You are on page 1of 2

LECTURE 4: ENVELOPE THEOREMS: EXTENSIONS AND APPLICATIONS

Reference: Pemberton & Rau Section 18.2

Everything we did in Lecture 3 can be extended to many variables and many constraints.
Let f (x, t ) be a function of n  r variables x1 ,..., xn , t1 ,..., tr . Suppose there are two

constraints: we take t as given and choose x to maximise f (x, t ) subject to the constraints
g (x, t )  0 and h(x, t )  0 . The maximal value of the objective function subject to the
constraints depends on the vector t: call it v(t ) . The envelope theorem states that
v L
 for k  1,..., r ,
t k t k
where L is the Lagrangian, defined as follows:
L(x,  ,  , t )  f (x, t )  g (x, t )  h(x, t ) .
A similar result with three multipliers holds when there are three constraints; and so on
provided the number of constraints does not exceed the number of components of x.

More on the cost-minimising firm


The firm’s problem is to minimise w1 x1  ...  wn xn subject to the constraint

F ( x1 ,..., xn )  q . The Lagrangian for the problem is

L( x1 ,..., xn ,  , w1 ,..., wn , q )  w1 x1  ...  wn xn  F ( x1 ,..., xn )  q .

The firm’s cost function gives the minimum level of cost subject to the output constraint;
it is denoted by C ( w1, ..., wn , q ) . The marginal interpretation of the multiplier gives

C / q   . This result may also be obtained from the envelope theorem:


C L
 .
q q
Also by the envelope theorem
C L

wi wi

when x1 ,..., xn and  take their optimal values. Thus we get

C
 xi for i  1,..., n
wi

where xi is the optimal amount of input i employed. This result is known as Shephard’s
lemma.

More on indirect utility functions


The consumer’s problem is to maximise U ( x1 ,..., xn ) subject to the constraint

p1 x1  ...  pn xn  m .

The Lagrangian for the problem is


L( x1 ,..., xn ,  , p1 ,... pn , m)  U ( x1 ,..., xn )   ( p1 x1  ...  pn xn )  m .

The consumer’s indirect utility function gives the maximum level of utility subject to the
budget constraint; it is denoted by V ( p1, ..., pn , m) . The marginal interpretation of the

multiplier gives V / m   . This result may also be obtained from the envelope
theorem:
V L
 .
m m
The envelope theorem also gives
V
 xi for i  1,..., n .
pi
Putting this together with the marginal interpretation of the multiplier, we get
V V
xi   for i  1,..., n .
pi m

This result is known as Roy’s identity.

Exercises: Pemberton & Rau 18.2.3-18.2.4

You might also like