Collective Learning about Systematic Risk

We present an investment-based asset pricing model in which firms' exposure to systematic risk is uncertain. Beliefs about this parameter are updated from collective observations of firms' peers, causing an endogenous shift in the discount rate that should affect firms' real decisions and market valuations. We empirically show that the mean belief about risk exposure, which evolves through this collective learning process, negatively predicts the investment-capital ratio and the market-to-book ratio and positively predicts the implied cost of capital. In addition, greater precision in beliefs about the risk exposure parameter lowers the cost of capital and, in turn, raises capital investment, consistent with the model predictions. In contrast, an alternative risk estimate based on firms' individual histories is only insignificantly connected to the firm observables, offering evidence of the collective nature of learning.