Savings, Asset Holding and Debt: New Evidence from Chinese Household Data

This thesis consists of three empirical studies that investigate contemporary topics related to household finance. Specifically, this thesis aims to contribute to the literature relating to household savings, household risky assets and household debt by examining three distinct but related topics in the context of China. The first empirical study (Chapter 2) examines the relationship between planning for overseas education and household saving behaviour in China by using a household-level dataset, the China Household Finance Survey (CHFS) covering 2011, 2013, 2015 and 2017. This chapter also examines the role of planning for overseas education on wider measures of household assets such as household financial assets and household net wealth. The results indicate that households where parents plan to send their children to study abroad hold more household savings, household financial assets and household net wealth than those who do not plan to do so for their children. Furthermore, such a positive effect of planning for overseas education on household savings is revealed after dealing with potential endogeneity issues. In addition, different effects of planning to send children to study abroad on household savings are found across the whole savings distribution. The second empirical study (Chapter 3) examines the relationship between financial literacy and risky asset holding in China using a panel dataset from the CHFS covering 2013, 2015 and 2017 in order to control for unobserved heterogeneity. Risky asset holding is captured in three ways: the probability of holding risky assets; the log level of risky assets; and the share of risky assets in total household financial assets. Then, household risky assets are split into high-risk assets and low-risk assets. In addition, this chapter explores the relationship between financial illiteracy and household risky asset holding. The findings indicate that financial literacy is positively associated with risky asset holding. The importance of the role of financial literacy on household risky asset holding remains once time-invariant effects have been accounted for. Furthermore, financial literacy has been found to be positively associated with high-risk asset holding and low-risk asset holding but the size of the positive effect of financial literacy differs across high-risk assets and low-risk assets. Finally, these findings have been found to be robust after dealing with the potential endogeneity of financial literacy and the results have also revealed a negative relationship between financial illiteracy and household risky asset holding. The third empirical study (Chapter 4) examines the association between risk attitudes and household debt using a household-level dataset from the CHFS (2011, 2013, 2015 and 2017). Household debt is captured by the probability of holding household debt and the amount of total household debt held. Then, household debt is split into housing debt and non-housing debt to explore how risk attitudes affect the two types of debt. In addition, households are split into urban and rural households in order to explore whether the effect of risk attitudes on household debt differs across urban and rural households. Finally, this chapter investigates the two-part process related to holding total household debt: (1) the decision to hold debt; and (2) the decision over the amount of debt held. The results indicate that risk tolerance is positively associated with household debt. The findings also indicate a positive relationship between risk tolerance and non-housing debt. In addition, we have found differences in the effect of risk tolerance across total household debt, housing debt and non-housing debt by rural and urban households. For example, the magnitude of the marginal effect of risk attitudes on the probability of holding total household debt is larger for rural households than for their urban counterparts. Finally, the findings are robust to using the double hurdle approach thereby providing further evidence that the risk tolerance of the head of household is positively associated with household debt.