Beyond linear risks: non-linear associations and gender asymmetry in spousal age differences and marital stability

Spousal age differences are a salient dimension of assortative matching in marriage markets, yet evidence on their relationship with marital stability remains mixed and often uses linear specifications. This study examines whether spousal age differences are associated with the risk of first divorce and whether large age gaps reflect distinct types of unions within gendered marriage markets. Using longitudinal data from Waves 1–20 of the Household, Income and Labour Dynamics in Australia (HILDA) Survey, we estimate Cox proportional hazards models of first divorce. Age differences are modelled both continuously and categorically to capture non-linearities and to distinguish between male-older and female-older unions. The results show a non-linear pattern across the specified age-gap categories that differs by gender direction. Moderate age differences (4–10 years) are associated with divorce risks similar to age-homogamous marriages, whereas very large age gaps (> 10 years) are linked to substantially lower divorce hazards. This pattern is concentrated in male-older unions, while age differences in female-older marriages show little association with divorce risk. The associations persist after covariate adjustment. Selective matching remains a possible explanation, but the analysis neither establishes selection as the cause nor rules out causal pathways related to age differences. More broadly, the results highlight the importance of non-linear associations and gender asymmetry in understanding how marriage markets structure union stability.