The “Fourteenth Five-Year-Plan” proposes to “implement the gradual deferment of the legal retirement age”. How to make the policy of deferring retirement widely accepted by the public is the key issue in the process of China’s endowment insurance system reform. Based on the theory of individual utility maximization, combined with China’s current endowment insurance system and S-type utility function in prospect value function, this paper constructs a comprehensive utility model associated with wage income, insurance payment, pension wealth and leisure, and determines the retirement age to maximize individual utility from the aspects of gender, leisure preference, wage income level, interest rate, insurance age and individual contribution rate of endowment insurance. Then it puts forward targeted suggestions on the policy of deferring retirement in combination with China’s current national conditions. The results show that the individual utility function under different retirement ages is a curve about retirement age, which first rises sharply and then decreases slowly. Male employees get the maximum utility when they retire at the age of 63, while female employees get the maximum utility when they retire at the age of 60. Meanwhile, the optimal retirement age of the insured and the insured does not change with the change of wage income level, which only affects the absolute value of utility obtained by insured workers. The higher the wage income level, the greater the utility value obtained by insured employees. In addition, the lower leisure preference, lower interest rate, older the insured age and lower the individual contribution rate of endowment insurance can result in the older the optimal retirement age of insured employees. In general, in the implementation of gradual retirement policy, only by considering gender factors, giving appropriate flexible operation space to the delayed retirement policy, improving the pension insurance payment method and creating a better employment environment for the elderly can make policy implication better.
