Bengt Holmström Prize Lecture: Pay for Performance and Beyond

428 

The Nobel Prizes

the commission rate on x

2

 will be α

2 

< p

2

, trading risk against incentives for a 

second-best outcome.

Now, assume that the tasks are substitutes: the more time the agent spends 

on one task, the higher is his marginal cost of spending time on the other task 

(the cross partial C

12

 > 0.) In this case it is no longer optimal to give the agent 

first-best incentives for choosing quantity (α

1

 = p

1

), because lowering α

1

 slightly 

will at the margin cost nothing in lost value on the first task, but will be strictly 

beneficial for providing incentives on the second task. The agent can be given 

the incentive to supply the same amount of quality for a lower α

2

, which reduces 

the cost of risk. Or alternatively, when α

1

 is lower the marginal cost of spending 

time on quality has gone down, making the agent spend more time on that task. 

Either way, lowering the incentive on quantity is advantageous for the supply 

of quality, by how much depends on the precision with which quality can be 

measured and how substitutable the tasks are in the cost function. If the two 

tasks are perfect substitutes (i.e., the agent just allocates his time between the 

two tasks so the cost function is C(e

1

 + e

2

)) and attention to quality is essential 

(this will require a nonlinear benefit function B), any incentive on quantity will 

have to be matched by a correspondingly strong incentive on quality so that α

1

= α

2

. This makes the agent indifferent between spending time on either task and 

he will choose whatever allocation is best for the principal. The cost is that the 

incentive for both tasks will have to be reduced because quality is poorly mea-

sured. If there is no quality measure at all (i.e., the variance of ε

2

 is infinite) then 

no incentive for either task is optimal.  If C′(0) < 0, the agent will still choose a 

positive level of total effort.

B. Misalignment and Manipulation

Misalignment is an important variation on the multitask theme (Baker 1992, 

2002). It can also be analyzed with the general multitask model.

Suppose again that the principal’s value is B(e

1

,e

2

) = p

1

e

1

 + p

2

e

2

, and there 

is just one performance measure x(e

1

,e

2

) = g

1

e

1

 + g

2

e

2

. Even though there is no 

uncertainty in the measure, there is a nontrivial incentive problem if the vec-

tors p and g are not aligned. For example, suppose x = e

1

 + e

2

 but B(e) = e

1

. 

The principal only values effort on the main task 1, but the agent can produce 

measured output x with e

1

 as well as e

2

. The second activity can be interpreted 

as “manipulation,” which the agent may feel a moral dislike for, but if the prin-

cipal pushes too strongly on measured performance x in the hope of getting the 

agent to work hard on the main task, the agent will engage in manipulation as 

well. The principal ends up compensating the agent for worthless performance, 

---

Pay For Performance and Beyond 

429

so misalignment causes waste, more so the higher α is. It will be optimal to set 

α < 1, because at α = 1 the marginal cost of reducing α is strictly negative (e

1

  is  

first-best while e

2

 is not).

Both variants of multitasking, disparities in measurement errors as well as 

misalignment between performance measures and the value created, are relevant 

variants of the general multitask model. It depends on the context which one is 

more natural to use.

C. “You Get What You Pay For”

Gross manipulation of performance measures was behind the recent Wells Fargo 

scandal (Tayan 2016). Wells Fargo had avoided the banking scandals associated 

with the financial crisis. It was known for prudence in lending, making profits 

by emphasizing its retail banking and customer service. Its branch managers 

were highly incentivized toward cross-selling: getting its regular banking cus-

tomers to buy a range of products, such as credit lines. This part of their banking 

business had steadily grown and been highly profitable. But continued growth 

also required new customers and eventually, as the sales goals were tightened 

(the branch managers’ performance was measured daily), some of the managers 

opened new accounts for its customers without the customers’ knowledge. Shell 

accounts were like a second activity in the two-task model described earlier. It 

improved measured performance and generated bonuses, but since there was no 

real activity in the accounts this activity generated minuscule  profits for Wells 

Fargo ($2.6 million according to Tayan 2016). Shell accounts caused minimal 

costs for customers (an estimated $2.50 per account), but they were of course 

illegal. Eventually the scam was discovered, causing the firing or resignation of 

thousands of employees and eventually also the resignation of Mr. Stumpf, the 

CEO. Wells Fargo had to pay $185 million in penalties, but the biggest cost by 

far was the enormous damage to their stellar reputation.

16

The explanations for the BP oil spill in the Gulf of Mexico point in many 

directions, but the fact that BPX—the exploration arm of BP—was encouraged 

to be more aggressive in its exploration activity was likely one of the culprits 

(Garicano and Rayo 2016). Measurable results were pushed hard (implicitly or 

explicitly), compromising safety. This is an example where there are many activi-

ties, some easy to measure (successful exploration), and others not so easy. While 

safety can be monitored, the intensity of monitoring whether rules are being 

strictly followed is not easy. Slight delays in service or in repair of faulty parts, 

especially in the case of backup systems and checks, may appear to carry minimal 

risk and therefore be subject to trade-offs under pressure.

---