A division in a company, division A, is trying to find talented people. They have a number of positions to fill and they can fill them now or wait to find better candidates. Once a position is filled, it is hard to terminate employment for a few periods at least. The profits of the division depend both on the quantity and quality of the people employed.
This system creates an option value to waiting. Given its payoff function, the division has a quality bar for hiring. The quality bar depends on the number of unfilled slots. It leaves some slots empty deliberately to capitalize on option value. There is another threshold for firing, also with an important option value component in its determination.
A CEO is hired to build value. He is focussed on the short term. For him, unused resources mean lower output – the quality is less important. So he starts allocating resources across divisions. If a division has unused resources or positions, the CEO gives them to another division if they can come up with a candidate that passes their bar.
Let’s assume the divisions have private values – they put zero value on the other division’s success. (Microsoft recently reorganized itself because different parts of the company were at war with each other.)
For each division, the option value of an empty slot declines – it can now be filled by someone from the other division. The incentives are obvious – both divisions lower their standards for hiring and firing. There is a race to the bottom.
Counterintuitive effects arise if there are common values – each division takes the other’s success into account. Suppose division A’s standards for hiring are originally higher than division B’s. Division A faces much the same incentives as in the case of private values – it will lower standards to pre-empt slot reallocation to division B. But division B, since it now wants to prevent the decline of division A, will raise standards so the principal will not give it division A slots. The fact that this seems so implausible argues in favor of private values.
In any case, even in the case of common values, the divisions’ equilibrium response to principal short termism will not achieve the first best. A better strategy is to reallocate slots between divisions ex ante. The reallocation should reflect the principal’s payoff function as well as those of the divisions. Then, let the divisions make their own decisions on hiring and firing.