Another issue addressed by this study is the size of the housing wealth effect, that is, the MPC. The FIES also contains information on the annual pre-tax household income of the previous year and on a wide variety of household characteristics, such as the age of the head of the household, the number of household members, etc. As for the estimation of the value of the houses of survey households: the FIES provides information on the floor area (in square meters), the structure of the house (wood, reinforced concrete, etc.) and the year of construction.
9 According to the 2009 National Survey of Family Income and Expenditure (NSFIE), the estimated average value of the household's home and the land it sits on (for households of two or more people) is approximately 20 million yen, whereas the estimated average value of non-owner-occupied housing (buildings and land) owned by households is about 5 million yen. To investigate housing wealth effects, we use the theoretical prediction of standard LC/PIH that current consumption depends on current asset holdings and human capital (ie, the discounted sum of expected future income). Since human wealth cannot be observed, it is related to household income,14 the industry in which the head of the household works, and the type of work15 of the head of household.
As already mentioned, as proxies for permanent household income, we include household income, the industry in which the head of the household works, and the type of work of the head of the household in the regression. The first is based on the fact that under the LC/PIH standard, life resources consist of (i) housing wealth, (ii) net financial wealth and (iii) human wealth, and that the share of human wealth becomes negligible after retirement. Despite the fact that the FIES contains information on the size category (in terms of number of employees) of the company in which the head of the household works, we decided not to use this variable in our regression, as it is only available for about 60% of households in the sample .
In specifications (1) and (5), which make full use of the large sample size, the MPC of housing wealth is 0.0010 for unaffordable and 0.0033 for total consumption.
Pseudo-panel analysis
The reason is that the volatility of housing wealth has historically been greater in cities than in rural areas, which is likely to lead to greater variation in the key explanatory variable (𝛥𝛥 𝑙𝑙𝑙𝑙 𝐻𝐻𝐻𝐻𝑖𝑖, 𝑡𝑡+1. in our case land area measured in square meters). constant to obtain a consistent estimator for housing wealth effects. In the following, we exploit our data structure, which allows us to break down the growth rate of housing assets (𝛥𝛥 𝑙𝑙𝑙𝑙 𝐻𝐻𝐻𝐻𝑖𝑖 ,𝑡𝑡+1) into two parts: the change in land prices per square meter (𝛥𝛥 𝑙𝑙𝑙𝑙 𝑃𝑃𝑖𝑖,𝑡𝑡+1)26 and land area, measured in
In specifications (1) to (4), in which the key independent variables are the rate of change in housing wealth (𝛥𝛥 𝑙𝑙𝑙𝑙 𝐻𝐻𝐻𝐻𝑖𝑖, 0.0049 for volatile and aggregate consumption, respectively. Focusing on States of Taken together, they instrument changes in housing wealth from price elasticities of housing supply using county-level micro data.As for changes in the responses of young and old, specifications (2) and (4) show that consumption of older households responds to changes in housing wealth to a greater extent than the consumption of young households and that MPCs for young people are slightly negative.
In summary, the finding that a greater effect of housing wealth on consumption is observed for older households remains unchanged even in the pseudo-panel analysis, which differs from the contradictory results obtained by Attanasio et al. In addition, the MPCs out of housing wealth in the pseudo-panel analysis for non-durable consumption and for total consumption are values that are almost equal to or slightly larger than the estimates obtained in the cross-sectional analysis and are smaller than the reported MPCs for other developed countries ( see section 2.1). That is, if households own other properties that are rented out, only the value of the house they live in is included in our estimated value of housing wealth.
In the previous sections, MPCs from housing assets were estimated for all sample households and compared across age groups, following the empirical strategy of Attanasio and Weber (1994). However, a more conservative method to examine the effects on housing wealth is to estimate the MPCs for homeowners and renters separately. In specifications (2) and (6), the MPCs of housing wealth are 0.0010 (non-durable consumption) and 0.0027 (total consumption), which are significant but slightly smaller than the corresponding estimates in Table 2, which are based on the sample including tenants.
To circumvent this issue, we replace it with the average housing wealth in the city where the tenants live. According to estimates using specifications (2) and (6), the MPCs outside the housing estate are 0.0002 (non-durable consumption) and 0.0011 (total consumption). In summary, since the estimation results remain largely unchanged even when we exclude renters from the sample, in both the cross-sectional and pseudo-panel analyses, we conclude that our main finding still holds that the most plausible explanation for the observed movement between housing wealth and household consumption in Japan appear to be pure wealth effects.
Macroeconomic implications
Although it would be worthwhile to investigate in more detail which of the possible interpretations is more reasonable, we leave this issue for future research, as renters' response to changes in house prices is not the main focus of our research. On the other hand, the number of renter households is too small to assemble a pseudopanel. Tables 7 and 8 show the regression results based on the constructed pseudo-panel data for homeowner households.
The second row in Table 5 shows the result of the same exercise for the period 1991−1994. Given that real household consumption continued to grow by 8.9% between 1991 and 1994, our estimate suggests that during this period the fall in land prices simply reduced the increase in household consumption; that is, household consumption would have increased by 2.8% more if land prices had remained unchanged. This paper investigated the extent to which household consumption responds to changes in housing wealth using Japanese microdata covering approximately 500,000 households over the period 1983-2012.
The cross-sectional analysis showed that households with higher housing wealth spend more if other factors remain unchanged. The estimated MPCs are 0.0012 for non-sustainable and 0.0035 for total consumption. The derived MPCs apply to non-sustainable consumption and to total consumption. Here too, it turned out that the consumption of older households responds better to changes in housing wealth than that of younger households. Judging from these regression results, it appears that housing wealth effects can be observed in Japan, but are weaker than in other developed countries.
Moreover, the pure wealth effect channel provides the most plausible explanation for the positive relationship between consumption and housing wealth, which is consistent with the findings by Campbell and Cocco using country-level data for the United States, concluding that nearly 40% of the decrease in total consumption relative to trend from 2006 to 2009 was due to the drop in home values. All regressions include the age, sex, occupational industry and type of work of the household head, the previous year's (pre-tax) income, the prefecture of residence, family size, the number of household members aged 65 and over, the number of household members aged 15 and under, year dummies, cohort dummies, an urban dummy and first month of the survey dummies. The dummy "Young" takes one if the head of the household is 40 years or younger and zero otherwise.
The results of the test examining whether the MPC from housing wealth for young households is statistically different from zero are shown at the bottom of the table. Notes: In all specifications, the real interest rate, the rate of change of the previous year's pre-tax income, family size, the number of working household members, the number of household members aged 15 and under, and the number of household members aged 65 and over are included. The MPCs are calculated by multiplying the estimated elasticity by the consumption-housing-wealth ratio.
At the bottom of the table are the results of the test that examines whether the elasticity of consumption with respect to the housing wealth of young households is statistically different from zero. Notes: In all specifications, the real interest rate, the first difference between the previous year's pre-tax income, the family size, the number of working members of the household, the number of household members aged 15 or younger and the number of household members the household aged 65 or younger above are included.
