ABSTRACT We examine how analysts’ information acquisition and processing differ when analysts have access to AI resources, focusing on investment banks’ AI investments. We propose and test a two‐step framework, which is informed by in‐depth interviews with analysts. First, consistent with AI facilitating automation‐assisted public information processing, we show that AI investments are associated with more timely earnings forecasts following 10‐K filings, particularly after the implementation of iXBRL, which increases the machine readability of filings. Second, we show that analysts reallocate the time and capacity freed by automation toward acquiring and incorporating private information, supported by several sets of evidence: AI investments (1) are associated with higher quality and bolder earnings forecasts, particularly when private information is more important and accessible to analysts; (2) are associated with an expansion of analyst coverage to new firms and industries; and (3) are associated with higher information‐seeking efforts, particularly greater participation in earnings conference calls. Additionally, exploiting the launch of AskResearchGPT at Morgan Stanley, an in‐house generative AI designed for research, we find results consistent with our main analyses. Overall, our study provides insights into the potential for AI to reshape the human value of sell‐side analysts.

We analyze the impact of the 2021 Child Tax Credit (CTC) expansion using detailed transaction data from nearly two million individuals and difference-in-difference analyses around monthly CTC payments. We find that recipients significantly increase their consumption in the week following the payment versus the prior week, with more pronounced effects for lower-income recipients and those with more children. We also find a significant reduction in liquidity constraints for lower-income recipients, with reductions in overdrafts, use of high-interest payday loans, and use of gig work for supplemental income. Building on these initial findings, our analysis further delves into more nuanced, policy-targeted aspects: We observe a greater alleviation of liquidity constraints from monthly CTC payments compared with annual tax refunds. In addition, monthly payments decrease consumption volatility and increase stability in consumption levels. These results inform considerations for payment frequency. Furthermore, we provide insights into the income thresholds for the policy's phase-in and phase-out ranges by identifying the most pronounced consumption effects among the lowest-income recipients-who benefit from full refundability under the plan-and noting an absence of significant consumption changes among higher-income recipients. Our results present the first large-scale, transaction-based, empirical archival evidence of the effects stemming from the 2021 CTC amendments and provide insights for policy-related discussions.

We build a novel comprehensive data set of new product trademarks as an output measure of product development innovation. We show that risk-taking incentives in CEO compensation motivate this type of innovation and that this innovation improves firm performance. Using an exogenous shock to executive compensation, we find that reductions in stock option compensation cause reductions in new product development. We also find that firms undertaking new product development experience increases in future cash flow from operations and return on assets. These findings suggest the importance of product development innovation to firms and new trademarks as a novel innovation measure.

The revised Markets in Financial Instruments Directive (MiFID II) requires the unbundling of research payments from trading execution, fundamentally changing the way in which investors typically pay for analyst research in Europe. We examine the effectiveness of the regulation in changing the link between analyst research and trading, the research-trading link, and the analyst response to this potential change in incentives. Using a difference-in-differences research design, we find that forecast frequency, optimism, and accuracy are less associated with the brokerage trading share after MiFID II, suggesting that MiFID II weakened the link between the brokerage share of trading and analyst research. Following MiFID II, analysts in Europe are less likely than analysts in the United States to continue high forecast frequency, optimism, and accuracy for stocks with high share importance for the analyst's brokerage house. We find similar results throughout for buy/sell recommendations. Overall, our evidence suggests that MiFID II is at least partially successful in unbundling research from execution, and impacts both the trading effects and the production of analyst research.

We examine the extent, tone, and impact of media coverage of employment discrimination violations. For a comprehensive set of discrimination cases brought by both the Equal Employment Opportunity Commission (EEOC) and private litigation, we manually identify news articles covering these cases and find that only 42% of violations receive news coverage. Firms with higher diversity prior to the violation announcement receive greater coverage, consistent with the media viewing violations by these firms as more newsworthy. Media coverage of discrimination appears informative to the market: the market response to violation announcements is significantly more negative for violations with higher coverage. This negative market response is concentrated in violations covered by left-leaning media and appears to reflect differences in dissemination audience rather than variation in coverage tone by outlets of different political leanings. Overall, our results highlight the media’s role in facilitating the investor response to discriminatory practices.

This data guide describes how to obtain the brokerage trading volume data used in Lehmer, Lourie, and Shanthikumar (2022) and Lourie, Shanthikumar, and Yoo (2022), from Bloomberg Terminal. The data is available at the broker-stock-day level, allowing for detailed analyses of brokerage trading volume. This data can be used to better understand the flow of trading volume through different brokerage houses, including examining brokerage trading volume relative to equity analyst research. We provide an overview of the brokerage trading volume data in Bloomberg and explain in detail how to download the brokerage trading volume data, including specific examples.

Using an experimental design, we examine whether CEO race and CEO gender affect investors’ willingness to invest in a company, their forecasts of future performance, and their evaluations of CEO performance, when given narrative and financial information about a firm. We deliberately avoid triggering race and gender stereotypes, in order to strengthen external validity. We collect data from a diverse group of participants and examine whether race- and gender-effects differ based on participant characteristics. We find that participants are significantly less likely to invest in the Black-CEO led company than the white-CEO led company, consistent with race-based implicit biases regarding leadership. Contrary to expectations, we find that participants are more likely to invest in the female-CEO led company than the male-led company. Both of these results are primarily driven by a low willingness to invest in firms led by Black male CEOs. This low willingness to invest in Black male led firms occurs in data collected from a diverse public University and from private historically white institutions but is absent in the data from Historically Black Colleges and Universities. Additional analyses exploiting participant race and gender suggest that both results are driven in part by in-group biases: White participants are more likely to prefer white CEOs, while female participants are more likely to prefer female CEOs.

Using unique new data, we examine whether brokerage trading volume creates a conflict of interest for analysts. We find that earnings forecast optimism is associated with higher brokerage volume, even controlling for forecast and analyst quality, recommendations, and target prices. However, forecast accuracy is also significantly associated with higher volume. When analysts change brokerage houses, they bring trading volume with them, influencing trading volume at the new brokerage. This indicates that analysts drive the volume effects we observe. Consistent with a reward for generating volume, brokerage houses are less likely to demote analysts who generate more volume. Finally, analysts strategically adjust forecast optimism based on expected volume impact. Analysts become more (less) optimistic if their optimistic forecasts in the prior year were more (less) successful at generating volume. However, consistent with higher costs to increasing accuracy, analysts do not update accuracy based on expected volume impact. Overall, our results are consistent with a brokerage trading volume conflict of interest moving analysts towards more optimistic earnings forecasts, despite the volume reward for accuracy.

The revised Markets in Financial Instruments Directive, known as MiFID II, requires the unbundling of research payments from trading execution. Using a difference-in-differences research design, we examine whether this regulation achieved its intended objective. We find that MiFID II weakened the link between the brokerage trading share and analyst research. Forecast frequency, optimism, and accuracy are less likely to be associated with the brokerage trading share after MiFID II. Analysts appear to respond to these changes. Forecast frequency and forecast optimism both decrease after MiFID II among brokers who earned the highest brokerage trading share due to these behaviors before MiFID II. Overall, our evidence suggests that MiFID II is at least partially successful in unbundling research from execution.

Using comment streams on Seeking Alpha articles, we examine whether interacting on social media increases or moderates the extremeness of investors’ opinions. Unlike some findings from political science that show social media increases extremeness of opinions, we find that interaction on Seeking Alpha moderates extremeness. Comments become less extreme over the sequence of comments for a given article, as well as within individual comment sub-threads, and over a single user’s comments for a given article. Extremeness reduction is stronger when the article itself is more moderate, and when more users are self-identified (i.e., not anonymous). Results also suggest that the extremeness reduction triggered by Seeking Alpha interaction has capital market implications. Differences of opinion captured by stock-based measures, abnormal volume and turnover, decrease significantly after the release of Seeking Alpha articles with comments. Our results provide the first evidence of the effect of social media interaction on the updating of individuals’ opinions.

New product development is critical for firms to achieve and maintain growth and performance. We build a novel dataset of 123,545 USPTO trademark registrations by S&P 1500 firms from 1993 to 2011 to study whether and how CEO compensation risk incentives motivate new product development. Using the OECD’s broad definition of innovation that includes new product development, our tests offer evidence on how risk incentives affect innovation of new products. We find that the number of trademarks increases with the fraction of compensation in the form of stock options, the convexity of incentives, and unvested stock options, both in low-patent (non-high-tech) and high-patent (high-tech) industries. Using a revised accounting rule, SFAS 123(R), as an exogenous shock, we find that reductions in stock option compensation cause reductions in trademark creation. Overall, the evidence indicates that CEO risk-taking incentives are important drivers of product development.

ABSTRACT We study how public firm misvaluation affects private peer firm investments. An economic competition hypothesis predicts a negative relation because misvaluation-induced new investment by public firms crowds out investment by private peers that share common input or output markets. An alternative shared sentiment hypothesis predicts a positive relation because private firm stakeholders share in the sentiment associated with misvaluation in public markets. Misvaluation is proxied using both the price-to-fundamental ratio and an exogenous instrument obtained from mutual fund flows. The evidence is consistent with the shared sentiment hypothesis, and robust to alternative treatments for growth opportunities. Private firms finance misvaluation-induced investment primarily internally or externally with debt, not equity. Finally, misvaluation-induced investment increases future return on investment for private firms, in contrast with public firms. Overall, these findings suggest that overvaluation in public markets increases private firm investments and has beneficial effects on private firm investments by relaxing financing constraints. JEL Classifications: G32; M41. Data Availability: Data are available from sources identified in the paper.

We address whether retail investors use SEC filings when making trading decisions. We find that retail investor trading, both buying and selling, is significantly related to EDGAR search for 10-K and 10-Q filings, more so than to Google search. This is true for firms with high or low visibility, firm-days with and without press coverage, and during or excluding earnings-announcement and filing windows. The results of lead-lag and two-stage-least square analyses are consistent with search leading to trade. In addition, the direction of retail investors’ trading is consistent with the direction of earnings changes reported in the downloaded filings, suggesting that retail investor trading direction is influenced by the accounting information they read in the filings. We also find that the significantly positive relation between retail trading and EDGAR search is strongest for the most easily readable 10-K and 10-Q filings. Finally, we find that retail investor trading-predicted returns are higher on days with heavier EDGAR search, consistent with retail investors making more profitable, or at least less loss making, trades when doing more research on EDGAR. Overall, our results provide strong evidence that retail investors use EDGAR filings data in making their trading decisions.

We examine the impact of distance on internet search, and the effect of the "local bias'' in search on the stock market response around earnings announcements. We find significant local bias in search behavior. Motivated by theories explaining local bias, local information advantage, and familiarity bias, we predict and find that firms with higher local bias in search experience higher bid-ask spreads, lower trading volumes, and lower earnings response coefficients at the time of earnings announcements, consistent with non-local investors relying more than locals on public information announcements. Consistent with local information advantage, we find that in the week prior to the announcement, firms with higher local bias have higher bid-ask spreads, higher trading volumes, and returns that are more predictive of the coming earnings surprise. Consistent with familiarity bias, firms with higher local bias in search experience stronger post-earnings announcement drift. We use unique predictions, propensity score matching, and two-stage least squares to identify the effects of local bias separately from the effects of overall visibility. Overall, we show there is significant local bias in search, and that this local bias has a significant impact on the market response around earnings announcements.

The spike in restatements in the early 2000s prompted an important question: Are retail investors able to react to the large volume of restatements, and in particular are they able to differentiate between non-fraud restatements and more serious fraud-related restatements? The latter are associated with more negative announcement-window and post-announcement returns. Using a unique dataset, we directly examine retail investor trading reactions to restatement announcements. We find that retail investors display significantly higher trading volumes during restatementannouncement windows, as well as significantly higher abnormal trading volume for fraud-related restatements than for non-fraud restatements. However, retail investors increase both their buying and selling, and do not sell any more strongly as a percentage of trade for fraudversus non-fraudrelated restatements. We examine the returns following retail investors’ trades and find that retail investors’ share purchases for fraud-related restatements earn significantly negative raw and abnormal returns (e.g., -8% raw returns over three months). This suggests that retail investors’ failure to differentiate between fraudand non-fraud-related restatements may have significant welfare consequences. We further examine the potential role of press coverage and find that retail investors sell more strongly for fraud-related restatements when restatements receive more press coverage. However they continue to buy significantly for fraud-related restatements even when press coverage is high, which leads to subsequent negative returns. Overall, our results suggest that retail investors fail to understand the negative implications of fraud-related restatements.

We introduce trademarks as a new measure of innovation output, and examine the relation between CEO incentives and trademarks in a broad set of industries. Our new dataset contains over 123,545 USPTO trademark registrations by S&P 1500 firms from 1993 to 2011. As compared with patents, trademarks measure innovation over a wider range of industries and focus on the development portion of innovation that culminates more immediately as new products and services. We find that, on average, firms with more new product trademarks have more volatile stock returns, sales, and earnings after relevant controls, consistent with new trademarks being an indicator of risky product development innovation. We find that the fraction of CEO pay in the form of stock options, the convexity of CEO incentives, and the amount of unvested stock options held by the CEO are strongly positively associated with future trademarks. We also examine subsets of industries based on their patent production, and find generally similar results for all levels of patent-intensive industries. Finally, we document a positive relation between changes in stock option compensation around the implementation of SFAS 123(R) and subsequent changes in trademark creation, suggesting that stock option compensation is an important driver of product development innovation.

Why do security analysts issue overly positive recommendations? We propose a novel approach to distinguish strategic motives (e.g., generating small-investor purchases and pleasing management) from nonstrategic motives (genuine overoptimism). We argue that nonstrategic distorters tend to issue both positive recommendations and optimistic forecasts, while strategic distorters speak in two tongues, issuing overly positive recommendations but less optimistic forecasts. We show that the incidence of strategic distortion is large and systematically related to proxies for incentive misalignment. Our two-tongues metric reveals strategic distortion beyond those indicators and provides a new tool for detecting incentives to distort that are hard to identify otherwise.

Prior research on equity analysts focuses almost exclusively on those employed by sell-side investment banks and brokerage houses. Yet investment firms undertake their own buy-side research and their analysts face different stock selection and recommendation incentives than their sell-side peers. We examine the selection and performance of stocks recommended by analysts at a large investment firm relative to those of sell-side analysts from mid-1997 to 2004. We find that the buy-side firm’s analysts issue less optimistic recommendations for stocks with larger market capitalizations and lower return volatility than their sell-side peers, consistent with their facing fewer conflicts of interest and having a preference for liquid stocks. Tests with no controls for these effects indicate that annualized buy-side Strong Buy/Buy recommendations underperform those for sell-side peers by 5.9% using market-adjusted returns and by 3.8% using four-factor model abnormal returns. However, these findings are driven by differences in the stocks recommended and their market capitalization. After controlling for these selection effects, we find no difference in the performance of the buy- and sell-side analysts’ Strong Buy/Buy recommendations.