Description: An overview of bias and fairness in machine learning models, as they apply to use cases in life insurance.

Description: This paper uses the ideal quasi-natural experiment of government-subsidized agricultural insurance program phased out in different regions of China since 2007 to study the effects and transmission mechanisms of the program on alleviating regional poverty using a multi-period Difference-in-Differences (DID) approach. Specifically, this paper identifies the causal effects on the incidence, depth and intensity of poverty among rural residents. The research has important policy implications. First, this paper uses the program as a case to provide a research paradigm to estimate effects of implementing policy instruments that combine markets and government mandates. Second, as common prosperity is now a government imperative, the research also provides insights into ways insurance can be used to hedge income risks, alleviate poverty and help achieve more equitable distribution. The paper has been published on WeChat account of Chinese Association of Actuaries on February 9, 2022. The full version has been published in CSSCI journal Insurance Research, issue 11, 2021.