A Study Of The Heterogeneous Effects Of Retirement On The Mental Health Of Older Adults

As China’s 1960s baby boomers gradually move into retirement,more and more people will reach retirement age.According to a report by China’s National Bureau of Statistics,by 2021,China will have 264.02 million people over the age of 60,accounting for 18.7%of the total population,and 190.64 million people over the age of 65,accounting for 13.5%of the total population.This shows that China’s current aging problem is already quite serious.And China has a huge population base;the aging population will bring more retirees.In order to alleviate the pressure of the pension and labor shortages,more and more people are calling for a delayed retirement policy.By 2022,the delayed retirement policy will already have been implemented in some regions of China.Therefore,sorting out the heterogeneous effects of retirement on mental health will not only help to remove individuals’ doubts about the delayed retirement policy and reduce resistance to its implementation but also facilitate a more targeted implementation of the policy.First,we combed through the relevant domestic and international literature and found that most current scholars have used breakpoint regression designs,two-stage least squares,instrumental variables,and propensity score matching methods,etc.These methods are efficient when dealing with low-dimensional data,but their estimation results may no longer be valid when the data are extended to high dimensions.Also,when performing heterogeneity analysis,researchers may add covariates to the model that are not significantly heterogeneous.Based on this,this paper introduces the generalized random forest algorithm of machine learning into the analysis of the effect of retirement on mental health.Second,we modeled the CFPS using data from 2014,2016,2018,and 2020 in each year.The results show that retirement has a significant improvement effect on mental health in all years,and this effect increases dramatically and then decreases slowly over time.We analyzed that this dramatic increase might be caused by the significant increase of residents’ time spent on the Internet in recent years.The best linear prediction test reveals that the generalized random forest can estimate satisfactory heterogeneity treatment effects.The results of the heterogeneity analysis indicated that the effect of retirement on improving mental health was more pronounced for older adults with suboptimal weight,aged 57-67 years,and with Internet access habits.Finally,we conclude the full paper and propose relevant policy recommendations from the government and individual perspectives based on the findings.