A recent study demonstrates that the use of artificial intelligence is significantly improving the ... [+]
A study recently published in the Review of Accounting Studies demonstrates that the use of artificial intelligence is significantly improving the quality and efficiency of financial statement audits, as well as displacing audit professionals.
Audits follow a set of auditing standards and are very reliant on the highly-skilled tasks of prediction and anomaly detection. Such settings are where machine learning or artificial intelligence (AI) should flourish.
In a study titled "Is Artificial Intelligence Improving the Audit Process?” researchers analyzed more than 310,000 employee resumes to measure AI investments for the 36 largest U.S. audit firms. The period examined by the study was from 2010 to 2019. The article is authored by Anastassia Fedyk from the University of California at Berkeley, James Hodson of the AI for Good Foundation, Natalia Khimich from Drexel University, and Tatiana Fedyk of the University of San Francisco.
To guide their analyses, the researchers first interviewed audit partners about how their firms used AI on their audits. The study then used detailed resume data to measure audit firms’ investments in AI. Specifically, AI investments were measured by how many AI workers the audit firm was hiring as a percentage of their total workforce. AI investment levels were subsequently linked to measures of audit quality, audit fees, and the extent to which the audit firms were reducing their human workforce.
Overall, the AI workers tended to be male, were relatively young, and held non-accounting degrees like engineering and computer science. Also, the AI employees were clustered in the states of New York, California, and to a lesser degree Washington D.C., Illinois, and Texas.
“Our main result is that when audit firms invest in AI, their audit quality goes up. There are fewer restatements, including material restatements, and fewer SEC investigations related to audits performed by AI-investing firms,” says Anastassia Fedyk. Interestingly, the study found the AI and audit quality link for both Big Four and non-Big Four firms. The link is also strongest in contexts where you would most expect it: the audits of older companies that have more data available and the audits of companies in the retail industry. Many of the audit partners interviewed by the researchers highlighted the retail industry as the most suitable for AI tools.
Fedyk further notes that, “We also observe that this improved audit quality is achieved more efficiently by the auditors. While we cannot directly measure auditor efficiency, we use audit fees with the assumption that more efficient and less costly audits can be priced lower. Greater investments in AI were associated with lower audit fees.”
The study further reports that the audit client’s investments in AI plays a much smaller role with respect to audit quality than the audit firm’s AI investments. “In retrospect, that’s intuitive. Clients invest in AI for other goals like new product creation, and the auditors are the ones who understand the intricacies of the audit process and how technology might help. But heading into the study, we expected client adoptions of AI to play a larger role than what we actually saw,” says Fedyk.
Last, the researchers find that greater levels of AI adoption were followed by reductions in the audit firms’ workforces. This effect was most prominent with audit professionals at the more junior levels of the firms. The article also illustrates that audit fees earned per employee have a positive relation with AI investment.
“Overall, I would say that our study has a very positive takeaway. Investing in technology can have real benefits for firms. This is not always the case. In a previous study, myself and a fellow researcher documented that hiring employees with popular and over-hyped skills, like IT in the early 2000s and Data Analysis in the 2010s, can increase companies’ immediate valuations but then be detrimental in the long term. But if the context is right, and the technology fits, then the benefits are there. This seems to be the case for AI in the audit setting,” concludes Fedyk.
