Who Faces Higher Prices? An Empirical Analysis Based on Japanese Homescan Data

• Author: Naohito Abe,Kyosuke Shiotani
• DOI: http://doi.org/10.1111/aepr.12050
• Date: 01 January 2014
• Published date: 01 January 2014

Who Faces Higher Prices? An Empirical

Analysis Based on Japanese Homescan Data

Naohito ABE1† and Kyosuke SHIOTANI2

1Hitotsubashi Univeristy and 2Bank of Japan

On the basis of household-level scanner data (called homescan data) for Japan, we construct a

household-level price index and investigate the causes of price differences across households. We

observe large price differentials across households, a result that is consistent with the previous

research based on the data in the USA. However, the differences across age and income groups are

small. In addition,we find that elderly people face higher prices than the younger ones, which is con-

trary to the results of the previous research. The most important determinant of the price level is the

extent to which households rely on bargain sales; doubling purchases of goods at bargain sales

decreases the price level by about 2%,while shopping frequency only has a limited effect on the price

level.

Key words: household consumption, price index, scanner data

JEL codes: D12, D30

1. Introduction

Owing to recent technological developments in data creation, numerous researchers of

commodity prices have begun to use not only traditional aggregates, such as the con-

sumer price index, but also information on micro-level commodity prices. To date,

commodity-level price information has been used in various economic fields, such as

macroeconomics (Nakamura & Steinsson, 2008), international economics (Haskel &

Wolf, 2001), and industrial economics (Baye et al., 2004). Recently, on the basis of

commodity-level homescan data, Aguiar and Hurst (2007) (hereafterAH) found a v iola-

tion of the law of one price across different age groups. More precisely, in the USA, com-

pared with younger families, elderly families face lower prices for the same commodities.

AH interpret their results in line with the standard life-cycle model of consumption with

endogenous decisions of shopping time. The mechanism of this model is straightfor-

ward. Since the opportunity costs of shopping for retired people are lower than those

for younger people, the elderly can spend more time for searching lower prices. AH’s

We are grateful to Andrew Leicester, Yukiko Abe, Sachiko Kuroda, Colin McKenzie, Tsutomu

Watanabe,Yukinobu Kitamura, Michio Suzuki, and participants at a seminar at Osaka University,

the Annual Meeting of the Japanese Economic Association, and the Seventeenth Asian Economic

Policy Review Conference for their helpful comments. This research is supported by the JSPS

Grants-in-Aid for Young Scientist (S) 21673001. The views expressed in this paper are those of the

authors and are not reflective of those of the Bank of Japan.

†Correspondence: Naohito Abe, The Institute of Economic Research, Hitotsubashi University,

Naka, Kunitachi,Tokyo, Japan. Email: nabe@ier.hit-u.ac.jp

bs_bs_banner

doi: 10.1111/aepr.12050 Asian Economic Policy Review (2014) 9, 94–115

© 2014 The Authors

Asian Economic Policy Review © 2014 Japan Center for Economic Research

94

findings have provided us with resolutions of several famous puzzles such as the retired-

saving puzzle and the excess sensitivity of expenditure to predicted income shocks.

Compared with standard consumption panel data, such as the Panel Studyof Income

and Dynamics, homescan data by AC Nielsen or Kantar provide us with detailed and fre-

quent information on purchases at the household-commodity level. However, most

homescan data do not update household characteristics such as income and employment

status regularly. In other words, there is no within variation in household data relating to

income and employment status in homescan data. For this reason, AH did not control

for household-level fixed effects; rather, they used age and income as instrumental vari-

ables (IV) to deal with the endogeneity in the determination of shopping frequency.

Because AH’s IV estimates are about 20 times larger than their non-IV estimates, a

careful examination of the effects of unobservable characteristics of households is

required.

This study considers the relationship between shopping behavior and the price level

using commodity-level homescan data for Japan. The advantage of using Japanese

homescan over those for the USA is that Japanese data updates household characteris-

tics every year, which enables us to obtain more robust estimates. It is also worth

noting that many previous researchers have confirmed the existence of the saving-

retirement puzzle and excess sensitivity in Japan.1Therefore, by investigating the

relationship between shopping behavior and the price level in Japan, we can check

whether the mechanisms proposed by AH play significant roles in economies outside

the USA.

As in the USA, we find that any particular commodity is traded at various different

prices in Japan. Figure 1 illustrates the distribution of the relative commodity price index

defined a la AH.2The index takes the value of unity if the recorded price is equal to the

regional average price. A value of 1.2 implies that the price is 20% larger than the

average.3Figure 1 clearly shows that the same products are sold at very different prices.

We also found that the price level increases with ages, which is in sharp contrast to AH’s

findings. We find the ratio of purchases at bargain sales is the most important factor

among several potentially important determinants of the price index. By doubling pur-

chases at bargain sales, households can enjoy a reduction in their price level of about 2%,

which is consistent with Griffith et al. (2009) who find a significant amount of savings

from purchasing at bargain sales in the United Kingdom. Other shopping behavior, such

as the frequency of shopping, the degree of mass purchasing, or a preference for high-

quality goods, are all statistically significant. However, this type of behavior is not quan-

titatively important.

Our empirical results suggest that the price reduction mechanism based on the

opportunity cost of shopping provided by AH is not observed in Japan. In contrast to the

USA, elderly people in Japan who are supposed to have lower opportunity costs for

shopping tend not to use bargain sales, which means they face higher prices than the

young. This suggests that further investigation into shopping strategies, particularly the

determinants of purchasing at bargain sales, is necessary to understand the mechanism

behind the price-level differential across families.

Naohito Abe and Kyosuke Shiotani Who Faces Higher Prices?

© 2014 The Authors

Asian Economic Policy Review © 2014 Japan Center for Economic Research 95