The Role of Financial Intermediaries

Afrasiab Mirza

Department of Economics
University of Birmingham

November 7, 2024

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Background I

Welfare Economics:
▶ systematic method of evaluating the economic implications
of alternative resource allocations
▶ welfare analysis answers the following questions:
▶ Is a given resource allocation efficient? pareto optamility
▶ Who gains and who loses under various allocations and by
how much?

Competitive economy:
▶ an economy which consists of many small economic units -
each with no market power
▶ conditions for a competitive economy:
▶ many buyers and sellers
▶ perfect information
▶ traded good is homogenous
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Background II

Pareto improvement:
▶ A reallocation of resources such that some individuals are
made better off while no individuals are made worse off

Pareto efficiency:
▶ We say we have satisfied Pareto efficiency when there are
no opportunities for Pareto improvements

Welfare Theorem I If all agents/traders have monotonic selfish

utility functions, then any competitive equilibrium is Pareto
optimal. more is better then competitive behavior pareto optimal
Welfare Theorem II Any Pareto optimal outcome can be
achieved through a competitive market with some prices if
money can be freely transferred between agents.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Financial Intermediaries: Definition

An economic agent who specializes in the activities of buying

and selling (at the same time) financial claims.
▶ Similar to a retailer that buys goods/services from
producers and sells them to customers:
▶ they buy the securities issued by borrowers (i.e. they make
loans)
▶ and they sell them to lenders (i.e. they collect deposits)
▶ Exist because of frictions in transaction technologies.
▶ Banks, brokers and dealers are examples of intermediaries.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Financial Intermediaries: Specialness

However, financial intermediares (e.g. banks) are more

complicated:
1. They deal with financial contracts (loans and deposits)
that may not be easily resold.
2. The characteristics of contracts or securities issued by firms
(borrowers) are different from the ones desired by investors
(depositors).
▶ Differences may arise in maturity, size and risk and so banks
are there to transform financial contracts and securities

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Financial Intermediaries: Existance

There are two main justifications:

1. Classical transaction costs:
▶ existence of economies of scale and economies of scope
▶ but physical and technological costs do not provide a
satisfactory justification given significant progress in
telecommunications and computers that can dramatically
reduce these costs
2. Informational Asymmetries:
▶ these asymmetries generate market imperfections that act
like transaction costs
▶ banks gather information by accepting deposits and making
loans so they have more information on demand for
liquidity and more information on corporate risk

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Classical Transaction Costs

Do economies of scope or scale exist between deposit and credit

activities?
▶ “Central location” story:
▶ because of transportation costs, it is efficient for the same
firm or branch to offer deposit and credit services in a single
location
▶ Portfolio theory:
▶ if two assets are positively correlated but one has positive
expected returns and the other negative expected returns,
then holding both at the same time is a way of diversifying
▶ (Pyle 1971): banks are interpreted as investors who hold a
long position in securities having a positive expected excess
return and a short position in securities that have a
negative expected excess return under the assumption that
the returns of these two categories of securities are
positively correlated
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Asymmetric Information I: Bryant (1980)
Starting with Bryant (1980), we think of banks as “pools of
liquidity.” Model:
▶ An economy with three dates (t = 0, 1, 2) and one good
that is to be consumed at t = 1 or t = 2.
▶ liquidity risk is modelled as uncertainty about the timing of
consumption don't know when the money is required
▶ A continuum of ex-ante identical agents that is endowed
with one unit of the good at t = 0.
▶ At time t = 0, agents don’t know if they want to consume
at t = 1 or t = 2
▶ At t = 1, agents learn their types.
▶ impatient agents (type 1): need to consume early and have
utility u(c1 ) + 0 · u(c2 ) = u(c1 )
▶ patient agents (type 2): only get utility from consuming in
period 2 so have utility 0 · u(c1 ) + u(c2 ) = u(c2 )
▶ Information about type is private . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Agents

At t = 0, the probability of being type i (i = 1, 2) is πi . Thus we

can write the ex-ante expected utility of each agent as:

U = π1 · u(c1 ) + π2 · u(c2 )

Assumptions:
▶ u′′ (c) < 0 < u′ (c) - that is the utility of consumption is
increasing and concave
▶ no discounting (to simplify the model) but can be added
without changing the basic results

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Investments

There are two investment technologies in this economy:

▶ Storage technology (liquid).
▶ For example, a bank deposit. Through storage, the good
can be moved from one period to the next.
▶ If you store an amount x at t = 0, then x can be used in the
following period.
▶ Long-run technology (illiquid).
▶ For example, a bank loan. The investment cannot be
returned immediately, and immediate liquidation comes at
a loss. In compensation, holding the investment to maturity
brings a reward.
▶ If you invest an amount I at t = 0, you get RI units of
consumption at t = 2 but only lI units if you have to
liquidate it at t = 1 where R > 1 and l < 1.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

The optimal allocation: best possible world
There is a unique symmetric Pareto-optimal allocation (c∗1 , c∗2 )
that is easily obtained by computing:
max π1 u(c1 ) + π2 u(c2 ) (1)
c1 ,c2

under the following resource constraints:

π1 c1 = 1 − I and π2 c2 = RI.
We can combine these constraints into a single one:
c2
π1 c1 + π2 = 1
R
The optimal allocation satisfies the following first-order
condition:
u′ (c1 )/u′ (c2 ) = R ratio of prices
which says that the agents would like to equate the marginal
rate of substitution between consumptions at dates 2 and 1
with the returns on the long-run technology. .
.
.
.
.
. . . . .
. . . .
. . . .
. . . .
. . . .
. . . . .
.
.
.
.
.
.
.
.
.

Mirza, Afrasiab EFI Part B: Lecture 1

Autarky

Let’s think about the case in which there is no trade between

agents, called autarky.
▶ each agents chooses independently the quantity I that will
be invested in the long-run technology
▶ if he has to consume early (at t = 1), then his investment
will be liquidated at t = 1, yielding

c1 = 1 − I + lI = 1 − I(1 − l)

▶ but if he has to consume late he gets:

c2 = 1 − I + RI = 1 + I(R − 1)

▶ in autarky each agent will select the consumer profile

(c1 , c2 ) that will maximize his ex-ante utility U under the
constraints above
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Autarky (continued)

The key result is that in autarky, the allocation is inefficient:

Proposition
The autarky allocation is inefficient because π1 c1 + π2 cR2 < 1.

Proof.
Notice that c1 < 1 (unless I = 0) and c2 < R (unless I = 1).
Then combining these two facts we have:
c2
π 1 c1 + π 2 < 1.
R
total value is less than 1, so worse off

by having autuarky everybody is poor

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Market Economy
Suppose a bond market is open at t = 1 allowing agents to
trade:
▶ the bond pays one unit of consumption in period t = 2
▶ p is the price at t = 1 of the bond that yields one unit of
good at t = 2
▶ p ≤ 1 otherwise the market does not clear: no one would
be willing to buy the bond, they would prefer to store
With a bond market, if the agent has to consume early, they get

c1 = 1 − I + pRI

where the agent sells RI bonds. If the agent can consume later
he gets
1−I
c2 = RI +
p
since he can buy (1 − I)/p bonds at t = 1. .
.
.
.
.
. . . . .
. . . .
. . . .
. . . .
. . . .
. . . . .
.
.
.
.
.
.
.
.
.

Mirza, Afrasiab EFI Part B: Lecture 1

Market Economy (continued)

The agent’s problem is then to choose I ex-ante to maximize

utility. Notice that:
c1
c2 =
p
and both are linear functions of I. Since I can be freely chosen
by agents, the only possible interior equilibrium price is:
1
p=
R
Otherwise, either an excess supply or excess demand of bonds
would occur (arbitrage opportunity would exist): I = 1 if
p > R1 and I = 0 if p < R1 .

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Allocations in a Market Economy

▶ Market clearing condition is

(1 − I)
π1 RI = π2
p
.
▶ Idea: impatient types sell his proportion of the long-term
asset in order to consume in period t = 1: π1 RI.
▶ But the amount must equal to the remaining resources of
the patient types: π2 (1 − I).
▶ The first-order condition, the budget constraint and the
market-clearing conditions can be used to solve for the
market equilibrium (cM M
1 , c2 ) .
▶ Generally Ru′ (cM ′ M
2 ) < u (c1 ) so the market allocation is not
Pareto-optimal.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Thus the market economy does not provide perfect insurance
against liquidity shocks and therefore does not lead to an
efficient allocation of resources.

This is because individual shocks are not publicly available and

we cannot trade securities contingent on these shocks, leading
to a problem of incomplete financial markets.

Next we show how a financial intermediary can improve the

situation.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Allocation via Financial Intermediation

The Pareto-optimal allocation (c∗1 , c∗2 ) can be obtained very

easily by a financial intermediary that offers a deposit contract
as follows:
▶ in exchange for a deposit of one unit at t = 0, individuals
get either c∗1 at t = 1 or c∗2 at t = 2.
▶ in order to fulfill its obligations, the FI stores π1 c∗1 and
invests I = 1 − π1 c∗1 in the illiquid technology
This leads to the following result:
Proposition
In an economy in which agents are individually subject to
independent liquidity shocks, the market allocation can be
improved by a deposit contract offered by a financial
intermediary.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Notes

▶ Banks can insure the agent against shocks.

▶ If there is only one bank in the economy but no market
then the Pareto-optimal allocation can be achieved and the
private information does not matter.
▶ There is no profitable deviation: someone who is impatient
does not gain from pretending he is patient and vice-versa.
▶ Financial intermediaries and markets cannot co-exist in
this model:
▶ The price of the bond would still be p = 1/R.
▶ So the optimal allocation is no longer a Nash equilibrium as
late consumers are better off withdrawing early and buying
bonds.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Asymmetric Information II: Leland and Pyle (1977)
Generally, adverse selection in markets occurs when:
▶ sellers have more information about product “quality” than
the buyers
▶ this makes buyers hesitant to pay a fair price
▶ in turn, high quality sellers cannot get a fair price and do
not sell
▶ only low quality sellers remain in the market
▶ buyers understand this and lower the prices they are
willing to pay
▶ overall the market is thus inefficient
In financial markets, adverse selection problems may be
overcome by banks:
▶ they can act as “information sharing coalitions”
▶ this is the idea of Leland and Pyle (1977)
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Model:
▶ There are a large number of entrepreneurs:
▶ each has initial wealth W0 and is endowed with a risky
project of size normalized to 1.
▶ gross return of the project is R̃(θ) = 1 + r̃(θ) where
r̃(θ) ∼ N(θ, σ 2 )
▶ θ differs across projects - we can think of it as an
entrepreneur’s type
▶ the entrepreneurs have an exponential utility function
u(w) = −e−ρw
▶ because of risk-aversion, even W0 > 1, the entrepreneurs
would prefer to sell their projects
▶ Investors:
▶ risk-neutral and have access to a storage technology
▶ Equilibrium in the project market when θ is observable:

P(θ) = E[r̃(θ)] = θ.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Equilibrium when θ is not observable by investors
▶ The price P is the same for all projects.
▶ If self-financed, each entrepreneur obtains:
1
E[u(W0 + r̃(θ))] = u(W0 + θ − ρσ 2 )
2
▶ If selling the project to the market, he gets utility

u(W0 + P)

▶ Sells only if

1
u(W0 + P) > u(W0 + θ − ρσ 2 )
2
▶ So you sell only if θ ≤ P + 12 ρσ 2 ≡ θ̂
▶ only entrepreneurs with low quality projects want to sell
▶ in equilibrium the price is thus: P = E[θ|θ < θ̂] < E[θ]
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

More intuition

▶ If there are too many bad entrepreneurs, then the discount

the buyers demand is too great.
▶ Then good entrepreneurs would not be willing to sell their
project.
▶ This is inefficient because even good entrepreneurs are
risk-averse and they must bear the risk of running their
projects.
▶ When P is sufficiently large it is possible that all good
entrepreneurs trade. Then, this is efficient since no
risk-averse agents hold risk. But here the good
entrepreneurs subsidize the bad ones.
▶ When there is asymmetric information, agents with more
information can sometimes signal their information to
those with less information.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Signalling
▶ Leland and Pyle (1977) propose that good entrepreneurs
can signal the quality of their projects by investing their
own wealth into the project
▶ let θ take two values: θ1 with prob. π1 and θ2 with prob. π2
with θ2 > θ1 (so there are only two types of projects)
▶ let α be the fraction of the project self-financed by the good
entrepreneurs; hence he sells a fraction 1 − α
▶ idea: entrepreneurs with higher quality projects should be
more willing to take debt than equity
▶ in order to make sure others do not mimic we need:

u(W0 + θ1 ) ≥ E [u(W0 + (1 − α)θ2 + αr̃(θ1 )]

▶ in the case of the exponential utility and the normal

distribution we have:
α2 2(θ2 − θ1 )
≥
1−α ρσ 2
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Equilibrium with signalling

Proposition
When the level of projects’ self-financing is observable, there is
α2 2 −θ1 )
a signalling equilibrium as long as 1−α ≥ 2(θρσ 2 , where there
is a low price P1 = θ1 for entrepreneurs that do not self-finance
and a high price P2 = θ2 for entrepreneurs who self-finance a
fraction α of their projects.

Notes:
▶ θ1 entrepreneurs get the same outcome as in the
full-information case
▶ θ2 entrepreneurs get lower utility i.e. u(W0 + θ2 − 12 ρσ 2 α2 )
instead of u(W0 + θ2 ), and the difference C = 12 ρσ 2 α2 is the
informational cost of capital
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

Mirza, Afrasiab EFI Part B: Lecture 1

Coalition of Borrowers
A “coalition of borrowers” or a bank can also overcome the
asymmetric information problem:
▶ notice first that α is decreasing in σ because the agent
wants to retain less equity when the project is more risky
▶ notice that the cost of borrowing is increasing with the
variance of the return:
1 2(θ2 − θ1 )
C = ρσ 2 α2 = (1 − α)
2 ρ
which is increasing in σ as α is decreasing in σ
▶ let N identical entrepreneurs of type θ2 form a partnership
and collectively issue securities in order to finance N
projects
▶ if individual returns are independent, the expected return
2
per project is still θ2 but the variance per project is now σN
▶ so a coalition of borrowers or bank can do better than
individual borrowers by “pooling” risk .
.
.
.
.
. . . . .
. . . .
. . . .
. . . .
. . . .
. . . . .
.
.
.
.
.
.
.
.
.

Mirza, Afrasiab EFI Part B: Lecture 1