Drew Allen and David Wood, Accounting
In order to ensure accurate financial reporting, deter fraud, and safeguard assets, companies implement internal controls throughout their organization. Internal controls can be something as simple as requiring a manager to sign off on all expenses or having two people present to handle and count cash receipts. Without these controls, companies would have to rely solely on the competency and morality of their employees or use costly and restrictive supervision and monitoring to prevent intentional and unintentional errors. While internal controls serve an important role in reducing many types of risk, controls may also have unintended consequences that can negatively affect a company and its employees.
In our study, we test for an unintended effect of internal controls on a previously unresearched dimension: self-control. According to the theory of ego depletion from the psychology literature, tasks that require an individual to use self-control will diminish an individual’s ability to exercise self-control in a subsequent self-control task. Like a muscle, self-control can become fatigued as it is used, and a period of rest is needed to restore self-control back to full capacity. Due to the fact that internal controls can limit the decision-making rights of employees and can provide feedback to employees in real time (factors which other studies have shown to cause depletion), we hypothesize that internal controls can induce self-control depletion.
If internal controls do in fact result in self-control depletion, such a finding would have several important downstream implications. First, if internal controls leave employees with less remaining self-control capacity, this would mean that employees have less remaining self-control for important tasks at work that require self-control. Self-control is needed to focus the mind, to act objectively and without bias, and in order to control emotions that may inhibit work. Secondly, if internal controls deplete self-control, this would mean that employees have less remaining self-control that would be needed to behave ethically. If this is the case, it would mean that internal controls (which are designed to prevent fraud) could actually contribute to an increased likelihood of fraud and other unethical behaviors. In order to test this theory, we designed an experiment to capture the self-control effects of internal controls. Our experiment began by having participants complete a typing task that either had a data-entry control placed on it or that had no data-entry control on it.1 Participants were told to type as quickly and accurately as possible. After completing this typing task, we had participants complete an anagram task in which they attempted to unscramble anagrams to form words (the anagrams actually had no solution). We measured persistence on this unsolvable anagram task as a measure of self-control (persisting longer indicates more self-control). Thus, by comparing the level of persistence on the second task, we could compare self-control levels between those that had a control placed on the first task and those that had no data-entry control placed on the first task. We predicted that those that were subjected to a data-entry control in the first task would have less self-control and would persist for less time on the anagram test compared to those that were not subjected to any data-entry control.
However, our results showed that there was no statistically significant difference in persistence between those that were subjected to the data-entry control and those that were not subjected to the data-entry control. We made several adjustments to our experiment and tried running modified versions of the experiment two different times, but failed to generate statistically significant results in any iteration of the experiment.2
Even though our results did not match our predictions, there are still valuable insights that may be drawn from them. First, our results indicate that internal controls do not affect self-control capacity. This means that managers can organize their companies in a way that relies on internal controls without the fear of having negative unintended consequences related to self-control loss. Secondly, our results are part of a growing body of work that challenges the basic assumptions of ego depletion. Several papers have been published in recent years calling into question the soundness of depletion theory, and our results add to the discussion by demonstrating that we still may not fully understand the underlying mechanics of self-control failure.
After obtaining our results, I worked with professors David Wood, Scott Summers, and Bill Heninger to write up our results, draft a paper, and submit the paper for review at an accounting journal. The paper is currently under review at a well-regarded accounting journal, and we are hopeful that the paper will be published at some point in the future. The experience of working with a mentor through the entire research process of coming up with an idea, figuring out how out how to test the idea, running an experiment, and interpreting the results was an incredibly informative and educational experience. There is no better way to learn about research than to do research yourself, and for that, I am very grateful for the opportunity I had to complete this project with the help of my ORCA grant.
1 In order to test how different types of internal controls affect self-control depletion, we included three different types of internal controls in the typing task in addition to the control group (a group that had no data-entry control on the typing task). These three different controls included a preventive control, a detective control with immediate feedback, and a detective control with delayed feedback.
2 We made several adjustments to our experiment after running it the first time. In our first adjusted experiment, we made the data-entry controls more obtrusive in the typing task (fewer mistakes were allowed in the typing task). We also decreased the number of anagrams in the anagram task and altered the wording of the instructions for the second task. There was no significant difference in our results in this experiment compared to our first. Our second adjusted experiment replaced the anagram task with a common psychology tool known as the Stroop Task. Even when using this other method to capture the effects of depletion, our results were no different.