Scribd
Upload
10 views

Envelope-theorem-F24

The document discusses the Envelope Theorem in the context of macroeconomic dynamic programming, particularly focusing on maximizing a function with respect to a parameter. It provides examples using the Bellman equation, illustrating how the theorem applies to optimal solutions and value functions. The document emphasizes the relationship between derivatives of the maximum achievable value and the parameters involved in the optimization problem.

Uploaded by

gimdongeun282
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
10 views

Envelope-theorem-F24

The document discusses the Envelope Theorem in the context of macroeconomic dynamic programming, particularly focusing on maximizing a function with respect to a parameter. It provides examples using the Bellman equation, illustrating how the theorem applies to optimal solutions and value functions. The document emphasizes the relationship between derivatives of the maximum achievable value and the parameters involved in the optimization problem.

Uploaded by

gimdongeun282
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

HSS310 MACRO & DP Hyung Seok E.

Kim

KAIST
Macro and Dynamic Programming
Fall, 2024
The Envelope Theorem

Consider the problem:

max f ( x, a)
x∈X

where f : X × A −→ R is differentiable and a ∈ A is viewed as a “parameter.” Let x ∗ ( a) be


the optimal choice conditional on a, and let M ( a) ≡ f ( x ∗ ( a) , a) be the maximum achievable value.
Suppose that x ∗ ( a) is single-valued and differentiable around a point ā . Then

d ∂
M ( a) = f ( x∗ ( a) , a) .
da a= ā ∂a a= ā

Example 1 Consider the Bellman equation: i.e. in any period t,

V (k t ) = max F (k t , k t+1 ) + βV (k t+1 ) (1)


k t +1 ∈ Γ ( k t )

Suppose that F and V are differentiable. Given k t , let k∗t+1 ≡ k∗t+1 (k t ) be an optimal solution to the right-hand
side of the Bellman equation and differentiable around k̄. Also let M (k t ) be the maximum achievable value given
the “parameter” k t :
M (k t ) ≡ F k t , k∗t+1 (k t ) + βV k∗t+1 (k t ) .
 

Then, by the envelope theorem,

d ∂
[ F k t , k∗t+1 (k t ) + βV k∗t+1 (k t ) ]
 
M (k t ) =
dk t k t =k̄ ∂k t k t =k̄
∂
F k t , k∗t+1 (k t )

=
∂k t k t =k̄
∂ ∗

= F k̄, k t+1 k̄ .
∂k t
Since V (k t ) is a unique value function, V (k t ) = M (k t ). Thus, we have

d ∂
F k̄, k∗t+1 k̄ .

V (k t ) =
dk t k t =k̄ ∂k t

Page 1
HSS310 MACRO & DP Hyung Seok E. Kim

Example 2 Consider the right-hand side of the Bellman equation (1) again. Assuming that the right hand side
of (1) is attained in the interior of Γ (k t ), its first-order condition for an optimal solution k∗t+1 is given by

∂ ∂
F k t , k∗t+1 + β V k∗t+1 = 0.
 
(2)
∂k t+1 ∂k t

Now consider the Bellman equation in period t + 1: given any k t+1 ∈ Γ (k t ) ,

V ( k t +1 ) = max F (k t+1 , k t+2 ) + βV (k t+2 ) .


k t +2 ∈ Γ ( k t +1 )

As in Example 1, given k t+1 , let k∗t+2 ≡ k∗t+2 (k t+1 ) be an optimal solution to the right-hand side of the Bellman
equation and differentiable but around k∗t+1 in (2). The envelope theorem implies

∂ ∂ ∂
V k∗t+1 = F k∗t+1 , k∗t+2 k∗t+1 .
 
V ( k t +1 ) = (3)
∂k t+1 k t+1 =k∗t+1 ∂k t+1 ∂k t+1

Combining (2) and (3), we obtain

∂ ∂
F k t , k∗t+1 + β F k∗t+1 , k∗t+2 k∗t+1 = 0.
 
∂k t+1 ∂k t+1

Page 2

You might also like