Do Heterogeneous Background Risks Matter to Household Asset Allocation?


  • Acknowledgments: The authors are grateful to Tom Smith and two anonymous referees for constructive suggestions, which greatly helped them improve the paper. Mingchao Cai acknowledges financial support from the National Natural Science Foundation of China Grant (70971088), and thanks Wang Sheng, Zhu Guomei and Ge Mengfei for their hard work in data collection during the survey. Mingchao Cai and Jing Shi acknowledge financial support from the National Natural Science Foundation of China Grant (71271135). All remaining errors are the authors'.

Corresponding author: Mingchao Cai, Antai College of Economics and Management, Shanghai Jiaotong University, 200052 No. 535, Fahua Zhen Road, Shanghai Jiaotong University, Shanghai, China. Tel: +86-21-54746492, Fax: +86-21-54746492, email:


This paper extends the literature reviews of Curcuru et al. (2009, Handbook of Financial Econometrics, Elsevier Science, Amsterdam), on the heterogeneity of background risk during investment into the following categories: income from labor and entrepreneurial firm, housing, wealth, health, professional knowledge and risk aversion attitude. Referring to the literature on asset allocation, the paper designs a survey and sets up a two-stage decision model to empirically test the relationship between households' risky assets demand and their heterogeneous background risks. Based on questionnaires administered to 770 Chinese households, the paper shows that a background risk substitution effect does not exist during households' market participation decisions and that their risky assets ratio mainly depends on subjective market expectations. Moreover, background factors are not properly considered; individuals with a higher income risk or older individuals exhibit a higher proportion of risky assets. Furthermore, households do not consider the background factors regarding housing and subjective risk preference attitudes significantly. The paper further finds that the value-at-risk of representative household's wealth in the sample is underestimated by approximately 29% when the housing background factor is not included.