Scrutinising financial information on Twitter & detecting misreporting
Scrutinising financial information on Twitter & detecting misreporting
Published 1 October, 2019
There's actually a lot of research in psychology and communications where they find that if you're lying to somebody, you're very intentional about what you say; you dance around the topics you don't want to talk about. We're taking this idea and incorporating it to companies as a whole.
Richard Crowley
Assistant Professor of Accounting
In brief
- Companies manipulate their revenues to make it look like everything is good for shareholders when it is not the case. They are adept at manipulating financial ratios so accounting does not work when trying to identify misreporting.
- Machine Learning algorithms are not perfect and may also pick out companies that are not committing fraud. However, it can be the first line of defence for trying to find companies committing fraud. Investigators then need to dig deeper.
- Companies are relatively transparent on Twitter when disclosing financial information. If users respond to the company disclosing information on Twitter, there is a 40% chance the company will do it again the next quarter. If users don't engage, then that number drops to 8%. This shows that companies seem to really care about engagement in relation to financial reporting on social media.
Assistant Professor Richard Crowley from SMU’s School of Accountancy examines financial accounting using both archival and analytical methods for his research. In this podcast, he discusses how fraud and misreporting can be detected by studying a company’s financial statement using a machine learning technique, and shares how companies disclose financial information on Twitter, and how feedback from investors and others can influence this.
Assistant Professor Richard Crowley from SMU’s School of Accountancy examines financial accounting using both archival and analytical methods for his research. Much of his archival work deals with large sets of unstructured data using high-powered computing algorithms to address accounting issues that are otherwise infeasible to approach.
He has recently written two research articles on using a machine learning technique to assess the content of companies’ disclosures. In this podcast, he discusses how fraud and misreporting can be detected by studying a company’s financial statement using a machine learning technique, and shares how companies disclose financial information on Twitter, and how feedback from investors and others can influence this.