**Unformatted text preview: **SimpleModelClassHomework April 19, 2018 1 Simulating a macroeconomic model in Python Recall that you can run all of this code yourself!
• First, open a web brower and go to the website:
– Click "Try Jupyter With Python"
• Second, after the notebook page has loaded, you can remove the pre-filled cells by clicking
them and hitting the "cut" button in the toolbar at the top of the page (it has a pair of scissors
on the button).
• Third, copy a code cell from this PDF, paste into a cell in the notebook, and then hit "run"
(again, in the toolbar at the top of the page)
– Note that each code cell that has been "run" has a number alongside, e.g. [1]
– You can create new cells by hitting "add" (it has a plus (+) symbol on the button).
• Finally, you can always save your work via: "File">"Download as">"HTML (.html)".
– Unfortunately, saving as PDF doesn’t seem to work from the web browser
– Note that using the web broswer version of the Notebook, it is not straightforward to
load your saved work.
– Instead, save as an HTML file, and next time you run the code, simply copy the code
cells across to a new Notebook that you load from the web browser.
1.0.1 Simulate Impulse Response Functions for the IS/MP/AS model
Consider the IS/MP/AS model with inflation, from class Lecture 10 slides (Slides 11-12). We also
include a simple Taylor Rule (slide 22).
The model is:
y∗t = ηat + x
1
rt∗ = r ∗ + (y∗t+1 − y∗t + dt )
σ
yt − y∗t = −σ (rtn − πt+1 − rt∗ ) + yt+1 − y∗t+1
πt = λ(yt − y∗t ) rtn = r ∗ + π 0 + ϕπ (πt+1 − π 0 )
where the final equation is the Taylor rule (using expected inflation targeting).
In this model we assume that the agents in the economy have perfect foresight over future output gaps and future inflation. This allows us to solve the model simply, using backward iteration.
1 2 Set model parameters In [1]: # Importing Python Libraries
import numpy as np
import matplotlib.pyplot as plt
In [2]: # Set model parameter values
rnstar = 0.01 # Natural rate of interest
phi_pi = 3 # coefficient in the Taylor rule pi0 = 0 # Steady state rate of inflation sigma = 0.5
lammbda = 0.75
eta = 0.75
x = -0.05 #
#
#
# rho = 0.75 3 risk aversion in the utility function
Philip's curve coefficient on the output gap
LRAS curve coefficient on productivity shocks
LRAS curve constant parameter # AR(1) persistence parameter Simulate demand and productivity shocks
• Suppose the shocks are generated by an AR(1) process:
dt = ρdt−1 + ε d,t
• ρ is a parameter describing the persistence of shocks to demand
• ε d,t is a shock to demand.
• Notice that a single shock to demand (ε d,t ) affects today’s demand, dt , but because of the
persistence of the shock, it will also affect tomorrow’s demand because:
dt+1 = ρdt + ε d,t+1 = ρ(ρdt−1 + ε d,t ) + ε d,t+1
• When simulating an Impulse Response Function, we suppose there is a single shock to demand, and we trace out its impact over time In [3]: # 1. Simulate a_t and d_t
T = 50 # number of simulation periods
a_t = np.zeros(T)
d_t = np.zeros(T)
eps_a_t = np.zeros(T)
eps_d_t = np.zeros(T)
# Consider a demand shock
eps_d_t[1] = 0.01
for tt in range(1,T):
2 d_t[tt] = rho*(d_t[tt-1] + eps_d_t[tt])
plt.plot(d_t)
plt.title('Demand shock')
plt.show() 4 Simulate the model
•
•
•
• We use backward iteration to solve the model
Assume the model has returned to steady state in the final period
Iterate backwards given the values of inflation and the output gap in the period ahead
Note that flexible price output and the natural rate of interest can be simulated immediately
since they only depend on the sequence of productivity and demand shocks In [4]: # 2. Simulate natural rate of output and interest
ystar_t = eta*a_t + x
rnstar_t = rnstar*np.ones(T)
for tt in range(0,T-1):
rnstar_t[tt] = rnstar + 1/sigma*(ystar_t[tt+1] - ystar_t[tt] + d_t[tt])
In [5]: # 3. Backwards iteration on the model taking current path of inflation as given 3 pi_t = pi0*np.ones(T+1)
x_t = np.zeros(T+1)
rn_t = (rnstar + pi0)*np.ones(T+1)
for ss in range(1,T+1):
rn_t[T-ss] = rnstar + pi0 + phi_pi*(pi_t[T-ss+1] - pi0)
x_t[T-ss] = -sigma*(rn_t[T-ss] - pi_t[T-ss+1] - rnstar_t[T-ss]) + x_t[T-ss+1]
pi_t[T-ss] = lammbda*x_t[T-ss] 5 Plot the Impulse Response Function for each of the model variables In [6]: plt.figure(figsize=[15,10])
plt.subplot(2,3,1)
plt.plot(a_t)
plt.title('Productivity shock')
plt.subplot(2,3,2)
plt.plot(d_t)
plt.title('Demand shock')
plt.subplot(2,3,3)
plt.plot(x_t)
plt.title('Output gap')
plt.subplot(2,3,4)
plt.plot(x_t)
plt.title('Inflation')
plt.subplot(2,3,5)
plt.plot(rn_t)
plt.title('Nominal interest rate')
plt.show() 4 5 ...

View
Full Document

- Fall '19
- Inflation, Monetary Policy, Impulse response, T-1