中国劳动经济学者论坛工作论文 China Labor Economists Forum Discussion Paper No. 76
Heterogeneous preferences and risk sharing at households level in China
Jennifer T. Lai; Isabel K. M. Yan; Xingjian Yi
发布时间 Public time: 2016/7/5 21:02:32 更新时间 Update time: 2016/7/5 21:02:32 阅读 Read: 370
摘要 Abstract: This paper investigate the degree of risk sharing across households in China. Standard tests assume homogeneous preferences across households, which may bias the true risk sharing degree to different directions, if risk and time preferences are correlated with variations of household income. We use Chinese Health and Nutrition Survey (CHNS) and China Family Panel Survey (CFPS) to show that in China, the incomes of less risk-averse and less patient households correlates more positively with the aggregate risk. These two correlations bias the true degree of risk sharing toward opposite directions if homogeneous preference is assumed. We apply the factor and GMM estimation techniques from Schulhofer-Wohl (2011) to the CHNS data. We find that around 30% of household income shocks would pass through to household consumption when households are assumed to have homogeneous preferences. When both risk and time preferences are allowed to differ across households, this number remain roughly the same, because risk preference heterogeneity would adjust the estimated risk sharing parameter downward while time preferences heterogeneity would adjust it upward, so the two effects roughly offset one another. By comparing this result with that of US, we find that the degree of risk sharing across households in China is worse than that in the US, where roughly less than 10% of household income shocks pass through to consumption, indicating that household consumption is more sensitive to idiosyncratic household income risk in China. In addition, we find that when transfer income is included into household income, the degree of risk sharing across households does not improve. This result remains robust when we further splitting household transfer income into public transfer and private transfer components. Our findings have important policy implications.