This paper proposes a novel approach to estimate the elasticity of intertemporal substitution (EIS) of firm owners using a unique quasi-natural experiment and new theoretical insights regarding the spending response to news about future dividend tax changes. We study the Norwegian dividend tax reform, announced in 2004 and implemented in 2006, which increased the dividend tax rate by 28 percentage points. Leveraging rich administrative data and a dynamic difference-in-differences framework, we find that exposed households increased spending after the reform was announced and reduced it following its implementation. This behavior is only consistent with an EIS above one. Using a structural model, we estimate the EIS to be around 1.6..