Qualitative, unmeasurable organizational goals - limitations of formal evaluation and control systems

It follows from the analysis above that: (i) many of the important variables in organizations are impossible to measure with any degree of accuracy; and (ii) there are several dimensions to the goals of any company. Under these conditions, it is impossible for formal evaluation and reward systems to push people in the right direction (Holmstrom and Milgrom, 1991; Baker, 1992; Gibbons, 1998). The basic idea in these analyses is that if there is an underlying variable (profit, firm’s value, or any other) that the firm should maximize, this variable is measurable only imperfectly, so any explicit variables being measured will only be imperfectly correlated with the underlying variable.

The phenomena analyzed in the previously cited papers of Baker (1992) and Gibbons (1998) largely explain the recent scandals. Specifically, suppose that the underlying variable the firm wants to increase (say, long-run firm value) is not perfectly measurable today, and that the performance variable measured is only imperfectly correlated with that underlying variable. Then, when considering two specific courses of action, a1 and a2, it may well be that a1 is better than a2 from the point of view of the performance measure but is worse from the point of view of the underlying value. Suppose now that the firm has an incentive system that rewards measured performance - for example, a linear compensation system where employees are paid a fixed amount independent of performance, plus an amount for each unit of measured performance. For the firm as a whole, a2 is better than a1, but the opposite is true for the decision-maker. Hence, the decision-maker is quite likely to choose a1. In the extreme case, an agent could earn higher incentive pay by manipulating performance measurements, lying or cheating.

Therefore, a system that was supposed to align the interests of the professional managers perfectly with the long-run interests of the stockholders has in fact achieved the opposite.⁴