RATIONAL CONSUMERS*

• DOI: http://doi.org/10.1111/iere.12154
• Author: Kohei Kubota,Mototsugu Fukushige
• Date: 01 February 2016
• Published date: 01 February 2016

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 1, February 2016

RATIONAL CONSUMERS∗

BYKOHEI KUBOTA AND MOTOTSUGU FUKUSHIGE 1

Yamagata University,Japan; Osaka University, Japan

The life cycle/permanent income hypothesis (LCPIH) entails two postulates: People have rational expectations and

people do not have problems with self-control. If either or both of these postulates do not apply, we cannot obtain a

testable implication for the LCPIH. We use Japanese representative panel data that include responses to self-reported

and retrospective questions in order to elicit behavior such as forward-looking and self-control problems. We test

the rational expectations hypothesis and the LCPIH implication and find that rational consumers do not change their

expenditure in response to expected income changes, which we restrict to fit the two LCPIH postulates.

1. INTRODUCTION

Understanding the mechanisms of consumer behavior is fundamental to evaluating the

macroeconomic impact of public policy. The life cycle/permanent income hypothesis (LCPIH)

is a basic model for describing consumption behavior over time, with many studies conducting

empirical analyses to test the LCPIH and its implications. For example, some studies have

examined the relation between predicted income changes and consumption growth and have

found that if the LCPIH holds, the predicted income changes do not help explain consump-

tion growth, the procedure for which is the excess sensitivity test.2However, despite the many

empirical studies testing the LCPIH using this method, no consensus has been reached on the

LCPIH (Browning and Lusardi, 1996; Attanasio, 1999; Attanasio and Weber, 2010; Jappelli

and Pistaferri, 2010).3

Our article focuses on the implicit and defined assumptions of the standard LCPIH.4The

standard LCPIH assumes that households construct consumption plans based on unbiased

predictions of their future income profile and that they can execute their plans. In other words,

we focus on two postulates of the LCPIH: (1) households have rational expectations and (2)

∗Manuscript received September 2009; revised August 2014.

1We thank referees for helpful comments that substantially improved the manuscript. We also thank Charles Yuji

Horioka, Hidehiko Ichimura, Miki Kohara, Fumio Ohtake, Masao Ogaki, Takashi Unayama, Kazuo Yamaguchi and

participants at 10th Macro Conference and 8th Panel Research Conference for helpful comments and suggestions. We

acknowledge financial support from the Osaka University Global COE program, “Human Behavior and Socioeconomic

Dynamics.” Kubota also acknowledge a grant obtained by the Nihon University Population Research Institute from the

“Academic Frontier” Project for Private Universities matching fund subsidy 2006–2010 from MEXT, and Grant-in-Aid

for JSPS Fellows Grant Number 20·56451. Please address correspondence to: Kohei Kubota, Faculty of Education,

Art and Science, Yamagata University, 4-12, Kojirakawa-machi, 1-chome, Yamagata City 990-8560, Japan. E-mail:

kubota@e.yamagata-u.ac.jp.

2Flavin (1981) originally proposed this test, known as the excess sensitivity test. Flavin (1981) modified Hall’s

(1978) model by assuming that income follows an autoregressive moving average (ARMA) process and found excess

sensitivity, meaning that the response of a change in consumption to a change in income was greater than that implied

by the LCPIH.

3In rejecting the LCPIH, several alternative hypotheses have been duly tested, including liquidity constraints,

precautionary motives, the inseparability of consumption and leisure, buffer-stock saving behavior, durability of goods,

and habit formation.

4In this article, we refer to the standard consumption model as the model proposed by Hall (1978). Hall’s (1978)

assumptions include that the utility function is additively separable over time and the quadratic (which makes the

instantaneous utility function concave), that the interest rate is equivalent to the subjective discount rate, and the

presence of a perfect capital market. Hall (1978) also introduced the uncertainty of future income.

231

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

232 KUBOTA AND FUKUSHIGE

households do not exhibit self-control problems. If either or both of these assumptions do not

apply, we cannot obtain a testable implication of the LCPIH. Moreover, if these assumptions are

invalid, the economic model will also not explain actual behavior. Consequently, to construct a

more adequate model, we must know which particular assumptions are reasonable.

Standard neoclassical models exploring household dynamic behavior implicitly and definitely

assume the rational expectations hypothesis (REH) as a fundamental premise, duly tested by

a large number of empirical studies.5Further, standard models also assume that people do not

exhibit self-control problems. In other words, under the LCPIH, consumers can execute their

ex ante optimal consumption plans.6Therefore, if the assumption of having no self-control

problems is not valid, the implications of the LCPIH could be rejected (Angeletos et al., 2001).

The purpose of this article is then to examine whether the LCPIH holds for a subsample that

confirms both postulates. If the LCPIH holds, the expected income changes should not explain

the consumption changes in the next period for rational consumers, who not only precisely

predict their future income changes but also execute their ex ante optimal consumption plans.

To examine this prediction, we employ Japanese panel data from the Preference and Life

Satisfaction Survey (PLiSS). The PLiSS is a longitudinal survey conducted from 2004 to 2010

that includes questions about individual preferences and future expectations of income and

consumption. The responses to the self-reported and retrospective questions enable us to elicit

each respondent’s forward-looking behavior, which is an element of the REH, and each behavior

relating to their self-control problems.7Based on the responses to these questions, we can then

test the LCPIH for the subsample that confirms the assumptions of the LCPIH.

A feature of this analysis is the use of subjective perceptions on own behavior to proxy for

the behaviors required by the LCPIH to hold. However, using self-reported and retrospective

questions is not only a novelty of this article but also a limitation. As discussed in the psychology

literature, such questions are subject to response bias and measurement error as general issues

concerning survey data (Kalton and Schuman, 1982; Furnham, 1986).8However, although there

is some criticism of these questions, some studies support the validity of responses about real

preferences for risk attitudes and patience in comparing survey and experimental data (Dohmen

et al., 2011; Vischer et al., 2013).9Although we cannot directly examine and address the effects

of response bias, we are able to use several self-reported and retrospective questions to provide

robustness.

Tests of both the REH and the LCPIH require the expectation of an income change. In

our study, we employ subjective expectations of income changes, which are useful when using

the excess sensitivity test because they contain rich information that an econometrician cannot

typically observe (Browning and Lusardi, 1996; Jappelli and Pistaferri, 2000). Moreover, our

data on subjective expectations have at least two advantages over previous studies, such as Das

and van Soest (1999), Jappelli and Pistaferri (2000), and Souleles (2004). First, the responses to

our questions on expected income changes are quantitative instead of qualitative. Second, the

responses have potentially greater variation entailed than those found in previous studies. For

example, Souleles (2004) specified the responses about the expectation of financial position with

just three variations, namely “will be better off,” “same,” and “will be worse off.”In contrast, our

measure includes 11 variations, ranging from less than –9% to more than 9%. Thus, our data

5For example, Das and van Soest (1999) and Souleles (2004) examine whether households forecast their future

income changes correctly using microdata.

6Some empirical studies have shown that a certain proportion of people do experience self-control problems. These

include Frederick et al. (2002) and DellaVigna (2009).

7Dohmen et al. (2011) and Vischer et al. (2013) show how survey measures could elicit individual preferences by

comparing them with experimental results.

8Response bias can affect the validity of the survey to which the participant responds. Existing studies have identified

several types of response bias, including, for example, anchoring and acquiescence bias (Hurd and Kapteyn, 1999) and

social desirability bias (Grimm, 2010).

9There are two arguments for and against the treatment of response bias. One is that the effect of response bias

is negligible, as these effects would simply disappear in a sufficiently large sample were the bias independently and

identically distributed. The other is that response bias has a significant effect.