Lecture Note For Elements of Economic Analysis IV1

Lecture Note for
Elements of Economic Analysis IV1
Seung Mo Choi
(choism@uchicago.edu)
November 27, 2006
Used for Elements of Economic Analysis IV (ECON 203), Section 03, Fall 2005 and Fall
2006, University of Chicago. The course is for eleven weeks including mid-term and nal exams.
Contents
1 Labor Market Policy
1.1
Unemployment Insurance
1.2
Minimum Wage, Labor Union and Firing Restriction
1
1
5
2 Fiscal Policy (I)

2.1
Lump-Sum Tax
2.2
Distortionary Tax: Capital Income Tax
18
18
22
3 Fiscal Policy (II)

3.1
Intertemporal Budget Constraint, Government Debt and
Lump-Sum Tax (Ricardian Equivalence)
3.2
Government Debt and Distortionary Tax (Ramsey Problem)
3.3
Social Security
32
4 Monetary Policy (I)

4.1
Seigniorage and In ation
4.2
Price Level, Output and (Un)Employment (Monetary
Neutrality)
55
55
5 Monetary Policy (II)

5.1
In ation, Unemployment and Optimal Monetary Policy
(Expectations-Augmented Phillips Curve)
5.2
Interest Rate and Optimal Monetary Policy (Friedman
Rule)
75
32
40
45
63
75
79
6 International Trade
6.1
Exchange Rate (Purchasing-Power Parity)
6.2
Gains from International Trade (Comparative Advantage)
6.3
International Trade and Economic Growth
89
89
95
101
7 Asset Pricing
105
ii
Contents
7.1
7.2
Fundamental Equation of Asset Pricing

Contingent Claims (Lucas-Tree Model)
8 Appendix: Tables and Figures
105
112
125
Chapter 1
Labor Market Policy
This chapter discusses labor market policies. The effects of (i) unemployment
insurance, (ii) minimum wage, (iii) labor union, and (iv) ring restriction are
discussed.
1.1
Unemployment Insurance
Question: Does a More Generous Unemployment Insurance Increases the

Long-Term Unemployment? (Reference: Chapter 10 of Barro. Nan Li's
ECON 203 Material.)
The unemployment insurance provides unemployed workers with income
while they are searching for new jobs. Before we start any theoretical analysis, it
is useful to understand related concepts clearly.
(1) Unemployment (U ): The number of people who are looking for work
but have no job. (On the other hand, employment (L) is the number of people
who have jobs.)
(2) Labor force: The sum of unemployment and employment. (Hence, people who neither have a job nor are looking for one (e.g., students) are classi ed
as outside of the labor force.)
(3) Unemployment rate (u): Unemployment divided by labor force (i.e.,
U=(U + L)).
- The unemployment rate is usually countercyclical. When the economy is
good (bad), the unemployment rate is low (high).
- A stable level of unemployment rate is often called the natural unemployment rate.
Some datasets regarding the unemployment can be downloaded at the Bureau of Labor Statistics website (http://www.bls.gov/ s/availability.htm). Also,
Bene ts and Wages: OECD Indicators (2002, 2004) published by the OECD
(available at the university library website) describes the unemployment insurance systems of various countries. [Show a few gures.] According to the
data, some economies in Western Europe (notably Spain, Italy and France) have
relatively higher unemployment rates. Also, the duration of unemployment is
1
Chapter 1 Labor Market Policy
much shorter in the U.S. relative to many European countries.

Higher rate of (long-time) unemployment in Europe may be because European countries, in general, provide more generous unemployment insurances.
This theory will become clearer after we introduce the following model.
1.1.1
Model Description and Solution
The generosity of unemployment compensation may means either more money

(amount) or a longer period (duration), or perhaps both. In our model, assume that the unemployment compensation is paid forever as long as a worker
stay unemployed. This enables us to focus on the amount rather than the duration.
(We will think about the duration in Exercises.)
Consider an in nitely-lived worker, who is initially unemployed at the current
period (say, period 0). She meets one employer at each period of job search. The
employer may offer a high wage wh or a low wage wl . The probability that a
high wage is offered is p. Looking at the wage offer, the worker may accept or
reject: If she accepts the offer, she receives the speci ed wage at each period,
starting from this period. To make the problem simple, assume there is no risk of
voluntary or involuntary job loss. If she rejects, she receives the unemployment
bene t b (where b < wl < wh ) at this period, and continues to search at the next
period. The environment in the next period (say, period 1) is exactly the same as
the environment in this period (say, period 0) if she rejects the offer. That is, she
again meets an employer who offers wh or wl in the next period. Our interest is
to see what happens to this unemployed worker's decision when b increases (i.e.,
the unemployment insurance becomes more generous).
Before we start any analysis, let us think about what this worker gets if she
accepts the wage offer. If a high wage wh is offered and accepted by this worker,
she gets wh dollars at each period. In terms of utility level, this is u(wh ), so the
discounted value of the utility levels is
u(wh ) + u(wh ) +
P1
using t=0 = 1=(1
and accepted, she gets
u(wh ) + ::: =
1
X
u(wh ) =
t=0
) for 0 <
u(wl ) + u(wl ) +
u(wh )
;
1
(1.1)
< 1. Similarly, if a low wage is offered
u(wl ) + ::: =
1
X
t=0
u(wl ) =
u(wl )
1
(1.2)
1.1 Unemployment Insurance
units of discounted value of utility.

Now let's think about the decision of this worker in detail. First, suppose a
high wage (wh ) is offered. The unemployed worker has two options: accept
or reject. But it is clear that she will accept the offer because it is the best
possible offer. If she rejects the offer, she will only receive b (unemployment
compensation) in this period, and she will have to draw another wage offer in the
future which can never be higher than wh . So it is not a reasonable solution for
her to reject a high wage offer.
Summary: In this model, if a high wage is offered, the worker will de nitely
accept it.
But things are somewhat different if a low wage (wl ) is offered. If the worker
rejects, she will only receive b at this period, but instead, she will have a chance
to draw a high wage offer (wh ) in the next period. That is: (i) Cost of rejecting:
Receives only b which is smaller than wl . (ii) Bene t of rejecting: Has a chance to
win a higher wage (wh ) in the future. Hence, to determine the worker's decision,
we have to compare the discounted utility values of accept and reject.
First, let us see her discounted utility level when she accepts the offer, denoted by Vaccept . From (1.2), it is clear that
Vaccept =
u(wl )
:
1
(1.3)
Second, consider her discounted utility level when she rejects the offer, denoted by Vreject . Assume the worker is consistent in her decision. That is, if
she chooses to reject a low-wage offer at this period, then she will continue to
do so when a low wage is offered in later periods. (This assumption is relevant
because the environment tomorrow is exactly the same as today's if she rejects.)
If she rejects, she gets b at this period, so we can write
Vreject = u(b) +
(discounted expected utility of the next period):
(1.4)
Then, at the next period, a high wage is offered with probability p, which will
give (1.1) in terms of utility level, and a low wage is offered with probability
1 p, which will, again, give Vreject (although we do not know the value of
Vreject yet). Then, (1.4) becomes
Vreject = u(b) +
u(wh )
+ (1
1
p)Vreject :
We can solve this equation for Vreject . That is,
[1
(1
p)]Vreject = u(b) + p
Vreject =
u(b) + pu(wh )=(1

1
(1 p)
u(wh )
;
1
or equivalently,
1.1.2
(1.5)
Discussion
The bottom line is that a worker with a low-wage offer will reject it if
Vreject > Vaccept ;

or equivalently, from (1.3) and (1.5),
u(b) + pu(wh )=(1

1
(1 p)
>
u(wl )
:
1
(1.6)
Whether this inequality holds depends on the values of parameters (e.g., , p,

wh , and wl ). But it is clear that this inequality is more likely to hold as b increases. That is, suppose (1.6) does not hold when b = 0 so that an unemployed
worker accepts the low-wage offer. As you increase b, the left-hand side of (1.6)
increases, so it may be the case that (1.6) now holds if b is big enough. Therefore,
a more generous unemployment bene t is expected to delay the acceptance of a
job offer.
Summary: In this model, if a low wage is offered, the worker's decision
depends on parameter values. But the worker is more likely to reject the offer as
the unemployment bene t becomes more generous.
In many European economies, it takes longer for an unemployed worker to
accept a new job relative to the U.S. Perhaps one of the reasons is because the
European system of unemployment insurance is more generous. You have to understand that the introduction of this unemployment insurance has two different
effects, positive and negative: (i) A positive effect is that an unemployed worker
can be protected while she is unemployed. (ii) A negative effect is that this insurance will delay the job acceptance of unemployed worker and thereby increase
the long-term unemployment. (Needless to say, this program also requires more
money collected from tax-payers, which will distort the economy. We will think
about distortionary effects of taxation in the next chapter.) We have to design the
1.2 Minimum Wage, Labor Union and Firing Restriction
optimal insurance system by considering these two effects together, which is one
of major challenges in macroeconomics.
The model introduced in this section is simple. (In fact, our conclusion is
so clear that we do not need such a mathematical model.) But this becomes a
nice tool to analyze more complicated problems. You will see some examples in
Exercises.
True or False?
1. We should not offer the unemployment insurance because it will only
distort the worker's decision.
2. The United States pays the unemployment bene ts for about six months
after a worker gets red. As in many European countries, the United States
should lengthen the duration of the bene ts, for example, to two years.
1.2
Minimum Wage, Labor Union and Firing Restriction
Question: How Do the Minimum Wage, Labor Union, and Firing Restriction Affect the Economy? (Reference: Chapter 10 of Barro.)
This section brie y discusses three other labor market policies.
1.2.1
Minimum Wage
Think about the labor market for low-skilled workers. Just as any prices, their
wage rates are determined in the market, based on a demand-supply story. A
graphical analysis makes it clear that if the minimum wage is higher than the
equilibrium wage, a smaller number of workers will be employed because rms
need to pay more for each worker hired. [Draw a graph.] Hence, the minimum wage policy may increase the wage rate of the employed workers who are
low-skilled, but it leaves more workers unemployed. A positive effect is that
some employed workers can be protected, but a negative effect is that it may
decrease the employment of low-skilled workers. The size of this negative effect
depends on the shapes of demand and supply curves. [Show elastic/inelastic
demand/supply curves.]
Although the systems are slightly different across states, the U.S. federal minimum wage is currently $5.15 per hour. Less than 3% of jobs pay the minimum
wage.
6
1.2.2

Labor Union
The labor union tries to protect union members. In some industries, unions are
strong enough to raise the wage rates (for example, by threatening strikes). Here
we will consider a simple model regarding the consequences of union activities.
The environment here is similar to the model about the minimum wage, in the
sense that the wage rate is set arti cially high relative to the equilibrium wage
rate. One difference is that labor unions are strong only in some industries, so the
wage rates are arti cially set in those industries, but are determined in the market
equilibrium in others. We will see that workers in not unionized industries may
have some side-effects.
Suppose there are two industries. Each industry has 5 rms. All 10 rm are
identical and have the following production function:
p
Y = 2 L;
(1.7)
where Y is the output (in terms of dollars) and L is the labor input (in terms of
workers). There are 1,000 workers, inelastically supplied. (This means that all of
them are employed and that the wage rate is determined so that all are employed.
This assumption makes the problem simple: Without it, we have to consider the
choice of workers on work and leisure and to determine which workers work
for how many hours, etc.) Finally, let us denote the wage per worker (in terms
of dollars) by w. We will rst think about how the equilibrium wage rates are
determined without unions. Then we introduce unions to see how the wage rates
are determined.
1. No Union: Assume there is no labor union. Each rm maximizes its pro t
(output minus cost),
p
2 L wL;
where w is treated as given. The rst-order condition implies
1
p
L
w = 0;
or equivalently,
1
:
(1.8)
w2
This implies that given the level of w, each rm hires L = w12 workers. So for
given w, the economy (i.e., all 10 rms) demands w102 workers. Since there are
L=
1,000 workers in this economy, a market clearing (quantity demanded equals

quantity supplied) implies
10
= 1000;
w2
or equivalently,
w = 0:1:
Under this wage rate of 0.1 dollars per worker, each rm hires 100 worker (i.e.,
L = 100) from (1.8) and produces $20 (i.e., Y = 20) from (1.7). Since there
are 10 rms each of which produces $20, the GDP is $200.
2. Unionization: Assume there is a labor union in one industry, and it forces
the industry (i.e., 5 rms) to raise the wage rate from $0:1 to $0:2. There are
no unions in the other industry. From (1.8), if the wage is given by w1 = 0:2,
the labor demand of each rm in this unionized industry becomes 25. So this
industry hires 25 5=125 workers. Each rm produces $10 from (1.7), so this
industry produces $50.
Then, the wage rate for the other (not unionized) industry is determined so
that the remaining 1000 125=875 workers are all hired in an equilibrium. The
labor demand for each rm is (1.8), and there are 5 rms in this industry, so the
market-clearing condition implies
5
= 875;
w22
1
or equivalently, w22 = 175
. Solving this equation, we have the equilibrium
wage of 0:08. Each rm in this industry hires 875=5=175 workers, so (1.7)
implies that each rm produces $26.5 in value. Hence, the industry produces
$26.5 5=$132.5.
Therefore, the entire economy (i.e., two industries combined) produces $50+$132.5=$182.5,
which becomes the new GDP. This level is lower than the previous GDP level of
$200.
3. Conclusion: The intuition is clear: There are labor unions in some industries, whose purposes are to protect the workers hired in those industries.
As the wages of union members arti cially increase due to union activities, less
workers will be hired in those industries. So more workers need to be employed
in other industries, which will reduce the wage rates there. Furthermore, this will
decrease the GDP level because the rms under unionized industries, forced to
pay higher wage rates, are not able to produce at optimal levels.
In sum, labor unions help the workers to get what they deserve, but their
activities may also have some side effects. We are not saying that union activities
should be discouraged. But we should remember that as always, there are positive
and negative effects of their activities.
1.2.3
Firing Restriction
To protect the employed workers, many countries enact restrictions so that ring
a worker is costly or dif cult. Obviously, a positive effect is that the employed
workers are protected with more secure jobs, but a negative effect is that the
rm's decision would be distorted: Because it is costly or dif cult to re workers
later, rms will hire less workers from the beginning. Let's consider a model to
make this intuition clear.
For our purposes, we move our attention from how equilibrium wage rates
are determined. Rather, we introduce a partial equilibrium model: There are a
lot of rms in the economy, and the wage is set to be w = 1 in the labor market.
We consider a rm, which is small in the sense that it cannot affect the wage level
of the market. (So in this model, we treat the wage rate as given.) There are two
periods in this economy, but the production functions are different:
p
Y1 = 4 L1 at period 1;
p
Y2 = 2 L2 at period 2;
where Y1 and Y2 are outputs (in dollars) at periods 1 and 2 and L1 and L2 are
labor inputs (in number of workers) at period 1 and 2. Why should we assume
different production functions for different periods? The idea is that we want to
introduce some uctuations so that there are booms and recessions. So the rm
makes a lot during booms (or at period 1) but less during recessions (or at period
2).
1. No Firing Restriction: If there is no ring restriction, and if the rm
maximizes the discounted value of pro ts, the objective function becomes
p
(4 L1 1
|
{z
p
1
L1 ) +
(2 L2 1
}
{z
1+r |
pro t at period 1
L2 ):
}
pro t at period 2
The rst-order conditions imply that
L1 = 4; L2 = 1:
So four workers are employed at period 1, but only one worker is employed at
period 2, so three workers are red at the end of period 1.
2. Firing Restriction: Now introduce a ring restriction that no one can be
red (i.e., L1 = L2 ) between the two periods. The rm maximizes
p
(4| L {z1
p
1
L}) +
(2| L {z1
1+r
pro t at period 1
L}):
pro t at period 2
To make the algebra simple, let's assume r = 0. (r is close to 0 anyway.) The

rst-order condition is
3
p
2 = 0;
L
or equivalently,
9
L= :
4
3. Conclusion: Workers are red when the business is bad, but everyone
wants to avoid the risk of getting red. The ring restrictions help them to have
more secure jobs. But under these restrictions, recognizing the costs of ring,
rms will hire less workers from the beginning. That is, the government has made
the labor market less exible, the bene t is that the employed become happier,
but the cost is that we may have more unemployed workers. In the previous
example, the rm without any ring restrictions hired 4 workers and 1 worker at
periods 1 and 2. In total, it hired 5 workers cumulatively. If ring is prohibited,
the rm only hires 9/4 workers at each period. So, cumulatively it only hired 9/2
workers, which is less than 5. Again, this economic policy of ring restrictions
have good and bad effects at the same time.
1.2.4
Final Comments
We all wish to make workers happier. All policies or institutions we considered

in this chapter are designed to help the workers. But as economists, you should
also know about the side effects. If the policy is too strong, the side effects
might dominate, and the workers may be eventually hurt. The unemployment
insurance is desirable to help the unemployed, but then they may delay the job
10
acceptance, which may further increase the unemployment. By the minimum

wage policy, low-skilled workers can get more money for their lives, but it might
discourage rms from hiring low-skilled workers. Labor unions help the hired
workers to have what they deserve, but they may decrease the employment (and
hence output) and may also affect the wage rates of the rms in other industries.
Firing restrictions make the jobs more secure, but rms may hire less workers
from the beginning because ring is costly. All economic policies have costs and
bene ts No exceptions!
True or False?
1. Labor unions help their member workers to get what they deserve. Hence,
the government needs to subsidize any activities of labor unions.
2. The United States labor market is too exible in the sense that the rms
can re workers easily compared to European countries. If the United States
government makes ring more costly (for example, by collecting ring taxes
from rms), the rms may be hurt, but de nitely the workers will always bene t.
3. Assume that there are two industries which produce goods by hiring workers, where all workers are identical and may work in either industry. Further,
in the market equilibrium, the wage per worker is $1 (of course, the same in
both industries). If the labor union in one industry arti cially increases the wage
per worker to $1.5 in that industry only, then the other industry will hire more
workers with a lower level of equilibrium wage per worker.
11
Exercises
1. (Unemployment Insurance) Consider a worker who maximizes the expected discounted value of consumptions. (This is equivalent to assuming u(C) =
P1 t
C . At period 0, she maximizes E
Ct for 0 < < 1.) While unemt=0
ployed, a worker meets one employer every period, who offers either a high wage
wh (with probability p) or a low wage wl (with probability 1 p). Looking at
the wage offer, the worker may accept or reject it. If she accepts, she receives
the speci ed wage at each period, starting from this period. There is no risk of
getting red, and it is not allowed to search for a new job once employed.
The unemployment bene t is provided only for one period. To be speci c, if
a worker starts a job search (say, at period 0) and rejects the rst wage offer, the
unemployment bene t b (where b < wl < wh ) is paid immediately. But if she
remains unemployed and rejects other offers in later periods (at periods 1, 2, ...),
no bene ts are provided.
(a) Will an unemployed worker accept a high-wage offer? Why or why not?
(b) Consider an unemployed worker who previously received the unemployment bene t. Show that the value (i.e., expected discounted value of consumptions) of this worker when a low wage is offered and is rejected by her is
Vreject =
pwh =(1
1
(1
)
:
p)
(The worker is consistent with her decision. That is, if she rejects a low wage
offer once, then she does so under the same situation in the future.) [Hint:
P1 t
= 1=(1
) for 0 < < 1.]
t=0
(c) Assume = 0:9, p = 0:2, wh = 2, and wl = 1:3. Will this worker

accept or reject a low-wage offer?
(d) Now consider an unemployed worker who starts the job search. (So
she is eligible for the unemployment bene t if she rejects an offer at the initial
period.) Show that the value of this worker when a low wage is offered and is
rejected by her is
Wreject = b + pwh =(1
) + (1
p)Vreject :
(e) Assume b = 1. Under the same parameter values, will this worker accept
or reject a low-wage offer?
12

(f) Summarize your ndings in the previous two questions in plain words.
(g) Suppose the government decreases the unemployment bene t to b = 0:1.

Will this worker (at the initial period of her job search) accept or reject a lowwage offer? What is the implication of this result?
(h) (OPTIONAL) Continue to assume b = 0:1. Suppose the government now
decides to increase the duration of unemployment bene t from one period to two
periods. Consider an unemployed worker at the initial period (period 0) of the job
search. Will she accept a low-wage offer? What about in the next period (period
1)? What about future periods (periods 2, 3, ...)? Summarize what happens when
the government increases the duration from one period to two periods.
2. (Unemployment Insurance) In our discussions on economic effects of
unemployment bene ts, we considered the following set-up (except that u(C) =
C is assumed here):
Consider a worker who maximizes the expected discounted value of
consumptions. (This is equivalent to assuming u(C) = C .) For example,
at period 0, she maximizes
#
"1
X
t
E
Ct ; 0 < < 1:
t=0
While unemployed, a worker meets one employer every period, who offers
either a high wage wh (with probability p) or a low wage wl (with probability 1 p). Looking at the wage offer, the worker may accept or reject.
If she accepts, she receives the speci ed wage at each period, starting from
this period. There is no risk of getting red, and it is not allowed to search
for a new job once employed. If she rejects, she receives the unemployment bene t of b (which is lower than wl ) in this period and continues to
meet an employer in the next. The unemployment bene t is paid forever,
as long as she is unemployed.
Continue to assume this set-up. In addition, assume that a low-wage job is not
secure, so an employed worker with wl (low wage) has a ring risk of probability
q at each period. To be speci c, at the end of each period (i.e., after the wage
is paid), an employed worker with wl is red with probability q . If red, she
becomes unemployed from the next period and meets a new employer making a
wage offer which the worker can accept or reject again (as in the set-up). With
probability 1 q , an employed worker with wl stays at a current position in the
13
next period. An employed worker with wh (high wage), on the other hand, keeps
the job forever without being red.
(a) Suppose that an unemployed worker always rejects the offer if the offered
wage is wl . Let V1 denote the value (i.e., expected discounted value of consumptions), at the beginning of each period (i.e., before wage or compensation is
paid), for an unemployed worker who has just rejected the wage offer. Show that
b + pwh =(1
)
:
1
(1 p)
P1
[You have to clearly explain why this is true. Hint:
t=0
0 < < 1.]
V1 =
= 1=(1
) for
(b) Suppose that an unemployed worker always accepts the offer if the offered wage is wl . Let V2 denote the value, at the beginning of each period, for
a worker who has just accepted the low-wage offer. Show that
V2 =
wl + pqwh =(1
)
:
1
q(1 p)
(1 q)
[Hint: She will receive wl in the current period. In the next period, it is uncertain
whether she will continue to be employed.]
(c) When does an unemployed worker accept a low-wage offer: V1 > V2 or
V1 < V2 ? As b increases, is she more likely to accept one? As q increases, is she
more likely to accept one?
(d) Why does an increase in a ring risk for low-wage jobs (q ) affect the
decision of an unemployed worker in accepting a low-wage offer in the way you
describe in the above question? Provide clear intuitions.
3. (Unemployment Insurance) In our discussions on economic effects of
unemployment bene ts, we considered the following set-up (except that u(C) =
C is assumed here):
Consider a worker who maximizes the expected discounted value of
consumptions. (This is equivalent to assuming u(C) = C .) For example,
at period 0, she maximizes
"1
#
X
t
E
Ct ; 0 < < 1:
t=0
14

While unemployed, a worker meets one employer every period, who offers
either a high wage wh (with probability p) or a low wage wl (with probability 1 p). Looking at the wage offer, the worker may accept or reject.
If she accepts, she receives the speci ed wage at each period, starting from
this period. There is no risk of getting red, and it is not allowed to search
for a new job once employed. If she rejects, she receives the unemployment bene t of b (which is lower than wl ) in this period and continues to
meet an employer in the next. The unemployment bene t is paid forever,
as long as she is unemployed.
Let us make the following changes:
(1) An unemployed worker meets two employers each period. (But of course,
the worker can accept only one if she chooses to accept.) Each employer still
offers wh and wl with probabilities p and 1 p, respectively. (If both employers
offer wh , then the worker will accept either one. If one employer offers wh and
the other wl , the worker will accept wh . If both employer offers wl , then the
worker will either accept one of them or reject both.)
(2) An employed worker with wl (low wage) can also search for a job. (Notice
that a worker with wh does not need to search because she already has the best
offer.) In particular, she meets only one employer at the beginning of every
period (i.e., before she works and receives the wage). An employer offers wh
and wl with probabilities p and 1 p, respectively. If wh is offered, then she will
de nitely accept a new job immediately, receiving wh from the current period.
Otherwise assume she stays at the current position.
(a) Suppose that an unemployed worker always accepts one offer if both
of offered wages are wl . Let V1 denote the value (i.e., expected discounted
value of consumptions), at the beginning of each period (i.e., before wage or
compensation is paid), for an unemployed worker who has just accepted one of
the offers where both offers are wl . Show that
V1 =
wl + pwh =(1
)
:
1
(1 p)
P1 t
[You have to clearly explain why this is true. Hint:
= 1=(1
)
t=0
for 0 < < 1. She will receive wl in the current period. In the next period, it is
uncertain whether she will stay at a current position.]
(b) Suppose that an unemployed worker always rejects both offers if both of
offered wages are wl . Let V2 denote the value, at the beginning of each period
15
(i.e., before wage or compensation is paid), for an unemployed worker who has
just rejected both offers where both offers are wl . Show that
V2 =
b + (1
(1
1
p)2 )wh =(1

(1 p)2
(c) When does a worker with two wl offers accept either offer: V1 > V2 or
V1 < V2 ? As the unemployment bene t (b) increases, is an unemployed worker
more likely to stay longer as unemployed?
(d) Assume p = 0. Do we have V1 > V2 or V1 < V2 ? Explain this result
in plain words. What do you think will happen as p increases (other things being
equal)? In light of your answers, explain good and bad effects to herself
when a consumer chooses to accept one of offers, relative to when she rejects
both offers, under the condition that she has two wl offers.
4. (Minimum Wage) Consider a rm that hires both skilled and unskilled
workers. The production function is given by
p
Y (LS ; LU ) = 2LS + LU + 2 LS LU ;
where Y is the output in dollars, and LS and LU are the numbers of skilled
and unskilled workers hired. There are 75 skilled and 75 unskilled workers who
want to be hired regardless of wages by this rm. (Hence, labor is inelastically
supplied for this rm.) Denote the wages by wS for the skilled and wU for the
unskilled.
(a) Set up the pro t-maximization problem of the rm, given the wages (wS
and wU ). Obtain the rst-order conditions.
(b) The market-clearing condition implies that equilibrium wages are determined so that all workers who want to be hired are hired. What are the
equilibrium wages for skilled and unskilled workers, respectively?
(c) Suppose the government sets the minimum wage to 2.5. If this is higher
than equilibrium wage, the market clearing is broken and the rm may not hire
all workers who want to be hired. How many unskilled workers are now hired?
Are some unskilled workers unemployed?
(d) Now what is the market-clearing wage for skilled workers? Has it increased or decreased? Why does this happen? [Hint: See marginal productivity
of skilled workers.]
16
(e) Compare the outputs in terms of dollars in two cases (no minimum wage
vs. minimum wage). By introducing the minimum wage of 2.5, has the output
increased or decreased? Why?
5. (Immigration of The Unskilled) Consider an economy with skilled and
unskilled workers. Each individual rm has an identical production function:
Y (LS ; LU ) = (LS )2=3 (LU )1=3 ;

where Y is the output in units of consumption goods, and LS and LU are the
numbers of skilled and unskilled workers hired by an individual rm, respectively. (LS has a higher power, 2/3, which re ects that those workers are skilled.)
Denote wage rates by wS for a skilled worker and by wU for an unskilled.
(a) An individual rm chooses how many skilled and unskilled workers to
hire (LS and LU ), taking wage rates (wS and wU ) as given. Set up a pro xmaximization problem of an individual rm. Obtain the rst-order conditions.
(b) (Since the production function is constant-returns-to-scale, the number
of rms existing in this economy does not matter to obtain the total output of
the economy.) Assume there is (only) one representative rm in this economy
with the above production function. Labor is inelastically supplied for this
rm, and there are 125 skilled and 125 unskilled workers who want to be hired.
A market-clearing condition implies that equilibrium wage rates are determined
so that all workers (i.e., 125 workers in each labor market) are hired. (That is,
LS = 125 and LU = 125.) What are the equilibrium wage rates for skilled
and unskilled workers, respectively? What is the GDP (i.e., total output) of this
economy? GDP per capita (i.e., GDP divided by 250)?
(c) Suppose there are 91 unskilled immigrants. So now there are 125 + 91 =
216 unskilled workers in this economy. Obtain equilibrium wage rates for skilled
and unskilled workers, GDP, and GDP per capita. (Hint: 216 = 63 : Feel free to
use 25=108 0:231 and 150=341 0:440.)
(d) Describe your ndings in plain words so that your friend (whose major
is not economics) can understand. Your solution should contain explanation on
all of the following questions:
- Who gains from immigrants? Why do they gain?
- Who loses? Why do they lose?
- What happens to GDP and why?

- What happens to GDP per capita and why?
17
Chapter 2
Fiscal Policy (I)
This chapter addresses several issues in taxation. In particular, we discuss how
some types of taxes - lump-sum tax and capital income tax - affect the economy.
2.1
Lump-Sum Tax
Question: How Do Lump-Sum Taxes Affect an Economy?

Section 12.1 of DLS.)
(Reference:
How taxes affect production, consumption, investment, capital stock or government revenue is an important question in macroeconomics. We start our
discussion with lump-sum tax. This type of tax simply collects a xed amount
of money from a taxpayer, regardless of any individual characteristics such as
income or wealth. It does not take 10% of your income (like income tax) or 5%
of your wealth (like property tax). It takes, for example, $1,000 from you, no
matter who you are, how much you earn, or how rich you are.
Of course, this type of tax is somewhat unrealistic because no government
in reality collects the same amount of money from everyone. But it is simple to
analyze, so it often becomes a starting point of an economic analysis on taxation. But more importantly, there is another reason why this type of taxation is
important, which will be revealed after
2.1.1
Model Description
Our goal is to compare two economies: one without any taxation, and the other
with $G of lump-sum tax per worker. There are no other taxes. We study how
productions, consumptions, investments, capital stocks and government revenues
come to be different between these two economies. To do this, we use a general
equilibrium setup, which is familiar to you from earlier courses.
In our ctional economy, there is only one type of consumption good. (For
example, there are only apples.) There is a representative consumer (so that we
can imagine a country with only one person), who produces and consumes at the
18
2.1 Lump-Sum Tax
19
same time. She has the following preferences:

1
X
(2.1)
U (Ct );
t=0
where Ct is consumption at time t, U is a utility function (that is increasing,

concave, and twice-differentiable), and 0 <
< 1 is a time discount factor.
Notice that (2.1) is an extended version of a two-period setup, U (C0 ) + U (C1 ),
so that the representative consumer can live forever.
Now consider a production function. The representative consumer has an
access the following technology:
Yt = (Kt ) ; 0 <
< 1;
(2.2)
in which the capital stock Kt is the only source of production. Here, Yt and
Kt are output and capital stock at period t, both in units of consumption goods
(apples). Notice that the marginal product is decreasing, which is clear if you
differentiate (2.2) with respect to Kt . [Draw a brief gure for this function.] (2.2)
is standard when the capital is the only input in the production function.
The government spends Gt units at each period t (where Gt < Yt ). For now,
assume that this is constant over time, i.e., Gt = G. This amount of spending is
solely nanced by collecting lump-sum taxes from the representative consumer
in each period. The government does not produce anything.
The production of this economy (or, everything produced by the representative consumer) is consumed (Ct ), invested (It ), or paid to the government as
lump-sum taxes (G). So the representative consumer's resource constraint is
Ct + It + G = Yt :
(2.3)
We further assume that the capital stock of the economy evolves from the following (standard) law of motion:
Kt+1 = (1
)Kt + It :
(2.4)
This means the following: Capital stock at period t was Kt (in terms of consumption goods, or apples). The ratio of this capital stock (apples) vanishes
due to depreciation, so only (1
)Kt units (apples) remain until t + 1. Then,
the new investment It is added by the representative consumer. These account
for the new stock of capital, Kt+1 . Once K0 is given, the consumer decides I0 ,
20
Chapter 2 Fiscal Policy (I)
then K1 is determined, then the consumer again decides I1 , and so on.

This completes the model description. At each period, the representative
consumer decides how much to consume (Ct ) and how much to invest (It ), given
the capital stock (Kt ) and the lump-sum tax (G). That is, the representative
consumer maximizes (2.1) subject to all other constraints, (2.2), (2.3) and (2.4).
We have one function to maximize, and three constraints binding it.
2.1.2
Solution and Discussion
An easy way to solve this problem is to eliminate the constraints as many as we

can. (2.2) and (2.3) yields
Ct + It + G = (Kt ) ;
(which eliminates Yt ,) and again with (2.4), it becomes
Ct + Kt+1
(1
)Kt + G = (Kt ) ;
(which eliminates It ,) or equivalently,
Ct = (1
)Kt + (Kt )
Kt+1
(2.5)
G:
We can even make this problem simpler by eliminating Ct and transforming it

to an unconstrained problem. That is, inserting (2.5) into (2.1), our problem
becomes
max1
fKt+1 gt=0
1
X
U ((1
)Kt + (Kt )
Kt+1
G):
(2.6)
t=0
Now all we have to do is to obtain the rst-order conditions.

We must be careful in doing this because there are both Kt and Kt+1 in one
parenthesis. That is, (2.6) is extended to
max fU ((1
K1 ;K2 ;:::
t
::: +
+
t+1
)K0 + (K0 )
U ((1
)Kt + (Kt )
U ((1
)Kt+1 + (Kt+1 )
K1
G)
Kt+1
G)
Kt+2
G) + :::
There are a lot of rst-order conditions because there are so many choice variables: K1 , K2 , ... But those conditions turn out to be in the same form anyway,
so let's obtain the one with respect to Kt+1 . Notice that there are two terms with
2.1 Lump-Sum Tax
21
Kt+1 . The rst-order condition is

t
U 0 (Ct )[ 1] +
t+1
U 0 (Ct+1 )[1
+ (Kt+1 )
] = 0; for all t = 0; 1; :::;
or equivalently,
U 0 (Ct ) = U 0 (Ct+1 )[1
+ (Kt+1 )
]; for all t = 0; 1; :::
(2.7)
Inserting t = 0; 1; ::: provides all the rst order conditions. This equation determines how fKt+1 g evolves over time.
It is not easy and beyond our scope to investigate (2.7) in detail. But the
analysis is simple if we look at the steady state only. It turns out that in the long
run, we have a steady state in which the capital stock stays at a constant level
KSS forever. (Proving this argument is dif cult. At this stage, simply assume it.)
From (2.5), we know that the consumption also stays at a constant level CSS . So,
in a steady state, (2.7) becomes
U 0 (CSS ) = U 0 (CSS )[1
+ (KSS )
];
or,
KSS =
1= +
1
1
(2.8)
This is the long run level of capital and we are now ready to examine how the
government spending G affects this level.
1. Capital Stock (KSS ): As you can see from (2.8), G does not affect the
steady-state level of capital. (We will see how other types of taxes might affect
this level later in this chapter.)
2. Investment (ISS ): From (2.4), we can easily see that
KSS = (1
)KSS + ISS ;
or equivalently,
ISS = KSS ;
so ISS does not depend on G, either.
3. Production (YSS ): The total output in the steady state becomes from
(2.2),
YSS = (KSS ) ;
(2.9)
22
and we know this is not affected by G because KSS is not affected by it. Now
think about this: The production is not affected by this lump-sum tax. But
the government is spending something. Then how can the economy nance
this? Output and investment stay at the same levels, and the only thing left is
consumption. Yes, the consumer decreases consumption to nance government
spending!
4. Consumption (CSS ): The consumption becomes from (2.5),
CSS =
KSS + (KSS )
G:
This means if we assume < 1, the private consumption decreases as the government spends more resources. This kind of effect a decrease in consumption
coming from an increase in government spending is called the crowding-out
effect.
Let us summarize our ndings. [Draw gures: (i) time vs. G, (ii) time vs.
KSS , (iii) time vs. ISS , (iv) time vs. YSS , and (v) time vs. CSS .] An important
feature of this lump-sum tax is that it is not distortionary in the sense that the
steady-state output stays at the same level, regardless of the size of lump-sum tax.
This property makes this type of tax interesting. This tax can be a benchmark case
for the analysis of other taxes to see how bad other taxes are. Lump-sum tax
is a starting point in the analysis of scal policy not only because it is simple but
also because it is not distortionary.
True or False?
1. Suppose the government nances its spending solely by lump-sum taxes.
Since this will take some resources from consumers, both the consumption and
investment of this economy in a steady state will decrease.
2. The crowding-out effect does not arise if the taxes are not distortionary,
i.e., if the government nances its spending by lump-sum taxes only.
2.2
Distortionary Tax: Capital Income Tax
Question: How Do Capital Income Taxes Affect an Economy? (Reference: Section 13.3 of DLS.)
Then why not the lump-sum tax? Ef ciency is not the only concern of economics. If Tom makes $10,000 and Albert makes only $100, then perhaps it is not
2.2 Distortionary Tax: Capital Income Tax
23
fair to force each of them to pay $100 for taxes. If someone earns more dollars,
we think he/she should pay more dollars. (Furthermore, in many countries, those
people pay a higher fraction (e.g., 40%) of their incomes than middle-income
workers do (e.g., 20%). Here, to make the analysis simple, we will assume that
everyone pays the same fraction of their incomes.) In this section, we analyze
the effects of (capital) income tax. If you have your money in bank accounts
or hold stocks in stock markets, you should pay some part of your income from
those sources to the government. At this point, you may guess that because your
capital income is now taxed, you want to hold less stock of capital. In that sense,
this type of tax is different from lump-sum tax. We now con rm this conjecture
in a clear way.
2.2.1
Model Description
The model is the same as before, but a new assumption here is that the government will take a fraction (say, 10%) of your capital income (and not a xed
amount of $G) as a tax. For convenience, we rewrite expressions (2.1), (2.2) and
(2.4). The preferences of a representative consumer are
1
X
(2.10)
U (Ct );
t=0
and private production and capital accumulation follows
Yt = (Kt ) ; 0 <
Kt+1 = (1
< 1;
)Kt + It :
(2.11)
(2.12)
Now the question is how to replace (2.3): Ct +It +G = Yt : In this equation, what
is produced in the economy (Yt ) is consumed (Ct ), invested (It ), or paid as lumpsum tax (G). But in our new setup with capital income tax, your entire income,
Yt , comes from your capital holding, i.e., all your income is the capital income.
The government collects the ratio of your (capital) income, or equivalently,
Yt :
Hence, (2.3) should be replaced by
Ct + It + Yt = Yt ;
24
or equivalently,
Ct + It = (1
(2.13)
)Yt ;
which becomes a new resource constraint. This completes a model description.

2.2.2
How to solve this model is similar to our previous setup. (2.11) and (2.13) yield
Ct + It = (1
(2.14)
)Kt ;
and (2.12) implies
Ct + Kt+1
(1
)Kt = (1
)Kt ;
or equivalently,
Ct = (1
)Kt + (1
)Kt
(2.15)
Kt+1 :
We can make the problem simpler by inserting (2.15) into (2.10): Given K0 , the
consumer chooses fKt+1 g for all t = 0; 1; 2; ::: to maximize
1
X
U ((1
)Kt + (1
)Kt
Kt+1 ):
t=0
The rst-order condition with respect to Kt+1 is

t
U 0 (Ct )[ 1] +
t+1
U 0 (Ct+1 )[1
+ (1
)(Kt+1 )
] = 0;
or equivalently,
U 0 (Ct ) = U 0 (Ct+1 )[1
+ (1
)(Kt+1 )
]:
(2.16)
This equation determines how fKt+1 g (and other variables) evolves over time.
As before, let us focus on a steady state. In a steady state, the capital stock
stays at a constant level KSS , and from (2.15), it turns out that the consumption
also stays at a constant level Css . In this steady state, (2.16) implies
U 0 (CSS ) = U 0 (CSS )[1
+ (1
)(KSS )
];
or equivalently,
KSS ( ) =
1= +
(1
1
)
1
1
(2.17)
25
On the left-hand side of this equation, I wrote KSS as a function of

size that KSS may depend on .
to empha-
1. Capital Stock (KSS ): As increases (for example, from 0% to 10%),

the nominator in the parenthesis of (2.17), (1
), decreases. So the entire
1= + 1
part in parenthesis, (1 ) , increases (because from since 0 <
< 1, i.e.,
1
1=
1 > 0, we have 1= +
1 > 0). Hence, as 1 < 0, the right-hand side
of (2.17) decreases. This result implies that as the government increases the tax
rate on capital income, the steady-state level of capital stock decreases. Recall
that the lump-sum taxation does not affect the capital stock. In this sense, we
often say the capital income tax is distortionary.
2. Investment (ISS ): Now the investment. From (2.12), we have
KSS ( ) = (1
)KSS ( ) + ISS ( );
or equivalently,
ISS ( ) = KSS ( ):
We know as increases, KSS decreases, so of course ISS also decreases. That
is, a steady-state investment level decreases as government raises the tax rate.
3. Production (YSS ): For production, (2.11) implies
(2.18)
YSS ( ) = KSS ( ) ;
so it is clear that the (private) production YSS also decreases as
increases.
4. Government Revenue (RSS ): Another interesting question is how the

government revenue is affected. Should we expect that the government revenue
would increase whenever the government increases the tax rate? The answer is
no and here is why. The government gets ratio of total production, so the
steady-state government revenue, RSS ( ), becomes from (2.18),
RSS ( ) =
YSS ( )
KSS ( )
1= +
(1
=
=
(1
)
1= +
)1
|
{z
constant
(2.19)
26
Hence, RSS ( ) may increase or decrease as increases. A clear way to see

whether it increases or decreases is to differentiate (2.19) with respect to and
see the sign of it. A differentiation yields
0
RSS
( ) =
(1
)1
(1
(1
1
1
(1
)1
(1
)1
)1
(constant)
(constant)
(constant)
where (constant) is positive (since 0 < < 1, i.e., 1=

1 > 0) and (1
1
0
is also positive (since 0 <
< 1). Then, it is clear that RSS
( ) is
)1
positive if
1
1
> 0;
or equivalently,
<1
(2.20)
0
Similarly, RSS
( ) is negative if > 1
. So as the government increases the
tax rate between 0 and 1
, its revenue increases. But if it increases it from
1
to a higher level, (i.e., if the tax rate becomes too high,) the government
revenue actually decreases. [Draw a graph.] The interpretation is as follows.
The government may get more revenue if it increases the tax rate because the
consumer will pay a higher share of her income. But this tax will distort the
decision of consumer, who will lower the private production, and it may be the
case that this decrease in production is so big that the government revenue (which
is a fraction of the private production) actually decreases. The curve that shows
how the government revenue reacts as the government increases the tax rate is
called a Laffer curve. Needless to say, too high level of taxation (for example, a
tax rate higher than 1
in this model) should be avoided because it will distort
the economy while the government will suffer from a loss of revenue at the same
time.
5. Consumption (CSS ): Finally, let us investigate how the steady-state consumption is affected by capital income tax. (2.15) implies
CSS ( ) = (1
)KSS ( ) + (1
)KSS ( )
KSS ( );
(2.21)
27
or equivalently,
)KSS ( )
CSS ( ) = (1
KSS ( ):
Plugging (2.17),
CSS ( ) = (1
1= +
(1
)
)
= (1
)1
)
1
(constant)
If you increase , the rst term (1
1= +
(1
1= +
1+ 1
= (1
(constant)
(1
1
1
1
1
1= +
) 1 hde nitely decreases. But it is not easy

i
1
to determine the sign of the second term, (constant) 1
(constant) 1 . It
depends on parameter values. So, depending on the sign of the second term,
CSS ( ) can either increase or decrease. Then why might consumption increase
or decrease? The answer lies in (2.21), or more fundamentally, (2.14). (2.14) can
be written in terms of steady-state variables,
CSS ( ) + ISS ( ) = (1
)KSS ( ) :
The key point is that an increase in has two different effects. First, it decreases
the right-hand side, (1 )KSS ( ) (since both (1 ) and KSS ( ) decreases),
which means now the consumer has less after-tax income, which will perhaps
decrease her consumption level. Second, on the other hand, it also decreases
the investment, ISS ( ), so the consumer want to consume rather than invest,
which will increase her consumption level. Which one dominates is an empirical
question.
In summary, capital income tax is distortionary. Unlike lump-sum tax, it
affects all macroeconomic variables: capital stock, investment, production and
consumption. [Draw gures. vs. variables: (i) vs. KSS , (ii) vs. ISS , (iii)
vs. YSS , (iv) vs. CSS .]
2.2.3
Challenges
We have analyzed the effects of lump-sum tax and capital income tax. You are
now equipped with theoretical tools to analyze other types of taxes. All you have
to do is to change the budget constraint, (2.3) or (2.13). For example, suppose the
1
1
28
government wants to tax the capital income, but it does not want to discourage
the investment. So it decides not to tax the portion of capital income that is used
for investment. The budget constraint in this case will be
Ct + It +
1 (Yt
It ) = Yt :
Examine how this type of tax affects macroeconomic variables for your exercise.
To take another example, suppose the government has decided to tax the
consumption only. You still have a full amount of capital income, but if you
want to spend $1 for your consumption, you should pay $ 2 to the government.
The budget constraint in this case becomes
Ct + It +
2 Ct
= Yt :
You can similarly analyze how this tax will affect the economy (or you can show
why this budget constraint is, in fact, the same as the previous one).
True or False?
1. Increasing the tax rate for capital income will serve for social justice in
a sense that richer people pay more taxes in terms of dollars, without any costs
that the economy should bear.
2. Suppose the government has nanced its spending solely by collecting
capital income taxes for a long time, and that the government revenue at each
period has been $100. If the government stops collecting capital income taxes
and starts collecting lump-sum taxes of the same amount ($100), then the investment, capital stock, and output in a steady state will increase.
3. Suppose the government nances all its spending by capital income taxes.
That is, the consumers need to pay a at rate of their capital income to the
government. As the government increases , the government revenue in a steady
state also increases because consumers need to pay more taxes.
29
Exercises
1. (Consumption Tax) Suppose the government nances its spending solely
by the consumption tax with a xed rate of . That is, if you consume $1, you
should pay $ to the government. No tax is levied on the investment. Your job is
to consider the effect of increasing this tax ratio (for example, from 10% to 20%)
on macroeconomic variables in a steady state.
Let us use the model described in the classes. The preferences of the representative consumer are to maximize
1
X
U (Ct );
t=0
where Ct is consumption at time t. Production (Yt ) and capital accumulation (Kt )

follows
Y t = Kt ; 0 <
Kt+1 = (1
< 1;
)Kt + It ;
where It is the investment at time t. The government does not produce anything.
The resource constraint is
Ct + It + Ct = Yt ;
which means you consume (Ct ), invest (It ), and pay taxes ( Ct ), from what you
have earned (Yt ).
(a) Set up the problem of representative consumer.
(b) Derive the rst-order condition(s).
(c) Assume the economy is in a steady state. That is, assume the capital stock
stays at KSS ( ) forever, depending on the tax rate, . How does the capital stock
in a steady state (KSS ( )) change as the government increases ?
(d) What about the investment in a steady state (ISS ( ))?
(e) What about the output in a steady state (YSS ( ))?
(f) What about the consumption in a steady state (CSS ( ))? Interpret your
answer in a plain English.
(g) What about the government revenue in a steady state ( CSS ( ))?
30
2. (Capital Holding Tax) Suppose the government nances its spending

solely by this capital holding tax with a xed rate of . That is, if you hold
$1 value of capital stock, you should pay $ to the government at each period.
The representative consumer maximizes
1
X
U (Ct );
t=0
where Ct is consumption (in terms of dollars) at time t. Production (Yt ) and stock
of capital (Kt ) (both in terms of dollars) follow
Yt = (Kt ) ; 0 <
Kt+1 = (1
< 1;
)Kt + It ; 0 <
< 1;
where It is the investment (in terms of dollars) at time t. ( is the capital depreciation ratio.) The resource constraint of the representative consumer is
Ct + It + Kt = Yt ;
which means the consumer consumes (Ct ), invests (It ), and pays taxes ( Kt ),
from what she earns (Yt ). Assume that the government does not produce anything
with tax revenues.
(a) Set up the problem of representative consumer and derive the rst-order
condition(s).
(b) Assume the economy is in a steady state. That is, assume the capital
stock stays at KSS ( ) forever, which may depend on the tax rate . Find an
explicit expression for KSS ( ). Does the capital stock in a steady state (KSS ( ))
increase or decrease as the government increases ?
(c) Does the output in a steady state (YSS ( )) increase or decrease as the
government increases ? In one of the problem set questions, you solved a similar
problem regarding the consumption tax. (If you consume $1, you should pay $t
to the government.) If you are a policy maker, which tax scheme would you prefer
between this capital holding tax and the consumption tax? Justify your answer.
(d) Suppose the government still nances its spending solely by this capital
holding tax and needs to increase the tax revenue per period. So it has decided
to double the tax rate (for example, from 15% to 30%). Will it increase the
tax revenue? Justify your answer. [You may defend your answer either by
mathematics or by intuitions. In either case, you should clearly show why your
31
Laffer curve should look like yours.]

3. (Labor Income Tax) Consider a representative consumer with a utility
function over consumption C (in terms of dollars) and hours worked L (in terms
of hours):
p
U (C; L) = 2 C L:
The wage rate per hour is w, so the labor income is
Lw
in terms of dollars. However, the government collects a ratio
income, so the consumer only gets
(1
of this labor
)Lw:
(a) Set up the problem of this consumer. (This is a one-period problem. The
consumer chooses C and L and takes w as given.)
(b) Derive the rst-order condition(s). How many hours does this consumer
work? If the government increases the tax rate , does this consumer work more
or less?
(c) Obtain the government revenue. Does the government revenue increase
as the government raise the tax rate? Explain.
Chapter 3
Fiscal Policy (II)
3.1
Intertemporal Budget Constraint, Government Debt and

Lump-Sum Tax (Ricardian Equivalence)
Question: In an Economy with Lump-Sum Taxes only, How Does an Introduction of Government Debt Affect the Consumption Level? (References:
Sections 14.1 and 14.2 of DLS.)
There are three main sources to nance government spendings:
(1) Collect taxes. We analyzed how taxes affect an economy in the last
chapter.
(2) Issue government debts. (Borrow at a nancial market.) How government debt affects an economy is of our main interests in this chapter.
(3) Print money. This will be discussed in a later chapter.
Two main results derived in the last chapter are that lump-sum taxes are not
distortionary and that capital income taxes (and most of other taxes in reality)
are. In this section, we model an economy with some combination of lump-sum
taxes and government debts. (We replace lump-sum taxes by distortionary taxes
in the next section.) The goal is to see how the size of government debt affects
our ctional economy. When the government borrows instead of taxing, will
consumers be happy because of lower taxes?
3.1.1
Model Description
Into our model we introduce a perfect nancial market available both to a representative consumer and to the government. This nancial market is perfect in
a sense that anyone can borrow or lend, (i) up to the amount he/she wishes, and
(ii) at a xed interest rate r. So you are free to borrow $100 in this period, but
of course, you should pay $100(1 + r) back in the next period. You can lend (or
invest) $500, then you will receive $500(1 + r) back in the next period. There
is no technology of storing goods, so the only way to store something is through
this nancial market.
Since we will assume that there are only lump-sum taxes, the outputs of this
32
3.1 Intertemporal Budget Constraint, Government Debt and Lump-Sum Tax (Ricardian
33
Equivalence)
economy are not affected by these taxes anyway. This allows us to forget about a
production procedure and to simply assume that a xed amount of consumption
goods are given. That is, our economy in this model is an endowment economy,
in which the representative consumer simply gets her income from the heaven
without working (i.e., providing labor) or accumulating capital. Thanks to this
assumption, we are able to concentrate on the role of perfect nancial market.
Now we describe the model in detail. Assume there are only two periods, 0
and 1. The representative consumer is endowed with Y0 and Y1 units of consumption goods at periods 0 and 1, respectively. These values are known in advance
to her. By borrowing and lending, she maximizes her discounted utility:
U (C0 ) + U (C1 ); 0 <
< 1;
(3.1)
where U is a utility function that is increasing, concave, and twice-differentiable,

and C0 and C1 denote the units of consumption goods consumed at periods 0 and
1.
1. No Government: First, suppose that there is no government. The consumer maximizes (3.1) with respect to the constraints:
C0 = Y0 + b0 ;
(3.2)
C1 + b0 (1 + r) = Y1 ;
(3.3)
and
where b0 is the number of units of consumption goods borrowed in the nancial

market (i.e., b0 is the size of consumer's one-period debt). (3.2) means that
at period 0, the consumer consumes (C0 ) what she is endowed with (Y0 ) plus
what she borrows at the nancial market (b0 ). (3.3) means that at period 1, the
consumer consumes (C1 ) and pays back her debt with interests (b0 (1 + r)) from
what she is endowed with (Y1 ). Notice that there are only two periods, so if you
borrow at period 0, you should pay this back at period 1. Since period 1 is the end
of the world, the consumer cannot borrow at period 1. Here, b0 can be negative:
If b0 = $5 (positive), that means the consumer borrows $5 today, so she should
pay $5(1 + r) back tomorrow. If b0 = $3 (negative), that means the consumer
today borrows $3, that is, lends $3, so she will receive $3(1 + r) back
tomorrow.
Given Y0 and Y1 , the consumer chooses C0 , C1 and b0 , in order to maximize
(3.1), subject to (3.2) and (3.3). We can eliminate one choice variable, b0 , and
34
Chapter 3 Fiscal Policy (II)
then the problem becomes to maximize (3.1) subject to
C0 = Y0 +
Y1 C1
;
1+r
or equivalently,
C0 +
C1
Y1
= Y0 +
:
1+r
1+r
(3.4)
This constraint is worth discussing in detail. The left-hand side of this equaC1
tion, C0 + 1+r
, is a present value of consumptions at periods 0 and 1. Similarly,
Y1
the right-hand side, Y0 + 1+r
, is a present value of endowments at periods 0 and 1.
So (3.4) implies that the present value of consumptions equals the present value
of endowments (i.e., incomes). This equation is often called an intertemporal
budget constraint. This constraint shows the role of a nancial market. Without
this market, the consumer's decision should satisfy
C0 = Y0 ;
C1 = Y1 :
That is, she should consume what she is endowed with at each period because
there is no way to borrow or lend (or store). Now, the consumer's problem
becomes somewhat exible in a sense that these two budget constraints, saying
that consumption equals endowment at each period, are replaced by only one intertemporal budget constraint, (3.4), saying that a present value of consumptions
equals a present value of endowments. Of course, this difference is made by an
introduction of the nancial market.
The representative consumer maximizes (3.1) with respect to (3.4). The problem becomes an unconstrained problem as C1 is eliminated. That is,
max U (C0 ) + U ((1 + r)Y0 + Y1

C0
(1 + r)C0 ):
The rst-order condition is
U 0 (C0 ) + U 0 (C1 )[ (1 + r)] = 0;

or equivalently,
U 0 (C0 ) = (1 + r)U 0 (C1 ):

Many economists assume
1
1+r
(i.e., a discount factor to the consumer is
35
Equivalence)
the same as a discount factor in the nancial market) in this type of setup. Then,
this result reduces to
U 0 (C0 ) = U 0 (C1 );
hence (since U is strictly concave),
C0 = C1 :
This means that the consumer wants to smooth the consumption over time
although endowments may be different between periods. This consumption
smoothing turns out to be important in many types of economic analyses.
2. Government Spendings Financed Solely by Taxes: Now we have a
public sector, so the government needs to spend G0 and G1 at periods 0 and
1. (The government does not produce anything from these.) G0 and G1 are
pre-determined constants. The government collects lump-sum taxes to nance
them: The consumer are forced to pay T0 and T1 as lump-sum taxes to the
government, where T0 and T1 are chosen by the government, so the consumer's
budget constraint (3.4) should be replaced by
C0 +
C1
= Y0
1+r
T0 +
Y1 T1
:
1+r
(3.5)
Notice that (Y0 T0 ) and (Y1 T1 ) are after-tax endowments. Since the government spending is nanced solely by the tax at each period, we should have
G0 = T0 ;
G1 = T1 ;
so (3.5) is replaced by
C0 +
C1
= Y0
1+r
G0 +
Y1 G 1
:
1+r
(3.6)
The consumer maximizes (3.1) with respect to (3.5). This is almost identical to
the problem of no government case, and we can similarly get the consumption
smoothing result from this consumer's problem.
3. Government Spendings Financed by Taxes and Government Debts:
Now the government has an access to the perfect nancial market. Suppose the
government borrows B0 at the nancial market at period 0. (So B0 is the size of
36
government debt.) Then, at period 0, the government constraint becomes

(3.7)
G0 = T0 + B0 ;
which means that the government nances its spending (G0 ) by collecting taxes
(T0 ) and by issuing debt (B0 ). At period 1, it is
(3.8)
G1 + (1 + r)B0 = T1 ;
which means the government spends (G1 ) and pays back ((1 + r)B0 ) by collecting taxes (T1 ). So as long as these two constraints are satis ed, the government
is free to choose T0 , T1 and B0 .
But notice that we can eliminate one constraint. B0 is automatically determined by (3.7) once T0 is chosen. Eliminating B0 , (3.7) and (3.8) reduce to
G0 = T0 +
T1 G1
;
1+r
(3.9)
or equivalently,
G0 +
T1
G1
= T0 +
:
1+r
1+r
(3.10)
The government had to satisfy two conditions, G0 = T0 and G1 = T1 , in the

previous setup, so government spending should equal tax revenue at each period.
But now, (3.10), saying that the present value of government spendings equals
the present value of taxes collected, is enough.
Our goal is to see how this new environment affects the consumer's problem.
The consumer still maximizes the same objective function, (3.1), subject to one
constraint, (3.5). The levels of T0 and T1 are selected by the government, which
satisfy (3.10). But wait. Consumer's constraint (3.5) can be written as
C0 +
C1
Y1
= Y0 +
1+r
1+r
T0 +
T1
1+r
(3.11)
which is rewritten using (3.10) as,
C0 +
C1
Y1
= Y0 +
1+r
1+r
= Y0
G0 +
G0 +
Y1 G 1
;
1+r
G1
1+r
(3.12)
37
Equivalence)
and this is the same as (3.6)! This implies that the consumer's problem is exactly the same whether the government borrows or not. This result is called the
Ricardian equivalence (also known as the Barro-Ricardo equivalence).
3.1.2
Discussion
1. Implication: Finally, the consumer's problem when the government is allowed to borrow reduces to maximizing (3.1) with respect to (3.12). ((3.12) is
the same as (3.6).) There are no taxes (T0 and T1 ) or debts (B0 ) in this problem.
G1
Only G0 and G1 are there. And they are in the form of G0 + 1+r
. This means that
only the present value of government spendings is important in the consumer's
decision on consumption. How much to be taxed today or tomorrow does not
matter at all as long as the present value of government spendings remains constant.
What should we learn from this result? In consumer's budget constraint
(3.11), what is important to the consumer is the present value of taxes: If the
T1
government decreases T0 (today's tax) by increasing debt, then 1+r
(discounted
value of tomorrow's tax) is increased by the same amount. A decrease in tax
payment today made possible by an increase in government debt offsets by an
increase in tax payment tomorrow. So the consumer has no reason to feel happy
in the news of tax cuts nanced by debts.
To understand the importance of this implication, it is useful to review some
history of macroeconomics. The idea of Ricardian equivalence originally came
from David Ricardo in the 19th century. But for a long time, economists and
policy makers had not paid so much attention to this argument. After the inuence of John Keynes in 1930s, many Keynesian economists thought that the
government could stimulate the economy by spending more money. Suppose
the government has borrowed one million dollars from Tom. Tom is not hurt at
all because he still has a claim to one million dollars. But the government now
has new one million dollars out of nowhere, so consumers do not have to pay one
million dollar taxes today, which makes them one million dollars richer. Hence,
consumers will spend more money, which will increase the consumption of an
economy. The Ricardian equivalence tells you why this idea is not effective.
Those one million dollars did not come out of nowhere. The taxpayers will
eventually pay for this. Since they need to pay more taxes in the future, they
will try to save more today.
2. Limitation: This result is derived from several assumptions which in-
38
cludes lump-sum taxation. So in reality (full of distortionary taxes), we do not

really expect that this result would perfectly hold. That is, a change in government debt will affect the tax rate, which will change output and eventually (and
indirectly) consumption. But an important implication of the model can never be
underestimated. Yes, the decision of government regarding how much to borrow
will perhaps affect the consumer's decision to some extent in reality, but she is
not foolish enough to take those tax cuts as additional incomes.
3. Extension of the Model: Our model here is a two-period model. It is not
so dif cult to extend our current model to an in nite-period model. At time t, the
government constraint is
Gt + (1 + r)Bt
= Tt + Bt ; for all t = 0; 1; 2; :::;
where B 1 = 0 is given (i.e., there is no government debt that should be paid at

period 0). Recall that Bt is a one-period government debt. Since this implies
B0 = G0
T0
if t = 0 and
G1 + (1 + r)B0 = T1 + B1
if t = 1, we can eliminate B0 to have
G0 +
1
1
1
G1 = T0 +
T1 +
B1 :
1+r
1+r
1+r
We can continue to do this algebra to eliminate B1 , B2 , ..., BS

S
X
t=0
1
1+r
Gt =
S
X
t=0
1
1+r
Tt +
1
1+r
to have
BS :
(3.13)
Now suppose we are looking at a very long time horizon. That is, suppose S
is very very large (S ! 1). In this case, we usually assume that
lim
S!1
1
1+r
BS = 0;
(3.14)
which is often called the transversality condition. This eliminates the second
term in the right-hand side of (3.13). This assumption means that a sequence
of the government debt levels, fBt g, does not grow too fast (i.e., to be more
speci c, the debt level should grow slower than (1 + r)). So in plain words, we
39
Equivalence)
do not want the government to accumulate the debt forever to an extremely huge
amount. No one can nance its spending by borrowing more and more forever.
If we take (3.14), we can simply write (3.13) as
1
X
t=0
1
1+r
Gt =
1
X
t=0
1
1+r
Tt :
(3.15)
This, again, means that the present value of government spendings should be
equal to the present value of tax revenues. The representative consumer's preferences are to maximize
1
X
U (Ct );
t=0
subject to the budget constraint

1
X
t=0
1
1+r
Ct =
1
X
t=0
1
1+r
[Yt
Tt ];
(3.16)
implying that the present value of consumptions equals the present value of aftertax endowments. ((3.16) can be obtained similarly just as we have derived the
government's constraint.) But wait. (3.16) can be written using (3.15) as
1
X
t=0
1
1+r
Ct =
1
X
t=0
1
1+r
[Yt
Gt ]:
Notice that neither fTt g nor fBt g appears hear. Only the present value of government spendings is important for the consumer. She does not even look at whether
government spendings are nanced by taxes or government debts, and she will
not be fooled by tax cuts made possible by issuing more government debts.
True or False?
1. Suppose the Ricardian equivalence holds in an economy. If the government
wants to increase the private consumption this year, then it is a desirable scal
policy to tax less this year by borrowing more (i.e., by issuing government bonds),
so that the consumer has more resources to be used for consumption.
3.2
Government Debt and Distortionary Tax (Ramsey
40
Problem)
Question: Can the Government use Government Bonds to minimize the
losses caused by Distortionary Taxes? (Reference: Barro, Robert J. (1979),
On the Determination of the Public Debt, Journal of Political Economy,
87(5), 940-971.)
Higher rates of distortionary taxes are costly. But the government sometimes
should spend a lot for many reasons for example, to nance a war or to recover
from a disaster. If the government cannot borrow at all, then the tax rate should
be high for those periods, which will increase the deadweight losses (i.e., costs
arising from taxation) for those periods. Can we do something if the government
debts are introduced? In other words, while we know distortionary taxes are
costly, can we use the government debts to minimize these deadweight losses?
3.2.1
Let us go back to the endowment economy in the previous section. Suppose

the endowments (Y0 , Y1 , Y2 , ...) are pre-determined prior to period 0, and the
government is free to choose the amount of taxes to be collected at each period.
That is, the government can determine T0 , T1 , T2 , etc. We say that T0 =Y0 , T1 =Y1 ,
T2 =Y2 ... are the overall tax rates for given periods. Higher tax rates will be more
costly, so let a function,
f (Tt =Yt );
denote the fraction of income that the economy loses by deadweight losses. f is
assumed to be increasing (f 0 > 0) and convex (f 00 > 0). [Draw a gure.]
The convexity of this loss function can be justi ed by our results in the last
chapter. That is, (2.17) and (2.18) imply that
YSS ( ) = (1
)1
(constant):
Here, the deadweight losses are, by de nition, YSS (0) YSS ( ). This is the
difference between two outputs, with and without capital income taxes. As you
increase , you will easily see that the deadweight losses are increasing and
convex.
Since f (Tt =Yt ) is a fraction of deadweight losses out of income, the deadweight losses in units of consumption goods are
Yt f (Tt =Yt ):
3.2 Government Debt and Distortionary Tax (Ramsey Problem)
41
To summarize, the environment of the model is as follows. Suppose at time t, the

government collects taxes of Tt units of consumption goods (or dollars). Then,
the consumer, whose original endowment is Yt , loses Tt because this amount
is transferred to the government. At the same time, she also loses Yt f (Tt =Yt )
because of distortions caused by taxation. These losses simply disappear from
this economy.
Assume that the government minimizes the present value of these deadweight
losses:
1
X
1
1+r
t=0
Tt
Yt
Yt f
(3.17)
And the constraint is that the present value of government spendings should be
smaller than or equal to the present value of tax revenues. That is,
1
X
t=0
1
1+r
1
X
Gt
1
1+r
t=0
(3.18)
Tt :
(This constraint is adopted from (3.15).) So we have a constrained maximization

problem of choosing fTt g to minimize (3.17) with one constraint (3.18).
The Lagrangian function is
L=
1
X
t=0
where
is
1
1+r
Yt f
"
Tt
Yt
1
X
t=0
1
1+r
Tt
1
X
t=0
1
1+r
Gt ;
is a Lagrangian multiplier, so the rst-order condition with respect to Tt
1
1+r
Tt
Yt
f0
1
1+r
= 0;
or equivalently,
f0
Tt
Yt
= :
The convexity of f function guarantees that there is one solution of Tt =Yt satisfying this equation, which is constant over time.
In plain words, the sequence of optimal taxes fTt g satis es that (i) (3.18)
42
holds with equality, i.e.,

1
X
t=0
1
1+r
Gt =
1
X
1
1+r
t=0
Tt :
(3.19)
(If (3.18) holds with inequality, then taxes collected are too much relative to
required spendings. Why collect unnecessary taxes?) It also requires that (ii) the
overall tax rate (Tt =Yt ) is constant over time. That is, regardless of the sequence
of government spendings fGt g, the government should keep the overall tax rates
as constant. Recall that the consumer loves consumption smoothing. Similarly,
the government loves tax smoothing.
3.2.2
Discussion
To understand the implication of this model, let us consider several examples.

Example 1. Suppose that both the endowment and government spending are
constant over time, i.e., Yt = Y and Gt = 1 < Y for all t = 0; 1; 2; ::: [Draw
gures for fYt g and fGt g.] Since Tt =Yt should be constant, it is clear that Tt is
also constant, i.e., Tt = T . From (3.19), we have
Tt = T = 1;
so the government spending in each period is always fully nanced by taxation
at the same period. [Draw a gure for fTt g.] Hence, there is no government debt
at any periods, i.e., Bt = 0 for all t = 0; 1; 2; ::: [Draw a gure for fBt g.]
Example 2. Suppose that the endowment is constant over time, i.e., Yt = Y ,

and the government needs to nance an immediate temporary spending. Let's
say G0 = 1 and G1 = G2 = ::: = 0. [Draw gures for fYt g and fGt g.] Since
Tt =Yt should be constant, it is clear that Tt is also constant, i.e., Tt = T . From
(3.19), we have
1=T
Using
P1
t=0
1
X
t=0
= 1=(1
) for 0 <
1 = T
= T
1
1+r
< 1, we can rewrite this as

1
1
1 1+r
1+r
r
3.2 Government Debt and Distortionary Tax (Ramsey Problem)
43
or equivalently,
T =
r
1+r
So in order to nance a temporary spending, the government collects this amount

of constant tax at each period.
Then, we can obtain a sequence of the optimal government debts. At period
0, it is
B0 = G0
T0 = 1
r
1
=
:
1+r
1+r
This means the following. The government nances its temporary spending, $1,
by taxing $r=(1 + r) and borrowing $1=(1 + r) at period 0. Then what will
happen at period 1? We have
B1 = G1 + (1 + r)B0
r
= 0+1
1+r
1
=
1+r
T1
and we can show that B0 = B1 = B2 = ::: = 1=(1 + r). So the government

will continue to have a debt amount of $1=(1 + r) forever at all periods. (That
is, the government rolls over forever.) [Draw gures for fTt g and fBt g.] (You
can easily extend the result to consider a more general case: G0 = G1 = ::: =
Gn = G and Gn+1 = Gn+2 = ::: = g where G > g .)
Example 3. Continue to assume that the endowment is constant over time,
i.e., Yt = Y , but at this time, the government should nance a future spending.
Let's say the government does not spend anything except for period 2. So G2 = 1
and Gt = 0 for t = 0; 1; 3; 4; ::: [Draw gures for fYt g and fGt g.] Since Tt =Yt
should be constant, it is clear that Tt is also constant, i.e., Tt = T . From (3.19),
we have
1
1+r
= T
1
X
t=0
= T
1
1+r
1+r
r
44
or equivalently,
1
T =
1+r
r
=
(1 + r)3
1+r
r
Hence, we have
B0 = G0
T0 = 0
B1 = G1 + (1 + r)B0
B2 = G2 + (1 + r)B1
B3 = G3 + (1 + r)B2
r
=
(1 + r)3
r
;
(1 + r)3
r
r
r(r + 2)
T1 = 0
=
2
3
(1 + r)
(1 + r)
(1 + r)3
r
1
r(r + 2)
=
T2 = 1
2
3
(1 + r)
(1 + r)
(1 + r)3
r
1
1
=
T3 = 0 +
2
3
(1 + r)
(1 + r)
(1 + r)3
and it is easy to see B4 = B5 = ::: = 1=(1+r)3 . So the government accumulates

assets up to period 2 by collecting taxes that are constant over time. At period 2,
the government debt jumps up to a positive level (because of the spending), and
then it stays there forever. (You can easily extend the result to consider a more
general case: Gn = G and Gt = g for all t = 0; 1; :::; n 1; n + 1; n + 2:::;
where G > g .)
The Ramsey problem asks how to design a tax system to make the consumer
the happiest, for a given schedule of government spendings. In this section, we
have seen a nice example of this Ramsey problem: the optimal taxation problem
across different periods. There is another interesting example of this problem,
that asks how to tax different goods. According to the result, goods with higher
demand elasticities should be taxed less, and this result is often called the Ramsey rule.
True or False?
1. If the government needs to nance a temporary increase in government
spending because of the hurricane, this should be nanced only by a temporary
increase in tax revenue so that the level of government debt continues to stay at
a similar level.
3.3 Social Security
45
2. Just as any economic agents, the government needs to pay interests for
borrowing. Hence, the government should not issue government bonds (i.e.,
should not borrow in the nancial market and should nance its spending solely
by collecting taxes).
3.3
Social Security
Question: (i) When Do (and Don't) the Consumers Prefer a Pay-as-you-go

Pension System? (Reference: Section 12.3 of DLS.) (ii) Does the Redistribution
Policy Affect the Output of an Economy? (Reference: Section 13.4 of DLS.)
3.3.1
Model 1: Pension
The pension system is administrated by the government in many countries. Broadly

speaking, there are two pension schemes: (i) In a funded pension scheme, the
government taxes young workers and saves tax revenues, and later when they
retire, it transfers the money back to them. So this scheme is a forced savings
system. (ii) The U.S. is roughly on the other system, which is the unfunded or
pay-as-you-go pension scheme. Here, payments to retirees are collected from
current young workers.
A brief discussion on these two systems will help you to understand the
debate around the social security. Suppose the consumers live for two periods
only. Generation-t consumers are born in period t and dies in period t + 1, and
they work when they are young (in their rst period, which is t) to obtain income.
When they are old (in their second period, which is t + 1), they are not able to
work, so their income become zero. (Assume generation-(t 1) consumers
always exist. That is, there is nothing like the rst-ever generation like Adam
and Eve.)
The population of generation-t consumers, denoted by Nt , evolves as
Nt+1 = (1 + n)Nt ; 0 < n < 1;
(3.20)
so that n becomes a constant population growth rate. This means that at period
t, there are Nt 1 old agents and Nt = (1 + n)Nt 1 young agents.
Assume a perfect nancial market with a xed interest rate, r. Suppose each
young worker pays T for taxes (to be used for the pension system). In the funded
system, each old consumer receives what she paid ($T ) in the previous period
46
plus interests ($rT ). That is, she receives
T (1 + r):
But in the pay-as-you-go scheme, old consumers receive what young consumers
pay at the same period. The sum of payments of young agents in period t is
T Nt ;
which is distributed equally among the old, so each old consumer receives
T Nt
= T (1 + n)
Nt 1
from (3.20).
Notice that in either system, each young consumer pays $T . But each old
consumer receives $T (1 + r) in the funded system, while she receives $T (1 + n)
in the pay-as-you-go system. Hence, if the consumer is in the pay-as-you-go
scheme and if n < r, she will prefer her own private savings rather than the
pension system operated by the government. (If the government promises to
pay a return rate of r even in the pay-as-you-go system, of course the government
will run out of money.) In the United States, n is about 1% and r is believed to
be higher than 1%.
3.3.2
Model 2: Redistribution
Suppose the government collects taxes from someone and transfers these to someone else. Will it still affect the output of economy? While this is the main
question here, we will also think about labor income taxes. As a footnote, it
is useful to recall from microeconomics that a decrease in a wage rate (which is
equivalent to an increase in labor income tax rate in our discussion) may increase
or decrease the hours worked of an agent. We will see something similar.
Assume that there are two types of agents A and B and they have common
utility functions over consumption Ci (in dollars) and labor input Li (in hours
worked) for i = A; B :
p
U (Ci ; Li ) = 2 Ci Li :
(3.21)
Suppose each agent has a technology transforming labor effort into the production:
Yi = wi Li ;
3.3 Social Security
47
for i = A, B , where Yi is the output (thus the pre-tax income) and wi is her
wage rate (or labor productivity). Assume wA > wB so that agent A is more
productive.
For redistribution purposes, the government collects a labor income tax from
agent A with a constant rate , and gives this to agent B as a lump-sum transfer.
So two agents face budget constraints of the form:
CA = (1
)wA LA ;
CB = wB LB + F;
(3.22)
(3.23)
where is the at tax rate and F is the lump-sum transfer. Of course, the
government should balance revenue and expenditure, so
wA LA = F:
Agent A maximizes (3.21) subject to (3.22). That is, she maximizes with
respect to LA ,
p
2 (1
)wA LA LA :
or equivalently,
or equivalently,
(1
)wA
(1
)wA LA
(1
)wA =
LA ( ) = (1
1 = 0;
p
LA ;
)wA :
(3.24)
This result implies that as the government increases the tax rate , agent A will
decrease her labor supply in this model.
In fact, the result really depends on the utility function. For other types of
utility functions, the hours worked may even increase as the tax rate increases.
(See Section 13.4 of DLS.) Recall from microeconomics courses that a decrease
in (after-tax) wage rate (i.e., an increase in ) has two effects: (i) income effect:
Now you have less money, so you want to consume less leisure (if leisure is a
normal good). That is, you want to work more. (ii) substitution effect: Now
your leisure is less expensive, so you want to consume more leisure. That is, you
48
want to work less. If the former effect dominates, an increase in will increase
the hours worked. If the latter effect does, it will decrease the hours worked. In
this model, the latter dominates.
Similarly, agent B maximizes (3.21) subject to (3.23), or equivalently, she
maximizes with respect to LB ,
p
2 wB LB + F LB :
wB
wB LB + F
1 = 0;
or equivalently,
(wB )2 = wB LB + F;
or equivalently,
(wB )2 F
:
(3.25)
wB
This implies that agent B decreases her hours worked as the transfer F increases.
This is because the transfer has only an income effect: Now you have more
money, so you want to consume more leisure. Notice that the lump-sum transfer
does not alter any price for agent B, so there is no substitution effect.
So both agents A and B decreases hours worked as the government strengthens the redistribution policy. A few remarks on this result:
(1) This result is based on a speci c example of utility function. A more
general analysis can be found in Section 13.4 of DLS. Still, for a broad family
of utility functions, the section shows that the income of an economy (i.e., A's
income plus B's income) decreases as the government strengthens the redistribution system.
(2) Notice that (3.25) implies that the agent responds to lump-sum transfers.
Similarly we can show that the agent responds to lump-sum taxes. While we
argued that lump-sum taxes are not distortionary when the model has only the
capital formulation in the previous chapter, we see that lump-sum taxes do affect
the hours worked when labor input is introduced in the model. Still, there is a
good reason why the distortion of lump-sum taxes is not discussed very seriously.
If you solve the model, the lump-sum tax actually increases the labor supply (if
leisure is a normal good). So it does not decrease the output of economy anyway.
LB =
3.3 Social Security
49
True or False?
1. Given the current levels of population growth and interest rate in the
United States, an average worker will prefer being engaged in the pay-as-you-go
pension system rather than using his/her own savings accounts or other investment methods.
2. The redistribution policy will serve for our social justice in the sense that
we can help old or poor people while it will not affect the GDP level.
3. Consider a consumer who has a utility function over consumption C (in
dollars) and labor input L (in hours worked), wherep
she prefers leisure to work.
(For example, you may want to assume U (C; L) = 2 C L.) Also, assume this
consumer gets a xed amount of wage for each hour worked. (So if w denotes
a wage rate, she earns wL dollars.) If the government introduces a lump-sum
tax so that a xed amount of tax is collected from this consumer, then the income
effect implies that this consumer will increase the number of hours worked, while
the substitution effect implies that this consumer will decrease it, so the total
effect is uncertain.
4. In a paper published in 2004, Edward Prescott (2004 Nobel-prize winner)
argues that the total number of hours worked of an economy (e.g., the U.S. or a
European country) decreases by 0.8% as the wage rate decreases by 1%. This
implies that the substitution effect dominates the income effect. (Assume that
leisure is normal.)
50

Exercises
1. (Optimal Taxation) Consider an endowment economy in which the endowment (i.e., income) at period t = 0; 1; 2; ::: is given by Yt . The government should spend Gt at period t = 0; 1; 2; ::: The values for Yt and Gt for
all t = 0; 1; 2; ::: are pre-determined. As discussed in the class, suppose the
government wants to minimize the present value of deadweight losses subject
to the (intertemporal) budget constraint. The optimal tax revenue at each period
t = 0; 1; 2; :::, denoted by Tt , should satisfy the following two conditions:
(i)
1
X
t=0
1
1+r
Gt =
1
X
t=0
1
1+r
Tt ;
i.e., the (intertemporal) budget constraint of the government is satis ed, and,
(ii)
Tt
is constant;
Yt
i.e., the overall tax rate is constant over time for all t = 0; 1; 2; ::: Assume Yt =
Y , where Y is constant and greater than 2. (So the endowment is always the
same.) Also, assume there is no government debt that should be paid at time 0.
Suppose G0 = 2 and G1 = G2 = ::: = 1. That is, the government faces a
temporary increase in government spending at time 0 in which the government
spending is $2, but after time 0, the spending stays at $1 forever.
(a) How much should the government collect as taxes at each period under
P1
).)
the optimal taxation? (Hint: If 0 < < 1, then t=0 t = 1=(1
(b) Draw a gure that shows how the government debt evolves over time. (To
do this, you may mathematically obtain the levels of government debts for several
periods, or you may qualitatively defend your gure based on your answer to (a)
and other information in the question. The former method might be clearer.)
(c) After time 0, (i.e., at periods 1; 2; :::;) does the government collects more
taxes than it spends for government consumption? (That is, is Tt > Gt for
t = 1; 2; :::?) If so, what does the government do with the extra revenue (i.e.,
Tt Gt ) at each period?
Now suppose G0 = G2 = G4 = ::: = 2 and G1 = G3 = G5 = ::: = 1.
That is, the government should spend more at even periods than at odd periods.
3.3 Social Security
51
(a') Answer the same question above. (Use the same hint. That is all you
need for the algebra.)
(b') Answer the same question above.
[End of the question. There is no (c').]
2. (Ramsey Problem with Labor Income Tax) The goal of this question is to
derive an optimal scal policy when government spendings are nanced solely
by labor income taxes and government debts. There are only two periods, 0 and
1, (and period 1 is the end of the world). Consider a representative consumer
with a discounted utility function over consumptions and hours worked:
U (C0 ; L0 ) + U (C1 ; L1 );
where C0 and C1 are consumptions (in units of consumption goods) at periods 0
and 1, and L0 and L1 are hours worked (in hours) at periods 0 and 1. In particular,
assume a speci c form of U so that
p
U (Ct ; Lt ) = 2 Ct Lt ;
for t = 0 and 1. That is, the consumer maximizes

p
p
2 C0 L0 + (2 C1
L1 ):
(3.26)
This representative consumer's output (which becomes the output of this economy) follows at each period,
Yt = L t ;
for t = 0 and 1, where Yt is the output in units of consumption goods. However,
income tax rates are 0 and 1 at periods 0 and 1 (which may be different between
these two periods), so the government collects a fraction 0 or 1 of these outputs.
Hence the consumer only gets
(1
t )Lt ;
as disposable income at period t = 0 and 1. Throughout the question, the

consumer does not have an access to the nancial market. So this disposable
income becomes the consumption at each period in any of the problems below.
(a) Set up the problem of this consumer. Derive the rst-order conditions.
How many hours does this consumer work, given tax rates 0 and 1 ? (De-
52
note these optimal hours worked by L0 ( 0 ) and L1 ( 1 ), respectively.) Does the

worker work more or less as the government increase the tax rate in either period?
(b) Obtain the government revenue at each period. Draw a period-0 Laffer
curve (that relates period-0 tax rate 0 and period-0 tax revenue). Draw a period1 Laffer curve. Suppose the government needs to spend $0.25 (=$ 41 ) and $0
(nothing) at periods 0 and 1, respectively. The government cannot borrow, so
government spendings should be nanced solely by labor income taxes. What
should be the tax rates at periods 0 and 1?
(c) Continue to assume that the government needs to spend $0.25 (=$ 14 ) and
$0 (nothing) at periods 0 and 1, respectively. But now the government has an
access to the perfect nancial market, which offers a xed interest rate, r. The
government debt at period 0, denoted by B0 (in dollars), should be paid back at
period 1 with interests. Period 1 is the last period, so it is not allowed to borrow.
That is, the government should choose 0 , 1 and B0 to satisfy
1=4 =
0 + B0 (1 + r) =
0 L0 ( 0 )
+ B0 ;
1 L1 ( 1 ):
The consumer does not borrow or lend, and hence she acts as you described in
question (a). The (benevolent) government wants to choose a scal policy ( 0 , 1
and B0 ) that maximizes the discounted utility of consumer, which is expression
(3.26). (Of course, the government knows that consumer will act as you described
in question (a).) Set up the problem of the government. Go as far as you can to
characterize the optimal scal policy. (Your characterization should be enough
to answer the questions in (d). Feel free to assume = 1=(1 + r).)
(d) Describe your ndings in plain words so that your friend (whose major
is not economics) can understand. Your solution should contain explanation on
all of the following questions:
- To make the consumer the happiest, how should the government design the
tax system? Are the tax rates the same across the periods?
- Should the government borrow?
- Is there anything discussed in class that is consistent with your ndings?
What is it?
3. (Redistribution) In the class, we discussed the effect of introducing redistribution policy on the economy. In this question, you are asked to follow the
3.3 Social Security
53
same steps, but with a slightly different utility function. There are two agents,
A and B, who have a common utility function over consumption Ci (in terms of
dollars) and labor input Li (in terms of hours worked) for i = A; B :
U (Ci ; Li ) = log(Ci )
Li :
Suppose each agent has a technology transforming labor effort into the production:
Yi = wi Li ;
for i = A, B , where Yi is the output (thus pre-tax income) and wi is her wage rate
(or labor productivity). Assume wA > wB so that agent A is more productive.
For redistribution purposes, the government collects the labor income tax
from agent A (with a constant rate ) and gives this to agent B as a lump-sum
transfer. So two agents face the budget constraints of the form:
CA = (1
)wA LA ;
CB = wB LB + F;
where is the at tax rate and F is the lump-sum transfer. Of course, the
government should balance revenue and spending, so
wA LA = F:
(a) How does agent A's hours worked (LA ) react to the tax rate ? Explain
why you have such a result using the following two concepts: income effect and
substitution effect.
(b) How does agent B's hours worked (LB ) react to the lump-sum transfer
F ? Explain why you have such a result using the following two concepts: income
effect and substitution effect.
(c) Suppose the government has just introduced this redistribution policy. So
agent A now faces the tax rate (which used to be 0) and agent B now obtains
F = wA LA (which used to be 0) as transfer. What are the changes to agent
A's hours worked (LA ), pre-tax income (wA LA ), and after-tax income (that is
equivalent to consumption, CA )? What are the changes to agent B's hours worked
(LB ) and consumption (CB )? What is the change to the aggregate income of the
economy (that is equivalent to aggregate consumption, CA + CB )?
54
(d) What do you think are good and bad effects of introduction of this
redistribution policy on the economy?
Chapter 4
Monetary Policy (I)
This chapter provides a rst half of our discussions on monetary policy. First,
we extend our analysis on scal policy to introduce seigniorage and see how
seigniorage may cause in ation. Second, we provide a theoretical model on how
money growth affects the output of an economy.
4.1
Seigniorage and In ation
Question: Why May Issuing Debts and Printing Money to Finance Government De cits Bring Some Unusual Episodes of Hyperin ation? That is,
Why Are Balanced Budgets Important? (Reference: Chapter 18 of DLS.)
The previous two chapters discussed two ways to nance government spending (Gt ): (i) collecting taxes (Tt ) and (ii) issuing government debts (Bt , which
denotes a one-period debt). At time t, the government constraint used to be
Gt + (1 + r)Bt 1 = Tt + Bt ;
{z
}
|
(4.1)
to be nanced
for all t = 0; 1; 2; :::, where B 1 = 0 is given (i.e., there is no government debt

to be paid at period 0). The goal of this section is to introduce a third way: (iii)
printing money.
Since money is now introduced for the rst time in this course, we have
to clearly distinguish real terms from nominal terms. Assume there is only one
type of consumption goods (or apples). In (4.1), all variables, Gt , Bt and Tt ,
are in units of consumption goods, i.e., in real terms. Now, let Mt denote the
amount of money existing in this economy, in dollars, i.e., in nominal terms, at
the end of period t. This means that the central bank prints Mt Mt 1 dollars
during period t. If this amount of dollars are provided to the government, the
government can additionally buy
Mt
Mt
Pt
(4.2)
units of consumption goods, where Pt is the nominal price of one unit of consumption good at period t. The amount (4.2), which is the units of government
55
56
Chapter 4 Monetary Policy (I)
spending nanced by printing money, is called seigniorage. With seigniorage,

the government constraint (4.1) is replaced by a new equation:
M t Mt 1
Bt
+
;
(4.3)
Gt + (1 + r)Bt 1 = Tt +
|{z}
|{z}
{z
}
|
Pt
|
{z
}
by taxation
by borrowing
to be nanced
by printing money
for all t = 0; 1; 2; :::, where B 1 = 0 is given. Notice that the government

spending nanced by seigniorage ultimately comes from consumers. Consumers
eventually nance the seigniorage.
Printing money causes in ation (an increase in the overall price level), which
sometimes develops to hyperin ation. For example, in Germany of the early
1920s, the price level doubled every 49 hours (i.e., the in ation rate was 3 million
% per month). In Hungary after World War II, it doubled every 15 hours (4
quadrillion (1016 ) % per month). Hyperin ation is destructive: Under hyperination, holding money is very costly because the purchasing power decreases,
and as a result, everyone is willing to nd ways not to hold money for example,
hoarding of real assets or extreme consumption.
But isn't money supply controllable by the central bank? Why do the central
bank want to print such an incredible amount of money only to bring hyperination? The following model, often called Some Unpleasant Monetarist Arithmetic (based on an article by Thomas Sargent and Neil Wallace), will help us to
understand it.
4.1.1
Model Description
We assume the following:

(1) Population grows at a constant rate, n. That is,
Nt
=1+n
Nt 1
(4.4)
for all t = 1; 2; :::, where Nt denotes the population size at period t.

(2) Output per capita (in real terms, i.e., in units of consumption goods), yt ,
is constant and equal to 1. That is,
yt = 1:
So total output of the economy in real terms, or real GDP, becomes
Yt = Nt y t = Nt ;
(4.5)
4.1 Seigniorage and In ation
57
which grows at n. Remember: Nt is both population size and real GDP.

(3) Government debt cannot exceed some given level. In particular, the debtto-GDP level cannot be higher than b, i.e.,
Bt
Nt
bt
b;
(4.6)
for all t = 0; 1; :::

(4) There are positive de cits at all periods (i.e., Gt
for simplicity, assume
Gt
Tt > 0). In particular,
Tt
= d > 0;
(4.7)
Nt
for all t = 0; 1; ::: This means that the deci t-to-GDP level is constant over time
and positive. (4.7) makes the model interesting: Spendings are not fully nanced
by taxation, so the government should determine how much to borrow and how
much to print.
(5) The quantity theory of money,
Mt Vt = Pt Yt ;
(4.8)
holds with a constant velocity, Vt = 1. This theory means the following. Yt is the
total output in real terms. So Yt units of nal goods are produced and purchased
by somebody in the market. Then Pt Yt dollars should be transacted, which is
called money demand. If money can be used only once, this money demand
should be equal to Mt , which is money supply. But money is used more than once
during a given period. If money turns over twice, we should have Pt Yt =2 = Mt .
Similarly, denoting by Vt the velocity of money (average number of times a
piece of money turns over in a period), we shoould have Pt Yt =Vt = Mt , or
equivalently, (4.8).
In this model, we further assume Vt = 1, so
Pt Yt = M t :
(4.9)
Then (4.5) and (4.9) imply that
Pt =
Mt
:
Nt
(4.10)
This equation determines the price level (Pt ), given money supply (Mt ) and real
GDP (or population) (Nt ).
58
(6) Monetary policy of the central bank is to determine the growth rate of
money supply. In particular, assume the central bank wants to keep this (net)
growth rate constant at . Then, the monetary policy is to pick in which
Mt
=1+ ;
Mt 1
(4.11)
whenever money growth is controllable. Notice that
Mt
Mt
Pt
Mt
Pt
Mt
=
Pt
= Nt
Mt 1 Mt
M t Pt
Mt 1
1
Mt
Mt 1
1
Mt
from (4.10)
So (4.3) can be written as
Gt + (1 + r)Bt
= Tt + Bt + Nt 1
Mt 1
Mt
(4.12)
In particular, under the monetary policy (4.11), this becomes
Gt + (1 + r)Bt 1 = Tt +
|{z}
{z
}
|
to be nanced
by taxation
that is, the seigniorage becomes Nt

GDP.
4.1.2
1+
Bt
|{z}
+ Nt
|
by borrowing
1+
{z
;
}
by printing money
, or equivalently, a fraction
1+
of the
Now we are ready to see how government debt (Bt ), money supply (Mt ), and
price level (Pt ) evolve over periods. First, divide both sides of (4.12) by Nt . That
is,
Gt
Bt 1
Tt
Bt
+ (1 + r)
=
+
+ 1
Nt
Nt
Nt Nt
Mt 1
Mt
59
Rearranging,
Bt
Bt 1 Gt Tt
Mt 1
= (1 + r)
+
1
Nt
Nt
Nt
Mt
Nt 1 Bt 1
Mt 1
= (1 + r)
+d
1
by (4.7)
N t Nt 1
Mt
1 + r Bt 1
Mt 1
=
+d
1
by (4.4)
1 + n Nt 1
Mt
Hence, the evolution of debt-to-GDP ratio, bt
bt =
1+r
bt
1+n
+d
Bt =Nt , is given by
1
Mt 1
Mt
(4.13)
We see what will happen in this ctional economy.

1. Evolution of Debt/GDP (bt ): Under the monetary policy (4.11), (4.13)
becomes
1+r
bt =
bt 1 + d
:
1+n
1+
Therefore, we have
;
1+
1+r
=
d
+d
1+n
1+
1+
1+r
=
1+
d
;
1+n
1+
"
#
2
1+r
1+r
= 1+
+
d
1+n
1+n
1+
b0 = d
b1
b2
so
bt =
=
d
d
1+
1
1= + 1
t
X
i=0
t
X
i=0
1+r
1+n
1+r
1+n
(4.14)
i
if
6= 0
(4.15)
for all t = 0; 1; ::: in which (4.11) is true. This is an explicit solution for debt-to-
60
GDP ratio, bt .
2. Choice of Monetary Policy ( ): Assume r > n. (We believe this is true
Pt
1+r i
in reality.) The term i=0 1+n
in (4.14) or (4.15) is explosive. [Draw a line
for .] Suppose is zero (i.e., the bank does not print any money so there is no
money growth). Then, the term d 1+ in (4.14) becomes d. As increases
from 0 (i.e., as the bank increases money growth), as is clear from (4.15), this
term d 1+ decreases, so bt decreases, which means the government borrows
less in the nancial market. The term d 1+ will reach 0 if increases to
in which
= 0;
1=
+1
1
:
1=d 1
or equivalently,
(4.16)
If the central bank increases even more from this level, then the term d 1+
becomes negative. But this means that bt is also negative, so the government is
lending resources. Rule out this case: Facing the scal de cit, the government
tries to print money and to borrow money (instead of lending), so assume that
is not higher than in (4.16).
3. Catastrophe Date (T ): Let's go back to (4.14). If r > n and if
, the
debt-to-GDP ratio, bt , grows over periods (except that it stays at 0 when = )
according to (4.14). But wait. This ratio cannot grow forever because of (4.6).
In other words, when < is given, bt increases until some period T in which
bt reaches b. After that, bt should stay at the same level forever. This period T is
called the catastrophe period. [Draw a time path for bt .] Mathematically, T is
roughly obtained by solving:
1+
T
X
i=0
1+r
1+n
= b:
4. Money Supply (Mt ) and Price Level (Pt ): Now consider the money
growth, Mt =Mt 1 . Before T , from (4.11), gross money growth is 1 + . After
T , issuing debts is restricted, so the central bank is forced to print a lot of money
61
to generate seigniorage. Formally, since bt = b, (4.13) implies that
b=
1+r
b+d
1+n
Mt 1
Mt
for t = T + 1; T + 2; :::, or equivalently,
Mt
=
Mt 1
1
1
d+ 1
1+r
1+n
for t = T + 1; T + 2; ::: But
1+
1+
=
<
1
from (4.16)
1=d 1
1
1
1
d+ 1
1+r
1+n
since r > n
which means
1| {z
+}
money growth before T
<
1
|
1
d+ 1
{z
1+r
1+n
money growth after T
b
}
So the money growth should de nitely increase after T . [Draw a time path for
Mt =Mt 1 . Also for Mt . This should be amazing.] Of course, from (4.10),
higher money growth implies higher in ation other things being equal. [Draw a
time path for Pt .]
Result 1: If there are scal de cits at all periods, they should be nanced
by accumulating issued debts and/or by printing more money. However, after
the debt level reaches the limit, the central bank is forced to increase the money
growth because issuing debts becomes dif cult. This bring higher, uncontrollable
in ation.
5. Dilemma of the Central Bank: The central bank can choose between
lower and higher values of (money growth). The second result of this model
comes from considering different values on :
(1) Lower money growth (lower value of ) now ! Lower in ation before
T (which is good now) ! A larger part of de cit to be nanced by borrowing
62
! Debt-to-GDP ratio, bt , reaching its limit b sooner ! Catastrophe period (T )

arriving early (which is bad later)
(2) Higher money growth (higher value of ) now ! Higher in ation before
T (which is bad now)! A smaller part of de cit to be nanced by borrowing
! Debt-to-GDP ratio, bt , reaching its limit b later ! Catastrophe period (T )
arriving late (which is good later)
And, going extreme, we can make the in ation rate constant forever:
(2') Very high money growth ( = ) now ! Much higher in ation ! No
part of de cit to be nanced by borrowing ! Debt-to-GDP ratio, bt , staying at
zero ! Catastrophe period (T ) never arriving ! In ation rate staying
Now the conclusion:
Result 2: If there are scal de cits at all periods, the central bank should
choose between low in ation today and later onset of hyperin ationary catastrophe.
6. Balanced Budget: To avoid this unpleasant trade-off, one of the assumptions needs to be broken. The most important assumption that brings this
dilemma is (4.7). That is, if budgets are balanced, these problems will not arise.
Result 3: Balanced budgets are important!
Thomas Sargent, The Ends of Four Big In ations, In ation: Causes and
Effects (1983), considers four episodes of hyperin ation in Poland, Hungary,
Austria and Germany in 1919-1924. He argues that (1) governments were running huge de cits after the war, which were nanced by borrowing and printing
money, and that (2) hyperin ation ended after regime changes: Governments
abandoned scal de cits and seigniorage nancing in favor of balanced budgets
and independent central banks. (DLS, p. 222) These observations are consistent
with our results.
True or False?
1. Assume that the real per-capita output is constant at one unit of consumption good and that the population grows at n (e.g., 1%) where n > 0. Also,
suppose that the quantity theory of money holds with a constant velocity. Then,
even though the money supply increases, i.e., Mt =Mt 1 > 1, it may happen that
the price level decreases, i.e., Pt =Pt 1 < 1.
2. Consider an economy in which the per-capita real output increases by 2%
per year, population grows at 3% per year, and money stock (Mt ) grows at 6%
4.2 Price Level, Output and (Un)Employment (Monetary Neutrality)
63
per year. If the quantity theory of money holds with a constant velocity of money
(V ), then the price level of this economy will stay at the same level over time.
3. Consider the environment in Some Unpleasant Monetarist Arithmetic.
Suppose the government spending exceeds the government tax revenue in an
economy in which the government is not legally allowed to issue any debts. Then,
the central bank in this country is not able to control the in ation.
4.2
Price Level, Output and (Un)Employment (Monetary

Neutrality)
Question: How Does Money Growth (or In ation) Affect the Real Output and (Un)Employment Rate? (References: Chapter 19 of DLS and
various papers including Lucas's (1995) Nobel lecture, Monetary Neutrality, Economic Sciences.)
If the central bank prints more money (and in ation arises), will it increase
or decrease the real output? To answer this question, we will introduce two
theoretical models. In the rst, the overall price level is perfectly predictable. In
the second, the information on the overall price level is imperfect, so consumers
are not sure about tomorrow's price level. The conclusion we will derive is that
the real output may be affected by unanticipated changes in money supply (or
in price level).
4.2.1 Model 1: Perfect Information on Price Level
There are N industries, and each industry has one representative consumer. Industry i produces good i with the following (symmetric) production function:
Yi = Li ; for i = 1; :::; N;
(4.17)
where Yi is the output in terms of units of good i, and Li is the hours worked by
representative consumer i. (So one-hour working produces one unit.) She sells
all her products in the market. If the price in terms of dollars is Pi , she makes
Pi Yi dollars, which becomes her nominal income.
But no one consumes money itself, so nominal income is not important. That
is, we have to control for the overall price level P to make it her real income.
There are several ways to measure P , but here let us use the following geometric
mean of all prices:
p
N
P
P1 ::: PN :
(4.18)
64
(This type of index makes our discussion more convenient.) Similarly, let us also
denote by Y the aggregate output, which is also de ned as a geometric mean of
all outputs:
p
N
Y
Y1 ::: YN :
(4.19)
The real income is, as usual, the nominal income (Pi Yi ) divided by the overall
price level. The real income, or equivalently consumption, becomes
Pi Yi
:
(4.20)
P
Assume that the utility function of representative consumer i is given by
Ci =
(Li )
U (Ci ; Li ) = Ci
> 1:
(4.21)
One way to interpret (4.21) is to consider U (Ci ; Li ) as a pro t of rm i. That

is, Ci = PPi Yi is the revenue, and (Li ) is the labor cost (which is convex). Representative consumer i maximizes her utility, given the price of good i (Pi ) and
overall price level (P ). That is, we assume:
Assumption for Model 1: Each consumer i perfectly observes not only the
price of good i (Pi ), but also the overall price level (P ).
From (4.17), (4.20) and (4.21), she maximizes
Pi L i
P
(Li )
(4.22)
by choosing Li . The rst-order condition is
Li =
Pi
P
From (4.17), this equation also gives the output of industry i. That is,
Yi = L i =
Pi
P
1
1
(4.23)
Since > 1, the power 1 1 is positive, which means that the labor input (Li )
or output (Yi ) increases in the relative price of good i to the overall price level
(Pi =P ). The interpretation is clear: As the good you produce becomes more
expensive (relative to other goods), you will produce more units. It is convenient
65
to take logs on each side of (4.23) to have
yiS =
(pi
(4.24)
p);
log(Yi ) and lower-case variables denote logs (e.g., pi

where yiS
(4.24) describes the supply of good i given pi and p.
log(Pi )).
We are looking for an equilibrium, so we also need an equation describing

the demand for good i. We will simply assume that the demand is given by
yiD = y + zi
p); where
(pi
> 0:
(4.25)
Here, y is a constant. zi represents preferences for good i, taking a higher value

PN
as it becomes more popular, satisfying i=1 zi =N = 0. And the nal term
( (pi p)) implies that as good i becomes relatively more expensive (i.e., as
pi p increases), the quantity demanded decreases. If we accept (4.25) and take
average over all goods, then we have
!
PN D
PN
PN
y
z
p
i
i=1 i
i=1 i
= y + i=1
p :
N
N
N
But we know the second term in the right-hand side becomes zero by assumption.
Furthermore, (4.18) implies
log P =
1
(log P1 + ::: + log PN );
N
that is,
PN
i=1
pi
= p;
N
so the third term also becomes zero. Hence, we nally have
PN D
i=1 yi
= y:
N
(4.26)
(4.27)
So just as p in (4.26) is the log of overall price level, the constant term (y )
of (4.25) is, in fact, the log of aggregate demand. ((4.19) introduces a similar
de nition for the aggregate output. In the equilibrium, aggregate demand equals
aggregate supply which equals aggregate output, so they are all the same.)
66

In the equilibrium, demand equals supply, so from (4.24) and (4.25), we have
1
1
(pi
p) = y + zi
(pi
p);
(4.28)
or equivalently,
1+
(pi
p) = y + zi ;
or equivalently,
pi =
(4.29)
(y + zi ) + p:
1+
This equation determines how the (log) price of good i is related to the taste (zi ),
(log) aggregate output (y ), and (log) overall price level (p).
Our goal is to relate the total output (Y ) to the price level (P ) to see how
changes in P affects Y . Averaging (4.29) over all i = 1; :::; N , we have
p=
by p =
Hence,
1
N
PN
i=1
1
1+
pi (which we showed!) and

1
1+
but since
1
1+
6= 0 from
y + p;
PN
i=1 zi =N
= 0 (by assumption).
y = 0;
> 1, we have y = 0, which implies

Y = 1:
(4.30)
So the aggregate output is constant at 1, and is not affected by the overall price
level, P . From (4.17), units of outputs are equal to units of labor inputs. If we
de ne some sort of employment index by
p
L = N L1 ::: LN
(4.31)
just as (4.18) and (4.19), then (4.30) implies L = 1. So the (un)employment level
is not affected by the overall price level, P .
Of course, the overall price level is determined by money supply controlled
by the central bank. We have M V = P Y . We assume V = 1, and we know from
(4.30) that Y = 1. This means M = P . So a 10% increase in money supply
67
(M ) only increases the price level (P ) by 10%, but does not change any real
variables such as the aggregate output or (un)employment index. This result is
called monetary neutrality.
Result 4: If the information for the overall price level is perfect, an increase
in money supply affects the overall price level but not the aggregate output or
(un)employment rate.
[Show gures from Lucas (1995).] This result appears to be supported by
many pieces of empirical evidence. If we look at many annual observations of
the United States, in ation rates are positively related to money growth rates, but
are not closely related to real GDP growth rates or unemployment rates.
However, it seems that some subsets of data (for example, 5 or 6 consecutive annual observations) sometimes reveal negative relationships between
in ation rates and unemployment rates. This implies that in the short run, contrary to our predictions, money growth may increase real outputs and decrease
unemployment. How can we explain the Phillips curve the curve relating
in ation rates to unemployment rates which is sometimes downward-sloping
in the short run?
4.2.2
Model 2: Imperfect Information on Price Level
A substantial number of economists are willing to accept the argument that money
growth has no long-run effects. But short-run effects of money growth are expected to exist for various reasons. The following model, known as Lucas island
model, provides theoretical explanation on these short-run effects.
The environment now is slightly different from Model 1. Assume different
industries (i.e., different representative consumers) live in different islands. A
representative consumer still knows the price of the good she produces, but does
not exchange information with others. So she only has imperfect information
on the overall price level (P ). That is, representative consumers only have their
own beliefs (which are not necessarily identical to one another) on the overall
price level.
To make the setup simple, we will assume that all representative consumers
believe that the price level would surely be P e , which is constant.
Assumption for Model 2: Each consumer i observes only the price of good
i (Pi ), but not the overall price level (P ). All consumers have a common forecast
of the overall price level (P e ).
68
Under this assumption, consumer i maximizes the following objective function (replacing (4.22)), by choosing Li :
Pi Li
Pe
(Li )
A new rst-order condition (replacing (4.24)) becomes
yiS =
(pi
pe );
(4.32)
which describes the supply of good i.

The interpretation of this equation is important. Suppose that the overall price
level increases, and because of this, the (nominal) price of good i also increases.
Although this is simply a nominal increase, the representative consumer i may
mistakenly interpret this as an increase in the real price (Pi =P ). This misperception increases her production. If all producers act like this, there will be an
increase in aggregate output.
On the other hand, assume that the demand function (4.25) is still valid. This
implies that the equilibrium is made by equating yiD of (4.25) (i.e., demand) and
yiS of (4.32) (i.e., supply), that is,
1
1
pe ) = y + zi
(pi
(pi
p);
(replacing (4.28)). Hence we have
1
1
pi = y + zi + p +
pe :
Taking average over i, we have
1
1
p=y+ p+
1
1
pe :
(See the counterpart algebra in the previous subsection.) Or equivalently,
y=
1
1
(p
pe ):
(4.33)
This is the result, and now y (log aggregate output) is related to p (log overall
price level).
69
The interpretation of (4.33) is as follows. If the forecast on overall price level

is perfectly correct, i.e., P = P e or p = pe , (4.33) implies y = 0. So even
though the overall price level changes due to some change in monetary policy,
the output is not affected as long as it is anticipated. The monetary policy is valid
only when it is unanticipated. If the actual price level becomes higher [lower]
than the forecasted level (i.e., P > P e or p > pe ), then (4.33) implies that output
will increase [decrease].
We have similar results for the employment level. If we de ne l
where L is de ned in (4.31), then (4.33) also implies
l=
(p
pe ):
log(L)
(4.34)
So unanticipated changes in the price level will affect (un)employment level of

the economy. [Draw gures similar to Phillips curves.]
(4.33) also implies
1
P
Pe
Y =
Inserting this into M V = P Y with V = 1,
M =P
P
Pe
1
1
or equivalently,
1
P =M
(P e ) :
Once the forecast (P e ) is given, the price level (P ) still increases in the money
supply (M ) since 1 > 0. In particular, if P = P e , we will have P = M .
Result 5: An increase in money supply affects the overall price level. But
only unanticipated movements in this overall price level affects real output or
(un)employment. That is, if the overall price level rises more than anticipated,
then the real output increases and the unemployment level decreases.
If this result is correct, the Phillips curve, sometimes downward-sloping in
the short run, can be explained. (4.34) can be written as
lt =
1
1
(pt
pet )
70
by introducing some time subscript t, or equivalently,

2
lt =
6
4(p| t {zpt }1 )
t
7
(pet pt 1 )5 :
| {z }
e
t
Log differences become the growth rates, so pt pt 1 and pet pt

and anticipated in ation rates, denoted by t and et here. That is,
lt =
1
1
e
t) :
are actual
(4.35)
So if et stays at a similar level for some period, then lt (log employment) and
t (in ation rate) are positively correlated, so we can say that unemployment
and in ation are negatively related. This justi es a possible negative relationship
between in ation rate and unemployment rate. Of course, if consumers change
their beliefs, this negative relationship may not necessarily hold. That is, in the
long run, the downward-sloping Phillips curve is not expected to be observed.
4.2.3
Discussion
The growth of money supply (M ) is closely related to changes in price level (P ),

both theoretically and empirically. But it is more complicated to relate money
growth or in ation rate (nominal term) to real GDP growth or unemployment rate
(real term). Although a clear, long-term relationship between these variables does
not seem to be observed (which implies monetary superneutrality), we cannot
completely rule out the possibility of some sort of negative relationship in the
short run.
One attempt to explain this is the island model that we have just discussed.
Producers only have imperfect information on the overall price level, so they
may mistakenly perceive nominal increases [decreases] in prices of the goods
they produce as real ones. If this happens, producers produce more [less] units,
increasing the aggregate output.
Although we do not cover in this lecture, another attempt to explain the effects of money growth on output and (un)employment is the sticky-price model.
Prices of various products in the real world do not necessarily change whenever
the overall price level moves. Perhaps it is costly to change the price tags every
day. (This cost is called the menu cost.) Also in some cases, long-term contracts
are made with xed prices. If these are important, price changes may nd their
71
ways to affect real variables.
True or False?
1. A higher level of money growth increases the real output and decreases the
unemployment rate.
2. Because in ation rate and unemployment rate are negatively correlated,
high in ation and high unemployment cannot coincide. (This is called stagnation.)
72

Exercises
1. (Seigniorage and In ation) Consider an economy which is identical to the

one discussed in Section 4.1, except that the government is not allowed to borrow
at all at any period. (Thus, in this economy, the government expenditure should
be nanced either by taxation or by seigniorage.)
i) Population grows at a constant rate, n > 0. That is, Nt =Nt
all t = 0; 1; :::, where Nt denotes the population size.
= 1 + n, for
ii) The real output per capita, yt , is constant and equal to 1. That is, yt = 1
for all t = 0; 1; :::
iii) The real government debt is always zero. That is, Bt = 0 for all t =
0; 1; :::
iv) There are positive de cits at all periods (i.e., Gt Tt > 0). In particular,
assume (Gt
Tt )=Nt = d > 0; for all t = 0; 1; :::; where Gt is the real
government consumption and Tt is the real tax revenue.
v) The quantity theory of money holds with a constant velocity, Vt = 1: That
is, Pt Yt = Mt ; for all t = 0; 1; :::; where Pt is the price level, Yt is the real
output of this economy, and Mt is the money stock.
vi) Seigniorage should be enough to satisfy the government budget constraint.
Answer the following questions. (You may assume period-( 1) values for
all variables are given.)
(a) Write down the government budget constraint in this economy and explain what each term means.
(b) Describe the evolution of Mt given Mt 1 and d. What is the money
growth rate (i.e., Mt =Mt 1 1) given d? How is this money growth rate related
to d? What is your economic explanation about this relationship?
(c) Show how Mt evolves over time. (That is, draw a graph with time t on
x-axis and Mt on y-axis.) Feel free to assume M 1 is given and start with time 0.
If d becomes smaller, what happens to the evolution of Mt ? What if d becomes
larger? Explain why in plain words.
(d) Find an explicit expression for the in ation rate, Pt =Pt 1 1. Suppose the
population growth rate is n = 1%. For what value of d is the in ation rate zero?
Is it possible that the government enjoy seigniorage even though the in ation rate
stays at zero? Why or why not? Explain in plain words. (If you want to, quote
73
the quantity theory of money.)

2. (Seigniorage and In ation) Consider an economy which is identical to the
one discussed in Section 4.1:
i) Population grows at a constant rate, n > 0. That is, Nt =Nt
all t = 0; 1; :::, where Nt denotes the population size.
= 1 + n, for
ii) The real output per capita, yt , is constant and equal to 1. That is, yt = 1
for all t = 0; 1; :::
iii) The debt-to-GDP ratio cannot exceed a given level, b. That is, bt
Bt =Yt
b for all t = 0; 1; :::; where Bt is the real government debt and Yt is
the real output of this economy.
iv) There are positive de cits at all periods (i.e., Gt Tt > 0). In particular,
assume (Gt
Tt )=Nt = d > 0; for all t = 0; 1; :::; where Gt is the real
government consumption, Tt is the real tax revenue, and d is a given constant.
v) The quantity theory of money holds with a constant velocity, Vt = 1: That
is, Pt Yt = Mt ; for all t = 0; 1; :::; where Pt is the price level and Mt is the
money stock at the end of t.
vi) The central bank keeps the money growth rate constant at whenever
possible (i.e., before bt reaches b). That is, Mt =Mt 1 = 1+ ; for all t = 0; 1; :::;
whenever possible.
vii) The government budget constraint is given by
Gt + (1 + r)Bt
= Tt + Bt +
Mt
Mt
Pt
where r is the real interest rate.

Based on this model, we derived:
bt =
1+r
bt
1+n
+d
Mt 1
Mt
for all t = 0; 1; ::: Throughout this question, assume b

assume r > n unless otherwise noted.
;
= 0 is given. Further,
Is the following statement TRUE or FALSE? Fully justify your answer by

elaborating some of given equations. Provide intuitions whenever possible.
(a) If the government de cit is always zero (i.e., if d = 0), the central bank
is able to keep the money growth rate at zero forever.
74
(b) If the government de cit is always zero (i.e., if d = 0), the central bank is
able to keep the money growth rate at zero forever, and in this case, the in ation
rate is always zero.
(c) Even though the government always has positive de cit (i.e., d > 0),
it is possible that the government debt stays at zero forever by choosing some
monetary policy.
(d) If the GDP growth rate is higher than the interest rate (r < n), then even
though the government de cit is high relative to the level of money growth rate
(i.e., d 1+ > 0), it is possible that the a catastrophe date never arrives (i.e., bt
P1
does not reach b forever). (Hint: t=0 t = 1=(1
) for 0 < < 1.)
(e) The condition d
than seigniorage.
1+
> 0 implies that the government de cit is higher
Chapter 5
Monetary Policy (II)
This chapter introduces theoretical approaches to nd the optimal monetary policy.
5.1
In ation, Unemployment and Optimal Monetary Policy

(Expectations-Augmented Phillips Curve)
Question: Knowing the Relationship between In ation and Unemployment, How Should the Central Bank Design the Monetary Policy? (Reference: Chapter 19 of DLS.)
In a previous chapter, we attempted to nd the optimal scal policy assuming
that the government minimizes the present value of deadweight losses. In this
section, our goal is to nd the optimal monetary policy, where the central bank
minimizes some sort of losses caused by in ation or unemployment. In particular, we will assume that the sum of squares of in ation rate and unemployment
rate (u2 + 2 ,) is the objective to be minimized.
We start from (4.35). This equation argues that the employment is positively
related to the actual in ation rate minus the expected. Let us write this as
u=u
b(
); where b > 0:
(5.1)
Here, u is the unemployment rate (in terms of %), u is some constant, is

the actual in ation rate (%), and e is the anticipated in ation rate (%). This
equation, often called the expectations-augmented Phillips curve, means that
the unemployment rate is negatively related to the actual in ation rate minus the
expected. Notice that if = e , we have u = u , so the constant term (u ) in
(5.1) is the unemployment rate in which the actual in ation is perfectly anticipated. This level is called the natural unemployment rate, which is assumed to
be constant in this setup.
The central bank is assumed to (equally) dislike unemployment and in ation.
In particular, it minimizes
u2 +
75
76
Chapter 5 Monetary Policy (II)
With (5.1), the central bank's problem can be written as
min [u
b(
)]2 +
(5.2)
We assume that the central bank can freely choose the level of in ation rate (for
example, by printing money). The optimal monetary policy is to set the in ation
rate which solves (5.2). Different assumptions on the anticipated in ation e
yield different results.
Case 1. Fixed Expectations: First, assume e is xed and observable. So e
is a given constant, which is known to the central bank. The rst-order condition
to (5.2) is
2b[u
b(
)] + 2 = 0;
or equivalently,
e
= b[u + b(
so the optimal in ation rate given
( e) =
, which is denoted by
b
u +
2
|1 +{zb }
positive intercept
)];
b2
2
{zb }
|1 +
( e ), satis es
e
(5.3)
positive slope, smaller than 1

2
b
[Draw a graph.] Notice that the slope ( 1+b
2 ) is positive. So if the anticipated
in ation is high [low], the central bank should keep the actual in ation high
[low]. Also notice that the intercept is positive but the slope is lower than 1.
This implies that when the anticipated rate of in ation is relatively low [high],
the optimal rate of in ation should be higher [lower] than the expectation.
Case 2. Nash Equilibrium: But this solution is based on xed expectations. Consumers may know that the central bank's target is ( e ) in (5.3).
This information will change their anticipated rate of in ation. If so, ( e )
is no longer optimal because e itself is changed. [Use the graph to make this
clear.] This environment becomes a game in which the central bank chooses
and consumers choose e . The objective of central bank is to solve (5.2),
and that of consumers is to anticipate the in ation rate as accurately as possible.
As in a usual perfect-information game, consumers in this game know about
the central bank's problem. Also, the central bank knows that consumers know
about the central bank's problem. Further, consumers know that the central bank
knows that consumers know about the central bank's problem. ... (This continues
5.1 In ation, Unemployment and Optimal Monetary Policy (Expectations-Augmented

77
Phillips Curve)
forever.) Similarly, the central bank knows about the problem of consumers,
consumers know that the central bank knows about the problem of consumers,
etc.
Let us solve this game. In a Nash equilibrium, players maximize their
payoffs (or minimize their costs) given the strategies of others. It is clear that
the central bank will choose ( e ) in (5.3) once the consumer's strategy e is
given. Consumers will set e to be ( e ) once the central bank's strategy is
given. That is, we should have
( e) =
in a Nash equilibrium. Hence, (5.3) implies
b
b2
u
+
1 + b2
1 + b2
or equivalently,
= bu ;
which means the optimal monetary policy is to keep the in ation rate constant at
bu %. In this Nash equilibrium, both expected and actual in ation rates become
this level at . So (5.1) implies that the unemployment rate becomes
u = u
b(
= u
So the unemployment rate will also be constant at the natural level.
Case 3. Ramsey Equilibrium: Can't we do better? We can also imagine
the following consequence, called the Ramsey equilibrium. Let the central
bank announce the target in ation rate to the public. (And assume that this
announcement is credible.) Suppose that the announced target is
. Also
suppose that consumers trust the central bank, so the anticipation of consumers
is at the same level,
. So (5.1) implies that
u = u
= u
b(
Hence, the unemployment rate is not affected by the target in ation rate as long
78
as it is credible. Recall that the government minimizes
u2 +
= ( |{z}
u )2 + (
constant
|{z}
)2 ;
target in ation rate
so the solution is obviously

= 0. To summarize, in the Ramsey equilibrium,
the in ation rate is zero, (that is, the central bank announces 0% in ation as a
target, consumers trust the central bank, and the central bank actually sets 0%
in ation as promised,) and the economy has the natural unemployment rate.
Of course, to obtain this result, the central bank needs to credibly commit
to the target. But notice that the central bank may choose to deviate. That is, if
consumers actually anticipate 0% in ation, (5.3) implies that the optimal in ation
b
rate is now 1+b
2 u %, so the central bank may break its promise to increase the
in ation rate. Because of such a worry, consumers will not trust the central bank's
0% target, so this Ramsey equilibrium (which requires consumers to accept the
target) will be hard to be attained again. This is similar to Prisoner's dilemma in
which both players can be better off if both choose to deny.
The credibility of monetary policy is therefore important. It is better for the
central bank to keep the in ation as low as possible and to announce this target to
the public to affect the anticipation. To do this, the central bank should show its
credibility by keeping promises for a long time. Perhaps, this is one reason why
the Federal Reserve tries its best to convince the people that it is very seriously
ghting against the in ation.
True or False?
1. If we look at annual observations of any four or ve consecutive years, the
in ation rate should be inversely related to the unemployment rate.
2. Suppose the price level has been constant for a long time until year 2006.
A news is released, at the beginning of 2006, that the central bank will increase
the money supply in that year. If all people expect that the price level would
increase by 1% (in 2006), and if it actually increases by 2% (in 2006), then the
unemployment rate will increase.
3. Page 235 of DLS says: Paul Volcker [in early 1980s] arrived as Chairman of Federal Reserve Board at a time of high in ation and high unemployment.
He announced that there would be low in ation in the future. The private sector
did not adjust its expectations, but Volcker followed through on his promise.
Then, the unemployment would have stayed above the natural rate for a while.
5.2 Interest Rate and Optimal Monetary Policy (Friedman Rule)
79
4. In the optimal monetary policy problem minimizing the sum of squares

of in ation rate (%) and unemployment rate (%), a Nash equilibrium gives the
optimal result.
5. Assume an expecations-augmented Phillips curve, ut = u
b( t
where ut is the unemployment rate at period t, u is a positive constant
equals to natural unemployment rate, b is a positive constant, t is the actual
in ation rate at period t, and et is the anticipated in ation rate at period t.
Suppose that the Federal Reserve wants to minimize the discounted value of sum
P1
of squares of in ation rate and unemployment rate (i.e., t=0 t ( 2t + u2t )) by
choosing the in ation rates. Further, suppose that the in ation rate anticipated
by consumers is given by 0% in period 0. Then, in period 0, the Federal Reserve
would choose an in ation rate which is strictly positive if the anticipated in ation
rates in periods 1, 2, ... depend on the in ation rates of previous periods (periods
0, 1, ..., respectively). [Correct intuitions are enough. You don't have to be formal
in your solution. If you want to be formal and if you need, you are allowed to
make any reasonable assumptions.]
e
t ),
6. Because the scal and monetary policies should harmonize, the central
bank should be under a perfect control of the government.
5.2
Interest Rate and Optimal Monetary Policy (Friedman

Rule)
Question: How Should the Central Bank Design the Monetary Policy to
Minimize the Cost of Holding Money? And What are the Tools of Monetary Policy in Practice? (References: Sections 17.1 and 19.4 of DLS
and the website of Board of Governors of the Federal Reserve System
(http://www.federalreserve.gov).)
We have assumed that the central bank simply hates in ation. But why?
In this section, we continue to think about the optimal monetary policy, but with
some theory about why in ation is bad. Savings provide interests, but holding
money doesn't. We should hold some amount of money for daily purchases,
which is costly because interests are not paid. Can the central bank do something
to minimize the cost of money holding? This is the main question of the section.
First of all, we need a story relating (nominal) interest rate to (anticipated)
80
in ation. The following Fisher formula provides this relationship:
r+
(5.4)
where R is the nominal interest rate, r is the real interest rate, and e is the
anticipated in ation rate. This formula means the following: If someone borrows
one unit of consumption good in this period, he/she should pay back (1 + r) (for
example, 1.02) units in the next period. But suppose the in ation is expected
to be e (for example, 0.03 or 3%). This means one unit of consumption good,
that is worth $1 in this period, will be worth $(1 + e ) in the next period. So in
terms of dollars, if someone borrows $1 (which is worth one unit of consumption
good) in this period, he/she should pay back $(1 + r)(1 + e ) dollars (which
is worth (1 + r) units of consumption goods) in the next period. This implies
that the (net) nominal interest rate for money should be r + e + r e . If r e (for
example, 0.02 0.03=0.0006) is negligible, (5.4) is a reasonable approximation.
In particular, assume that (5.4) holds with an equality (for simplicity) and that
the in ation rate is correctly forecasted. That is,
R=r+ :
(5.5)
Notice that under this assumption (on correct forecast of in ation), the unemployment always stays at its natural level. This enables us to forget about the
labor market and to concentrate on minimizing the costs arising (directly) from
in ation. We further assume that r in (5.5) is determined in the nancial market
and is not controlled by the central bank. The central bank can in uence the
nominal interest rate (R) by affecting the in ation rate ( ) only.
5.2.1
Model Description
Now let us describe the model. A consumer has all her asset in the bank, which
pays the nominal interest of R at each period. She goes to the bank x times
per period to withdraw money for daily transactions. Assume that there is a xed
(constant) cost of dollars to withdraw any amount of money. (This may include
time cost, transportation cost and transcation cost.) That is, withdrawing x times
costs
(5.6)
dollars. Also, there are opportunity costs from holding money. Assume this
consumer withdraws c dollars in one period. (For simplicity, assume c is given.)
81
She goes to the bank x times in one period, so she withdraws c=x dollars every
time. So she starts living with c=x dollars right after each withdrawal, constantly
spends money until her money holding becomes 0, when she withdraws another
c=x dollars. This means that she will, on average, hold c=2x dollars and lose the
interests paid to this amount. That is, she loses
c
R
2x
(5.7)
dollars. Therefore, her total cost of holding money becomes
x+
c
R
2x
(5.8)
dollars. The consumer chooses x to minimize this. There is a trade-off. If x is

small (i.e., if she rarely goes to the bank), then cost of withdrawal, (5.6), is small,
but opportunity cost of holding money, (5.7), is large.
5.2.2
The rst-order condition in this minimization problem is
cR
= 0;
2x2
or equivalently,
x=
cR
:
2
In this case, (5.8) implies that the total cost of this consumer from holding money
becomes
s
p
c
cR
+ q R = 2cR ;
(5.9)
2
2 cR
2
given c (amount of dollars required for consumption),

and R (nominal interest rate).
(unit cost of withdrawal)
The central bank, able to control R, should try to minimize this cost, (5.9).
Obviously, the solution to minimize this cost is
R = 0:
82
In this solution, (5.5) implies that the optimal in ation rate becomes
r:
This result that the in ation should be negative is called the Friedman rule.
Under the Friedman rule, the opportunity cost of holding money, (5.7), becomes
zero, which means consumers hold all assets as money in order to minimize costs
from withdrawal, (5.6).
Perhaps this result is not really practical for an actual monetary policy. When
the nominal interest rate is zero, no one will save his/her money at banks, so
nancial intermediaries do not exist anymore. Also, de ation is harmful in
production because simply holding money is sometimes better than investing.
However, this result is interesting at least theoretically. The model implies that
there are costs to consumers when they have in ation (and in fact, even mild
de ation). Consumers are always losing something because they should go to
the bank and because holding money does not pay interests. These costs become
zero only under the Friedman rule.
5.2.3
Monetary Policy in Practice
In this subsection, we discuss three main tools of monetary policy in practice.

1. Reserve Requirements: Banks in the United States should hold a given
fraction of their deposits, as cash in their vaults or as deposits with Federal
Reserve Banks with no interests. (Banks may hold even more reserves than
requirements to restock ATMs, to clear overnight checks, etc.) Changing the
required fraction of these reserve requirements affects the money stock (M ) in
the following way.
Let us consider a hypothetical economy in which consumers always just save
their money in banks. Assume that the required fraction is set to 10% by the Fed.
The Fed printed $100 for circulations. Consumers save this money in banks. The
banks need to reserve 10%, or $10, while $90 are lent out to other consumers.
These consumers, again, save this amount ($90) at banks. Banks again reserve
$9, and lend out $81 to the people. This continues forever. So the printed amount
of dollars, $100, now become the money stock of
100 + 100
by
P1
t=0
= 1=(1
0:9 + 100
(0:9)2 + ::: = 100
) for 0 <
1
= 1000
1 0:9
< 1. As the reserve requirements become
83
higher [lower] (i.e., as the policy becomes contractionary [expansionary]), the

money supply becomes smaller [larger] because more [less] money should be
reserved.
2. Open-Market Operations: Banks, in order to satisfy reserve requirements, may borrow or lend among themselves at the close of business each
day. (That is, banks with reserve surpluses will lend money to others.) The
interest rate used in these transactions is called the federal funds rate. The Fed
can affect this federal funds rate by purchasing national securities from banks.
This procedure is called the open-market operation. If the Fed purchases them,
banks come to have more reserve in response with the Fed. This will decrease
the federal funds rate because banks, in general, have more federal funds. At the
same time, it will increase the money supply because the additional amount of
federal funds can be traded for money. This type of policy the Fed's purchasing
national securities is expansionary in the sense that it decreases the interest rate
and increases the money supply. (On the other hand, selling national securities is
contractionary.)
3. Discount Rates: The discount rate is the interest rate charged to commercial banks and other depository institutions on loans they receive from their
regional Federal Reserve Bank's lending facility the discount window. (Federal Reserve website) The Fed may increase (which is contractionary) or decrease
(which is expansionary) this rate to affect market interest rates.
True or False?
1. The Friedman rule is obtained by minimizing the costs of in ation and
unemployment.
2. A low rate of in ation is not costly at all.
3. Consider the (informal) setup of the Friedman rule, in which a consumer
minimizes the sum of (i) the cost of withdrawals and (ii) the opportunity cost from
losing interests, by choosing number of withdrawals. Even though the consumer
confronts de ation of 1% (where the price level of the next period is 99% of that
of this period), the sum of two costs may be positive.
4. Let us consider the setup that we considered when discussing the Friedman
c
rule. A consumer minimizes a total cost, x + 2x
R, which is the sum of withdrawal costs ( x) and opportunity costs from holding cash (cR=2x), by choosing
the number of costly withdrawals (x). Here, is the cost of each withdrawal (in
dollars), x is the number of costly withdrawals, c is the consumption (in dollars)
84
in one period, which is treated as given, and R is the nominal interest rate. As the
nominal interest rate (R) rises by 1%, the number of withdrawals (x ) optimally
chosen by the consumer decreases by 0.5%.
5. C.A.E. Goodhart, an economic advisor to the Bank of England for many
years, suggests having a lottery based on cash serial numbers. If we introduce
this lottery, the Friedman rule does not necessarily imply that the nominal interest
rate should be zero.
85
Exercises
1. (In ation and Unemployment) Suppose that the following expectationsaugmented Phillips curve holds:
ut = u
e
t );
for all t = :::; 2; 1; 0; 1; 2; ::: (So b in the lecture note is assumed to be 1.)
Here, ut is the unemployment rate (%), u is the natural unemployment rate (%),
e
t is the actual in ation rate (%), and t is the anticipated in ation rate (%).
Assume all these four are non-negative.
We are at time 0, and the in ation rate before time 0 (at times 1, 2, ...)
was constant at 10% (i.e.,
1 =
2 = ::: = 10). The central bank, able to
fully control actual in ation rates, minimizes at time 0
1
X
(u2t +
2
t );
t=0
by choosing 0 , 1 , 2 , ... (There is no time discount factor ( ) in this expression.

That is, we take
= 1. We consider a general value of in question (e).)
Notice that the optimal in ation rate does not grow forever. If it does, the above
expression will explode.
(a) Suppose that consumers have full information, so they always anticipate
the correct level of in ation rate, i.e., et = t for all t = 0; 1; 2; ::: Obtain the
optimal in ation rate. Obtain the unemployment rate. How do they evolve over
time?
(b) Suppose that consumers have adapted expectations so that
e
t
t 1.
i) Show that the rst-order condition satis es

t+1
t 1
=0
for all t = 0; 1; 2; :::

ii) Guess that the central bank sets t =
t 1 , for all t = 0; 1; 2; :::, for
some constant > 0. This cannot be higher than 1 if we follow the assumption
in the question. (That is, t does not grow forever.) Obtain .
iii) Describe how the optimal in ation rate and the unemployment rate move
over time.
86
(c) Why do your answers to these two questions differ? In particular, why
don't the central bank change the in ation rate more aggressively in Question 2?
Explain in plain words.
(d) [OPTIONAL: The solution will not be provided.] Suppose that consumers
are between the above two cases so that et = ( t 1 + t )=2. This means
that consumers do not 100% correctly predict the in ation rate, but at the same
time, they are not foolish enough to just adapt today's in ation rate to predict
tomorrow's. Describe how the optimal in ation rate and the unemployment rate
evolve over time. Is the solution to this question somewhere between those to the
previous questions?
(e) [OPTIONAL: The solution will not be provided.] How should we change
the answers to the previous questions if the central bank minimizes
1
X
(u2t +
2
t );
for 0 <
<1
t=0
instead?
2. (A Revised Baumol-Tobin Model for Friedman Rule) In our discussion on

the Friedman rule, we considered a model in which it is costly for a consumer to
withdraw money. But withdrawal technologies have evolved, and now we have
ATMs in many places from which we can withdraw money with almost no costs.
Our goal in this problem is to understand how the model's predictions on cash
holding change with this new environment.
Let us consider the original setup that we discussed once. A consumer minimizes a total cost, which is the sum of withdrawal costs ( x) and opportunity
costs from holding cash (cR=2x):
c
x + R;
2x
by choosing the number of costly withdrawals (x). Here, is the cost of each
withdrawal (in dollars), x is the number of costly withdrawals, c is the consumption (in dollars) in one period, which is treated as given, and R is the nominal
interest rate.
(a) Obtain the rst-order condition of the problem.
(b) As the nominal interest rate rises, does the number of costly withdrawals
(x) also increase? Why? (Provide intuitions.) What is the interest-rate elasticity
87
of number of withdrawals? (That is, when the interest rate rises by 1%, by what
% does the number of withdrawals increase?)
(c) As the nominal interest rate rises, does the average cash holding (i.e., the
average amount of cash that this consumer holds at any moment) also increase?
Why? (Provide intuitions.) And what is the interest-rate elasticity of cash holding?
(d) This original setup needs to be changed re ecting the development of
withdrawal technologies. As before, the consumer can still go to the bank to
withdraw money with unit cost . In addition, she meets the ATM p times
from which she can withdraw money without costs. The value of p is given
exogenously (and p becomes higher as withdrawal technologies develop). In sum,
this consumer's free withdrawals are limited to p times, while she makes costly
withdrawals (with cost of dollars each) x times. Set up the problem of the
consumer to minimize the total cost. Obtain the rst-order condition. (Assume
in each of costly or free withdrawals, she withdraw the same amount of money.
And assume she can make free or costly withdrawal at any moment she wants. If
you need, make some reasonable assumptions.) (Hint: She chooses x
0, the
number of times she withdraws costly.)
(e) Suppose a consumer's optimal solution is to choose x = 0 (that is, she
does not make costly withdrawals at all) for some given parameter values. How
many free withdrawals does she make? Why? (Provide intuitions.)
(f) For simplicity, assume that the entire development of withdrawal technologies is represented by increasing availability of ATMs. Hence, consumers
without ATM cards do not enjoy any bene ts from this development while those
with ATM cards do. Let us consider an economy with many consumers who
have identical parameters except p. That is, this economy consists of identical
consumers but the number of free withdrawals allowed is different across consumers.
Is the following statement TRUE or FALSE? In each of these, you should
explain the reason using both mathematical results and interpretations/intuitions.
i) An unpublished paper written by Fernando Alvarez and Francesco Lippi
reports the following statistics based on the survey conducted in Italy in year
2004. Reported numbers are sample means across many households.
Table 1: Number of (free or costly) Withdrawals per Year
88

Consumers without ATM cards
23.0
Consumers with ATM cards

63.1
As shown in Table 1, the development of withdrawal technologies (i.e., the

overall increase in individual p's) will increase the average number of (free or
costly) withdrawals.
ii) The same paper also reports:
Table 2: Average Cash Holding to Daily Consumption (c) ratio
Consumers without ATM cards
17.8
Consumers with ATM cards

13.6
As shown in Table 2, the development of withdrawal technologies (i.e., the

overall increase in individual p's) will decrease the Average Cash Holding to
Daily Consumption ratio.
iii) Interest-rate elasticity of number of withdrawals for an average consumer
will be lowered as p rises.
iv) Interest-rate elasticity of cash holding for an average consumer, in absolute value, will be lowered as p rises.
Chapter 6
International Trade
This chapter discusses some issues on international interactions among economies.
We rst consider economic theories on the determination of exchange rates.
Then, we discuss comparative advantages to understand gains from international
trade. A broader discussion on gains from international trade is provided as we
consider how international trade is related to the overall economic growth.
6.1
Exchange Rate (Purchasing-Power Parity)
Question: How Are the Exchange Rates Determined? (References: Chapters 15 and 16 of Barro.)
Different economies use different currencies. Let Ei denote the exchange
rate of economy i's currency:
US$1 = Ei in country i's currency
(6.1)
For example,
EU S = 1 (i.e., 1 U.S. Dollar = 1 U.S. Dollar),

EEU = about 0:82 (i.e., 1 U.S. Dollar = 0.82 Euros),
EJapan = about 114:07 (i.e., 1 U.S. Dollar = 114.07 Yens),
EU K = about 0:57 (i.e., 1 U.S. Dollar = 0.57 Pounds).
The equation EJapan = 114:07 means that you can exchange $1 for U114.07 in
the exchange market. Equivalently, U114.07 becomes the price of one dollar
1
in Japanese yens. This also means that you can exchange U1 for $ 114:07
. (That
is, just as one dollar gives Ei units of foreign currency i, one unit of currency i
gives 1=Ei dollars.) What if you want to exchange yens for pounds? You can
1
0:57
exchange U1 for $ 114:07
, which is worth 114:07
in pounds.
Let us introduce some terminology.
1. Devaluation: The value of dollar decreases, or the dollar is devalued, if
one dollar now gives less foreign currencies, i.e., Ei 's decrease for all i. Similarly, the value of a foreign currency i decreases, or the foreign currency i is
89
90
Chapter 6 International Trade
devalued, if one unit of foreign currency i now gives less foreign currencies, i.e.,
if 1=Ei decreases, or equivalently, if Ei increases, other things being equal.
2. Revaluation: This is the opposite of devaluation. A currency is revalued
if it now gives more units of foreign currency.
In the U.S. and many other developed economies, the movement of exchange
rate is determined in the exchange market. Just as for any goods, the value of a
dollar will increase (decrease) if there is a higher (lower) demand for it. Here we
discuss a few models on how exchange rates are determined in the market.
6.1.1 Model 1: Purchasing-Power Parity
Suppose there are only one type of consumption goods (e.g., apples). The unit
price of this good is $PU S in the U.S. And it is UPJapan in Japan. Disregard
all the barriers (geographic distance, tariff, etc.) between the U.S. and Japan.
Then, there are two ways for a consumer in the U.S. to buy one unit of this good.
First, she can simply pay $PU S in the U.S. market. Second, she can exchange
her U.S. dollars for Japanese yens and buy the product in the Japanese market.
To do this, she should pay UPJapan , which is equivalent to $PJapan =EJapan .
(Remember (6.1)!) By the law of one price, these two values should be the same
if the two markets exist at the same time. That is,
PU S =
PJapan
:
EJapan
(Otherwise, everyone will go to a cheaper market.) We can extend this relationship to a multiple-economy setup:
PU S =
PJapan
PEU
=
= :::
EJapan
EEU
(6.2)
This relationship is called the purchasing-power parity. Once the price levels
of all economies are given, the exchange rates are determined through (6.2).
What happens if the U.S. price level goes up (i.e., if there is in ation in the
U.S.) while the price levels in all other economies stay at the same levels? Then
according to (6.2), all Ei 's will decrease, which means the dollar is devalued.
(The intuition is clear: A higher price level in the U.S. means that one dollar can
buy only less units of goods. Obviously, the dollar is devalued.) The opposite
is true if the U.S. price level goes down. Similarly, if the Japanese price level
goes up (down), then EJapan will rise (fall), which means the yen is devalued
(revalued).
6.1 Exchange Rate (Purchasing-Power Parity)
91
Let me add a few remarks.

(1) Of course, this model is a simpli ed version of the real world. We do
not imagine that consumers can buy apples freely across international markets.
Barriers across economies forbid it. This may somewhat distort (6.2).
(2) Further, there are so many types of goods in reality. The overall price level
can be used to apply (6.2), but it is based on a consumption basket that may
be internationally different. Another way to evaluate (6.2) is to consider a good
which is identical in every country. Perhaps Big Mac is close to such a product.
We can compare the actual exchange rates and the exchange rates predicted by
(6.2) with Big Mac prices. It turns out that these two do not coincide, but are at
least positively related. [See the Big Mac indexes. (http://www.economist.com/markets/Bigmac/Index.cfm)]
Some investors use this type of information for pro ts.
6.1.2
Model 2: Interest-Rate Parity
The purchasing-power party is based on the goods market. What about the nancial market? Suppose that (nominal) interest rates for the U.S. and Japan are
given by constant RU S and RJapan: So if you invest $1 in period t in the U.S.,
you get $(1 + RU S ) in period t + 1. If you invest U1 in period t in Japan, you
get U(1 + RJapan ) in period t + 1. Assume there are no barriers between these
two economies.
Then there are two ways for an investor in the U.S. to invest $1. First, it can
simply invest in the U.S. nancial market in period t, which gives $(1 + RU S ) in
period t + 1. Second, she can exchange $1 for UEJapan;t in period t and invest in
the Japanese nancial market. This gives U(1 + RJapan )EJapan;t in period t + 1,
which is $(1 + RJapan )EJapan;t =EJapan;t+1 in dollars. Therefore,
1 + RU S = (1 + RJapan )
EJapan;t
:
EJapan;t+1
In a multi-economy setup, we have
1 + RU S = (1 + RJapan )
EJapan;t
= (1 + REU )
EJapan;t+1
EEU;t
= ::: (6.3)
EEU;t+1
This relationship is called the interest-rate parity.

As the interest rate in Japan (RJapan ) rises (other things being equal), the
EJapan;t
term EJapan;t+1
should drop from (6.3). The intuition is as follows. Now the
Japanese nancial market is more attractive because investments will provide a
92
higher return. Then every investor in the world will come to Japan, trying to buy
yens. To make this market less attractive, the exchange rate should be adjusted.
Then, all nancial markets in the world will be equally attractive.
Of course, there are some uncertainties in the real world (although this model
does not capture any of those). The interest rates in many economies are not
constant (i.e., some assets are risky). Furthermore, this model disregards the
barriers between economies. While the model provides a big picture, there are
some aspects left unexplained (just as in any economic model).
6.1.3
More on Exchange Rates
Some more discussions on exchange rates are provided.

1. Exchange Rate and Trade De cit: Some policy makers worry about
trade de cits of the U.S., so they sometimes argue that the U.S. dollar needs
to be devalued. What is the logic behind this argument? Suppose a U.S. rm
produces a good which is worth $1. It used to be U120 in the Japanese market,
but after devaluation of dollars, it is only, say, U110. So Japanese consumers
are more likely to buy U.S. products. Furthermore, if a Japanese rm produces a
good which is U120, then it used to be $1 in the U.S. before, but now the price
is higher than $1. Therefore, U.S. exports are promoted, while U.S. imports are
discouraged.
Nevertheless, devaluation does not always bring a decrease in the trade de cit.
The reason is as follows. By imports, the U.S. pays $P Q, which is price times
quantity imported. By the logic above, Q will decrease (i.e., the U.S. consumers
will decrease the consumption of imported goods). But wait: P increases, which
means the direction of P Q is unclear. Hence, the actual impact on trade de cit
is unclear.
2. Flexible vs. Fixed Exchange Rate: Until the early 1970s, most countries
were under the Bretton Woods System, which was based on xed exchange
rates. In this system, the U.S. promised to exchange US$35 for one ounce of
gold so that the value of the U.S. currency was xed. (So it was a gold-exchange
standard.) Other countries pegged their exchange rates to the U.S. dollar within
some narrow bands. But in 1960s, the U.S. price level had risen substantially,
and Richard Nixon nally decided in 1971 to raise the dollar price of gold. After
that, many countries stopped following the Bretton Woods, adopting exible
exchange rates.
The xed exchange rate system helps to decrease uncertainties around the
6.1 Exchange Rate (Purchasing-Power Parity)
93
exchange rate. But then, there arise restrictions on the monetary policy. (Recall
(6.2): To keep Ei constant, the central bank should keep Pi constant.) Furthermore, the economy may be more vulnerable to the currency crisis. To think
about this, assume that the Mexican Peso is pegged to the dollar (which used
to be true), and that the Mexican in ation is higher than the U.S. in ation. The
of cial ( xed) exchange rate is 3 Mexican pesos per dollar, but people think
it is 5-6 pesos per dollar in actual values. Then few people will want to hold
pesos because of the possibility of a sudden devaluation made by the exchange
authority. There also may be some speculative attacks. If you are sure that the
Mexican authority would decide to devalue pesos, there is an easy way to make
money: You borrow 3 pesos in Mexico, exchange them for one dollar, and simply
wait. After devaluation, you can now exchange your one dollar for, say, 5-6
pesos, so you pay your debt, and then 2-3 pesos become your net pro t. If many
speculators act like this, this will accelerate the crisis even further because there
will be higher demand in selling pesos.
In fact, a similar event happened in the 1994 Mexican currency crisis. Within
one week, the exchange rate had been raised from 3 pesos per dollar to 10 pesos.
There are many similar examples: the 1992 European Monetary System (EMS)
crisis in the U.K. and the 1997 Asian currency crisis, to name a few.
Bene ts and costs of exible exchange rates are, of course, the opposite to
what we have discussed on the xed exchange rate. The risk of currency crisis
may be somewhat lower, which is a bene t, but there are a lot of uncertainties in
future exchange rates, which is a cost.
True or False?
1. Suppose the purchasing-power parity holds. If the price level in Japan
rises relative to other economies, then the exchange rate for Japanese yen (i.e.,
EJapan in US$1=UEJapan ) will drop.
2. Assuming a theory of purchasing-power parity with a Starbucks Tall Latte
as a consumption good, the following information suggests that euro is overvalued against dollar.
- Unit price of a Starbucks Tall Latte in the U.S.: 2.80 dollars
- Unit price of a Starbucks Tall Latte in the E.U.: 2.93 euros
- Exchange rate: 1 dollar = 0.82 euros
3. There are two economies, the U.S. and the U.K., and two periods, today
94
and tomorrow. The exchange rate today is 1 pound per dollar. If the in ation
rates between today and tomorrow are 10% and 0% in the U.S. and in the U.K.,
respectively, then tomorrow's exchange rate will be 1.1 pounds per dollar, other
things being equal.
4. Suppose the interest-rate parity holds. If the (nominal) interest rate in
Japan rises relative to other economies, then the exchange rate for Japanese yen
(i.e., EJapan in US$1=UEJapan ) will drop.
5. Suppose that one-year interest rates are 1% in the U.S. and 2% in the
E.U., respectively. If the current exchange rate is 1 euro per dollar, then it is
expected that the exchange rate after one year would be higher than 1 euro per
dollar according to the interest-rate parity.3. In July 2005, the exchange rate of
the Chinese currency (yuan) was changed from 8.28 yuans per dollar to 8.11.
This would decrease the trade de cit of the United States against China.
and tomorrow. All investors are risk-neutral. The exchange rate today is 1 pound
per dollar. The exchange rate tomorrow will be 1.5 or 0.5 pounds per dollar with
equal probabilities, and it is known to all investors today. Then, the nominal
interest rate between today and tomorrow should be higher in the U.S. than in
the U.K.
and tomorrow. The exchange rate today is 1 pound per dollar. The nominal
interest rate between today and tomorrow is 10% in either economy. To avoid
tomorrow's exchange rate risk, a forward market opens today, in which the
forward exchange rate is 1.1 pounds per dollar. That is, you can nd your partner
today who wants to x tomorrow's exchange rate between you and her at 1.1
pounds per dollar regardless of the actual exchange rate tomorrow. To make
money, you should start by borrowing pounds, exchanging those for dollars and
investing in the U.S. market.
8. Suppose China's exchange rate is currently xed at 8.11 yuans per dollar.
A speculator, who believes that China will shortly revalue its currency to 8.00
yuans per dollar, would borrow yuans in China to exchange them for dollars,
wait until the currency is revalued, and repay the debt after exchanging dollars
back to yuans. (Disregard the interest the speculator needs to pay for debt.)
9. The exchange rate system of China is close to the xed exchange rate
system. Since there is a higher risk of currency crisis in this system, China should
6.2 Gains from International Trade (Comparative Advantage)
95
instead take the exible exchange rate system.
6.2
Gains from International Trade (Comparative

Advantage)
Question: Do Two Economies Gain from International Trade (Even Though

One Economy is More Productive in Producing All Goods)? (References:
Sections 15.1 and 15.2 of DLS.)
The material covered in this section is essential to understand gains from
international trade.
6.2.1
In our ctional world, there are two economies, U.S. and E.U., and there are two
types of goods, A and B. A representative consumer in each economy has an
identical utility function,
U (CA ; CB ) = log(CA ) + 2 log(CB );
(6.4)
where CA and CB are the units consumed for goods A and B. While each economy has 20 units of labor, the U.S. requires 4 units of labor to produce 1 unit of
either good A or good B. The E.U. requires 5 units of labor to produce 1 unit of
good A and 10 units of labor to produce 1 unit of good B. To summarize,
To produce one unit of:
U.S. labor required:
E.U. labor required:
Good A
4 units
5 units
Good B
4 units
10 units
Notice that the U.S. has absolute advantages for both goods: to produce
either good, the U.S. requires less units of labor inputs. Our question is what will
happen if two economies are allowed to trade. In particular, we are interested in
whether either or both of economies can be better off in terms of consumptions.
To jump to the conclusion, both economies become better off. But if the U.S. is
more productive in both goods, then why should the U.S. import anything from
the E.U.? Let us see how the U.S. gains.
1. Autarky: First, suppose there is no interaction between two economies.
Assume the U.S. provides LU S;A units of labor to produce good A. So LU S;B =
(20 LU S;A ) units of labor are provided to produce good B. Since 4 units of
labor produce each unit of goods A and B, such an allocation of labor inputs will
96
produce LU S;A =4 units and (20

That is,
LU S;A )=4 units of goods A and B, respectively.
CU S;A = LU S;A =4;
(6.5)
CU S;B = (20
(6.6)
LU S;A )=4:
From (6.4), the U.S. maximizes
log(LU S;A =4) + 2 log[(20
LU S;A )=4]
with respect to LU S;A , so the rst-order condition becomes
1=4
+2
LU S;A =4
(20
1=4
= 0;
LU S;A )=4
or equivalently,
1
LU S;A
20
2
;
LU S;A
so
LU S;A = 20=3:
This implies that the U.S. consumptions under autarky are, using (6.5) and (6.6),
CU S;A = 5=3;
CU S;B = 10=3:
The same procedure for the E.U. gives the following maximization problem:
log(LEU;A =5) + 2 log[(20
LEU;A )=10];
so the rst-order condition becomes
LEU;A = 20=3:
Hence, we have
CEU;A = 4=3;
CEU;B = 4=3:
2. Trade Allowed: Now suppose that the two economies are allowed to trade.
From the law of one price, denote by PA and PB the international prices, in dollars, for goods A and B. (We assume the dollar currency is used in international
97
trade.) Then, the U.S. can supply one unit of labor to produce 1/4 units of good
A, which is worth PA =4. The U.S. can also supply one unit of labor to produce
1/4 units of good B, which is worth PB =4. This implies that the U.S. will choose
to specialize in good A if
PA =4 > PB =4;
or equivalently,
PA
> 1:
PB
(6.7)
Otherwise (i.e., if PA =PB < 1, and let's disregard the equality for simplicity),
the U.S. will choose to produce only good B. Similarly, the E.U. can supply one
unit of labor to produce 1/5 units of good A, which is worth PA =5. The E.U.
can also produce 1/10 units of good B with one unit of labor, and that is worth
PB =10. So the E.U. will choose to specialize in good A if
PA =5 > PB =10;
or equivalently,
PA
1
> ;
PB
2
(6.8)
and will choose the other option otherwise (i.e., if PA =PB < 1=2). Then, we
have the following:
(1) If 0 < PA =PB < 1=2, that is, if the relative price of good A to good
B is too low, then neither economy will produce good A. This cannot be an
equilibrium because both economies produce the same goods so there will be no
trade at all. Both economies will go back to the autarky case.
(2) If PA =PB > 1, both economies will produce good A. Again, this cannot
be an equilibrium with trades.
(3) If 1=2 < PA =PB < 1, then the U.S. produces good B only and the
U.K. produces good A only. In this case, we say the U.S. has a comparative
advantage in good-B production. Similarly the E.U. has a comparative advantage in good-A production. Each economy specializes in a good in which it
has a comparative advantage. So in our discussion to nd an equilibrium with
international trade, assume 1=2 < PA =PB < 1, which is the case in which there
can be any international trade.
98
3. Equilibrium with International Trade: Let us consider the problem of

the U.S. more carefully. The U.S. specializes in good B, the U.S. has 20 units of
labor, and 4 units of U.S. labor makes 1 unit of good B, so the U.S. produces 5
units of good B. Since the price of good B is PB , the income of the U.S. is 5PB
in dollars. So the U.S. maximizes with respect to CU S;A and CU S;B ,
log(CU S;A ) + 2 log(CU S;B )

subject to
PA CU S;A + PB CU S;B = 5PB :
(6.9)
Equivalently, the U.S. maximizes with respect to CU S;B ,
log[(5
CU S;B )PB =PA ] + 2 log(CU S;B );
(5
PB =PA
2
+
= 0;
CU S;B )PB =PA CU S;B
or equivalently,
CU S;B = 10=3:
And the U.S. consumption of good A becomes from the budget constraint (6.9),
CU S;A = (5
CU S;B )
PB
5
=
PA
3
PB
:
PA
(6.10)
We do not know the value of PB =PA yet. This will be determined by a marketclearing condition. Similarly, the problem of the E.U. is to maximize
log(CEU;A ) + 2 log(CEU;B )
subject to
PA CEU;A + PB CEU;B = 4PA :

The unconstrained problem is to maximize
log(CEU;A ) + 2 log[(4
CEU;A )PA =PB ];
CEU;A = 4=3;
(6.11)
99
and hence,
8 PA
:
(6.12)
3 PB
Then how is the relative price PA =PB decided? The production of good A in the
world (i.e., U.S.+E.U.) should be the same as the consumption of good A in the
world, i.e., the market for good A needs to be cleared, so
CEU;B = (4
CEU;A )PA =PB =
CU S;A + CEU;A = 4;
(because only the E.U. produces 4 units of good A), and using (6.10) and (6.11),
5
3
PB 4
+ = 4;
PA 3
or equivalently,
PA
5
= :
(6.13)
PB
8
Recall that we are assuming that 1=2 < PA =PB < 1, which is satis ed here.
We also can check whether our calculation is correct using the market clearing
condition for good B, which is omitted here. (6.13) implies that (6.10) becomes
8
CU S;A = ;
3
and that (6.12) becomes
5
CEU;B = :
3
We summarize the result:
Consumption of:
U.S. under autarky
U.S. with trade
E.U. under autarky
E.U. with trade
6.2.2
Good A
5/3
8/3
4/3
4/3
Good B
10/3
10/3
4/3
5/3
Discussion
Both U.S. and E.U. consume more units of either good. De nitely, both economies
become better off.
In fact, this theory doesn't need a speci c form of the utility function, (6.10).
In the U.S. autarky problem, we can draw a budget line in the (CA ; CB ) plane,
100
which is a straight line between (5; 0) and (0; 5). (These two points are two extreme cases in which the U.S. specializes in good A or good B. The relative price
becomes PA =PB = 1.) The U.S. chooses one point to maximize its utility. [Draw
a gure.] We can draw a similar gure for the E.U. (A straight line between (4; 0)
and (0; 2). with PA =PB = 1=2.) [Draw a gure.]
We have seen that the two economies trade if 1=2 < PA =PB < 1. And the
U.S. will specialize in Good B when 1=2 < PA =PB < 1. In this case, the U.S.
has a new budget line: The U.S. produces 5 units of good B, and it may choose
to consume those 5 units of good B, which means (0; 5) is still on the budget
line. But now the slope is different. Before, the U.S. had PA =PB = 1, but now
it has 1=2 < PA =PB < 1. If you draw a graph with the new budget line, you
will see now the U.S. has a wider area for budget set. We can do similarly for the
E.U. This is the gain from international trade. Any reasonable utility functions
will yield gains to both economies. Further, our results hold with any assumed
units of labor required to produce each good.. The gain from trade that we have
discussed so far is pretty much general.
True or False?
1. Suppose there are two countries, the U.S. and Mexico. Assume that the
labor is the only input in the production of all goods. If the U.S. is more productive (i.e., requires less units of labor to produce one unit of good) in producing
any goods, then the U.S. has nothing to gain from the international trade with
Mexico.
2. Assume there are two countries and two goods. If the U.S. requires one
unit of labor to produce each unit of goods A and B, and if the E.U. requires four
units and two units of labor to produce each unit of goods A and B respectively,
then the U.S. has a comparative advantage in the production of good A, i.e., the
U.S. will specialize in good-A production if two countries trade.
3. Assume that there are two countries (the U.S. and Mexico) and two goods
(A and B). Both goods can be produced solely by labor input. In particular, the
U.S. can produce one unit of either good with one unit of labor, but Mexico can
produce one unit of A and B with two and three units of labor, respectively. If the
relative price of good A to good B is between 1 and 1.5, then two countries are
willing to trade.
6.3 International Trade and Economic Growth
6.3
101
Question: Does International Trade Affect Economic Growth?

ences: Various recent papers.)
(Refer-
If the theory of comparative advantage is correct, international trade is bene cial to all countries. The European Union in reality was, at least partly, built
under the motivation to maximize such gains from trade. Of course, our theory
is based on a simple model. There are a lot of things regarding gains/losses from
international trade that this model is missing, which is the subject of this section.
Studying the effects of international trade is also important in growth perspectives. While most of economies that were initially rich in 1960 grow around 2%
per year, the annual growth rates in 1960-2003 of initially poor economies vary
from 4% (which is negative!) in Liberia to 7% in Taiwan. Why are there several
miracle economies, including Taiwan, South Korea and Singapore? And why
are there disaster economies? While various factors (such as education, savings
rate, and institutional quality) are thought to matter, many economists believe that
openness to international trade is very important.
6.3.1
There are mixed effects, positive and negative, of international trade on economic
growth. A substantial number of economists seem to believe that positive effects
dominate at least in terms of economic growth. Here, we list some of theories on
gains/losses from international trade.
1. Negative?: Here are some losses from trade.
(1) Disadvantage in Learning By Doing: Rich economies tend to have advanced technologies while poor economies do not. So rich economies are more
likely to specialize in high-quality goods, for example, computer chips and satellites. Poor economies may continue to export agricultural products. Then, by
producing sophisticated products, rich economies will learn how to make them
more ef ciently and will develop more advanced technologies (which is known as
the learning-by-doing effect). And poor economies may still specialize in producing the same products in which there are not that much of learning-by-doing
effect. (That is, productivity improvements in agriculture are usually slower than
those in manufacturing.) This means poor economies have disadvantages in
improving productivities when they trade with rich economies.
(2) Transition Cost: Suppose you are a farmer in the U.S. All of a sudden, the
102
U.S. government decides to trade with a country that has a comparative advantage
on agricultural products. The U.S. consumers will consume imported products
instead of your crops because yours are more expensive. According to our model
of comparative advantages, labor inputs can be used to produce either product.
But in reality, it will take some time for you, a farmer, to transform to an engineer.
If it is costly to change the structure of an economy facing the new environment
of openness, growth rates may be lowered at least for a short run.
2. Positive?: Here are some gains from trade.
(1) Improvements from Competition: Opening an economy will provide a
more competitive environment. Suppose a car industry is under monopoly in a
low-income economy. We know that a monopolistic producer does not have a lot
of incentives to invest in quality improvements because there is no competition.
If this country opens itself, then this producer needs to compete with German
or Japanese producers, and this higher degree of competition will force her to
produce more ef ciently or to invest more to make better cars. This may help the
economy to grow faster. (Of course, due to this very competition, the domestic
producer might go out of business. This means that it does not have a comparative advantage in cars. Workers will be allocated into other industries which are
internationally more competitive.)
(2) Imports of Foreign Technologies: Through international trade, it becomes
possible to import technologies from developed economies. An introduction
of computers will substantially improve the productivities of many industries.
But for a poor and isolated economy to develop its own technologies to invent
computers may take several decades. It is cheaper to import computers or to
learn the technologies required to make computers from developed economies.
Needless to say, this will accelerate economic growth.
6.3.2 Some Terminology
Before we close this chapter, it will be useful to introduce more terminology.
The balance of payment is a summary statement of all international transactions
and capital ow of a country within a speci ed period. It consists of the current
account and the capital account, while the current account tracks transactions
of currently produced goods and services, including
- Exports of goods and services (+)
- Imports of goods and services (-)
(Upto here (i.e., exports minus imports in terms of domestic currency) is
6.3 International Trade and Economic Growth
103
called the trade balance, net exports, or trade surplus. The negative of this
is the trade de cit.)
- Net factor payment from abroad (+): This includes compensation for labor
or capital.
- Net transfer received (+): This includes development aids.
(Upto here is called the current-account balance.)
On the other hand, the capital account tracks
- Net increase in foreign assets held by home country (-)
- Net increase in home assets held by foreigners (+)
(The latter minus the former becomes the capital-account balance. A capital in ow (out ow) means a positive (negative) capital-account balance.)
The current-account balance plus the capital-account balance is close to zero
because most transactions enter each balance. Suppose the U.S. exports a good
and receives cash dollars. The export is a positive item in the current account,
but the fact that a foreign country paid dollars means a decrease in dollars (home
assets) held by foreigners, which is a negative term in the capital account.
It is also useful to introduce some equalities. Recall that
(6.14)
Y = C + I + G + N X;
where Y is the GDP, C is the consumption, I is the investment, G is the government spending, and N X is the net export (export minus import, trade balance,
or trade surplus). Basically, this equality means what an economy produces (Y )
is consumed (C ), invested (I ), spent by the government (G) or exported (N X ).
As is discussed above, the current-account balance (CB ) is
(6.15)
CB = N X + F P + T R;
where F P is the net factor payment from abroad and T R is the net transfer
received. National savings (S ) is de ned as
S = Y
P + T R}
| + F{z
national income
C
+ G}
| {z
national spending
= I + N X + F P + T R (by (6.14))
Then, (6.16) implies with (6.15) that
CB = S
I:
(6.16)
104
This is another way of interpreting the current account. The current-account

balance is the difference between the national savings and investments. If an
economy runs a de cit on the current account, this means the savings are not
enough relative to investment.
True or False?
1. A closed, low-income economy should immediately open itself to international trade in order to increase the growth rate.
2. As a producer produces more units of products, she will learn more on
how to produce them more ef ciently or how to make better products, etc. This
learning-by-doing effect will help a low-income economy to grow faster, especially after it becomes open to international trade with a high-income economy.
3. A de cit in the current account implies that the savings are higher than the
investment.
Chapter 7
Asset Pricing
This chapter discusses theoretical approaches on the determination of asset prices
in the market.
7.1
Fundamental Equation of Asset Pricing
Question: How Is the Price of An Asset Determined in a Macroeconomic

Model? (Reference: Chapter 1 of Cochrane (2005)'s textbook, Asset Pricing.)
Most of the economic models covered so far assumed a single interest rate,
treated as exogenous and constant. But the interest rate is endogenously determined in the market. This section shows how. Further, we observe different
returns from various types of investments in reality. What makes this difference?
In this section, we try to understand how asset prices are determined in the context of a macroeconomic model.
7.1.1
The (stochastic) endowment of a representative consumer at time t = 0; 1; 2; :::

is Yt units of consumption goods. And the consumer faces some uncertainty
about these endowments. (This uncertainty is introduced for the rst time in this
course.) So Yt 's are random variables, following some stochastic distributions.
The consumer is allowed to buy or sell some assets. There is no production,
and there is no storage technology. Trading assets is the only way to change the
consumption level at any given time.
An asset is modeled in the following way. The unit price of an asset is P0 (in
units of consumption goods) at time 0. If a consumer buys this asset at time 0, she
will be paid X1 units of consumption goods at time 1. We assume a one-period
asset, so the payoff are zero after time 1. This X1 is, of course, allowed to be
stochastic. It may follow some stochastic distribution such as normal or uniform.
Or it may be constant. The payoff may be high when the economy is good and
low when it is bad. Or, it may be the opposite. This stochastic distribution of X1
is known to the consumer at time 0. She decides how many units of this asset
she will buy at time 0. This amount of units is denoted by a. If a is negative,
105
106
Chapter 7 Asset Pricing
that means she sells this asset to somebody else, so she needs to pay X1 units of
consumption goods at time 1 to that person.
Now we solve for this a for our representative consumer. At time 0, she
maximizes her expected utility of consumption,
"1
#
X
t
E0
U (Ct ) ; 0 < < 1:
(7.1)
t=0
Here, E represents a mathematical expectation, where a subscript 0 emphasizes

that she is at time 0, so this expectation is based on all relevant information
available at time 0. The constraints of the consumer are
aP0 ;
(7.2)
C1 = Y1 + aX1 ;
C2 = Y2 ; :::
(7.3)
C0 = Y0
Again, notice that she is at time 0, so Y0 , the amount of endowment at time

0, is observed. And P0 , the unit price of an asset, is also known (although we
will solve for it later). In sum, the consumer's problem is to maximize (7.1),
by choosing how many units of assets she would like to hold (i.e., a), subject to
budget constraints, (7.2), (7.3), etc.
This is equivalent to solving the following unconstrained maximization problem:
max E0 U (Y0
a
U (Y2 ) +
E0 [U (Y2 )] +
aP0 ) + U (Y1 + aX1 ) +
U (Y3 ) + ::: ;
or equivalently,
max U (Y0
a
aP0 ) + E0 [U (Y1 + aX1 )] +
E0 [U (Y3 )] + :::;
because U (Y0 aP0 ), which is observed at time 0, can come out of E0 . Under
some regularity condition (that enable us to differentiate in the expection), the
rst-order condition is
P0 U 0 (Y0 aP0 ) + E0 [X1 U 0 (Y1 + aX1 )] = 0:

| {z }
| {z }
=C0
Equivalently,
P0 = E0
=C1
U 0 (C1 )
X1 :
U 0 (C0 )
(7.4)
7.1 Fundamental Equation of Asset Pricing
107
In asset pricing, this equation is known as the fundamental equation of asset

pricing or Euler equation. Through (7.4), we can price any given asset characterized by some stochastic nature of X1 , once we know C0 and C1 .
In discussing this fundamental equation, we often de ne a (stochastic) variable
U 0 (C1 )
;
(7.5)
m1 = 0
U (C0 )
which is called a stochastic discount factor, the pricing kernel, or marginal
rate of substitution in asset pricing. With this notation, we often write (7.4) as
P0 = E0 [m1 X1 ] :
(7.6)
Another way to write this equation is
1 = E0 m1
X1
:
P0
1
is the ratio of what you get in the next period to what you pay in this
Here, X
P0
period. That is, it is simply a gross return between periods 0 and 1. We have
denoted this by 1 + r1 . So we have
1 = E0 [m1 (1 + r1 )] :
(7.7)
(7.4), (7.6) and (7.7) are all equivalent equations.

Sometimes we need to go further than this. What is the equilibrium of this
model? The equilibrium is attained when the demand for this asset equals its
supply. So we are looking for a solution in which the net demand becomes zero.
In this model, we are considering one representative consumer, who is the only
consumer in this economy. The only way to clear the market is for her to choose
a = 0. If a = 0, we clearly have C0 = Y0 and C1 = Y1 from (7.2) and (7.3).
Then, (7.5) is replaced by
m1 =
U 0 (Y1 )
;
U 0 (Y0 )
so that the movement of GDP matters in determining the price of any asset.
7.1.2
Some Applications
This fundamental equation determines the price of any given asset once the property X1 is known. A well-known Black-Scholes option-pricing model is also
108
a special case of this fundamental equation (although we will not cover option
pricing here). Here, we particularly consider the pricing of some basic assets,
such as risk-free bonds and stocks.
1. Risk-Free Rate: Bonds (such as treasury bonds or corporate bonds)
promise to pay certain amounts of money (i.e., X1 in the above model) in the
future. As coupon bonds, bonds may provide coupon payments for several periods.
A risk-free bond is a special case of bonds, which pays a constant amount
of consumption goods as payoff at time 1. It should satisfy from (7.7) that
1 = (1 + r1f )E0 [m1 ]
(7.8)
since a (net) risk-free rate, r1f , is constant so that (1 + r1f ) can come out of the
expectation. This implies that the gross risk-free rate, 1 + r1f , is determined by
1 + r1f =
1
:
E0 [m1 ]
(7.9)
We can compute the value of the right-hand side with reasonable assumptions
on the utility function and the movement of consumption (or endowment) levels.
This becomes the model's prediction on the risk-free rate.
A natural question is how well the model works. In reality, there is nothing
like risk-free assets. Sometimes we use the real returns on U.S. government
securities as approximates for risk-free rates. In fact, empirical studies suggest
that the risk-free rate predicted by (7.9) is too high relative to the rates we observe
in U.S. government secutiry markets (which are about 1% per year). This failure
to predict the risk-free rate is called the risk-free rate puzzle.
The U.S. government securities have never defaulted, but corporate bonds
carry some risk of default. Risky bonds should provide higher expected returns
because otherwise all investors will choose to hold less risky assets. The difference between the returns on a risky asset and on a riskless asset (like treasury
bond) is called the risk premium. Bonds with the highest levels of risk premiums
are called junk bonds.
2. Stock Return: Let P0 be the unit price of a stock today. At time 1, this
stock pays a dividend D1 . The new price at time 1 is by P1 . (All these variables
are in units of consumption goods.) That is, you buy this stock paying P0 units
109
today and you will get P1 + D1 units tomorrow. Inserting these to (7.6) gives
P0 = E0 [m1 (D1 + P1 )] :
Or course, a stock return is de ned by
1 + r1s =
D1 + P 1
;
P0
and (7.7) is still valid when r1 is replaced by r1s . That is,
1 = E0 [m1 (1 + r1s )] :
(7.10)
Stock returns are on average higher than risk-free rates because they are risky.
It is often useful to consider an excess return of a stock, a difference between
stock return and risk-free return. Two equations, (7.8) and (7.10), are combined
to
0 = E0 [m1 (r1s r1f )]:

| {z }
excess return
Reasonable assumptions on the utility function and the movement of consumption (or endowment) level will provide rough estimates for excess returns. The
equity premium, average returns on stocks minus those on treasury bonds, has
been about 7-8% since World War II. This is higher than what this theory predicts,
which is called the equity-premium puzzle.
3. Asset Pricing with a Relationship between Consumptions and Payoffs:
Our fundamental equation provides an interesting implication on asset pricing in
terms of the covariance between consumptions and asset payoffs. Let us start
with (7.6). We know by de nition that
cov(X; Y ) = E[(X
E(X))(Y
E(Y ))] = E(XY )
E(X)E(Y ):
So
P0 = E0 [m1 X1 ] = E0 (m1 )E0 (X1 ) + cov0 (m1 ; X1 ):

Hence, using (7.8), we have
P0 =
E0 (X1 )
1 + r1f
+ cov0 (m1 ; X1 ):
(7.11)
The rst term in the right-hand side is the expectation of future payment discounted to today's value using the risk-free rate. As this expected value increases,
110
the price of an asset also rises other things being equal. Then what about the
second term, cov0 (m1 ; X1 )? We have
cov0 (m1 ; X1 ) = cov0

=
U 0 (C0 )
U 0 (C1 )
; X1
U 0 (C0 )
by (7.5)
cov0 (U 0 (C1 ); X1 )
since cov(AX; Y ) = Acov(X; Y ) when A is constant. So the second term

in (7.11) is proportional to the covariance of future marginal utility from consumption (U 0 (C1 )) and future payoff (X1 ). But the marginal utility is usually
assumed to decrease as consumption increases. Then, this second term becomes
bigger when the covariance between future consumption (C1 ) and future payoff
(X1 ) becomes smaller. That is, an asset which pays a higher output when the
consumer feels that the economy is bad (or consumption is low) is a good one
because you are more desperate at that time, so it should have a higher price!
In sum, the price of an asset is the sum of the two terms, as (7.11) shows:
discounted expected value of future payment, and (ii) cov0 (m1 ; X1 ),
(i)
which is ultimately about how future consumption and future payoff comoves.
An asset has a higher value, other things being equal, if it is expected to pay
more in the future and if it pays more when consumption level is lower.
E0 (X1 )
,
1+r1f
7.1.3
Some Concepts
We start with some concepts about interest rates that are useful to know.
1. Compounding: The same 10% interest rates may be eventually different
according to how to compound them. Suppose that you borrow [or lend] $100
with a xed annual nominal interest rate of 10%. This is what you have to pay
[receive] back in the next year:
(i) if you follow yearly compounding: $100
(1 + 0:1) = $110:00
(ii) if you follow bi-annual compounding: $100

(iii) if you follow monthly compounding: $100
(1 + 0:1=2)2
12
(1 + 0:1=12)
$110:25
$110:47
(iv) if you follow daily compounding: $100
(1 + 0:1=365)365 $110:52
(v) if you follow continuous compounding: $100 limn!1 (1+0:1=n)n =
$100 e0:1 $110:52 from ex = limn!1 (1 + x=n)n .
Continuous compounding provides a useful result called the rule of 70. Denote by R a continuously-compounded nominal interest rate. If you invest $1 for
111
one year, the above formula suggests that you will receive eR dollars in the next
year. If for n years, you will eventually receive (eR )n dollars. Suppose you are
interested in when your asset will become twice in value (i.e., $2 in this case).
You solve
eRn = 2;
so the solution is
Rn = log 2;
or equivalently,
70%
0:7
=
:
R
R
This means the doubling time of your investment when the interest rate is R is
70%
. For example, if the annual interest rate is 10% for example, it will double
R
roughly in 7 years.
2. Yield to Maturity: We are at year 0. Consider a hypothetical treasury
bond, paying $B1 in year 1 and $B2 in year 2, without any uncertainty. Suppose
the price is given by P0 . (Exercise: Obtain this price. That is, nd a counterpart to
(7.11) in this example.) Using the above method, the annual interest rate implied
by this asset should satisfy
n
P0
|{z}
today's price
B2
B1
+
1 + r (1 + r)2
|
{z
}
present value of future payments
We can similarly obtain the interest rate implied by any types of bonds. Such an
interest rate, obtained through this way, is called the yield to maturity. (This
is simply called the interest rate in many cases.) The term structure of interest rates refers to the relationships among (annual) interest rates on bonds with
different maturity lengths (e.g., 3 or 6 months, 1, 2, 3, 5, 10, or 30 years.) The
curve that shows these relationships (with maturity length on the horizontal axis
and annual interest rate on the vertical axis) is called the yield curve. The yield
curve is typically, but not always, upward-sloping.
3. Ef cient-Market Hypothesis: Our model in this section is related to
the ef cient-market hypothesis. Rational-expectations theory assumes that the
expectations of investors are identical to optimal forecasts using all available
information. The ef cient-market theory, as a special case, argues that stock
112
prices (that are determined based on expectations of investors) already re ect all
available relevant information. Under this hypothesis, a good strategy is to buy
index funds because you cannot make that much of money buy studying very
carefully about individual stocks anyway. According to Mishkin, the S&P 500
index outperformed more than two thirds of portfolios that were professionally
managed in 1970s-90s. Of course, this hypothesis is under severe attack by many
economists or practitioners. So-called January effects or momentum effects are
sometimes used as evidence against this hypothesis, while there are also on-going
efforts to explain these anomalies within the ef cient-market framework.
7.2
Contingent Claims (Lucas-Tree Model)
Question: How Is the Price of An Asset Determined in a Macroeconomic

Model? (Reference: Chapter 8 of Ljungqvist and Sargent's (2003) textbook, Recursive Macroeconomic Theory.)
This section introduces an alternative (and essentially identical) approach to
pricing an asset. The model is known as a Lucas-tree model.
7.2.1
Model Description
The model description follows.

1. Endowments: We assumed an economy with stochastic endowment in
the previous section. We follow the same setup, but we further specify how the
endowment level is associated with each state of the economy. For example,
there may be two states, rainy (denoted by s =rainy) and shiny (denoted by
s =shiny). The state is associated with endowment. [Draw a Lucas tree with
this setup. Shiny with a 2/3 probability.]
There are many consumers (unlike in the previous section in which there is
only one representative consumer), and agent i's stochastic endowment at time
t = 0; 1; 2; :::, denoted by yti , is given by
yti = y i (st );
(7.12)
where st is the state of the economy (rainy or shiny) at time t, which is common to
all consumers, and y i ( ) is a function which relates the current state to the current
endowment level for consumer i. (So consumers may have different endowments
even though the state of the economy is the same. Even though the state is shiny,
the outputs of two farmers may be different, for example, if they have different
7.2 Contingent Claims (Lucas-Tree Model)
113
lands.)
Notice that the endowment does not depend on the history. That is, only the
current state matters in (7.12). The state of yesterday, for example, does not
matter in our setup.
2. Consumptions: Consumer i's consumption at time t is denoted by cit (st ).
Here, st denotes the history upto time t, i.e., st
(s0 ; s1 ; :::; st ). Regarding
consumptions, the resource constraint suggests that at any given time t under
the history st , the aggregate endowment (which is the sum of all individual
endowments) should be the same as the aggregate consumption (which is the
sum of all individual consumptions). That is,
X
X
cit (st )
y i (st ):
(7.13)
i
3. History: Suppose we are at time 0, observing the initial state, s0 . (So we

have either s0 =rainy or s0 =shiny.) We denote the (conditional) probability
of time-1 state, s1 , by
prob(s1 js0 ):
For example, prob(s1 =rainyjs0 =shiny) means the probability that tomorrow's state is rainy conditional on today's state, shiny. Similarly, since the history
upto time 2 is denoted by s2
(s0 ; s1 ; s2 ), the conditional probability that we
2
will have s , given today's status s0 , is
prob(s2 js0 )
prob(s2 ; s1 js0 ):
For example, prob(s2 =shiny, s1 =rainyjs0 =shiny) means the probability that tomorrow's status is rainy and the day after tomorrow's status is
shiny, conditional on today's shiny status. It is known that by the properties
of conditional probability,
prob(s2 ; s1 js0 ) = prob(s2 js1 ; s0 )prob(s1 js0 ):

For simplicity, we assume a Markov chain, so that tomorrow's state depends on
today's state only (and not on the history upto yesterday). This means prob(s2 js1 ; s0 ) =
prob(s2 js1 ), so the above expression simpli es to
prob(s2 ; s1 js0 ) = prob(s2 js1 )prob(s1 js0 ):
114
We can generalize this. The probability of st conditional on s0 is denoted by
prob(st js0 ) = prob(st ; :::; s2 ; s1 js0 ):

Under our assumption of Markov chain, we also have
prob(st js0 ) = prob(st jst 1 ):::prob(s2 js1 )prob(s1 js0 ):

4. Arrow-Debreu Market: This is the essence of this model. There is an
Arrow-Debreu market at time 0, in which agents trade claims to consumption
goods for all possible histories upto all periods. Let qt0 (st ) be the price in units
of time-0 consumption goods, of a claim to one unit of consumption good at time
t, when the history upto time t is speci ed by st . That is, if you buy one unit of
this Arrow-Debreu claim today, then you will receive one unit of consumption
good at time t when st actually happens (and nothing otherwise). Similarly, if
you sell one unit of this claim today, you receive qt0 (st ) units of consumption
goods today, but you have to pay one unit of consumption good to the owner of
this claim when st actually occurred upto time t.
Then, the problem of consumer i is to maximize the expected discounted
utility of consumptions,
"1
#
X
t
E0
U (cit ) ; 0 < < 1;
t=0
(which is the same as (7.1)), which can be written as

1 X
X
t=0
st
U (cit (st ))prob(st js0 ); 0 <
(7.14)
< 1:
Recall that she cannot change her stochastic endowments. The only way to affect
the consumption level is through trading Arrow-Debreu claims. Then imagine
this. You sell all your future endowments at today's market, and then you zerobase all your future contingent consumptions. Then, your budget constraint becomes
1 X
X
t=0
st
qt0 (st )cit (st )

{z
value today of all future consumptions
1 X
X
t=0
st
qt0 (st )y i (st )

{z
value today of all future endowments
(7.15)

7.2.2
115
The problem is to maximize (7.14) by choosing fcit (st )g, subject to (7.15). The
Lagrangian function is
#
"1
1 X
1 X
X
XX
X
t
qt0 (st )cit (st ) :
qt0 (st )y i (st )
U (cit (st ))prob(st js0 )+ i
t=0
st
t=0
st
t=0
st
The rst-order condition with respect to cit (st ) is

t
U 0 (cit (st ))prob(st js0 ) =
i 0 t
qt (s );
(7.16)
for all i, t and st .

A competitive equilibrium of this Arrow-Debreu economy is:
(i) the solution fcit (st )g for all i, t and st (satisfying (7.15) and (7.16)) that is
feasible so that (7.13) is satis ed, and
(ii) the price system fqt0 (st )g for all t and st , such that given this price system, the solution fcit (st )g solves each consumer i's problem. We can solve for
fqt0 (st )g and fcit (st )g under this de nition. (Examples will be provided later.)
We discuss some implications.
1. Fundamental Equation: We can obtain the fundamental equation of asset
pricing using (7.16). A time-0 version of (7.16) is
U 0 (ci0 (s0 )) =
i 0
q0 (s0 );
since prob(s0 js0 ) = 1. (Recall that we are at time 0, so we know the time-0 state,
s0 .) But wait. What is q00 (s0 )? This is the price in units of time-0 consumption
goods for a claim to one unit of time-0 consumption good. So this should be
simply 1. This means that we now have
U 0 (ci0 (s0 )) =
(7.17)
On the other hand, a time-1 version of (7.16) is
U 0 (ci1 (s1 ))prob(s1 js0 ) =
i 0 1
q1 (s ):
Then, (7.17) and (7.18) give
U 0 (ci1 (s1 ))prob(s1 js0 )

=
U 0 (ci0 (s0 ))
i 0 1
q1 (s )
;
i
(7.18)
116
or equivalently,
q10 (s1 ) =
U 0 (ci1 (s1 ))prob(s1 js0 )

;
U 0 (ci0 (s0 ))
or equivalently,
U 0 (ci1 (s1 ))
s0 :
U 0 (ci0 (s0 ))
q10 (s1 ) = E
This is the same as (7.4).
2. Pricing Securities: Once we solve for fqt0 (st )g for all t and st , we can
price any given asset. Any assets can be broken into a set of many Arrow-Debreu
claims. For example, consider a risk-free security that pays one unit of consumption good at time 1 regardless of the state. This is the same as buying each unit
of all possible contingent claims on time-1 consumption goods. That is, the price
of this risk-free security is the sum of the prices of time-1 contingent claims:
X
q10 (s1 ):
s1
In general, a time-0 price of a security that pays d(st ) when time-t state is st is
1 X
X
t=0
7.2.3
qt0 (st )d(st ):
st
Examples
Discussions so far can be better understood under examples.

Example 1. Each consumer has identical preferences. In speci c, consumer
i maximizes the expected discounted log-utility function at time 0,
"1
#
X
t
E0
log(cit ) :
t=0
There are two states, good and bad. The initial state at time 0 is good.
There is no uncertainty in this economy: If today was good, tomorrow is bad.
If today is bad, tomorrow is good. (So it is good at time 0, 2, 4, ... and
bad at time 1, 3, 5, ...) Assume that when the state is good, each consumer is
endowed with 2 units of consumption goods. If bad, 1 unit.
117
There is no uncertainly, so (7.14) is replaced by

1
X
log(cit );
t=0
and (7.15) is replaced by

1
X
1
X
qt0 cit
|t=0{z }
value today of all future consumptions
qt0 yti
|t=0{z }
value today of all future endowments
where we know yt = 2 if t = 0; 2; 4; ::: and yt = 1 if t = 1; 3; 5; :::. Here,

qt0 is the price in units of time-0 consumption goods, of a claim to one unit of
consumption good at time t. (Here, the part (st ) is omitted because we do not
need it when there is no uncertainty.) The Lagrangian function is
!
1
1
1
X
X
X
t
qt0 cit ;
log(cit ) +
qt0 yti
t=0
t=0
t=0
so the rst-order condition with respect to cit is

t
1
= qt0 :
cit
We know
1
= i
ci0
at time 0, just as in (7.17). The above two equations suggest that
t
1
1
= i qt0 ;
i
ct
c0
or equivalently,
qt0 =
i
t c0
:
cit
(7.19)
This is the price of a claim. But we didn't solve for cit 's yet. In fact, at any given
time t, the aggregate consumption should equal the aggregate endowment (which
(7.13) shows). Hence,
X
X
yti ;
cit =
i
118
for all t, which replaces (7.13). But everyone is identical in this example, so we
ultimately have
cit = yti :
(7.20)
Then (7.19) becomes
qt0 =
i
t y0
:
yti
So the prices are
q00 = 1;
2
=2 ;
q10 =
1
2
q20 = 2 = 2 ;
2
2
0
q3 =
= 2 3;
1
42
0
q4 =
= 4 ; :::
2
So we have priced all claims in this economy.
Now suppose that there is a stock that pays one unit of consumption good in
every period from time 1, as a dividend. This is equivalent to buying each claim
for all t = 1; 2; ::: So its price is
q10 + q20 + q30 + ::: = 2 +
= 2 (1 +
2+
=
2
1
P1
+2
2
+ :::) +
+ :::
2
(1 +
+ :::)
= 1=(1
) when 0 < < 1 with = 2 .
Notice that (7.20) implies that everyone consumes what she is endowed with.
This means that nobody holds any claims at all. (The intuition is clear. Everyone
is identical. The only way to make the net demand zero is for everyone to hold
zero assets.)
by
t=0
Example 2. Everything is the same as before, but at any given time t, the state
is good or bad with 50% probability. The status of time t is independent of
the history upto t 1. The discussion from (7.14) to (7.16) still applies. Since
119
we have log utility, (7.16) is replaced by
cit (st )
prob(st js0 ) =
i 0 t
qt (s ):
We know
1
ci0 (s0 )
at time 0, just as in (7.17). The above two equations suggest that
qt0 (st ) =
i
t c0 (s0 )
prob(st js0 ):
cit (st )
(7.21)
Since everyone is identical,
cit (st ) = yti (st );

just as in (7.20). So (7.21) becomes
qt0 (st ) =
i
t yt (s0 )
prob(st js0 ):
yti (st )
Recall that time-0 status is good. So the prices are
q00 (s0 = good) = 1;

21
= ;
22
2
21
= bad, s0 = good) =
= ;
12
q10 (s1 = good, s0 = good) =
(7.22)
q10 (s1
(7.23)
q20 (s2 = good, s1 = good, s0 = good) =

:::
21
= ;
24
4
If we compare (7.22) and (7.23), we see the price of a claim is higher when
it is contingent on a bad state with lower endowment. Just as we discussed in
the previous section, this is because when the economy is bad, you are more
desperate to consume additional one unit of consumption good (because your
marginal utility is decreasing in units consumed).
Now suppose that there is a risk-free asset that pays one unit of consumption
120
good at time 1, regardless of the state. The price is
q10 (s1 = good, s0 = good)+q10 (s1 = bad, s0 = good) =
+ =
3
:
2
Similarly, we can price any given asset. Again, nobody holds any claims at all in
this example because everyone is identical.
What if the initial state at time 0 is bad instead of good? This is left as
an exericse. In this case, claim prices become lower. This is because the market
opens when the economy is bad. Since the endowment level today is lower, one
unit of consumption good today values more.
121
Exercises
1. (Fundamental Equation of Asset Pricing) The following statistical results
are provided based on 206 quarterly observations from 1952Q1 to 2003Q2:
i) average real net rate of return on 30-day treasury bill: 0.3% per quarter
ii) average real net growth rate of consumption: 0.5% per quarter (So average
real gross growth rate of consumption, i.e., average of Ct+1 =Ct , is 1.005 per
quarter.)
Let the representative consumer's endowment at time t = 0; 1; 2; ::: be Yt
units of consumption goods. The consumer faces some uncertainty about these
endowments, so Yt 's are random variables following some stochastic distributions. The representative consumer at time 0 maximizes her expected discounted
(log-)utility:
"1
#
X
t
E0
log(Ct ) ; 0 < < 1:
t=0

that this expectation is based on all relevant information available at time 0. (So
Y0 is observed when she makes this decision.) The constraints of the consumer
are
C0 = Y0
aP0 ;
C1 = Y1 + aX1 ;
Ct = Yt ; for t = 2; 3; 4; :::
where a is the number of units of a given asset held by this consumer between
time 0 and 1, P0 is a time-0 price of this asset (which is observed by this consumer
at time 0), and X1 is a time-1 stochastic payment from this asset.
(a) Set up the problem of the consumer, obtain the rst-order condition, and
obtain the fundamental equation of asset pricing in this version of a model. (Hint:
This model is identical to the one we covered in classes except that a speci c
utility function is assumed.)
(b) Consider a risk-free asset that pays constant number of units of consumption goods at time 1. (Of course, this constant payoff is known at time 0.) Find
an explicit expression for a net rate of return, r1f , for this asset.
122
(c) Suppose the consumption growth (C1 =C0 ) is constant and known at
time 0. Let us use the number reported above: C1 =C0 = 1:005. Then, what is
the prediction of this model on a risk-free rate? If the actual risk-free rate can be
proxied by the rate of return on 30-day government securities, does this model
correctly predict the actual level? Explain. Assume = 0:99.
2. (Comovement of Consumption Growth and Returns) Let the representative consumer's endowment at time t = 0; 1; 2; ::: be Yt units of consumption
goods. The consumer faces some uncertainty about these endowments, so Yt 's
are random variables following some stochastic distributions. The representative
consumer at time 0 maximizes her expected utility of consumption,
"1
#
X (Ct )1
t
E0
; 0 < < 1 and 0 < < 1:
1
t=0

Y0 is observed when she makes this decision.) The constraints of the consumer
are
C0 = Y0 aP0 ;
C1 = Y1 + aX1 ;
Ct = Yt ; for t = 2; 3; 4; :::
where a is the number of units of a given asset held by this consumer between
time 0 and 1, P0 is a time-0 price of this asset (which is observed by this consumer
at time 0), and X1 is a time-1 stochastic payment from this asset.
(a) Set up the problem of the consumer, obtain the rst-order condition, and
obtain the fundamental equation of asset pricing in this version of a model. (Hint:
This model is identical to the one we covered in classes except that a speci c form
of a utility function is assumed.)
e1 =
(b) Denote a gross rate of return of an asset between time 0 and 1 by R
1 + r1 where r1 is a net rate of return. Obtain the fundamental equation of asset

e1 . (That is, nd a version of 1 = E0 [something] where R
e is
pricing regarding R
inside the expectation.) Also, nd an explicit formula for a gross risk-free rate,
e1f .
R
e1 on
(c) Use your results in (a) and (b) to show that a gross rate of return R
123
any given asset satis es
"
e1 ) = R
e1f 1
E0 (R
(Hint: cov0 (X1 ; Y1 ) = E0 (X1 Y1 )

Acov0 (X1 ; Y1 ) for some constant A.)
cov0
C1
C0
e1
;R
!#
E0 (X1 )E0 (Y1 ) and cov0 (AX1 ; Y1 ) =
(d) What does the equation in (c) imply for the expected rate of return on an
e1 ))? That is, is this expected rate of return high or low when future
asset (E0 (R
e1 ) covary (i.e., move together)
consumption growth (C1 =C0 ) and future return (R
and do not covary, respectively? What is your interpretation/intuition behind this
nding?
3. (Capital Asset Pricing Model) Let the representative consumer's endowment at time t = 0; 1; 2; ::: be Yt units of consumption goods. The consumer
faces some uncertainty about these endowments, so Yt 's are random variables following some stochastic distributions. They are simply endowed, so the consumer
cannot do anything to change the nature of Yt 's. There is no storage technology.
If there is anything she can do, that is to buy/sell some assets. The representative
consumer at time 0 maximizes her expected (log-)utility of consumption,
"1
#
X
t
E0
log(Ct ) ; 0 < < 1:
t=0

Y0 is observed when she makes this decision.)
Your job is to obtain the price of a hypothetical security with the following
property: As long as an investor holds one unit of this in nitely-lived security, it
pays the units of consumption goods which are exactly the same as endowments.
If our representative consumer buys this security at time 0, this starts paying
payoffs from time 1 on. For example, if the endowment at time 1 turns out to be
5 units of consumption goods, the representative consumer will receive additional
5 units. If the investor continues to hold it at time 2, it will again give the payoffs
which are the same as the endowment at time 2, and so on. (This security is
called the wealth portfolio.)
Recall that in an equilibrium of this type of a model, there should be no excess
124
demand for any security, so we should have Ct = Yt for all t = 0; 1; 2; :::

(a) Suppose this representative consumer, at time 0, is choosing how many
units of this security to buy to maximize her expected (log-)utility of consumption
above. (She is at time 0, so Y0 is observed.) Set up the problem. Denote by
P0W the unit price of this security at time 0 (where the superscript W implies
wealth). Denote by a how many units she buys at time 0. (Hint: Try not to use
P1W in your answer. This will make (b) easier.)
(b) Obtain the rst-order condition. Assume that this economy is in an
equilibrium (a = 0) and obtain the price of this security at time 0. Your answer
should have P0 in the left-hand side. The right-hand side should be simpli ed.
P1 t
= 1=(1
) for 0 < < 1:) Does this price rise as time-0
(Hint:
t=0
endowment increases?
(c) Suppose an investor buys this security at time 0. At time 1, right after
she receives the payoff from this security, she sells it. Obtain the expected gross
return (at time 0) of this security (i.e., E0 1 + r1W where the superscript W
implies wealth) when the investor does it. Does this expected return increase
or decrease when the economy is expected to grow faster? Why? (Hint: You
know P0W from your answer to (b).)
(d) Recall that in general, we can use the following stochastic discount factor
to price any asset i that pays some payoff at time 1:
m1 =
U 0 (C1 )
:
U 0 (C0 )
Argue that assuming our log-utility environment, we can instead use
m1 =
1
:
1 + r1W
This model is called the capital asset pricing model (CAPM).
Unemployment Rates
(Year 2005)
Source: CIA
Unemployment Rates
(Year 2005 or the Closest)
Incidence of Long Term Unemployment (%)

(continued)
Source: ILO
(http://www.ilo.org/public/english/employment/strat/kilm/kilm10.htm)
Incidence of Long Term Unemployment (%)

(% of the Unemployed for more than 1 year out of the Unemployed)
(Year 2001 or the Closest)
Source: OECD
(http://ocde.p4.siteinternet.com/publications/doifiles/012005061T017.xls)
2002 DISPOSABLE INCOME* OF AVERAGE PRODUCTION WORKER

AS % OF GROSS PAY
(*: Gross earnings minus personal income tax and employees' social security
contributions plus family benefits)
Source: OECD, Benefits and Wages: OECD Indicators (2002)
Duration of Unemployment Insurance Payments (Months)
Source: OECD
2004 TOTAL GOVERNMENT EXPENDITURE

(% of GDP)
Source: OECD
2004 TOTAL GENERAL GOVERNMENT REVENUE

(% of GDP)
Source: OECD
(Figure downloaded at http://en.wikipedia.org/wiki/List_of_countries_by_public_debt)
Source: CIA
PUBLIC DEBT, 2005 (% of GDP)
2004 TOTAL GOVERNMENT DEFICITS

(% of GDP)
Source: OECD
Source: Casey Mulligans website
C, I, AND G, 1939-53 (% of GNP)
SOCIAL SECURITY TRANSFERS, 2004 (% of GDP)
German Hyperinflation, 1923-24

A woman in Germany feeds her tiled stove with money.
The money is worth less than firewood. [Source: wikipedia.com]
Source: Casey Mulligans website
GOVERNMENT PURCHASES AND SOCIAL SECURITY TRANSFERS,

1930-92 (% of GNP)
(Source: Lucas (1995), Figure 1. Taken from McCandless and Weber (1995))
LONG-RUN EFFECT OF MONEY GROWTH ON INFLATION

(110 countries, 30-year average of 1960-90)
500,000,000,000 (500 billion) Yugoslavia dinar banknote, 1993

The largest nominal value ever officially printed in Yugoslavia,
the final result of hyperinflation.
[Source: National Bank of Serbia (www.nbs.yu)]
(Source: Lucas (1995), Figure 2. Taken from McCandless and Weber (1995))
LONG-RUN EFFECT OF MONEY GROWTH ON REAL OUTPUT

(100 countries, 30-year average of 1960-90)
(Source: Benati (2005), Chart 4(d).)
U.S. M2 GROWTH AND GNP DEFLATOR INFLATION
ACROSS COUNTRIES (2003)
INFLATION RATE (S ) vs. NOMINAL INTEREST RATE ( R ),
(Source: Lucas (1995),

Figure 3.)
INFLATION AND
UNEMPLOYMENT
(U.S. Annual Obs.)
Pi / PUS
Column 3: E i / Ei 1
E i
Column 2: (A)=(B) implies
Column 1: $ Pi / Ei
: $ PJapan / E Japan
(B) PJapan ($1= E Japan )
(A) $ PUS
No Barriers?
Problem: One Consumption Good?
(BURGERNOMICS)
BIG MAC INDEX
U.S. TIME-SERIES OBS (1948-2000)
INFLATION RATE (S ) vs. NOMINAL INTEREST RATE ( R ),
Note: Annual Data.
(Source: Barro (2006))
GROWTH RATES AND RATES OF RETURN (1880-2004)
(Source: WDI)
Initial Level (1960) vs. Annual Growth Rate (1960-2003), 81 Economies
REAL PER-CAPITA INCOME:
U.S. RATE OF RETURN ON STOCKS (1952Q1-2003Q2)

(Real, 206 Quarterly Obs)
U.S. RATE OF RETURN ON 3-MONTH T-BILL (1952Q1-2003Q2)

(Real, 206 Quarterly Obs)

Lecture Note For Elements of Economic Analysis IV1

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

Lecture Note For Elements of Economic Analysis IV1

Uploaded by

Copyright:

Available Formats

Lecture Note for

Elements of Economic Analysis IV1

November 27, 2006

2 Fiscal Policy (I)

3 Fiscal Policy (II)

4 Monetary Policy (I)

5 Monetary Policy (II)

Fundamental Equation of Asset Pricing

8 Appendix: Tables and Figures

Question: Does a More Generous Unemployment Insurance Increases the

Chapter 1 Labor Market Policy

much shorter in the U.S. relative to many European countries.

Model Description and Solution

The generosity of unemployment compensation may means either more money

< 1. Similarly, if a low wage is offered

1.1 Unemployment Insurance

units of discounted value of utility.

(discounted expected utility of the next period):

Chapter 1 Labor Market Policy

We can solve this equation for Vreject . That is,

u(b) + pu(wh )=(1

Vreject > Vaccept ;

u(b) + pu(wh )=(1

Whether this inequality holds depends on the values of parameters (e.g., , p,

1.2 Minimum Wage, Labor Union and Firing Restriction

Minimum Wage, Labor Union and Firing Restriction

Chapter 1 Labor Market Policy

1.2 Minimum Wage, Labor Union and Firing Restriction

1,000 workers in this economy, a market clearing (quantity demanded equals

Chapter 1 Labor Market Policy

1.2 Minimum Wage, Labor Union and Firing Restriction

The rst-order conditions imply that

To make the algebra simple, let's assume r = 0. (r is close to 0 anyway.) The

We all wish to make workers happier. All policies or institutions we considered

Chapter 1 Labor Market Policy

acceptance, which may further increase the unemployment. By the minimum

1.2 Minimum Wage, Labor Union and Firing Restriction

(c) Assume = 0:9, p = 0:2, wh = 2, and wl = 1:3. Will this worker

Wreject = b + pwh =(1

Chapter 1 Labor Market Policy

(g) Suppose the government decreases the unemployment bene t to b = 0:1.

1.2 Minimum Wage, Labor Union and Firing Restriction

Chapter 1 Labor Market Policy

1.2 Minimum Wage, Labor Union and Firing Restriction

p)2 )wh =(1

Chapter 1 Labor Market Policy

Y (LS ; LU ) = (LS )2=3 (LU )1=3 ;

1.2 Minimum Wage, Labor Union and Firing Restriction

Question: How Do Lump-Sum Taxes Affect an Economy?

2.1 Lump-Sum Tax

same time. She has the following preferences:

where Ct is consumption at time t, U is a utility function (that is increasing,

Chapter 2 Fiscal Policy (I)

then K1 is determined, then the consumer again decides I1 , and so on.

Solution and Discussion

An easy way to solve this problem is to eliminate the constraints as many as we

(which eliminates It ,) or equivalently,

We can even make this problem simpler by eliminating Ct and transforming it

Now all we have to do is to obtain the rst-order conditions.

2.1 Lump-Sum Tax

Kt+1 . The rst-order condition is

] = 0; for all t = 0; 1; :::;

U 0 (Ct ) = U 0 (Ct+1 )[1