EC413 - EXAM - ANSWERS

1 Short questions

ANSWER ALL QUESTIONS. Each question carries 10 marks.

Question 1 [10 marks] Consider the following optimization problem:

max ln c1 + ln c2 + v(1 h1 ) + v(1 h2 )

c1 ;c2 ;h1 ;h2

s.t. (1)
c2 w2 h2
c1 + = w1 h1 + ;
1+r 1+r

where ct is the consumption in period t, ht is the fraction of available time worked in

period t, wt is the wage rate in period t, and r is the interest rate. The function v (1 ht )
describes the utility of leisure and satis…es standard regularity properties.

1. Suppose that we increase w1 and w2 by the same percentage; say we multiply both
with > 1. Show that the solutions for h1 and h2 are not a¤ected. (To simplify the
math you can set r = 0 and = 1). You do not have to provide any intuition.
Answer: The …rst-order conditions are the following:

1
= (2)
c1
1
= (3)
c2 1+r
@v(1 h1 )
+ w1 = 0 (4)
@h1
@v(1 h2 ) w2
+ = 0 (5)
@h2 1+r
c2 w2 h2
c1 + = w1 h1 + (6)
1+r 1+r

1
 Using r = 0 and = 1, we can rewrite this as

@v(1 h1 ) 1
+ w1 = 0 (7)
@h1 c1
@v(1 h2 ) 1
+ w2 = 0 (8)
@h2 c2
c1 = c2 (9)

c1 + c2 = w1 h1 + w2 h2 (10)

Using c1 = c2 , the life-time budget constraint can be written as

w1 w2
2= h1 + h2 (11)
c1 c2

Following a proportional increase in both w1 and w2 , then this system will remain
satis…ed if h1 and h2 remain the same and c1 and c2 increase proportionally. Why is
this the case? Obviously it is still the case that c1 = c2 . Moreover, the proportonal
increase in consumption imply that w1 =c1 and w2 =c2 remain the same. Then it
follows directly from the FOCs of labor that h1 and h2 remain the same.

2. Now suppose that only w1 increases and w2 remains unchanged. Compare the
changes in h1 and h2 with the changes found in part 1.
Answer: We know that we can decompose the increase in w1 into a substitution
and an income e¤ect. The substitution e¤ect of the increase in w1 is the same as
in the …rst part of this question. The income e¤ect will be smaller if the wage rate
only increases in the …rst period. Since the substitution e¤ect increases h1 and the
income e¤ec reduces h1 we know that the increase in h1 will be smaller when only
w1 increases. In terms of the equations, since it is still the case that c1 = c2 , we
know that c1 will not increase by as much if w1 only increases in period 1. Con-
sequently, w1 =c1 increases which means that h1 increases as long as the utility of
leisure increases as leisure decreases, which is the case for regular preferences.

Question 2 [10 marks] Small …rms and the share prices of small …rms typically su¤er
more when an economy enters a recession. Nevertheless, average stock returns are higher
for small …rms during recessions. Discuss possible reasons behind these two empirical

2
…ndings and discuss how can they can be reconciled.
Answer: During recessions, uncertainty increases and investors demand a higher com-
pensations for risky investment. So both the amount of risk and the price of risk increases.
This is especially true for small …rms. Why especially for small …rms? One reason is that
small …rms are more subject to information and moral hazard type frictions which tend
to intensify during recessions. Another possibility is that small …rms simply operate in
sectors that become riskier during recessions. These arguments imply that the required
rate of return has to increase during recessions, especially for small …rms. There is no
inconsistency between the two statements. In fact, they are related. The observed price
drops occur at the beginning of the recession (the …rst part of the statement) or when the
market realizes a recession is imminent. This is then followed by higher expected returns
during the recession (the second part of the statement). Above, we already gave reasons
why expected returns are higher during the recession. Now let’s turn to the question why
there is an initial drop in asset prices. There are two reasons why this is the case. First,
expected pro…ts are adjusted downwards when it becomes known that the economy enters
a recession. Second, from the earlier discussion we know that there is an increase in (the
price of) risk and an associated increase in the expected return. Common sense tells you
(hopefully) that the …rst has to lead to a reduction in the price. But so does the increase
in expected rate of return. After the additional price drop, prices can recover over time
at a higher expected rate which does indeed lead to a higher expected rate of return. So
it is the initial drop in asset prices (…rst part of the statement) that makes it possible to
have a higher average return during the recession (second part of the statement).

Question 3 [10 marks] Consider the standard OLG model in which there are two
generations in each period, the young and the old. There is no population growth. The
young solve the following optimization problem

3
 max ln(cy ) + ln(co+1 )
cy ;co+1 ;k+1

s.t.
cy + k+1 = ;
co+1 = k+1 + (1 ) k+1 +
where cy is consumption when young, co+1 is consumption when old, and k+1 is the amount
of capital carried over into old age.

1. Derive an expression which characterizes the solution for k+1 .

Answer:
1 k+1 1 + 1
= (12)
cy co+1
where cyt and cot+1 are given by the budget constraints above.

2. What do we mean with overaccumulation of capital?

Answer: A level of capital that is above the Golden Rule capital stock or alterna-
tively a level of the capital stock at which net returns are less than zero.

3. Is overaccumulation of capital possible in this model? You can focus on the steady
state. This question can be answered very fast with just a little bit of mathematics.
Answer: Imposing steady state values and combining the budget constraints with
the FOC gives

1 k 1+1
=
k k + (1 )k +
=)
1
k + (1 )k + = k +1 ( k + (1 ) k)

=)
1 1 + k + (1 )k + ( k + (1 ) k)
k +1 =

1 k + (1 )k + ( k + (1 ) k)
= 1+

> 1

4
 The LHS is the gross return on capital. There is no overaccumulation, since the gross
return on capital is bigger than 1. The intuition is the following. Overaccumulation
happens when young have lots of resources relative to what they have when old
and/or when you have a crummy savings technology. Neither is happening here. If
= 0, then we are back in the standard case. If > 0; then you get more when
young and when old. So that is not a reason to save. People save and invest in
capital because it gives them a good return. Remember that the marginal product
of capital is in…nite for the …rst unit invested.

Question 4 [10 marks]

1. Explain why the risk premium of a country could increase when this country’s pri-
mary de…cit turns from being a de…cit into a surplus.
Answer: If a county has a primary surplus, then it can …nance all government ex-
penditures (excluding those related to debt) by current revenues. This means that
it does not need the bond market anymore and is not a¤ected if investors do not
want to lend to this country after a default.

2. Explain why forward guidance may have a positive impact on the economy even if
it does not a commitment to a "lower-for-longer" type of forward guidance.
Answer: There are several possibilities. First, it could indicate to the private sector
that the central bank intends to follow a more expansionary policy than the market
is thinking. Second, by following unusual policies it could signal that the central
bank is willing to go out of its way to deal with the crisis.

3. What are outright monetary transactions (OMTs)?

Answer: These are purchases of the European Central Bank (ECB) of sovereign
assets in secondary markets. These are only possible if the country in question has
o¢ cially asked for assistance from the Eurozone bailout funds (EFSF/ESM). This is a
new policy initiative announced during the recent …nancial crisis. The announcement
led to an immediate decrease of the sovereign debt premia of Eurozone periphery

5
 countries. This is especially intriguing since so far there have been no OMTs. That
is, just announcing the possibility had an impact on markets.

Question 5 [10 marks] What are the main di¤erences between an agent-based business
cycle model and a DSGE type business cycle model. Be concise.
Answer: See slides

6
Long question

6. [50 marks] Consider an economy with the following characteristics.

General characteristics

All …rms are identical and can be modelled with a representative …rm.

Households are identical and can be modelled with a representative household.

The (representative) household consumes a "market" good, cm

t , and an endowment

good, cxt . These two goods are perfect substitutes.

The markets for labour, capital, and the endowment good are competitive mar-
kets without frictions. There is a search friction in the market for the "market"
consumption good.

all prices are in terms of the endowment good.

The "market" consumption good

There is a search friction in the market for the market consumption good. Speci…-
cally, the total amount traded depends on the search e¤ort put in by consumers, et ,
and the total amount put on the market by …rms. The latter is equal to yt . So we
get the amount sold (and consumed by households), cm
t is given by

1
cm
t = 0 et
1
yt 1
:

No relationships are formed in this search market. That is, every period, consumers
and producers face a search friction in …nding the right trading partner.

The number of goods acquired per a unit of search e¤ort, c, is given by

t

1
c et 1

t = 0 :
yt

7
 f
The fraction of produced goods sold, t, is given by

f et 1

t = 0 :
yt

Representative household problem

Each period, the (representative) household receives a …xed supply x of the endow-
ment good. These goods are suitable for consumption and for investment in capital.

Capital, kt , is owned by the household and rented out to …rms at rate rtk .

The household supplies labour, ht , on a competitive market at rate wt .

The budget constraint of the household is given by

pt cm x
t + ct + kt+1 rtk kt + wt ht + (1 ) kt + x; (13)

where pt is the price of the market good in terms of the endowment good. The
Lagrange multiplier of this constraint is denoted by 1;t .

The household faces an additional constraint, because it has to put in costly e¤ort
et to acquire consumption goods. The total amount of market consumption goods
is given by the following equation:

cm
t =
c
t et : (14)

The Lagrange multiplier of this constraint is denoted by c

2;t . t is taken as given by
the representative household, since each household is small relative to the size of the
market. The household does not have to put in any e¤ort to acquire capital goods
or (additional) units of the endowment good.

The household’s utility function is given by

(zt cm x 1
t + ct ) 1 h1+
t e1+
t
; (15)
1 1+ 1+

where zt is an exogenous random variable that determines the preference of the

market good relative to the endowment good.

8
 Representative …rm problem

The representative …rm has a standard production function that de…nes output yt
as
yt = kt ht1 (16)

and capital and workers are obtained on competitive spot markets.

f
The …rm only sells a fraction t of total output, yt . Thus, the …rm’s sales are given
f f
by t yt . The …rm takes t as given. Goods that are not sold fullly depreciate and
are not carried over into the next period as inventories.

Firms do not make any intertemporal decisions. We assume that …rms maximize
current period pro…ts. That is,

f
max pt t kt ht1 rtk kt wt ht : (17)
kt ;ht

The price at which the market consumption good is sold is given by

pt = zt with 0 < < 1: (18)

Exogenous random variable

The law of motion for zt is given by

2
ln (zt ) = ln (zt 1) + "t with "t N 0; : (19)

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Important general comments regarding exam questions and their answers:
When answering a question like this, it is very important that you

1. read the question carefully

2. take the time to understand this economy, the agents, their options, the physical
environment. This will save time in answering the questions.

9
 3. do not think that the properties of this model will be the same or similar to the ones
discussed in the lectures. That is, do not simply write down memorized answers.

Many students did not do this. The consequences are the following:

1. By not doing this, they lost a lot of points. For example, in this question, the
exogenous random variable is a preference disturbance not the usual productivity
disturbance of the models discussed in the lecture. So if the question asks what will
happen following a shock to this preference disturbance and you answer instead what
will happen in response to a productivity shock, then you are answering a completely
di¤erent question. Since the economy will respond quite di¤erently to a preference
shock than to a productivity shock, your answer will not be correct.

2. By not doing this, they did not realize that some questions are incredibly simple and
could be answered very quickly and very concisely.

Some examples are given below

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Questions

1. [5 marks] What are the state variables of this model and why?
Answer:

[5 marks] zt because it a¤ects preferences this period and because it has pre-
dictive power about future preferences.

[5 marks] kt because it a¤ects the wealth of the agent

Quite a few students wrote down that zt is a state variable because it directly a¤ects
resources of the household. These students basically wrote down which was the
correct answer for some of the homework problems in which the exogenous random
variable is a productivity shock. But there is no productivity disturbance in the
model. So this is not the right answer for this model.

Some other mistakes made by students

10
 Parameters are not state variables. Consequently, x is not a state variable.

Similarly, cxt is not a state variable. It is an endogenous variable. Just because

it is determined by a simple equilibrium condition, namely cxt = x does not
mean it is no longer an endogenous variable.

If an agent in the model takes something as given, say prices or matching

probabilities, then these variables do not become state variables. State variables
must be either exogenous random variables or predetermined.

Although, one could do this in some cases, it is better to not think of di¤erent
set of state variables for di¤erent agents. It is better to write down the complete
set of state variables. This complete set should be su¢ cient to determine all
the period-t outcomes for the endogenous variables.

2. [5 marks] What are the …rst-order conditions for the …rm?

Answer:

f 1 1
pt t kt ht = rtk (20)
f
(1 ) pt t kt ht = wt (21)

Comments:

!!! If agents takes a variable like a matching probability as given, then this
means that they cannot a¤ect it with their choices (typically, the justi…cation
is that each agent is small relative to the aggregate). Some students made
this question much more di¢ cult by using the expression for the matching
probability (which through yt depends on capital and labor) and then came up
with a …rst-order condition that contained changes in this matching probability.
This is wrong if agents take the matching probability as given.

If a question asks for the …rst-order condition, then there is no reason to do a

bunch of algebraic manipulations to get it in a particular form, say as a solution
for capital and labor. Note that it is not wrong to substitute in the expression
for the matching probability after you have obtained the FOCs. But there is

11
 no point in doing so. You only waist time and end up with complex looking
equations.

3. [5 marks] Write down the optimization problem of the household using the Bellman
equation. Denote the maximized value with the function v ( ) :
Answer:The utility function is monotonically increasing in consumption. Conse-
quently, the budget constraint will always be binding. If you do not make this
point explicitly in your answer, then your FOCs should keep the original inequality
speci…cation and allow for the possibility that the constraint is not binding.

v (zt ; kt )

=
(zt cm x 1
t + ct ) 1 h1+
t e1+
t
max + Et [v (zt+1 ; kt+1 )]
cm x
t ;ct ;kt ;et ;ht 1 1+ 1+
s.t.

pt cm x
t + ct + kt+1 rtk kt + wt ht + (1 ) kt + x

cm
t =
c
t et

or

v (zt ; kt )

=
(zt cm x 1
t + ct ) 1 h1+
t e1+
t
max min + Et [v (zt+1 ; kt+1 )]
cm x
t ;ct ;kt ;et ;ht 1;t ; 2;t 1 1+ 1+
k
+ 1;t (rt kt + wt ht + (1 ) kt + x pt cm
t cxt kt+1 )
c
+ 2;t ( t et cm
t )

s.t.

1;t 0; 2;t 0

You only have to give one of these two formulations. The second one is better, be-
cause it includes the info of the …rst representation and makes clear how to implement
the constraints.

12
4. [5 marks] Give the …rst-order conditions associated with the Bellman equation of the
household.
Answer:

(zt cm x
t + ct ) = 1;t (22)

zt (zt cm x
t + ct ) = pt 1;t + 2;t (23)
c
et = t 2;t (24)

pt cm x k
t + ct + kt+1 = rt kt + wt ht + (1 ) kt + x (25)

cm
t = c
t et (26)

ht = wt 1;t (27)
@v (zt+1 ; kt+1 )
1;t = Et (28)
@kt+1

Here it is assumed that the budget constraint is binding because preferences are such
that agents always prefer more consumption. Note that the …rst-order conditions are
part of the set of …rst-order conditions and these should be included in your answer.

5. [5 marks] Use the envelope condition to obtain a set of …rst-order conditions that
does not contain v ( ).
Answer:

@v (zt ; kt )
= 1;t rtk + 1 (29)
@kt
=) (30)
h i
k
1;t = Et 1;t+1 rt+1 +1 (31)

plus equations (23) through (28).

6. [5 marks] Discuss how the price setting rule for pt relates to Nash bargaining? Hint:
Ask yourself what the net-gain of an agreement is for the consumer and for the …rm.
Answer: Nash bargaining is about the splitting the surplus across two agents.
The key ingredients are the outside options of the two agents, the joint surplus,
and bargaining power. To see how the speci…ed price setting rule relates to Nash

13
bargaining we have to investigate these four ingredients. Just going through this
process would have already given you a lot of ponts.
A question like this becomes much easier if you have taken the time to understand
the actual economy and understand the transactions and the timing. The key aspect
of this economy that there is a friction in consumers and …rms …nding each other.
That takes e¤ort. So when they meet and any bargaining takes place the e¤ort
choice is sunk cost. Moreover, they also cannot bargain about how much the …rm
should produce. If that was the case, then …rms would not have unsold products.
So the bargaining is only about the price. The situation in this model is the same
as the one you typically face in practice. A buyer …nds a seller who has (an already
produced) good for sale.
The outside option for the …rm if it doesn’t reach an agreement about selling a
particular product is zero, since (as stated in the description of the economy) an
unsold good fully depreciates. So the …rm’s net gain is pt 0 = pt . What is the
outside option for the consumer? If the consumer doesn’t buy the market good,
then she can use her budget on the good that is not a¤ected by the search friction.
The question is particularly easy because the two consumption goods are perfect
substitutes. So the consumer’s net gain is zt pt : Note that the joint surplus is z.
If no agreement is reached then a good worth z is lost, if an agreement is reached
the consumer gets something worth z. The di¤erence between these two outcomes
is z. Finally, since relationships do not persist, this is a static bargaining problem.
You could directly apply the Nash bargaining outcome or derive it. Under Nash
bargaining the price would be solved from

max pt (zt p t )1 ; (32)

pt

or equivalently (33)

max ln pt + (1 ) (zt pt ) (34)

pt

where is the bargaining weight of the …rm. The FOC is the following:
1
: (35)
pt zt pt

14
 Rewriting gives
pt = zt ; (36)

so the price rule can be derived using Nash bargaining.

7. [5 marks] Consider a one-time increase in "t when = 0: How does this a¤ect the
cost of investing in capital? And how does this a¤ect the household’s choice of kt+1 ?
Hint: It is unlikely that manipulation of equations will provide insights. Use eco-
nomic mechanisms and possibly indicate which part of the model/equations capture
these aspects.
Answer: Many students wrote down paragraphs with interesting economic reason-
ing. Unfortunately, it was often about a di¤erent question, because zt was treated
as a productivity shock, which it is not in this question. More importantly, this is a
very simple question and doesn’t need paragraphs. The question is about “the cost
of investing in capital." So the …rst thing you should ask yourself is the following:
who is investing and what is the price paid? In this economy, it is the household
that invests. Note that investing means paying for something at a certain time and
getting something in return in the future. That is done by the household. The …rm
is simply renting capital in this environment.

So what does the household pay for capital. Well, the price of a unit of capital in
terms of the endowment good is equal to 1. So the …rst part of the answer is that
the cost of investing in capital in terms of the price in units of the endowment good
doesn’t change. Since this is an advanced course, you understand that there could be
a more sensible de…nition of “cost." That is, what is the “burden" for the household?
We know that the right way to measure this burden is the Lagrange multiplier of
the budget constraint times the price. The latter is 1 so we simply have to …gure
out how the Lagrange multiplier, 1;t , responds to an increase in zt . There are two
ways to get the answer. First, you can look at the FOCs. The expression for t is
equal to

15
 (zt cm x
t + ct ) = 1;t (37)

This expression tells you immediately that the direct e¤ect of an increase in zt is
a reduction in 1;t , that is, a reduction in the cost of capital. Note that the direct
e¤ect could be dampened somewhat if cm
t would decrease in response to an increase

in zt .1 Direct e¤ects often dominate. So 1;t increases. As long as the reduction in

1;t is stronger than a reduction in 1;t+1 , then this should lead to an increase in
capital according to the FOC of capital.2

You can also …gure out the right answer without the math. Ask yourself whether
the household is going to feel rich or poor when zt increases. If zt doubles, then this
is like get two goods for the price of one. If zt increases, then consumers can get the
same utility out of the market good at a lower level of cm
t . This is like a relaxation

of the budget constraint. This implies a reduction in the Lagrange multiplier,which

in turn implies that the cost in terms of utility of buying other things, like future
consumption (through an increase in capital), is getting smaller.

Please note that the actual answer to this question is very short. It could have been
the following.

Model answer: The price of a unit of capital in terms of the endowment good is
equal to 1. Thus, the cost of investing in capital in terms of the price in units of the
endowment good doesn’t change. However, in terms of utility the cost would be the
price times the Lagrange multiplier.The expression for t (see FOCs) indicates that
the direct e¤ect of an increase in zt is a reduction in t, that is, a reduction in the
cost of capital. This should lead to an increase in capital (according to the FOC of
capital). Intuitively, an increase in zt is like a relaxation of the budget constraint,
because the consumers can get the same utility out of the market good at a lower
1
It is possible that the consumer might reduce cm m
t since what matters is zt ct
2
Since zt is an AR(1), the biggest impact will occur in the …rst period. This is likely to carry over to
other variables such as 1;t . But the decrease in 1;t must be followed by an increase at some point, since
variables return to their steady state value.

16
 level of cm
t . A relaxation of the budget constraint means that the cost of investing

is smaller.
Price e¤ect: Quite a few students thought that an increase in zt is bad because it
raises the price. It is true that the seller “grabs" some of the increased valuation of
the good for its customer in terms of charging a higher price. But as long as <1
and zt pt increases, then the consumer is still better o¤. That is, with our setup
which resembles bargaining both buyer and seller share in the increased valuation.

8. [10 marks] In this economy, the price level varies with zt . Now consider the case in
which prices are sticky. Explain why …rms’output levels may change in response to
an increase in zt even if prices are sticky.
Important observation: The key di¤erence between the economy described here
and the matching model described in the lecture is that here the matching friction
is in the goods market instead of the matching model. So it makes sense that this
is going to matter in some of the subquestions. At least, you should consider this
possibility.

Answer - general comment I: Assuming that prices are sticky does not turn this
model into the New Keynesian model (with monopolistic competition, etc.). Recall
that with a positive surplus prices can be sticky as long as prices are not such agents
would prefer not to trade. In this environment this means that 0 < pt < zt .

Answer - general comment II: zt is still a preference disturbance and not a

productivity disturbance. When you answer the question as if zt is a productivity
disturbance, then you are not answering this question.

Answer: So consider an increase in zt , that is, consumers like the market good more.
How would that a¤ect …rms? In answering a question like this you want to start –if
possible –with direct e¤ects. If households like a good more, then this typically leads
to an increase in the price level. That cannot happen in this subquestion because
prices are assumed to be sticky. Now before you start looking at other possible
e¤ects, you should ask whether this increased demand can have a signi…cant direct

17
 f
e¤ect on …rms. Well, in this economy what matters is not just pt but pt t. In
fact, it is only the product that matters so the …rm only cares about the product of
these two terms and doesn’t care whether this product increases because the price
f
increases or the sell probability increases. So what happens with t? The value of
f
t will change if the household’s e¤ort level will change. If you would intuitively
argue that e¤ort in obtaining the market good should go up if household like the
market good better than you are already quite smart. But this is a 10 point question
and to get full points you have to use a bit of math. The …rst key equation is, of
course, the FOC for e¤ort:
c
et = t 2;t : (38)

An increase in 2;t (i.e. an increase in the RHS) would lead to an increase in e¤ort
(an increase in the LHS).3 What will happen with 2;t ? Combining the FOCs for
the two consumption goods, gives us a simple expression for 2;t ,namely

(zt p) (zt cm x
t + ct ) = 2;t (39)

It shows that there are two e¤ects because zt shows up twice. The increase in (zt p)
changes the relative value of cm x
t relative to ct . This increases the value of 2:t and,
thus, e¤ort. That is, …rms get compensated for having sticky prices by consumers
putting in more e¤ort. This e¤ect is larger when prices are sticky. But the increase
in zt also leads to a reduction in the overall value of consumption. This is captured
by the (zt cm x
t + ct ) term. This terms is there when prices are sticky and when
they are not. So there are o¤setting e¤ects on e¤ort and, thus, on the matching
probability of the …rm.4
Other e¤ects: The change in zt may also a¤ect labor demand and investment,
which in turn a¤ects the wage rate and the rental rate of capital. These are obviously
important for the …rm as well.
3
This increase in e¤ort will be dampened by the reduction in the matching probability caused by the
increase in e¤ort.
4
The increase in zt pt is like a change in the relative price and related to the substitution e¤ect. The
change in the marginal utility is related to the income e¤ect.

18
 9. [5 marks] Is the household’s choice for the level of search e¤ort e¢ cient from a social
welfare point of view?

Answer: The issues are similar to the e¢ ciency issues in the matching model discussed
in the lectures. One di¤erence is that in this model both sides endogenously choose their
activity in the matching market. The consumers by their e¤ort choice and …rms by their
productivity level choice. By contrast, in the model discussed in the lecture only vacancies
are endogenously chosen, whereas all unemployed workers will search for a job. The issues
for the household are the following.

Households take the matching probability as given, but it depends on households’

e¤ort level. Speci…cally, an increase in households’ search e¤ort lowers the overall
probability of a successful search. The social planner would take this change in the
matching probability into account. This aspect would imply that search e¤ort in the
competitive equilibrium is high relative to the level that is socially e¢ cient.

When choosing e¤ort levels, households ignore that a successful match does not only
bring them bene…ts but also to …rms. Speci…cally, a match means that a good will
not be destroyed. This aspect would imply that search e¤ort in the competitive
equilibrium is low relative to the level that is socially e¢ cient.

These two aspects could cancel each other out. However, for e¢ ciency evaluation
it is not su¢ cient to only look at the household. The same two aspects are present
for …rms. With their production level choice, they also a¤ect activity levels in the
matching market.

19
=====
Additional general comments

=====

ALL endogenous variables are functions of state variables. If a variable is a simple

function of some state variables, say Pt = zt , then Pt does not become a state
variable.

There are no separate state variables for households and …rm.

If an agent takes a variable as a given, then it does not necessarily become a state
variable. Only predetermined and exogenous variables can be state variables.

Being predetermined or exogenous is necessary but not su¢ cient for a variable to
be a state variable. kt 20 is predetermined but not a state variable. The rain fall in
France last year is predetermined but typically not a state variable. For a variable
to be a state variable you must be able to give a reason why it will a¤ect model
outcomes.

If a question asks to give …rst-order conditions, then do not manipulate them to get
them in a particular form. You can only loose points by making mistakes.

Similarly, if the question asks for the Bellman equation you do not have to give and
the equation with constraints listed and the associated Lagrangian. Give the best
representation, which would be the Bellman equation.

Similarly, do not waste time by giving obvious comments such as “this is the Bellman
equation of the household” when the question asks you to write down the Bellman
equation of the household.

Manipulation of equations sometimes but often do not give you intuitive insights.
Start with key known economic mechanisms and check whether equations are con-
sistent with your reasoning.

20
If you have one equation with multiple endogenous variables, then you cannot say
anything on what will happen to individual variables, E.g., suppose the model says

xt + ct + kt 2 + ct+1 = A:

If A increases, then you cannot conclude anything about changes in individual vari-
ables based on this equation. So again, start with key known economic mechanisms
and check whether equations are consistent with your reasoning.

Whenever an agent’s utility function is given, then you have been told how to dis-
count future revenue streams for this agent.

The question whether the correct discount factor is or 0 =U 0 is not relevant

Ut+1 t

for the Bellman equation. The Bellman equation has utility terms on the LHS and
RHS.

Number equations so you can refer to them without rewriting them.

Constraints are part of the FOCs of a constrained optimization problem.

If a constraint is an inequality constraint, then you (i) either have to use the two-part
Khun-Tucker condition or (ii) explicitly state a reason/assumption why you would
turn an inequality into an equality (such as non-sanitation which would imply that
budget constraints are binding).

Use common sense. The beauty of economics is that common sense can give you
good ideas on how get initial ideas on solving the model.

Ask yourself how a change in a parameter or exogenous variable …rst a¤ects the
system you are analysing. That is, what are the direct and indirect e¤ects. Direct
e¤ects often dominate.

If a question ask you to compare an object with a broken, red, car. You should
discuss whether the object is (i) broken, (ii) red, and (iii) a car. So when a question
asks to compare a price setting rule with Nash bargaining you should discuss the
four elements of Nash bargaining.

21