Using a game-theoretic approach, the benefits of macroprudential policy coordination are evaluated in a two-country model of a currency union with financial frictions. The gains from coordination are measured by comparing outcomes under a centralized regime, where a common regulator sets a macroprudential tax on loans to maximize union-wide welfare, and a decentralized regime, where each regulator sets the tax independently to maximize own-country welfare. Numerical experiments show that, in response to financial shocks, coordination involves increased activism in the country where the shock originates (keeping one’s house in order). However, while union-wide gains from coordination are positive when country-specific instrument rules are set, a one-size-fits-all policy makes the union worse off when member countries are asymmetric in size and structure. Under both the centralized and the decentralized regimes, getting monetary policy to lean aggressively against the financial cycle may be suboptimal. The broad implications of the analysis for macroprudential policy coordination and the central bank’s mandate in the euro area are also discussed.