This paper proposes a novel approach to estimate the elasticity of intertemporal substitution (EIS) of firm owners using a quasi-natural experiment and new theoretical insights on spending responses to anticipated dividend tax changes. We study Norway‘s dividend tax reform, announced in 2004 and implemented in 2006, which raised the dividend tax rate by 28 percentage points. Using administrative data and a dynamic difference-in-differences framework, we find that exposed households increased spending after the announcement and reduced it following implementation. This pattern is consistent only with an EIS above 1. Using a structural model, we estimate owners‘ EIS to be 1.5.