LECTURE NOTES ON THE PRINCIPAL AGENT MODEL · PRINCIPAL AGENT MODEL 1. SIMPLIFIED MODEL 2....

37
LECTURE NOTES ON THE PRINCIPAL AGENT MODEL 1

Transcript of LECTURE NOTES ON THE PRINCIPAL AGENT MODEL · PRINCIPAL AGENT MODEL 1. SIMPLIFIED MODEL 2....

LECTURE NOTES ON THE

PRINCIPAL AGENT MODEL

1

SIMPLIFIED MODEL

2

SIMPLIFIED MODEL

The Set-Up Two players: a firm and a worker.

SIMPLIFIED MODEL

The Set-Up Two players: a firm and a worker.

The sequential move-structure of the game:

1. Firm proposes a wage contract.

SIMPLIFIED MODEL

The Set-Up Two players: a firm and a worker.

The sequential move-structure of the game:

1. Firm proposes a wage contract.

2. Worker decides between accepting or reject-

ing the wage contract offered. If the worker re-

jects it, game ends with each player receiving

a payoff of zero. But if worker accepts wage

contract, then game moves to next stage.

SIMPLIFIED MODEL

The Set-Up Two players: a firm and a worker.

The sequential move-structure of the game:

1. Firm proposes a wage contract.

2. Worker decides between accepting or reject-

ing the wage contract offered. If the worker re-

jects it, game ends with each player receiving

a payoff of zero. But if worker accepts wage

contract, then game moves to next stage.

3. Workers chooses effort level e.

SIMPLIFIED MODEL

The Set-Up Two players: a firm and a worker.

The sequential move-structure of the game:

1. Firm proposes a wage contract.

2. Worker decides between accepting or reject-

ing the wage contract offered. If the worker re-

jects it, game ends with each player receiving

a payoff of zero. But if worker accepts wage

contract, then game moves to next stage.

3. Workers chooses effort level e.

4. Output is realized, wage is paid to worker,

and the game ends.

Given effort, e: with probability ηe, output is

high and revenue associated with that is v. But

with probability 1 − ηe, no output is produced

and zero revenue obtained. The former is a

case of the ”project” on which worker works

being a success, while the latter a failure.

So note that worker’s effort generates ”ran-

dom” output.

3

Given effort, e: with probability ηe, output is

high and revenue associated with that is v. But

with probability 1 − ηe, no output is produced

and zero revenue obtained. The former is a

case of the ”project” on which worker works

being a success, while the latter a failure.

So note that worker’s effort generates ”ran-

dom” output.

All players are risk neutral.

Given effort, e: with probability ηe, output is

high and revenue associated with that is v. But

with probability 1 − ηe, no output is produced

and zero revenue obtained. The former is a

case of the ”project” on which worker works

being a success, while the latter a failure.

So note that worker’s effort generates ”ran-

dom” output.

All players are risk neutral.

Expected profit to firm is: Eπ = (ηe)v − w,

where w is wage.

Given effort, e: with probability ηe, output is

high and revenue associated with that is v. But

with probability 1 − ηe, no output is produced

and zero revenue obtained. The former is a

case of the ”project” on which worker works

being a success, while the latter a failure.

So note that worker’s effort generates ”ran-

dom” output.

All players are risk neutral.

Expected profit to firm is: Eπ = (ηe)v − w,

where w is wage.

Expect Utility to worker is: EU = w− ce2

3 , where

c > 0.

FIRST-BEST EFFORT LEVEL

4

FIRST-BEST EFFORT LEVEL

Maximize Eπ such that EU ≥ u.

FIRST-BEST EFFORT LEVEL

Maximize Eπ such that EU ≥ u.

That is, choose e to maximize social surplus:

FIRST-BEST EFFORT LEVEL

Maximize Eπ such that EU ≥ u.

That is, choose e to maximize social surplus:

maxe

ηev − ce2

3.

FIRST-BEST EFFORT LEVEL

Maximize Eπ such that EU ≥ u.

That is, choose e to maximize social surplus:

maxe

ηev − ce2

3.

First-Order condition:

ηv =2ce3

.

FIRST-BEST EFFORT LEVEL

Maximize Eπ such that EU ≥ u.

That is, choose e to maximize social surplus:

maxe

ηev − ce2

3.

First-Order condition:

ηv =2ce3

.

Hence, first-best effort level is:

e∗ =3ηv2c

.

FIRM CANNOT OBSERVE EFFORT, BUT

ONLY OUTPUT LEVEL.

5

FIRM CANNOT OBSERVE EFFORT, BUT

ONLY OUTPUT LEVEL.

Hence, firm’s wage contract cannot be condi-

tioned on e.

FIRM CANNOT OBSERVE EFFORT, BUT

ONLY OUTPUT LEVEL.

Hence, firm’s wage contract cannot be condi-

tioned on e.

But instead, firm’s wage contract is condi-

tioned on observable and verifiable output.

FIRM CANNOT OBSERVE EFFORT, BUT

ONLY OUTPUT LEVEL.

Hence, firm’s wage contract cannot be condi-

tioned on e.

But instead, firm’s wage contract is condi-

tioned on observable and verifiable output.

So, wage contract is a pair: (wS, wF), where

wS is wage when output is high (ie., project is

a success) and wF is wage when output is low

(zero – project is a failure).

FIRM CANNOT OBSERVE EFFORT, BUT

ONLY OUTPUT LEVEL.

Hence, firm’s wage contract cannot be condi-

tioned on e.

But instead, firm’s wage contract is condi-

tioned on observable and verifiable output.

So, wage contract is a pair: (wS, wF), where

wS is wage when output is high (ie., project is

a success) and wF is wage when output is low

(zero – project is a failure).

Worker’s problem:

maxe

EU = (ηe)wS + (1 − ηe)wF − ce2

3.

6

Worker’s problem:

maxe

EU = (ηe)wS + (1 − ηe)wF − ce2

3.

FOC:

η(wS − wF) =2ce3

.

Worker’s problem:

maxe

EU = (ηe)wS + (1 − ηe)wF − ce2

3.

FOC:

η(wS − wF) =2ce3

.

This implies that the subgame perfect equilib-

rium (SPE) effort level is:

e =3η(wS − wF)

2c. (1)

RESULT 1:

If wS − wF = v, then e = e∗. That is, in that

case, SPE effort equals first-best effort level.

7

RESULT 1:

If wS − wF = v, then e = e∗. That is, in that

case, SPE effort equals first-best effort level.

Assume liquidity constrained.

That is: wS ≥ 0 and wF ≥ 0.

RESULT 1:

If wS − wF = v, then e = e∗. That is, in that

case, SPE effort equals first-best effort level.

Assume liquidity constrained.

That is: wS ≥ 0 and wF ≥ 0.

Proceeding backwards, we now choose wage

contract (wS, wF) to maximize firms’ expected

payoff subject to e = e, as derived above in

equation 1.

RESULT 1:

If wS − wF = v, then e = e∗. That is, in that

case, SPE effort equals first-best effort level.

Assume liquidity constrained.

That is: wS ≥ 0 and wF ≥ 0.

Proceeding backwards, we now choose wage

contract (wS, wF) to maximize firms’ expected

payoff subject to e = e, as derived above in

equation 1.

Thus:

maxwS,wF

Eπ = ηe(v − wS) + (1 − ηe)(0 − wF),

subject to e = e.

RESULT 1:

If wS − wF = v, then e = e∗. That is, in that

case, SPE effort equals first-best effort level.

Assume liquidity constrained.

That is: wS ≥ 0 and wF ≥ 0.

Proceeding backwards, we now choose wage

contract (wS, wF) to maximize firms’ expected

payoff subject to e = e, as derived above in

equation 1.

Thus:

maxwS,wF

Eπ = ηe(v − wS) + (1 − ηe)(0 − wF),

subject to e = e.

First, we claim that in the solution, wF = 0.

Argue by contradiction.

8

First, we claim that in the solution, wF = 0.

Argue by contradiction.

Suppose instead at the optimum, wF > 0. Then

one could decrease both wS and wF by same

and small amounts, keeping e constant (and

hence satisfying equation 1), and in turn in-

crease Eπ. Hence, a contradiction.

First, we claim that in the solution, wF = 0.

Argue by contradiction.

Suppose instead at the optimum, wF > 0. Then

one could decrease both wS and wF by same

and small amounts, keeping e constant (and

hence satisfying equation 1), and in turn in-

crease Eπ. Hence, a contradiction.

Substituting wF = 0 into the maximand above,

and also for e = e, using equation 1, we need

to now solve:

maxwS

3η2wS(v − wS)

2c.

First, we claim that in the solution, wF = 0.

Argue by contradiction.

Suppose instead at the optimum, wF > 0. Then

one could decrease both wS and wF by same

and small amounts, keeping e constant (and

hence satisfying equation 1), and in turn in-

crease Eπ. Hence, a contradiction.

Substituting wF = 0 into the maximand above,

and also for e = e, using equation 1, we need

to now solve:

maxwS

3η2wS(v − wS)

2c.

First Order Condition:

3η2v2c

− 3η2wSc

= 0.

9

First Order Condition:

3η2v2c

− 3η2wSc

= 0.

Solving for wS, we get wS = v2.

Conclusion: In the unique SPE, the wage con-

tract offered is (wS, wF) = (v2 , 0) and effort is

e = 3ηv4c .

First Order Condition:

3η2v2c

− 3η2wSc

= 0.

Solving for wS, we get wS = v2.

Conclusion: In the unique SPE, the wage con-

tract offered is (wS, wF) = (v2 , 0) and effort is

e = 3ηv4c .

NOTE, the SPE effort is less than first-best

effort.