[go: up one dir, main page]

0% found this document useful (0 votes)
48 views9 pages

Marshallian vs Hicksian Demand Functions

CUHK Microeconomic Theory Class Lecture 3

Uploaded by

liuyunshu93
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
48 views9 pages

Marshallian vs Hicksian Demand Functions

CUHK Microeconomic Theory Class Lecture 3

Uploaded by

liuyunshu93
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Lecture 3

0.1 Relationship Between Marshallian and Hicksian De-


mand Functions
1. The indirect utility function and expenditure function are intimately con-
nected. So are the Marshallian and Hicksian demand functions. We have
already seen that the two demand functions share a common tangency
condition. Regardless of whether one’s objective is to maximize utility or
minimize expenditure, one should consume in such a way that the mar-
ginal utility per dollar spent on each good is the same.
2. Theorem 1.8. Suppose u is strictly increasing.

(a) If x is a solution to the maximization problem max u (x) s:t: px y,


then it is a solution to the minimization problem min px s.t. u (x)
u (x ). Hence, e (p; v (p; y)) = y:
(b) If x is a solution to the minimization problem min px s:t: u (x)
u , then it is a solution to the maximization problem max u (x) s.t.
px px . Hence, v (p; e (p; u)) = u:

3. Part 1 says that if the maximum utility the consumer can obtain given p
and y is v, then the minimum expenditure needed to achieve v is y. Part
2 says that if the minimum expenditure needed to achieve utility u given
p is e, then maximum utility one can obtain given p and e is u.
4. Proof: I will only prove part a. Suppose x is not a solution to the
minimization problem, and there exists x0 such that

px0 < px

and
u (x0 ) u (x ) :
Then, for some small " >> 0;

p (x0 + ") < px;


0
u (x + ") > u (x ) ;

a contradiction. Note that v (p; y) = u (x ) and px = y.


5. Part 2 can be proven similarly.
6. It follows that when both x (p; y) and xh (p; u) are unique,

(a) x (p; y) = xh (p; v (p; y)).


(b) xh (p; u) = x (p; e (p; u)).

1
7. The …rst relation says that the Marshallian demand at prices p and income
y is equal to the Hicksian demand at prices p and the maximum utility
level that can be achieved at prices p and income y. The second says that
the Hicksian demand at any prices p and utility level y is the the same as
the Marshallian demand at those prices and an income level equal to the
minimum expenditure necessary to achieve the utility level.
8. Note that the theorem is incorrect if the utility function is not strictly
increasing. For example: if u (x) = 10 for all x, then

v (p; 5) = 10;
e (p; 10) = 0:

9. There are two ways to derive the Hicksian demand. The direct way is to
solve expenditure minimization problem. But we can also get the Mar-
shallian demand by solving the utility maximization problem. Substitute
the demand back into the utility function. This gives the indirect utility
function. From the indirect utility function, we know how much income
need to change to keep utility constant as price change. Substitute this
back into the Marshallian demand gives the Hicksian demand.
10. Cobb-Douglas example: u(x1 ; x2 ) = x y , + =1

y y
x1 = ;y =
p1 p2
The indirect utility function is

y y
v (I; p1 ; p1 ) = u(x (p1 ; p2 ; y) ; y (p1 ; p2 ; y)) =
p1 p2

= y
p1 p2

Rearranging the terms

u p1 p2
e(px ; py ; u) = = u:
p1 p2

The Hicksian demand for Cobb-Douglas utility function is

1 1
p1 p2 p1 p2
xh1 = u; xh2 = u:

2
0.1.1 Income and Substitution E¤ects
1. Di¤erentiating Thm 1.9.2 with respect to pj yields the Slutsky equation:

@xhi @xi @xi @e


= +
@pj p;v(p;y) @pj p;y @y p;y @pj p;v(p;y)
@xi @xi
= + xhj (p; v (p; y)) (by Sheppard’s lemma)
@pj p;y @y p;y
@xi @xi
= + xj (p; y) (by Thm 1.9).
@pj p;y @y p;y

Rearranging the terms, we have


@xi @xhi @xi
= xj (p; y) :
@pj p;y @pj p;v(p;y) @y p;y

The Slutsky equation decompose the e¤ect of the price of good j on the
quantity demand of good i into a substitution e¤ect and an income e¤ect.
The former captures the e¤ect of relative price, and the latter the e¤ect
of purchasing power. (Illustrate the two e¤ects in a diagram.)
2. To separate the income e¤ect from the substitution e¤ect, we can imagine
that the consumers are compensated for the change in purchasing power
due to a change in price so that their utility remain unchanged, and study
how their demand for a good will change purely as a result of relative
price changes. Hence, the Hicksian demand is also called the compensated
demand, and the Marshallian Demand uncompensated demand.
3. Recall that the own price substitution e¤ect is always negative, and the
income e¤ect can be positive or negative. Thus, (??) tells that for normal
goods ( @x
@y
i
> 0),
p;y
@xi @xhi
<
@pj p;y @pj p;v(p;y)

meaning that the Marshallian demand has a steeper slope. The opposite
is true for inferior goods.

0.1.2 Gi¤en Goods


@xi
1. Note that in the above equation there is no guarantee that @pj must
p;y
be negative. We refer to a good with a positive-sloped demand curve at
some price as a Gi¤en good at that price. (Obviously no good could have
a demand curve that is always positive.)
@xi @xi
2. Thus, @pj can be positive only when @y is negative. Put di¤er-
p;y p;y
ently, a Gi¤en good must be an inferior good. You can see from (??) why

3
Gi¤en good is hard to …nd. For the income e¤ect to dominate the negative
substitution e¤ect, in addition to a negative income e¤ect, the quantity
demand must be su¢ ciently large as well.
3. The idea of Gi¤en goods was …rst raised by Marshall in the third editions
of Principles of Economics (1895):
“There are however some exceptions. For instance, as Mr. Gi¤en has
pointed out, a rise in the price of bread makes so large a drain on the
resources of the poorer labouring families and raises so much the marginal
utility of money to them, that they are forced to curtail their consumption
of meat and the more expensive farinaceous foods: and, bread being still
the cheapest food which they can get and will take, they consume more
and not less of it. But such cases are rare; when they are met with they
must be treated separately.”
4. Two things that are interesting. 1. Marshall’s comment captures the right
intuition of the Slutsky equation. 2. The Mr. Gi¤en Marshall referred to
is the economist Sir Robert Gi¤en. There is actually no written record
that Gi¤en had ever written anything about the goods that now bear his
name.
5. Here is a simple example that captures Marshall idea. Suppose Wangwang
only consumes two types of food: beef and rice. For any consumption
bundle (b,r), the total satisfaction (S) and calories (C) are S(b,r)=ab+r
and C(b,r)=b+r, with a>1. That is, for equal amount of calories, beef
generates more satisfaction than rice. To survive Wangwang needs at
least q calories. So a consumption bundle that contain q calories or more
is always strictly better to one that contain less than q calories. If both
bundles contain more than q calories, then he strictly prefers the one that
generates a higher satisfaction. Analyze Wangwang’s behavior graphically.
6. Examples of Gi¤en goods in real life are hard to come by. Economists
used to cite potatoes during the Irish Potatoes (1845-1849) famine caused
by a fungus that destroyed a large part the potato crop. Modern research
has cast doubt on the claim. One reason is that since the total output
of potato dropped signi…cantly, it would be impossible for the poor to
consume more potato as a group.
7. Previous research also …nd evidence of Gi¤en behavior in rats.
8. In a recent paper (Gi¤en Behavior and Subsistence Consumption, AER,
September 2008) Robert Jensen and Nolan Miller present for the …rst time
evidence of Gi¤en good in Hunan and (to a weaker extent, Gansu).
9. The authors o¤er …ve months of price subsidies of a stable food to poor
urban families (with an income of one to two hundred yuan per person
per month) in Hunan and Gansu. The families use a voucher to purchase
rice (in Hunan) and wheat ‡our (in Gansu) at a discount. The authors

4
conduct three rounds of survey on food consumption and income. The …rst
round takes place before the subsidy scheme begins; the second round two
months into the scheme, and the …nal round several months after. They
then estimate a panel regression on the relationship between %change in
price and %change in consumption of the stable food.

10. They …nd that in the case of Hunan the coe¢ cient is signi…cantly positive
for those who consume less than 80% of calories from sources other than
rice. Speci…cally, a one percent increases lead to a 0.45% increase in the
consumption of rice. Those who consume more than 80% of calories in the
form of rice essentially consume only rice and do not (or cannot) exhibit
Gi¤en behavior.

0.1.3 The Slutsky Substitution Matrix


1. Given a Marshallian demand function x (p; y), let

@xi @xi @e
sij + :
@pj p;y @y p;y @pj p;v(p;y)

2. The Slutsky matrix is de…ned as

s11 s12 : : s1n


s21 :
: : :
: :
sn1 snn

3. Theorem: Suppose x (p; y) is a Marshallian demand function generated


by some continuous, strictly increasing utility function. Then the Slutsky
matrix of x is symmetric and negative semide…nite.
4. Proof: The proof is not hard, but it makes use of several results that we
have just learnt. From the Slutsky equation, we have

@xhi
sij = :
@pj

(The reason we need the utility function to be strictly increasing or at


least locally non-satiated, is that without it the Slutsky equation does not
hold.) By the Sheppard’s lemma, we have for any i and j

@2e @xhi @xhj


= = :
@pi @pj @pj @pi

5
This implies that the Slutsky matrix is symmetric. Finally, since e is
concave, its Hessian matrix:
@xh @xh @xh
1
@p1
1
@p2 : : 1
@pn
@xh
2
@p1 :
: :
: :
@xh
1 @xh
n
@pn @pn

must be negative semide…nite.


5. Recall that that the Hessian matrix of a function f is
2 @f 1 (x ) @f 1 (x ) 1 3
@x1 @x2 : : @f@x(xn )
6 @f 2 (x ) @f 2 (x ) 7
6 7
6 @x1 @x2 7
H 6 : : 7
6 7
4 : : 5
@f m (x ) @f m (x )
@x1 @xn

It is negative semide…nite if for all y


y T Hy 0:

0.2 Compensating Variation, Equivalent Variation, and


Consumer Surplus
1. The expenditure function provides two ways to measure how a change in
price a¤ects consumer welfare.
2. Let p0 and p1 denote the initial and …nal prices, and let u0 and u1 denote
the corresponding utility level. The income y is held constant.

(a) Compensating Variation (CV) associated with a change in the price


of good i is the income compensation required to keep the utility of
a consumer constant. That is,
CV p0 ; p1 ; y e(p1 ; u1 ) e(p1 ; u0 )
= e(p0 ; u0 ) e(p1 ; u0 )
= y e(p1 ; u0 ):
@e
By Shephard’s Lemma, @p i
= xh . It follows that when only the price
of good i is changed
Z p0i
@e
CV = pi ; p i ; u0 dpi ;
pi @p i
Z p0i
= xh pi ; p i ; u0 dpi :
pi

6
Thus, the compensating variation is the area to the left of the Hick-
sian demand between pi and p0i . (Draw diagram)
(b) Equivalent Variation (EV) associated with a change in the price of
good i is the income compensation that the consumer considers as
equivalent to the price change. That is

EV p0 ; p1 ; y e(p0 ; u1 ) e(p0 ; u0 )
= e(p0 ; u1 ) e(p1 ; u1 )
= e(p0 ; u1 ) y:

When only the price of good i is changed


Z p0i
@e
EV = (pi ; p i ; u1) dpi ;
pi @pi
Z p0i
= xh pi ; p i ; u1 dpi :
pi

See diagram. EV is commonly used to measure the deadweight loss


of taxation. More below.

3. Note that for any p0 and y, we can take EV as an indirect utility function.
For any (p0 ; y 0 ) and (p00 ; y 00 ),

e(p0 ; v (p0 ; y 0 )) y e(p0 ; v (p00 ; y 00 )) y

if and only if v (p0 ; y 0 ) v (p00 ; y 00 ) :


4. But the same is not true of CV. See the following example from an old
mid-term exam.
5. Jingjing consumes only co¤ee and rice, and her utility function is

0 if c 5;
u (c; r) =
r if c > 5;

where c denotes the number of cups of co¤ee she drinks and r the number
of bowls of rice she eats. Jingjing’s indirect utility function v (p; y) and
expenditure function e (p; u) are

0 if 5pc y;
v (pc ; pr ; y) y 5pc
pr if 5pc > y;

and
0 if u = 0;
e (pc ; pr ; u)
5pc + upr if u > 0:
The compensating variation for Jingjing is de…ned as CV p0 ; p1 ; y
y e(p1 ; v p0 ; y ), where p0 is the initial price and p1 is the …nal price.

7
Suppose initially the price of 1 cup of co¤ee and the price of a 1 bowl of
rice are both 1, and her income is 10. Her initial utility is then 5. Now
consider (p0c ; p0r ) = (2; 1) and (p00c ; p00r ) = (1; 3). In this case, v (p0 ; 10) =
0 < 5=3 = v (p00 ; 10) :But, CV (p0 ; 10) = 5 > 10 = CV (p00 ; 10). What’s
the intuition behind this example?
6. Deadweight Loss. The deadweight loss associated with a tax t on good i
is commonly measured by
EV (pi ; pi + t; y) txi (pi + t)
= y e (pi ; v (pi + t; y)) txi (pi + t) ;
that is, the di¤erence between the lump sum a consumer is willing to pay
in lieu of the tax and the actual of tax collected. This is the e¢ ciency
gain if the govt. replaces t by a lump sum tax that generates the same
amount of tax. See …gure. This is often called the Harberger Triangle.
For a small tax, the DWL is proportional to the square of the tax rate.
7. Since Hicksian demand is not directly observable, it is more convenient to
measure welfare change using Marshallian rather than Hicksian demand.
The consumer surplus (at pi ), CS, is the area bound by the Marshallian
demand and pi . The change in CS when the price of good i changes from
pi to p0i is
Z p0i
4CS p0 ; p1 ; y = x (pi ; p i ; u) dpi :
pi
Recall from the Slutsky equation that
@xi @xhi @xi
= xi (p; y) :
@pj p;y @pj p;v(p;y) @y p;y
0
For normal goods, xi (p0i ; y) < xhi pi ; v (pi ; p i ; y) when p0i > pi ; and
0
xi (p0i ; y) < xhi pi ; v (pi ; p i ; y) when p0i < pi . The converse is true for
normal goods. Hence, for normal goods
EV < j4CSj < CV
when p0i > pi , and
EV > j4CSj > CV
when p0i < pi .
8. Although, j4CSj does not equal CV or EV, in practice the di¤erence
is pretty small and is insigni…cant compared to measurement errors. For
1 1
example, consider the Cobb-Douglas utility function x12 x22 . The Marshal-
lian demand x, the compensated demand xh , and the expenditure function
e are given by
1
I p2 2
1
x1 = , xh1 = u and e = 2u(p1 p2 ) 2 :
2p1 p1

8
p00
Suppose p1 increases from p01 to p001 , where 1
p01 = 1:5:

Z p00
1
1 p2 2
y 0 1
CV = u dp1 where u = (p p2 ) 2

p01 p1 2 1
!
1=2
p2 1
= y 1 = y((1:5) 2 1)
p1
: 0:225y;

whereas
Z p00
1 y
CS = dp1
p01 2p1
p001
= y=2(ln ) = y=2(ln 1:5)
p01
; 0:20y:

9. Suppose p1i > p0i and p1j = p0j for all j 6= i. Note that downward sloping
Hicksian demand implies that
Z p1i
1 0 0 0 @e
e p ;u e p ;u = dpi
p0i @pi
< xhi p0 ; u0 p1i p0i .

The increase in expenditure is always less than the amount that is neces-
sary to keep the original bundle a¤ordable under the new price.

You might also like