mechanism (and weakens the Phillips correlation). Interestingly, the decline in

price responsiveness in the services sector propagates to the goods sector through the

general equilibrium effect, consequently weakening the Phillips correlation in that

sector as well, as has been observed in Japan. In this sense, the quantitative results

cast some doubt on the macroeconomic implications obtained by empirical analyses

based only on goods prices such as those using retail scanner datasets.

Price-setting behavior at the micro level has become a prominent issue recently

both in research and practice, particularly amid the recent prolonged deationary

pressure experienced in many developed countries. Several studies using micro data

KAIHATSU,KATAGIRI,AND SHIRAKI :1273

for pricing behavior provide interesting insights into price-setting behavior,including

Bils and Klenow (2004), Klenow and Kryvtsov (2008), and Nakamura and Steinsson

(2008, 2013) for the United States, Higo and Saita (2007) and Kurachi, Hiraki, and

Nishioka (2016) for Japan, and Dhyne et al. (2006) for the euro area. However, anal-

ysis of the macroeconomic implications of price-setting behavior remains an under-

explored research area.

The relationship between the Phillips correlation and menu costs is a classical topic

within the monetary economics literature. Ball, Mankiw, and Romer (1988) discuss

this issue and argue that, based on the menu-cost hypothesis, a decline in trend ina-

tion increases price rigidities and weakens the Phillips correlation. Recently, several

studies have shed new light on this topic using a general equilibrium model with

a menu-cost element (e.g., Golosov and Lucas 2007; Midrigan 2011; Vavra 2014;

Kehoe and Midrigan 2015; Watanabe and Watanabe 2017). The most closely re-

lated studies to the current articles are Enomoto (2007) and Nakamura and Steins-

son (2010), both of which examine price-setting behavior and monetary neutrality

using a multisector general equilibrium model with menu costs.1The current article

advances this line of research by assessing whether a multisector general equilibrium

model with menu costs can explain the observed shift in price-change distributions

by sector as well as the weakening in the Phillips correlation observed during the

deationary period in Japan.

The remainder of the article is organized as follows. Section 2 presents empirical

ndings on shifting price-change distributions and the changes in the Phillips cor-

relation in Japan. Section 3 describes an empirical procedure to test the menu-cost

hypothesis. In Section 4, we construct a multisector general equilibrium model with

the heterogeneous menu-cost structure across sectors. Section 5 provides quantitative

analyses and Section 6 presents our conclusion.

1. PRICE-CHANGE DISTRIBUTIONS AND THE PHILLIPS CORRELATIONIN

JAPAN

In this section, we examine price-change distributions, the frequency of price ad-

justments, and the Phillips correlation in Japan since the 1980s. The Japanese econ-

omy experienced two notable changes during this period. First, trend ination has

declined signicantly. Panel (1) of Figure 1 shows that the median price changes for

all goods and services before the mid-1990s were positive and around 2%, but they

have subsequently been close to 0%. Second, the share of the services sector in pri-

vate consumption has increased by around 10 percentage points since the 1980s, as

1. In terms of methodology and motivation, Enomoto (2007) and Nakamura and Steinsson (2010) use

a classic monetary model with money supply shocks and focus on monetary neutrality. Conversely, in the

current article, we use a model with an interest rate feedback rule by central banks and focus on the Phillips

correlation over the business cycle, which is drivenby demand shocks in line with the recent quantitative

monetary economics literature.