The Slutsky equation implies
Substitution e?ect:
?ˉx
i
(p,u)
?p
i
; Income e?ect: ?
?x
?
i
(p,I)
?I
· x
?
i
.
Proposition 1.22. (Compensated Demand).
(1) ˉx(p, u) is zero homogeneous in p.
(2) substitution matrix: D
p
ˉx(p,u) ≤ 0.
(3) symmetric cross-price e?ects:
?ˉx
j
(p,u)
?p
i
=
?ˉx
i
(p,u)
?p
j
.
(4) decreasing ˉx :
?ˉx
i
(p,u)
?p
i
≤ 0.
Proposition 1.23. (Ordinary Demand).
(1) Homogeneity: x
?
(p,I) is zero homogeneous in (p,I).
(2) Compensated Symmetry:
?x
?
j
?p
i
+ x
?
i
?x
?
j
?I
=
?x
?
i
?p
j
+ x
?
j
?x
?
i
?I
.
(3) Adding-up:
null
p
i
x
?
i
(p,I)=I.
3. Uncertainty Theory
3.1. Introduction
Example 1.8. Flip a coin once:
if tail, get $0; if head, get $100.
i.e., lottery nullx ≡ (0,
1
2
;100,
1
2
). The expected income is E(game)=50. Let u(x)=
√
x
be the personal value of x. The expected utility is Eu(nullx)=5. Let u(e)=Eu(nullx). Then,
e =25.eis the certainty equivalent.
1—13
3.2. Expected Utility
Lottery x =(x
1
,p
1
; x
2
,p
2
; ...; x
n
,p
n
): x
i
with probability p
i
and
null
p
i
=1.
Axiom 1. (x,1; y,0) = x.
Axiom 2. (x,p; y,1?p)=(y,1?p; x,p).
Axiom 3 (RCLA). [(x,p; y,1?p),q; y,1?q]=(x,pq; y,1?pq).
Using Axioms 1-3, the lottery space L is well defined. Suppose the consumer has
a preference relation null on L.
Axiom 4. null is complete.
Axiom 5 (Continuity). {p ∈ [0, 1] | (x,p; y,1?p) null z} and {p ∈ [0, 1] | (x,p; y,1?
p) null z} are closed sets in [0, 1].
Axiom 6 (Independence). x ~ y =? (x,p; z,1?p) ~ (y,p; z,1?p).
The expected utility property:
u[(x,p; y,1?p)] = pu(x)+(1?p)u(y).
Theorem 1.1. (Expected Utility Representation). If (null,L) satisfies Axioms 1—6,
there exists a utility representation u : L → R that has the expected utility property.
Example 1.9. For x ≡ (x
1
,p
1
;...;x
n
,p
n
), the expected utility is
u(x) ≡
null
p
i
u(x
i
).
With a perfect capital market, the firm cares only about the expected money value:
u(x) ≡
null
p
i
x
i
.
Any monotonic linear transform v : L → R preserves the expected utility property:
v(x)=au(x)+b, a > 0.
Proposition 1.24. (Uniqueness). Expected utility is unique up to a monotonic linear
transformation: u(·) and v(·) are expected utility representations of the same preference
relation i? there exist a>0 and b ∈R such that v(·)=au(·)+b.
1—14
Example 1.10. (Allais Paradox). Consider the following four lotteries:
0.1 0.01 0.89
x 5m 0 0
y 1m 1m 0
z 5m 0 1m
w 1m 1m 1m
Example 1.11. (Common Ratio E?ect). Consider the following four lotteries:
x = (3000,1),
y = (4000,0.8; 0,0.2),
z = (3000,0.25; 0,0.75),
w = (4000,0.2; 0,0.8).
3.3. Risk Aversion
Definition:
u[E(?x)] >Eu(?x) if risk averse
u[E(?x)] = Eu(?x) if risk neutral
u[E(?x)] <Eu(?x) if risk loving
For ?x ≡ (x, p; y, 1?p),y>x,he can buy an insurance z : he pays z in any
event, and is paid y?x when the bad event happens. The maximum insurance P that
he is willing to buy is the insurance premium, satisfying
E[u(?x)] = u(y?P).
Then,
E(?x) >y?P if risk averse
E(?x)=y?P if risk neutral
E(?x) <y?P if risk loving
1—15
Therefore,
u is concave if risk averse
u is linear if risk neutral
u is convex if risk loving
Themoreconcaveu is, the higher P is, or the more risk averse the consumer is.
Thus, we propose two measures of risk aversion:
absolute risk aversion: R
a
(x) ≡?
u
00
(x)
u
0
(x)
,
relative risk aversion: R
r
(x) ≡?
xu
00
(x)
u
0
(x)
.
They are for two types of fluctuations: ˉx ± ε or ˉx(1 ± ε).
Let ˉx ≡ E(?x). Define the risk premium π
a
:
u(ˉx?π
a
)=E[u(?x)]. (3.1)
Let nullx ≡ ˉx + ε. Then,
π
a
≈
1
2
σ
2
ε
R
a
(ˉx).
Similarly, define the relative risk premium π
r
:
u[ˉx(1?π
r
)] = E[u(?x)].
Let nullx =ˉx(1 + ε). Then,
π
r
≈
1
2
σ
2
ε
R
r
(ˉx).
Example 1.12. 14. The Demand for Insurance. For (w?l, p; w, 1?p), the consumer
can buy insurance q at price π :
Prob. p 1?p
Before w?lw
After w?l + q?πqw?πq
The consumer’s problem is
max
q
pu(w?l + q?πq)+(1?p)u(w?πq).
1—16
Assume a competitive market: π = p. With u
00
< 0, we have q
?
= l.
Alternatively, let I
1
≡ w?πq and I
2
≡ w?l+(1?π)q. Then, the problem becomes
?
?
?
?
?
max (1?p)u(I
1
)+pu(I
2
)
s.t. (1?π)I
1
+ πI
2
=(1?π)w + π(w?l).
In a competitive insurance market, π = p, implying I
?
1
= I
?
2
, i.e., full insurance.
4. Equilibrium Theory
See MWG (1995), Chapters 15—17 and 19.
Arrow-Debreu world:
1. Complete market: any consumption bundle is obtainable.
2. Perfect market: no frictions such as transaction costs, taxes, etc.
3. Perfect competition: economicagentstakepricesasgiven.
4. Symmetric Information: same information for all.
5. Private consumption.
4.1. Equilibrium in a Pure Exchange Economy
A pure exchange economy:
(1) k commodities j =1, 2,...,k;
(2) n consumers i =1, 2,...,n, with (u
i
,w
i
), where u
i
: R
k
+
→ R and w
i
∈R
k
+
.
Any x =(x
1
,x
2
,...,x
n
),x
i
∈R
k
+
, is an allocation. An allocation x is feasible if
n
null
i=1
x
i
≤
n
null
i=1
w
i
.
The Edgeworth box contains all the allocations satisfying
x
1
A
+ x
1
B
= w
1
A
+ w
1
B
,x
2
A
+ x
2
B
= w
2
A
+ w
2
B
,x
j
i
≥ 0.
It has two key features:
? Each point represents a feasible allocation.
? Both persons share the same budget line.
1—17