Traditional alpha-based tests face substantial noise when estimated on single stocks. Using diversified test portfolios help to reduce this noise, at the cost of inducing aggregation error and reducing cross-sectional variability. To address these shortcomings, we propose a more efficient statistic, a sharper alpha, that reduces estimation noise for single stocks. We find that, while sharper alphas estimated on portfolios are similar to traditional OLS alphas, they provide significant noise reduction at the stock-level and reveal cross-sectional patterns that were not visible before.
We develop a tractable general equilibrium framework providing a direct mapping between (i) the supply and demand for capital at the firm level and (ii) the cross-section of stock returns. Investor behavioural tilts and hedging needs drive capital supply, while firm profitability drives demand. Heterogeneity in supply and demand factors determines the sign of the risk-return relation and generates anomalies such as betting-against-beta, betting-against-correlation, size, value, investment, and profitability. We estimate the supply and demand schedules of over 4,000 U.S. firms and verify that the model accurately predicts the sign of the risk-return relation conditional on characteristics.
We document that the variance risk premium in asset returns decreases firms’ investments. We theoretically model the premium; we find that it increases the value of the real option to delay an investment and, thus, influences investments negatively. Empirically, we verify the negative link between the variance risk premium and investments. Cross-sectionally, the link is more important for investment-grade firms, which have relatively higher exposure to systematic variance risk. This premium helps us understand an otherwise surprising pattern investments are lower for investment-grade firms with better investment conditions than speculative-grade. Investment-grade firms basically hedge variance risk by delaying investment.