MGED06: ADVANCED MACROECONOMIC THEORY

Problem Set-1

Question 1
Suppose we have an agent who lives two periods. They earn income in both periods given by w 1
and w2 respectively. Further this agent has a period utility function with standard properties
(positive & diminishing marginal utility). Their lifetime utility function is additively separable and
is discounted.

V  u(c1 )    u (c2 ) , where 0 < β < 1

a) Suppose the (single) real interest rate paid to savers & by borrowers is given by r. Write down
and interpret the Intertemporal Budget Constraint (IBC).
b) Write down the optimal conditions for lifetime utility maximization. Explain where these have
come from. What is the Euler equation for this problem? Interpret what the Euler equation
represents.
c) In the optimization problem above what terms (parameters, functions and/or variable) are/were
exogenous and which ones are endogenous? Briefly explain why. Interpret the parameter β.
d) When, if at all, would this agent find it optimal to perfectly smooth their consumption over
time?
e) Suppose the real interest rate were to suddenly change. What impact would this have to the
choice set of this agent? Explain with the use of a diagram, math and words how this works.
f) When determining the impact of a change in the real interest rate on the optimal consumption
bundle (pattern) we have to consider both the income effect and a substitution effect (the
change causes). Explain these effects in the case where the agent initially was a net saver. Now
do the same for the case where the agent is a net borrower. Lastly, do the same for the case
where the agent is neither saving/borrowing in net at the initial optimal consumption bundle.
Question 2

Here we are again considering the agent whose information is given in question 1 above.

One of the important determinants of the response of saving and consumption to the real interest
rate is the Elasticity of Intertemporal Substitution (EIS). That elasticity can be measured by the
responsiveness of the growth rate of consumption to changes in the (expected) real interest rate.
Thought of this way the EIS is defined as the percentage change in consumption (in year 2 versus
year 1) per percentage increase in the real interest rate.

d ln (c 2 / c1 )
Thus, EIS  , where natural logs are essentially equivalent to growth rates.
dr

d ln (c 2 / c1 )
Show that EIS can also be written as: EIS   , so that EIS is equivalent to the
d ln (u ' (c 2 ) / u ' (c1 ))
elasticity of consumption growth with respect to marginal utility growth.

Question 3
Suppose we return to the agent from question 1 above. Suppose their period utility function is given
by the isoelastic (power) utility function:

c
1

u (ct )  t , where θ is a positive constant.

1
a) Determine the optimal consumption bundle for this agent.
u ' ' (c )
b) The measure of Absolute Risk Aversion (ARA) is defined as: ARA   . Determine
u ' (c )
ARA for this agent. Explain any interesting features of their measure of ARA.
u ' ' (c )
c) The measure of Relative Risk Aversion (RRA) is defined as: RRA   c  . Determine
u ' (c )
RRA for this agent. Explain any interesting features of their measure of RRA.
d) The measure of Elasticity of Intertemporal Substitution (EIS) is defined as:
u ' (c ) 1
EIS    . Determine EIS for this agent. Explain any interesting features of
c  u ' ' (c) RRA
their measure of EIS.

Question 4
Suppose we return to the agent from question 1 above. Suppose their period utility function is given
by the exponential utility function:

U ( c t )   e   c t , where α is a positive constant.

2
a) Determine the optimal consumption bundle for this agent.
u ' ' (c )
b) The measure of Absolute Risk Aversion (ARA) is defined as: ARA   . Determine
u ' (c )
ARA for this agent. Explain any interesting features of their measure of ARA.
u ' ' (c)
c) The measure of Relative Risk Aversion (RRA) is defined as: RRA   c  . Determine
u ' (c )
d) The measure of Elasticity of Intertemporal Substitution (EIS) is defined as:
u ' (c ) 1
EIS    . Determine EIS for this agent. Explain any interesting features of
c  u ' ' (c) RRA
their measure of EIS.

Question 5
Suppose we return to the agent from question 1 above. Suppose their period utility function is given
by the (natural) logarithmic utility function:

U (ct )  ln(ct )

a) Determine the optimal consumption bundle for this agent.

u ' ' (c )
b) The measure of Absolute Risk Aversion (ARA) is defined as: ARA   . Determine
u ' (c )
ARA for this agent. Explain any interesting features of their measure of ARA.
u ' ' (c)
c) The measure of Relative Risk Aversion (RRA) is defined as: RRA   c  . Determine
u ' (c )
RRA for this agent. Explain any interesting features of their measure of RRA.
d) The measure of Elasticity of Intertemporal Substitution (EIS) is defined as:
u ' (c ) 1
EIS    . Determine EIS for this agent. Explain any interesting features of
c  u ' ' (c) RRA
their measure of EIS.

Question 6
Suppose we have an agent who lives two periods. Their period utility function is given by U(c) = c 0.5
and their personal (utility) discount factor (β) is equal to 0.95. This person earns income of $25,000
and $20,000 in periods one and two respectively. The market real interest rate is 5 percent.

a) Solve for this agents’ optimal levels of consumption and savings in both periods. Are they a net
saver or borrower? Explain.

b) What is their optimal period one utility equal to? What about their optimal period two level of
utility? What is their optimal level of lifetime utility equal to (in period one)?

c) Is this person smoothing their consumption? Explain.

3
d) Is this person perfectly smoothing their consumption? Explain. What value of the real interest rate
(r) would see this person perfectly smooth their consumption? Explain. Hint: Solve the problem for
generic values of r & β and then answer this question.

e) Is there a value of the real interest rate that would see this person save none of their period one
income? Explain.

f) Go back to the initial problem with r = 5% & β = 95% etc. Suppose r rises by 1 percent (not 1
percentage point). Determine the new optimal levels period one and period two consumption and
saving. Determine and explain the income and substitution effects this increase in the real interest
rate creates.

g) Using the answers you calculated for parts A & F above determine the elasticity if intertemporal
substitution (EIS) we have observed here. Hint: Parts of the answer to question 2 above might prove
useful in determining the EIS.

Question 7 - True/false/uncertain
An agent is consuming at the point where their lifetime utility is maximized. Suppose their level of
income for period one suddenly rises by 5%. As a consequence following this change at their new
optimal intertemporal consumption bundle this agent will definitely now be a net saver.

4
 Problem Set-1 (Solutions)

Question 1
a)
[ w1(1+r) + w2 ] ≥ [ c1(1+r) + c2 ] Equation (1a)

This equation says the future value (FV) of lifetime resources is greater than or equal to the future
value of (FV) of lifetime consumption (i.e. in future value terms, during their lifetimes the
representative agent’s expenditure on consumption can’t exceed their available resources).
Alternatively,

[ w1 + w2/(1+r) ] ≥ [ c1 + c2/(1+r) ] Equation (1b)

This equation says the present value (PV) of lifetime resources is greater than or equal to the
present value of (PV) of lifetime consumption (i.e. in present value terms, during their lifetimes the
representative agent’s expenditure on consumption can’t exceed their available resources). Of
course we should notice that both versions of the IBC are identical, as such either can be used.

Aside:
1) The assumption that the period utility function always has positive MU implies the IBC holds
with equality (i.e. the representative agent will never have leftover resources).
2) The assumption of diminishing MU provides an incentive for (some) consumption smoothing.
How much consumption smoothing we get exactly depends on the sizes of β & r.

b)

The Lagrangian for this problem is:   V  [ w1 (1  r )  w2  c1 (1  r )  c2 ]

Where V  U (c1 )    U (c2 ) , where 0 < β < 1

with first order conditions (F.O.C.s), given by:


 w1 (1  r )  w2  c1 (1  r )  c 2  0 IBC, Equation (1)



 u ' (c1 )   (1  r )  0 Equation (2)
c1

   u ' (c 2 )    0 Equation (3)
c 2

5
Combining FOCs (2) & (3), we get the Euler equation, u ' (c1 )  (1  r )    u ' (c2 ) . The Euler
equation shows how consumption is traded from one period to another (adjacent) period. 1 At
the optimum, when the Euler equation holds, the consumer is indifferent between consuming
an extra unit today and saving an extra unit today.

Aside:
1) The Euler equation and the IBC are the two equations that we can solve for the levels of
consumption in periods 1 & 2.
2) Since the IBC holds with equality we can rewrite it as an equation for c 2 and substitute this
value into the second period utility function.

c2  [w1 (1  r )  w2  c1 (1  r )]

V  U (c1 )    U (c2 )  U (c1 )    U ( w1 (1  r )  w2  c1 (1  r ))

V
 U ' (c1 )  (1  r )  U ' ( w1 (1  r )  w2  c1 (1  r ))  0
c1
We can solve this FOC for c1 and the substitute this into the above equation (IBC) for c2. The
solutions we obtain via this route should be identical to those arrived at via using the
Lagrangian approach.
c)
Exogenous terms are those that are given (fixed constants) in the model (i.e. they are terms we do
not need to solve for). In this model we treat the parameters β, w1, w2 & r as exogenous along with
the period utility function U(c).
Endogenous terms are those that are solved for in the model. In this model the terms for first and
second period consumption (c1 & c2) are endogenous as they are being determined by solving the
representative agent`s intertemporal utility maximization problem.

Interpretation of the β term

We have written the representative agent`s lifetime utility function as:
V  U (c1 )    U (c2 ) , where 0 < β < 1
Here the term β represents a personal discount factor that works to convert units of second period
utility into units of first period utility. Sometimes the term β is instead written (or replaced by) an
equivalent form using the term ρ, where
U (c2 )
V  U (c1 )  , where 0    1
1 

u ' (c1 )
1
If the Euler equation is re-written as follows:  (1  r ) , where the left-hand-side term is the Marginal
  u ' (c 2 )
Rate of Substitution between period 2 & 1, then the equation is referred to as the Keynes-Ramsey rule.

6
 1
Here, clearly   , where ρ represents a personal rate of discount (that helps to convert period
1 
two units of utility into units of period one utility). The term ρ is often called the rate of time
preference.

So the larger is ρ (and thus the smaller is β), then it takes more units of period two utility
(consumption) in order to get (convert into) one unit of period one utility. In a case such as this the
agent is said to be (more) impatient and will, ceterius paribus, prefer to consume (more) sooner
rather than later (in terms of earning lifetime utility). Of course the levels of income in both periods
and the real interest rate all matter as well.

If ρ is smaller (and thus β is larger), then it takes fewer units of period two utility (consumption) in
order to get (convert into) one unit of period one utility. In a case such as this the agent is said to be
(more) patient and will, ceterius paribus, prefer to consume (more) later rather than sooner (in terms
of earning lifetime utility). Of course w1, w2 & r still matter as well.

d)
This agent would find it optimal to perfectly smooth their consumption (i.e. set c 1 = c2), if and only
if, β(1+r) = 1. As when this occurs the Euler equation simplifies to: u ' (c1 )  u' (c2 ) , which implies
the utility maximizing consumption pattern must have c1 = c2 .

Note:
If β(1+r) = 1, then this implies ρ = r. So the representative agent will perfectly smooth their
consumption over time, if and only if, their personal rate of discount (ρ) is exactly equal to the
market rate of discount (r). This result is independent of the sizes of w 1, w2 & r.

Recall that at the optimum the consumer is indifferent between consuming an extra unit today and saving
an extra unit today. Perfectly smoothing consumption makes sense when ρ = r as the market return to
saving a bit more equals the personal rate of time preference (and thus altering the levels of
consumption & savings away from a perfectly smooth consumption pattern will not make utility
rise).

Other cases:

Case 2: If ρ < r, then β(1+r) > 1, so then the Euler equation has the form u ' (c1 )  u ' (c2 ) , which
implies the utility maximizing consumption pattern must have c1 < c2 (this is due to diminishing
marginal utility). This makes sense as the personal rate of time preference is smaller than the
market rate of interest (i.e. the personal rate of discount from period 2 to 1 is smaller than the
market rate of discount). As such the agent is willing to save more and consume less in period 1 - in
order to consume more in period 2 than they consume in period 1. This agent is more patient than
the market is and thus is willing to wait to consume more later in their life in order to earn higher
lifetime utility.

7
 Case 3: If ρ > r, then β(1+r) < 1, so then the Euler equation has the form u ' (c1 )  u ' (c2 ) , which
implies the utility maximizing consumption pattern must have c1 > c2 (this is due to diminishing
marginal utility). This makes sense as the personal rate of time preference is larger than the market
rate of interest (i.e. the personal rate of discount from period 2 to 1 is larger than the market rate of
discount). As such the agent is willing to save less (even borrow more if they have to) and consume
more in period 1 - in order to consume less in period 2 than they consume in period 1. This agent is
less patient (i.e. more impatient) than the market is and thus is not willing to wait to consume
more later in their life in order to earn higher lifetime utility.

e)

C1

w2 + ↑r
w1(1+r)

E
w2

C1
w1 w1 + w2/(1+r)
8
The initial IBC (in blue) goes through the point of incomes (w 1,w2) with intercepts equal to the PV
of lifetime income on the c1 axis and the FV of lifetime income on the c2 axis. If the real interest
rate (r) were to suddenly increase this would make the IBC pivot to a new IBC (now in red) that
still goes through the point of incomes but that is now steeper.
Note: The slope of the IBC is equal to – (1+r) and by construction the IBC must go through the
point of incomes (point E).

This shift may result in an expansion or a reduction in the choice set of the (representative) agent.
This really depends on the location of the agent`s indifference curves in the c 1 & c2 plane. If the
indifference curves are located up near the portion of the IBC that sweeps outwards (i.e. above
point E) then the agent has experienced an expansion of their choice set. If the indifference curves
are located down near the portion of the IBC that sweeps inwards (i.e. below point E) then the agent
has experienced a reduction in their choice set.

f)
As noted in the question there are two effects, a substitution effect and an income effect.

Substitution Effect
In all cases an increase in the real interest rate (r) makes the price of first period consumption rise
relative to consuming in the second period. As such, the substitution effect (of an increase in r)
causes the agent to consume less in the first period (and save more in that period) and as a result to
consume more in period two. So the agent has substituted away from the now relatively more
expensive first period consumption towards the now relatively less expensive second period
consumption.
Substitution effect: ↑r causes ↑s1 (& ↓c1) and in turn ↑c2

Income Effect
Consumption in all (both) periods is assumed to be a normal good. The income effect depends on
whether the agent was or was not initially a net saver or net borrower prior to the increase in the
real interest rate. So there are three sub-cases:

Net Saver (i.e. initially s1 = (w1 – c1) > 0, indifference curve tangency above & left of point E)
An agent who is initially a net saver experiences an increase in their income as a result of the higher
real interest rate (which returns more future investment income to them in period two). As a result
such an agent will increase their level of consumption in both periods (of course as c 1 ↑ then s1 ↓, so
that they save less).
Income effect (net saver): ↑r causes ↑income (over whole lifetime) which causes ↑c 1 & ↑c2

Net Borrower (i.e. initially s1 = (w1 – c1) < 0, indifference curve tangency below & right of point E)
An agent who is initially a net borrower experiences a decrease in their income as a result of the
higher real interest rate (which requires more future resources in order to pay back any period one

9
borrowing by them). As a result such an agent will decrease their level of consumption in both
periods (of course as c1 ↓ then s1 ↑, so that they borrow less).
Income effect (net saver): ↑r causes ↓income (over whole lifetime) which causes ↓c 1 & ↓c2

Neither (i.e. initially s1 = (w1 – c1) = 0, indifference curve tangency at point E)

An agent who is initially neither borrowing nor saving in net sees no change in their income as a
result of the higher real interest rate. As a result such an agent will not increase or decrease their
level of consumption in both periods (due to the income effect).
Income effect (neither net saver or borrower): ↑r causes no change in income (over whole
lifetime) which does not change either c1 or c2

Total Effect (of ↑r)

Type of consumer Substitution Effect Income Effect Total Effect
Net saver ↓c1 & ↑c2 ↑c1 & ↑c2 Δc1 = ? & ↑c2
Net borrower ↓c1 & ↑c2 ↓c1 & ↓c2 ↓c1 & Δc2 = ?
Neither ↓c1 & ↑c2 Δc1 = Δc2 = 0 ↓c1 & ↑c2

So focusing on one case as a representative case lets discuss the net saver. In the case of period two
consumption (c2) both the income and substitution effects move in the same direction (i.e. these
effects reinforce each other) so second period consumption rises for the net saver as a result of the
increase in the real interest rate. For the net saver the substitution effect and income effect work in
opposite directions on first period consumption (c 1) and as such the total (net) effect on first period
consumption is ambiguous.
Note: The actual impact on first period consumption for the net saver depends on which effect
dominates – the income effect or the substitution effect. So clearly the size of the income change
that results due to the change in r matters (this is proportional to the income effect). But what also
matters is the strength of the substitution effect – this is determined by the degree of curvature of
the indifference curves (which of course comes from the exact nature of the agent`s preferences). A
straighter more linear indifference curve has a much higher degree of substitution from c 1 to c2 (as r
changes). As the indifference curves get more and more curvature the preferences exhibit more and
more complementarity between c1 and c2 and thus less and less substitutability between them. So
the curvature of the indifference curves and the strength of the substitution effect are all about the
degree of intertemporal substitutability and thus the Elasticity of Intertemporal Substitution (EIS)
which is what the next question is about.

10
Question 2

The Elasticity of Intertemporal Substitution (EIS) measures by how many percent the relative
demand for consumption next period, relative to demand for consumption in this period changes as
1
the relative price of consumption in next period to consumption in this period, q  , changes
1 r
by one percent. So EIS is given by:

  c 2    1    c    c    c 
 d       d ln 2    d ln 2    d ln 2  
 
  c1    1  r     
c1     
c1    c
EIS       1 
     
  d ln 1  r   
 
 d  1    c 2    d ln 1   dr 
  1  r   c     1  r      
   1       

Any of the above formulae are correct and can be used to determine the EIS. Above we have used
the following facts about natural logarithms (ln):
dx
1)  d ln(x) = instantaneous percentage change in x
x
a
2) ln    ln( a )  ln(b)
b
3) ln(1) = 0; &
4) ln1  r   r , for 0  r 1

Answer (source wiki http://en.wikipedia.org/wiki/Elasticity_of_intertemporal_substitution )

If the real rate rises, future consumption may increase due to increased return on savings; but future
consumption may also decline as the saver decides to save less given that she can get a higher return on
what she does save. The net effect on future consumption is the elasticity of intertemporal substitution.

Suppose total lifetime utility is given by

In this setting, the real interest rate will be given by the Euler equation:

Qu'(ct) = QβRu'(ct + 1)

A quantity of money Q invested today costs Qu'(ct) units of utility, and so must yield exactly that
number of units of utility in the future when saved at the prevailing gross real interest rate R. (If it
yielded more, then the agent could make herself better off by saving more.)

Solving for the real interest rate, we see that

11
  u ' (c1 ) 
1 r   
 u ' (c2 ) 

In logs, we have

 u ' (c 2 ) 
r   ln    ln( )
 u ' (c1 ) 

Since β is constant this implies,

 u ' (c 2 ) 
dr  d ln  
 u ' (c1 ) 

The elasticity of intertemporal substitution is defined as the percent change in consumption

growth per percent increase in the net interest rate:

 c 
 d ln  2  
  c 
EIS    1 

 dr 
 
 

By substituting in our expression for dr from the log equation above, we can see that this definition
is equivalent to the elasticity of consumption growth with respect to marginal utility growth:

d ln (c2 / c1 )
EIS  
d ln (u ' (c2 ) / u ' (c1 ))

Either definition is correct, however, assuming that the agent is optimizing and has time separable
utility

NOTE:
The question did not ask about this next part which is provided for additional learning. At a utility
maximizing allocation we have the Euler equation & IBC holding. Thus in the case of additively
separable intertemporal lifetime utility functions gives us an Euler equation of the form:

u ' (c1 )  (1  r )    u ' (c2 )

u ' (c1 )
Which can be re-written as: MRS (c 2 , c1 )   (1  r ) , the so-called Keynes-Ramsey rule.
  u ' (c 2 )

12
Of course MRS stands for the Marginal Rate of Substitution as we move along a particular
indifference curve. Our lifetime utility function V is given by, V  u (c1 )    u (c2 ) , so along an
indifference curve for these preferences we have:

dV  0  u ' (c1 )  dc1    u' (c2 )  dc2

  MRS c 2 , c1 
dc 2 u ' (c1 )
Which implies: 
dc1 dV  0
  u ' (c 2 )

 MRS c 2 , c1 
dc 2 u ' (c1 )
Or,  
dc1 dV  0
  u ' (c 2 )

d ln (c2 / c1 )
Our last EIS equation above is: EIS  
d ln (u ' (c2 ) / u ' (c1 ))

d ln (c2 / c1 )
This is the same as: EIS 
d ln (u ' (c1 ) / u ' (c2 ))

Since β is a constant we can substitute as follows: d ln(MRS )  d ln(u ' (c1 ) / u ' (c2 )) , so we get

d ln (c2 / c1 )
EIS 
d ln ( MRS (c2 , c1 ))

The EIS then measures the degree to which we substitute away from period 2 consumption (c 2) and
towards period 1consumption (in percentage change of the ratio c2/c1) as the real interest rate rises by
one percentage point. In terms of geometry, we can see this as the degree to which the ratio (c 2/c1)
changes as we change the slope of the IBC (via changes in the real interest rate). This is the change in
the slope of the ray from the origin to the optimal tangency (at which point MRS = (1+r) ) that occurs
as we change the slope of the indifference curve (which happens as we change the real interest rate
by 1 percent). This is depicted in the diagram below. From the diagram we can easily notice how the
degree of curvature of the indifference matters in determining the degree of intertemporal
substitution. It is also worth noting the higher the curvature of u(c), the higher the risk aversion.

13
 C2

slope=C12/C11

C1 E1
|slope|=MRS(C12:C11)
slope=C02/C01

E0
indifference curve
C0

|slope|=MRS(C02:C01)

C1
C11 C01

14
Question 3
a)

The Lagrangian for this problem is:   V  [ w1 (1  r )  w2  c1 (1  r )  c2 ] ,

Where
V  u(c1 )    u (c2 ) , where 0 < β < 1

with first order conditions (F.O.C.s), given by:


 w1 (1  r )  w2  c1 (1  r )  c 2  0 IBC, Equation (1)



 u ' (c1 )   (1  r )  0 Equation (2)
c1

   u ' (c 2 )    0 Equation (3)
c 2

Combining FOCs (2) & (3), we get the Euler equation, u ' (c1 )  (1  r )    u ' (c2 ) . For this utility
function the Euler equation becomes: (c1 )  (1  r )    (c 2 )
 

Cross multiplying we get: (c 2 *)  (1  r )    (c1 *)

Taking the θth root we get: c 2 *  (1  r )1 /    1 /   c1 *
Substituting this into the IBC gives: w1 (1  r )  w2  (1  r )  c 1*  (1  r )1 /    1 /   c1 *

w1 (1  r )  w2  (1  r )  c 1*  (1  r )1 /    1 /   c1 *

w1 (1  r )  w2 [ w1 (1  r )  w2 ]  (1  r )1 /    1 / 
c1 *  c2 * 
(1  r )  (1  r )1 /    1 /  & (1  r )  (1  r )1 /    1 / 

Aside:
1 w (1  r )  w2
1) When,   which implies  (1  r )  1 , c1 *  c 2 *  1
1 r 2r
1
2) When,   which implies  (1  r )  1 , c1 *  c2 * , as a “large” β relative to the market
1 r
real interest rate implies the agent is relatively “patient” and is thus willing to wait until later
periods to consume (more in order to maximize their utility).

15
 1
3) When,   which implies  (1  r )  1 , c1 *  c2 * , as a “small” β relative to the market
1 r
real interest rate implies the agent is relatively “impatient” and is thus not willing to wait
until later periods to consume (more in order to maximize their utility).
b)

For this period utility function we have:

u ' (c)  c  & u ' ' (c)  c  1

u ' ' (c) c  1 

This implies: ARA     
u ' (c ) c c

The measure of absolute risk aversion (ARA) varies as consumption (c) varies (as c rises/falls ARA
falls/rises) and is proportional to the positive (constant) preference parameter θ.

c)

u ' ' (c )
For these preferences: RRA   c  
u ' (c )
The measure of relative risk aversion (RRA) is constant for these preferences (such preferences are
said to be/exhibit CRRA). It equals the positive constant preference parameter θ.

d)

The Euler equation implied by these preferences is given by: (c1 )  (1  r )    (c 2 )

 

 c2 

u ' (c 2 ) 1
This can be re-written as: u ' (c )  (1  r )    c 
1  1
 u ' (c 2 )   c2 
Taking the natural log of both sides gives: ln u ' (c )     ln c 
 1   1 
c 
 ln 2 
EIS    c1   1  1
Using the earlier definition of the EIS we get:  u ' (c 2 )   RRA
 ln 
 u ' ( c )
1 

For these preferences the elasticity of intertemporal substitution (EIS) and the measure of relative
risk aversion (RRA) are both constant (independent of r, c etc) and are inverses of each other.

So when we do intertemporal choice problems with the isoelastic (power) utility function we are
unable to disentangle the elasticity of intertemporal substitution (EIS) from the coefficient of

16
relative risk aversion (RRA). Knowledge of either is sufficient to know the other (they are not
separately identified).

Here the intertemporal elasticity of substitution is given by . In general, a low value of theta (high
intertemporal elasticity) means that consumption growth is very sensitive to changes in the real
interest rate. For theta equal to 1, the growth rate of consumption responds one for one to changes
in the real interest rate. A high theta implies an insensitive consumption growth. In the limit when
theta equals to infinity consumption growth (the ratio of c 2/c1) is completely insensitive to changes
in r (this case is for preferences for perfect complementarity between c 2 & c1 – so that the agent
must consumption these in fixed proportions at the corner of their L-shaped indifferences curves).
A few typical indifference curves are plotted below for several special values of theta.

C2

indifference curves

Θ
=→0 ∞ Perfect Complements

0< <∞
0<Θ<∞

Θ=0
 Perfect Substitutes

C1

Aside:
When θ = 1 we use l’Hopital’s rule2 to derive the limiting form of the period utility function,
this simplifies to the case of log utility (i.e. u(c) = ln(c)). With these preferences the income &
substitution effects on saving can exactly cancel out. So log utility is a form of CRRA
preferences and thus should have the same sort of properties for RRA & EIS as described above
in this question (Q3). Log utility is dealt with in question 5 below.

Question 4
a)

The Lagrangian for this problem is:   V  [ w1 (1  r )  w2  c1 (1  r )  c2 ] ,

Where

c1
If we want the limit of u (c )  as θ tends towards 1 we use l’Hopital’s rule. This rule says
2

1
f ( ) f ' ( )  c 1  ln(c )
lim lim
c 1
 1 g ( )

 1 g ' ( )
, so lim
 1
u ( c )  lim
 1 1  
 lim
 1 1
 ln(c ) , u(c) = ln(c).

17
 V  u(c1 )    u (c2 ) , where 0 < β < 1

with first order conditions (F.O.C.s), given by:


 w1 (1  r )  w2  c1 (1  r )  c 2  0 IBC, Equation (1)



 u ' (c1 )   (1  r )  0 Equation (2)
c1

   u ' (c 2 )    0 Equation (3)
c 2

Combining FOCs (2) & (3), we get the Euler equation, u ' (c1 )  (1  r )    u ' (c2 ) . For this utility
function the Euler equation becomes: a  e 1  (1  r )    a  e 2
 ac  ac

Cancelling out the common term a & taking the natural log of both sides yields:
ln(e  ac1 )  a  c1  ln[(1  r )  ]  ln(a  e  ac2 )  ln[(1  r )  ]  a  c 2
 a  c1  ln[(1  r )  ]  a  c 2
1
c1 *  c 2 *    ln[(1  r )  ]
a
Aside:
1 w (1  r )  w2
1) As before when,   which implies  (1  r )  1 , we get c1 *  c 2 *  1
1 r 2r
1
2) When,   which implies  (1  r )  1 , c1 *  c2 * , as a “large” β relative to the market
1 r
real interest rate implies the agent is relatively “patient” and is thus willing to wait until later
periods to consume (more in order to maximize their utility).
1
3) When,   which implies  (1  r )  1 , c1 *  c2 * , as a “small” β relative to the market
1 r
real interest rate implies the agent is relatively “impatient” and is thus not willing to wait
until later periods to consume (more in order to maximize their utility).

1
Substituting the Euler equation c1 *  c 2 *    ln[(1  r )  ] into the IBC yields:
a
w1 (1  r )  w2  c1 * (1  r )  c2 *
1
w1 (1  r )  w2  (c 2 *    ln[(1  r )  ]) x (1  r )  c 2 *
a

w1 (1  r )  w2    ln[(1  r )  ]x1  r 
1
c2 *  a
2r

18
 w1 (1  r )  w2    ln[(1  r )  ]x1  r 
1
   ln1  r  
a 1
So, c1 * 
2r a
1
w1 (1  r )  w2    ln[(1  r )  ]
c1 *  a
2r

b)

For these preferences we have:

u ' ( c )  a  e  a c
u ' ' ( c )   a 2  e  a c
u ' ' (c )   a 2  e  ac 
ARA       a
u ' (c )  ae 
 ac

These preferences have Constant Absolute Risk Aversion (CARA), they are the only preferences
with this feature.

c)

u ' ' (c )   a 2  e  ac 
RRA   c   c   a  c
u ' (c )  ae 
 ac

These preferences exhibit increasing RRA as c rises.

d)
1 1
So we get EIS   . This does NOT result in an EIS that is constant! With exponential
RRA a  c
utility the elasticity of intertemporal substitution varies as the level of consumption c varies. As the
level of consumption tends towards infinity the EIS tends towards zero.
Aside: For some utility functions the EIS expression is not nice (& we would not ask for it on our
exams).

w1 (1  r )  w2    ln[(1  r )  ]x1  r 
1
 c2 *  a
  
c *
 1  1
w1 (1  r )  w2    ln[(1  r )  ]
a

19
 c  1 1
 ln 2  w1    ln[(1  r )  ]   
EIS   c1   a a
w1 (1  r )  w2    ln[(1  r )  ]x1  r 
r 1
a
 1  1 
w1    
 a  1  r 

1
w1 (1  r )  w2    ln[(1  r )  ]
a

Question 5
a)

The Lagrangian for this problem is:   V  [ w1 (1  r )  w2  c1 (1  r )  c2 ] ,

Where
V  u(c1 )    u (c2 ) , where 0 < β < 1

with first order conditions (F.O.C.s), given by:


 w1 (1  r )  w2  c1 (1  r )  c 2  0 IBC, Equation (1)



 u ' (c1 )   (1  r )  0 Equation (2)
c1

   u ' (c 2 )    0 Equation (3)
c 2

Combining FOCs (2) & (3), we get the Euler equation, u ' (c1 )  (1  r )    u ' (c2 ) . For this utility
1 1  c2 
function the Euler equation becomes:    (1  r )      , or    (1  r )  
 c1   c2   c1 

c *  (1  r )    c1 * w (1  r )  w2  c1 * (1  r )  c2 *
Rearranging we get 2 along with the IBC 1 we
have two equations that can be solved for consumption in both periods. Plug the first into the IBC
yields:
w1 (1  r )  w2  c1 * (1  r )  (1  r )    c1 *
Rearranging gives

c1 * 
w1 (1  r )  w2
c2 * 
w1 (1  r )  w2
  1  r 
(1  r )  (1  r )   & (1  r )  (1  r )  
These expressions are equivalent to:

20
 w2
w1  [ w1 (1  r )  w2 ]  
c1 *  1 r c2 * 
(1   ) & (1   )
w2 w
w1 (1  r )  w1  w1   2
s1 *  w1  c1 *  1 r  1 r
(1   ) 1 
Aside:
1 w1 (1  r )  w2
1) When,   which implies  (1  r )  1 , c1 *  c 2 * 
1 r 2r ,
(same as in Q3 &
Q4).
1
2) When,   which implies  (1  r )  1 , c1 *  c2 * , as a “large” β relative to the market
1 r
real interest rate implies the agent is relatively “patient” and is thus willing to wait until later
periods to consume (more in order to maximize their utility).
1
3) When,   which implies  (1  r )  1 , c1 *  c2 * , as a “small” β relative to the market
1 r
real interest rate implies the agent is relatively “impatient” and is thus not willing to wait
until later periods to consume (more in order to maximize their utility).

b)

For these preferences we get:

1
u ' (c)   c 1
c
1
u ' ' (c )   2   c  2
c
u ' ' (c ) (c 2 ) 1
ARA     1 
u ' (c ) c c
For these preferences ARA declines as c rises.
u ' ' (c)
a) The measure of Relative Risk Aversion (RRA) is defined as: RRA   c  . Determine
u ' (c )
RRA for this agent. Explain any interesting features of their measure of RRA.

u ' ' (c)  1

RRA   c    c   1
u ' (c )  c 
These preferences represent a Constant Relative Risk Aversion (CRRA) preference ordering!!

c)

21
 c2 
   (1  r )   (Equation A)
 c1 
c 
ln 2   ln(1  r )  ln( ) (Equation B)
 c1 
c 
 ln  2 
EIS   c1   1
1
r RRA
For the natural logarithm utility function the elasticity of intertemporal substitution (EIS) is
constant and equal to 1.

This means that the income and substitution effects (of a change in the real interest rate) exactly
cancel out and leave a total effect of 1 (as the effect of a change in r on the ratio (c 2/c1) ). This
answer is independent on the initial levels of s 1, w1, w2, r, & β (i.e. independent of whether the
agent is initially a net saver or borrower or to what degree).

ASIDE: As described in question 2 (above) the EIS can be written as follows

  c2    1             
 d        d ln c2    d ln c2    d ln c2  
  c   1  r    c  
 1  
c     c 
EIS     1       1   1
  d ln 1  r    dr 
 1   c2     1 
 d       d ln     
  1  r    c     1  r      
   1        

Using the first expression on the right-hand-side & equation A (directly above) re-written to
c   1 
1

become  2   (1  r )        , we get:
 c1  1 r 

  c2    1     1  
 d       
 
  c1    1  r     1  2
   1  r  
EIS         x  1
 1    c2     1  r    1  1
 d      
  
  1  r    c    1 r  
   1     

Similarly, if we use equation B (directly above) & the second last version of the EIS we get:

      
 d ln c2    d ln c2  
 c    c  
EIS     1   1  1
 d ln 1  r    d ln 1  r  
   
   
   

22
Question 6
a)
The Lagrangian for this problem is:   u (c1 )    u (c2 )  [ w1 (1  r )  w2  c1 (1  r )  c 2 ]

With FOCs that simplify to the IBC & the Euler equation u ' (c1 )  (1  r )    u ' (c2 ) .

For these preferences we have u (c)  c & u ' (c)  0.5c 0.5
0.5

For the given r & β the Euler becomes: 0.5(c1 ) 0.5  (1.05)  0.95  0.5(c 2 ) 0.5
 c 2 0. 5 
 0.5   (1.05)  0.95  0.9975
c 
 1 
 c2 
   [(1.05)  0.95] 2  [0.9975]2  0.99500625
 c1 
Plugging this into the IBC yields: w1 (1  r )  w2  c1 (1  r )  c2
25,000(1.05)  20,000  c1 (1.05)  c1 [(1.05)  0.95]2
26,250  20,000  46,250  c1 (1.05  [(1.05)  0.95] 2 )  c1 (2.04500625)
46,250
c1 *   22,616 .0677993
2.04500625
c 2 *  [0.9975] 2  c1 *  22,503.128811

Savings
Period One Saving  s1 *  w1  c1 *  25,000  22,616.0677993  2,383.932201
Period Two Saving  s 2 *  w2  c2 *  20,000  22,503.128811  2,503.128811
Since period one savings is positive the representative agent is a net saver in period one. In period
two, their net savings is negative as they dis-save as they spend their period one savings (plus any
interest earned on it).

In a two period model, unless a specific period is referred to, we normally use the label “saving” to
refer to first period saving. So in this case the agent is a net saver!

ASIDE: Since there is no bequest motive over the agent’s whole life saving equals zero – in either
present or future value terms.
s *  2,503.128811
PV of Lifetime Saving  s1 *  2  2,383.932201  0
1 r 1.05
FV of Lifetime Saving  s1 * (1  r )  s 2 *  2,383.932201x(1.05)  2,503.128811  0
So in period one this agent is a net saver, in period two they are a net borrower (from themselves) and
over their whole lifetime they neither save or borrow in net (as they consumer all of their lifetime
resources & leave no bequest).

23
b)
Optimal first period utility = u (c1 *)  22,616 .0677993 0.5  150 .386395

Optimal second period utility (in 2nd period units) = u (c 2 *)  22,503 .1288110.5  150 .010429

Optimal second period utility (in 1st period units) = u (c2 *)  0.95(150.010429)  142.509908

Optimal level of lifetime utility (in 1st period units) = u (c1 *)  u (c2 *)  292.520337

c)

Yes, this person is certainly smoothing their consumption. From period 1 to 2, and vice versa, their
income fluctuates 20 to 25% while their consumption only fluctuates less than 1%. Given this, our
representative agent has smoothed their consumption over time (they were a net saver during their
high income period & a net borrower during their low income period).

d)

The representative agent has not perfectly smoothed their consumption as their consumption is not
exactly the same in each period (i.e. since c1* ≠ c2*).

We know that consumption will be perfectly smoothed by the agent who maximizes their level
of lifetime utility if, and only if, β(1+r) = 1.

For the generic r & β the Euler is: 0.5(c1 ) 0.5  (1  r )    0.5(c 2 ) 0.5
 c 2 0.5 
 0. 5   (1  r )    1
c 
 1 
 c2 
   [(1  r )   ] 2  [1] 2  1
 c1 
c2 *  c1 * Perfectly smooth consumption (over t)

Plugging this into the IBC yields: w1 (1  r )  w2  c1 (1  r )  c2

25,000(1  r )  20,000  c1 * (1  r )  c1 *
25,000(1  r )  20,000
c1 *   c2 *
2r

Since lifetime consumption will only be perfectly smoothed if, and only if, β(1+r) = 1, then since
β=0.95 for this agent’s preferences then their lifetime consumption will only be perfectly smoothed if
the real interest rate r = (1/β) – 1 = 0.05263157 = 5.263157%.

24
e)

 c 0. 5   c2 * 
For this agent the Euler equation is:  2 0.5   (1  r )   , or    [(1  r )   ] 2  [(1  r )  0.95] 2
c  c *
 1   1 
The IBC is: 25,000(1  r )  20,000  c1 * (1  r )  c2 *

If first period savings is zero then: s1 *  w1  c1 *  0  25,000  c1 , or c1  25,000 . Plugging this

into the IBC we get: 25,000(1  r )  20,000  (25,000)(1  r )  c 2 * , so c2 *  20,000 . Plugging this
into the Euler equation and we get: c 2 *  [(1  r )  0.95]2 c1 *
20,000  [(1  r )  0.95] 2 x 25,000
 20,000 
[(1  r )  0.95] 2   
 25,000 
 20,000 
0.5

[(1  r )  0.95]   
 25,000 
 
(1  r )  0.80 0.5 / 0.95

 
r  { 0.80 0.5 / 0.95}  1  0.941502  1  0.058497693  5.8497693%
The representative agent will only save (exactly) none of their first period income if the real
interest rate is (approximately) equal to -5.8497693%.

f)

 c 2 0. 5   c2 * 
For this agent the Euler equation is:  0.5   (1  r )   , or    [(1  r )   ] 2  [(1  r )  0.95] 2
c 
 1   c1 * 

The IBC is: 25,000(1  r )  20,000  c1 * (1  r )  c2 *

The Euler implies: c2 *  [(1  r )  0.95]2 xc1 *

Plugging this into the IBC yields: 25,000(1  r )  20,000  c1 * (1  r )  [(1  r )  0.95] xc1 *
2

25,000(1  r )  20,000
So, c1 *  s1 *  w1  c1 *
(1  r )  [(1  r )  0.95] 2
 25,000(1  r )  20,000 
c 2 *  [(1  r )  0.95]2 x  2
 (1  r )  [(1  r )  0.95] 

Initially r1 = 5%

25
 25,000(1.05)  20,000 46,250
c1 *    22,616 .067799
(1.05)  [(1.05)  0.95] 2
2.04500625
s1 *  w1  c1 *  25,000  22,616.067799  2,383.932201

 0.99500625x22,616.067799  22,503.128811
 25,000(1.05)  20,000 
c 2 *  [(1.05)  0.95] 2 x  2 
 (1.05)  [(1.05)  0.95] 

 c 2 *   22,503.128811 
      0.99500625
 c1 *   22,616.067799 

Later r2 = r1(1.01) = 5.05%

25,000(1.0505)  20,000 46,262.50
c1 *    22,606.175231
(1.0505)  [(1.0505)  0.95] 2
2.046454
s1 *  w1  c1 *  25,000  22,606.175231  2,393.824769
c 2 *  [(1.0505)  0.95] 2 x22,606.175231  22,514.714027
 c 2 *   22,514.714027 
      0.9959514956
 c1 *   22,606.175231 

Percentage changes (due to the 1% increase in r):

 c1   22,606.175231  22,616.067799    9.892568 
       0.0437%
 c1   22,616.067799   22,616.067799 
 c 2   22,514.714027  22,503.128811  11.585216 
       0.0515%
 c2   22,503.128811   22,503.128811 
  c2  
   
  c1    0.9959514956  0.99500625   .0009452456 
     0.095%
 c 
  2   0.99500625   0.99500625 
 c  
  1 

Now try r3 = r1 + 0.01 = 6%

25,000(1.06)  20,000 46,500
c1 *    22,419 .913898
(1.06)  [(1.06)  0.95] 2
2.074049
s1 *  w1  c1 *  25,000  22,606.175231  2,580.086102
c 2 *  [(1.06)  0.95] 2 x22,419.913898  22,734.891268
 c 2 *   22,734.891268 
      1.014048999999
 c1 *   22,419.913898 

Percentage changes (due to the 1 percentage point increase in r):

26
  c1   22,419.913898  22,616.067799    196.153901 
       0.86732%
 1  
c 22,616.067799   22,616.067799 
 c 2   22,734.891268  22,503.128811  231.762457 
       1.0299%
 c2   22,503.128811   22,503.128811
  c2  
   
  c1    1.0140489999  0.99500625   0.01904275 
     1.91%
  c 2    0.99500625   0.99500625 
 c  
  1 
Determine and explain the income and substitution effects this increase in the real interest rate
creates. Solving the lifetime utility max problem AND the min future value of lifetime income
problem generate the same optimal c1 & c2 values for the same indifference curve & IBC. So the
compensated and uncompensated demand curves have to give identical values for c 1 & c2. Thus,

c1*,comp ( r ,V )  c1* (r , I * (r ,V )) , where I = future value of lifetime income = (1+r)c1 + c2

c1*,comp c1* c1* I

Differentiating wrt r:   
r r I r

c1* c1,comp c1* I

*

Rearranging we get:   
r r I r

c1* c1,comp c1*

*

   c1
r r I
 Total   Sub   Income 
       
 Effect   Effect   Effect 
Substitution effect
If the utility function is smooth and differentiable the substitution effect is negative due to the
assumption of diminishing marginal rate of substitution (i.e. a higher real interest rate r makes c 2
relatively cheaper & c1 relatively more expensive – than before r changed – this tends to make c 1
fall & c2 rise – the substitution effect).
Income effect
Of course c1 is positive since it is a good (not a bad). Since c1 is a normal good the sign of the
c1*
derivative depends on whether the agent is a net saver or borrower at the initial optimal
I
consumption point. If the agent is a net saver (and they are at the initial interest rate r 1 = 5%) an
increase in r drives up their future value of lifetime income – this causes a positive income effect
(i.e. the higher r drives their future value of lifetime income up which makes them increase their
consumption of both c1 & c2 – the income effect). In this case, the substitution & income effects on
c1 are moving in opposite directions, so the over-all impact on c1 is ambiguous.

27
g)

For these preferences we observe:3

u ' (c)  a  c a 1  0, Positive MU
u ' ' (c)  a(a  1)  c a 2  0, Di min ishing MU

u ' ' (c) 1  a 0.5

ARA    
u ' (c ) c c

 1  a   0.5
u ' ' (c )
RRA   c
u ' (c )

1 1
EIS   2
RRA 1  a

Note: Our empirical estimate of EIS for a 1 percentage point increase in r is 1.91 which is fairly
close to the EIS figure of 2 (percent) calculated from the definitions.

Any power utility function written as u (c )  c , where a is a constant positive fraction exhibits not just positive &
3 a

diminishing marginal utility but also constant relative risk aversion (CRRA).

28
Question 7
The answer is uncertain. The representative agent maybe a net saver or a net borrower at the
initial utility maximizing point. If she’s initially at a level of consumption where she’s neither of
the above (i.e. neither a net saver of net borrower), an increase in income may make her a net saver
or a net borrower depending on her preferences. An increase in current period income (or even
future period income) will increase both current and future consumption (due to, or using, the
consumption smoothing motive – along with the fact that consumption in both periods is a normal
good).

29