The three primary research questions addressed in this study are: 1) does past project performance history affect the risk level of future project selections; 2) do two common bonus incentive pay schemes affect the risk level of the projects selected, and; 3) does past performance history and the specific bonus incentive pay schemes employed combine in an additive manner to affect the risk level of the future project selected? These questions are important because firms trying to use bonus incentives to encourage a level of risk taking that is optimal from the firm's perspective may not incorporate the psychological effects that results from their managers' past experience with risk. The end result may be too much risk taking if managers given a hurdle bonus scheme (i.e. designed intentionally to not compensate for taking on additional risk) have had prior negative experience that induces risk seeking. In contrast, too little risk taking may result if managers are given a graduated bonus scheme that encourages risk taking, but the managers have had positive past experience which induces risk aversion.
In our experimental context, a hypothetical firm was looking for outstanding breakthrough new products that would earn large returns. Subjects, acting as new product managers had to select among a portfolio of projects, all with equal expected returns, but varying levels of risk. Riskier projects had a greater probability of earning far above average returns (i.e. corresponding to outstanding products). We manipulated the past project performance history of subjects by having them go through two practice periods where they selected new products and then learned whether their selections earned above or below target returns. Under negative (positive) past history, both products selected earned below (above) average returns. Bonus incentive compensation was manipulated by having half the subjects face a hurdle bonus scheme and the other half face a graduated bonus scheme. Under the hurdle scheme, subjects earned a fixed bonus if their product's return was at or above the target return rate. Under the graduated bonus, subjects earned incrementally increasing bonuses depending on if and how far above the target return their product's return was.
We hypothesized that subjects with negative (positive) past product performance history would select a riskier product in the third (measured) period than subjects with positive past history. Our results are consistent with this prediction and hold even after controlling for attitude and innate risk aversion. The importance of this result is that past history, although having no direct effect on future project risk or returns, is a significant predictor of the risk level of future projects selected by managers.
Our second hypothesis examined whether varying the structure of managerial bonus pay affected the level of risk taken on in new product selection. We predicted that subjects under the graduated bonus scheme (i.e. where higher returns resulted in higher bonus pay) would take on more risk than subjects under the hurdle scheme. We found results generally consistent with this prediction, although at a somewhat lower level of statistical significance.
Our third hypothesis predicted and our results show that past performance history and the bonus pay structure combine in an additive manner. Specifically, the greatest amount of risk was undertaken under negative history and a graduated bonus scheme. The least amount of risk was undertaken under positive history combined with a simple hurdle bonus scheme.
The current study contributes to the growing literature in accounting that examines what factors affect the level of risk taken on by managers. This topic is of high current interest in light of recent financial crises related to risk taking, and attempts to regulate compensation systems to induce appropriate behavior related to risk. Our study contributes to the debate on how firms can design compensation systems by stressing that, in addition to the rational predicted effects of an incentive scheme, the past performance of the managers affected needs to be taken into account.
