Evaluating the heterogeneous treatment effects of retirement on the mental health of older adults

In this paper, we investigate the heterogeneous treatment effects of retirement on the mental health of older adults using the generalized random forest (GRF) method. GRF has high functional flexibility and can explore complex treatment heterogeneity by selecting covariates from high-dimensional data that contribute to heterogeneity. Results show that retirement significantly reduces the elderly group's depression as well as increases subjective well-being, implying that it will improve mental health. The best linear predictor test suggests that the effect of retirement on depression is heterogeneous among the data for 2014 and 2020 but not in the analysis of well-being. Retirement is more effective in improving depression for males who are at an unhealthy weight, are between the ages of 57 and 67, and have an Internet habit. The results of this paper provide support for promoting the optimization and individualized implementation of relevant policies and improving the lives of the elderly in their later years.