In this paper, I consider the role of news provided by the media as signals used by investors to learn the underlying degree of partisan conflict. Partisan conflict is relevant for investment decisions because it affects the intensity of legislative effort aimed at (i) improving the institutional environment in which firms operate and (ii) instituting tax reforms. Higher partisan conflict makes tax reforms less likely but increases the probability of crises. These, in turn, affect the after-tax returns to investment. Whether the uncertainty and gridlock induced by partisan conflict is beneficial or detrimental for the economy depends on the status-quo level of taxes, on the identity of the party proposing policy reforms, and on the expected severity of crises. This is the case because even though a higher likelihood of bad economic outcomes is always negative for investment, stalemate makes tax-hikes less likely under some scenarios, and this may increase expected returns to investment. Agents do not observe the true degree of political disagreement (and hence the quality of policies), but can create expectations based on the observation of informative signals. Using a Bayesian learning model, I illustrate how these signals affect investment decisions by changing agents' expectations. I show that, to the extent crises are severe enough, an increase in the partisan conflict index (a summary of the signals observed) reduces expected returns and induces lower investment. Interestingly, investors react to news through changes in expectations even when there is no change in fundamentals.