This paper proposes a novel measure of the reliance on generative AI of investment companies and utilizes it to study the adoption and implications of generative AI tools in the asset management industry, particularly hedge fund companies. We document a sharp increase in the use of generative AI by hedge fund companies after ChatGPT was introduced in 2022. In a difference-in-differences test, we find that hedge fund companies adopting generative AI produce superior raw and risk-adjusted returns relative to nonadopters, with a gain of 3 to 5% in annualized abnormal returns. We further identify this effect by exploiting ChatGPT outages as exogenous shocks. The outperformance originates from investment in AI talent, and more from firm policy and performance information than from macroeconomic information. Unlike hedge funds, non-hedge fund companies do not produce significant returns from their adoption. Large and more active hedge fund companies adopt the technology early and achieve higher returns than others, indicating that utilizing generative AI effectively as an investment tool may require a combination of other resources, such as data and expertise. Overall, our findings suggest that generative AI may exacerbate existing disparities among investors rather than mitigate them, further widening the gap between market participants.

The decentralized nature of blockchain markets has given rise to a complex and highly heterogeneous market structure, gaining increasing importance as traditional and decentralized finance become more interconnected. This paper introduces the DeFi intermediation chain and provides theoretical and empirical evidence for private information as a key determinant of intermediation rents. We propose a repeated bargaining model that predicts that profit share of Ethereum market participants is positively correlated with their private information, and employ a novel instrumental variable approach to show that a 1% increase in the value of intermediaries’ private information leads to a 1.4% increase in their profit share.

This study leverages earnings conference call transcripts and the FinBERT machine learning model to measure greenwashing (GW) intensity across a broad sample of U.S. public firms from 2005 to 2021. We document an economy-wide increase in GW intensity following the 2015 Paris Agreement, with a significant rise in GW among fossil fuel and stranded asset industries. Higher GW intensity is linked to more future environmental incidents, EPA enforcement actions, and higher carbon emissions, but not to increased green innovation. GW is associated with lower cumulative abnormal stock returns post-earnings calls and poorer future operating performance, especially in firms with greater information asymmetry and weaker institutional monitoring. GW firms receive higher future environmental ratings, face lower forced CEO turnover, exhibit reduced CEO pay-for-performance sensitivity, and are more likely to link CEO pay to corporate environmental performance. Additionally, these firms show reduced risk-taking behaviors. Our findings suggest an agency motivation for GW, where managers engage in GW to enhance their job security and compensation at the expense of shareholders.

Decomposing inflation into core and non-core components (e.g., energy) sheds new light on the nature of inflation risk and risk premia. While stocks have insignificant exposure to headline inflation in the U.S., their core inflation betas are negative and energy betas are positive. Conventional inflation hedges such as currencies and commodities only hedge against energy inflation risk but not the core. These hedging properties are reflected in the prices of inflation risks: only core inflation carries a negative risk premium and its magnitude is consistent both within and across asset classes, whereas the price of energy inflation risk is indistinguishable from zero. The relative contribution of core and energy inflation varies over time, which helps explain why the correlation between stock and bond returns appears to switch sign in the data. We develop a two-sector New Keynesian model to account for these facts.

This study explores the risk preferences of Large Language Models (LLMs) and how the process of aligning them with human ethical standards influences their economic decision-making. By analyzing 30 LLMs, we uncover a broad range of inherent risk profiles ranging from risk-averse to risk-seeking. We then explore how different types of AI alignment, a process that ensures models act according to human values and that focuses on harmlessness, helpfulness, and honesty, alter these base risk preferences. Alignment significantly shifts LLMs towards risk aversion, with models that incorporate all three ethical dimensions exhibiting the most conservative investment behavior. Replicating a prior study that used LLMs to predict corporate investments from company earnings call transcripts, we demonstrate that although some alignment can improve the accuracy of investment forecasts, excessive alignment results in overly cautious predictions. These findings suggest that deploying excessively aligned LLMs in financial decision-making could lead to severe underinvestment. We underline the need for a nuanced approach that carefully balances the degree of ethical alignment with the specific requirements of economic domains when leveraging LLMs within finance.

The cryptocurrency market is connected to the traditional financial market through reserve-backed stablecoins. A one standard deviation ($330 million) increase in the issuance of major stablecoins (Tether and USD Coin) on a given day results in an 11% increase in the commercial paper issuance quantity and an 18 basis point decrease in the commercial paper yield the following day. This shows that the exponential growth of stablecoins created an excess demand for short-term money-like safe assets. I also study the fiat cryptocurrency market’s effect on the financial market.

Sustainable investing in financial markets can reduce carbon emissions if it lowers the cost of capital of green investments. We show that the cost of capital perceived by greener firms was close to that of browner firms until 2016 and then fell disproportionately after the post 2016 surge in sustainable investing. The average post-2016 difference between brown and green firms is roughly 1 percentage point. Within firms, we find that some of the largest energy and utility firms have started to apply a lower perceived cost of capital to their greener divisions since sustainable investment surged. Firms facing a higher relative cost of green capital in their sector pledge to reduce emissions by more. Our measures of the perceived cost of capital are hand-collected and associated with firm-level realized returns to capital and long-run investment. Taken together, the findings support the view that sustainable investing is associated with a reallocation of capital toward greener investments through a cost of capital channel, both across firms and within firms.

We study investors’ perceptions of inflation through the lens of a high-frequency event study and document that they have a stagflationary view of the world. In response to higher-than-expected inflation, investors expect firms' nominal cash flows to remain stagnant while discount rates increase, resulting in lower stock prices. Both the equity risk premium and nominal risk-free yields rise. However, longer-term real yields remain unchanged, and policy-sensitive real yields even decline, with increases in nominal yields offset by larger increases in inflation expectations. Consistent with a stagflationary view in which investors interpret inflation as a marginal cost shock, investors expect firms with low market power to suffer larger declines in cash flows. Cash flow expectations of equity investors are aligned with those of professional earnings analysts, both in the time series and across the market power distribution.

How good is artificial intelligence (AI) at persuading humans to perform personally costly actions that carry a moral valence? We study the effectiveness of phone calls made to persuade delinquent consumer borrowers to repay their debt. Both a regression discontinuity design and a randomized experiment reveal that AI is substantially less able than human callers to get borrowers to repay. Substituting human callers for AI six days into delinquency closes much of the collection gap, but one year later, borrowers initially assigned to AI and then switched to humans have repaid 1% less than borrowers who were called by humans from the beginning. Even accounting for wage costs and assuming zero costs for AI, using AI is less profitable (with the caveat that we do not observe non-wage costs of labor). AI’s lesser ability to handle complex situations and extract payment promises that feel binding may contribute to the performance gap.

We explore an optimal token holding and staking problem for cryptocurrency investors. Our investigation revolves around understanding the tradeoff between staking rewards and the consequent illiquidity that emerges from the distinct structure of blockchain platforms or Decentralized Autonomous Organizations (DAOs). We present comprehensive analytic solutions, which enable us to examine the novel implications stemming from the staking mechanism for trading and staking policies and the dynamics of risk-taking behaviors. Our model provides the insights distinguishing between token investments with staking rewards and conventional investment avenues, such as stocks and commodities.

We show that U.S. firms cut imports by 29.9% when their international suppliers

experience environmental and social (E&S) incidents. These trade cuts are larger for

publicly listed U.S. importers facing high E&S investor pressure and lead to crosscountry supplier reallocation, suggesting that E&S preferences in capital markets

can be privately costly but have real effects in far-flung economies. Larger trade cuts

around the incident result in higher supplier E&S performance in subsequent years,

and in the eventual resumption of trade. Our results highlight the role of investors

in ensuring suppliers’ E&S compliance along global supply chains.

Inflation risk premium is hard to identify in the data, because inflation induced by real shocks and that by nominal shocks carry risk premiums with opposite signs. We show that in the Calvo model of price rigidity, exposure to monetary-policy induced inflation risk is a monotonic function of firm markup. Using markup sorted portfolios around pre-scheduled FOMC announcements, we identify an inflation risk premium from the cross-section of equity returns that supports the Calvo mechanism of price adjustment. We develop a continuous-time Calvo model to guide our empirical exercise and to quantitatively account for the inflation risk premium in the data.

Machine learning algorithms have shown to be remarkably successful tools for predicting asset returns. However, the underlying economic mechanisms behind their performance remain unclear. This paper proposes a model-based dynamic arbitrage trading strategy that combines economic and statistical nonstationarity to demystify this black box. In predicting stock returns based on 153 firm characteristics (anomalies), our strategy ranks anomalies similarly to neural networks in the cross-section. Overall, it accounts for approximately 87.9 bps monthly alphas of the high-minus-low portfolios selected by neural networks in the time series. When unpublished anomalies and microcap stocks are excluded from trading, this strategy can fully explain the performance of neural networks. Our results reveal three economic sources of neural-network performance: a time varying strategy analogous to dynamic arbitrage, a tendency to weight portfolios on unpublished anomalies, and exposure to microcaps.

Confirming the conventional wisdom that China is a top-down economy with policydriven markets, we document a significant pre-Govt return of 42 basis points on Chinese equity over the 48-hour window before the announcement of the top government meetings. Similar to the pre-FOMC drift in U.S. equity, this pre-Govt drift in China is significantly larger than that before the macro announcements, demonstrating the unique importance of top government meetings in China. Explaining the pre-Govt returns, we identify two distinct drivers. Under high market volatility, the heightened uncertainty channel of Hu et al. (2022) dominates and the premium for heightened uncertainty (i.e., the pre-Govt drift) escalates to 91 basis points. Meanwhile, institutional investors over-sell equity as the heightened uncertainty builds up, and then over-buy upon its subsequent resolution before the announcements. Under low market volatility, the pre-Govt drift disappears and we document instead an information channel, where the pre-Govt returns are predictive of the post-Govt returns.

This paper uses textual analysis and plausibly exogenous instruments based on out-of-industry geographic connections and director networks to assess the role of trade associations in forming corporate strategies. Companies are most likely to join trade associations when innovative opportunities have declined, and they are older and larger. Joining associations then induces members to increase profits and markups, improve risk management, find acquisition partners and improve efficiency. To assess mechanisms regarding higher profits, we consider product pricing strategies and high dimensional analysis of market-exclusivity in how firms expand geographically across U.S. states. Overall, we find strong support for the conclusion that associations bring mutually beneficial gains to their members and their industries, and some evidence of an externality in the form of anti-competitive pricing and market-exclusion strategies.

Biodiversity loss as a global concern requires drastic shifts in conservation efforts that carry substantial costs. We investigate how the financial market prices such conservation costs, exploiting the “Green Shield Action” (GSA) — an unexpected regulatory change aimed at preserving biodiversity in national nature reserves in China — as an exogenous shock to local public financing. We document that GSA, while improving biodiversity, augments the yields of municipal corporate bonds by around 24 basis points. The effects are more pronounced for bonds with shorter maturities and for local governments in weak fiscal positions. The increased cost of public capital can be largely attributed to transition pressure resulting from preexisting economic activities within reserves and the growth in local public spending on biodiversity following the reform. Investors show little consideration beyond financial payoffs towards endeavors counteracting biodiversity loss. Our findings also point to the role of investor education and policy interventions in closing the financing gap for biodiversity conservation.

We examine the potential of ChatGPT and other large language models (LLMs) to predict stock market returns using news headlines. ChatGPT scores significantly predict daily stock returns, outperforming traditional methods. A model involving information capacity constraints and LLMs rationalizes this predictability, which strengthens among smaller stocks and following negative news and further predicts LLMs' wide availability will improve market efficiency. Only advanced LLMs accurately interpret hard-to-read news and deliver higher Sharpe ratios, while basic models cannot accurately forecast returns, suggesting return forecasting is an emerging capacity of bigger LLMs. We introduce an interpretability technique to evaluate LLMs' reasoning.

This paper examines the role of investor learning in facilitating price dis- covery following important macroeconomic announcements. Leveraging a unique feature of unscheduled macroeconomic news releasing outside of regular trading hours of Chinese stock markets, we investigate the impact of investor learning, in the absence of trading, on the price discovery process. We show that investors’ overnight learning activities on social media during the non-trading period – from the release of macro news to the market opening time – significantly enhance the price discovery once trading resumes. This overnight learning helps level the information playing field among heterogeneous investors, leading to reduced information asymmetry at market opening.

Recent trends show US firms reducing their investments in scientific research while relying increasingly on scientific breakthroughs for patent development. This shift raises a critical question: How do firms access scientific ideas to fuel their innovations? Our study examines the role of top decision-makers in firms, specifically Scientific Directors (SciDs), defined as outside directors with scientific expertise, in bridging the gap between scientific research and patent development. These SciDs are increasingly prevalent, especially in firms whose patents depend heavily on basic scientific research. We find a positive relationship between the influence of SciDs' publications on a firm’s innovation activity and the quality of its patents. Utilizing deep learning techniques, we categorize patents based on the SciDs' areas of expertise, revealing that firms produce high-quality patents in fields where SciDs are actively conducting research. We also provide evidence that SciDs contribute to developing a firm’s innovation talent by tapping their professional networks to help connect top-tier inventors. To address endogeneity concerns, we explore the impacts of scientific breakthroughs due to the Human Genome Project. We observe an increase in both the numbers and economic benefits of SciDs at pharmaceutical firms, highlighting their pivotal role in transforming basic science into patented innovations.

We analyze 18 quadrillion models for the joint pricing of corporate bond and stock returns. Only a handful of factors, behavioural and nontradable, are robust sources of priced risk. Yet, the true latent stochastic discount factor is dense in the space of observable factors. A Bayesian Model Averaging Stochastic Discount Factor (BMA-SDF), combining the corporate bond and stock factor zoos, explains risk premia better than all existing models, both in- and out-of-sample. We show that multiple factors are noisy proxies for common underlying sources of risk, and the BMA-SDF aggregates them optimally. The SDF, as well as its conditional mean and volatility, are persistent, track the business cycle and times of heightened economic uncertainty, and predict future asset returns. Finally, we show that stock factors price the credit component of corporate bond excess returns well, while the Treasury component is priced almost exclusively by the bond factors.

I study the optimal integration of humans and technologies in multi-layered decision-making processes. Each layer corrects errors made by previous layers but also introduces new errors. The ratio between these two aspects serves as a one-dimensional quality measure that uniformly ranks all technologies. Optimal integration requires higher quality technologies to be applied later. Furthermore, technologies asymmetrically influence human effort. Initial players specialize in correcting errors, while final players focus on avoiding new errors. When error costs depend on unobservable states, a uniform ranking is not generally possible. Consequently, the introduction of new technologies can significantly disrupt the integration of existing ones.

Major political events, such as Elections and the delivery of the State of the Union (SOTU) address, reveal subsequent course of government policy and set future political agenda in the U.S. Similar to traditional macroeconomic and monetary policy announcements, political announcements are associated with abnormal stock market returns of 50 bp on Election and 30 bp on SOTU days, relative to 3 bp daily average. Political announcement returns increase in adverse times of low economic growth and high aggregate volatility, and exhibit pre-announcement drift. We develop an illustrative model for political announcements and show that channels of risk premium/early resolution of uncertainty and optimal choice of government policy can account for the evidence in the data.

We study how workforce composition and labor cost affect the recovery of financially distressed firms. We use China’s 2008 Labor Contract Law (LCL) as a laboratory, which increases the cost for a firm to lay off its workers. Post LCL, financially distressed firms decelerate the layoff rate of less skilled employees and avoid mass layoffs. The inability to offload low-quality labor is particularly pronounced among distressed firms located in regions with stringent law enforcement, employer-friendly courts, or having more labor unrest. Non-SOE firms suffer more than their SOE counterparts. Following an increase in the cost to restructure their workforce, distressed firms with an ex-ante larger share of less skilled employees experience a higher cost of debt financing and lower ROA growth; they speed up assets sales and cut wages to a greater extent. Consequently, labor-intensive firms, especially those with a lower-quality workforce, have a significantly lower survival rate and are more likely to turn into zombie firms. These firms take longer to recover and suffer from productivity setbacks. The delay in financial recovery leads to a shift in the firm’s workforce composition, exacerbating the departure of highly skilled employees. Our findings highlight the mechanism by which labor costs can hinder a firm’s ability to emerge from financial distress.

The bankruptcy process in many countries is lengthy and cumbersome, which prevents creditors from recovering their claims in distress. Exploiting the staggered introduction of specialized bankruptcy courts appointing professional judges in China, we find that they lead to a 7.6% reduction in the cost of debt financing, with a stronger effect for issuers of higher default risk. To establish the direction of causality, we focus on differences in court enforcement, namely bankruptcy court versus civil court, across otherwise similar bordering counties located within the same province. By analyzing manually collected bankruptcy filings, we uncover the mechanism with reductions in liquidations, shorter bankruptcy proceedings, less government interference, and higher creditor recovery rates. We further exploit the heterogeneous debt enforcement across Chinese prefectures, together with the bankruptcy reform to estimate the elasticity of bond prices to local enforcement. Improvement in court efficiency translates into significant increases in debt capacity, bond maturity, and investment by bond issuers. These results indicate that stronger court enforcement is a necessary precondition in handling the USD 120.2 billion bond defaults.

We test whether forecast bias affects individual investors’ stock trading by combining bias measures from laboratory experiments with administrative trade-level data. On average, subjects exhibit positive forecast bias (extrapolators), while a large minority exhibit negative bias (contrarians). Forecast bias is positively associated with past excess returns of purchased stocks: Extrapolators (contrarians) purchase past winners (losers). Forecast bias is negatively associated with capital gains of sold stocks. Forecast bias explains investor heterogeneity in the relation between market returns and net flows to stocks. Our study shows that forecast bias links past returns to trading decisions for purchases, sales, and net flows.

Currencies in emerging markets with stricter capital controls exhibit lower average returns, unexplained by standard currency risk factors. This relation is pronounced in debtor countries with high foreign currency liability shares. Capital controls mitigate currency risk by preventing depreciation during market turmoil. We propose an intermediary asset pricing model incorporating an occasionally binding credit constraint for borrowing countries. Capital controls lower crises probability and reduce currency crashes. The model replicates the empirical findings and quantifies the financial impact of pecuniary externality. Based on the model, currency risk premia serve as a tool for policy evaluation.

This paper studies how individual attention to investments affects consumption out of capital gains. We leverage granular account-level data of 200,000 active investors from a leading mobile payment platform. The platform connects mutual fund investments with expenditure records of investors. We find that investments that attract more attention elicit a greater consumption response. We first show that the marginal propensity to consume (MPC) out of capital gains is larger for more recent investments and better-performing ones. This heterogeneity in MPCs across investments suggests that selective attention may render money non-fungible for investors. We then exploit two quasi-experiments that introduce exogenous variations in what funds are displayed. First, the platform’s default setting presents the fund that was most recently acquired by investors at the top of their fund holding page. Additionally, mutual funds from the same fund company as the recently invested fund, even if the former were acquired a long time ago, will be grouped together with the recently invested fund in a single block. We show that investors’ consumption response is stronger to investment proceeds of the funds that were from the same fund company as the recently invested fund than funds that were not. Second, we exploit a change in the display setting on the platform that allowed investors to sort their portfolio funds based on holding period performance. We find that after the change, the consumption response to proceeds from top-performing funds relative to worse-performing funds are more different.

Mutual funds face the risk of withdrawals if they perform poorly in the short term, which

encourages manager myopia. We show that fund families can insulate managers from

this funding pressure via compensation tied to long-term fund performance. Managers

with long-horizon contracts are more likely to undertake long-term investments and

outperform their constrained peers. Since long-horizon pay does not shut off the

funding pressure, it simply insulates the manager from it; not all families can offer these

contracts. Long-horizon contracts are more prevalent in families that cater to patient

investors and have more resources to buffer liquidity shocks.

Despite of the regulatory restriction on mortgage refinancing, Chinese households

prepaid an unprecedented amount of mortgage loans between 2021 and 2023, when the

government cut interest rates to combat economic slowdown. Using loan-level data

from a major commercial bank in China, we find that prepayments are most likely to

occur when the gap between households’ own mortgage rate and the current rate

becomes positive. Evidence further suggests that households use precautionary savings

to prepay (rather than refinancing), and prepayment is associated with household

deleverage and consumption reduction. Combining with data on total lending and

UnionPay consumption, we find macro-level evidence that as the national lending rate

decreases, cities with more existing borrowers having a positive gap tend to have

greater prepayment, lending contraction, and consumption reduction, suggesting

counter-productive monetary policy transmission.

We develop a continuous time general equilibrium model with intermediaries at the heart of international financial markets. Global intermediaries bargain with households and extract rents from providing access to foreign claims. By tilting state prices, intermediaries’ market power breaks monetary neutrality and makes international risk- sharing inefficient. Despite having zero net positions, markups charged by intermedi- aries significantly distort international asset prices and exchange rate dynamics and their response to shocks. Our model can reproduce patterns consistent with several well-known exchange rate puzzles, such as deviations from Uncovered and Covered Interest Parity. All equilibrium quantities are derived in closed form, allowing us to pin down the underlying economic mechanisms explicitly.

“Buy Now, Pay Later” (BNPL) is a key innovation in consumer payments in recent years. It enables merchants to engage in price discrimination by providing free finance. Using a randomized experiment at an e-commerce company, we document that BNPL increase sales by 20%, driven primarily by the intensive margin. The merchant is more inclined to approve BNPL when customers have high credit scores or high margin products in their shopping cart. We find that payment defaults inflict moderate costs and the benefits of offering BNPL significantly outweigh the costs for the merchant. Our findings help to explain the surge in popularity of BNPL in e-commerce around the world.

Should ESG-motivated investors screen out non-responsible firms from their portfolio, or should they rather invest in them and engage with the management? This paper evaluates the effectiveness of these exit and voice strategies and uncovers the conditions under which each strategy is most impactful. Using a novel classification framework based on US mutual funds’ portfolio holdings and votes on shareholder proposals, I show that voice funds are generally more effective than exit funds at pushing firms towards more socially responsible behavior. The exit strategy relies on the threat of lower stock prices and is effective only in firms with high CEO wealth-performance sensitivity. Voice funds threaten directors’ reelection and are more effective when elections are approaching. Taken together, my results point to the financial and career concerns of the top management as driving pro-social change when shareholders demand it.

Recent survey evidence suggests that personal experiences have a strong influence on

the individual’s attitude towards time and impatience. Embedding this feature in a general

equilibrium model, we show that even a small fraction of time-inconsistent investors with

time-varying degree of short-termism cause variation in discount rates over time. As a

result, the consumption of time-consistent agents is negatively correlated with discount

rates, and therefore they require a compensation for this extra risk. In a setting with

Epstein-Zin utility and constant risk preferences this risk helps to explain a plethora of

asset pricing puzzles, even when consumption and dividend growth are i.i.d. In particular,

it generates a substantial time-varying risk premium, variance premium, cross-sectional

predictability, upward sloping yield curve, and high trading volume. A unique prediction

of our model is that the forecast error of expected consumption growth, proxied by the

subjective consumption growth mean and standard deviation, can forecast the excess market

return and real risk-free rate, which we show finds support in actual US survey data.

We document that there is a (re)connection between exchange rate movements

and relative changes in aggregate quantities, such as consumption and output growth, once wealth changes are controlled for. After controlling for these wealth effects, the conditional correlation between exchange rates and macroeconomic quantities becomes negative, as predicted by risk-sharing models. We find that relative wealth changes positively correlate with aggregate macroeconomic quantities. We interpret this as an indication that the economic forces driving these patterns are imperfectly shared productivity shocks. These observations motivate our analysis of how technological innovation affects exchange rate movements. We introduce a minimal deviation from the standard endowment economy model of exchange rate: in an economic boom, new firms are created, but they are randomly distributed to a small part of the population. Our calibrated model does a good job replicating key features of the data, specifically, the joint dynamics of exchange rates, stock returns, real output and consumption growth, and trade flows.

Contrary to the prevailing wisdom that green investors willingly accept lower returns for sustainable investment, our analysis of account-level data from a major Indian bank indicates the opposite. We find that investors with a higher proportion of green stocks in their portfolios achieve superior risk-adjusted portfolio returns. To explain this surprising observation, we hypothesize—and empirically verify—that green investments may help investors mitigate detrimental behavioral bias, such as the disposition effect and under-diversification. Alternative mechanisms related to stock selection ability, aggregate demand shocks, and risk mitigation fail to explain green performance. Instead, tests utilizing abnormal temperatures as exogenous shocks support a causal interpretation of our findings. These results suggest a novel behavioral channel for fully understanding the implications of green preferences.

This paper examines how mutual funds’ portfolio holdings respond to environmental regulations. Using county-level ozone nonattainment designations induced by discrete policy changes in the National Ambient Air Quality Standards as a source of exogenous variation in local regulatory stringency, we find that funds underweight (overweight) those polluting stocks whose cash flows covary negatively (positively) with the regulatory shock. Our results are consistent with active portfolio rebalancing in response to expected changes in firm fundamentals due to negative cash flow shocks stemming from the costs of nonattainment regulation. Further analyses in the post-nonattainment period show that stocks with high exposure to nonattainment designations exhibit worse operating performance and increased regulatory compliance costs. The most underweighted of such firms also exhibit worse abnormal stock return performance. Funds that reduce their portfolio exposure to nonattainment designations see an improvement in their investment performance.

In the context of mergers and acquisitions, this paper analyzes a maturity overhang problem that is due to shorter debt maturities creating higher rollover risk. Using bond transaction data, we develop a market-based measure of rollover risk and find that i) rollover risk dampens merger activities at the firm and aggregate levels; ii) acquirers facing higher rollover risk are more sensitive to changes in cash reserves and prefer equity as a payment method over cash; and iii) positive market reactions to cash payment are observed only when firms have low rollover risk. To shed light on our empirical findings, we study a dynamic investment model that underscores the importance of precautionary savings and rollover risk for maturity overhang.

We introduce a new channel through which US money market funds (MMFs) affect the pricing of near-money assets and measured convenience yields. Our theoretical model reveals that MMFs' strategic interactions create frictions that are exacerbated by T-bill market illiquidity. Using instrumental variables, we show that MMFs have an economically significant price impact in the T-bill market. Consistent with strategic behavior, they internalize this price impact when setting repo rates and allocating portfolios. Our evidence suggests that these frictions drive a sizeable part of common measures of T-bill convenience yields. Our results have implications for monetary policy transmission, government debt issuance, and the regulation of MMFs.

We explore after-hours trading (AHT) in U.S. equity markets. We collect a large set of news releases during AHT and document their effect on AHT activity and market quality. There are three major triggers of AHT. After-hours corporate news catches the attention of retail participants. We use a machine learning method on retail trades to construct a new measure of retail trading activity during AHT. Zero-commission platforms facilitate a reaction of retail traders to positive after-hours corporate news that increases after-hours trades and return volatility. Insider sales are more likely to be announced after markets close allowing faster correction of company overvaluations. Index rebalancing and reconstitutions during AHT lead to large liquidity-driven trades contributing to the negative CAPM slope.

We analyze the stock market response to a comprehensive international sample of 1,560 corporate green bond announcements between January 2013 and January 2022. We do not find any significant market response to these green bond announcements. We conduct a battery of tests to check the robustness of this result, including a focus on the power of event study methodology as it relates to event window length. Also, we find that in the year of the green bond announcements, the abnormal operating performance of these announcing firms is significantly negative, which is consistent with the argument that managers of these firms are using the green bond announcements as a cover for their poor business performance.

This paper proposes a new explanation for the observed co-movement in default risk across borrowers. In our equilibrium model, a negative idiosyncratic shock to one borrower reduces its creditworthiness but also makes another borrower a relatively larger player in the economy, increasing its systematic risk. This results in higher debt costs for the latter borrower, tilting its decision towards an earlier default, especially when debt needs to be refinanced more rapidly. We thus identify a novel source of endogenous default risk linkages, which cannot be explained by commonality in fundamentals alone. Given the embedded leverage in equity, this model jointly generates positive excess co-movement in default probability, equity excess returns, and equity return volatility, aligning with our empirical evidence across U.S. industries. These results offer important insights into the interplay between credit and equity markets in a multi-borrower economy.

We document positive and significant returns on long-term U.S. Treasury bonds on

the day before the FOMC announcements and attribute this pre-FOMC drift to the premium for heightened uncertainty. Unlike the pre-FOMC drift in U.S. equity, which is realized mostly on the day of the FOMC announcement, the pre-FOMC drift in long-term bond occurs earlier. On the day before the FOMC announcement, the 10-year bond yield drops by a significant 0.68 bps and contributes importantly to the secular decline in interest rates documented by Hillenbrand (2021). Unique to the day before the FOMC is a severe disconnect between the long- and short-term yields – an indication that the pre-FOMC pricing of long-term bonds is dominated by the risk-premium channel, not the monetary-policy decision on the target rate. We further capture the pre-FOMC heightened uncertainty using the ex-ante Macro Attention Index (MAI) of Fisher et al. (2022). Conditioning on above-median MAI on unemployment rates, the pre-FOMC reduction in 10-year yield increase significantly to 1.50 bps and is predictive of the subsequent pre-FOMC drift in equity. We further find a strong and positive relation between the pre-FOMC reduction in 10-year yield and the ratio of dissent among the FOMC committee.

In this paper, we provide the first analysis of credit flow dynamics in the Chinese market for wealth management products (WMPs) when their information sensitivity changes. Precipitated by bond market volatility, a shift in risk perceptions amplifies WMPs’ sensitivity to performance, triggering outflows when their net asset value falls below the initial value. As WMPs become less money-like, investors flows exhibit a flight-to-cash pattern. Products with lower redemption frequencies experience smaller outflows, consistent with the notion that they are poorer substitute for cash. Banks respond to outflows by imposing redemption gates, which are effective on the day of implementation, and by shifting issuance away from fragile WMPs.

This paper investigates the allocative efficiency of green finance instruments through a general equilibrium model with heterogeneous firms and financial frictions. We emphasize the impact of the timing of financial instruments—'ex-post', such as carbon taxes, versus 'ex-ante', like green credit schemes—on the distribution of dirty capital and its environmental implications. Our study reveals that ex-post instruments inadvertently direct dirty capital towards financially constrained firms with higher utilization, potentially exacerbating economy-wide emission intensity. Such theoretical prediction explains empirical observations of Hartzmark and Shue (2023), indicating such strategies may be counterproductive. Conversely, ex-ante instruments yield beneficial redistributions. The study emphasizes the significance of incorporating the distributive effects of green finance tools into their design and advocates for a general equilibrium viewpoint to evaluate their effectiveness comprehensively, highlighting the pivotal role of instrument timing.

This paper documents the rise of "poison bonds" corporate bonds that allow bond- holders to demand immediate repayment in a change-of-control event. The share of poi- son bonds among new issues has grown substantially in recent years, from below 20% in the 1990s to over 60% since the mid-2000s, predominantly driven by investment-grade issues. We show that a key factor behind this rise is the pervasive shareholder aver- sion to poison pills, leading firms to issue poison bonds as an alternative. Moreover, our analysis suggests that this practice potentially entrenches incumbent managers and destroys shareholder value. Holding a portfolio of firms that remove poison pills but promptly issue poison bonds results in negative abnormal returns of −7.3% per year. Our findings have important implications for the agency theory of debt: (i) more debt may not discipline the management; and (ii) even without financial distress, managerial entrenchment can lead to agency conflicts between shareholders and creditors.

We document that the sectoral composition and marginal buyers of government debt differ notably across jurisdictions and have evolved significantly over time. Focusing on the United States, we estimate the yield elasticity of demand across sectors using instrumental variables constructed from monetary policy surprises. Our estimates point to a 11% increase in the demand by non-central-bank players for a 1 percentage point increase in long-term yields. Hence, a hypothetical reduction in the central bank balance sheet of around $215 billion increases long-term yields by 10 basis points. We find commercial banks, foreign private investors, pension funds, investment funds, and insurance companies to be the sectors whose demand is most sensitive to changes in long-term yields, but to varying degrees. Heterogeneous elasticities imply compositional shifts in the holders of government debt as central banks normalize balance sheets, which has policy implications.

A limit order executes only if the price moves against the order. To mitigate this limit order pick-off risk, investors can monitor the execution process and cancel their unexecuted stale orders in time. We propose an investor-level indicator for the susceptibility to the stale limit order risk using the average time interval from limit order submission and cancellation. The time-to-cancellation is related to cognitive ability, trading experience, and other behavioral biases, such as the disposition effect. Investor-level time-to-cancellation is negatively associated with their limit order performance. Individual investors may earn higher returns if they cancel the stalest limit orders early or if their canceled orders were successfully executed at the market price prevailing at either the time of submission or cancellation.

Using comprehensive auto loan data, we identify a gap in financing terms between Electric Vehicles (EVs) and non-EVs. EVs, compared to their non-electric counterparts in the same make-model or make-model-power category, are financed with higher interest rates, lower loan-to-value ratios, and shorter loan durations. The primary driver of this financing gap is the risk associated with EVs. The rapid and uncertain progress in EV-specific technologies accelerates obsolescence, reducing EVs' resale value and thus increasing the cost associated with loans for these vehicles. Factors such as car buyers' willingness to pay, socioeconomic characteristics, government incentives for EVs, lenders' market power, and macroeconomic conditions play minimal roles in explaining the higher cost of EV loans. Our findings highlight that technological carbon-transition risk is priced in financing terms of green durable assets consumption.

Fixed income funds carry significant duration risk from their use of interest rate derivatives (IRDs). This duration risk is hidden, as funds typically disclose portfolio duration weighted by market values instead of notionals, concealing their true risk. We find substantial variation in IRD duration, both across funds and over time. The primary motives behind funds' use of IRDs are not to hedge interest rate risk or manage flow risk; rather, they are driven by risk-taking and extending duration risk exposure. During our sample period from 2019 to 2023, returns on funds' IRD positions were the major driver of their performance, overshadowing returns on their non-IRD positions. The performance of funds' IRD positions is not associated with manager skill—funds that outperformed due to high IRD duration in early 2020 performed particularly poorly during interest rate hikes in 2022 and 2023.

The common practice to include firm fixed effefcts in corporate finance research may eliminate the explanatory power of important economic factors that are persistent. To illustrate this point, we review the intuitive R&D-patent relation in recent studies and surprisingly find that R&D input does not always positively explain patent output. This missing link can be attributed to the persistence of R&D and patents that causes the between-firm variation to be absorbed by firm fixed effects. We use Hausman-Taylor and advanced machine learning methods to recover or to modify the firm fixed effect models. We find that Hausman-Taylor correction and Machine learning-based models restore the positive R&D-patent relation. These methods thus offer a second opinion for empirical researchers working with explanatory variables that strongly correlate with between-individual unobservables.

We measure firm-level deviation salience (DS) as the return divergence between individual stocks and their peers. We find that the predictive ability of the past month's stock performance for future returns strongly depends on the level of DS. High-DS stocks exhibit short-term reversals with a return spread of -1.30% per month, whereas low-DS stocks display return continuation with a return spread of 1.41% per month. The result is robust after controlling for the effects of size, illiquidity, volatility, and turnover. Our finding is consistent with the story that investors are prone to overreact to salient information but underreact to non-salient information.

A dynamic model featuring multiple job tasks shows the fundamental role of human capital associated with Information and Communications Technology (ICT) in shaping firms’ remote work practices, asset prices, and labor policies. In the model, ICT human capital enables firms to adopt remote work policies by sustaining high labor productivity despite the associated adoption costs. Consequently, firms with high ICT human capital are more likely to adopt the remote work policy, which in turn makes their firm value, labor policies and output less affected by the pandemic shocks. Empirically, consistent with the model’s predictions, we demonstrate that ICT human capital is a key driver of the dynamics of firms' remote work policy. Furthermore, industries with high ICT human capital have higher fractions of remote work adoption and experience smaller declines in asset prices, employment, and output levels during and after the pandemic.

This paper investigates the anticipation and informational impact of monetary policy shocks through the lens of actively managed government bond funds. Analyzing their performance on FOMC announcement days, we identify a stronger “Fed information effect” when fund managers’ expectations align with the Fed’s. In these cases, the signaling effect of monetary policy prevails, prompting economists to adjust forecasts of future macroeconomic fundamentals. Conversely, in cases of misaligned beliefs, the real interest rate aspect dominates, resulting in a negative albeit insignificant relationship between policy shocks and future fundamentals. Consistent with funds’ ability to forecast macroeconomic news, pre-FOMC changes in fund portfolio duration can predict post-FOMC shocks, leading to an average outperformance of funds on announcement days.

Firms invest intermittently, and a significant part of capital formation occurs during ``investment spikes''. Industry-level ``spike waves'', during which growth opportunities in an entire industry surge and a large fraction of firms in the industry generate investment spikes, also occur quite regularly. We define financial flexibility as the capacity to accommodate large shortfalls between a firm's investment needs and cash flows, as is the case when it has an investment spike. We develop an index for financial flexibility (FF) based on which firm-specific variables differentiate firms that generate investment spikes and those that do not during industry spike waves. In out-of-sample tests, our FF Index outperforms five popular financial constrained (FC) indices in predicting investment spikes, as well as regular investment and R&D investment. We use ChatGPT to evaluate the financial constraint status of a large sample of firms based on management statements in 10-Ks and generate new financial constraint measures; however, FF also outperforms these measures. The superior performance of our FF measure suggests that firms' actions are more revealing of their financial status and its determinants than what is disclosed in annual reports. We validate our empirical approach using data simulated in a model adapted from Gao, Whited, and Zhang (2021). As an application, we show that the FF Index predicts the capacity of firms to sustain investment during economic downturns, and again outperforms the FC Indices.

Mining 29,000 accounting ratios for t-statistics over 2.0 leads to cross-sectional return predictability similar to the peer review process. For both methods, about 50% of predictability remains after the original sample periods. Predictors supported by peer-reviewed risk explanations or equilibrium models underperform other predictors post-sample, suggesting peer review systematically mislabels mispricing as risk, though only 20% of predictors are labelled as risk. Data mining generates other features of peer review including the rise in returns as original sample periods end and the speed of post-sample decay. It also uncovers themes like investment, issuance, and accruals---decades before they are published.

We study how increasing whistleblower incentives affects a firm's communication decisions, price informativeness and real efficiency. An informed manager, who can divert cash for private benefit, privately communicates with his employee about project fundamentals and chooses investment. Given her information, the employee maximizes internal alignment and can leak the manager’s message with some noise. Stronger whistleblower incentives lead to more informative leaks, less misconduct and higher price informativeness. However, they can decrease firm value and real efficiency by increasing the manager's manipulation of internal communication. More targeted policies (e.g., mandating more public disclosure) improve both price informativeness and real efficiency.

Mutual funds face risks of dilution from trading costs when investors place purchase or redemption orders. To deal with this risk, the SEC in 2018 started allowing U.S. mutual funds to change their net asset value (NAV) up or down by a prescribed amount in response to abnormally large flows—a practice known as swing pricing. However, no U.S. fund has thus far chosen to adopt this practice. This paper provides evidence that funds can employ an alternative way to change the value of their portfolios in response to flows, namely by changing the valuation of their underlying holdings. We refer to this phenomenon as “silent swing pricing,” as these swings in valuations are not announced and lack transparency, but still effectively achieve the same goal. Focusing on active fixed-income funds from mid-2008 to 2022, we find that a fund’s valuation gap of a particular bond relative to peer funds’ valuations is positively related to that fund’s same-day flows. The sensitivity of valuations to flows is greater when a fund experiences outflows than when it has inflows, and when it holds more illiquid securities. The extent of silent swing pricing is attenuated, however, by return smoothing incentives when funds have poor past performance and fragile investor base. We show this practice has persisted even after the 2018 SEC rule change.

We develop a dynamic structural model to show how an economy's production network affects firms' financial decisions, default risks and contagion. We analytically characterize the equilibrium and derive novel implications for the impacts of network characteristics. Firm leverage and default probabilities increase (decrease) with network concentration in larger (smaller) industries, and increase with network sparsity. Our calibrated model shows that network characteristics have quantitatively significant effects. Expected and unexpected shocks to variables influencing firms' financial decisions have contrasting directional and quantitative effects. We exploit our calibrated model to explore the susceptibility of the U.S. production network to default contagion.

A disproportionately large fraction (70%) of stock momentum reflects return continuation on the same weekday (e.g., Mondays to Mondays), or the same-weekday momentum. Even accounting for partial reversals in other weekdays, the same-weekday momentum still contributes to a significant fraction (20% to 60%) of the momentum effect. This pattern is robust to different size filters, weighing schemes, time periods, and sample cuts. The same-weekday momentum is hard to square with traditional momentum theories based on investor mis-reaction. Instead, we provide direct and novel evidence that links it to within-week seasonality and persistence in institutional trading. Overall, our findings highlight institutional trading as an important driver of the stock momentum.

In dynamic models of asset markets with asymmetric information and endogenous screening, higher originator effort generally leads to more signaling through delay of sale. We test this theoretical prediction using the mortgage market as a laboratory and processing time as a measure of screening. Our findings are threefold: First, and in line with the theory, mortgage processing time and the delay of sale after origination are strongly positively related in the data. Second, processing time is longer for mortgages with higher ex ante credit risk, i.e., observably riskier loans are processed slower. Finally, both processing time and delay of sale are negatively related to conditional mortgage default, indicating that more screening effort leads to unobservably higher quality loans that are also sold with a longer delay.