Research Seminars and Workshops
Our research seminars and workshops feature invited speakers from leading institutions, while also facilitating the exchange of research ideas within the department. We aim to foster meaningful dialogue and build a collaborative environment that supports the development of accounting research and practice.
Upcoming Seminars
Details to Come
Past Workshops
CEO Pay and Customized Performance Measures: Evidence from the First Inline XBRL Based Executive Compensation Disclosure
Speakers: James Gong and Shuqing Huang, California State University, Fullerton
Friday, April 10, 2026 | 10:00–11:00 AM
Location: Zoom
In 2022, the Securities Exchange Commission (SEC) issued the pay-versus-performance (PvP) disclosure rule that requires public companies to use Inline XBRL (iXBRL) to "tag" executive compensation and company performance data in the PvP table in their annual proxy statements. In addition, the same rule requires companies to disclose a Company Selected Measure (CSM), the primary performance measure they use to determine CEO pay. This allows researchers to have access to machine readable data on all Compustat companies and study the pay-performance relationship with company specific performance measures instead of universal measures like Return on Assets (ROA) and Total Shareholder Return (TSR). We find a significantly positive relationship between changes in the CSM and changes in two measures of executive pay: total pay reported in the summary compensation table (SCT pay) and compensation actually paid (CAP), consistent with the pay for performance claims made by nearly all companies. We then split CSM into GAAP and non-GAAP measures. For CAP, the pay-performance association is concentrated in firms using non-GAAP CSMs; GAAP-based CSMs show no statistically significant association with CAP. In contrast, SCT is positively associated with both GAAP and non-GAAP measures. This divergence indicates that non-GAAP measures are associated with the market value of equity awards, while both GAAP and non-GAAP measures are associated with the grant date value of equity awards. Our study provides new insights into the debate on pay for performance.
AI-Generated Earnings Summaries and Information Processing
Speakers: Elizabeth Blankespoor, University of Washington
Wednesday, April 22, 2026 | 10:30–11:30 AM
Location: SGMH 4270
We examine how AI-generated earnings summaries affect retail investors' information processing and trading decisions. We use the staggered introduction of AI-generated summaries on StockTwits to compare user behavior with and without the summary. We find that AI summaries reduce users' browsing and engagement with user-generated content, while users' own posts converge towards the summary's framing. These effects are concentrated among users with the most prior engagement with the firm. Turning to capital markets, retail trading volume increases after summaries appear, and post-summary retail order flow becomes significantly more predictive of future returns, but these informativeness gains vary across settings. Our findings suggest that AI-generated summaries can improve retail investors' information processing of and trading around earnings news.
Harnessing Large Language Models for Core Earnings Measurement
Speakers: Matthew Shaffer, Associate Professor, University of Virginia – Darden School of Business
Time: Wednesday, March 4, 2026 | 10:30 AM
Location: SGMH 4270
Estimating core earnings—the persistent profitability from a firm’s main business activities—is fundamental to financial analysis and valuation, but it has become increasingly difficult as financial disclosures have grown longer and more complex and non-recurring items flowing through net income have proliferated. The task has historically resisted automation, as it requires integrating unstructured textual and narrative information across financial reports, and context-specific judgments. We study whether and how large language models (LLMs) can aid in this task, by developing four strategies for prompting LLMs to estimate core earnings using 10-K filings, and systematically evaluating the results. In our baseline approach, in which we provide a leading LLM with a precise definition of core earnings but no further procedural scaffolding, the model fails systematically, in ways that suggest its outputs were influenced by related but distinct financial analyses present in the training data. A leading reasoning model does not fully remedy the problem. However, harnessing the model in a workflow that leverages its language-understanding capabilities while constraining scope for such biases yields core earnings measures that are highly performant on standard tests, on par or better than available alternatives. Out-of-sample tests for lookahead bias indicate that textual understanding, not memorization, drives its performance. We show how LLMs can excel and fail in tasks of this type, and suggest avenues for improvement.
Local-Thinking Bias
Speakers: Charles Lee, Stanford University and University of Washington
Friday, May 9, 2025 | 8:45–9:45 AM
Location: SGMH 1308
Local-thinking bias, wherein agents overweight information that comes readily to mind, is a prominent finding in cognitive psychology. In this study, we investigate local-thinking bias in the context of sell-side analysts and measure each analyst’s “local” information as news stemming from their individual coverage portfolio. Tests examining multiple analysts forecasting on the same focal firm at the same time show that individual analysts overweight idiosyncratic local news and underweight news from economically linked firms that are not in their coverage portfolios. Market prices track the analyst bias from local news, leading to predictable and economically significant return reversal patterns in the future. A trading strategy that adjusts for analysts’ biases earns meaningful abnormal returns. We discuss the implications of these findings for three literatures: (a) cognitive psychology, (b) analyst behavior, and (c) behavioral asset pricing.
Interactive Rulemaking and Corporate Innovation: Evidence from Energy Conservation Standards
Speakers: Xiaoli Tian, Georgetown University
Friday, April 25, 2025 | 10:30 AM - 11:30 AM
Location: SGMH 4270
We examine whether interactive rulemaking, specifically the Department of Energy’s “Process Rule,” influences corporate innovation. The Process Rule is implemented to foster a more interactive rulemaking process by increasing public input and consensus-building in developing Energy Conservation Standards. We find that firms subject to energy standards issued under this more interactive regime file significantly more patents in regulated products covered by the standards. This effect is absent when the energy standards are issued under the traditional regime. Focusing on the interactive regime, the effect of energy standards on innovation is more pronounced when the interactive process is likely more influential in setting the standards. Firms subject to energy standards file a significantly higher percentage of energy-related patents under the interactive regime. A placebo test indicates that the effect of interactive rulemaking on innovation is driven by corporate filers, but not by individuals or governmental agencies. Overall, these findings suggest that interactive rulemaking can encourage corporate innovation by enhancing public participation.
Inside the Blackbox of Firm Environmental Efforts: Evidence from Emissions Reduction Initiatives
Wednesday, March 19, 2025 | 11:45 AM
Location: SGMH 3210
Speaker: Aaron Yoon, Associate Professor, Northwestern University - Kellogg School of Management
Description: Although firms face a tremendous amount of pressure to reduce greenhouse gas emissions, little is known about how they actually achieve reductions. Using detailed project-level data, we show that most U.S. firms reduce emissions by focusing persistently on projects with short payback periods, particularly if they face performance pressure and commit to short decarbonization horizons. Such projects require small amounts of investment and predominantly address energy efficiency. While they yield greater NPVs and generate higher CO2 savings per annum, they save significantly less CO2 over their lifetime. A greater share of such short-term projects predicts higher environmental ratings. However, firms implementing a mix of short- and long-term projects exhibit the most total CO2 savings. Overall, corporate climate efforts so far emphasize cost-effective and short-term outcomes.