Problem Set 5
TA: Markus Münch,
Graduate Macroeconomics I, ECONS427
November 20, 2024
Problem 1. [40%] Consider an P economy populated by infinitely
lived agent. Assume that
their utility is expressed by E0 t β t log ct − BI ht = h̄ . These agents derives utility
from consumption and disutility from labor. When they work for h̄ (constant) the disutility
of labor is B. If they do not work, we can set ht to zero, meaning that the disutility of labor
is equal to zero. The indicator function I is equal to one if the event in question is true, and
zero if it is not.
Let us assume that the agent’s wage in a given period is the product wϵt , where w is the
marginal product of labor from a Cobb-Douglas technology with capital and labor as the
inputs, and ϵt is the idiosyncratic component, you may assume that it is i.i.d with mean 1
and a constant variance.
Finally, markets are incomplete: It is assumed that the agent can save in non-state
contingent assets, essentially claims on the aggregate capital stock, and also that she cannot
borrow.
1.1 Write down the Bellman equation(s) of this agent. Hint: Based on the fact that labor
is at the extensive margin you should note that in every period the agent will observe ϵt
and decide whether its worthwhile to work or not. If ϵt is low, then it may be optimal
to set ht = 0 and rest for a few periods until ϵt increases again. Therefore your Bellman
equation(s) should be explicit about this option value.
1.2 From the hint in 1.1 it is clear that this model possesses the reservation wage policy.
According to this, there is a unique threshold ϵ̄ such that if ϵt < ϵ̄ht = 0 and if
ϵt > ϵ̄ht = h. Is it true that ϵ̄ is unique, or should it be a function of other attributes?
1.3 Define the Recursive Competitive Equilibrium in this economy, when there is a unit
mass of agents solving the above problem and the markets should clear in the standard
fashion (e.g., aggregate capital and labor must give the returns w and r in equilibrium.)
1.4 Let us assume that there is no heterogeneity in ϵ. Everyone in the economy has the
same idiosyncratic endowment which stays constant over time. In equilibrium we
should expect that (1 + r)β = 1. Use this and the assumption of a Cobb Douglas
technology in capital and labor, to show that in equilibrium if the long run wealth
distribution has a mass point with probability 1 at this point then it must be that
all individuals are working or they are indifferent between working and not working.
Hint: Think of what consumption would be equal to in this case.
1
Problem 2. [20%] Consider the python code aiyagari.py on UVirtuelle, which represents
on the computer the Aiyagari model that we have seen in class.1 Treat as your benchmark
the given parameterization.
2.1 Assume that the period utility is given by log(c). What are the effects of lower rel-
ative risk aversion on interest rate, capital, and the wealth distribution, compared to
benchmark? Are those in line with what we should find theoretically?
2.2 Starting from the parameterization of the benchmark case, decrease the standard de-
viation of the earnings shocks σϵ to 0.45. What is the effect on interest rate, capital,
and wealth distribution? Are those in line with what we should find theoretically?
Problem 3. [20%] Consider the centralized business cycle model covered in class with Cobb-
1−γ 1+ϵ
Douglas technology and the following time utility function: U (c, ℓ) = c1−γ − ℓ1+ϵ , where ℓ is
labor.
3.1 Write down the problem recursively, using the Bellman equation.
3.2 Derive the first-order conditions of the problem, then use them to obtain the inter-
temporal optimality condition (the Euler equation) and the intra-temporal condition
(the labor supply condition). Interpret these two conditions.
Problem 4. [20%] Consider the python code rbc.py on UVirtuelle, which represents on
the computer the Real Business Cycle model that we have seen in class with preferences
1−σ 1+1/ϑ
U (C, N ) = C1−σ − N1+1/ϑ and Cobb-Douglas technology. The code generates graphs that
describe the impulse response function of increasing by 1% productivity At . Treat as your
benchmark the given parameterization.
4.1 Starting from the parameterization of the benchmark case, increase the coefficient of
relative risk aversion (sigma in the code = σ in U (C, N )) to 2. Describe and interpret
the results of the impulse response function obtained with the code.
4.2 Starting from the parameterization of the benchmark case, increase the Frisch elasticity
(vartheta in the code = ϑ in U (C, N )) to 2. Describe and interpret the results of the
impulse response function obtained with the code.
1
See https://python-programming.quantecon.org/intro.html for instruction about how to set-up python
on your laptop. Ask for help if you need it.
