Knowledge that Transforms

Abstract Generative artificial intelligence (AI) is increasingly consuming and producing huge amounts of data. We propose a social learning model of AI, emphasizing a data-AI feedback loop: data quality affects AI productivity, which influences AI adoption and, consequently, the composition (AI versus human-generated) and quality of future data. Calibrated to evidence on synthetic training loops, the model predicts hump-shaped labor dynamics—short-term displacement that partially reverses as data quality deteriorates. A Grossman–Stiglitz-style externality emerges: AI adopters free-ride on the human-generated actions that supply the novel information on which AI itself relies. In a competitive market, AI should be taxed to correct the data-quality externality; a concentrated AI industry overcorrects, making a subsidy optimal. (JEL O33, D62, D83, J24)

Abstract There are two major institutional investors in the syndicated loan market: collateralized loan obligations (CLOs) and loan mutual funds. CLOs are closed-end funds while loan funds are open-end funds that issue claims that are redeemable on demand. In this paper, we examine whether CLOs provide arbitrage capital that contributes to the resilience of loan funds. We find that CLOs act as shock absorbers, providing liquidity through par-building trades when loan funds experience large outflows. However, CLO-provided liquidity is limited to par build–eligible loans, leading to potential flow-induced fire sales among par build–ineligible loans. (JEL G23, G38

Abstract We show that positive flows to active mutual funds with high recent returns partially reverse at longer horizons. This outcome is robust across a broad range of alternative specifications. Reversal occurs from greater outflows associated with high prior returns, not reduced inflows. We test theories to explain the reversal: investment life cycles, tax loss selling, and a behavioral “disappointment” hypothesis based on investors’ overreaction to positive returns. While both tax loss selling and short investor life cycles can contribute, the evidence supports a role for investor disappointment, whereby investors redeem their capital when return performance fails to meet expectations.

Abstract A fundamental concern about green investing is that it may crowd out political support for public policies addressing negative externalities. We examine this concern in a preregistered experiment conducted shortly before a real referendum on a climate law in Switzerland. We find that offering an opportunity to invest in a climate-friendly fund does not reduce individual support for climate regulation, measured by political donations and voting intentions. A replication of the experiment in the United Kingdom yields similar results. Our estimates reject a crowding-out effect, suggesting instead a modest crowding-in effect of green investing on political support for green policies.

Abstract We develop a model-free measure of the variance premium by constructing option portfolios whose returns are highly correlated with realized stock variance. This effectively decomposes returns into realized variance minus implied variance. We apply this decomposition to document a novel quarterly cross-sectional continuation pattern in both realized variance and implied variance of individual stocks. Implied variance underanticipates the seasonality of realized variance, so options that performed well at quarterly lags continue to earn high returns in the future. Quarterly periodicity in realized stock variance only occurs on days with analyst earning revisions, suggesting an informational channel for this pattern. (JEL G12, G13, G40)

Abstract This study examines whether mutual fund managers generate alpha through thematic investment strategies that select stocks poised to benefit from specific themes. Using textual analysis of 10-K filings, we identify stocks’ thematic exposures and construct each fund’s thematic concentration index (TCI) from its holdings. High-TCI funds significantly outperform, with a top-minus-bottom decile spread of 4.26% in annualized four-factor alpha. Managers’ thematic expertise is related to their undergraduate field of study. Outperformance arises from superior stock selection rather than theme-related timing, with an informational advantage on firm earnings, particularly in stocks exposed to themes related to managers’ undergraduate training.

Abstract We examine the informativeness of fund risk disclosure by combining fund returns and textual data. We develop a novel measure, INF, to capture the explanatory power of disclosed risks for fund returns. Average INF is 55% after controlling for market risk and remains 26–29% after excluding risks related to fund name or strategy. We explain variations in INF through disclosure costs and benefits: (1) Funds with less informative disclosures face SEC comment letters and reduced flow, (2) informative disclosures attract institutional flows, and (3) performance concerns may disincentivize disclosure. Overall, funds face trade-offs between transparency and maintaining their competitive advantage.

Abstract We analyze optimal cross-country risk sharing and bank capital requirements amid collective moral hazard by governments and banks. Transfers provide insurance but weaken fiscal discipline. Since lenient fiscal policies amplify banks’ risk-shifting, optimal support must be contingent on banking system health. For fiscally weak countries, support is larger in sovereign crises with solvent banks. For fiscally strong countries, support is larger in joint sovereign and bank crises, requiring deposit insurance mutualization. Optimal contracts feature lower capital requirements than without cross-country transfers, raising the cost of joint sovereign and bank crises to strengthen fiscal discipline and enable greater risk sharing.

Abstract Using a novel measure of investment companies’ reliance on generative artificial intelligence (GenAI), we document a sharp increase in GenAI usage by hedge funds after ChatGPT’s 2022 launch. A difference-in-differences test shows that hedge funds adopting GenAI earn 2-4% higher annualized abnormal returns than nonadopters, while non-hedge funds do not benefit. The outperformance originates from funds’ AI talent and ChatGPT’s strength in analyzing firm-specific information. We conduct a new survey of fund managers’ GenAI usage to provide direct validation of our measure and offer additional new insights on how managers adopt GenAI tools in their practice.

Abstract We examine the role of race and ethnicity in the mutual fund context at two distinct levels. At the fund manager level, we document a co-racial tilt—funds managed by minority-dominant (White-dominant) teams allocate larger portfolio weights to minority-led (White-led) firms. This tilt is not associated with superior performance. It diminishes as fund managers gain experience, suggesting the presence of inaccurate statistical discrimination. At the investor level, we find that minority-dominant funds are penalized similarly to White-dominant funds for poor performance but are not rewarded as much for superior performance. Overall, our results uncover race-related investment choices at both levels.