Voluntary retirement savings in China: A spatial ordered probit approach

Abstract

This paper employs a spatial ordered probit model to study people’s voluntary retirement savings decisions using survey data collected in 2017 in China, where savings are recorded in a set of discrete ordered categories. To account for spillover effects and cross-sectional correlations, induced by for example omitted factors, we construct spatial connectivity matrices based on age and education for people located in the same province. We estimate the model using a Bayeisan scheme and discuss how to calculate various marginal effects, including those from nonstandard covariates like squared, binary, and multicategorical variables. Our empirical results indicate strong evidence of positive correlation in voluntary retirement savings decisions among high-income and low-income workers in both rural and urban areas. We find that participation in the government-managed basic pension is the strongest predictor of having high voluntary retirement savings, while the effects of income and gender vary across income-area groups.