When I was CEO of the Monsanto Company, I occasionally read business academic journals, but sometimes lost interest because they were often written in opaque “journal speak” or else the subject matter didn’t seem particularly relevant to real-life business situations.
Three years ago, as Executive in Residence at the Olin Business School at Washington University in St. Louis, I initiated “The Olin Award,” designed to showcase the excellent research by the Olin faculty and to recognize the faculty’s journal articles that have the greatest potential to affect business results.
A distinguished panel of senior executives voted on the candidate’s papers, which were accompanied by a short essay as to why the research had potential to impact either as written, or if very basic research, how later authors might use it as input for their own practical applications. The winner receives a $10,000 cash award and a handsome engraved crystal trophy at a recognition dinner.
The results have been very positive based on comments from the judges as well as the increasing number of faculty who ask opinions about the potential relevance of an intended study.
This year’s winner is Judi McLean Parks, professor of Organizational Behavior at the Olin School. Her paper, with co-author James Hesford, titled “Winners and Losers: Rethinking Performance Based Pay,” greatly impressed the judges with its methodology and unexpected conclusions. And even though a small, controlled experiment, its potential to directionally, at least, encourage practitioners to reconsider compensation schemes is a timely topic for Directorship readers.
Winners and Losers: Rethinking Performance-Based Pay
By Judi McLean Parks and James Hesford
It is well established in the academic literature—and certainly in practice—that properly aligned performance incentives can have a positive effect on business results. What has been given little academic attention, however, is the potential cost related to the unintended negative consequences of incentive compensation. These can come both in the form of accounting fraud – overstating actual results to achieve an award (think Enron et al) – and in property theft from disgruntled employees. Warren Buffet and Alan Greenspan – among others – have warned about the potential for executives to “cook the books” to achieve performance awards.
This paper seeks to quantify those negatives through a controlled behavioral study with students from an East Coast university. It was designed to simulate, as best one can in studies of this kind, potential actual behavior in a work setting. The advantage of this methodology is that it carefully controls virtually all aspects of the behavioral setting, eliminating confounding factors. While the outcome of a controlled study is only indicative of potential behavior, these results were so robust that they offer strong support for explaining the fraudulent behaviors observed in real business settings.
Recruited to Perform: The Procedure
Individuals were recruited to perform measurable tasks with the potential to earn significant (for students) compensation. One of these tasks was to unravel 15 complicated anagrams (e.g., uioettms, drinpomlea, and gliebapaarn for titmouse, palindrome, and plea bargain). Each person was given a high-quality pen to use and a sealed envelope with solution key and scoring sheet. Participants were told they would self-score performance at the end, turn in their score along with the pens, and place worksheets in designated bins. The obvious implication for participants was that they would be on the honor system regarding their scores.
Participants were assigned to one of three forms of compensation. In the first group (“Salaried”) each participant received $30 for their efforts in the anagram task, regardless of score. The second group of participants (“Performance Bonus”) each earned $2 for each correct answer, but received no base pay. Participants in the third group (“Performance Penalty”) were each given $30 at the outset, but had to return $2 for each anagram they failed to solve.
To obtain further behavioral insights, half the participants in each group were given an “ethics primer,” which asked that they check a box on the score sheet attesting to the accuracy of their results.
The researchers had a coding system for the worksheets and pens to compare actual results to reported results, and to identify which participants returned the high-quality pens as requested.
The table (opposite) gives the outcomes, scaled to 100, in terms of actual performance (anagrams correctly solved), reported performance (on their scoring sheets) and fraud values (how much they claimed, over and above their actual performance), with comparisons to the salary (no incentive) participants.
Incentive Plans Outperform Salary
There were a number of noteworthy findings. First, both incentive plans outperformed salary on an actual basis. As the underlying rationale of incentive schemes predicts, there are performance gains. Yet our second observation is cautionary: incentive plans – penalty especially so – induced fraudulent behaviors. Third, the compensation form (salary, bonus or penalty) also affected whether or not the participants stole the pens (which they were asked to return). While being primed to think about ethical concerns had no effect on cheating, it did suppress pen theft, as shown in the table below.
Hazards of Incentive Comp
From these results, it appeared that incentives worked. Participants produced better actual scores (performance) but at a significant cost (cheating), especially among those who had penalty-based pay (claw-backs, where bonuses are earned but subject to being returned if later results negate the earlier ones) group. In addition to excess performance claims, the penalty group stole two to three times as many pens as the others. Perhaps they were angered at having a reward taken away once the money was in their hands. Determining whether the net effect of incentives is positive or negative requires additional analysis.
Imagine that, as part of the experiment, we would have been able to sell each participant’s output at three times the amount each was paid, for actual output, and that the experimental “firm” had fixed costs of $100. The table below illustrates the effect each compensation scheme would have had, on average, for the firm:
Here we see that bonus-based compensation, increased both productivity and misrepresentation in a way that improved firm performance. However, penalty (claw-back) pay increased misrepresentation so much that the losses due to misrepresentation virtually wiped out the productivity gains. Different figures and assumptions will lead to different results, but the principal point we want to make is that accountants and board members need to be aware of the potential hazards that performance-based pay may bring.
While these types of methodologies only simulate real work-site conditions, the results are so striking that they should issue a warning of caution in common bonus systems and in the increasingly popular “claw-back” systems. Such return arrangements are already common in medicine (emergency medical technicians fined for submitting late or incorrect forms, physicians with capitation-withhold contracts), workers penalized based on defect rates, and sales people who pay back salary draws if sales are under quota. More recently, the outrage caused by the failure of financial institutions’ instruments after already having been paid hefty bonuses for “performance” has prompted demands for “claw-backs” in future compensation plans.
As much as we wish it were otherwise, when people are given performance incentives and the opportunity to cheat without apparent consequences, for many individuals the temptation is too great.
Developers of incentive systems might well consider the negative outcomes of this study and design around them if possible. It is good to remember Ben Franklin’s sage advice: “He’ll cheat without scruple, who can without fear.”
[1] While the salaried participants over-claimed their performance, there was no financial benefit to them in doing so. Because the anagram task was extremely difficult, over-claiming by salary participants could have been motivated by pride (not wanting to turn in a scoring sheet with few, if any, correct answers), knowing no harm would result from over-claiming (no monies paid).
The full paper can be obtained by emailing the authors at mcleanparks@WUSTL.edu
Very powerful research that validates an all too common pattern within organizations. Kudos to the Olin Award for showcasing this.