SSRN Id3350684

Download as pdf or txt
Download as pdf or txt
You are on page 1of 52

Liquidity Shocks and Interbank Market Failures: The

Role of Deposit Flights, Non-Performing Loans, and


Competition
Demian Macedo and Victor Troster

Abstract

Banks may be reluctant to remove bad loans from their portfolios during liquidity
shortfalls, giving rise to a moral hazard problem. In this paper, we analyze how
liquidity shortages affect the ability of the interbank market to provide liquidity in
a moral hazard setting. We distinguish two types of liquidity shocks that arise due
to a deposit flight (a contraction in the deposit supply) or to a cash-flow shock (an
increase in the non-performing loans). We show that the source of a liquidity shortfall
is the main determinant of the decision of banks to stop lending in the interbank
market, rather than the extra amount of funds that banks need to cover. An increase
in the non-performing loans has more adverse effects on balance sheets than a deposit
flight. We also demonstrate that competition has a dual effect on financial stability.
Interbank competition enhances financial stability by reducing the liquidity provision
cost, whereas credit market competition worsens financial stability by inducing banks
to take riskier profiles.

Keywords Liquidity shocks; Interbank market; Deposit flights; Non-performing


loans; Credit market competition; Moral hazard.
JEL Classification D82; G00; G01; G21.

Electronic copy available at: https://ssrn.com/abstract=3350684


1 Introduction
Banks with high proportions of non-performing loans (NPLs) experienced severe complica-
tions to meet their liquidity needs in the interbank market during the subprime mortgage
crisis in the US from 2007 to 2009 (Brunnermeier, 2009; Afonso et al., 2011; Purnanandam,
2011). Unlike the subprime mortgage crisis in the US, European banks suffered liquidity
shortfalls due to deposit flights and increases in the NPLs during the financial crisis. Deposit
flights affected countries such as Greece, Italy, Ireland, Portugal, and Spain that suffered a
reduction of retail and institutional deposits at an unprecedented rate (Moro, 2014; Whelan,
2014; Bibow, 2015; Priftis and Rousakis, 2017). European banks also coped with an increase
in their non-performing exposures (Cocco et al., 2009; Cucinelli, 2013; Makri et al., 2014;
Roman and Sargu, 2015; Cai and Zhang, 2017; Ozili, 2019). The proportion of NPLs to
loans exceeded 15% in many countries with levels around 50% in Greece and Cyprus, under-
mining an important source of bank liquidity such as the revenue streams (Aiyar et al., 2015;
Magnus et al., 2018). As a result, European and North-American banks had to raise their
interest rates to attract deposits that eroded their profitability and led to weaker balance
sheets (Messai and Jouini, 2013; Van Rixtel and Gasperini, 2013; Allen et al., 2014; Acharya
and Mora, 2015; Dimitrios et al., 2016; Grigorian and Manole, 2017).
Banks tend to hide loan losses and be reluctant to remove bad loans from their portfolio
when exposed to financial distress.1 Liquidity shortfalls may induce a moral hazard problem
in the interbank market by exacerbating such behavior by banks. In this paper, we analyze
the reaction of banks to liquidity shortages stemming from both sides of the balance sheet.
We investigate whether banks react differently to liquidity shocks on the asset (a reduction
in the revenue stream due to an increase in the NPLs) and liability (a contraction in the
deposit supply) side of the balance sheet.
We develop a theoretical model to examine how deposit flights or decreases in the value
of the assets limit the access of banks with weak balance sheets to interbank funding. Both
types of liquidity shocks weaken the balance sheet and reduce the incentives of banks to
restructure their portfolio and to restore solvency. Nevertheless, our model suggests that
“NPLs” shocks affect the balance sheets of banks more than “deposit shocks” because the
1
The literature provides evidence of this behavior during a financial crisis (Sheng, 1996; Gunther and
Moore, 2003; Niinimaki, 2012). Banks may be reluctant to remove bad loans from their portfolio because
they face a reputational cost, and a bad reputation worsens their funding conditions in the market (Gabrieli,
2011). Further, banks restructuring their portfolios may have to sell their assets at fire-sale prices, leading
to considerable losses for their shareholders (Diamond and Rajan, 2011).

Electronic copy available at: https://ssrn.com/abstract=3350684


former leads to more leveraged institutions. As a result, the debt capacity of banks is more
affected under a liquidity shock coming from the asset side. Therefore, the source of a
liquidity shortfall is the main determinant of the decision of banks to stop lending in the
interbank market, rather than the extra amount of funds that banks need to cover.
Our model features a three-period economy populated by two banks that are monopolists
in their own deposit markets. Banks invest in loans that mature in the last period, which
generate some cash in the interim stage. This cash is lower than the amount of deposits
invested, giving rise to a maturity mismatch among assets and liabilities that forces banks
to renew part of their deposits. Banks differ in their probability of loan maturity. The bank
with a lower probability of success has a chance of increasing it by removing bad loans in the
interim stage. However, improving the probability of success is costly. Then, the behavior
of this bank depends on the strength of its balance sheet.
This bank is also impaired by two mutually exclusive liquidity shocks, which represent
relevant events that took place during the recent European financial crisis. The first shock
is a “deposit flight” that arises from the liability side of the balance sheet of banks, when
depositors require a higher interest rate to keep their savings in banks. The second one is a
“cash-flow” shock due to an increase in the NPLs, as in Holmström and Tirole (2000). An
increase in the NPLs reduces the cash flow of banks so that they need to renew a higher
amount of deposits to fulfill its financial duties. A major implication of our model is that the
reason for which banks need extra funding is crucial to explain a breakdown of the interbank
market.
If retail deposits are the unique source of funding for the impaired bank, then an increase
in the NPLs cuts off the “cash-in-hand” of the bank so that it needs to offer higher rates to
gather extra deposits, for fulfilling its obligations. This increased amount of deposits and of
interest rates expand its liabilities, reducing its equity.
A deposit flight worsens the equity of the impaired bank as deposits are renewed at a
higher cost, but their liabilities remain constant. Nonetheless, the increment in liabilities of
the bank due to cash-flow shocks exceeds the reduction in the bank equity due to deposit-
flight shocks. As a result, cash-flow shocks harm the debt capacity of banks more than
deposit shocks do, since the former ones lead to more leveraged institutions.
We examine the role of interbank relationships in the liquidity provision. We extend our
model to a large number of banks under Cournot competition. The results of the base model
are still valid under this setting. Liquidity shortfalls imply a greater cost for the impaired
bank when they arise from the asset side of the balance sheet, regardless of the degree of

Electronic copy available at: https://ssrn.com/abstract=3350684


interbank competition.
We also investigate whether credit market competition increases the exposure of banks
to an interbank failure. Unlike the traditional literature that focuses on the impact of com-
petition on the risk-taking incentives of banks (Allen and Gale, 2004a; Boyd and De Nicoló,
2005; Carletti, 2008; Berger et al., 2009; Martinez-Miera and Repullo, 2010), we analyze the
role of credit market competition on the ability of the interbank market to allocate liquidity
during a crisis. To our knowledge, Carletti and Leonello (2019) is the only work that ex-
amines this issue. They find that credit market competition improves financial stability by
reducing the opportunity cost of liquid reserves.
Boyd et al. (2007), De Nicoló and Loukoianova (2007), Schaeck et al. (2009), Uhde and
Heimeshoff (2009), Soedarmono et al. (2013), Anginer et al. (2014), Schaeck and Cihák
(2014), Akins et al. (2016), Leroy and Lucotte (2017), and Goetz (2018) provide empirical
evidence indicating that interbank competition is associated with greater financial stabil-
ity. In contrast, the positive relationship between interbank competition and risk-taking
incentives of banks is well documented (Brewer III and Saidenberg, 1996; Demsetz et al.,
1996; Beck et al., 2006; Yeyati and Micco, 2007; Ariss, 2010; Beck et al., 2013; Jiménez
et al., 2013; Kabir and Worthington, 2017; Danisman and Demirel, 2019). Nonetheless,
Berger et al. (2009) and Martinez-Miera and Repullo (2010) accommodate both literature
strands by showing that interbank competition and risk-taking incentives have a U-shaped
relationship.
In this paper, we examine how both interbank and credit market competition affect finan-
cial stability. Our model implies that interbank competition reduces the liquidity provision
cost, enhancing financial stability. On the other hand, credit market competition decreases
the cash flow of banks so that they become more dependent to external funding. As a result,
banks lower the supply of interbank funding, undermining financial stability. Therefore, we
show that competition has a dual effect on financial stability.
Several papers have pointed out the inefficient provision of liquidity by the interbank
market as a consequence of information problems. Wagner (2007) argues that the interbank
market may not allocate liquidity properly under aggregate liquidity shocks. Allen et al.
(2009) assert that banks hoard liquidity to face aggregate liquidity shocks that withhold
liquidity in the market. Bruche and Suarez (2010) show that deposit insurance may decrease
the level of trade of interbank loans due to a rise in the counterparty risk. Acharya et al.
(2012) demonstrate that banks with liquidity surplus have a significant degree of market
power that allows them to withhold liquidity, inducing inefficient asset sales.

Electronic copy available at: https://ssrn.com/abstract=3350684


Acharya et al. (2011) show that a change in the information available may reduce the
ability to use assets as collateral, even tough assets have a high fundamental value. Bolton
et al. (2011) demonstrate that the adverse selection problem of the value of assets may ac-
celerate inefficiently asset liquidation during a liquidity shortage. Baglioni (2012) establishes
that the interaction between liquidity and credit risks may collapse the interbank market
when the adverse selection problem is severe.
Heider et al. (2015) highlight how asymmetric information can amplify the consequences
of the counterparty risk. Accordingly, asymmetric information impairs the role of the in-
terbank market as a liquidity provider. Boissay et al. (2016) applied a Dynamic Stochastic
General Equilibrium (DSGE) model to evaluate the effect of banking crises on macroeco-
nomic variables. Moral hazard and asymmetric information in the banking sector may induce
credit crunches and abrupt interbank market freezes.
In contrast to previous literature, we focus on the source of a liquidity shortfall as a
determinant of an interbank failure. To this purpose, we define two types of liquidity shocks
that represent relevant events that took place during the recent European financial crisis, and
analyze their effects on the balance sheet of banks. Overall, our framework helps understand
the mechanisms of liquidity redistribution, and it has relevant implications for policy makers
when regulating capital and liquidity requirements of banks.
The rest of the paper proceeds as follows. Section 2 describes our model. Section 3
explains the setup of the model. Section 4 presents a solution to the analytical problem
of the agents. Section 5 defines the equilibria of the model, whereas Section 6 discusses
their main implications. Section 7 examines the impact of interbank and credit market
competition on the ability of impaired banks to withstand liquidity shocks. Finally, Section
8 concludes the paper.

2 The model
We consider a three-period model, t = {0, 1, 2}, with two banks (A and B) that are monop-
olists in their short-term deposit markets. Each market contains a unit mass continuum of
identical agents. For simplicity, we assume that there is no discounting between the periods.
In this economy, banks A and B can invest in a long-term portfolio of loans that takes I
units of deposits at date 0 and produces Cf < I and R > I units of consumption at dates
1 and 2, respectively. Nevertheless, the probability of loan maturity differs between banks,
being such probability lower for Bank A.

Electronic copy available at: https://ssrn.com/abstract=3350684


We assume that both banks lack any income of their own to invest. As a result, they use
short-term deposits to carry out their projects, and thus a maturity mismatch among assets
and liabilities arises.
For simplicity, we assume that only Bank A (the impaired bank) is affected by two
mutually exclusive liquidity shocks: either a deposit and a cash-flow shock. The former
leads to a contraction in the deposit supply, whereas the later cuts off the cash flow of
the bank, raising its need for external funds. Since liquidity shocks do not affect Bank B,
both banks are heterogeneous with respect to their intermediation ability, which creates the
incentive to trade interbank loans.
Nevertheless, a classic moral hazard problem arises in the market. Bank A has the
opportunity to remove its bad loans in the interim stage. This action increases the probability
of project success, but it is unobservable for Bank B. Thus, the moral hazard problem
introduces the possibility that banks do not lend to each other, resulting in an inefficient
allocation of funds within the banking system.
In our setting, the moral hazard problem cannot be solved through contracts, since
outside financiers cannot observe the choices of Bank A. Conversely, limited liability prevents
it from being punished when the return on the long-term investment is zero. To illustrate
how the model operates, we present below the decision problems of depositors and banks.

2.1 Depositors
There are two independent deposit markets, labelled as A and B. Each market contains a
continuum of identical depositors with endowments of I and ω in periods 0 and 1, respec-
tively. Depositors consume only in periods 1 and 2, and their savings are their only resource
for consumption in period 2.
We assume that the bank is the only mean available for depositors to allocate consumption
from one period to another. They lend funds to the bank as short-term deposits to save,
which are held from t to t + 1. Since deposits are protected by deposit insurance, depositors
receives gross interest rates of r0 and r1 in periods 1 and 2, respectively.
We assume that individual preferences are specified by the following constant relative
risk-aversion utility function:

cθ1 cθ2
U (c1 , c2 ) = + , (1)
θ θ
where ct denotes the consumption at period t, and the risk-aversion parameter θ determines

Electronic copy available at: https://ssrn.com/abstract=3350684


the inter-temporal elasticity of substitution between consumption in any two periods. It also
shapes the inverse-deposit supply curve. This curve depicts the minimum gross interest rate
that depositors are willing to receive at the following period to supply a certain amount of
deposits at the current period. We denote the inverse-deposit supply curve at time t by Γt .

2.2 Banks
We assume that both banks carry out the same long-term investment that is funded only
through retail deposits at period 0. The investment yields an income of Cf at period 1 and
either R > I or no income at period 2. The probability of producing an income of R differs
across banks A and B. Bank A gets R with probability PL , while the probability of getting
R is PH for Bank B, where PH > PL .
In the interim stage, Bank A suffers a liquidity shock but also has the chance of removing
bad loans from its portfolio. Then, Bank A can keep its original portfolio, with a probability
of repayment of PL , or it can restructure its portfolio to improve the quality of the loans, at
a cost of τ . In the latter case, the probability of success increases to PH , where we assume
that PH = 1, without loss of generality.
Bank A also faces a liquidity shortage that stems from one of two mutually exclusive
equiprobable shocks: a decrease in the deposit supply or a reduction in the cash flow caused
by an increase in its NPLs. Both shocks decrease the present value of the bank, but they
have different effects. An increase in the NPLs reduces the cash flow of the bank by α > 0,
and the bank receives Cf − α instead of Cf .
Under a deposit shock, Bank A suffers a contraction of its deposits and, consequently,
an increase in its financing cost. Then, both shocks do not affect neither the payment R
nor the probability of the success of the project {PL , PH }. These liquidity shocks, however,
change the incentives of Bank A to remove bad loans, making this investment riskier.

2.3 The interbank market


In our model, banks A and B may trade short-term loans denoted by x. The interbank
market helps the banking system achieve a better allocation of liquidity when the problem of
asymmetric information is moderate. This market is a competitive one so that the interbank
interest rate, RI , is given for both banks.
Trading funds between banks does not produce any benefit at period 0 because both
banks and deposit markets are symmetric. However, under liquidity shocks, the reallocation

Electronic copy available at: https://ssrn.com/abstract=3350684


of funds through this market allows Bank A to offset the effect of the shocks. Besides, since
Bank B is not affected by any shocks, it acts as the lending bank.

3 Setup of the model


This section presents the setup of the model. Fig. 1 displays the timeline of the model. In
period 0, depositors receive an income of I > 0 that is allocated in savings. The savings
of the depositors are supplied to the both banks as short-term deposits that pay a gross
interest rate of r0 = 1.2 Both banks demand these liquid deposits to finance an illiquid loan
portfolio that requires an initial investment of I.
At the beginning of period 1, banks A and B lack enough cash relative to the maturity of
deposits (Cf < r0 I), and thus they face an asset-liability mismatch. To meet their current
obligations, both banks should offer depositors an interest rate high enough to renew their
short-term deposits or to liquidate their investment. At the same period, Bank A suffers
a liquidity shock that deteriorates its financial situation. This results in a worse financial
situation for Bank A than that of Bank B, encouraging both banks to engage in interbank
loans.
After the shock, Bank A must decide either to keep its original portfolio or to manage
its loans diligently to get a repayment with a higher probability. As the decision of Bank
A is unobservable for Bank B, if the liquidity shock is severe enough, the interbank market
may collapse. Then, an inefficient allocation of funds may arise.
At period 2, the investment projects of banks A and B are liquidated (paying R or zero),
and the proceeds are split up according to the commitments made at period 1.
2
Banks A and B are monopolists in their own deposit markets, and the depositors’ utility of consuming
at t = 0 is null. Then, it follows that r0 = 1.

Electronic copy available at: https://ssrn.com/abstract=3350684


t=0 t=1 t=2

- Depositors supply short-- Both banks should renew- Project return is


term deposits. their short-term deposits realized, paying
- Both banks make their to operate. either R or zero.
illiquid investment. - Bank A suffers either a
- Bank A uncovers the cash-flow or a deposit shock
quality of its assets. - Bank A may restructure
its loan portfolio.
- Interbank loans may
be signed.

Fig. 1 Timeline of the model

4 The decision problem


In this section, we investigate the optimal decisions of depositors and banks A and B. We
apply market-clearing conditions to obtain the equilibria of the model. We illustrate how
liquidity shocks operate under asymmetric information, affecting the behavior of the agents.
The model has two states of nature indexed by j = {C, D}, which refer to cash-flow and
deposit-flight shocks, respectively.
In state D, depositors of market A suffer a reduction in their income, and they withdraw
their deposits to cover consumption needs. This leads to a lower volume of funds supplied
by depositors, but the cash flow of Bank A is not affected. Conversely, Bank A experiences
a drop in its cash flow in state C, whereas the income of its depositors remains unchanged.
We derive the inverse-deposit supply curve and the optimal allocation of interbank loans and
deposits that Bank A demands. We also present the lending problem of banks A and B, and
we describe our criteria for making both types of shocks comparable.

4.1 The maximization problem of the depositors


There are two independent deposit markets. Each market contains a continuum of identical
depositors with endowments of I and ω in periods 0 and 1, respectively. They obtain
utility from inter-temporal consumption, and their preferences are represented by the utility

Electronic copy available at: https://ssrn.com/abstract=3350684


function of Eq. (1). The only difference between markets A and B is that depositors of
market A face two states at period 1.
In state D, all depositors are affected by an income shock of size ρ > 0, whereas their
incomes remain unchanged in state C. Hence, depositors of market A receive a payment of
ωD = ω − ρ and ωC = ω at states D and C, respectively, in period 1. We assume that
ρ ∈ [0, ρmax ] for some ρmax > 0, which is precisely defined in the Appendix.
We specify the deposit shocks exogenously following the model of Diamond and Dybvig
(1983). For simplicity, we assume that depositors suffer a sudden income shock instead of
an unpredictable shock on their needs for cash. Then, a representative depositor chooses the
optimal level of consumption and saving for each state of nature by solving:

cθ1j cθ2j
max + ,
ctj ,stj θ θ


c1j + s1j = ωj + I,
s.t.
c = s r ,
2j 1j 1j

where ctj and stj denote the consumption and saving allocations, respectively, at period
t and under shock j. We derive the consumption and saving functions of period 1 from
the maximization problem above. We denote the consumption and saving functions by
C1j (r1j ; ωj ) and S1j (r1j ; ωj ), respectively. In addition, the inverse-deposit supply at t = 1
after the occurrence of a shock of type j = {C, D} is given by:

  θ−1
(ωj + I − s) θ
Γ1j (s; ωj ) = .
s

The inverse-deposit supply Γ1j (s; ωj ) determines the interest rate that Bank A must pay
to gather an amount s of funds from the deposit market. Since the endowment of depositors
in Market B is not affected by any type of shocks, the inverse-deposit supply in that market
is given by

  θ−1
(ω + I − s) θ
Γ1 (s; ω) = .
s

Electronic copy available at: https://ssrn.com/abstract=3350684


4.2 The maximization problem of Bank A
To maximize profits, both banks make a long-term investment in their retail deposit markets
at period 0. To maintain a volume of deposits equal to I from period 0 to period 1, they
pay a gross interest rate of r0 = 1 since depositors do not consume at period 0. At period
1, banks A and B need to renew their liability. They find it optimal to use their cash flow
to cancel deposits, and thus to reduce the payment of interest. Hence, both banks need to
renew their deposits by an amount of L = I − Cf to continue with their current operations.
Bank A faces two mutually exclusive liquidity shocks. An increase in the NPLs reduces
the cash flow of the investment to Cf − α. Then, the liabilities of Bank A rise by an amount
of α: LC = I − (Cf − α). If Bank A does not have access to the interbank market, the cost
of renewing its liabilities from the deposit market is:

Γ1C (LC ; ωC )LC .

Deposit shocks arise because of a sudden drop of size ρ in the income of depositors.
Nevertheless, the liabilities of Bank A that should be renewed are LD ≡ L. As a result,
deposit shocks increase the cost of obtaining new financing from the deposit market so that
Γ1D (s, ωD ) > Γ1C (s, ωC ). Thus, the cost of renewing its obligation in autarky (without
interbank liquidity provision) is:

Γ1D (LD ; ωD )LD .

We assume now that Bank A may borrow from the interbank market short-term funds
to substitute retail deposits. Then, after the occurrence of a shock of type j ∈ {C, D}, Bank
A should choose the optimal allocation of deposit and interbank funds x that minimize its
funding cost. Bank A chooses the level of x that solves the following problem:

min CjA (x) ≡ Γ1j (sA A


j (x); ωj )sj (x) + RI x,
x≥0

where sA
j (x; Cf ) ≡ Lj − x is the amount of deposits that Bank A needs to gather from
the deposit market after obtaining x units from the interbank market. Thus, the first term
of CjA (x) is the cost of obtaining sA
j (x; Cf ) deposits, while the second one is the cost of
borrowing an amount of x funds from the interbank market. By solving the minimization
problem above, we obtain the demand of Bank A for interbank funds:

10

Electronic copy available at: https://ssrn.com/abstract=3350684


∂Γ1j (sA A
j (0);ωj ) ∂sj (0) A

 0,
 if RI > ∂s ∂x
sj (0) + Γ1j (sA
j (0); ωj ),
XjA (RI ) =
 ∂Γ1j (sA A A A
j (xj );ωj ) ∂sj (xj ) A
xA A A
j > 0, if RI = Γ1j (sj (xj ); ωj ) − sj (xA
j ),

∂s ∂x

where XjA (RI ) = 0 if the interbank interest rate is higher than the marginal cost of renewing
the obligations of Bank A in the deposit market. Accordingly, we define the inverse demand
for interbank funds in state j as follows:
 A
∂Γ1j sA
j (x); ωj ∂sj (x) A
RIj (x) ≡ Γ1j sA

j (x); ωj − s (x). (2)
∂s ∂x j

4.3 Equivalence between liquidity shocks


To compare the effect of the shocks on the liquidity provision of the interbank market, we
want them to produce the same liquidity needs for Bank A under autarky. Therefore, we
first need to determine the liquidity shortfall produced by a shock of size ρ in state D, and
then we equalize it to the decrease in the cash flow.
Let r̂1 be the deposit rate that Bank A offers to depositors to obtain the amount of
L = I − Cf in the absence of shocks, r̂1 = Γ1 (L; ω). If Bank A offers the rate r̂1 when there
is a deposit-flight shock, the amount of liquidity needs generated by ρ is:

α(ρ) ≡ L − S1D (r̂1 ; ωD ), (3)

where S1D (r̂1 ; ωD ) is the volume of deposits supplied by consumers at the rate r̂1 in state D.
Accordingly, the liquidity needs of Bank A are the difference between the obligations that
should be renewed and the amount of deposits obtained at the interest rate r̂1 . Then, if we
define the cash-flow shock equal to α(ρ), the liquidity needs for Bank A are the same in both
states.
In state C, Bank A should renew an amount of obligations given by LC = L + α(ρ).
Bank A can only obtain S1C (r̂1 ; ωC ) = L at the deposit rate r̂1 . Therefore, Bank A suffers
a liquidity shortfall of size α(ρ) regardless the type of shock, for any value of ρ.

4.4 The interbank market supply


The ability of banks to withstand liquidity shocks depends on the alternative sources of funds
that they have access to. The interbank market appears as the main alternative source of

11

Electronic copy available at: https://ssrn.com/abstract=3350684


liquidity for banking institutions. It provides banks with funds to offset the impact of a
liquidity shortage.
In our model, Bank B acts as a lending bank because it is unaffected by both shocks.
Thus, Bank B is able to gather cheaper funds than Bank A does, which creates the incentives
to trade funds. Nevertheless, interbank trading can be affected because Bank A may be
reluctant to remove bad loans from its portfolio if they are costly. As a result, the pledgeable
income of Bank A decreases. We assume that the investment project has a positive net
present value in any state of nature j = {C, D} if Bank A restructures its portfolio by
removing bad loans. Hence,

PH R − (Γ1j (Lj ; ωj )Lj + τ ) > 0,

where PH R are the loan returns when they have a high probability of success (PH ), τ is the
cost of restructuring its portfolio, and Γ1j (Lj ; ωj )Lj + τ is the total cost of the bank under
autarky. If Bank A does not restructure its loan portfolio, then the net present value of its
investment will be negative regardless of the occurrence of shocks. This assumption can be
written as follows:

PL R − Γ1 (L; ω)L < 0,

where Γ1 (L; ω) and L are the inverse-deposit supply and the liabilities of both banks that
should be renewed in period 1 if Bank A is not affected by any shocks, respectively. Therefore,
if Bank A takes liquidity from the interbank market keeping its original portfolio, lenders
will lose money in expectation. Accordingly, interbank trading is zero.
We assume that Bank B can identify the type of shock that occurs, but it is unable
to observe the behavior of Bank A. Therefore, Bank B lends to Bank A an amount of x
in state j only if Bank A has incentives to restructure its portfolio such that the following
incentive-compatibility constraint holds:

PH R − Γ1j (sA A A A
 
j (x); ωj )sj (x) − RI x − τ ≥ PL R − Γ1j (sj (x); ωj )sj (x) − RI x . (4)

In addition, we assume that Bank B only provides funds in the interbank market after it
has renewed all the obligations incurred by its investment, L = I − Cf . Then, the net cost
of providing an amount x of interbank loans is:

12

Electronic copy available at: https://ssrn.com/abstract=3350684


CIB (x) = Γ1 (sB (x); ω)sB (x) − Γ1 (sB (0); ω)sB (0),

where Γ1 (sB (x); ω)sB (x) is the cost of obtaining an amount of sB (x) deposits from Market
B, Γ1 (sB (0); ω)sB (0) is the cost of the investment of Bank B, and sB (x) ≡ L + x is the
amount of deposits that Bank B has to raise if it supplies x units of interbank funds.
As in Carletti and Leonello (2019), the cost of interbank funds does not depend on the
state j. Although the depositors of Market B are unaffected by any type of shock, liquidity
shocks affect the compatibility constraint of (4) and the availability of funds in the interbank
market. Accordingly, the volume of funds, supplied at the interbank rate RI , is obtained
from the following maximization problem:

max π B (x) ≡ PH RI x − CIB (x)


x≥0

s.t. CCj (x; RI ) > 0,

where CCj (x; RI ) ≡ R − (τ /∆P ) − Γ1j (sA A


j (x); ωj )sj (x) − RI x is the incentive-compatibility
constraint, and ∆P ≡ (PH −PL ) = 1−PL , since we assume PH = 1 for simplicity. Then, from
the first-order conditions obtained for an interior solution to π B (x), the supply of interbank
funding is:

B B
X̃ (RI ), if CCj (X̃ (RI ); RI ) > 0,


X B (RI ) =


0, otherwise,

where X̃ B (RI ) = xB is the supply of interbank funds in the absence of a moral hazard
problem, and xB is implicitly defined as

∂Γ1 (sB (xB ); ω) ∂sB (xB ) B B


RI = s (x ) + Γ1 (sB (xB ); ω).
∂s ∂x

5 Equilibria of the model


This section analyzes the market allocations that arise in this economy, in which banks A
and B finance their illiquid assets with both short-term deposits and interbank funds. Moral
hazard introduces the possibility of obtaining three different equilibria.

13

Electronic copy available at: https://ssrn.com/abstract=3350684


We denote the first equilibrium as an “Active Interbank Market Equilibrium” (AIE).
Under this equilibrium, Bank A is able to meet the necessary funds from the interbank
market to offset the negative effects of liquidity shortages. We define this equilibrium as
follows.

Definition 1 Active Interbank Market Equilibrium: An AIE exists in this economy



if, given the interbank rate RIj for j = {C, D}, there is a volume of interbank funding x∗j > 0
such that the following conditions hold:

∗ ∗
(1) XjA (RIj ) = X̃ B (RIj ) = x∗j ;

   
(2) PH R − Γ1j (sA ∗ A ∗ ∗ ∗ A ∗ A ∗ ∗ ∗
j (xj ); ωj )sj (xj ) − RIj xj − τ ≥ PL R − Γ1j (sj (xj ); ωj )sj (xj ) − RIj xj .


An AIE is possible as long as the demand of Bank A for interbank funds, XjA (RIj ), does

not exceed its debt capacity. Consequently, an AIE arises only if XjA (RIj ) satisfies condition
(2) of Definition 1 for j = {C, D}. In this equilibrium, banks and depositors maximize their
profits and utility, respectively, and all markets clear. Hence, funds are allocated efficiently,
and the impaired bank can partially reduce the financial cost produced by a liquidity shortage
of size α(ρ).
The model may also produce a “Market Freeze Equilibrium” (MFE), where there is no
RI at which interbank lending occurs. Trading is not possible under this equilibrium. If this
equilibrium occurs, both banks are unable to gather funding from external sources so that
an inefficient allocation arises.

Definition 2 Market Freeze Equilibrium: A MFE exists in this economy if, given a

liquidity shortfall of size α, there is no interbank interest rate RIj , for j = {C, D}, such that
both conditions (1) and (2) of Definition 1 are simultaneously satisfied.

In our analysis, a MFE occurs when the borrowing debt capacity is impaired because of
the asymmetric information across banks A and B. Under a MFE, the impaired bank funds
all its obligations through its own deposit market, regardless of whether it is affected by a
deposit flight or a cash-flow shock. Finally, we define a third possible equilibrium by “Mixed
Equilibrium” as follows.

Definition 3 Mixed Equilibrium: A Mixed Equilibrium (ME) exists in this economy if



there exists an interbank interest rate RIj , for only one type of shock j = {C, D}, such that
both conditions (1) and (2) of Definition 1 hold.

14

Electronic copy available at: https://ssrn.com/abstract=3350684


The Mixed Equilibrium arises when the source of the liquidity shortfall determines the
ability of the impaired bank to access interbank funding. Then, the lending bank provides
funds to the interbank market only under one type of shock if a ME occurs. We explain the
economic intuition of these outcomes in the following section.

6 Analysis of the equilibria of the model


This section derives the necessary conditions under which different equilibria arise in our
model. We analyze the implications that liquidity shocks have on the financing costs and
balance sheets of the impaired bank, as well as on trading in the interbank market.

6.1 Autarky
We first analyze the impact that each shock has on the financial costs of Bank A, when it
funds all its liquidity needs exclusively in the deposits market.

Proposition 4 For any ρ ∈ [0, ρmax ], liquidity shortfalls of size α(ρ) have a greater cost
for the impaired bank if they come from the asset side of the balance sheet. In particular,
a cash-flow shock always incurs higher costs than those generated by a deposit-flight shock:
Γ1C (LC ; ωC )LC ≥ Γ1D (LD ; ωD )LD .

Proposition 4 implies that cash-flow and deposit shocks have different effects on the
financing costs of Bank A, although they generate the same liquidity needs. This result arises
because the mechanism behind each shock is different. Under a deposit flight, depositors
need to be compensated by a higher interest rate to keep their money in the bank for another
period. Then, Bank A has to pay a higher interest rate to the depositors, decreasing their
overall profitability. The increase in the interest rate caused by this shock worsens the equity
of the bank as deposits will be renewed at a higher cost. In this case, the total cost under
autarky is the following:

Γ1D (LD ; ωD )LD = Γ1 (L; ω) L + ∆EQD ,


| {z }
r̂1

where the first term on the right-hand side of the equation above is the total cost of renewing
the liabilities of both banks when there is no shock, and the second term is the decrease in
equity due to the increase in interest paid, which is defined as:

15

Electronic copy available at: https://ssrn.com/abstract=3350684


 
∆EQD ≡ LD Γ1D (L; ωD ) − r̂1 > 0,

where the term in parentheses is the increase in the interest rate generated by a deposit
shock.
Suppose now that the bank suffers an increase in its NPLs. Then, the cash-flow level
of the bank is reduced, generating a decrease of α in the cash flow. This drop in the cash
flow should be compensated by an increase in the liabilities of the bank from L to LC . By
attempting to cover this funding shortage, Bank A has to raise the interest rate on deposits
to attract more deposits. The total cost can be decomposed into the sum of the costs in
the absence of shocks, the increment in the interest paid (∆EQC ), and the increase in the
liabilities due to the shock (α(ρ)):

Γ1C (LC ; ωC )LC = Γ1 (L; ω) L + ∆EQC + α(ρ) > 0,


| {z }
r̂1

where the total amount of interest paid under a cash-flow shock is given by:

∆EQC ≡ LC (Γ1C (LC ; ωC ) − r̂1 ) + α(ρ) (r̂1 − 1) .

Proposition 4 states that cash-flow shocks are costlier than deposit shocks: Γ1C (LC ; ωC )LC ≥
Γ1D (LD ; ωD )LD . Therefore,

α(ρ) + ∆EQC ≥ ∆EQD .

As a result, banks that undergo an increase in their NPLs have weaker balance sheets
than those affected by a deposit-flight shock.

6.2 Interbank trading


We allow now Bank A to supplement traditional retail deposits with interbank loans as an
alternative source of funding. We analyze the implications that Proposition 1 has on the
interbank relationships. The role of the interbank market as a distributor of liquidity is well
recognized in the literature (Bhattacharya and Gale, 1987; Allen and Gale, 2000; Freixas
et al., 2000; Huang and Ratnovski, 2011; Acharya and Merrouche, 2013; Acharya and Mora,

16

Electronic copy available at: https://ssrn.com/abstract=3350684


2015). Interbank markets allow banks to fulfill their liquidity needs from banks with a
liquidity surplus.
In particular, if the problem of asymmetric information is moderate, interbank trading
helps the banking system reach a better allocation of funds. In this subsection, we show that
the type of shock is the main determinant of the decision of banks to stop lending to other
banks, rather than the amount of extra funds that banks need to cover.
We first present an auxiliary result. The following Lemma states that if there was not
a moral hazard problem in the interbank market, then there would be always a unique
market-clearing interest rate of R̃Ij .

Lemma 1 For all values of ρ, it is possible to find an interest rate R̃Ij such that XjA (R̃Ij ) =
X̃ B (R̃Ij ) = x̃j . Then, condition (1) of Definition 1 always holds.

Lemma 1 shows that if the moral hazard problem does not arise, then interbank trading
will be always possible. Accordingly, Lemma 1 states that the moral hazard problem is the
unique reason that explains an interbank market freeze in our model. We now determine
under which conditions the interbank market is able to provide liquidity to the impaired
bank.

Proposition 5 There is an upper bound ρC such that, for all ρ ∈ [0, ρC ], both conditions
(1) and (2) of Definition 1 simultaneously hold; thus, the banking system reaches an Active
Interbank Market Equilibrium.

The intuition behind Proposition 5 is straightforward. As long as ρ ∈ [0, ρC ], liquidity


shortfalls are moderate with a small impact on the balance sheets of banks. Hence, liquid-
ity shocks do not entail a high cost for the impaired bank, and the incentive-compatibility
constraint of Definition 1 holds. Therefore, an AIE is reached, and Bank A replaces de-
posits with interbank loans to offset the negative effect of liquidity shortfalls. The trade of
funds between banks A and B improves financial stability because the incentive-compatibility
constraint (4) is relaxed when Bank A borrows interbank funds to supplement deposits.
We now analyze under which conditions the interbank market reaches a mixed equilibrium
(ME). The existence of a ME implies that the source of a liquidity shock is crucial to explain
a market freeze.

Proposition 6 There exists a value of ρD such that, for all ρ ∈ (ρC , ρD ], the banking system
reaches a Mixed Equilibrium. In this equilibrium, the interbank market only provides funds

17

Electronic copy available at: https://ssrn.com/abstract=3350684


when the liquidity shortages are caused by a deposit shock, whereas the market freezes if a
cash-flow shock occurs. Moreover, if ρ > ρD , the interbank market freezes regardless of the
type of shock.

Proposition 1 establishes that cash-flow shocks are more adverse than deposit shocks
when Bank A has only one source of financing. Proposition 6 entails that a cash-flow shock
is still costlier than a deposit shock if Bank A has access to the interbank market because
CIB (x) is not affected by shocks.
In addition, an increase in the interest rate produced by a deposit flight can be partially
offset by replacing deposits with a cheaper source of funding such as interbank loans. Nev-
ertheless, the increment in liabilities of the banks due to a cash-flow shock, α(ρ), cannot be
offset by taking more debt from the interbank market. Thus, a cash-flow shock harms more
the debt capacity of Bank A than a deposit shock does. As a result, for all ρ in (ρC , ρD ], the
following inequalities simultaneously hold:

CCD (x∗D , RID



) > 0 and CCC (x̃C , R̃IC ) < 0,

where R̃IC denotes the interbank market-clearing rate in the absence of a moral hazard
problem. These inequalities above show that Bank A has an incentive to reduce its exposure
to bad loans only under a deposit shock, while it keeps its original portfolio if a cash-flow
shock occurs. Hence, Bank B supplies no funds under a cash-flow shock because of the
different impact of this shock on the balance sheet of Bank A.
Finally, the second statement of Proposition 6 asserts that a MFE is reached when
liquidity shortfalls are sufficiently large (if ρ > ρD ). This result follows since the incentive-
compatibility constraint of Definition 1 is not satisfied regardless of the shock type.
Fig. 2 displays the interbank market dynamics for different sizes of liquidity shocks. The
curves XjA and CCj are the demand for interbank funds and the incentive-compatibility
constraint for states of nature j = {C, D}, respectively, and X̃ B is the supply of funds in
the absence of a moral hazard problem.
If a shock of type j occurs, then the incentive-compatibility constraint (4) is satisfied on
and below the curve CCj . The points A and B represent the equilibrium in the interbank
market under a cash-flow and a deposit shock, respectively, if the problem of moral hazard
does not arise. Accordingly, the points A and B stand for an equilibrium only if the inter-
section between X̃ B and XjA lies on or below the curve CCj . All of the mentioned curves
are functions of the volume of interbank funds x.

18

Electronic copy available at: https://ssrn.com/abstract=3350684


Fig. 2.(a) illustrates the case where ρ ≤ ρC . The points A and B lie below or on the
curves CCJ . The liquidity shortfall of the bank is lower than or equal to α(ρc ); therefore,
the interbank market is able to allocate liquidity regardless of the type of shock. Then, both
conditions (1)-(2) of Definition 1 are satisfied so that an AIE is reached.
Fig. 2.(b) represents a ME. If ρC < ρ ≤ ρD , Bank A suffers a liquidity shortfall of size
α(ρ) ∈ (α(ρC ), α(ρD )]. The point B lies on or below the curve CCD , whereas A is above
CCD . Only under a deposit-flight shock the compatibility constraint (4) holds, and a ME is
reached. Otherwise, the interbank market breaks down.
Fig. 2.(c) depicts a situation where ρ > ρD . Bank A experiences strong liquidity shortfalls
that harm its incentives to behave prudently, collapsing the interbank market.

(a) If ρ 6 ρC , then an AIE arises (b) If ρC < ρ 6 ρD , then a ME is reached

(c) If ρ > ρD , the interbank market collapses (a MFE occurs) (d) Type of equilibrium as a function of liquidity shortages

Fig. 2 Possible equilibria of the model

Fig. 2.(d) delineates the types of equilibrium as a function of liquidity shortages in three
regions. Each one of them contains the values of ρ such that the model leads to an AIE, a
ME, or a MFE.
Overall, our model highlights the importance of moral hazard for interbank trading. The

19

Electronic copy available at: https://ssrn.com/abstract=3350684


interbank market helps banks reach an optimal allocation of funds when the opportunity
cost of removing bad loans is low or when liquidity shortfalls are moderate. Nevertheless,
if the impaired banks have no incentives to remove bad loans from their portfolio under a
liquidity shock, then the interbank market collapses.

7 Competition and financial stability


This section studies how interbank and credit market competition affect financial stability.
The banking industry consists of the retail deposit, credit, and interbank market. In the
retail deposit market, banks collect the savings of the individuals to lend them in the credit
market as loans. Banks interact with savers in the retail deposit market, whereas they deal
with consumers, investors, and firms in the credit market. Finally, banks provide liquidity
to each other in the interbank market to mitigate liquidity shortfalls.
We now analyze the effect that the interbank and credit market structure have on the
resilience of banks to liquidity shoks. We first extend our base model to a large number
of banks to examine the role of interbank competition on financial stability. We consider
that there is an oligopoly in the interbank market under Cournot competition. Next, we
investigate the effect of credit market competition on the ability of banks to withstand
liquidity shortfalls. Following Freixas and Ma (2014) and Carletti and Leonello (2019), we
assume that the cash flow of the loans is a sufficient statistic for the level of credit market
competition.

7.1 Interbank trading under Cournot competition


In this subsection, we expand our model to a large number of banks to investigate the role
of interbank relationships in the liquidity provision. We consider that there is an oligopoly
in the interbank market by assuming that banks compete à la Cournot. As in the model
with two banks, after suffering a liquidity shock of size ρj , Bank A searches for liquidity in
the interbank market, which it is now composed of N identical banks index by i = 1, . . . , N .
These banks provide liquidity at a cost Ci (x), where Ci (x) belongs to the set of convex
cost functions that satisfy some regularity conditions, Ω, defined in Assumption 5 in the
Appendix.
After Bank A suffers a liquidity shock of size ρj for j = {C, D}, it requires an amount
XjA (RI ) of liquidity from the interbank market. In state j, given the volume of interbank
loans of other banks x−ij , each bank supplies an amount xij of liquidity that solves the

20

Electronic copy available at: https://ssrn.com/abstract=3350684


following maximization problem:

max πijB (xij , x−ij ) = RIj (xj ) xij − Ci (xij )


xij ≥0

s.t. CCj (xj ; RIj (xj )) > 0,

where

A
 A
∂Γ 1j s j (x j ); ω j ∂sj (xj ) A
RIj (xj ) = Γ1j sA

j (x j ); ωj − s (xj )
∂s ∂xij j

ωj + I − θsA j (xj ) A
1
− θ−1
= Γ 1j s j (x j ); ω j ,
sAj (xj )
PN
with xj = i=1 xij . The first-order condition for an interior profit-maximizing level of liq-
uidity is
∂πijB (xij , x−ij ) ∂RIj (xj ) dCi (xij )
≡ xij + RIj (xj ) − = 0. (5)
∂xij ∂xij dxij
Given N banks and a liquidity shock with type j = {C, D}, a pure-strategy Nash equi-
librium is a set of interbank loans (x∗1j (N ), . . . , x∗N j (N )), in which each one of the banks
maximizes its profits and

CCj x*j (N ); RIj (x*j (N )) > 0,




PN
where x*j (N ) = i=1 x∗ij (N ). For simplicity, we consider only a symmetric equilibrium
x∗ij (N ), for all i, in which all banks follows the same strategy in the equilibrium. Let
(x∗1j , . . . , x∗N j ) be the competitive equilibrium, where limN →∞ x∗ij (N ) = x∗ij , limN →∞ x∗j (N ) =
x∗j , RIj (x∗ ) = dCi (x∗ij )/dxij , and CCj (x∗j ; RIj (x∗j )) > 0. The following proposition summa-
rizes how the interaction between banks under Cournot competition affects the liquidity
provision through the interbank market, in a moral hazard setting.

Proposition 7

(i) Liquidity shortfalls have a greater cost for the impaired bank if they come from the
asset side of the balance sheet, regardless of the degree of competition in the interbank
market:

CCD (x; RID (x)) > CCC (x; RIC (x)) , for all x.

21

Electronic copy available at: https://ssrn.com/abstract=3350684


(ii) If banks have constant marginal costs, dCi (xij )/dxij = c, and the competitive equi-
librium holds, CCj (xj ∗ ; RIj (x∗j )) > 0, then the interbank equilibrium with N banks
(x∗1j (N ), . . . , x∗N j (N )) always exists such that RIj (xj ∗ (N )) > RIj (x∗j ).

(iii) If banks have convex costs, d2 Ci (xij )/dx2ij > 0, and the competitive equilibrium holds,
CCj (xj ∗ ; RIj (x∗j )) > 0, then the interbank equilibrium (x∗1j (N ), . . . , x∗N j (N )) can be
ensured only if the number of banks satisfies N > Nj such that RIj (xj ∗ (N )) > RIj (x∗j ),
after a shock of size ρj for j = {C, D}.

(iv) A higher number of banks is necessary to restore the interbank market under a cash-flow
shock than under a deposit shock: NC > ND . Hence, the liquidity provision through the
interbank market is more sensitive to the level of competition when cash-flow shocks
occur.

Part (i) of Proposition 7 follows because the cost of interbank funds is independent of
the type of shock. Therefore, cash-flow shocks are costlier than deposit shocks regardless of
the level of competition in the interbank market.
The existence of a competitive equilibrium, CCj x∗j ; RIj (x∗j ) > 0, is a necessary condi-


tion for reaching an interbank equilibrium because the liquidity cost is the lowest possible
in a competitive market. Then, the more competitive the interbank market, the higher the
amount of liquidity that Bank A can gather to mitigate liquidity shocks, which reduces its
incentives to misbehave. Hence, if Bank A has incentives to misbehave in a competitive
market, then it will also misbehave in a less competitive market in which liquidity is costlier.
Part (ii) of Proposition 7 entails that banks always offer an amount of liquidity to restore
the interbank trading with a higher interbank rate, when they have constant marginal costs
and the competitive equilibrium holds. The presence of constant marginal is crucial to
restore the interbank trading, regardless of the number of banks in the market. A single
bank acting as liquidity provider is able to provide the competitive amount of liquidity with
non-negative profits. Hence, a lower level of competition leads to a higher interbank rate,
without collapsing the liquidity provision of the interbank market.
Part (iii) Proposition 7 indicates that a low level of bank competition may collapse the
interbank liquidity provision, when banks have convex costs, d2 Ci (xij )/dx2ij > 0, and the
competitive equilibrium holds. This result follows because a small number of banks do not
find it optimal to supply a huge amount of liquidity to induce Bank A to behave, since
the convexity of the cost function leads to negative profits. As a result, the interbank
market breaks down. The aggregate liquidity supply increases with the number of banks,

22

Electronic copy available at: https://ssrn.com/abstract=3350684


and interbank trading is restored with a higher interbank rate when the interbank market
becomes sufficiently competitive (when N > Nj ). Thus, interbank competition mitigates the
moral hazard problem, enhancing financial stability.
Finally, part (iv) of Proposition 7 states that bank competition affects more the interbank
trading under a cash-flow shock than under a deposit shock. Since cash-flow shocks involve
a greater cost for the impaired bank than deposits shocks do, a higher amount of liquidity
is necessary to avoid that Bank A misbehaves under a cash-flow shock. As a result, a higher
number of banks is necessary to restore the interbank market under a cash-flow shock than
under a deposit shock.

7.2 Credit market competition and financial stability


This subsection analyzes the effect of credit market competition on the ability of banks
to withstand liquidity shortfalls. As in Freixas and Ma (2014) and Carletti and Leonello
(2019), we consider that the cash flow of the loans of the banks, determined by R and Cf ,
is a sufficient statistic for the extent of competition in the credit market. We interpret a
reduction in both parameters as in increase in the level of credit market competition.
Credit market competition reduces the incentives of the bank facing liquidity shortfalls to
remove bad loans. This result arises for two reasons. First, credit market competition erodes
the payment of loans by reducing the banking spread. Further, more intense competition
implies a higher financing cost for banks. A lower cash flow Cf increases, in the interim stage,
the liabilities of the banks so that ∂Lj /∂Cf < 0. Accordingly, banks become more dependent
on external funding and their demand for deposit rises to fulfill their liabilities, which leads
to an increment in the deposits rate. Thus, competition results in greater financing costs for
banks, impairing the incentives of Bank A to behave prudently.
Credit market competition also raises the cost of interbank funds CIB (x) of bank B,
leading to a lower supply of funds in that market. The demand for deposits in Market B
increases because of the reduction in Cf . Hence, credit market competition undermines the
ability of interbank market to provide liquidity by both harming the incentives of Bank A
to remove bad loans and by raising the cost of interbank funds. The following proposition
summarizes these findings.

Proposition 8 An increase in credit market competition reduces the ability of the interbank
market to act as a liquidity provider because:

(i) Credit market competition increases the cost of interbank funds so that ∂ X̃ B (RI ; Cf )/∂Cf >

23

Electronic copy available at: https://ssrn.com/abstract=3350684


0, and it intensifies the dependence of the impaired bank on external funding in that
∂XjA (RI ; Cf )/∂Cf < 0;

(ii) Credit market competition exacerbates the moral hazard problem by eroding the margin
of profits of Bank A:
∂CCj (x; RI ; Cf ) ∂CCj (x; RI ; Cf )
> 0 and > 0.
∂Cf ∂R

Fig. 3 illustrates the impact of credit market competition on the interbank market. Fig.
3.(a) displays an interbank market equilibrium if a shock of type j occurs and the level of
credit market competition is low. The intersection between XjA and X̃ B occurs at the point
A below the curve CCj . Eq. (4) is satisfied, and Bank B provides an amount of liquidity x∗j
to Bank A in the interbank market.
Fig. 3.(b) shows that the same liquidity shock has different effects on financial stability as
the level of credit market competition increases. An increase in credit market competition
undermines the cash flow of banks A and B, making them more dependent on external
funding. This leads to an increase in the demand for deposits and, hence, to an increment
in the deposits rates.
Consequently, a higher deposits rate in Market A raises the demand for interbank funds
0
from XjA to XjA since Bank A is willing to substitute retail deposits for interbank loans, to
mitigate the impact of greater competition on deposit rates.
Credit market competition also raises the deposit rate in Market B. This implies an
0
upward shift of the supply in the absence of the moral hazard problem from X̃ B to X̃ B .
Hence, the intersection of these curves moves from A to A0 , implying a greater cost for Bank
A. Both the higher cost and the reduced cash flow of Bank A exacerbate the moral hazard
0
problem, and the compatibility constraint curve shifts downwards from CCj to CCj . As a
0 0 0
consequence, the intersection between XjA and X B lies above the curve CCj so that the
interbank market fails, harming financial stability.

8 Conclusions
This paper develops a model to explain the relationship between the type of liquidity shock
and the ability of the interbank market to provide liquidity. We employ a moral hazard
model with limited liability to study the mechanism by which liquidity shocks operate. A
key element in our model is that banks are monopolists in their own deposit markets, but
they may trade funds in the interbank market when a shock occurs.

24

Electronic copy available at: https://ssrn.com/abstract=3350684


Based on events that took place during the recent European financial crisis, banks may
suffer a liquidity shortfall due to a deposit flight or an increase in their NPLs. After a liquidity
shock, the impaired bank may access interbank funds to mitigate the shock. Nevertheless,
a problem of moral hazard arises because banks may be reluctant to remove bad loans from
their portfolio under a liquidity shortfall.
Our model implies that the strength of the balance sheet is a key determinant of the
market freeze, when banks undergo liquidity shocks. Even in the presence of an interbank
market, the effect on the financing cost of cash-flow shocks is greater than that of deposit
shocks. Unlike a deposit-flight shock, an increase in the NPLs reduces the “cash-in-hand”
of the banks, expanding their liabilities. Therefore, an increase in the NPLs leads to weaker
balance sheets than when a deposit flight occurs.
We analyze the role of interbank relationships in the liquidity provision, by extending
our model to a large number of banks under Cournot competition. The results of the base
model remain valid under this setting. Cash-flow shocks are costlier for the impaired bank
than deposit shocks, regardless of the degree of interbank competition.
We also study how both interbank and credit market competition affect financial stability.
Our model implies that interbank competition reduces the liquidity provision cost, enhancing
financial stability. Conversely, credit market competition decreases the cash flow of banks so
that they become more dependent to external funding. As a result, banks lower the supply
of interbank funding, undermining financial stability. Therefore, we show that competition
has a dual effect on financial stability.
A possible direction for future research is an empirical validation of our model. Our
theoretical results can be tested by regressing a measure of bank-individual risk on a mea-
sure of competition after controlling for the level of NPLs, bank-individual characteristics,
and macroeconomic variables. The sign of the coefficient of the measure of competition is
expected to be positive if the effect of credit market competition on risk-taking incentives
is stronger than that of interbank competition, in line with Beck et al. (2006), Beck et al.
(2013), and Jiménez et al. (2013). Otherwise, the expected sign of the coefficient of the mea-
sure of competition is expected to be negative, consistent with Boyd et al. (2007), Schaeck
and Cihák (2014), and Leroy and Lucotte (2017).
Our model can also be extended to a systemic perspective. Recent literature on contagion
shows that networking relationships help banks cope with liquidity shocks, increasing the
interdependence between financial institutions with a potential for contagion (Freixas et al.,
2000; Allen and Gale, 2004b; Upper and Worms, 2004; Elsinger et al., 2006; Nier et al., 2007;

25

Electronic copy available at: https://ssrn.com/abstract=3350684


Mistrulli, 2011; Grilli et al., 2015; Green et al., 2016; Temizsoy et al., 2017; Ahnert and
Georg, 2018; Biondi and Zhou, 2019). Many papers also highlight that the structure of the
interbank market determines the stability of the banking system (Boss et al., 2004; Müller,
2006; Degryse and Nguyen, 2007; Georg, 2013; Ladley, 2013; Capponi and Chen, 2015).
In addition, a strand of the literature provides evidence that the channels through which
shocks operate determine their propagation to the market (Dornbusch et al., 2000; Boyson
et al., 2010; Bai et al., 2012; Paltalidis et al., 2015; Shen and Li, 2019). Our model entails
that deposit and cash-flow shocks have distinct effects on the balance sheet of banks. Hence,
these shocks may propagate to other banks differently. Thus, further research might explain
the effect of different types of shocks on contagion in the banking system to complement this
field.

References
Acharya VV, Merrouche O (2013) Precautionary hoarding of liquidity and interbank markets:
Evidence from the subprime crisis. Review of Finance 17:107–160

Acharya VV, Mora N (2015) A crisis of banks as liquidity providers. The Journal of Finance
70:1–43

Acharya VV, Gale D, Yorulmazer T (2011) Rollover risk and market freezes. The Journal of
Finance 66:1177–1209

Acharya VV, Gromb D, Yorulmazer T (2012) Imperfect competition in the interbank market
for liquidity as a rationale for central banking. American Economic Journal: Macroeco-
nomics 4:184–217

Afonso G, Kovner A, Schoar A (2011) Stressed, not frozen: The federal funds market in the
financial crisis. The Journal of Finance 66:1109–1139

Ahnert T, Georg CP (2018) Information contagion and systemic risk. Journal of Financial
Stability 35:159–171

Aiyar S, Bergthaler W, Garrido J, Ilyina A, Jobst A, Kang K, Kovtun D, Liu Y, Monaghan


D, Moretti M (2015) A Strategy for Resolving Europe’s Problem Loans. Tech. rep., IMF
Staff Discussion Note SDN/15/19

26

Electronic copy available at: https://ssrn.com/abstract=3350684


Akins B, Li L, Ng J, Rusticus TO (2016) Bank competition and financial stability: Evidence
from the financial crisis. Journal of Financial and Quantitative Analysis 51:1–28

Allen F, Gale D (2000) Financial contagion. Journal of Political Economy 108:1–33

Allen F, Gale D (2004a) Competition and financial stability. Journal of Money, Credit and
Banking 36:453–480

Allen F, Gale D (2004b) Financial intermediaries and markets. Econometrica 72:1023–1061

Allen F, Carletti E, Gale D (2009) Interbank market liquidity and central bank intervention.
Journal of Monetary Economics 56:639–652

Allen F, Hryckiewicz A, Kowalewski O, Tümer-Alkan G (2014) Transmission of financial


shocks in loan and deposit markets: Role of interbank borrowing and market monitoring.
Journal of Financial Stability 15:112–126

Anginer D, Demirguc-Kunt A, Zhu M (2014) How does competition affect bank systemic
risk? Journal of Financial Intermediation 23:1–26

Ariss RT (2010) On the implications of market power in banking: Evidence from developing
countries. Journal of Banking & Finance 34:765–775

Baglioni A (2012) Liquidity crunch in the interbank market: Is it credit or liquidity risk, or
both? Journal of Financial Services Research 41:1–18

Bai J, Julliard C, Yuan K (2012) Eurozone sovereign bond crisis: Liquidity or fundamental
contagion. Federal Reserve Bank of New York Working Paper

Beck T, Demirgüç-Kunt A, Levine R (2006) Bank concentration, competition, and crises:


First results. Journal of Banking & Finance 30:1581–1603

Beck T, De Jonghe O, Schepens G (2013) Bank competition and stability: Cross-country


heterogeneity. Journal of Financial Intermediation 22:218–244

Berger AN, Klapper LF, Turk-Ariss R (2009) Bank competition and financial stability. Jour-
nal of Financial Services Research 35:99–118

Bhattacharya S, Gale D (1987) Preference shocks, liquidity, and central bank policy. In:
Barnett WA, Singleton K (eds) New Approaches to Monetary Economics: Proceedings of
the Second International Symposium in Economic Theory and Econometrics, New York:
Cambridge University Press, pp 69–88

27

Electronic copy available at: https://ssrn.com/abstract=3350684


Bibow J (2015) The Euro’s savior? Assessing the ECB’s crisis management performance
and potential for crisis resolution. Tech. rep., Levy Economics Institute of Bard Col-
lege Working Paper No. 845. Available at SSRN: https://ssrn.com/abstract=2671877 or
http://dx.doi.org/10.2139/ssrn.2671877

Biondi Y, Zhou F (2019) Interbank credit and the money manufacturing process: a sys-
temic perspective on financial stability. Journal of Economic Interaction and Coordination
14:437–468

Boissay F, Collard F, Smets F (2016) Booms and banking crises. Journal of Policial Economy
124:489–538

Bolton P, Santos T, Scheinkman J (2011) Outside and inside liquidity. The Quarterly Journal
of Economics 126:259–321

Boss M, Elsinger H, Summer M, Thurner 4 S (2004) Network topology of the interbank


market. Quantitative Finance 4:677–684

Boyd JH, De Nicoló G (2005) The theory of bank risk taking and competition revisited. The
Journal of Finance 60:1329–1343

Boyd JH, Nicoló GD, Jalal AM (2007) Bank risk-taking and competition revisited: New
theory and new evidence. Manuscript, Carlson School of Management, University of Min-
nesota

Boyson NM, Stahel CW, Stulz RM (2010) Hedge fund contagion and liquidity shocks. The
Journal of Finance 65:1789–1816

Brewer III E, Saidenberg MR (1996) Franchise Value, Ownership Structure, and Risk at
Savings Institutions. Tech. rep., Federal Reserve Bank of New York, Research Paper 9632

Bruche M, Suarez J (2010) Deposit insurance and money market freezes. Journal of Monetary
Economics 57:45–61

Brunnermeier MK (2009) Deciphering the liquidity and credit crunch 2007-2008. Journal of
Economic Perspectives 23:77–100

Cai R, Zhang M (2017) How does credit risk influence liquidity risk? Evidence from
Ukrainian banks. Visnyk of the National Bank of Ukraine 241:21–33

28

Electronic copy available at: https://ssrn.com/abstract=3350684


Capponi A, Chen PC (2015) Systemic risk mitigation in financial networks. Journal of Eco-
nomic Dynamics and Control 58:152–166

Carletti E (2008) Competition and regulation in banking. In: Thakor AV, Boot AW (eds)
Handbook of Financial Intermediation and Banking, Elsevier/North Holland, Amsterdam,
the Netherlands, pp 449–482

Carletti E, Leonello A (2019) Credit market competition and liquidity crises. Review of
Finance 23:855–892

Cocco JF, Gomes FJ, Martins NC (2009) Lending relationships in the interbank market.
Journal of Financial Intermediation 18:24–48

Cucinelli D (2013) The determinants of bank liquidity risk within the context of Euro area.
Interdisciplinary Journal of Research in Business 2:51–64

Danisman GO, Demirel P (2019) Bank risk-taking in developed countries: The influence of
market power and bank regulations. Journal of International Financial Markets, Institu-
tions and Money 59:202–217

De Nicoló G, Loukoianova E (2007) Bank ownership, market structure and risk. In: The
Mixing of Banking and Commerce, Federal Reserve of Chicago, Proceedings of the 43rd
Annual Conference on Bank Structure and Competition, pp 377–395

Degryse H, Nguyen G (2007) Interbank exposures: An empirical examination of contagion


risk in the Belgian banking system. International Journal of Central Banking 3:123–171

Demsetz RS, Saidenberg MR, Strahan PE (1996) Banks with something to lose: The disci-
plinary role of franchise value. Economic Policy Review 2:1–14

Diamond DW, Dybvig PH (1983) Bank runs, deposit insurance, and liquidity. The Journal
of Political Economy 91:401–419

Diamond DW, Rajan RG (2011) Fear of fire sales, illiquidity seeking, and credit freezes. The
Quarterly Journal of Economics 126:557–591

Dimitrios A, Helen L, Mike T (2016) Determinants of non-performing loans: Evidence from


Euro-area countries. Finance Research Letters 18:116–119

Dornbusch R, Park YC, Claessens S (2000) Contagion: Understanding how it spreads. The
World Bank Research Observer 15:177–197

29

Electronic copy available at: https://ssrn.com/abstract=3350684


Elsinger H, Lehar A, Summer M (2006) Risk assessment for banking systems. Management
Science 52:1301–1314

Freixas X, Ma K (2014) Banking Competition and Stability: The Role of Leverage. Tech.
rep., CEPR Discussion Papers 10121

Freixas X, Parigi BM, Rochet JC (2000) Systemic risk, interbank relations, and liquidity
provision by the central bank. Journal of Money, Credit and Banking 32:611–638

Gabrieli S (2011) The Functioning of the European Interbank Market during the 2007-08
Financial Crisis, CEIS Working Paper No. 158

Georg CP (2013) The effect of the interbank network structure on contagion and common
shocks. Journal of Banking & Finance 37:2216–2228

Goetz MR (2018) Competition and bank stability. Journal of Financial Intermediation 35:57–
69

Green C, Bai Y, Murinde V, Ngoka K, Maana I, Tiriongo S (2016) Overnight interbank


markets and the determination of the interbank rate: A selective survey. International
Review of Financial Analysis 44:149–161

Grigorian D, Manole V (2017) Sovereign risk and deposit dynamics: Evidence from Europe.
Applied Economics 49:2851–2860

Grilli R, Tedeschi G, Gallegati M (2015) Markets connectivity and financial contagion. Jour-
nal of Economic Interaction and Coordination 10:287–304

Gunther JW, Moore RR (2003) Loss underreporting and the auditing role of bank exams.
Journal of Financial Intermediation 12:153–177

Heider F, Hoerova M, Holthausen C (2015) Liquidity hoarding and interbank market spreads:
The role of counterparty risk. Journal of Financial Economics 118:336–354

Holmström B, Tirole J (2000) Liquidity and risk management. Journal of Money, Credit and
Banking 32:295–319

Huang R, Ratnovski L (2011) The dark side of bank wholesale funding. Journal of Financial
Intermediation 20:248–263

30

Electronic copy available at: https://ssrn.com/abstract=3350684


Jiménez G, Lopez JA, Saurina J (2013) How does competition affect bank risk-taking?
Journal of Financial Stability 9:185–195

Kabir MN, Worthington AC (2017) The ‘competition–stability/fragility’ nexus: A compara-


tive analysis of Islamic and conventional banks. International Review of Financial Analysis
50:111–128

Ladley D (2013) Contagion and risk-sharing on the inter-bank market. Journal of Economic
Dynamics and Control 37:1384–1400

Leroy A, Lucotte Y (2017) Is there a competition-stability trade-off in European banking?


Journal of International Financial Markets, Institutions and Money 46:199–215

Magnus M, Deslandes J, Dias C (2018) Non-performing loans in the Banking Union: Stock-
taking and challenges. Tech. rep., European Parliament PE 614.491

Makri V, Tsagkanos A, Bellas A (2014) Determinants of non-performing loans: The case of


Eurozone. Panoeconomicus 61:193–206

Martinez-Miera D, Repullo R (2010) Does competition reduce the risk of bank failure? The
Review of Financial Studies 23:3638–3664

Messai AS, Jouini F (2013) Micro and macro determinants of non-performing loans. Inter-
national Journal of Economics and Financial Issues 3:852–860

Mistrulli PE (2011) Assessing financial contagion in the interbank market: Maximum entropy
versus observed interbank lending patterns. Journal of Banking & Finance 35:1114–1127

Moro B (2014) Lessons from the European economic and financial great crisis: A survey.
European Journal of Political Economy 34:S9–S24

Müller J (2006) Interbank credit lines as a channel of contagion. Journal of Financial Services
Research 29:37–60

Nier E, Yang J, Yorulmazer T, Alentorn A (2007) Network models and financial stability.
Journal of Economic Dynamics and Control 31:2033–2060

Niinimaki JP (2012) Hidden loan losses, moral hazard and financial crises. Journal of Finan-
cial Stability 8:1–14

31

Electronic copy available at: https://ssrn.com/abstract=3350684


Ozili PK (2019) Non-performing loans and financial development: New evidence. The Journal
of Risk Finance 20:59–81

Paltalidis N, Gounopoulos D, Kizys R, Koutelidakis Y (2015) Transmission channels of


systemic risk and contagion in the European financial network. Journal of Banking &
Finance 61:S36–S52

Priftis R, Rousakis M (2017) Deposit flight, precautionary consumption, and capital con-
trols: A tale from Greece. Economics Series Working Papers 822, University of Oxford,
Department of Economics

Purnanandam A (2011) Originate-to-distribute model and the subprime mortgage crisis.


Review of Financial Studies 24:1881–1915

Roman A, Sargu AC (2015) The impact of bank-specific factors on the commercial banks liq-
uidity: Empirical evidence from CEE countries. Procedia Economics and Finance 20:571–
579

Schaeck K, Cihák M (2014) Competition, efficiency, and stability in banking. Financial


Management 43:215–241

Schaeck K, Cihak M, Wolfe S (2009) Are competitive banking systems more stable? Journal
of Money, Credit and Banking 41:711–734

Shen P, Li Z (2019) Financial contagion in inter-bank networks with overlapping portfolios.


Journal of Economic Interaction and Coordination pp 1–21

Sheng A (1996) Banking fragility in the 1980s: An overview. In: Sheng A (ed) Bank Re-
structuring: Lessons from the 1980s, Washington, D.C.: The World Bank, pp 5–24

Soedarmono W, Machrouh F, Tarazi A (2013) Bank competition, crisis and risk taking:
Evidence from emerging markets in Asia. Journal of International Financial Markets,
Institutions and Money 23:196–221

Temizsoy A, Iori G, Montes-Rojas G (2017) Network centrality and funding rates in the
e-MID interbank market. Journal of Financial Stability 33:346–365

Uhde A, Heimeshoff U (2009) Consolidation in banking and financial stability in Europe:


Empirical evidence. Journal of Banking & Finance 33:1299–1311

32

Electronic copy available at: https://ssrn.com/abstract=3350684


Upper C, Worms A (2004) Estimating bilateral exposures in the German interbank market:
Is there a danger of contagion? European Economic Review 48:827–849

Van Rixtel A, Gasperini G (2013) Financial Crises and Bank Funding: Recent Experience
in the Euro Area, BIS Working Paper No. 406

Vives X (1999) Oligopoly Pricing: Old Ideas and New Tools. MIT Press

Wagner W (2007) Aggregate liquidity shortages, idiosyncratic liquidity smoothing and bank-
ing regulation. Journal of Financial Stability 3:18–32

Whelan K (2014) TARGET2 and central bank balance sheets. Economic Policy 29:79–137

Yeyati EL, Micco A (2007) Concentration and foreign penetration in Latin American banking
sectors: Impact on competition and risk. Journal of Banking & Finance 31:1633–1647

Appendix

Technical Assumptions
We first present a set of technical assumptions under which our model is well specified. We
define an upper-bound of PL as follows.

Γ1 (L; ω)
P̄L ≡ .
R
The following assumption is necessary to ensure that the net present value of the invest-
ment of Bank A is negative when it does not restructure its portfolio.

Assumption 1
PL < P̄L .

We also need to impose a condition to ensure that if a bank behaves prudently, then its
investment always has a positive net present value. We define a lower bound for R as follows:

R ≡ Γ1C (L − α(ρ̄); ωC ) + τ.

Assumption 2
R ≥ R.

33

Electronic copy available at: https://ssrn.com/abstract=3350684


Let us define the threshold ρ̄¯ and the upper-bound ρmax for ρ as follows:

I +ω
ρ̄¯ ≡ Cf ,
I − Cf

and

ρmax ≡ min{ρ̄(θ), ρ̄¯},

where ρ̄(θ) is implicitly defined in the Proof of Proposition 1. The threshold ρ̄¯ avoids negative
cash flows in period 1. That is, if ρ ≤ ρ̄¯, then Cf ≥ α(ρ). Besides, the threshold ρ̄(θ)
guarantees that an increase in the NPLs is costlier than a deposit shock for the bank.
Proposition 4 does not hold when ρ > ρ̄(θ) because depositors become reluctant to renew
their deposits. Since there is a negative relationship between θ and the saving elasticity to
the deposit rate, then ρmax is equal to ρ̄(θ) only for small values of θ.

Assumption 3
ρ ≤ ρmax .

Finally, let ω be a lower bound of ω as follows:

ω ≡ L + ρ̄¯.

The following assumption is a sufficient condition to ensure that CjA (x) is a convex
function and π B (x) is a concave function of x. Thus, the second-order conditions for the
optimization problems of Bank A and Bank B are satisfied.

Assumption 4
ω ≥ ω.

Let Ω be the family of functions f (x) such that:


df (x)
(i) dx
> 0;
df (x) d2 f (x)
(ii) Either dx
= c, for all x, or dx2
> 0, where c > 0 is a constant;
df (0) df (0)
(iii) Γ1j (L, ωj ) > dx
, which ensures that RIj (0) > dx
.

34

Electronic copy available at: https://ssrn.com/abstract=3350684


The following assumption is a sufficient condition to ensure that RIj (xj ) is log-concave.

Assumption 5

(i) ω > max{ω C , ω}, where ω C ≡ θ + (1 − θ + θ2 )1/2 ;

(ii) All liquidity suppliers have the same cost function Ci (xij );

(iii) The cost function of liquidity suppliers Ci (xij ) belongs to the family of functions Ω;

(iv) The endowment of the depositors is I = 1.

Proofs

Proof of Proposition 1:

To prove that cash-flow shocks incur a higher cost than deposit shocks in autarky, we
proceed as follows. First, we derive a result that holds for several utility functions. This
result states that for small values of ρ, a cash-flow shock always generates a higher financial
cost than a deposit shock. Next, we find an upper-bound ρ̄(θ) such that if the preferences
of depositors are represented by the utility function of Eq. (1), then cash-flow shocks are
costlier than deposit shocks on the interval [0, ρ̄(θ)].

a) If the consumption and saving functions of depositors in the absence of shocks are
differentiable and satisfy
∂S1 (r̂1 ; ω) ∂C1 (r̂1 ; ω)
≥ > 0,
∂ω ∂ω
then it is a sufficient condition to ensure that a cash-flow shock always generates a higher
financial cost than a deposit shock for small values of ρ.

We know that Γ1 (s; ω) denotes the inverse-deposit supply if no shocks occurs in period
1. Hence, the bank’s cost of obtaining an amount of saving s = L from the deposit market
is:

Γ1 (L; ω)L.

Let us see how a deposit shock and a cash-flow shock affect such cost. A deposit shock
arises because of a sudden drop of ρ in the income of depositors. Hence, ωD ≡ ω − ρ is the

35

Electronic copy available at: https://ssrn.com/abstract=3350684


income of depositors after a shock of size ρ that reduces the deposit supply. Accordingly,
the cost of obtaining an amount of deposit L increases from Γ1 (L; ω)L to

Γ1D (L; ωD )L.

Conversely, a cash-flow shock does not affect the deposit supply, but it increases the
deposit needs of Bank A. According to Eq. (3), for any value of ρ, Bank A suffers a cash-
flow shock of size

α(ρ) ≡ L̂ − S1D (r̂1 ; ωD ).

Bank A must obtain the following amount of savings from the deposit market to keep
operating at t = 1:

LC (ρ) ≡ L + α(ρ).

Since we are analyzing the effect of small shocks, we define a linear approximation of LC (ρ)
at the point ρ = 0 as follows:

dα(0)
L̃C (ρ) ≡ L + ρ, (A.1)

where dα(0)/dρ = ∂S1D (r̂1 ; ω)/∂ω. Then, the financial cost of Bank A after suffering a small
cash-flow shock is given by:

Γ1C (L̃C (ρ); ω)L̃C (ρ).

To determine which shock is more adverse to the bank, we define the function

γ(ρ) ≡ Γ1D (LD ; ωD )LD − Γ1C (L̃C ; ωC )L̃C

as the difference between the financial costs produced by each type of shock. We show
now that this function is strictly decreasing in the neighborhood of ρ = 0, which implies
that cash-flow shocks are costlier than deposit shocks when ρ is small. To this purpose, we
differentiate γ(ρ) with respect to ρ, and we evaluate it at ρ = 0 as follows:

36

Electronic copy available at: https://ssrn.com/abstract=3350684


" #
dγ(0) ∂Γ1D (L; ωD (0)) ∂ωD (0) ∂ L̃C (0) ∂Γ1C (L̃C (0); ωC )
= L− L̃C (0) + Γ1C (L̃C (0); ωC ) .
dρ ∂ω ∂ρ ∂ρ ∂s

As ∂ω(ρ)/∂ρ = −1, ωD (0) ≡ ωC = ω, and L̃C (0) = L, it follows that

∂Γ1D (L; ωD (0)) ∂Γ1 (L; ω)


=
∂ω ∂ω
and
∂Γ1C (L̃C (0); ωC ) ∂Γ1 (L; ω)
= .
∂s ∂s
Hence, dγ(0)/dρ < 0 if and only if

" #
∂Γ1 (L; ω) ∂Γ1 (L; ω) ∂ L̃C (0) ∂ L̃C (0)
L + + Γ1C (L̃C (0); ωC ) > 0.
∂ω ∂s ∂ρ ∂ρ

We can also rewrite the slope of Γ1 (s; ω) with respect to s as

∂Γ1 (s; ω) 1
= ∂S1 (r1 ;ω)
,
∂s
∂r1

and we can rewrite the slope of S1 (r1 ; ω) with respect to the interest rate as

∂S1 (r1 ; ω) ∂C1 (r1 ; ω)


=− .
∂r1 ∂r1
Then,

" #
∂Γ1 (L; ω) 1 ∂ L̃C (0) ∂ L̃C (0)
L − ∂C1 (r̂1 ;ω)
+ Γ1C (L̃C (0); ωC ) > 0.
∂ω ∂ρ ∂ρ
∂r1

Moreover, from Eq. (A.1), we have that ∂ L̃C (ρ)/∂ρ = ∂S1 (r̂1 ; ω)/∂ω. Hence,

" #
∂Γ1 (L; ω) 1 ∂S1 (r̂1 ; ω) ∂S1 (r̂1 ; ω)
L − ∂C1 (r̂1 ,ω)
+ Γ1C (L̃C (0); ωC ) > 0. (A.2)
∂ω ∂ω ∂ω
∂r1

Since the consumption at period t is a normal good, then

37

Electronic copy available at: https://ssrn.com/abstract=3350684


∂C1 (r̂1 ;ω)
∂Γ1 (L; ω) ∂ω
= ∂C1 (r̂1 ;ω)
< 0,
∂ω
∂r1

and, therefore, the inequality (A.2) can be written as:

" ∂C #
1 (L;ω) ∂S1 (r̂1 ;ω)
∂ω ∂ω ∂S1 (r̂1 ; ω)
L ∂C1 (r̂1 ,ω)
− ∂C1 (r̂1 ;ω)
+ Γ1C (L̃C (0); ωC ) > 0.
∂ω
∂r1 ∂r1

Hence, the inequality (A.2) holds if

∂S1 (r̂1 ; ω) ∂C1 (r̂1 ; ω)


≥ > 0,
∂ω ∂ω
and dγ(0)/dρ < 0. As a result, a cash-flow shock is more adverse than a deposit-flight shock
for small values of ρ.

b) If the utility function of consumers is given by Eq. (1), then γ(ρ) ≤ 0 for any ρ ∈
[0, ρ̄(θ)]. Hence, a cash-flow shock generates a greater financial cost than a deposit shock.

If the preferences of depositors are represented by Eq. (1), then

  θ−1   θ−1
ω − ρ + Cf θ ω + Cf − α(ρ) θ
γ(ρ; θ) = (I − Cf ) − (I − Cf + α(ρ)),
I − Cf I − Cf + α(ρ)
| {z } | {z }
Γ1D (LD ; ωD )LD Γ1C (LC ; ωC )LC

where γ(ρ; θ) is a convex function because


 − θ1  − θ1 
ω−ρ+Cf ω+Cf −α(ρ) 2
d2 γ(ρ; θ)

1−θ  I−Cf I−Cf +α(ρ) ω+I
= −  > 0,

dρ 2 θ 2 } ω − ρ + Cf

(ω + Cf − α(ρ)) I +ω+ρ
| {z
+ | {z }
+

where α(ρ) = [ρ(I − Cf )]/(I + ω). We also know by item (a) that γ(ρ) is negative around
ρ = 0. We can solve the equation γ(ρ̄; θ) = 0 for θ to implicitly define the threshold ρ̄(θ):

ln(I − Cf + α(ρ̄)) − ln(I − Cf )


θ(ρ̄) ≡ 1 − = 1 − H(ρ̄). (A.3)
ln(ω + Cf − α(ρ̄)) − ln(ω − ρ̄ + Cf )
| {z }
H(ρ̄) > 0

38

Electronic copy available at: https://ssrn.com/abstract=3350684


Since the denominator of H(ρ̄) increases faster than its numerator with ρ̄, then H(ρ̄) is
a decreasing function and the implicit function ρ̄(θ) is well defined. Finally, as γ(ρ; θ) is
continuous and negative for small values of ρ, we have that γ(ρ; θ) ≤ 0 if ρ ∈ [0, ρ̄(θ)].
Besides, the threshold ρ̄(θ) increases with θ because

dρ̄(θ) 1
= dH(ρ̄) > 0,
dθ − dρ̄

as dθ(ρ̄)/dρ̄ = −dH(ρ̄)/dρ̄ > 0. In addition, we have that

 
∂α(ρ̄)
dθ(ρ̄; Cf ) µ1 1 1− ∂Cf
=− 2 −  < 0, (A.4)
d Cf µ2 ω − ρ̄ + Cf ω + Cf − α(ρ̄)
| {z }
+

where for simplicity we write

µ1 ≡ ln(I − Cf + α(ρ̄)) − ln(I − Cf ) > 0,

and

µ2 ≡ ln(ω − ρ̄ + Cf ) − ln(ω − α(ρ̄) + Cf ) < 0.

We apply this result in the Proofs of Propositions 2, 3, and 4 since it implies that the
threshold ρ̄ increases with Cf , for a given θ. 

Proof of Lemma 1:

Lemma 1 states that if the moral hazard problem does not arise, then there is a unique
interbank rate R̃Ij such that XjA (R̃Ij ) = X̃ B (R̃Ij ) = x̃j and, hence, this market clears.
To prove this result, let λj (x) for j = {C, D} be the difference between dπ B (x)/dx and
dCjA (x)/dx as follows:

∂Γ1 (sB (x); ω) ∂sB (x) B


λj (x) ≡ − s (x) − Γ1 (sB (x); ω) + Γ1j (sA
j (x); ωj )
∂s ∂x
∂Γ1j (sA A
j (x); ωj ) ∂sj (x) A
− s (x),
∂s ∂x j
where ∂sA B
j (x)/∂x = −1 and ∂s (x)/∂x = 1.

39

Electronic copy available at: https://ssrn.com/abstract=3350684


There is an excess of demand for interbank funds over supply if λj (x) < 0; otherwise,
there is an excess of supply over demand. Since λj (x) is a decreasing function of x, we
apply the Bolzano’s Theorem to prove the existence and uniqueness of x̃j , which satisfies
λj (x̃j ) = 0. For this purpose, we show that λj (0) > 0 and λj (Lj ) < 0, for all ρ > 0.
Let us see that λD (0) > 0 and λD (LD ) < 0 under a deposit shock. If x = 0, it follows
that

 
 ∂Γ1D (sA B B
D (0); ωD ) ∂Γ1 (s (0); ω) 

λD (0) = Γ1D (sA
D (0); ωD ) − ΓB B
(s (0); ω) +  −  L > 0,
| {z 1 } | ∂s {z ∂s }
+
+

where sA B
D (0) = LD , s (0) = L, and LD = L. The first term of the right-hand side of the
equation above is positive because ωD < ωC so that Γ1D (L; ωD ) > Γ1 (L; ω). In addition, the
term inside brackets is given by:

" − θ1  − θ1 #
1−θ ω − ρ + Cf ω + Cf
(I + ω − ρ) − (I + ω) > 0,
θ(I − Cf )2 I − Cf I − Cf
| {z } | {z }
+ +

which is positive for any ρ > 0. Hence, it follows that λD (0) > 0. If Bank A meets all its
liabilities from the interbank market x = LD , then

λD (LD ) = Γ1D (sA B


D (LD ); ωD ) −Γ1 (s (LD ); ω) +
| {z }
=0
| {z }

 
 ∂Γ1D (sA
D (LD ); ωD ) A ∂Γ1 (sB (LD ); ω) B
sD (LD ) − s (LD ) < 0,

∂s ∂s

| {z }
=0
| {z }

where sA B
D (LD ) = 0. Moreover, s (LD ) = 2L because Bank B needs to renew its deposits by an
amount L, and it lends L in the interbank market. Since Γ1j (0; ωj ) = 0, ∂Γ1j (0; ωj )/∂s = 0,
and ∂ΓB B
1 (s (Lj ); ω)/∂s > 0 for j = {C, D}, we have that λD (LD ) < 0.
Under a cash-flow shock, if x = 0, then

40

Electronic copy available at: https://ssrn.com/abstract=3350684


 
 ∂Γ1C (sA
C (0); ωC ) A ∂Γ1 (sB (0); ω) B 
λC (0) = Γ1C (sA
C (0); ω C ) − Γ 1 (s B
(0); ω) +  s C (0) − s (0) > 0,
| {z } | ∂s {z ∂s }
+
+

where sA
C (0) = LC . The first term of the right-hand side of the equation above is positive
because LC > L implies Γ1C (LC ; ωC ) > Γ1 (L; ω). Moreover, the term inside brackets is given
by:

" − 1  − θ1 #
(1 − θ)(I + ω) ω + Cf − α(ρ) θ I − Cf ω + Cf
− > 0,
θ(I − Cf ) I − Cf + α(ρ) I − Cf + α(ρ) I − Cf
| {z }| {z }
+ +

which is positive for any ρ > 0. As a result, λC (0) > 0. If x = LC , we have that

λD (LC ) = Γ1C (sA B


C (LC ); ωC ) −Γ1 (s (LC ); ω) +
| {z }
=0
| {z }

 
 ∂Γ1C (sA
C (LC ); ωC ) A ∂Γ1 (sB (LC ); ω) B
sC (LC ) − s (LC ) < 0,

∂s {z ∂s

| }
=0
| {z }

where sA B
C (LC ) = 0 and s (LC ) = L + LC since Bank B needs to gather L + LC deposits
from the market (L units to renew its own liabilities and LC units to lend in the interbank
market).
Since λj (0) > 0 and λj (Lj ) < 0, there is a unique x̃j ∈ (0, Lj ) such that λj (x̃j ) = 0.
Finally, the clearing interbank rate R̃Ij is given by:


∂Γ1 sB (x̃j ); ω B
R̃Ij = s (x̃j ) + Γ1 (sB (x̃j ); ωC ) , (A.5)
∂s
which is unique because R̃Ij is an increasing function of x. 

Proofs of Propositions 2 and 3:

We prove Propositions 2 and 3 in two steps. First, we show that a cash-flow shock is
still costlier than a deposit shock for the bank, under the presence of an interbank market.

41

Electronic copy available at: https://ssrn.com/abstract=3350684


In a second step, as cash-flow shocks incur a greater financial cost than deposit shocks do,
then the compatibility constraint of Definition 1 holds under both type of shocks only if
ρ ∈ [0, ρc ] . Nevertheless, if ρ ∈ (ρc , ρD ], then this compatibility constraint is satisfied only
when a deposit shock occurs.

Step I: Under the presence of an interbank market, cash-flow shocks are still costlier than
deposit shocks.

As defined before, CjA (x; RI ) is the total cost of a shock of type j = {C, D} when the
impaired bank replaces retail deposits with an amount x of interbank loans. Then, let Ψ(x)
A
be the difference between CD (x; RI ) and CCA (x; RI ) as follows:

Ψ(x; RI ) ≡ Γ1D sA
 A  A
 A 
D (x); ωD sD (x) + RI x − Γ1C sC (x); ωC sC (x) + RI x ,
| {z } | {z }
A (x; R )
CD A (x; R )
CC
I I

which can be written as

C̃f
z }| { ! θ−1 C̃f
θ
ω − ρ + (Cf + x) z }| {
Ψ(x; RI ) = (I − (Cf + x))
I − (Cf + x)
| {z }
C̃f
| {z }
Γ1D (sA A
D (x); ωD )sD (x)

C̃f
z }| { ! θ−1 C̃f
θ
ω − α(ρ) + (Cf + x) z }| {
− (I + α(ρ) − (Cf + x)) . (A.6)
I + α(ρ) − (Cf + x)
| {z }
C̃f
| {z }
Γ1C (sA A
C (x); ωC )sC (x)

Now, let us show that Ψ(x; RI ) < 0 for any x > 0. This is an instrumental result to
prove that CCA (x∗C ; RIC
∗ A ∗
) > CD ∗
(xD ; RID ). By defining C̃f ≡ Cf + x > Cf , then Ψ(x; RI )
approximates the function γ(ρ) when the cash flow is equal to C̃f instead of Cf . Recall that
γ(ρ) is the difference between the financial costs produced by each type of shock in autarky.
However, Ψ(x; RI ) differs from γ(ρ) because α(ρ) = ρ(I − Cf )/(I + ω) cannot be rewritten
as a function of C̃f . By Proposition 1, it follows that in autarky (x = 0)

γ(ρ; C̃f ) = Γ1D s̃A


 A A
 A
D (0); ωD s̃D (0) − Γ1C s̃C (0); ωC s̃C (0) < 0,

42

Electronic copy available at: https://ssrn.com/abstract=3350684


where s̃A A
D (0) = I − C̃f , s̃C (0) = I − C̃f + α̃(ρ), and α̃(ρ) ≡ ρ(I − C̃f )/(I + ω). Since α(ρ) is
decreasing in Cf and C̃f > Cf , we have that α̃(ρ) < α(ρ). Therefore, the following inequality
holds:

  θ−1
ω − ρ + Cf + x θ
0 > γ(ρ; C̃f ) > (I − (Cf + x))
I − (Cf + x)
| {z }
Γ1D (sA A
D (x); ωD )sD (x)
  θ−1
ω + (Cf − α(ρ) + x) θ
− (I − (Cf − α(ρ) + x)) = Ψ(x; RI ). (A.7)
I − (Cf − α(ρ) + x)
| {z }
Γ1C (sA A
C (x); ωC )sC (x)

The inequality (A.7) follows because CCA (x; RI ) increases with α(ρ). Thus, CCA (x; RI ) >
A
CD (x; RI ). Finally, let us show that CCA (x∗C ; RIC
∗ A ∗
) > CD ∗
(xD ; RID ). We know that

Ψ(x∗C ; RIC
∗ A ∗
) = CD ∗
(xC ; RIC ) − CCA (x∗C ; RIC

) < 0.

Under a deposit shock, the financial cost of the bank is minimized at x∗D = XD ∗
(RID ).
A ∗ ∗ A ∗ ∗
Hence, CD (xD ; RID ) < CD (xC ; RIC ). Therefore,

A ∗ ∗
CD (xD ; RID ) < CCA (x∗C ; RIC

),

which implies that cash-flow shocks are still costlier than deposit shocks when the impaired
bank obtains funding from the interbank market.

Step II: Existence of different equilibria

As discussed before, Bank A can access interbank funds only if the conditions (1) and
(2) of Definition 1 are simultaneously satisfied. From Lemma 1, condition (1) always holds.
Hence, the compatibility constraint (condition (2)) determines whether Bank A can get
external financing from the interbank market under any type of shock.
Let ρj be the highest value of ρ such that the compatibility constraint of Definition 1
holds if a shock of type j = {D, C} occurs. The upper bounds for ρ are implicitly defined
from the incentive-compatibility constraint of Eq. (4) as follows:

τ ∗ ∗ ∗ ∗
= R − Γ1C (sA A
C (xC ); ωC )sC (xC ) − RIC xC ,
∆P | {z }
A (x∗ ; R∗ )
CC C IC

43

Electronic copy available at: https://ssrn.com/abstract=3350684


and

τ ∗ ∗ ∗ ∗
= R − Γ1D (sA A
D (xD ); ωD )sD (xD ) − RID xD .
∆P | {z }
A (x∗ ; R∗ )
CD D ID

By Step I, cash-flow shocks are costlier than deposit shocks so that ρC < ρD . Hence,
the banking system reaches an AIE if ρ ∈ [0, ρC ] because the compatibility constraint holds
for both type of shocks. Nevertheless, if ρ > ρC , then the compatibility constraint is only
satisfied under a deposit shock, implying that a ME arises when ρ ∈ (ρC , ρD ]. Finally, the
interbank market freezes under both type of shocks if ρ > ρD . 

Proof of Proposition 4:

Before proceeding to the Proof of Proposition 4, we derive two auxiliary results. Facts
1-2 below ensure the existence and uniqueness of the Cournot equilibrium in the absence
of a moral hazard problem. Part (ii) of Fact 2 also implies that the incentive-compatibility
constraint CCj (xj , RIj (xj )) is increasing in N , for a liquidity shock ρj , because the interbank
rate is decreasing with the number of banks.

Fact 1:

(i) RIj (xj ) is log-concave;


dRIj (xj )
(ii) dxj
< 0;

(iii) There exists a value of x̄j such that RIj (x̄j ) = 0.

Proof of Fact 1:

Part (i) follows by differentiating log(RIj (xj )) twice with respect to xj :


" #
2 3
d log (RIj (xj )) 1 θ−1 1 θ
2
= 2 + 2 − 2 < 0.
dxj θ sj (xj )
A A
ω + I − sj (xj ) ω + I − θsA j (xj )

By Assumption 5-(iv), the endowment level is I = 1 so that sA


j (xj ) ∈ [0, 1]. Then, we
can rewrite the terms in brackets as follows:
θ−1 1 θ3
H sA

j (xj ); ω; θ ≡ 2 + 2 − 2 ,
sA
j (x j ) ω + 1 − s A
j (x j ) ω + 1 − θs A
j (x j )

44

Electronic copy available at: https://ssrn.com/abstract=3350684


where

∂H sA
j (xj ); ω; θ θ2
 
1
=2 − < 0,
∂ω 1 + ω − θsA
j (xj ) 1 + ω − sA
j (xj )

and
 

∂H sA 4
 
j (xj ); ω; θ  1−θ 1 θ 
A
= 2 3 + A
− A
 > 0,
∂sj (xj ) 1 + ω − s (x ) 1 + ω − θs (x )

 sA (x )
 j j

j j j j
| {z }
+

since 0 < θ < 1 and θsA A


j (xj ) < sj (xj ). Then,

1 θ3
H(y, ω, θ) 6 H(1, ω, θ) = − (1 − θ) − ,
ω2 (1 − θ + ω)
where H(1, ω, θ) is strictly decreasing in ω. Hence,

H(y; ω; θ) 6 H(1; ω; θ) < 0,

for any ω ≥ ω C = θ + (1 − θ + θ2 )1/2 , where ω C satisfies H(1, ω C , θ) = 0. Thus, part (i) of


Fact 1 follows by applying Assumption 5-(i).
Assumption 5-(i) implies part (ii) of Fact 1. Given the interbank rate RIj (·) of Eq. (2),
we have that:
1+θ
 1−θ
dRIj (xj ) (I + ωj )2 Γ1j sA
j (xj )
= (θ − 1) 3 < 0.
dxj | {z } θ 2 sA
j (xj )
− | {z }
+

Finally, part (iii) follows by solving RIj (xj ) = 0 on Eq. (2):


ωj + I − θL
xj = .
θ


Fact 2

PN
(i) Let (x̂∗1j (N ), . . . , x̂∗N j (N )) and x̂j = i=1 x̂∗ij (N ) be the Cournot equilibrium and the
total amount of liquidity supplied in the market in the absence of moral hazard, respec-
tively. Then, for any ρj , the equilibrium (x̂∗1j (N ), . . . , x̂∗N j (N )) exists and is unique;

45

Electronic copy available at: https://ssrn.com/abstract=3350684


(ii) The Cournot equilibrium tends to the competitive one as the number of banks in the
interbank market increases to infinity. Formally,

∂ x̂∗j (N ) ∂ x̂∗ij (N ) ∂πij x̂∗j (N ) dRIj x̂∗j (N )


 
> 0, < 0, < 0, and < 0.
∂N ∂N ∂N dN
Moreover, limN →∞ x̂∗j (N ) = x̂∗j and RIj (x̂∗j ) = dCi (x̂∗ij )/dxij , where x̂∗j is the competi-
tive equilibrium;

(iii) The incentive-compatibility constraint CCj x∗j (N ), RIj (x∗j (N )) is increasing in the


number of banks N .

Proof of Fact 2:

According to Vives (1999), the log-concavity of RIj (xj ) together with

d2 Ci (xij ) dRIj (xj )


− > 0, for all i,
dx2ij dxij

are sufficient conditions for the existence and uniqueness of the Cournot equilibrium in
the absence of a moral hazard problem. Besides, Vives (1999) shows that Assumption 5
and Fact 1 are sufficient conditions for part (ii) of Fact 2. Finally, part (ii) of Fact 2 and
∂CCj (xj , RIj (xj ))/∂RIj < 0 entail that CCj (x∗j (N ), RIj (x∗j (N ))) is increasing in the number
of banks N . 

The Proof of Proposition 7 follows from Facts 1-2. Part (i) of Proposition 7 is straightfor-
ward. Proposition 4 states that cash-flow shocks are costlier than deposit shocks in autarky.
The cost of interbank funds Ci (xij ) is independent of shocks; then, the marginal cost of pro-
viding liquidity in the interbank market is the same under both types of shocks. Therefore,
cash-flow shocks are costlier than deposit shocks regardless of the level of competition in the
interbank market, and part (i) of Proposition 7 holds.
To prove parts (ii)-(iii), we need to distinguish two cases determined by the compatibility
constraint CCj (xj ; RIj (xj )) > 0. If x̂j (N ) satisfies CCj (x̂j (N ); RIj (x̂j (N ))) > 0, then part
(i) of Fact 2 implies that the interbank equilibrium is given by x∗j (N ) = x̂∗j (N ), for all i,
because there is no moral hazard problem in the market.
Nevertheless, if CCj (xˆj (N ); RIj (x̂j (N ))) < 0, then x̂∗j (N ) cannot be an equilibrium be-
cause Bank A has incentives to misbehave. In this case, the existence of a competitive
equilibrium CCj (x̂∗j ; RIj (x̂∗j )) > 0 is a necessary condition to ensure an interbank equilib-
rium with N banks. If CCj (x̂∗j ; RIj (x̂∗j )) > 0, then there exists a value x0j ∈ (x̂∗j (N ), x̂∗j ]

46

Electronic copy available at: https://ssrn.com/abstract=3350684


such that CCj (x0j ; RIj (x0j )) = 0. The existence of x0j follows because CCj (xj ; RIj (xj )) is
continuous and increasing in xj . Hence, x0ij for all i is a possible interbank equilibrium.
A sufficient condition for the existence of an interbank equilibrium follows when banks
have constant marginal costs or when a high number of banks have convex costs. For
simplicity, we consider only a symmetric equilibrium.
We first show that the competitive equilibrium exists with CCj (x̂j ; RIj (x̂j )) > 0 >
CCj (x̂∗j (N ); RIj (x̂∗j (N ))) when banks have constant marginal costs. Let us show that x0ij =
x0j /N , for all i, is an equilibrium. As x0j > x̂∗j (N ), then no bank supplies an amount of
liquidity higher than x0ij because the first-order condition evaluated at x0ij in Eq. (5) does
not hold, leading to ∂πijB (x0ij , x0−ij )/∂xij < 0. A bank may reduce the supplied liquidity from
x0ij to x xij , x0−ij )/∂xij = 0. This reduction in
eij to increase the interest rate such that ∂πijB (e
the supplied liquidity, however, leads Bank A to misbehave because x0j > m6=i x0mj + x
P
eij ,
implying that
!!
X X
CCj x0mj + x
eij ; RIj x0mj + x
eij < 0.
m6=i m6=i

The existence of a competitive equilibrium implies that πijB (x0i , x0−i ) > πijB (x̂∗i , x̂∗−i ) = 0.
Thus, no bank has incentives to deviate from x0ij . As a result, part (ii) of Proposition
7 holds; if banks have constant marginal costs and the competitive equilibrium satisfies
CCj (x̂∗j ; RIj (x̂∗j )) > 0, then x∗ij (N ) = x0ij , for all i, is a Cournot equilibrium in the interbank
market with x̂∗j > x0j > x∗j (N ).
To prove part (iii), suppose now that banks have convex costs, d2 Ci (xij )/dx2ij > 0, and
the competitive equilibrium holds,

CCj x̂∗j ; RIj (x̂∗j ) > 0 > CCj x̂∗j (N ); RIj (x̂∗j (N )) .
 

If the competitive equilibrium holds, parts (ii)-(iii) of Fact 2 imply that there is a number
of banks N j such that x̂∗j (N j ) = x0j . As a result, for all N > N j , we have that

CCj x̂∗j (N ); RIj (x̂∗j (N )) = 0 and πijB x0ij , x0−ij > 0.


 

Then, part (iii) of Proposition 7 follows, and the equilibrium in the interbank market is
x∗ij (N ) = x̂∗ij (N ) for all i and for all N > N j .
Finally, part(i) of Proposition 4 together with part (ii) of Fact 1 imply that x0C > x0D .
Under a cash-flow shock, a higher amount of liquidity is necessary to avoid that Bank A
misbehaves than under a deposit shock, which entails that N C > N D . Then, part (iv)

47

Electronic copy available at: https://ssrn.com/abstract=3350684


of Proposition 7 holds, and the liquidity provision through the interbank market is more
sensitive to the level of competition when cash-flow shocks occur. 

Proof of Proposition 5:

Before proceeding to the proof of the Proposition 5, we derive the following auxiliary
result.

Fact 3:

(a) The objective functions of Bank A and Bank B are convex and concave, respectively.
Then, the second-order conditions for an interior solution of xA B
j and x are satisfied.

∂π B (xB ;Cf ) ∂CjA (xA ;Cf )


(b) ∂x∂Cf
> 0 and ∂x∂Cf
> 0.

∂Γ1j (sA A
j (x);Cf ) ∂sj (x)
(c) ∂s ∂Cf
< 0.

Proof of Fact 3:

(a) The second derivative of CjA (x) with respect to x is:

∂ 2 CjA (x) ∂ 2 Γ1j (sA A


j (x); ωj ) ∂sj (x) A ∂Γ1j (sA A
j (x); ωj ) ∂sj (x)
= − s (x) − 2
∂x2 ∂s2 ∂x j ∂s ∂x
∂ 2 Γ1j (sA
j (x); ωj ) A ∂Γ1j (sA
j (x); ωj )
= s j (x) + 2 ,
∂s2 ∂s
since ∂sA
j (x)/∂x = −1. Then,

 − θ1
ωj
∂ 2
CjA (x) (1 − θ) ωj2 sA (x) −1
j
= 2  > 0,
∂x2 θ2 sj (x) ωj − sA
A
j (x)

In particular, ∂ 2 CjA (xA 2


j )/∂x > 0 implies that the second-order condition of the problem
of Bank A is satisfied. Given ∂sB (x)/∂x = 1, we have that π B (x) is a concave function
because

48

Electronic copy available at: https://ssrn.com/abstract=3350684


∂ 2 π B (x) ∂ 2 Γ1 (sB (x); ω) B ∂Γ1 (sB (x); ω)
 
=− s (x) + 2
∂x2 ∂s2 ∂s
 
 − θ1
 (1 − θ) ω 2 ω
 sB (x)
− 1 

= −
 θ2 (sB (x))2 (ω − sB (x))  < 0.

| {z }
+

Hence, the second-order condition of the maximization problem of Bank B is also satisfied.

(b) From the maximization problem of Bank A, we have that

∂ 2 CjA (x; Cf ) ∂ 2 Γ1j (sA A


j (x); ωj ) ∂sj (x) A ∂Γ1j (sA A
j (x); ωj ) ∂sj (x)
=− s (x) − 2
∂x∂Cf ∂s2 ∂Cf j ∂s ∂Cf
∂ 2 Γ1j (sA
j (x); ωj ) A ∂Γ1j (sA
j (x)); ωj )
= 2
s j (x) + 2 > 0,
∂s ∂s
since ∂sA
j (x)/∂Cf = −1. From the maximization problem of Bank B, it follows that

∂ 2 π B (x) ∂ 2 Γ1 (sB (x); ω) ∂sB (x) B ∂Γ1 (sB (x), ω) ∂sB (x)
=− s (x) − 2
∂x∂Cf ∂s2 ∂Cf ∂s ∂Cf
∂ 2 Γ1 (sB (x); ω) B ∂Γ1 (sB (x); ω)
= s (x) + 2 > 0,
∂s2 ∂s
because ∂sB (x)/∂Cf = −1.

(c) Finally, we have that

− 1 
∂Γ1j (sA A
j (x); Cf ) ∂sj (x)
 
θ − 1 ωj + Cf + x θ ωj + I
= < 0.
∂s ∂Cf θ sA
j (x) (sA
j (x))
2

According to Fact 3, the Proof of Proposition 5 is straightforward:

(i) By applying the implicit function theorem, we have that

∂ 2 CjA (xA ;Cf )


∂XjA (RI ; Cf ) ∂x∂C
= − ∂ 2 C A (xAf;C ) < 0,
∂Cf j f
∂x2

49

Electronic copy available at: https://ssrn.com/abstract=3350684


and

∂ 2 π B (xB )
∂ X̃ B (RI ; Cf ) ∂x∂C
= − ∂ 2 πB (xfB ) > 0.
∂Cf 2
∂x

Hence, credit market competition increases the cost of interbank funds, and it intensifies
the dependence of the impaired bank on external funding.

(ii) The effect of credit market competition on the incentive-compatibility constraint is


given by:

∂CCj (x; RI ; Cf )
= 1,
∂R

and

 
 ∂Γ (sA (x), ω ) ∂sA (x) ∂sA
∂CCj (x; RI ; Cf ) j (x) 

 1j j j j
= − +Γ1j (sA
j (x), ωj )  > 0,
∂Cf  ∂s ∂Cf ∂Cf 
| {z }

as ∂sA
j (x)/∂Cf = −1. The sign of the derivative above implies that credit market
competition leads to a downward shift in the curve CCj (x). Therefore, credit market
competition exacerbates the moral hazard problem in the interbank market. 

50

Electronic copy available at: https://ssrn.com/abstract=3350684


(a) Low level of credit market competition

(b) High level of credit market competition

Fig. 3 The impact of credit market competition on the interbank market

51

Electronic copy available at: https://ssrn.com/abstract=3350684

You might also like