Another Determinant of Household Leverage: Evidence from Japan's Mortgage Loan Data

• Author: Dong‐Ho Yeom,Mamoru Nagano
• Published date: 01 March 2014
• Date: 01 March 2014
• DOI: http://doi.org/10.1111/irfi.12025

Another Determinant of Household

Leverage: Evidence from Japan’s

Mortgage Loan Data*

MAMORU NAGANO†AND DONG-HOYEOM‡

†Faculty of Economics, Seikei University, Tokyo, Japan

‡Institute of Comparative Economic Studies, Hosei University, Tokyo, Japan

ABSTRACT

We investigate determinants of household leverage in Japan, which did not

experience the sharp rise in real estate prices and dramatic securitized mort-

gage market developments in the 2000s, and prove that these determinants

are not universal. We employ household sample data collected during 2001–

2010 by the Japan Housing Finance Agency and the Nikkei NEEDs Radar

Financial Survey, totaling to 28,561 samples. We f‌ind that the degree of

household interest rate risk preference, which proxies the household consti-

tutional and behavioral factors including risk tolerance, positively relates to

the household debt to income ratio. Further, as regards residential mortgage

loans, these interest rate risk-preferring households buy higher-priced houses

relative to their income, hold fewer f‌inancial assets, and tend to be headed by

young males. We also f‌ind similarities between household mortgage debt

determinants in Japan and the United States: the degree of regional bank

market competition and the state of bank management soundness inf‌luence

the aggressiveness of the residential mortgage loan business.

I. INTRODUCTION

A number of articles analyzing the causes and consequences of the residential

mortgage loan market collapse in the United States have been published since

2008. Pennington-Cross and Chomsisengphet (2007) pointed out that 65

percent of additional loan borrowing households that were stimulated by the

real estate property price rise already owned residences in the pre-bubble period.

Other studies such as Demyanyk and Van Hemert (2011) empirically concluded

that the creditworthiness of the securitized credited mortgage loan borrowers

was overrated by the residential property asset price rise, because creditworthi-

ness was measured by the secondary market interest rate spread from the

* An earlier version of this paper has been presented at several conferences, including the Inter-

national Review of Finance and the Nippon Finance Association Conference, Japanese Financial

Markets: Corporate Finance, Institutions, and Investments. We thank the Issue Editor, Sheridan

Titman, Jay Hartzell (discussant and referee), Sudipto Dasgupta, Takao Kobayashi, and the confer-

ence participants for valuable suggestions. This research is f‌inancially supported by KAKENHI,

Grant-in-Aid for Scientif‌ic Research (C) 24530357.

bs_bs_banner

International Review of Finance, 14:1, 2014: pp. 105–139

DOI: 10.1111/irf‌i.12025

© 2014 International Review of Finance Ltd. 2014

government bond yield. Thus, recent articles analyzing the U.S. household

leverage assert that the dramatic increase in household debt in the mid-2000s

was caused by the increase in additional external borrowing accompanied by

the price rise in real estate and excessive expansion of the mortgage loan

securitization business.

Meanwhile, many researchers have also observed a signif‌icant household

leverage increase in other countries where the rise in mortgage asset prices is not

dramatic or where mortgage securitization markets are less developed, although

these facts are hard to spot in the shadow of the U.S. mortgage loan market

crisis. For example, in December 2011, the Japanese government off‌icially

announced that the leverage of Japanese households increased on average by 10

percent compared to the observed value in 2001. Recently, the Chosun Daily

News (June 18, 2012) also reported that the number of Korean credit card

debtors who applied to the government bailout program exceeded the number

of applicants in 2011 by 1.2 million. Cecchetti et al. (2011) reported that

Denmark, Finland, the Netherlands, and Norway increased their household

debt to GDP ratio by more than 30 percent in 2000–2010. Thus, even though

some of the abovementioned countries did not experience soaring prices in the

real estate market or dramatic securitization market expansion during the last

decade, they did experience a household leverage increase.

This paper focuses on Japan, where neither the sharp rise in real estate prices

nor dramatic securitized mortgage market developments were experienced in

the 2000s. We believe that the household leverage determinants derived from

analyses of U.S. households are not universal, that is, they do not always apply

to other countries, and that other determinants can explain the recent house-

hold debt increase in other countries. Specif‌ically, we examine our overall

hypothesis that a change in the household constitutional factors inf‌luences the

household debt even under the zero interest rate def‌lationary economy. The

term household constitutional factor includes risk preference, willingness to

utilize external f‌inance, and an optimistic view of future repayment capability,

and we assume that these are positively related to the household time discount

rate. Therefore, we def‌ine these constitutional factors as personal and household

characteristics that encourage immediate house purchase to increase utility

because the utility of an immediate purchase is larger than that of a future

purchase for these households, assuming the time discount rate is not constant.

Further, we hypothesize that these constitutional factors inf‌luence the house-

hold leverage in Japan in addition to the factors of real estate price rise and

securitization market development already pointed out in the existing literature.

We would have ordinarily employed household time discount rate f‌igures

directly estimated by an experimental survey but cannot do so owing to the lack

of such survey data for household mortgage contractors. Therefore, we employ

variables representing interest risk preference in mortgage f‌inance, risky house-

hold asset investment patterns, and frequency of credit card use as household

constitutional factors including risk tolerance, in line with the studies by Van

Praag and Booji (2003) and Hiruma and Ikeda (2007) that present empirical

International Review of Finance

106 © 2014 International Review of Finance Ltd. 2014

evidence that these variables and the personal discount rate are positively

related based on their experimental surveys. Therefore, we def‌ine that variables

representing interest risk preference include risk preference, willingness for

external f‌inance dependency, an optimistic view of future repayment capability,

and other constitutional factors that are positively related to time discount rate

utility.

On the basis of this research motivation, this study focuses on identifying

other determinants of household leverage, using household loan contracts and

survey data from Japan. In this paper, we set two goals. The f‌irst goal is to

discover a household leverage determinant not found in the existing literature.

The second goal is to conf‌irm whether household leverage determinants listed

in the existing literature are consistent in the Japanese context. In the next

section, we review the existing literature to show how it relates to our study and

the types of empirical methodologies employed therein. In Section III, we

provide a background of the recent mortgage loan market in Japan and intro-

duce our hypothesis explaining the recent increase in household leverage in

Japan. In Section IV, we present the data used to examine our hypothesis. In

Section V, we discuss our empirical model and introduce new empirical evi-

dence. In Section VI, we present our conclusions.

II. EXISTING LITERATURE

As noted in the previous section, numerous studies have analyzed the causes

and consequences of the rise and fall of the U.S. residential mortgage market.

Among these, some researchers have studied the wealth effect of residential

property asset value. This topic was also studied before the subprime crisis,

because researchers are interested in studying this effect over a long-term period

(see Table 1). Naturally, post 2007, the number of related articles increased

exponentially. Mian and Suf‌i (2011), Bostic et al. (2009), Bostic et al. (2007),

Lustig and Nieuwerburgh (2005), and Benjamin et al. (2004) are examples of

studies that belong to what we shall term as the f‌irst group in the literature

relevant to our discussion. On the basis of their empirical analysis using house-

hold survey data from the U.S. Inland Revenue Service, Mian and Suf‌i (2011)

concluded that a sharp fall in residential mortgage property asset values induces

a decrease in household demand for durable consumer goods. Bostic et al.

(2007) sourced data from Survey of Consumer Finances and Consumer Expenditure

Survey published by the Federal Reserve Board and the Ministry of Labor,

respectively, and provided empirical evidence that the wealth effect caused by

the change in the residential mortgage asset price is larger than that caused by

the change in household f‌inancial asset value. The result is consistent with that

of Lustig and Nieuwerburgh (2005), who also provided empirical evidence of

the same by analyzing household long-term quarterly aggregated data between

1952 and 2001. Benjamin et al. (2004) also asserted that the wealth effect from

owned residential property is four times as large as that from the rise in

household f‌inancial asset value. Bostic et al. (2009) shared the abovementioned

Another Determinant of Household Leverage

107© 2014 International Review of Finance Ltd. 2014