原文地址：https://zhuanlan.zhihu.com/p/19841390
我校 Hui He, Feng Huang, Dongming Zhu 与旧金山联储的 Zheng Liu 合作的文章：《Breaking the “Iron Rice Bowl” and Precautionary Savings: Evidence from Chinese State-Owned Enterprises Reform》使用上世纪末的国企改革作为自然实验，估算了中国家庭预防性储蓄的规模，摘要如下：
We use China’s large-scale reform of state-owned enterprises (SOE) in the late 1990s as a natural experiment to identify and quantify the importance of precautionary savings for wealth accumulation. Before the reform, SOE workers enjoyed the same job security as government employees. After the reform, a cumulative of over 35 million SOE workers have been laid off, although government employees kept their “iron rice bowl.” The change in unemployment risks for SOE workers relative to that of government employees before and after the reform provides a clean identification of income uncertainty that helps us estimate the importance of precautionary savings. In our estimation, we correct a self-selection bias in occupational choices and we disentangle the effects of uncertainty from pessimistic outlook. We obtain evidence that precautionary savings contribute to about one-third of the wealth accumulations for SOE workers between 1995 and 2002. Precautionary savings motive is thus an important factor that drives the observed rising Chinese saving rate.
文献中识别预防性储蓄是非常困难的，因为识别家庭收入风险的原因本身就很困难。文章使用上世纪 90 年代的国企改革作为外生的冲击，比较了政府工作人员与国有企业工人的储蓄行为。因为政府工作人员几乎不受到这次改革的影响，而国有企业工人的铁饭碗被打破，因而收入风险大大增加。利用这一外生的政策变化，使用（类似于）DID 的方法，识别除了预防性储蓄的规模。
此外，此类文献一般受到 self-selection biases 的影响。为了去除自选择效应，作者将样本限制在政府分配的工人中，这样就不会存在风险偏好不同的工人选择不同种类工作的问题。
根据作者估计的结果，预防性储蓄大约构成了30%的家庭金融资产积累。