Incentives and Behavior Prof. Dr. Heiner Schumacher KU Leuven 14. - - PowerPoint PPT Presentation

Incentives and Behavior

• Prof. Dr. Heiner Schumacher

KU Leuven

• 14. Multitasking/Crowding
• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

1 / 20

Introduction

In the last chapters, we discussed how we can solve problems of asymmetric information with monetary incentives. However, are monetary incentives always good for performance? In reality, employment contracts often specify …xed wages. Incentives within …rms frequently are muted. In this chapter, we discover two reasons why monetary incentives may harm performance: Multiple tasks and motivational crowding out.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

2 / 20

Introduction

Overview Multitasking Motivational Crowding Out

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

3 / 20

Multitasking

In most real-world work environments, an agent has not only one task, but many tasks. For example, the CEO of a company has to hire managers, search for new markets, communicate with investors and so on. In general, when there a multiple tasks, the contract design does not

• nly in‡uence incentives and risk-allocation, but also the allocation of

the agent’s attention among her duties. In the following, we study optimal incentives in a setting with multiple

• tasks. In particular, we consider a simpli…ed version of Holmström

and Milgrom (1991).1

1Holmström, Bengt, and Paul Milgrom (1991): “Multitask Principal-Agent Analyses:

Incentive Contracts, Asset Ownership, and Job Design,” Journal of Law, Economics and Organization 7, 24 - 52.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

4 / 20

Multitasking

Assumptions (1) A has to perform two independent tasks. The e¤ort levels are a1 and a2, respectively. She has CARA preferences U(w, a) = eη[w ψ(a1,a2)], where η is her coe¢cient of absolute risk aversion. The cost function is ψ(a1, a2) = 1 2(c1a2

1 + c2a2 2) + δa1a2,

where 0 δ pc1c2. For δ = 0 the two e¤orts are technologically independent, for δ = pc1c2 the two e¤orts are perfect substitutes.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

5 / 20

Multitasking

Assumptions (2) P observes the outputs q1, q2 from the two tasks. The production function of task i is given by qi = ai + εi, where εi is normally distributed with variance σ2

i . The covariance

between outputs of di¤erent tasks is 0. P o¤ers the following linear incentive contract to A: w = t + s1q1 + s2q2. Hence, t is a …xed payment and si the incentive pay for task i. P’s …nal payo¤ is q1 + q2 w.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

6 / 20

Multitasking

We derive the optimal incentive scheme. A’s certainty equivalent from contract ft, s1, s2g can be calculated as t + s1a1 + s2a2 η 2 (s1σ2

1 + s2σ2 2) 1

2(c1a2

1 + c2a2 2) δa1a2.

(1) From this expression we can derive A’s unique best-response to contract ft, s1, s2g: a1 = s1c2 δs2 c1c2 δ2 and a2 = s2c1 δs1 c2c1 δ2 . (2) P’s objective is to maximize a1(1 s1) + a2(1 s2) t subject to the constraints that the expression in (1) weakly exceeds the value of A’s outside option (PC), and A’s e¤ort is given by (2) (IC).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

7 / 20

Multitasking

The solution to the maximization problem (make sure that you can derive it) is given by s

1

= 1 + (c2 δ)ησ2

2

1 + ηc2σ2

2 + ηc1σ2 1 + ησ2 1σ2 2(c1c2 δ2),

s

2

= 1 + (c1 δ)ησ2

1

1 + ηc2σ2

2 + ηc1σ2 1 + ησ2 1σ2 2(c1c2 δ2),

and t is such that the (PC) holds with equality. In the following, we assume that an interior solution exists, i.e., s

i 2 (0, 1) for i 2 f1, 2g.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

8 / 20

Multitasking

We analyze what happens to incentives if σ2

2 increases (i.e., it

becomes more di¢cult to distinguish e¤ort from noise). Clearly, s

2 decreases in σ2

• 2. The more noise there is in the production

function, the more di¢cult it is to provide incentives. More surprisingly, s

1 also decreases in σ2

• 2. We can calculate that

sgn ∂s

1

∂σ2

2

• = sgn
• δη(ησ2

1(δ c1) 1)

= sgn [s

2 ]

and s

2 is positive by assumption.

Note that there is an e¤ort-substitution problem: When task i is highly incentivized, while the other task i is not, A allocates too much of her e¤ort to task i. In order to reduce the e¤ort-substitution problem, both tasks are optimally less incentivized when the quality of the output measure for one task decreases.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

9 / 20

Multitasking

When P cares about a task where it is very hard to relate output to input (a task i with σ2

i = ∞), then it may be best to give no

incentives at all (for this result Holmström and Milgrom assume that A provides some e¤ort in the absence of any incentives). This is a very important result. It shows that in many circumstances it is optimal to provide no incentives even if some dimensions of the job have very precise link between input and output. A good example are contracts for teachers (which typically have “low-powered” incentives). While some dimensions of their “output” are simple to measure (students’ test scores), others are not (creativity, oral communication skills). Linking pay to test scores would then bias teachers’ e¤ort to skills that are tested on standardized exams.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

10 / 20

Multitasking

Exclusion from other tasks. When designing the optimal

• rganizational design, P also has to consider what A can do outside

his job (outside work). If the market starts valuing outside work more highly, P either has to increase incentive pay or exclude A from

• utside work.

Allocating tasks among agents. If P can allocate tasks among several (identical) agents, it may be useful that some agents specialize in tasks that are hard to monitor (and get low-powered incentives), while others only engage in tasks that are easy to monitor (and get high-powered incentives).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

11 / 20

Motivational Crowding Out

In psychology, people di¤erentiate between two categories of motivation. Extrinsic Motivation occurs when people are compelled to do something or act a certain way because of external factors. For example, a manager works hard for his bonus, or a student studies hard in order to obtain a good grade. Intrinsic Motivation occurs when people are internally motivated to do something because it brings them pleasure, or they think that it is important, or they feel that what they are doing is signi…cant. For example, a manager may exert e¤ort even in the absence of monetary incentives if she is convinced that she is working or a good cause. Or a student studies hard, because she likes the subject.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

12 / 20

Motivational Crowding Out

The crowding-out e¤ect suggests that raising monetary incentives crowds out intrinsic motivation and therefore may reduce supply (of e¤ort). In certain circumstances, it is therefore not advisable to use monetary incentives to elicit e¤ort, but intrinsic motivation. Now even economists admit that the crowding-out e¤ect may play an important role. Gibbons (1998) says:2 “A more troubling possibility is that management practices based on economic models may dampen (or even destroy) non-economic realities such as intrinsic motivation and social relations.”

2Gibbons, Robert (1998): “Incentives in Organizations,” Journal of Economic

Perspectives 12(4), 115 - 132.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

13 / 20

Motivational Crowding Out

The e¤ect of external rewards on intrinsic motivation has been attributed to two psychological processes: Impaired Self-Determination. When individuals perceive an external intervention as reducing their self-determination, intrinsic motivation is substituted by extrinsic control. Impaired Self-Esteem. When outside intervention carries the notion that the actor’s motivation is not acknowledged, his or her intrinsic motivation is e¤ectively rejected. An intrinsically motivated person is deprived of the chance of displaying his or her own interest and involvement in an activity when someone else o¤ers a reward.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

14 / 20

Motivational Crowding Out

There is some empirical evidence for the crowding out e¤ect. In the following, we consider the two experiments from Gneezy and Rustichini (2000).3 In the …rst experiment, 160 students from the University of Haifa were recruited to answer a set of 50 questions from an IQ test. Students were paid a …xed amount of 60 New Israeli Shekel (NIS). They were divided into four groups. The …rst group had no further monetary incentives. The second group additionally earned 10 cents for each correctly answered question. The third group earned 1 NIS for each correctly answered question, and in the fourth group the corresponding incentive was 3 NIS per question.

3Gneezy, Uri, and Aldo Rustichini (2000): “Pay Enough Or Don’t Pay At All,”

Quarterly Journal of Economics 115(3), 791 - 810.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

15 / 20

Motivational Crowding Out

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

16 / 20

Motivational Crowding Out

In the second experiment, 180 high school children collect monetary donations that households make to societies for cancer research. They were divided into three groups. All groups heard a speech about the importance of the task. The …rst group had no further monetary

• incentives. The second group additionally earned 1 percent of the

total amount collected. The third group earned 10 percent of the total amount collected.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

17 / 20

Motivational Crowding Out

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

18 / 20

Motivational Crowding Out

If we compare the treatments with no monetary incentives and those with small monetary incentives, we observe that monetary compensation produces a reduction in the performance (crowding out e¤ect). If we compare treatments with monetary rewards, then we observe that performance increases in monetary incentives. Hence, we have discontinuity: “Pay enough or don’t pay at all”.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

19 / 20

Motivational Crowding Out

In both experiments, the activity has intrinsic motivation, and the introduction of monetary rewards (extrinsic motivation) displaces it. If the monetary rewards are su¢ciently small, performance is reduced. The authors interpret this …nding as follows. If there are no monetary rewards, subjects feel obliged to reciprocate the …xed payment. If there are monetary rewards, the …xed payment becomes meaningless, only the variable payment matters for e¤ort. If the variable payment is small, so is the e¤ort.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 14. Multitasking/Crowding

20 / 20