| Peter Klein |
I was flipping channels last night and came across a Jimi Hendrix biopic. Lots of concert footage, with Hendrix doing his usual amazing Hendrix things — singing, crazy guitar riffs, playing with his teeth. Then I noticed something I hadn’t seen before: He’s doing all this while chewing gum. How can he sing without choking or spitting it out? How does he pluck guitar strings with his teeth and not leave a big wad on the pickups? Some people have trouble walking and chewing gum at the same time. How is he doing this?
Naturally, this got me thinking about multitask principal-agent problems. I liked one article that appeared in 2005 but didn’t generate much attention: Besanko, Regibeau, and Rockett’s “A Multi-Task Principal-Agent Approach to Organizational Form” (Journal of Industrial Economics, December 2005). Multitasking has often been applied performance evaluation, specifically the use of objective versus subjective performance measures. (If the agent is assigned multiple tasks, only some of which are measurable, than a quantitative evaluation scheme biases the agent’s effort toward more easily measurable tasks.) Besanko et al. apply this logic to divisionalization, examining the choice between product-line and functional organization:
This paper studies the choice of organizational forms in a multi-task principal-agent model. We compare a functional organization in which the firm is organized into functional departments such as marketing and R&D to a product-based organization in which the firm is organized into product lines. Managers’ compensation can be based on noisy measures of product-line profits. Measures of a functional area’s contribution to total profits are not available, however. This effect favors the product organization. However, if there are significant asymmetries between functional area contributions to organizational success and cross-product externalities within functions, organizing along functional lines may dominate the product organization. The functional organization can also dominate when a function is characterized by strong externalities while the other is not.
An obvious example: university faculty who are tasked with research, teaching, and service. Research is (in principle) easy to measure: publications, citation counts, grants, speaking invitations, etc. Teaching and service outputs are fuzzier. Even well-intentioned administrators tend to reward what they can count, which means refereed journal articles (and, in some fields, grant dollars) end up being the primary performance metric. Need I point out what this does to teaching?
The same issue is explored in Williamson (1975, pp. 155-75), where the ability to reward division managers based on standalone, divisional profit figures is cited as an advantage of the M-form structure.