Knowledge that Transforms

This paper examines the information content of firm-specific sentiment extracted from Twitter messages. We find that Twitter sentiment predicts stock returns without subsequent reversals. This finding is consistent with the view that tweets provide information not already reflected in stock prices. We investigate possible sources of return predictability with Twitter sentiment. The results show that Twitter sentiment provides new information about analyst recommendations, analyst price targets and quarterly earnings. This information explains about one third of the predictive ability of Twitter sentiment for stock returns. Taken together, our findings shed new light on whether and why social media content has predictive value for stock returns.

In this study, we investigate whether investors are willing to trade off wealth for societal benefits. We take advantage of unique institutional features of the municipal securities market to provide insight into this question. Since 2013, states and other governmental entities have issued over $23 billion of green bonds to fund eco-friendly projects. Comparing green securities to nearly identical securities issued for non-green purposes by the same issuers on the same day, we observe economically identical pricing for green and non-green issues. In contrast to a number of recent theoretical and experimental studies, we find that in real market settings investors appear entirely unwilling to forgo wealth to invest in environmentally sustainable projects. When risk and payoffs are held constant and are known to investors ex-ante, investors view green and non-green securities by the same issuer as almost exact substitutes. Thus, the greenium is essentially zero.

Within the UK's proactive financial-reporting enforcement regime, we examine the effect of increased regulatory scrutiny on equity values. We find that a fourfold increase in the likelihood of regulator-initiated reviews of financial reports reduces equity values by 1.3% on average. Reductions in equity values are largest for firms with strong private oversight that likely ensures that they are closer to their equity-value-maximizing level of transparency. Additional evidence suggests that competition increases and that managers' investment horizons decrease in industries selected for increased regulatory scrutiny, consistent with direct compliance costs not fully explaining the reduction in equity values.

Psychology and neuroscience studies document that facial trustworthiness perceptions may affect observers' decision-making process. Our study examines whether auditors' perceptions of client executives' facial trustworthiness are associated with their audit fee decisions. We employ a machine-learning-based face-detection algorithm to measure executives' facial trustworthiness. We find that auditors charge 5.6% less audit fee to firms with trustworthy-looking CFOs than to those with untrustworthy-looking CFOs in initial audit engagements. Auditor tenure weakens the negative association between CFOs' facial trustworthiness and audit fee. Further evidence shows that CFO's facial trustworthiness is associated with neither financial reporting quality nor litigation risk.

We assess the role of human discretion in lending outcomes using a randomized, controlled experiment. The lenders in our sample utilize a third party, machine-generated credit model as an input in their decision. We design a new feature for the credit-scoring platform – the slider feature – which invites lenders to incorporate additional discretion in their decision by adjusting the machine-based recommendation. We compare the loan outcomes for treatment lenders that randomly get the slider, relative to a control group. The treatment group's adjustments are predictive of forward looking portfolio characteristics – they show larger declines in future portfolio-level credit risk and larger increases in future sales orders, relative to the control group. The effects of our intervention are more pronounced when borrowers do not have social media accounts and in competitive markets. Our study provides insights about the role of human decisions, given the rapid evolution of machine-based lending models.

Chordia and Miao (2020) provide evidence that low-latency trading (LLT) improves the long-term informational efficiency of stock prices. This discussion raises two primary concerns with their analysis. First, the mechanism through which LLT enhances long-term efficiency is unclear. Second, CM's measure of LLT trading activity is correlated with non-LLT trading activity, which may in turn cause the documented improvements in efficiency. We close by proposing an alternative explanation—changes in market microstructure have had a bifurcated impact on liquidity, enhancing efficiency for large and liquid stocks, but not for small and illiquid stocks.

We connect conservative accounting to the cost of capital by developing an accounting model within an asset pricing framework. The model has three distinctive features: (1) transaction-cycle-conformity, where the book value equals the value of cash at the beginning and the end of a cash-to-cash transaction cycle; (2) a revenue recognition principle, where uncertainty affects the amount of revenues recognized; (3) a matching principle, where expenses are matched with revenue with a conservative bias due to uncertainty. We demonstrate how the growth rate of expected earnings, the accruals-to-cash ratio, and the expected earnings yield relate to the expected stock return.

We document that earnings acceleration, defined as the quarter-over-quarter change in earnings growth, has significant explanatory power for future excess returns. These excess returns are robust to a wide range of previously documented anomalies and a battery of risk controls. The future return predictability appears to be consistent with investors assuming a seasonal random walk model for quarterly earnings and missing predictable implications of earnings acceleration for future earnings growth. Finally, the excess returns from the basic earnings acceleration strategy can be enhanced further by focusing on profit firms, low earnings volatility firms and on specific patterns of earnings acceleration.

We examine how cultural differences among agents influence the aggregate outcome of a common forecasting task. Using both exogenous shocks to sell-side analyst diversity and panel regression methods, we find that increases in analyst cultural diversity positively affect the quality of the consensus earnings forecast. We further provide evidence on the potential mechanisms underlying this result by showing that cultural diversity is associated with improvements in individual analyst forecasts, greater analyst conference call participation and interaction, and greater diversity in analyst education backgrounds and professional interests. Overall, our results indicate that greater cultural differences among agents producing an aggregate forecast are associated with a higher quality consensus forecast.

We study how friendly boards design the structure of optimal compensation contracts in favor of powerful CEOs. Our study yields unexpected results. First, powerful managers receive higher pay and a contract with a higher pay-performance sensitivity (PPS) if firm performance is low and vice versa. Moreover, we identify conditions where expected pay and expected PPS are both increasing in the friendliness of the board. Second, we show that friendly boards provide managers with higher salaries, more shares, but less options. Third, friendly boards offering contracts with a higher PPS also make more intensive use of relative performance evaluation (RPE). Overall, our results suggest that frequently used indicators of poor (or sound) compensation practices should be interpreted with care. Extending the scope of our model beyond executive pay, we show that powerful managers underinvest in capital but have less incentives to manage earnings.