Housing wealth appreciation and consumption: evidence from China

Abstract

This article examines the effect of housing wealth appreciation on household consumption by using two rounds of China Household Finance Survey data from 2013 and 2015. The empirical results show that housing wealth appreciation leads to an increase in household consumption, especially for consumption goods with higher expenditure elasticities, suggesting that increase in housing wealth not only promotes household consumption but also improves the composition of the consumption. Estimations on different subsamples suggest that precautionary saving acts as the primary mechanism through which housing wealth appreciation affects consumption in China. The analysis further reveals that household consumption lags behind housing wealth appreciation. We conclude that drastic changes in house prices are likely to exhibit augmented effects on the real economy via the consumption channel. Policymakers in China should prevent large swings in housing prices while taking regulatory steps to control the housing market.

Keywords:

• Housing wealth appreciation
• consumption
• China

Acknowledgements

The authors would like to thank the participants in the workshop on ‘The Rise of China: 40 Years of Reform and Opening-up and Implications for Other Countries’.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 We applied the formula ln(x + 1) to variables with a large number of zero values, such as spending on traveling. Exceptionally, we applied the natural log transformation to our key variable of interest, housing appreciation. Zero or negative housing appreciation suffers from severe measurement error in light of the recent housing booms all around China.

2 We exclude housing-related consumption, eg utility fees and housing renovation costs in our analysis to avoid reverse causality.

3 The CHFS questionnaire contains detailed information about housing attributes, such as homeownership, self-assessed home value, housing purchase price, housing tenure, housing size and so on. But it only records the detailed information of up to three housing units, so n ≤ 3.

4 Considering the time-value factor, we inflated the original purchase price using the yearly consumer price index in urban areas, which we obtained from the National Statistics Bureau.

5 We thank workshop participants for pointing out the time-discount factor of annualized housing appreciation. From a psychological perspective, the impact of annualized housing appreciation should decline with the length of housing tenure. That is to say, the household will value the annualized housing appreciation less as housing tenure increases since previous household consumption could have incorporated part of the housing effect. There are two ways to account for the time-discount factor. One is to propose a subjective discount rate. However, it is difficult to choose a discount rate that is not arbitrary. So we adopt an alternative way, which is to control housing tenure in the regression analysis. Note that we include the housing tenure of the primary house (mostly referring to the currently occupied house) as a control variable in all estimations, not the weighted housing tenure of the individual housing units.

6 To confirm the differences in the coefficients among the subsamples, we ran an auxiliary Seemingly Unrelated Regression model incorporating the three subsample estimations and then performed a Wald test of the null hypothesis with no difference between the subsamples. The test rejected the null hypothesis (chi-squared, 12.38; p value, 0.0021) and indicated that the differences among the three coefficients were statistically significant.

7 As in Table 5a, we applied a Wald test to check the significance of the differences in the coefficients between columns (1) and (2), and between columns (3) and (4), respectively. For the first two columns, we obtained a chi-squared value of 7.4 and a p value of 0.0065, while for the last two columns, chi-squared was 11.22 and the p value was 0.0008. Both sets of test results rejected the null hypothesis that the coefficients were the same within the two groups.

Additional information

Funding

This work is financially supported by the 111 Project under Grant No. B16040".

Notes on contributors

Changyan Peng

Changyan Peng joined the Survey and Research Center for China Household Finance, Southwestern University of Finance and Economics in September 2019. Her research interests include household finance and financial inclusion. Her work has been published in several Chinese refereed journals, including Management World and Agricultural Economic Review.

Weisong Qiu

Weisong Qiu is a PhD candidate at the Research Institute of Economics and Management, Southwestern University of Finance and Economics. His research interests include household finance and SMEs.

Quanyun Song

Quanyun Song joined the School of Finance, Southwestern University of Finance and Economics as an assistant professor in March 2017. Her research interests include household finance, bank loans, and SMEs. Her work has been published in Journal of the Asia Pacific Economy and International Review of Economics & Finance.

Bihong Huang

Bihong Huang joined the Asian Development Bank Institute as a research fellow in February 2016. Previously, she was the academic staff of Renmin University of China and University of Macau. Her research interests include environment, development, and financial economics. Her work has been published in books and refereed journals, such as China Economic Review, Economic Modelling, Energy Economics, Journal of Banking and Finance, Journal of Corporate Finance, Review of Development Economics, World Economy, etc.