PS 9

University of California Economics 202A
Department of Economics Fall 2014

P. Gourinchas/D. Romer
Problem Set 9
Due in Lecture Thursday Nov. 6
1 Romer 8.14
2 Romer 8.15
3 Consumption with State-Contingent Goods
Consider a consumer whose labor income (which he or she takes as exogenous) is uncertain.
Specifically, the consumers labor income in state s in period t is yst where s denotes the state
of the world
P and t the period. The probability that the state in period t is s is π st . Thus, for
each t, s π st = 1. The realization of the state each period is independent of the realization in
all other periods.
The consumer seeks to maximize
" #
X
E β t u(ct )
t
where 0 < β < 1 is the discount factor and u satisfies the usual conditions. The consumer can
purchase and sell state-contingent goods. Denote pst for the price of the consumption good at
time t in state s. Thus, we can write the consumers objective function as
XX XX
pst cst ≤ pst yst
t s t s
1. Set up the consumers maximization problem, and find the first-order condition for cst .
2. Consider two states in some period t, s0 and s”. Under what conditions is consumption
the same in the two states? (That is, under what conditions is cs0 t = cs00 t ?)
3. Consider state s0 in period t0 and state s” in period t”. Under what conditions is cs0 t0 =
cs00 t00 ?
4. Consider 2 consumers who differ only in their yst ’. Show or provide a counterexample to
the following claim: If Consumer 1s consumption in one period is greater than Consumer
2s consumption in that period, Consumer 1s consumption in each period is greater than
Consumer 2s consumption in the same period.
1
5. Suppose that both consumers have constant relative risk aversion utility, with the same
coefficient of relative risk aversion θ > 0. What, if anything, can one say about how the
ratio of Consumer 1s consumption to Consumer 2s consumption behaves over time?
6. In practice, we often see consumption reversals (that is, one consumer initially having
consumption higher than another, but later having lower consumption). List 2 or 3 ways
the assumptions of this problem could fail that could make such reversals possible; explain
each possibility in no more than a sentence.
7. Suppose that in some period, the realization of s is the one that has the highest value
of pst yst for that period for the consumer. How, if at all, will that affect the consumers
consumption in later periods?
4 Consumption and Liquidity Constraints

You have a feeling that liquidity constraints may matter for intertemporal consumption. How-
ever, you also know that that certain specializations of the conventional unconstrained model of
consumption may perhaps equally well explain the facts. You’d like to investigate the empirical
importance of these liquidity constraints. To make your research more precise you write down
the following alternative hypotheses on what is important in the way people go about choosing
their consumption paths:
-H0: Individuals are expected lifetime utility maximizers (i.e. choose consumption paths
according to the canonical model) without credit market constraints.
versus
-H1: Individuals maximize lifetime utility subject to a constraint on borrowing
To enable you to reject one of these hypotheses (usually the null), you must know some
testable implications of H0 or H1. They way to find such testable implication is to write down a
model which, when solved, implies some restrictions on how data should behave. To keep things
simple and parsimonious, you decide to model H1 using the simplest possible extension of the
canonical model:
T
X −1
E0 β t u(ct )

max
t=0
subject to:
at+1 = R(at + ỹt − ct )
and
at ≥ 0
This is identical to the standard problem except for the constraint at ≥ 0 which says that
the household may never run into debt. The only source of uncertainty is future income and we
assume that ymin > 0.
1. Defining ‘cash on hand’ xt = at + ỹt , express the liquidity constraint in t in terms of ct
and xt .
2
2. Argue that, if the liquidity constraint is binding at time t, the standard Euler equation
is violated. Explain the direction of the inequality. Now assume that the Euler condition
holds between t and t + 1. Could liquidity constraints still affect the consumption decision
between the two periods? Finally, suppose that, say, the next 5 period liquidity constraints
are binding. What is the effect on consumption at time t (today) of an increase in income
in period t + 6?
3. Now let us see how the Euler equation is modified in the presence of a liquidity constraint.
Define vt (xt ) the value function at time t. Why is xt rather than at the state variable of
the problem? Write down the Bellman equation, defining µ as the (nonnegative) Lagrange
multiplier on the liquidity constraint. Find the first order necessary condition for optimal-
ity. Use the envelope condition to show that the Euler equation becomes an inequality
involving marginal utility of consumption today u0 (ct ), expected marginal utility of con-
sumption tomorrow Et [u0 (ct+1 )], the discount factor β, the interest rate R and the value
of cash on hand xt . Explain why you obtain an inequality
4. From the above, argue informally that there exists some cash-on-hand level xt such that
ct (x) = x if x < xt and ct (x) < x if x ≥ xt .
5. Explain why the relationship between consumption and current income implied in the
preceding question cannot be used to test H0 against H1. Following Zeldes, now consider
the following way of testing H0 against H1. Suppose we had a panel of consumers which
we could split into two groups: G1 is a group of people who are likely to be liquidity
constrained whereas G2 is a group of people who the theory says should not be liquidity
constrained. Then, estimating and testing the significance of µ should, if the theory is
right, give us a significant µ for G1 but not for G2.
6. What criterium does the theory suggest for splitting the sample? To what extent would
you trust this criterium to, say, get you only liquidity constrained people into G1? On
what features of preferences might this depend?
7. Now work your way from the first order condition to an equation we can use to estimate
and test µ for each group
(a) rewrite the first order condition as
Et [Rβu0i,t (ct+1 )/u0it (ct )](1 + λit ) = 1
where you will define λ in terms of µ.

(b) assume the following utility function: uit (c) = c1−θ /(1 − θ) exp(γ i t) where γ it is a
taste shifter that depends on time and demographic variables (family size, age) and
some other influences that may not be observable
(c) From the previous two equations, derive an empirical specification with log change of
consumption as the left hand side, fixed effects, gross interest rate, observable family
variables and an error term on the right hand side. [Hint: you may want to define
γ it = zit + ω i + η t + uit where zit are observable demographics, ω i is a household fixed
effect, η t is a time effect and uit captures any unobservable determinants.]
3
8. Using the equation derived above, show that everything else a higher λ corresponds to a
higher rise in expected consumption between t and t + 1.
9. Now consider the right hand side of the regression you specificed. With which variables
might you have an endogeneity problem? Why? How would you correct for this?
10. Indicate qualitatively how you would estimate λit

PS 9

Uploaded by

Copyright:

Available Formats

PS 9

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PS 9

Uploaded by

Copyright:

Available Formats

University of California Economics 202A

Department of Economics Fall 2014

4 Consumption and Liquidity Constraints

Et [Rβu0i,t (ct+1 )/u0it (ct )](1 + λit ) = 1

where you will define λ in terms of µ.

10. Indicate qualitatively how you would estimate λit

You might also like