Arguing about performance-based pay is running neck-and-neck with grousing about bonuses, whether they be too big or too small, among popular topics this holiday season. In their upcoming paper, “Give & Take: Incentive Framing in Compensation Contract,” Judi McLean Parks (Washington University in St Louis) and James W. Hesford (Cornell University) test out their hunch that certain compensation packages may be linked to rising fraud, losses from which are currently estimated by the Association of Certified Fraud Examiners to total something like $994 billion annually.
Since compensation packages are considered a central tool in managerial control systems, managerial accounting research has long taken an interest in how compensation packages influence behavior. A primary foundation of such research is agency theory, a model that assumes agent and principal are self-interested, but in divergent goals, that agents will shirk, “if necessary, [with] guile and deceit,” and that principals will attempt to control agents through monitoring or by aligning the interest of the agents with their own. In theory, performance-based compensation systems are one way to accomplish the latter. This may have worked well at GM in the 1980s, when line-workers were put on performance-based pay, so that when GM did well, they did well, but when the agents themselves are reporting the results, such contingency packages may instead encourage financial mismanagement and deceit.
In an effort to supplement empirical studies of performance-based pay, and to include penalty-contingencies, which are actually quite common, McLean Parks and Hesford undertook a controlled study. Rather than the obvious choice of rats as participants, the authors brought in a random sample of students, paying them for solving anagrams under three compensation packages: flat salary, performance-based bonus, and performance-based penalty. Each student was given a package with instructions, a “high-quality attractive pen” (keep your eye on these), and self-evaluation forms. Once they turned in the self-scored performance sheets, they threw away their actual work, allowing plenty of opportunity for fraud.
Basic results: those receiving flat salaries were the most honest in their reporting, those on bonus-contingent schedules were less honest, and those on penalty-contingent schedules were the least honest. Even worse, when no ethics statement was signed, those on penalty-contingent pay were three times as likely as those on salary, and twice as likely as those on bonus-contingent pay, to steal those attractive pens.
The authors unearthed a concern about the use of ethics statements, such as the attestations all CEOs must sign under Sarbanes/Oxley. Although, overall, 46% of those who did not sign such statements stole their pens, and only 29% of those who did sign statements did not “misappropriate assets,” the details are more complicated. Those facing performance-based penalties were more likely to misrepresent their performance if they had signed such statements than if they had not. The authors suspect that the existence of the statements themselves suggested to the agents that the principals were weak on apprehending fraud. Why else would they be required to sign such statements?
McLean Parks sums up: “For years we have touted the basic mantra of pay for performance because that's the way you get the best performance. Maybe you get the best performance reported, but what's the underlying performance?"
Not really in the holiday spirit, but you can read the full study, still under review, here: