This paper investigates the impact of marital traditions on women's welfare. It constructs and structurally estimates a theoretical model to understand how cultural preferences influence a husband's decision to abuse his wife. Within the context of Pakistani exchange marriage custom - Watta Satta and the practice of child brides, the estimated latent preferences imply that abusive husbands have a negative perception of time invested in home production and prefer to have a relatively lower contribution to it than wives. Contrary to the literature, the main policy takeaway from the counterfactual policy experiment conducted in this paper is that women's empowerment schemes for financial independence will fail to impact marital abuse in patriarchal societies, while male-centric policies - favourable perceptions about daughters by men and education for boys have significant returns.