In an investment-based asset pricing model, we build a collective-learning framework in which decision-makers learn a target firm's exposure to systematic risk from its peers' observations. This learning mechanism endogenously creates a time-variation in the discount rate that significantly affects firms' real decisions and market valuations. We provide supportive empirical evidence that the posterior estimate of systematic-risk exposure from collective learning, rather than from individual firm's observations, negatively predicts firms' investment-capital ratio and market-to-book ratio, and positively predicts firms' implied cost of capital. Moreover, the precision of parameter beliefs also empirically predicts these variables in accordance with our model implications.