Macroeconomic Theory: Assignment 5: Value

Function Iterations

Exercise 1. [Value Function Iterations] Consider an economy where

the household chooses an in…nite sequence of consumption and next period’s
capital stock fct ; kt+1 g1
t=0 to solve the following sequential problem

1
X
t
maxfct ;kt+1 g1
t=0
u(ct )
t=0
ct + kt+1 = g(kt )
subject to ; for t = 0 : : : 1
ct ; kt+1 0

The utility function and the production function take the following form:

c1 1
u(c) = ,
1
g(kt ) = Akt + (1 )kt

The associated contraction mapping is de…ned as

T v(k) = max u(g(k) k 0 ) + v(k 0 )

k0 2X;k0 g(k)

Let A = 1, = 0:5, = 0:8, = 0:9, = 0:5. Follow the instructions

below to …nd the …xed point of the contraction mapping by using value function
iterations:

1. Follow instructions below to construct the …xed point of the contraction

mapping T :

Step 1. [De…ne the domain and image of the value function]: Dis-
cretize the domain of state variable by de…ning a vector
smaller grid => higer accuracy
X = 0 = k0 ; k1 ; k2 ; : : : ; kn = k : more times => higher accuracy

(Hint: the suggested grid size is 0.005.). Construct a vector v =

fv0 ; v1 ; : : : ; vn g to represent the initial value function v(k); that is,
vi = v(ki ), for i = 0; 1; : : : ; n.

1
 Step 2. [Given v(k), solve for T v(k)]: construct a vector

T v = fT v0 ; T v1 ; : : : ; T vn g;

where T vi = T v(ki ) such that

T v(ki ) = max u(g(ki ) k 0 ) + v(k 0 ) (?)

k0 2X;k0 g(ki )

Moreover, construct a policy vector pol = fpol0 ; pol1 ; : : : ; poln g, where

poli is the index of k 0 that solves (?).
Step 3. [Repeat Step 2. Conduct value function iterations]: Let
maxi fjT vi vi jg denote the sup norm. After computing , replace
v by T v, and replace the old policy vector by the new policy vector.
Repeat Step 2 for 100 times. What is the you obtain after 100 times
of iterations?

2. Apply the policy vector pol to …nd the dynamics of the economy. Let
A = 1, = 0:5, = 0:8, = 0:95, = 0:5. Suppose that the economy
was initially at the steady state (For t = 5; 4; 3; 2; 1). At time
t = 0, the time discount factor decreases from 0:95 to 0:9 unexpectedly
and permanently. Characterize the dynamic path of capital, consumption,
output, and investment before and after the TFP shock.
3. Compare the dynamic paths you drew in Question 1.2 (by using value
function iteration) with the dynamic paths you drew in Assignment 4,
Part 2 (by using steady state linearization). To compare, simply put
the dynamic paths under these two approaches in the same …gure and
check whether they are close enough.

Question 2: Suppose that T is a contraction mapping with modulus on

a space (V; ). Let vn = T n v0 and (v1 ; v0 ) = 1, and the …xed point of T is
denoted by v . Given > 0, what is the smallest number of iterations that
guarantees that (vn ; v ) < .