Salesforce Compensation: Revisiting and Extending the Agency-Theoretic Approach

Since the papers of Basu et. al. (1985) and Lal and Srinivasan (1993), marketing academics have been interested in the design and implementation of optimal compensation plans. The literature has focused on agency theory as a foundation to help describe and understand this process. Although there has been much theoretical work on this topic, empirical evidence to support this theory remains sparse. Studies by Coughlan and Narasimhan (1992) and John and Weitz (1988, 1989) have found some early evidence that supports agency theory. In this paper we revisit the issue of salesforce compensation on both theoretical and empirical fronts. On the theory side we build a game theoretic model of salesforce compensation that accounts for risk aversion on the part of both the principal and the agent. We find that the results obtained from our model, while substantiating past findings, offer some new insights into the compensation design process. In particular we find that firm demographics play an important role in the design of the optimal compensation scheme. We then use two datasets collected ten years apart by the Dartnell Corporation to investigate and test hypotheses generated by our model and the extant literature. We advance the sales compensation literature methodologically by simultaneously estimating the effects of independent variables on our two compensation measures of interest: the ratio of salary to total pay, and total pay itself. We find that simultaneous estimation improves our results considerably, and is logically and theoretically motivated by the natural correlation between the dependent variables of interest. Our research thus adds to our substantive knowledge of the drivers of salesforce compensation, while adding to the theoretical structure through taking account of the possibility of principal risk aversion, and adding methodological insight through simultaneous estimation of the equations of interest.