This paper studies whether employee reviews on the social media platform, Glassdoor, are useful for predicting accounting fraud committed by their employers. I find that the rating of the employer overall and the average value of subsector ratings have predictive ability for accounting fraud incremental to F-score and other market participants. Similar inferences apply to the rating of employers’ culture values and the textual assessment of employers’ social and governance quality. In addition, I find that employee reviews are most useful for predicting fraud one year before fraud revelation. Within the fraud sample, I show that above ratings decrease and the textual assessment of social and governance quality deteriorates in the two years before fraud revelation. The declines are more...[ Read more ]

This paper studies whether employee reviews on the social media platform, Glassdoor, are useful for predicting accounting fraud committed by their employers. I find that the rating of the employer overall and the average value of subsector ratings have predictive ability for accounting fraud incremental to F-score and other market participants. Similar inferences apply to the rating of employers’ culture values and the textual assessment of employers’ social and governance quality. In addition, I find that employee reviews are most useful for predicting fraud one year before fraud revelation. Within the fraud sample, I show that above ratings decrease and the textual assessment of social and governance quality deteriorates in the two years before fraud revelation. The declines are more pronounced when fraud firms’ business operations are less complex, internal controls are poorer, reviewing employees are not managers and reviewing employees have already left the firm. The study documents an important channel through which fraud-related information is revealed to the public and the findings have implications for investors and regulators.