Abstract: Numerous studies have revealed the gender difference in the relationship between socioeconomic status (SES) and obesity:for women, the higher the SES, the lower the likelihood of obesity; for men,the higher the SES, the more likely it is to be obese. According to the "social determination hypothesis",this gendered correlation can be attributed to people's SES,the fundamental cause of obesity. The causal relation presents a direction from SES to obesity. On the contrary,the "health selection hypothesis" proposes that the labor market makes discriminatory selection based on people's body size. As a result,the causal relation is from obesity to SES. This study aims to examine the above mentioned gendered causal relationship between SES and obesity. Specifically,the fixed effects model and the dynamic structural equation model are applied to address the two major challenges of casualty establishment:confounding bias and reciprocal/reverse causality. The analysis based on longitudinal data of the China Family Panel Studies (CFPS) of 2010,2012,2014 and 2016 suggests a clear gendered causal pattern:for men,the positive association between SES and obesity results from both social determination and health selection hypotheses:higher SES likely leads to weight gain in two years;while higher BMI is also related to SES achievement in two years. For women,however,the negative correlation is mainly due to the health selection process:higher BMI seems to significantly reduce individual SES after two years.

Key words: health inequality, obesity, causal relationship, gender heterogeneity, cross-lagged panel models