1.

Review of Demand Theory

1 Preferences
Let X be the consumption set, where X RL+ : An element of X is a con-
sumption bundle. Usually L = 2; which allows for a graphical representation
of bundles. Consumers will be able to rank consumption bundles by means of
their individual-speci…c preferences, denoted by % :
Preferences % on X are rational if they are:
1) Complete: the consumer is able to rank any two bundles x and y. 8x; y 2
X; x < y or y < x or both.
2) Transitive: 8x; y; z 2 X; if x < y and y < z; then x < z:
The set of bundles y such that both x < y and y < x is the indi¤erence
curve that contains bundle x: The MRS is the slope of the indi¤erence curve at
any point.
3) Re‡exive: 8x 2 X; x < x:
If a preference relation % on X satis…es all three properties, then it is ra-
tional.
4) Continuity: & on X is continuous if for any sequence f(xn ; y n )gn=1 with
1

xn < y n 8n; and x = lim xn ; y = lim y n ; then x < y:

n!1 n!1
Lexicographic preferences do not satisfy continuity. For instance, take xn =
( n1 ; 0) and y n = (0; 1): In this case, it is true that xn < y n 8n: However, lim xn =
n!1
(0; 0); whereas lim y n = (0; 1) and (0; 1) (0; 0):
n!1
5) Strong Monotonicity: & on X are strongly monotone if for all x; y 2 X;
if x y but x 6= y; then x y: Monotonicity: & on X are monotone if for
all x; y 2 X; if x > y; then x y: Local non-satiation: & on X are locally
nonsatiated if 8x 2 X, 8" > 0 9y 2 X s.t. ky xk < " and x y:
6) Convexity: & on X are convex if 8x 2 X; the upper contour set fy 2 X : y < xg
is convex, i.e. if y < x; z < x; y + (1 )z < x; 8 2 [0; 1]:
In words, the “preferred to” set is convex, i.e. a convex combination of any
two points in that set also belongs to that set.
Preferences that exhibit increasing MRS do not satisfy the convexity as-
sumption.
If 8x 2 X; y < x; z < x; y 6= z y + (1 )z < x; 8 2 [0; 1]; then the
consumer’s preferences are strictly convex.
7) & on X are homothetic if x y ) x y; 8 0:
8) & on X are quasilinear with respect to commodity 1 if
i) if x y ) (x + e1 ) (y + e1 ); where e1 = (1; 0; :::; 0) 8 2 R
ii) x + e1 x 8x; > 0:

1
2 Utility functions
If the preference relation is continuous, then there is a continuous utility function
u(x) that represents < :
Lexicographic preferences do not satisfy continuity, and thus, have no utility
representation.
The utility function that represents < is not unique. Any strictly increasing
transformation f (u(x)) also represents < : This implies that u(x1 ; x2 ) = x1 x2
and v(x1 ; x2 ) = ln x1 + ln x2 represent the same preferences. This is a conse-
quence of the fact that preferences are ordinal (and not cardinal) rankings of
bundles. We will normally assume u( ) to be twice continuously di¤erentiable.
If preferences are convex, then the utiliy function is quasiconcave, i.e. fy 2
RL+ : u(y) u(x)g is a convex set, 8x: Increasing MRS violates convexity and
hence, quasiconcavity. Quasiconcavity does not necessarily imply concavity.

3 Choice and demand

Given the utility function that represents the consumer’s preferences, we want
to determine his optimal choice. In order to properly de…ne the consumer’s
utility maximization problem, we need to specify his budget set, i.e. the set of
feasible bundles, given prices and income. Once prices and income are known,
the consumer makes his choice by solving his utility maximization problem:

maxu(x)
x 0
s.t. p x w

with p >> 0 and w > 0: The Walrasian budget set (the set of feasible alloca-
tions) is Bp;w = x 2 RL + :p x w
The solution to the consumer’s utility maximization problem is the Wal-
rasian demand correspondence (function if single-valued) x(p; w): If preferences
are not strictly convex, the Walrasian demand may fail to be a correspondence.
The indirect utility function is v(p; w) = u(x(p; w)), where x(p; w) is the Wal-
rasian demand.
Equivalently, we can de…ne the consumer’s expenditure minimization prob-
lem as

minp x
x 0
s.t. u(x) u

with p >> 0; u > u(0): The solution to this problem is the Hicksian demand cor-
respondence (or function) h(p; u): This is the Hicksian or compensated demand,
in the sense that the level of utility is maintained constant. When comparing
the Walrasian and Hicksian demands, slopes are di¤erent due to the income
e¤ect. The expenditure function is e(p; u) = p h(p; u).

2
 If w = e(p; u), then it holds that h(p; u) = x(p; w). At this point, the
Hicksian and Walrasian demands intersect.
There is a relationship between the two formulations of demand, which is
known as the Slutsky Equation:
@hl (p; u) @xl (p; w) @xl (p; w)
= + xk (p; w)
@pk @pk @w
which relates the slopes of the Walrasian demand function and the Hicksian
(compensated) demand function when l = k. In particular, the last term takes
into account the existence of income e¤ects. However, we will usually focus on
the utility maximization problem and on the Walrasian demand correspondence.

Example 1 Given a Cobb-Douglas utility function, the UMP becomes

max kx1 x12

x1 ;x2
s.t. p1 x1 + p2 x2 w

By applying a strictly monotonic transformation to the utility function, we

can rewrite it as

max ln x1 + (1 ) ln x2
x1 ;x2
s.t. p1 x1 + p2 x2 w

The Lagrangian of this problem is

$(x1 ; x2 ; ) = ln x1 + (1 ) ln x2 (p1 x1 + p2 x2 w)

The …rst-order conditions are

p1 0 = 0 if x1 > 0
x1
1
p2 0 = 0 if x2 > 0
x2
(p1 x1 + p2 x2 w) = 0

which implies that

1
= = ) p1 x1 = p2 x2
p1 x1 p2 x2 1
Making use of the budget constraint,
1 (1 )w w
p2 x2 = w ) x2 = ; x1 =
1 p2 p1
In some other cases, the nonnegativity constraints will be binding. An ex-
ample of this is the case of perfect substitutes. The consumer’s optimal choice
will typically be a corner solution, unless the consumer is indi¤erent between
consuming either good in which case, the solution is a correspondence.

3
Example 2 Suppose that the consumer’s utility function is given by u(x1 ; x2 ) =
x1 + x2 ; and that income is 10: Then, the consumer’s utility maximization prob-
lem is
max x1 + x2
x1 ;x2
s.t. p1 x1 + p2 x2 10
and the Lagrangian is $(x1 ; x2 ; ) = x1 + x2 (p1 x1 + p2 x2 10) : Then, the
…rst-order conditions are, respectively,
1 p1 0, = 0 if x1 > 0
1 p2 0, = 0 if x2 > 0
(p1 x1 + p2 x2 10) = 0
This leads us to consider four possible cases:
1) x1 = x2 = 0: Impossible, since the utility function is increasing.
2) x1 = 0; x2 > 0: Then, it must be true that 1 p1 < 0; = p12 =) pp12 > 1:
Using the budget constraint yields x2 = p102 ; x1 = 0:
3) x1 > 0; x2 = 0: Then, it must be true that 1 p1 = 0; 1 p2 < 0 =)
p1
p2 < 1: Using the budget constraint yields x1 = p101 ; x2 = 0:
4) x1 > 0; x2 > 0: Then, it must be true that 1 p1 = 0; 1 p2 = 0 =)
p1
p2 = 1 (a necessary condition). From the budget constraint we get x1 + x2 =
10 10
p2 ) x1 = p2 x2 , with x2 2 [0; p102 ]. In this case, the Walrasian demand is a
correspondence.
The Walrasian demand correspondence (function) has the following proper-
ties:
1) x(p; w) is homogeneous of degree zero in (p; w) : x( p; w) = 0 x(p; w):
2) x(p; w) satis…es Walras’law: p x = w8x 2 x(p; w):
3) If u( ) is strictly quasiconcave, then x(p; w) is a function.
In economic terms, the MRS is the ratio of marginal utilities, and the in-
terpretation of the multiplier is the marginal utility of wealth. The multiplier
is the marginal utility of wealth because the change in utility from a marginal
increase in w is given by

ru(x(p; w)) Dw x(p; w) = p Dw x(p; w) =

because p x(p; w) = w ) p Dw x(p; w) = 1:

4 Welfare analysis
A welfare analysis evaluates the e¤ects of changes in the consumer’s environment
on his well-being. Consider the welfare e¤ect of a price change, from p0 to p1 ,
given a constant wealth level w > 0. Thus, for any indirect utility function
v(p; w), and for an arbitrary price vector p, the expression
e(p; v(p1 ; w)) e(p; v(p0 ; w))

4
is itself an indirect utility function and provides a measure of the welfare change
expressed in monetary units. Depending on our choice of p; we have the equiv-
alent variation or the compensating variation:

EV (p0 ; p1 ; w) = e(p0 ; u1 ) e(p1 ; u1 ) = e(p0 ; u1 ) w

CV (p0 ; p1 ; w) = e(p0 ; u0 ) e(p1 ; u0 ) = w e(p1 ; u0 )

Using the fact that the Hicksian demand function is the derivative of the
expenditure function and that w = e(p0 ; u0 ) = e(p1 ; u1 ), we may rewrite, for
the case in which one of the prices decreases:
Z p01
EV (p0 ; p1 ; w) = h1 (p1 ; p 1; u
1
)dp1
p11
Z p01
0 1 0
CV (p ; p ; w) = h1 (p1 ; p 1; u )dp1
p11

If income e¤ects are absent, compensating and equivalent variations coincide,

and we call this the change in the Marshallian consumer surplus. Indeed, this
is the usual procedure when making welfare calculations, since they allow for
the use of the Walrasian demand function, which is observable. In this case, the
variation in consumer surplus is
Z p01
CS(p0 ; p1 ; w) = EV (p0 ; p1 ; w) = CV (p0 ; p1 ; w) = x1 (p1 ; p 1 ; w)dp1
p11

Otherwise, the consumer surplus will be bounded by the equivalent and

compensating variations. For the case of a decrease in p1 , we get: EV > CS >
CV .