1. Introduction
Consider an economy with 2 consumers and 1 consumption good. There are 2 periods, t = 0 (today) and t = 1 (tomorrow). The agents do not know the state of the world at t = 1. To simplify, we will assume that there are two possible states (alternatives), e = e1, e2. The probabilities of each state are: p(e1) = π, p(e2) = 1 − π.
ππππ 1111−−−− ππππ
e1 e2 t = 0
t = 1
w11
w12
w21
w22 Agent 1 Agent 2
x11
x12
x21
x22 Agent 1 Agent 2
w11
w12
w21
w22
x11
x12
x21
x22 +
+
+
+
= w1
= w2
=
=
ππππ
1111−−−− ππππ
We consider that there is a unique good in the economy that may be consumed only at t = 1 (that is, there is no consumption at t = 0.) Let us use the following notation:
• xsi is the amount of the good that agent i consumes in state es.
• wsi are the initial endowments that agent i has in state es.
• ws = ws1+ w2s, s = 1, 2.
• The utility functions of the agents are
ui(x1i, x2i) = πui(x1i) + (1 − π)ui(x2i) i = 1, 2 i.e. agents maximize expected utility.
1
We may represent this situation using Edgeworth’s box
x11 x21
x12 x22
x11 x21
(w11+ w12, w21+ w22)
2. Pareto Efficiency
Let us compute the Pareto efficient allocations. These are solutions of the following optimization problem
max πu1(x11) + (1 − π)u1(x21) subject to πu2(x12) + (1 − π)u2(x22) ≥ ¯u
x11+ x12 = w1 x21+ x22 = w2 The Lagrangian is
L = πu1(x11)+(1−π)u1(x21)+λ ¯u − πu2(x12) − (1 − π)u2(x22)+µ1 w1− x11− x12+µ2 w2− x21− x22 The first order conditions are
∂L
∂x11 : πu01(x11) = µ1
∂L
∂x12 : λπu02(x12) = µ1
∂L
∂x21 : (1 − π)u01(x21) = µ2
∂L
∂x22 : λ(1 − π)u02(x22) = µ2
πu01(x11) = λπu02(x12) (1 − π)u01(x21) = λ(1 − π)u02(x22)
Observe that the marginal rate of substitution for agent i = 1, 2 is
πu0i(x1i) (1 − π)u0i(x2i)
Simplifying the above equations, we see that
u01(x11) = λu02(x12) (2.1)
u01(x21) = λu02(x22) (2.2)
and dividing the above equations, we obtain that
(2.3) u01(x11)
u01(x21) = u02(x12) u02(x22)
These equations, together with the feasibility conditions
x11+ x12 = w1 x21+ x22 = w2
determine the Pareto efficient allocations.
Example 1. Suppose that
u1(x) =√
x, w11 = 0, w12 = 1 u2(x) =√
x, w21 = 1 = w22 π = 1/4
w11= 0
w12= 1
w21= 1
w22= 1
π = 1 / 4
π = 1 / 4 π = 1 / 4 π = 1 / 4 1111−−−− π π π π
= 3 / 4Initial endowments Agent 1 Agent 2
u1(x) = x u2(x) = x The conditions for Pareto optimality
u01(x11)
u01(x21) = u02(x12) u02(x22) become
x21 x11 = x22
x12 which together with the feasibility condition
x11+ x12 = 1 x21+ x22 = 2 determine the Pareto efficient allocations. Taking
x = x11 y = x21 we have
x12 = 1 − x11 = 1 − x x22 = 2 − x21 = 2 − y and substituting in the first order condition, we obtain
y
x = 2 − y 1 − x
that is x = y/2 and the Pareto efficient allocations are of the form x11 = y/2 x12 = 1 − y/2
x21 = y x22 = 2 − y 0 ≤ y ≤ 2
In the Edgeworth box, we see that the Pareto efficient allocations determine the straight line x11 = x21
2
(1,2)
x11 x21
2.1. One risk neutral agent and one risk averse agent. Suppose now that
• agent 1 is risk averse: u001 < 0. (Equivalently, u01 is strictly decreasing)
• agent 2 is risk neutral: u02 is constant.
In this case,
u02(x12) u02(x22) = 1 So, the first order Pareto optimality condition 2.3 is
u01(x11) u01(x21) = 1 that is,
u01(x11) = u01(x21)
and since u01 is strictly decreasing, the first order Pareto optimality condition becomes x11 = x21
That is, agent 1 insures himself completely.
Example 2. In example 1, suppose that agent 2 is risk neutral. That is, u1(x) =√
x, w11 = 0, w12 = 1 u2(x) = x, w21 = 1 = w22
π = 1/4
w11= 0
w12= 1
w21= 1
w22= 1
π = 1 / 4
π = 1 / 4 π = 1 / 4 π = 1 / 4 1111−−−− π π π π
= 3 / 4Agent 1 Agent 2
u1(x) = x u2(x) = x Initial endowments
Thus, the Pareto efficient allocations are
(1,2)
x11 x21
1
1
3. Exchange Economies
Now, we introduce prices that allow to contract (at t = 0) the delivery of 1 unit of the good at each state.
ππππ 1111−−−− ππππ
p1 p2
t= 0 t= 1
The model is the as follows.
• At t = 0 the ‘Walrasian auctioneer’ announces prices p1, p2.
• For each state of nature s = 1, 2, the price ps allows to purchase one unit of the good to be delivered if state es occurs at t = 1.
• That is, agent i = 1, 2 can purchase the bundle (x1i, x2i) at t = 0 for the price p1x1i + p2x2i. This bundle allows agent i to consume x1i units of the good if state e1 occurs and x2i units of the good if state e2 happens.
• The bundle (x1i, x2i) is a contingent consumption plan.
• These prices are paid at t = 0.
• The income of agent i = 1, 2 at t = 0 is
p1w1i + p2wi2
• Thus, at t = 0 agent i = 1, 2 chooses the bundle that maximizes his utility subject to his budget constraint,
max πui(x1i) + (1 − π)ui(x2i) (3.1)
subject to p1x1i + p2x2i = p1wi1+ p2wi2
• The solution to the above optimization problem determines the demand of each agent: x1i(p), x2i(p).
• The equilibrium prices (p1, p2) are those that clear the market, state by state x11(p) + x12(p) = w11+ w12
(3.2)
x21(p) + x22(p) = w21+ w22 Observation 3. The first order condition of problem 3.1 is
MRS = p1 p2 that is,
πu0i(x1i)
(1 − π)u0i(x2i) = p1
p2 i = 1, 2
In particular, the marginal rates of substitution for both agents coincide. Hence, the competitive equilibria are Pareto efficient.
Example 4. Let us compute the competitive equilibrium in example 1. The economy is described by
u1(x) =√
x, w11 = 0, w12 = 1 u2(x) =√
x, w21 = 1 = w22 π = 1/4
w11= 0
w12= 1
w21= 1
w22= 1
π = 1 / 4
π = 1 / 4 π = 1 / 4 π = 1 / 4 1111−−−− π π π π
= 3 / 4Initial endowments Agent 1 Agent 2
u1(x) = x u2(x) = x
Let p = (p1, p2) be the equilibrium prices. The condition for Pareto efficiency is x21 = 2x11
On the other hand, the first order conditions for utility maximization, πu0i(x1i)
(1 − π)u0i(x2i) = p1
p2 i = 1, 2 for agent 1, imply that
p1 p2 = 1
3 s
x21 x11 =
√2 3
(In the second equality, we have used the Pareto efficiency condition, x21 = 2x11). We see that the equilibrium prices are
p1 =√
2, p2 = 3 The budget restriction for agent 1 is
p1x11 + p2x21 = p2
and, since the equilibrium allocations are Pareto efficient, we have that p2 = p1x11+ 2p2x11 = (p1+ 2p2)x11 so,
x11 = p2
p1+ 2p2 = 3
√2 + 6, x21 = 6
√2 + 6 and the demand of agent 2 is
x12 = 1 − 3
√2 + 6, x22 = 2 − 6
√2 + 6
Observation 5. Coming back to the general model, if, in addition one of the agents, for example agent 2, is risk neutral, then
u02(x12) u02(x22) = 1 so, the equilibrium prices verify that
π
1 − π = p1 p2
And if, in addition, agent 1 es strictly risk averse, the above observation implies that in the competitive equilibrium
x11 = x21 Example 6. In example 1, suppose that
u1(x) =√
x, w11 = 0, w12 = 1 u2(x) = x, w21 = 1 = w22
π = 1/4 That is, agent 2 is risk neutral.
w11= 0
w12= 1
w21= 1
w22= 1
π = 1 / 4
π = 1 / 4 π = 1 / 4 π = 1 / 4 1111−−−− π π π π
= 3 / 4Agent 1 Agent 2
u1(x) = x u2(x) = x Initial endowments
Then, the equilibrium prices are
p1 = 1
4, p2 = 3 4 the demand of agent 1 verifies that
x11 = x21 = x and the budget restriction
p1x11+ p2x21 = p1w11+ p2w12
is 1
4x + 3 4x = 3
4 so the equilibrium allocation is
x11 = x21 = 3
4 For agent 1
and
x12 = 1
4, x22 = 5
4 For agent 2
3.2. Interpretation. One way to interpret 6 is in terms of exchange economies. In state e2 agent 1 transfers the amount
w12− x21 = 1 − 3 4 = 1
4 (= x22− w22) to agent 2. In exchange, agent 1 receives
x11− w11 = 3
4 (= w12− x12) from agent 2, in state e1.
W11= 0
W12= 1
W21= 1
W22= 1
ππππ
1111−−−− ππππ
Initial endowments Agent 1 Agent 2
Consumption Agent 1 Agent 2
x11= 3 / 4
x12= 3 / 4
x21= 1 / 4
x22= 5 / 4 3 / 4
1 / 4
4. Insurance markets
Another possible interpretation of example 6 is in terms of insurance. Agent 1 has initially a wealth w10 = 1 in t = 0. And with probability
π = 1 4
may suffer a loss of D units of wealth. That is, he faces the following situation
(1) In state e1, which happens with probability π = 1/4, agent 1 loses D = 1 units, so w11 = w01− D = 0.
(2) In state e2, which happens with probability π = 3/4, agent 1 does not suffer any loss, so w21 = w10 = 1.
Graphically,
w10 - D
ππππ
1111−−−− ππππ
Agent 1
w10
w20
ππππ
1111−−−− ππππ
Agent 2
w20
w20 w10
w10 - D -q
αααα + α + α + α + α ππππ
1111−−−− ππππ
w10– qαααα
w20 + q
αααα
-αααα ππππ
1111−−−− ππππ
w20+ qαααα
w20 + q
αααα
w10– qαααα
Suppose now that agent 2, who is risk neutral, represents an insurance company. Initially, the insurance company owns one unit w02 = 1 of the good at t = 0 and it is not subject to risk. That is, in both states e1 and e2 keeps its wealth w12 = 1 = w22.
Agent 2 offers to agent 1 the possibility of insuring α units of wealth at the price q, per each insured unit. From the point of view of agent 1, the situation is as follows
• At t = 0 agent 1 buys α units of the insurance and pays the amount qα to agent 2. He keeps w01− qα = 1 − qα.
• If the state e1 happens (with probability π = 1/4) agent 1 suffers a loss of D = 1 units of his wealth. Therefore agent 2 pays to agent 1 the amount insured α. The remaining wealth of agent 1 is x11 = w01− qα − D + α = α − 1.
• If the state e2 occurs (with probability π = 3/4) gente 1 does not suffer any loss. Therefore, he does not receive any compensation from agent 2. The remaining wealth of agent 1 is x21 = w10− qα = 1 − qα.
Suppose also that the insurance market is perfectly competitive. Therefore, the expected benefit for insurance companies is 0. From the point of view of the insurance company (agent 2), if agent 1 purchases α units of the insurance, the situation is the following,
• At t = 0 it receives qα units from agent 1 and starts with w02+ qα = 1 + qα.
• If state e1 occurs agent 1 loses D = 1 units and agent 2 pays the amount α insured. The remaining wealth of agent 2 is w21 = w02 + qα − α = 1 + qα − α.
• If state e2 occurs, agent 1 does not suffer any loss and agent 2 does not have to pay any compensation. The remaining wealth of agent 2 is w22 = w02+ qα = 1 + qα.
The condition of expected 0 zero profit is
w20 = π(w20+ qα − α) + (1 − π)(w20+ qα) = w20+ qα − πα that is, the competitive price of each insured unit is
q = π This is the actuarially fair price.
Given the price
q = π
agent 1 chooses the amount α of insurance that maximizes his expected utility, πu1(w10− qα − D + α) + (1 − π)u1(w10− qα)
The first order condition is
π(1 − q)u01(w10− qα − D + α) = (1 − π)qu01(w01− qα) and since q = π this condition implies
u01(w10− qα − D + α) = u01(w10− qα) But, since agent 1 is risk averse, we have that u01 is decreasing. Hence,
w10− qα − D + α = w10− qα and we see that
α = D = 1
that is, agent 1 insures himself completely. The consumption of the agent is
ππππ 1111−−−− ππππ
w10–ππππ D
w10–
ππππ D
w10–ππππ D
Note that EV = w10 − πD is the expected consumption of the agent when he buys no insurance.
That is, when there is perfect competition among the insurance companies, the expected utility of the agent is u(EV).
On the other hand, the amount paid by to the insurance company by the agent is pe= πD. Letting Ie be the amount of insurance bought by the agent, we see that the insurance policy in the competitive equilibrium is (pe, Ie) = (πD, D).
Plugging in the values
α = D = 1, q = π = 1 4 we see that at t = 0, agent 2 charges the amount
1 4
to agent 1, in exchange for giving him full insurance. In either state the wealth of agent 1 is 1 − 1
4 = 3
4 = EV Agent 2 has
1 4
1 + 1
4− 1 = 1 4 and in state e2 does not pay anything to agent 1 and has
1 + 1 4 = 5
4
5. Perfectly Discriminating Monopoly The model is the as follows.
• Agent 2 is a monopolist. He knows the utility function of agent 1.
• At t = 0, agent 2 proposes the feasible ‘contract’ (x11, x21) to agent 1. That is, he proposes agent 1 the previous contingent consumption plan.
• In above contract, agent 2 consumes the rest of the resources (x12, x22) = (w11+ w21, w12+ w22) − (x11, x21)
• If, at t = 0, agent 1 accepts the contract, then consumption at t = 1 is the one described by the above contract.
• If, at t = 0, agent 1 rejects the contract, then at t = 1 each agent consumes his initial endowments (wi1, wi2).
The monopolist (agent 2) chooses a contract (x1i, x2i) such that he maximizes his utility and keeps agent 1 at his reservation utility
max πu2(x12) + (1 − π)u2(x22)
subject to πu1(x11) + (1 − π)u1(x21) ≥ πu1(w11) + (1 − π)u1(w12)
This is coincides exactly with the problem of computing the Pareto efficiency allocations keeping the utility of agent 1 equal to ¯u = πu1(w11) + (1 − π)u1(w12). The condition of Pareto optimality is again
u01(x11)
u01(x21) = u02(x12) u02(x22)
Example 7. Suppose in example 1, agente 2 is a perfectly discriminating monopoly u1(x) =√
x, w11 = 0, w12 = 1 u2(x) =√
x, w21 = 1 = w22 π = 1/4
w11= 0
w12= 1
w21= 1
w22= 1
π = 1 / 4
π = 1 / 4 π = 1 / 4 π = 1 / 4 1111−−−− π π π π
= 3 / 4Initial endowments Agent 1 Agent 2
u1(x) = x u2(x) = x
Then, agent 1 is kept at his reservation utility
¯ u1 = 1
4
√ 0 + 3
4
√ 1 = 3
4
On the other hand, this utility is attained at an allocation of the form
x11 = x21 2
That is,
3 4 = 1
4 q
x11+3 4
q
x21 = 1 4
q x11+3
4 q
2x11
That is,
3 = q
x11
1 + 3√ 2
and we obtain that
x11 = 9 1 + 3√
22
(1,2)
x11 x21
1
1
p1x12+ p2x22= c
Competitive allocation
Competitive Economy: General Case
inititial endowments
Agent 2 is a perfectly discriminating monopoly
5.1. One risk neutral agent and one risk averse agent. Suppose now that agent 2 is risk neutral. Assume this agent is an insurance company that behaves as a monopolist and agent 1 purchases insurance from agent 2. We define the certainty equivalent CE as the consumption level that leaves the agent indifferent between accepting the gamble and consuming CE. That is,
u1(CE) = πu1(w10− πD) + (1 − π)u1(w01) And the risk premium is
RP = EV − CE The amount paid by agent 1 to the monopolist is
w01− CE = EV +pe− CE = pe+ RP
Therefore, the insurance policy that the agent 1 purchases from the monopolist is (pm, Im) = (pe+ RP, D)
Example 8. Suppose that in example 1, agent 2 is a perfectly discriminating monopoly, who in addition is risk neutral.
u1(x) =√
x, w11 = 0, w12 = 1 u2(x) = x, w21 = 1 = w22
π = 1/4
w11= 0
w12= 1
w21= 1
w22= 1
π = 1 / 4
π = 1 / 4 π = 1 / 4 π = 1 / 4 1111−−−− π π π π
= 3 / 4Agent 1 Agent 2
u1(x) = x u2(x) = x Initial endowments
Pareto efficiency requires that
x11 = x21
The reservation utility of agent 1 is
3 4 = 1
4 q
x11+3 4
q x11 =
q x11
and we obtain that
x11 = CE = 9 16
and
x21 = 7
16 x22 = 23 16
(1,2)
x11 x21
1
1
p1x12+ p2x22= c p1=
ππππ
p2= 1 -ππππ
p1 x11 + p2 x21= c
Actuarially fair insurance
Competitive Economy: Agent 2 is risk neutral
inititial endowments
Agent 2 is a perfectly discriminating monopoly
EV CE
6. Labor Contracts Example 8 may be interpreted in the following way.
• In example 8 we may consider that agent 2 (the principal) has 1 unit of the good and hires agent 1 to do some work. In state e1, the worker is not efficient and produces 0 units of the good. In state e2, the worker is efficient and produces 1 unit of the good.
• The principal offers a wage w to the worker in a ‘take it or leave it’ offer. The worker accepts if his reservation utility uR1 = 3/4 is below the utility from accepting the contract.
• The principal pays to agent 1 the minimum amount that leaves the agent indifferent between accepting and his reservation utility, uR1 = 3/4. The utility function of agent 1 on monetary amounts is √
x. His utility will be equal to uR1 = 3/4 if the salary is w = 3
4
2
= 9 16 Agent 1 will accept this salary from the principal.
• On the other hand, agent 2, receives in each state the amount initial endowment + value of production − salary
so, in state e1 he receives
1 − 9 16 = 7
16 and in state e2 he receives
1 + 1 − 9 16 = 23
16