Financial Economics: Tutorials second half 1

Financial Economics
Problem Set 2
SOLUTIONS

Problem 1: Insurance: how much to buy

This is similar to what I have done in the lecture but now not with somebody choosing the
level of insurance instead of the excess again.

• An agent can decide how much insurance to purchase.

• The agent chooses p ∈ [0, D], which is the amount of insurance the agent wants to buy.

• D is the damage that occurs with probability π.

• p = 0 means no insurance bought, while p = D is full-insurance.

• Also suppose that for each Dollar insured, the insurer charges a premium of r.

• The wealth of the agent is denote c̃.

• Denote consumption in the loss state as cl and consumption in the no-loss state as cnl .

(a) Write down the expression for the consumption in the two states of the world, i.e. cnl and
cl as a function of how much insurance the agent buys.
cl = c̃ − D + p(1 − r) and cnl = c̃ − pr.

(b) Find the combinations of cnl and cl the agent can feasibly choose by choosing p. Hint: I am
looking for the budget line cl as a function of cnl .
c̃ 1−r
Solve cnl for p, substitute into cl and simplify! This yields cl = r
−D− c .
r nl

(c) Interpret the slope of the budget line. Draw the budget line in a cl -cnl diagram and comment
on how the price of insurance r impacts the slope.
You can trade off consumption in the two states. If you want one more Dollar of consump-
tion in the no accident state you have to give up 1−r r
Dollars in the accident state. Or
equivalently, if you want 1 − r Dollars more in the accident state you have to give up r
Dollars in the non-accident state.
Financial Economics: Tutorials second half 2

cl

c̃ − rD cf ull

c̃ − D cno
cnl
c̃ − rD c̃

Changes is r make the budget line pivot around cno . Lower r implies steeper budget line.

(d) Suppose we have a risk-averse agent. Describe how we can graphically represent such
preferences and how the graphic representation incorporates π and the degree of the agent’s
risk aversion.
We can represent the preferences using indifference curves representing all cl -cnl combina-
tions the agent is indifferent between. Indifference curves are downward sloping and convex
for risk-averse agents. Higher indifference curves mean that the person is better off. The
curvature of the indifference curve is determined by the risk preferences. The stronger the
curvature the higher the level of risk aversion. A straight line is risk neutral. The proba-
bility of an accident tilts the indifference curve around the certainty line, where the slope is
− 1−π
π
.

(e) Add some indifference curves of a risk-averse agent and show the optimal allocation. (As-
sume that π < r for your graph.) How can you read off your graph how much insurance
the agent is going to buy?

cl

c̃ − rD cf ull
c∗l c∗
(1 − r)p∗

c̃ − D cno
cnl
c∗nl c̃
Financial Economics: Tutorials second half 3

The graph shows everything, because of r > π we have an interior solution. The red
distance is proportional to the amount of insurance bought p∗ .

(f) Discuss the conditions for full insurance and buying no insurance at all to be optimal.
Full insurance: Indifference curve in the full insurance point being flatter than the budget
line, which requires r ≤ π.
No insurance: The indifference curve at the no insurance point′ need to be steeper than than
u (c̃)
the budget line. The condition is a bit clumsy 1−r
r
< 1−π
π u′ (c̃−D)
. The indifference-curve
interpretation is more important.
Financial Economics: Tutorials second half 4

Problem 2: Complete Markets for State-Contingent Claims

1. Explain what Complete Markets for State-Contingent Claims means.

If complete markets for state-contingent claims exist, then an individual can transfer as
much consumption across different stochastic states as she likes at a given price ratio. pi
is the nominal price for one unit consumption in state i. Then the agent can get one more
unit of consumption in state j for giving up pj /pk units in state k.

2. What are the requirements for a complete market of contingent claims to exist? Is this
likely to be true in reality?
We require n linearly independent financial instruments, where n is the number of states.
Hence, given the many states it is probably unlikely that complete markets exist. However,
the now very sophisticated system of financial products means that we are much closer to
this than in the past. Hence complete markets are a very useful benchmark model for us.
Market imperfections are more likely for less important states.

3. Sketch the condition that has to hold for consumption in any two states to be optimal for
an agent and interpret it.

π1 u′ (c∗1 ) π2 u′ (c∗2 ) πn u′ (c∗n )

= = ... =
p1 p2 pn
The expected marginal utility πi u′ (c∗i ) per dollar spent (i.e., divided by pi ) is equalized
across states, such that the expected marginal gain of well-being cannot be increased by
transferring consumption from one state to another by trading. This is the Fundamental
Theorem of Risk Bearing.

4. Draw the optimal choice for n = 2 with an arbitrary initial allocation c̃ and explain
indffernce curves and the budget constraint.

c2

c∗2
c∗

c˜2
c̃

c1
c∗1 c˜1
Financial Economics: Tutorials second half 5

Problem 3: Some violations of Expected Utility Theory

1. What are the underlying assumptions for optimal individual choices to look like if som-
body maximised expected utility?

(a) Complete preferences over lotteries: for any two lotteries can be compared (either
one is strictly preferred or there is indifference.
(b) Transitivity of preferences: If A is weakly preferred to B and B is weakly preferred
to C then A must be weakly preferred to C
(c) Independence Axiom: If A is weakly preferred to B then mixing A and B with
the same third lottery implies that the compount lottery containing A is weakly
preferred. (This basically implies that changes in probabilities are perceived the
same regardless of the probability being high or low.)

2. People’s choices often depend on irrelevant alternatives (like in the decoy effect). Explain
how this violates the assumptions underlying EUT.
If someone’s preferences between A and B change, dependent on whether an irrelevant
alternative is on offer or not, then preferences are not transitive, which violates rationality
of preferences (complete and transitive).

3. Provide some evidence from introspection and the lectures for people typically being
motivated by changes in wealth or consumption rather than by wealth or consumption
levels.
Here are lot can be said: Small losses like losing a 20-Dollar bill feel quite bad, while
they do not change lifetime wealth a lot; People are willing to take more risk if the same
situation is framed in terms of losses than if it is frames in terms of gains (lecture slides);
Rabin’s calibration theorem that shows that if utility is on absolute wealth then a tiny bit
of risk aversion in the small implies unrealistically high levels of risk aversion for large
gambles.

4. Explain why the often observed behaviour of the same person buying insurance and lot-
tery tickets at the same time (and for all wealth levels) is not consistent with expected
utility theory.
The expected win playing the lottery is smaller than the ticket price. Hence, a lottery
has a negative expected value. T be willing to accept playing it a person needs to be
risk-loving. Buying insurance, the premium is typically higher than the expected damage
(higher than the actuarially fair premium), which means that people are willing to give up
some expected payoff to eliminate risk. This implies that they must be risk-averse. People
cannot be risk-loving and risk-averse at the same time.

5. What is probability weighting, and how can it explain the lottery-insurance paradox.
Probability weighting takes into account that people typically over-weigh low probability
and under-weigh high probabilities. Probabilities are replaced by weights that differ from
the actual probability. In other words, psychologically low probabilities feel to make a
higher impact than they actually do, while high probabilities are felt to have a lower
Financial Economics: Tutorials second half 6

impact than they actually do. Over-weighing low probabilities mean that the likelihood of
winning the lottery feels bigger than it is, as well as the likelihood that your house burns
down. Hence, to gamble on the small probability in the case of a lottery and insuring
against the small probability of the house burning down become rationalizable.
Financial Economics: Tutorials second half 7

Problem 4: Discounting Utility

1. Suppose you have an income of $1000 per month and you are given the choice between
and additional
√ $225 today and $y in one year’s time. Your utility function over money
is u (x) = x. Supposing you are an exponential discounter with a discount factor δ
between years of 5/6, find the amount y that would make you indifferent between taking
the money now or in a year’s time.
√ √
Solve u(1000 + 225) = δu(1000 + y) for y, =⇒ 1225 = 5/6 1000 + y or y = 764

2. Suppose instead the choice is between an additional $225 today and $3900 in t years’
time. Find the lowest value of t such that you would prefer to take the $225 today.
We need to solve u(1225) = (5/6)t u(4900), which implies 35 = 70(5/6)t . Taking logs
on both sides and solving yields t ≈ 3.8 So you will take the money right now, if you have
to wait at least 4 years.

3. Without calculating, discuss how the magnitude of the discount factor δ influences your
answer to the two questions above.
A higher δ means that you are more patient. Hence, your are discounting future consump-
tion less, which implies that you are willing to accept a lower future amount in (1), and
you are willing to wait longer to receive the higher future payment in (2).

4. Again without calculating anything, comment on how your answers to (1) and (2) would
change if your base income increased.(Tricky!)
The crucial insight here is that additional income becomes less valuable the richer you
are, given that the utility function is concave. Hence, you need a higher additional income
to be willing to wait in (1) and are willing to wait less time in (2). The formal proof is
straight-forward if you know implicit differentiation, which I do not expect. So let’s not
do the proof.
Financial Economics: Tutorials second half 8

Problem 5: Time Consistency and Going to The Gym

Imaging the following situation. Suppose you have got the once-off chance to book a world-
famous personal trainer. For this you will have to pay some money today (t = 0). Then next
month (t = 1), if you have booked her, you can either train with her or not. Training will incur
an effort cost. Then the month after (t = 3) you will enjoy a health benefit if you booked the
trainer and trained the period before. Period utilities are:

time action 1 action 2

t = 0 u0 (book) = −1 u0 (notbook) = 0
t = 1 u1 (train) = −1 u1 (nottrain) = 0
t = 2 u2 (trained) = 3 u2 (nottrained) = 0

1. Denote the discount factor by δ and write down the intertemporal utilities U 0 for the three
possible course of action (book and train, book and not train, not book and not train)

U 0 (b + t) = −1 + δ(−1) + δ 2 3
U 0 (b + nt) = −1
U 0 (nb + nt) = 0
p
2. For which value of δ will you plan to buy and to train. Recall that x = −b/2 ± b2 /4 − c
gives the solutions to x2 + bx + c = 0
We require U 0 (b + t) to greater than the inter-temporal utilities for the other two course
of actions, which implies U 0 (b + t) > 0 or δ > 0.768.

3. In period 0, for which δ’s do you plan to work out in t = 1 in case you book the trainer.
Compare this to the critical delta required once you arrive in t = 1 to actually work out.
Comment.
For t = 0: Compare U 0 (b + t) and U 0 (b + nt) = −1, which yields δ(−1) + δ 2 > 0. For
period t = 1, U 1 (train) > U 1 (ntrain) = 0 =⇒ −1 + 3δ > 0. This is the exactly same
condition as for t = 0 as multiplying both sides by δ gives the latter. So in both cases
we have δ > 1/3. Hence, a person who finds working out optimal in t = 0 still does in
period t = 1. Exponential discounting implies time-consistency.

4. for which ranges of δ will you follow the three course of action: (a) not book and not
work out even if you had booked, (b) not book but work out if you had booked, (c) book
and work out, (d) book but do not work out.
(a) δ < 1/3; (b) 1/3 < δ < 0.768; (c) δ > 0.768; (d) never.
Financial Economics: Tutorials second half 9

Problem 6: What to do with a lottery win?

Suppose there are two periods t = 1 representing now and t = 2 is later. Your income is c̃
Dollars every period. Your mum gave you a lottery ticket for your birthday and you won 50, 000
Dollars. Now you have to decide how much of the win you are spending right now and how
much (denoted by x) to put in a term deposit that pays interest r and matures in period t = 2.
You have a concave period utility function u(c) and discount future utility with discount factor
δ.

1. Describe the factors that impact the optimal amount you will save. Comment on the
direction the factors will impact your saving decision.
patience δ : +; interest rate r : +; curvature of the utility function (smoothing motive): +;
continuous income ĉ : − by intuition but we will see that it depends on the relation
between r and δ.

2. Show the optimal saving decision in a graph with consumption in t = 1 on the x and
consumption in t = 2 on the y-axis. Label all points.

c2

c∗

c̃
c1
x∗

3. What is the slope of an indifference curve? What is the slope of the budget line. What do
the slopes actually tell you.
Budget line:

c2 = c̃ + (1 + r)x
x = c̃ + 50, 000 − c1
=⇒ c2 = c̃ + (1 + r)(c̃ + 50, 000 − c1 )

Slope: −(1 + r), The amount of consumption in period two you have to give up in or-
der to receive one more unit of consumption in period one. In other words, how much
consumption in period two you loose if you safe one less dollar in period one.
Financial Economics: Tutorials second half 10

Indifference curve: totally differentiate utility equalling a constant ū and solve for the
slope dc2 /dc1 :
u(c1 ) + δu(c2 ) = ū
u′ (c1 )dc1 + δu′ (c2 )dc2 = 0
dc2 1 u′ (c1 )
=− ′
dc1 δ u (c2 )
Most students will not be familiar with the total differential. Tell them that they do not
need to be able to derive this but that they need to know it. This is the amount of consump-
tion in period one you are willing to give up in order to get one more unit of consumption
4. Assume u(c) = ln c, δ = 0.8 and r = 0.25. Calculate the optimal saving amount x∗ .
This can be attempted by brute force or by smarts. The smart student realizes that here
we have 1 + r = 1/δ, which implies u′ (c1 ) = u′ (c2 ) and therefore full consumption
smoothing with x∗ = 22, 222. Otherwise,
max {log (c̃ + 50, 000 − x) + 0.8 log (c̃ + 1.25x)}
x
=⇒ x∗ = 2222.

5. Now suppose that instead r = 0.1. Recalculate x∗ and comment.

You will save less and the impatience dominates the lure of the interest rate, the such that
c1 > c2
max {log (c̃ + 50, 000 − x) + 0.8 log (c̃ + 1.1x)}
x
2c̃
=⇒ x∗ = 2222 − .
33
We also see that now a higher income reduces saving.
6. Finally, assume that the bank pays a massive interest rate of r = 0.4. Recalculate x∗ and
comment again.

max {log (c̃ + 50, 000 − x) + 0.8 log (c̃ + 1.4x)}

x
2c̃
=⇒ x∗ = 2222 + .
12
Now a higher income increases saving.
7. Give an intuition for why the impact of wealth on saving depends on the the sign of
(1 + r) − 1/δ.
For (1 + r) > 1/δ the lure of the interest rate for delaying gratification dominates the
impatience. Then having more income allows you to put more towards earning interest,
which you will do. If (1 + r) < 1/δ the effect of impatience dominates. Then having
higher income reduces the lost utility from having unequal consumption across periods,
which then allows you t bring more consumption forward and save less.
Financial Economics: Tutorials second half 11

8. Draw a graph showing indifference curves budget lines and optimal allocations for the
the cases above.

c2

c c1

x∗
Financial Economics: Tutorials second half 12

Problem 7: Saving and Borrowing in Imperfect markets

Suppose, we have a complete but imperfect market for shifting consumption between two
periods denoted by t = 1 and t = 2. The interest rates an agent faces differ for borrowing and
saving, with the interest rate for borrowing rb exceeding that for saving rs .

1. Draw the budget constraint for an agent with arbitrary endowment ĉ and explain where
saving and where borrowing occurs and what the corresponding slopes are.

c2

ĉ

c1

Borrowing is right of ĉ with slope −(1 + rb ) and saving is left of the endowment, with
slope −(1 + rs )

2. Add indifference curves for three different people with endowment ĉ leading to one per-
son saving, one borrowing and one doing neither. Comment on what determines if some-
body borrows, saves or stays put.

c2

c∗

c∗
ĉ

c∗

c1
Financial Economics: Tutorials second half 13

• Neither saving nor borrowing:

1 u′ (cˆ1 )
1 + rs ≤ ≤ 1 + rb
δ u′ (cˆ2 )

• Borrowing:
1 u′ (cˆ1 )
1 + rb <
δ u′ (cˆ2 )
• Saving:
1 u′ (cˆ1 )
1 + rs >
δ u′ (cˆ2 )

It is the relation of the slope of the indifference curve to the two interest rates. Both
influence factors, incentive to consumption smooth and patience play a role. The higher δ
is the more likely it is that we get a saver. The less equal the endowment incomes are, the
likelier it is that we get somebody who saves or borrows to smooth consumption. People
where none of these influence factors is particularly strong will get stuck in the kink, and
will neither save nor borrow.

3. Now suppose competition increases in the intermediary sector, and banks are now forced
to increase interest rates on deposits and decrease interest rates on loans. Further suppose
that they now apply one interest rate r to all transactions which is between rs and rb .
What happens to the three agents with respect to their behaviour and with respect to their
welfare.
First of all, nobody will be worse off, as all points on the new budget line are further up
and to the right (only the endowment point stays the same). Hence, everybody who has
been borrowing or saving will end up on higher indifference curve. Almost all people who
have not been saving or borrowing before will also be better off. The type of person that
will stay at the same well-being level is the one that even now does not save or borrow
money. The condition for that is

1 u′ (cˆ1 )
=1+r
δ u′ (cˆ2 )

The direction the borrowers and savers go is governed by two effects (substitution and
wealth effect). The substitution effect pushes savers to save more and borrowers to borrow
more. The wealth effect goes into the opposite direction. As in the whole chapter, for
reasonable preferences such as CRRA, the substitution effect dominates.

4. Draw a graph showing, where the three individuals end up.

Financial Economics: Tutorials second half 14

c2

c∗

c∗
ĉ

c∗

c1

5. What are the policy implications following from this analysis?

Talk about how a competitive banking industry is welfare enhancing, as it both yields
lower cost of debt and higher returns and also allows consumers to better smooth their
consumption (no consumer gets stuck at the endowment point, if she does not want to.)

6. A banker tells you that interest rates on loans are higher because there is a higher risk that
the borrower defaults than that his bank defaults on deposits. Therefore, the interest rate
spread between loans and deposits is actually not because of lack of competition. Does
this argument have merit? Discuss.
That is clearly a valid point, as then the expected break-even interest rate for loans is
higher than that for deposits. Our model does not take that into account as we are not
modeling in default risk in these simple models. In reality we would have to combine our
models for risk and uncertainty and for intertemporal choice and also model the banking
sector in more detail, if we wanted to incorporate that. Still there is clear evidence for
some lack of competition, as the banks do pay less interest on deposits than the RBA’s
cash rate, which is the cost of funds for the bank.
Financial Economics: Tutorials second half 15

Problem 8: Market Equilibrium

Suppose, there are two friends with very different jobs. Penny is a partner at a consulting
firm, while Garry is a public servant working with a government agency. There are two possible
states of the world, state n (for normal) is business as usual, while state c (crisis) represents a
bad shock for the economy. The likelihood of a bad shock occurring is πc = 0.1, which makes
πn = 0.9. Garry’s public servant job pays 1.5 regardless of the state realised (think in hundred
thousand dollars per year). Penny makes much more money in normal times, i.e. 4 in state n,
but her business will be badly hit in case of a crisis with income of 0.5 in state c. Both Penny
and Garry are risk averse with constant relative risk aversion (CRRA).

1. Observe that Garry does not face any risks. So he does not require any insurance. Does
it still make sense to meet with Penny and discuss potential state-contingent claim con-
tracts?
Yes, it does. Remember that people are willing to take a risk if the expected return is large
enough. Hence, he would be willing to provide some insurance to Penny against a crisis
occurring, if she is willing to compensate him with a sufficiently high payment in case a
crisis does not occur.

2. Draw one diagram for each actor depicting their endowment and the indifference curve
through the endowment point. Does a point with lower expected income than in the
endowment point but higher utility exist for Penny? What about Garry?

Penny Garry
cc cc

1.5 c

0.5 c
cn cn
4 1.5

Whenever a risk-averse person has unequal incomes in different states they are willing to
give up some expected value to reduce the spread (i.e. buy insurance at the actuarially
fair rate.) Hence, there are such points for Penny but not for Garry. The red lines depicts
consumption points with the same expected value as the endowment.

3. Draw an Edgeworth box with endowments and bargaining lens.

See below the full graph.

4. Show all points that might be outcomes of bargains and explain how a contingent contract
signed by both could look like.
Financial Economics: Tutorials second half 16

See the graph below. The black line in the lens is where the indifference curve of the two
would touch. Indifference curves are omitted for clarity of the picture.

5. Are these points efficient? Explain the concept of efficiency you are using and explain
why it is satisfied in the predicted outcomes or not.
Pareto Efficiency: No other feasible allocation exists where at least one person is better
off, without making anybody worse off. Yes all points on the contract curve are Pareto
optimal, since otherwise they would negotiate a different contract where they are better
off.

6. Explain what “decentralisation” of an allocation means. Moreover, show in a drawing

how decentralisation leading to a Pareto optimal outcome would look like in our example.
Explain!

cn
5 4 3 2 1 0
2.0 0.0
Garry

1.5 0.5

1.0 1.0
cc

cc

0.5 c 1.5
Penny

0.0 2.0
0 1 2 3 4 5
cn

The green line’s slope gives the price ratio for a decentralised equilibrium, which is
pareto optimal, the dot on the contract curve. The corresponding indifference curves are
omitted for clarity. See the lecture notes for cases where we also show the indifference
curves.

7. Under which condition will a decentralized market lead to the pareto efficient outcome
you drew above?
First Fundamental Welfare Theorem. The conditions are: complete markets, complete
information, price taking and no externalities.

8. Give two reasons why the First Fundamental Welfare Theorem might fail in our example.
Market power of firm charging higher than actuarilly fair premium or moral hazard.
Penny might shirk and herself increase the probability of the crisis outcome without the
insurer being able to observe that.