Posted on March 17, 2026 by Wendy Frost
What if auditors could predict when errors are more likely to occur in financial reporting? Instead of simply improving techniques for detecting errors, they could focus on how to stop them from happening.
Abigail Zhang Parker
This is the focus of work by Chanyuan (Abigail) Zhang Parker, assistant professor of accounting in the Carlos Alvarez College of Business. Parker recently had a paper, “Predicting Material Misstatements Using Machine Learning,” accepted in The Accounting Review on this topic.
Material misstatements are errors, intentional or unintentional, that materially affect an investor’s judgment regarding a financial report.
Because these types of errors must be disclosed in SEC filings, Parker utilized over 59,000 observations in filings from 2001 to 2014 to develop a machine learning model that could forecast these misstatements. She and her co-authors found that their model outperformed benchmarks for both one-year and two-year predictions.
“Investors and other stakeholders are more interested in forecasting the future rather than looking back to detect the mistakes that already happened,” said Parker, who joined the college’s faculty in 2023. “This model can help stakeholders make informed decisions about a company’s future financial health.”
The model also determined which variables were most predictive. These included auditor-client tenure, comprehensive income, foreign firm status, accrued interest and penalties, percentile rank of non-audit fee and stock return volatility.
“We ran an academic simulation to see how the model could have benefited various stakeholders,” said Parker. “Scenarios included whether investors could use this tool to make money in the market; if managers could use it to prevent internal control weaknesses; and whether it would benefit SEC regulators in identifying which firms to inspect. We found evidence that the model provided a meaningful contribution.”
Parkers sees three distinct categories of artificial intelligence (AI) currently: prediction, automation and generative (like ChatGPT). Each of which she said could be instrumental in improving the efficiency and effectiveness of the accounting profession.
“AI can improve accuracy within the profession as well as be used by accountants to brainstorm and explore cases and topics together,” she said. “The downside is that people worry about workforce implications. But, people also said that Excel was going to replace accountants. It didn’t. In reality, it creates more work for accountants.”
According to Parker, she believes that AI will shift the types of work accountants do from mundane and rule-based tasks to those that require higher-level professional judgment.
Parker noted that there is still much gray area and client specific situations where human judgment and interpretation are irreplaceable by AI. Setting up the proper governance and monitoring systems will be crucial as the world moves toward working more with AI.
Specializing in the intersection of AI and accounting, Parker’s research focuses on the transformative impact of AI on accounting and auditing practices. Current projects that she is exploring include using machine learning to improve the way abnormal accruals are calculated as well as utilizing it for real-time fraud detection.
“I’m proud of my research portfolio, because most of my papers are very practice relevant,” she said. “I am fascinated by this type of work that allows me to solve problems found within industry and improve accounting constructs.”
— Wendy Frost
Share this story:
Read more stories like Dan's in our 2025 Alvarez Alumni Magazine.
