Knowledge that Transforms

How do recent advances in Generative Artificial Intelligence (AI) affect firm value? We construct the first measure of firms' workforce exposures to Generative AI and show that an “Artificial‐Minus‐Human” portfolio earned 5% in the two weeks following the release of ChatGPT. The labor‐exposure effect is more pronounced for firms with greater data assets and is distinct from the effect of firms' product exposures to AI. We assess whether exposed workforces are substituted or complemented by Generative AI based on whether their exposed tasks are core or supplemental. Examining firms' labor demand and profitability following the release of ChatGPT supports a labor‐technology substitution channel.

Firms tend to compete more aggressively when facing financial distress. This intensified competition reduces profit margins, pushing them further into distress and adversely affecting their industry peers. To study such feedback and contagion effects, we incorporate strategic competition in a dynamic model with long‐term defaultable debt, exploring various peer interactions such as predation and price wars. The feedback effect represents a novel source of financial distress costs associated with leverage, which helps explain the negative profitability‐leverage relation across industries. Owing to the contagion effect, firms' optimal leverage is often excessively high from an industry perspective, undermining the industry's financial stability.

When managers revise earnings guidance between earnings announcements, they choose not only what to disclose but also where to disclose—in a standalone release, a guidance call, an investor conference, or an analyst/investor (A/I) day. These venues differ in audience, format, and visibility, raising the question of whether venue choice is a strategic communication decision. In a large sample of unbundled revisions, we find that managers use calls for relatively large negative revisions and when litigation risk is higher and that they use conferences and A/I days for smaller revisions and when analyst disagreement is greater. Textual analysis reveals that calls feature cautious, litigious language for both upward and downward revisions, whereas conference and A/I day communications remain optimistic. In line with these framing differences, calls are associated with amplified reaction to negative news and are followed by more favorable earnings surprises, suggestive of effective expectation management, whereas conferences and A/I days are associated with muted reactions and reduced analyst dispersion and are followed by less favorable surprises. Our findings suggest that venue choice is a dimension of disclosure strategy for managers and an informative signal for investors.

I show how firms optimize the allocation of charitable funds around investments that require external stakeholder support, such as mergers and acquisitions. Acquirers increase donations to target‐related charities before making a takeover offer, but scale funds back after withdrawing from a deal. Firms redistribute funds from nontakeover‐related areas to target insider–affiliated charities, employee communities, and counties with a regulatory presence. Targeted donations that encourage blockholder reciprocity and employee commitment are associated with better deal performance through decreased merger premiums and lower integration costs. Overall, the results highlight firms’ use of philanthropy to manage stakeholder relations in takeover decisions.

To study competition, innovation, and industry dynamics that arise as a result of their interaction, we develop a new oligopolistic general-equilibrium Schumpeterian growth model. This model ties together the endogenous growth, oligopolistic competition, and dynamic industrial organization literatures in a single unified framework. Within each industry, there are an endogenously determined number of large firms (“superstars”) that compete á la Cournot and a continuum of small firms which collectively constitute a competitive fringe. Firms dynamically choose their innovation strategies, cognizant of other firms' choices, and their entry and exit are endogenous. The model is consistent with the macroeconomic trends observed in the United States since the 1970s, such as the domination of industries by a small number of superstar firms, the rise of markups, market concentration, profits, and R&D spending, and the decline in business dynamism, productivity growth, and the labor share. It replicates the empirical relationship between innovation and competition within and across industries. As an application, we estimate the model to disentangle the effects of separate mechanisms on the structural transition observed in the United States, which yields striking results: (1) While the increase in the average markup causes a significant static welfare loss, this loss is overshadowed by the dynamic welfare gains from increased innovation in response to higher profit opportunities. (2) The increasing costs of innovation are found to be the primary determinant of lackluster productivity growth, i.e., ideas are getting harder to find.

Central banks increasingly disclose their private information regarding the economy’s current and future state (CBET). Consistent with CBET shifting investors’ collection and pricing of aggregate-level (where managers do not have an information advantage) toward firm-level (where managers do have an information advantage) information, we find that CBET reduces managers’ learning from stock price. The results are (1) pronounced when firms do not provide guidance and when noise trading in firms’ stock is low, and (2) robust to using the 2013–2014 amendments to the Bank of England’s Inflation Report as an exogenous shock. Additional evidence suggests that CBET lowers investment efficiency.

Using résumé data on over 20 million U.S. workers, we find that the flow of employees between a pair of firms sharply drops by about 20% during periods in which the firms share a board director. This relationship is stronger in settings where firms have greater potential gains from limiting worker mobility and is most pronounced for higher-skilled employees. Moreover, we find more connected firms have lower rates of internal promotion and no improvements in employee satisfaction. The evidence is consistent with shared directors serving as a mechanism for labor market coordination.

Large language models are increasingly used in social sciences, but their training data can introduce lookahead bias and training leakage. A good chronologically consistent language model requires efficient use of training data to maintain accuracy despite time-restricted data. Here, we overcome this challenge by training a suite of chronologically consistent large language models, ChronoBERT and ChronoGPT, which incorporate only the text data that would have been available at each point in time. Despite this strict temporal constraint, our models achieve strong performance on natural language processing benchmarks, outperforming or matching widely used models (e.g., BERT), and remain competitive with larger open-weight models. Lookahead bias is model and application-specific because even if a chronologically consistent language model has poorer language comprehension, a regression or prediction model applied on top of the language model can compensate. In an asset pricing application predicting next-day stock returns from financial news embeddings, we find that ChronoBERT and ChronoGPT’s real-time outputs achieve Sharpe ratios comparable to a much larger and potentially biased Llama model. Our results demonstrate a scalable, practical framework to mitigate training leakage, ensuring more credible backtests and predictions across finance and other social science domains.

We study branch openings and closures in U.S. banking. Both are concentrated where depositors are rate-sensitive and place less value on physical access, patterns driven by financially sophisticated households with greater digital adoption. The effects are strongest for large banks, with lending playing a lesser role. Incumbents retain branches where depositors are less sensitive and more reliant on proximity; entrants avoid such markets because sticky customers are difficult to attract. The Pandemic accelerated closures while allowing banks to retain deposits through online channels. Overall, depositor sophistication — through rate sensitivity and reduced branch reliance — drives modern branch restructuring.