Our recent paper uses data from Glassdoor to investigate changes in employees’ perceptions of firms and managers during periods of financial misconduct and after the public announcement of misconduct. Employees’ perceptions are critical for a few reasons. First, employee satisfaction is important for firm productivity. Second, employee satisfaction acts as a proxy for a firm’s ability to attract and retain talented human capital; it is reasonable to expect that a public announcement of misconduct will negatively impact employees’ perceptions of a firm. Moreover, because employees are insiders, they may observe nonpublic and value-relevant information about their employers. Thus, employees’ perceptions may help predict financial misconduct.
Glassdoor and Employee Perceptions
Our primary data are drawn from Glassdoor, a large crowd-sourcing company that gathers information on both compensation (wage and non-wage benefits) and perceptions of workplace practices. We use data collected between 2008 and 2016. Individuals who visit the Glassdoor website are asked to take surveys regarding their salary, non-wage benefits, and ratings of their company across several dimensions. The majority of employer ratings follow a one to five scale, although a few questions are binary (e.g., CEO approval – yes/no).
Although one of our concerns is non-random selection into their site – that is, only very happy or very upset employees post their opinions – recent experimental research suggests that providing formal or informal incentives can significantly reduce the potential for bias in data from self-reported reviews. For example, Glassdoor’s “give to get” model requires individuals to contribute information about their company if they want to receive information about other companies’ compensation and ratings.
Misconduct Decreases Employee Satisfaction
In the first part of our paper, we find that the years following the public announcement of misconduct are associated with a 0.32 standard deviation decline in an employee’s overall rating of the company.
Glassdoor also provides a range ofspecific dimensions about firm ratings that are useful to examine in greater detail. These ratings include career opportunities, compensation and benefits, work-life balance, senior management, and culture and values.
We find a systematic negative association between the public announcement of misconduct and each specific dimension. However, our effects are the strongest for employee ratings of career progression, benefits, and leadership. These ratings drop by approximately 0.35 to 0.40 standard deviation after the announcement of misconduct. The negative association with benefits is consistent with misconduct affecting a firm’s ability to provide the same level of compensation packages, potentially because of a decline in credit supply. This finding is in line with past research showing that corporate default is associated with a large decline in employee compensation and career prospects. Finally, the decline in employee ratings of leadership is consistent with the fact that employees attribute the misdeed to the leader.
How Misconduct Hurts Employee Satisfaction
There are at least three possible explanations for the decline in employee satisfaction following the announcement of misconduct. First, employee satisfaction may decrease because of the short-term litigation costs and the related stock return drop. Penalties and litigation settlement costs following the public disclosure of misconduct lead to an immediate negative stock return and increase in financial distress, which may cause employees to be concerned about firm performance.
However, our analysis shows that although employees adjust their ratings downward right after the announcement of misconduct, the magnitude of this adjustment is small and their ratings continue to decrease afterwards. This suggests that changes in employee ratings do not mainly reflect the short-term litigation cost or related stock return drop. Moreover, we also find that employee ratings are not positively associated with yearly abnormal stock return, which implies that employees do not adjust their ratings mainly based on employers’ stock return.
Second, employee satisfaction may also decrease due to reduced compensation and higher risk of layoff. But our analysis shows that an announcement of misconduct does not have a long-term effect on salaries. This outcome indicates that decreased compensation only affects employee ratings in the short-term, if there are any effects at all.
The last possible explanation is long-term reputation damage. Companies care about how the public and the media perceive their brands, which represents an important subset of intangible capital. Firm reputation is damaged after the public disclosure of financial misconduct, and this damage would adversely affect a firm’s long-term performance. Damaged reputation may also be the reason we find much stronger negative effects for employees with low education. Given that these lower-skilled workers have less ability to differentiate themselves in the marketplace, the announcement of financial misconduct may further undermine their bargaining power in the labor market because working at a company that engaged in misconduct is a negative signal.
Based on the above evidence, litigation cost, stock price declines, and compensation/layoff channels may affect changes in short-term employee satisfaction after the disclosure of misconduct. However, reputation damage is likely to be the primary driving force in long-term employee ratings.
Employees as Insiders
We also considered whether financial misconduct has a direct effect on employee ratings during the years in which it is being committed. Green et al. document that rank-and-file employees possess valuable information. They find that firms experiencing improvements in employer ratings significantly outperform firms with declines. Employees likely have private information during the misconduct period and adjust their ratings at that time.
We find that employees’ overall rating drops 0.20 standard deviations during the misconduct period. Employees lower their ratings in every aspect during the period when the firm is committing misconduct. Employees’ response when the misconduct is being committed is weaker than their response to the misconduct announcement. This is predictable, since only a subset of individuals are likely to have private information about the ongoing misconduct.
Our findings demonstrate that employees who are exposed to more private information while the misconduct is taking place are more likely to report a decrease in satisfaction during the years that misconduct is being committed. First, we find strong negative effects for employees working in the state in which the headquarters is located. During the misconduct period, their overall ratings decrease by about 1.89 standard deviations. In contrast, employees in other states slightly increase their ratings. Employees in the headquarters state are more likely to acquire private information about firms; they are also more likely to have higher positions in the company. Second, we find strong negative effects for older employees. They decrease the overall ratings by about 0.75 standard deviations when firms commit financial misconduct. Similar to employees working in the same state as the headquarters, older employees are more likely to hold higher positions in the company and have private information during the misconduct period.
Finally, given that some employees have private information while firms are committing fraud, we investigate whether employee ratings can help predict misconduct. Indeed, we find that misconduct prediction models that include these employee ratings hold significant predictive power over conventional sources of financial data. Accordingly, our results show how employee rhetoric and perceptions can serve as a signal for the integrity of a company.