Problem set 1

Consumption, saving, and the business cycle

Markus Roth
Chair for Macroeconomics
Johannes Gutenberg Universitt Mainz
November 5, 2010
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 1 / 58
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 2 / 58
Problem 1 (review of the traditional consumption function)
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 3 / 58
Problem 1 (review of the traditional consumption function)
The consumption function
Consider the consumption function
C = c
0
+ c
1
Y
d
. (1)
This function should be familiar to you from undergraduate
courses in macroeconomics.
C denotes the aggregate consumption level in the economy.
It is sometimes called Keynesian consumption function.
Y
d
is disposable income in the economy.
c
0
and c
1
are parameters characterizing the function.
Usually we restrict the parameters such that
c
0
> 0
0 < c
1
< 1.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 4 / 58
Problem 1 (review of the traditional consumption function)
Discussing the consumption function
c
0
is called autonomous consumption.
I.e. consumption if disposable income is zero.
We can have positive consumption in the presence of zero
disposable income because of dissaving.
Disposable income is dened as income net of taxes, i.e.
Y
d
Y T.
For consumption only the disposable income is relevant because
only this part of income can actually be consumed.
c
1
is called the marginal propensity to consume (MPC).
It is a natural assumption that 0 < c
1
< 1 because this means
consumption is a fraction of income.
We cannot consume more than we earn.
On the other hand, higher income should yield higher
consumption.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 5 / 58
Problem 1 (review of the traditional consumption function)
Problems with the function
The Keynesian consumption function is the traditional way
economists think of consumption.
However, there are some problems to this function.
The most pronounced problem is that it links consumption to
current income and disregards potential future earnings.
Lifetime income should be relevant for individuals consumption
saving decision.
Another point is that the function per se is not micro-founded, it is
set up by (reasonable) assumptions and empirical support.
By contrast, modern macroeconomic theory is based on
optimization problems of households, rms, central banks, ...
In the tutorials we will derive a micro-based justication of the
consumption function, where we substitute current income by
permanent (lifetime) income.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 6 / 58
Problem 2 (consumer maximization, 2 periods)
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 7 / 58
Problem 2 (consumer maximization, 2 periods)
Recall for problem 2 and 3 from the lecture
The capital accumulation equation is given by
k
s+1
= k
s
+ i
s
k
s
.
The national accounting identity (for a closed economy) is
y
s
= c
s
+ i
s
.
Putting both together yields
y
s
= c
s
+ k
s+1
(1 )k
s
c
s
+ k
s+1
= y
s
+ (1 )k
s
. (2)
The consumer can shift resources intertemporally by investing in
the capital stock.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 8 / 58
Problem 2 (consumer maximization, 2 periods)
The objective function
The objective function reads
max
c
1
,c
2
,k
2
V
1
= max
c
1
,c
2
,k
2
log c
1
+ log c
2
. (3)
This means that the household chooses consumption in period 1
(c
1
) and period 2 (c
2
) in order to maximize its lifetime utility
function which we call V
1
.
In this problem the lifetime of the household is two periods only.
One can think of two periods where the household is young and
old respectively.
0 < < 1 is the discount factor.
It represents impatience of the household.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 9 / 58
Problem 2 (consumer maximization, 2 periods)
The objective function/optimization problem
The utility the household gets from consuming in period 1 or 2 is
determined by the period utility function U().
In our particular case the objective function is logarithmic.
We usually assume that the objective function is concave.
Why we make this assumption will be discussed in later tutorials.
The household now chooses in every period of life how much to
consume and how much to save for the next period.
There is a trade-off between consumption today (period 1) and
tomorrow (period 2).
The household could consume more (save less) today but then it
has to consume less tomorrow (because of the low savings).
The trade-off comes by the budget constraints.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 10 / 58
Problem 2 (consumer maximization, 2 periods)
The constraints
We maximize the lifetime utility function V
1
with respect to the
budget constraints of the household.
If there would not be any constraint the household could simply
maximize its lifetime utility by consuming an innite amount of c
1
and c
2
.
However, this is not reasonable hence we maximize subject to
c
1
+ k
2
= k

1
+ (1 )k
1
(4)
c
2
= k

2
+ (1 )k
2
. (5)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 11 / 58
Problem 2 (consumer maximization, 2 periods)
Solution of the problem
In general there are three methods to solve such a dynamic
optimization problem.
1 Substitute the constraints into the objective function and compute
the rst derivative with respect to k
2
.
2 Set up the Lagrangian function L and compute the rst derivative
with respect to c
1
, c
2
and k
2
.
3 Use the Bellman equation to solve the problem (in this case trivial).
We will usually use the Lagrangian to solve the problem.
L = log c
1
+ log c
2
+
1
[k

1
+ (1 )k
1
c
1
k
2
]
+
2
[k

2
+ (1 )k
2
c
2
] (6)
Note, that there are more than one possibility to set up the
Lagrangian. We will see different approaches throughout the
tutorials.
However, the solution will always be identical, only the
interpretation of the Lagrange multipliers (
1
,
2
) changes.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 12 / 58
Problem 2 (consumer maximization, 2 periods)
FOC
The rst-order conditions are
L
c
1
=
1
c
1

1
!
= 0 (I)
L
c
2
=
1
c
2

2
!
= 0 (II)
L
k
2
=
1
+
2
_
k
1
+ (1 )
_
!
= 0. (III)
Rearranging (III) yields

1
=
2
_
k
1
+ 1
_
.
Substituing (I) and (II) gives
1
c
1
=
1
c
2
_
1 + k
1
2

_
(7)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 13 / 58
Problem 2 (consumer maximization, 2 periods)
Euler equation
Equation (7) is known as the Euler equation.
The Euler equation describes the optimal consumption path.
It equates the marginal consumption today with the marginal
consumption tomorrow (from saving) discounted by .
Note that 1 + k
1
t+1
can be interpreted as an interest rate.
In the optimum the consumer cannot improve her utility by
shifting consumption intertemporally.
We can rewrite the equation to
U

(c
1
)
U

(c
2
)
= 1 + k
1
t+1
. (8)
Here we equate the marginal rate of substitution between
consumption today and tomorrow (LHS) and the marginal rate of
transformation (RHS).
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 14 / 58
Problem 3 (consumer maximization, innite periods)
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 15 / 58
Problem 3 (consumer maximization, innite periods)
Generalizing the problem
We now turn to a problem where the lifetime of households has
innitely many periods.
This is more general than the model in the previous question.
We can rationalize the innite horizon setup with household
bequests to younger generations.
Hence, we view the representative household as a family
dynasty.
However, the basic solution strategy and the economic
implications do not change.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 16 / 58
Problem 3 (consumer maximization, innite periods)
The problem
Objective
max
{c
t+s
,k
t+s+1
}

s=0
V
t
= max
{c
t+s
,k
t+s+1
}

s=0

s=0

s
log c
t+s
(9)
subject to
c
s
+ k
s+1
= k

s
+ (1 )k
s
for all s 0. (10)
Again we use the Lagrangian to solve the problem.
The only difference is that we have an innite number of
rst-order conditions.
The Lagrangian to this problem is
L =

s=0

s
{logc
t+s
+
t+s
[k

t+s
+ (1 )k
t+s
c
t+s
k
t+s+1
]} .
(11)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 17 / 58
Problem 3 (consumer maximization, innite periods)
The Lagrangian
We have to take partial derivatives with respect to c
t+s
and k
t+s+1
.
We write the sum as
L =
0
{log c
t
+
t
[k

t
+ (1 )k
t
c
t
k
t+1
]}
+
1
{log c
t+1
+
t+1
[k

t+1
+ (1 )k
t+1
c
t+1
k
t+2
]} +
+
s
{logc
t+s
+
t+s
[k

t+s
+ (1 )k
t+s
c
t+s
k
t+s+1
]}
+
s+1
{log c
t+s+1
+
+
t+s+1
[k

t+s+1
+ (1 )k
t+s+1
c
t+s+1
k
t+s+2
]} + .
Now we can see where we nd c
t+s
and k
t+s+1
.
c
t+s
appears only in the third line.
k
t+s+1
appears in the third and fourth line.
For a moment we can forget the following lines.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 18 / 58
Problem 3 (consumer maximization, innite periods)
FOC
The rst order conditions are
L
c
t+s
=
s
_
1
c
t+s

t+s
_
!
= 0 (I)
L
k
t+s+1
=
s+1

t+s+1
_
k
1
t+s+1
+ 1
_

s

t+s
!
= 0, (II)
s = 0, 1, 2, . . . , .
Note that actually we have innitely many FOCs.
Rewriting (II) yields

t+s
=
t+s+1
_
k
1
t+s+1
+1
_
. (12)
We can forward (I) in order to get

t+s
=
1
c
t+s

t+s+1
=
1
c
t+s+1
. (13)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 19 / 58
Problem 3 (consumer maximization, innite periods)
Euler equation
Substituting (13) into (12) gives
1
c
t+s
=
_
1 + k
1
t+s+1

_

1
c
t+s+1
. (14)
Equation (14) is the Euler equation for the innite horizon case.
The interpretation is analogously to the two period case.
The Euler equation describes the optimal consumption path.
It equates the marginal consumption today with the marginal
consumption tomorrow (from saving) discounted by .
In the optimum the consumer cannot improve her utility by
shifting consumption intertemporally.
Again we could rewrite it to equate the MRS and MRT.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 20 / 58
Review: Phase diagrams
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 21 / 58
Review: Phase diagrams
Important properties of functions
For this kind of analysis and also for other considerations we need
to know some basic properties about utility- and production
functions.
A utility function U() usually is assumed to be concave.
U

() > 0 means that more is always better.

U

() < 0 means that an additional unit of the argument (e.g.

consumption) increases utility but to a smaller extent than the unit
before.
Marginal utility is positive but diminishing in the argument.
We impose a similar assumption on the production function.
We assume that F(k
t
) is also concave.
Hence, an increase in k
t
increases production F() but decreases
marginal production F

().
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 22 / 58
Review: Phase diagrams
Some examples
If you have problems with the general denition of concavity
consider the following examples:
Log: U(c
t
) = ln(c
t
)
Power: U(c
t
) =
c
1
t
1
1
(with > 0)
Exponential: e
c
t
(with > 0)
Production: F(k
t
) = k

(with 0 < < 1)

Try to plot those functions with a program or by using pen and
paper to get intuition for the shape.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 23 / 58
Review: Phase diagrams
f (x)
x
Figure: Concave function
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 24 / 58
Review: Phase diagrams
Approximating the Euler equation
Consider the expression of the optimality conditon (Euler
equation) you have derived in the lecture
U

(c
t+1
)
U

(c
t
)
_
F

(k
t+1
) +1

= 1. (15)
Furthermore consider the capital accumulation equation
k
t+1
k
t+1
k
t
= F(k
t
) k
t
c
t
. (16)
Note that you have seen both equations in the lecture.
We take a rst order Taylor approximation of U

(c
t+1
) around c
t
U

(c
t+1
) U

(c
t
) +c
t+1
U

(c
t
).
Such an approximation is good in the neighborhood of c
t
.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 25 / 58
Review: Phase diagrams
Rearranging
Dividing by U

(c
t
) yields
U

(c
t+1
)
U

(c
t
)
1 +
U

(c
t
)
U

(c
t
)
c
t+1
. (17)
We know that by (reasonable) assumption
U

(c
t
)
U

(c
t
)
0.
We rearrange (15)
U

(c
t+1
)
U

(c
t
)
=
1
[F

(k
t+1
) +1 ]
. (18)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 26 / 58
Review: Phase diagrams
Substituting
We substitute (17) for the left hand side in (18) and get
c
t+1
=
U

(c
t
)
U

(c
t
)
_
1
1
[F

(k
t+1
) + 1 ]
_
. (19)
(typo in [Wickens, 2008]!)
The capital accumulation equation was given by
k
t+1
= F(k
t
) k
t
c
t
. (16)
With equations (19) and (16) we have a two-variable system of
two (still nonlinear) difference equations.
Since the system consists of two nonlinear difference equations
there is no easy way to solve them analytically.
However, we can use phase diagrams to understand the system.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 27 / 58
Review: Phase diagrams
Zero motion line 1
First we determine the loci where k
t+1
= 0 and c
t+1
= 0.
We call the result zero motion lines.
0 =
U

(c
t
)
U

(c
t
)
_
1
1
[F

(k
t+1
) +1 ]
_
F

(k
t+1
) =
1

. .

+
F

(k
t+1
) = +
This equation implicitly denes a constant zero motion line where
c
t+1
= 0, i.e. consumption does not change.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 28 / 58
Review: Phase diagrams
Digression: discounting the future
There are two different ways to express that agents are impatient.
Usually we assume a discount factor 0 < < 1.
Thus future period utility functions are multiplied by this factor
expressing that utility tomorrow is worth less than utility today.
However, sometimes it is convenient to think of impatience as
discounting the future.
Thus, we need a concept similar to the concept of interest rates,
where future values are dicounted by (1 + )
1
.
We can write
=
1
1 +
=
1

.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 29 / 58
Review: Phase diagrams
Zero motion line 1
Recall that on the zero motion line c
t+1
= 0.
c
t+1
=
U

(c
t
)
U

(c
t
)
_
1
1
[F

(k
t+1
) + 1 ]
_
. (19)
In equation (19) suppose starting from the zero motion line we
increase k
t
a little bit.
What sign does c
t+1
then have?
It is negative.
Why is this?
If k
t+1
increases F

(k
t+1
) decreases (F

() is concave).
Then the fraction (without minus sign) increases.
Since U

/U

is negative, the whole expression decreases.

c
t+1
< 0 when we increase k
t+1
.
The opposite is true when we decrease k
t+1
.
We bring this information into a k
t+1
-c
t
-diagram.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 30 / 58
Review: Phase diagrams
k
t+1
c
t
k

c
t+1
> 0 c
t+1
< 0
c
t+1
= 0
Figure: c
t+1
diagram
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 31 / 58
Review: Phase diagrams
Zero motion line 2
The second zero motion line is found by setting k
t+1
= 0
0 = F(k
t
) k
t
c
t
c
t
= F(k
t
) k
t
.
This is a concave function.
Consider fromk
t+1
= 0 an increase in c
t
in equation (16)
k
t+1
= F(k
t
) k
t
c
t
. (16)
We nd that k
t+1
< 0.
We bring this two another diagram in the same k
t+1
-c
t
-space.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 32 / 58
Review: Phase diagrams
k
t+1
c
t
k
t+1
> 0
k
t+1
< 0
k
t+1
= 0
Figure: k
t+1
diagram
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 33 / 58
Review: Phase diagrams
Combining both diagrams
We combine both diagrams.
Therefore we use all information we have accumulated by our
analysis.
What we can see from the resulting diagram is that there is an
intersection point where k
t+1
= c
t+1
= 0.
We usually call this point the steady state.
Furthermore we can draw a stable arm which has the property
that the system moves towards the steady state.
If we are not on this line, the system does not converge to the
steady state.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 34 / 58
Review: Phase diagrams
k
t+1
c
t
k

Figure: Combined diagram

Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 35 / 58
Problem 4 (empirical relevance)
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 36 / 58
Problem 4 (empirical relevance)
Agenda
Open the dataset and inspect the content.
Take logs of variables.
Detrend the data using the HP lter.
Compute the long-run and short-run component.
Plot both components.
Replicate the stylized facts of business cycles.
Compute standard deviations and correlation coefcients.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 37 / 58
Problem 4 (empirical relevance)
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
1970 1975 1980 1985 1990 1995 2000 2005
Time
HP-filter
y
y (long run)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 38 / 58
Problem 4 (empirical relevance)
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
1970 1975 1980 1985 1990 1995 2000 2005
Time
Business cycles
y
c
i
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 39 / 58
Problem 4 (empirical relevance)
Some stylized facts of business cycles
1 Investment is much more volatile than output.
2 (Private) consumption and investment are strongly correlated
with output.
There is co-movement between the three variables.
3 There is persistence in output and consumption.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 40 / 58
Problem 5 (HP-lter)
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 41 / 58
Problem 5 (HP-lter)
The minimization problem
The minimization problem of the HP lter is
min
{y
lr
t
}
T

t=1
_
y
t
y
lr
t
_
2
+
T1

t=2
__
y
lr
t+1
y
lr
t
_

_
y
lr
t
y
lr
t1
__
2
. (20)
The HP lter is a compromise between the two objectives.
Minimize the square deviation of the short-run component to trend.
Minimize the square change in the growth rate of the long-run
component
We are free to choose the relative weight of both objectives.
For illustration we consider the following two extreme cases
= 0
.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 42 / 58
Problem 5 (HP-lter)
= 0
For = 0 the second objective is switched off.
This means that we are only interested in minimizing the squared
deviation between y
t
and its long-run component y
lr
t
.
Since we choose y
lr
t
in order to achieve our objective we set
y
lr
= y
t
.
This means in turn that we interpret the actual time series y
t
as
consisting solely of a long-run component.
At the same time we decide that there is no short-run component
in the actual time series y
t
.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 43 / 58
Problem 5 (HP-lter)

For the rst objective is switched off.
This means that we are only interested in minimizing the squared
change in y
lr
t
.
In the limiting case this means that we assume a constant change
in y
lr
t
.
Thus also the growth rate of y
lr
t
is assumed to be constant.
The long-run component follows a linear time trend.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 44 / 58
Problem 5 (HP-lter)
Conclusion
Of course, we choose some value between both extreme cases.
Hence, we nd the optimal compromise between both objectives.
There is no right choice of but most researches agree with
= 1600 for quartely data.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 45 / 58
Problem 5 (HP-lter)
Solving the minimization problem
In order to solve the problem we have to distinguish ve different
cases.
This means that the derivatives for some periods are different.
More precisely, we compute the derivatives with respect to y
lr
1
, y
lr
2
,
y
lr
t
, y
lr
T1
and y
lr
T2
.
The rst order condition with respect to y
lr
1
is given by
2(y
1
y
lr
1
) + 2(y
lr
3
2y
lr
2
+ y
lr
1
)
!
= 0.
We solve this expression for y
1
, this yields
y
1
= y
lr
1
(1 + ) 2y
lr
2
+ y
lr
3
. (21)
We do the same for the remaining rst order conditions.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 46 / 58
Problem 5 (HP-lter)
The remaining periods
The derivative with respect to y
lr
2
is given by
2(y
2
y
lr
2
) +2(y
lr
3
2y
lr
2
+ y
lr
1
)(2) +2[y
lr
4
2y
lr
3
+ y
lr
2
]
!
= 0.
Solving this for y
2
gives
y
2
= 2y
lr
1
+ (1 + 5)y
lr
2
4y
lr
3
+ y
lr
4
. (22)
The derivative with respect to y
lr
t
reads
2(y
t
y
lr
t
) +2(y
lr
t
2y
lr
t1
+ y
lr
t2
) +
2(y
lr
t+1
2y
lr
t
+ y
lr
t1
)(2) + 2(y
lr
t+2
2y
lr
t+1
+ y
lr
t
)
!
= 0.
We solve again for y
t
y
t
= y
lr
t2
4y
lr
t1
+ (1 + 6)y
lr
t
4y
lr
t+1
+ y
lr
t+2
. (23)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 47 / 58
Problem 5 (HP-lter)
The remaining periods
We do not derive the remaining two derivatives (since the
problem is symmetric), they are given by
y
T1
= y
lr
T3
4y
lr
T2
+ (1 + 5)y
lr
T1
2y
lr
T
(24)
y
T
= y
lr
T2
2y
T1
+ (1 + )y
lr
T
. (25)
Having computed the rst order conditions, we can state them in
matrix notation.
We dene the following T 1 (column) vectors
y
_
_
_
_
_
y
1
y
2
.
.
.
y
T
_
_
_
_
_
and y
lr

_
_
_
_
_
y
lr
1
y
lr
2
.
.
.
y
lr
T
_
_
_
_
_
. (26)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 48 / 58
Problem 5 (HP-lter)
Matrix notation
In addition we dene the T T matrix
A
_
_
_
_
_
_
_
_
_
1 + 2 0 0 0 0 0
2 1 + 5 4 0 0 0 0
4 1 +6 4 0 0 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 0 0 4 1 + 5 2
0 0 0 0 0 2 1 +
_
_
_
_
_
_
_
_
_
.
The rst two rows of this matrix contain the derivatives with
respect to y
lr
1
and y
lr
2
, the last two rows contain the derivative with
respect to y
lr
T1
and y
lr
T2
and the remaining T 4 rows contain the
derivative with respect to y
lr
t
on the diagonal band.
We can write the system as
y = Ay
lr
.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 49 / 58
Problem 5 (HP-lter)
Matrix notation
Solving the system for y
lr
yields
y
lr
= A
1
y.
we have derived a closed form solution for the HP-lter.
Recall that y is a given data vector and A
1
only depends on
which we are free to choose (recall the previous discussion).
We are now prepared to implement the HP-lter into a
matrix/vector based programming language.
However, most statistical software packages already contain the
HP-lter.
If we are interested in the time series of the cyclical component
{y
sr
t
}
T
t=1
we simply use the residual
y
t
= y
lr
t
+ y
sr
t
y
sr
t
= y
t
y
lr
t
. (27)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 50 / 58
Problem 6 (consumer maximization with labor )
Contents
1 Problem 1 (review of the traditional consumption function)
2 Problem 2 (consumer maximization, 2 periods)
3 Problem 3 (consumer maximization, innite periods)
4 Review: Phase diagrams
5 Problem 4 (empirical relevance)
6 Problem 5 (HP-lter)
7 Problem 6 (consumer maximization with labor )
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 51 / 58
Problem 6 (consumer maximization with labor )
Interpretation of the problem
Similar to the maximization problems discussed above the
representative household maximizes lifetime utility.
Regarding consumption c
t
we consider a logarithmic period
utility function which is concave.
The diffference to the usual problem is that in addition we have
labor n
t
in the utility function.
This means that the consumer also has to choose the optimal
amount of labor she/he will supply.
Hence, in addition to previous problems we have to differentiate
with respect to n
t
.
What actually enters positively in the utility function is not labor
but (1 n
t
) which can be thought of leisure time.
Labor enters negatively in the utility function.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 52 / 58
Problem 6 (consumer maximization with labor )
Solving the problem
We substitute the production function into the capital
accumulation equation and get the period budget constraint
k
t+1
+ c
t
= (1 )k
t
+ (a
t
n
t
)

k
1
t
. (28)
Again we use the Lagrangian to solve the problem
L =

s=0

s
_
ln(c
t+s
) +
(1 n
t+s
)
1
1
_
+

t+s
_
(a
t+s
n
t+s
)

k
1
t+s
k
t+s+1
c
t+s
_
.
The rst order conditions are given by
L
c
t+s
=
s
1
c
t+s

t+s
!
= 0 (I)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 53 / 58
Problem 6 (consumer maximization with labor )
Solving the problem
L
k
t+s+1
=
t+s+1
_
1 + (1 + )
_
a
t+s+1
n
t+s+1
k
t+s+1
_

t+s
!
= 0 (II)
L
n
t+s
=
s
_
(1 n
t+s
)

+
t+s
a

t+s
_
k
t+s
n
t+s
_
1
!
= 0 (III)
Combining (I) with (II) and (III) yields
1
c
t+s
=
1
c
t+s+1
_
1 + (1 + )
_
a
t+s+1
n
t+s+1
k
t+s+1
_

_
(29)
(1 n
t+s
)

=
1
c
t+s
a

t+s
_
k
t+s
n
t+s
_
1
. (30)
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 54 / 58
Problem 6 (consumer maximization with labor )
Interpretation of the results
Equation (29) is the usual Euler equation.
The Euler equation is an intertemporal optimality condition.
Since in the objective function we assumed that labor is additive
seperable, labor does not inuence the Euler equation, the
interpretation stays the same as in previous problems.
Equation (30) is an implicit expression for optimal labor supply of
households.
It is also independent of consumption because we have assumed
additive seperability (instead of a multiplicative specication) in
the lifetime utility function.
Equation (30) determins how much labor households want to
supply in a given period.
Note that in contrast to (29) (30) is an intratemporal optimality
condition.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 55 / 58
Problem 6 (consumer maximization with labor )
The RBC model
Having derived those optimality conditions and assuming a
stochastic process for technology, one can set up the so-called real
business cycle (RBC) model.
In this model business cycles are generated by technology shocks
alone.
The RBC model is a dynamic stochastic general equilibrium
model.
This means that...
... that it can describe a time path of variables
... it has a stochastic component (technology shocks)
... prices in the model (such as the interest rate) are determined by
agents in the model
This kind of models can explain basic business cycle facts such as
volatility, correlations and autocorrelations.
However, it has been found that technology shocks are not the
source of business cycles.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 56 / 58
Problem 6 (consumer maximization with labor )
The RBC model
In addition to the households who maximize their lifetime utility
by choosing their consumption path and labor supply, there are
rms in the model who produce the consumption good.
In the standard RBC-model equations are log-linearized.
Then one can wirte a computer program to simulate the model.
Analysis in this model is usually done by inspecting impulse
response functions (IRFs) of main economic variables.
In addition we can generate articial data of output, the real
interest rate, investment, consumption and labor.
Descriptive statistics of those series will be close to the results we
obtained by the HP-lter.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 57 / 58
References
References
Hodrick, R. J. and Prescott, E. C. (1997).
Postwar U.S. Business Cycles: An Empirical Investigation.
Journal of Money, Credit and Banking, 29(1):116.
Srensen, P. B. and Whitta-Jacobsen, H. J. (2005).
Introducing Advanced Macroeconomics: Growth & Business Cycles.
McGraw-Hill Education.
Wlde, K. (2009).
Applied Intertemporal Optimization.
Lecture Notes, University of Mainz, Available at
http://www.waelde.com/aio.
Wickens, M. (2008).
Macroeconomic Theory: A Dynamic General Equilibrium Approach.
Princeton University Press.
Markus Roth (Advanced Macroeconomics) Problem set 1 November 5, 2010 58 / 58