This paper replicates Kosowski, Naik, and Teo (2007) and Jagannathan, Malakhov, and Novikov (2010), two seminal studies of hedge fund performance persistence. We discuss the challenges of constructing archival samples which effectively duplicate those used in earlier work, exacerbated by the design and evolution of hedge fund databases. Although we do not have access to the same data, we can replicate both papers quite well. Our extension shows that top Bayesian alpha persists even in a novel,” real-time” sample that reflects when observations become available to the econometrician. Our strong evidence on hedge fund persistence contrasts with recent literature documenting that mutual fund performance does not persist anymore.

This paper examines how leverage affects asset prices via delegated portfolio management. We develop a stylized model in which a hedge fund raises capital from investors in order to benefit from its leverage advantage in the spirit of Berk and Green (2004). Deteriorating funding conditions tighten hedge fund leverage and induce the fund to invest more in high-beta stocks so that hedge fund holding betas reflect the tightness (and thus the shadow price) of leverage in equilibrium. Consistent with the model, we observe that: 1) hedge funds with top-quintile asset-implied leverage reap 38% more economic rents (fees) than bottom-quintile funds, 2) hedge funds simultaneously adjust leverage and holding beta during adverse funding conditions (proxied by recent poor performance), and 3) a leverage-tightness factor constructed from hedge fund holding betas can significantly predict the cross-section of asset returns. Its prediction power goes beyond traditional factors (including mutual fund holding betas) and concentrates in periods with hedge fund leverage reductions. Our results suggest that the leverage choice of hedge funds plays a fundamental role in delegated portfolio management and asset pricing.

Basic asset pricing theory predicts high expected returns are a compensation for risk. However, high expected returns might also represent an anomaly due to frictions or behavioral biases. We propose two complementary, simple-to-use tests to assess whether risk can explain differences in expected returns. We provide general-equilibrium foundations for the tests and show their properties in simulations. The tests take into account any risk disliked by risk-averse individuals, including high-order moments and tail risks. The tests do not rely on the validity of a factor model or other parametric statistical models. Empirically, we find risk cannot explain a large majority of variables predicting differences in expected returns. In particular, value, momentum, operating profitability, and investment appear to be anomalies.

We provide evidence that arbitrageurs face a complex signal processing problem that delays the elimination of mispricing. Using a powerful database containing the precise timing of information announcements, we find that returns to many anomalies are concentrated in the 30 days after announcements and disappear soon thereafter. We find that prices incorporate information more quickly when portfolio rebalancing is less complex. Moreover, anomaly profits vanish more quickly and arbitrageurs trade more quickly as signal processing costs decline. Our evidence shows the timing of anomaly returns yields important insights about their existence, magnitudes, and relation to computational costs.

We find that retail investors' consideration patterns can explain mutual fund fee dispersion via the networked structure of competition they induce between funds. We measure investors' consideration sets by prospectus views on the SEC's EDGAR system and document limited and heterogeneous patterns of consideration, consistent with consumer behaviour studied elsewhere. We formulate a micro-theory model in which heterogeneous and overlapping consideration of investment alternatives grants funds differing degrees of (local) market power. We then calibrate the model, showing its predicted fees are significantly and positively associated with observed mutual fund fees. Our framework explicitly links investors' incomplete information and constrained choice sets to fund-level competition and outcomes, refining the notion of investor search costs.

We analyze the impact of affiliation on survival, performance and flows of affiliated hedge funds, conditional to moderate and extreme market states. We merge Lipper TASS data with ADV information about funds’ legal and operational issues. There are many different types of financial institutions hedge funds can be affiliated to, which makes affiliated funds a highly heterogenous group. We find that certain type of affiliations, that we refer to as the good type, significantly reduce the probability of fund liquidation while other type of affiliations have the opposite effect or have no effect at all, which we refer to as the bad type. Good affiliation types significantly reduce liquidation probabilities even after controlling for conflict of interest variables. Finally, we find a significantly reduced response of flows to the performance of good-affiliated funds compared to bad-affiliated ones. These two facts together support the view that certain types of financial affiliation may have a stabilizing effect in the financial system.

Theoretical models explaining the short-duration premium typically introduce additional risk factors for short-duration assets. We provide high-frequency evidence that duration premia associated with revisions of economic growth and interest rate expectations are consistent with these models. However, they fail to explain the pronounced short-duration premium at earnings announcements. We show that the trading activity of sentiment-driven investors raises prices of long-duration stocks, which lowers their expected returns and results in the short-duration premium. Long-duration stocks have the lowest institutional ownership, exhibit the largest earnings forecast errors, and are most overvalued. This overvaluation is consequently corrected at earnings announcements.

I show that investors’ beliefs play an important role in the valuation of private equity (PE) funds. According to finance principles, the value of PE equals the expected cash flows discounted for time and risk. Thus, pricing errors can originate from an incorrect stochastic discount factor (SDF) or from a discrepancy between investors’ beliefs and the true probability distribution of cash flows. I propose an estimation method to back out investors’ beliefs from funds’ cash flows. I validate recovered beliefs using investors’ sentiment surveys. I find that investors’ overoptimism about the public market and pessimism about the PE industry, rather than SDF misspecification, offers a potential explanation for the outperformance of PE funds.

We document the existence of commonality in short selling loan fees, and we show that the common component of loan fees explains a high degree of their variation. Stocks with high loan fees tend to exhibit high sensitivity to the common component. The common component is highly correlated with several asset pricing factors, suggesting loan fee commonality is associated with consequential states. Loan fee commonality is an important limit to arbitrage, as evidenced by a strong relationship between loan fee commonality and both low future returns and decreased price efficiency. Loan demand is the primary driver of the observed commonality.

The extant literature concludes that binding short-sale constraints restrict negative information revelation. We theoretically refute this conclusion. In our model, some long institutions have greater incentives to acquire information than short sellers, thus revealing information through their sales, and short-sale constraints do not constrain sales. We empirically support our theory predictions and mechanism. First, sufficiently large informed institutional sales eliminate the negative return predictability of short-sale constraints in the cross section and time series. Second, informed institutions lead those less informed ones in sales, implying that all long institutions join forces in impounding negative information, while short-sale constraints and short sales only rise following informed institutional sales. Finally, we support sellers' information advantage in cases where information can be concretely identified. Our results highlight the importance of sales rather than short sales for understanding market inefficiency.

We examine the value of skilled foreign labor for hedge funds by leveraging on two natural experiments. We find that hedge funds that secure more H-1B visas in random lotteries deliver higher alphas, Sharpe ratios, and information ratios. Moreover, an unexpected reduction in the H-1B quota undermined the performance of hedge funds that were dependent on H-1B workers. The superior performance of funds with high H-1B allocations can be attributed to highly-educated STEM majors operating systematic strategies. Notwithstanding the valuable skills that foreign workers possess, racial and ethnic homophily induces some fund managers to eschew foreign labor.

This paper examines the voting behavior of women-led mutual funds. We document

four main stylized facts. First, women-led mutual funds are more likely to support environmental and social (ES) proposals, consistent with stronger pro-social preferences. Their voting support is even more pronounced for ES proposals explicitly related to ES risks, consistent with a greater risk aversion. Within ES proposals, they support more environmental proposals and those related to diversity. Second, women-led mutual funds are more likely to vote with management in firms headed by female CEOs, consistent with the existence of an in-group bias. Third, women-led mutual funds are more likely to support female candidates in director elections. This support is even stronger when there is a shortage of female directors, consistent with women-led mutual funds promoting gender diversity in the boardroom. Fourth, women-led mutual funds are not more likely to follow ISS recommendations than other funds. Our results suggest that gender differences in fund management teams influence their voting behavior.

We investigate the impact of Exchange-Traded Funds (ETFs) on the comovements of their constituent securities using a novel identification that exploits the switch from synthetic to physical replication of a large French ETF. After the switch, constituent stocks experience greater commonality, in both returns and liquidity. For both the full sample of ETF constituents and the least liquid ETF constituents, a larger part of the variation in individual stock returns or liquidity is explained by market-wide variations. We present evidence that ETF arbitrage is the transmission mechanism of the comovements. Moreover, we show that the comovements do not appear excessive.

Asset managers play a dual role by simultaneously managing funds and increasingly providing investment model recommendations to third-party financial advisors. Using a novel data set on recommendations by ETF issuers and strategists, we show that the $4.8 trillion recommendation market has a substantial impact on ETF flows. Model providers recommend their affiliated ETFs more frequently. These funds tend to have higher fees and lower performance than recommended unaffiliated ETFs. In addition, investors following the recommendations exhibit weaker sensitivity to funds’ returns. We fail to find evidence that recommendations are driven by private information about the future performance of affiliated funds.

Minority operated hedge funds deliver higher alphas, Sharpe ratios, and information ratios than do non-minority operated hedge funds. Moreover, minority fund managers attended more selective schools and are more likely to hold post-graduate degrees. Yet, minority managers raise less start-up capital and attract lower investor flows. Racial homophily fuels investors' appetite for non-minority funds. To address endogeneity, we leverage on an event study of minority manager fund transitions and an instrumental variable analysis that exploits racial imprinting during childhood. Our results extend to actively managed equity mutual funds and suggest that racial minorities face significant taste-based discrimination in asset management.

This paper investigates to what extent proxy advice allows funds to vote as if they were informed. A fund’s vote is classified as “informed” if the fund visited and downloaded information about a company from the SEC’s Edgar website before voting. A fund’s proxy advisor is identified from the format of the regulatory filing of its votes. Our main finding, for the period 2004-2017, is that proxy advice did not always help funds vote as if they were informed. While advice from Glass Lewis generally moved funds to vote in the same direction as information, advice from ISS more often than not pushed them to vote in the opposite direction. We rule out several potential confounds using a battery of fixed effects, instrumental variables, and targeted tests. We also show that ISS recommendations distorted votes toward policies favored by socially responsible investors (SRI), and conjecture that ISS may have slanted its recommendations in response to pressure from SRI activists that wanted to influence the votes of ISS’s robo-voting customers.

This paper demonstrates that intense discussion on the Reddit social media platform reduces stock price informativeness. Increased social discussion reduces stock liquidity, especially during earnings announcements. Earnings response coefficients increase with discussion intensity, suggesting reduced price informativeness prior to announcements. Social discussion results in a delayed price correction of up to two months of well-documented anomalies; a corresponding trading strategy earns about 1% monthly. Traditional media do not generate similar effects, while use of emojis intensifies it. The findings suggest reduced production of fundamental-value-relevant information in the presence of intense social discussion, highlighting the importance of understanding the emergence of such discussion.

We hypothesize that when price correction requires more capital than any one investor can provide, institutions coordinate trading via crowd-sourcing in the media. When the crowd reaches a consensus, synchronized trading occurs, prices are corrected, and anomaly returns result. We use over one million Wall Street Journal articles from 1980 to 2020 to develop a novel textual measure of institutional investors making predictions in the media (InstPred). We show that (i) both value and momentum anomaly returns are 34% to 63% larger when InstPred is higher; (ii) these effects are driven by stocks whose institutional investors are highly cited in WSJ articles; and (iii) institutional investors collectively trade the anomalies more aggressively when InstPred is higher. Our results cannot be explained by existing measures such as document tone.

Rational expectations models of active management make precise predictions about how expectations are formed. We recover a forward-looking distribution of expected abnormal returns (alphas) for active equity mutual funds from analyst ratings. We then contrast analysts' subjective expectations with expectations from a standard rational expectations learning model. Analysts and the rational learner respond similarly to perceived managerial skill and fees, but analysts' expectations do not decrease with fund size. The absence of such decreasing returns to scale in analysts' expectations and their presence in actual fund returns make it difficult to reconcile analysts' expectations with existing rational expectations models of active management.

We find that a significant subset of Hedged Mutual Funds (HMFs) and smart-beta Exchange-Traded Funds (ETFs) tilt their portfolios towards well-known anomaly characteristics, especially on the short side, and that such tilts are highly predictable. Moreover, factor-based HMFs outperform ETFs with corresponding factor tilts, which is driven by short positions and higher factor-related returns. Perversely and in contrast to HMFs, large factor tilt ETFs underperform those with contrary tilts. Overall, our results demonstrate the importance of using portfolio holdings rather than stated objectives for benchmarking factor tilts, and indicate the superior factor replication ability of HMFs over ETFs.