I study how biases in managerial beliefs affect firm performance and the macro-economy. Using confidential survey data to test whether US managers have biased beliefs, I establish three facts. (1) Managers are neither over-optimistic nor pessimistic: their forecasts for future sales growth are correct on average. (2) Managers are overconfident: they underestimate future sales growth volatility. (3) Managers overextrapolate: their forecasts are too optimistic or pessimistic depending on whether the firm is growing or shrinking at the time of the forecast. To quantify the micro and macro implications of these facts, I build and estimate a general equilibrium model in which managers of heterogeneous firms may have biased beliefs and make dynamic hiring decisions subject to adjustment costs. Biased managers in the model overreact to changes in their firm’s profitability because they believe profitability is more persistent and stable than it really is. The model thus implies that a typical firm’s value would increase by 1.9 percent if it hired a rational manager. At the macro level, pervasive overreaction results in too many resources spent on reallocation. Welfare would be higher by 1 percent in an economy with rational managers.