(a) ‘o’ 33% E} I 1. Consider the deterministic BrockMirman model studied in class. The discount factor is ,3 E (0,1), the per-period utility function…

1. Consider the deterministic Brock-Mirman model studied in class. The discount factor is

β ∈ (0, 1), the per-period utility function is u(c) = log(c) and the production function is

f(k) = kα + (1 − δ)k, where δ = 1 and α ∈ (0, 1). Assume β = 0.99 and α = 0.3.

(a) Compute the model’s steady-state k∗.

(b) Use the F and G that we derived in class by solving the Bellman equation by guess and

verify to plot the optimal decision rules for k and c in the range [0.005, ¯k], where ¯k is

the upper bound on the state space. Can you figure out such ¯k. If not, set ¯k to twice

the steady-state capital stock.

(c) Let the initial capital stock be k0 = 0.5k∗. Compute the time series for kt, t = 1, 2, …, 200.

Plot the time series. Now do the same for k0 = 1.5k∗. Do the sequences converge to k∗?

2. Now solve the model in question (1), with the same parameter values, by discretising the state

space and iterating on the Bellman equation (use convergence criterium for the sup norm of

1e−5). For each iteration store the value of the sup norm (you may also want to print it on

the screen to see how the procedure is doing at each iteration). Use the state space you used

in (1) (b) and set the number of grid points (N + 1) to 401.

(a) Plot the optimal decision rules for k and c in the same graph as those in (1) (b). If you

have done it correctly, they should approximately agree.

(b) Plot the vector of the values of the sup norm against the the index denoting each

iteration. You should see that the sup norm declines monotonically. Would you expect

such a result? Why or why not?

(c) Now that you know your code works, solve the model again for a realistic depreciation

rate δ = 0.025 (you need to re-calculate the steady-state and adjust the state space

accordingly). In a new graph, again plot the decision rules. They should look more

linear than before as the resource constraint is more linear. The decision rules should be

especially close to linear in the neighborhood of k∗, the steady-state capital stock. (Note:

This justifies the use of LQ approximation methods we will discuss later in the module.)

Don’t worry about the step-like features in the optimal consumption rule—these can be

‘smoothed’ out by increasing N.

3. Now solve the model with f(k) = zkα + (1 − δ)k, where z ∈ {1.1, 0.9} and the transition

matrix between these two values is

P =

0.95 0.05

0.05 0.95


Plot the value function and the decision rules.

Looking for a Similar Assignment? Our Experts can help. Use the coupon code SAVE30 to get your first order at 30% off!