1

Macro-prudential policy
~examination of Basel Capital Accords~

14110064 ID

Abstract
We incorporate the banks, defined as maturity-mismatching financial
intermediaries by Diamond and Rajan (2001a,2012), into an
overlapping-generations model where capital good is reproducible. We show
that, in our model, the laissez-faire banks take on undue risks, compared to
the social optimum, owing to the pecuniary externalities. Further, the model
replicates rare but severe crises without assuming any large exogenous
shocks because systemic bank runs take place endogenously followed by
sharp contractions in output. We also make policy assessment PCA based
on the model. We discuss about the efficiency of introducing PCA, then we
deliberate whether strengthening the required minimum level of capital is
effective for stability of the banking system or not.

1,Introduction
Since the global financial crisis for 2007-08, one of the challenges posed for
macroeconomists has been how to replicate systemic financial crises and
ensuring sharp contraction in macroeconomic activity in dynamic stochastic
general equilibrium (DSGE) models. The 2007-08 systemic financial crisis
could be interpreted as an unavoidable and unfortunate accident arising
from tail risks. In the meantime, Bernanke(2012) argues that the triggers of
the 2007-08 crisis were quite modest in size while heavy dependence on
short-term funding, high leverage, and inadequate risk management in the
(shadow) banking sector that engaged in maturity transformation was key
vulnerability of the system for the devastating outcomes of the crisis. In a
related context of the underlying vulnerabilities, others argue that there
might have been erosion of discipline owing to the anticipated bank bailouts.
This paper develops a dynamic general equilibrium model with
maturity-mismatching banks and explores how individual banks could take
My special thanks go to Yoshiaki Shikano(Professor, Doshisha University),Takayuki
Tsuruga(Associate Professor, Kyoto University), Soichi Shinohara(Professor, Doshisha
University), Kenichi Kaminoyama(Assistant Professor, Doshisha University), Yusuke
Mitsui(M1, Kyoto Univesity) and all Shikano and Shinohara seminar members for their
valuable comments and suggestions. Of course, any remaining errors are all my own.
1

on excessive systemic risks that could result in a devastating contraction in

macroeconomic activity. Based on the dynamic general equilibrium, we
discuss how the bank overleverage leads to inefficiently high crisis
probabilities and sharp declines in economic activity after a crisis. We also
explore how existing policy measures to enhance the resilience of banking
systems affect crisis probabilities.
Key elements of the model are (1)issuance of non-state-contingent debt by
banks and (2)the banks risk exposure to capital prices which affects the
bank solvency. The first element enables banks to raise funds and to promote
liquidity creation in the absence of the completed markets, as discussed in
Diamond and Rajan(2001a,b). The second element is incorporated with our
OLG framework that explicitly includes factor markets. These elements,
however, may lead to excessive risk-taking in the banking sector as a whole.
In this model, bank overleverage arises due to pecuniary externalities.
Recent studies on the pecuniary externalities have reached a broad
consensus that pecuniary externalities, or credit externalities, tend to
prevent the laissez-faire (LF) economy from achieving the social optimum. In
line with these previous studies, our result of the bank overleverage is
juxtaposed with the second-best allocations which a social planning (SP)
agent achieves subject to the same financial constraint as the LF economy.
Each atomic LF bank is selfish and lack incentives to coordinate with each
other to affect the capital prices. Pecuniary externalities create the wedge
between the social and private marginal cost of increasing the leverage and
ill-incentivizes banks to take on excessive risks systemically. In our
benchmark simulations, the crisis probability in the LF economy is 6.6%,
compared to 4.5% in the constrained social optimum.
We got the ground for government intervention to rein in excessive leverage
of banks, because such regulatory risk reduction can improve welfare. We
introduce prompt corrective action (PCA) as the policy intervention. First, we
argue that banks capital requirement with PCA may be less exposed the
risks of exacerbation of excessive risk-taking in the banking system. Then we
deliberate whether strengthening the required minimum level of capital is
effective for stability of the banking system or not.
The rest of the paper proceeds as follows. Section2 illustrates the
macroeconomic model with maturity-mismatching banks and characterizes
the banks optimal leverage in the competitive equilibrium. In section3 , we

compare the LF equilibrium in section2 with the allocation achieved by the

SP banks and explain why a competitive banking sector tends to be
overleveraged. Section4 assesses policy measure PCA aimed at reducing
crisis probabilities and discusses policy implications. In section5, we
deliberate extension of the model further more. Section6 concludes.

2. A Macro-economy with Banks

2.1 Setting
We consider an infinite-horizon OLG model incorporating banks with a
maturity mismatch. Each generation of agents consists of households,
entrepreneurs, and banks. Each period, generation t is born at the beginning
of period t and lives for two periods, t and t+1. Each agent is identical and
constant in the population. Furthermore, an initial old generation lives for
one period and the subsequent generations live for two periods. We explain
about each agent particularly later.

2.2 Agents
Households are risk averse and subject to a liquidity shock that affects their
preference for consumption over two periods. The liquidity shock is an
aggregate shock and the only source of the uncertainty in the model. They
aim to smooth their consumption intertemporally. Following DR, they are
endowed with a unit of consumption good at birth and do not consume the
initially endowed consumption goods at the beginning of period t. they
deposit all initial endowments at banks operating in the same generation.
They receive wages in the competitive labor market by supplying one
unit of labor in both periods, t and t+1.
Under the competitive banking sector, each household accepts the banks
offer on deposit face value at the beginning of period t, and observes the
liquidity shock in the middle of period t. the liquidity shock is common
across all households in the same generation and has the probability density
function f( ) with a support of [0,1]. This shock represents households
preference for consumption when young and signals the need for liquidity in
period t.
After the realization of , households make their decisions for consumption
smoothing without uncertainty. Given that a crisis does not take place,

households then choose withdrawal amount to maximize their utility,

U(1,, 2,+1 ) = 1, + (1 )2,+1 (1)
s. t. 1, = +
(2)
2,+1 = +1 + ( ) (3)
where 1, 2,+1 denote the consumption of households born in period t
when young and old, respectively. Each household supplies a unit of labor in
each period and receives wage income in period t and +1 in period t+1.
Here denotes the one-period gross interest rate from period t to t+1.
In our model, a financial crisis takes place with the endogenous probability
, depending on the realization of .with the probability 1t , a financial
crisis is not taking place and households can withdraw g t in period t and all
the remaining deposits in period t+1. With the probability t , however, a
financial crisis and households withdrawals amount to the liquidation value
of premature projects, X(<1), in period t and nothing is left in period t+1. In
the case of a crisis, households fail to smooth out their consumption and end
up with C1,t = wt + X and C2,t = wt+1 .
When the households can smooth out their consumption, the intertemporal
first-order condition for consumption is satisfied:
t
1t

1,

2,+1

(4)

Given the Euler equation (4), the withdrawals in the absence of a crisis can
be written as
+1

g t = t (

+ ) (1 t )

(5)

The withdrawal function implies that large t and Dt are likely to

precipitate a financial crisis.
Entrepreneurs are risk neutral and maximize their expected lifetime utility
represented by E(1, +2,+1 ),where1, 2,+1 denotes entrepreneurs
consumption when young and old. They launch long-term investment
projects at the beginning of period t, by borrowing households endowments
via the banks in the same generation. The investment project requires one
period for gestation, and capital goods are produced in period t+1. .
They use a unit of consumption goods financed from banks for their capital
goods production, and this production technology takes one period for
gestation before its completion. In period t+1, the project yields a random

capital goods output

, which is uniformly distributed over [ , ]with the
probability density function h(
). They sell their output in the capital goods
market for the capital price q t+1 . However, if this project is prematurely
liquidated in period t, the transformation from the consumption goods into
capital is incomplete. As a result, the output is reduced X units of
consumption goods and is repaid fully to banks in period t.
Each entrepreneur can borrow a unit of consumption goods from a bank who
has ,or learn until the project to mature, knowledge about an alternative, but
less profitable, method to operate the project. The banks specific knowledge
allows it to generate q t+1
from a project outcome with 0 < < 1. Once a
bank has lent, no one else can learn this alternative way to operate the
project. As a result, entrepreneurs accept the financing contract with each
bank and repay q t+1
. They are left with 1 of the share of their profit
and enjoy their own consumption based on their linear utility. We assume
that entrepreneurs are endowed with I units of capital goods at the
beginning of period t+1. They sell this endowment capital together with the
newly created capital made from the consumption goods. For simplification,
we assume a 100% depreciation rate in the law of motion for capital. The
introduction of the endowment of capital goods here guarantees a finite
capital price in the aftermath of a financial crisis in which all projects are
scrapped due to full liquidation.
Banks are risk neutral and maximize their expected lifetime utility
E(1, + 2,+1 ), where 1, 2,+1 denote consumption of banks when
young and old. Banks have no initial endowment at birth and thus they need
to raise funds from households to lend to entrepreneurs. As the relationship
lender, a bank has the knowledge to operate the entrepreneurs project that
cannot be transferred to households. Banks issue demand deposits
(short-term debt) Dt as a commitment device to compensate for the luck of
transferability of their knowledge. In this model, Dt is predetermined at the
beginning of period t, and a liquidity shock t is realized in the middle of
period t.
Each bank attracts many entrepreneurs through a competitive offer on the
loan, resulting in an identical portfolio shared by all symmetric banks. This
setup effectively leads to a convenient outcome in the model: each bank and
the aggregate economy face an identical distribution of entrepreneurs. In

period t, the banks receive signals that perfectly predict the realized
value
in period t+1. With this information and the households
liquidity demand observed in period t, each bank can choose one of the
options: (i) to liquidate projects in period t, obtaining X of consumption goods
per project; or (ii) to collect a fractionq t+1
from a completed project in
period t+1. The bank liquidates the project if the outcome of a project falls

short of
+1 , defined as a function of +1 :

+1 =

(6)

+1

Otherwise, the bank continues the project, and then receives repayment
q t+1
and entrepreneurs consume the remaining fraction of outcome,(1
)+1 , per project. After repaying the full amount of the households
withdrawals, the banks consume the consumption goods.

Next, we consider the banks asset A ( +1 ). The banks asset at the

beginning of period t can be expressed as

( +1 ) =

+1

() +

= (

+1

)+

+1

+1

+1
()

+1

(7)

Note that h() is interchangeable with h(

) owing to the perfect signaling.
The banks asset denoted in (7) can be decomposed into two components: the

values of the prematurely liquidated projects denoted as L = (

+1

+1

(), which is used to meet the liquidity demand from the

households, and the banks share of the investment output (measured in the
present value of consumption goods) denoted as

+1

,where = (

+1

(). The banks are subject to the solvency constraint

+1

)=

( +1 ). Owing to the uniform distribution assumption for , it can be

easily shown that A() monotonically decreases with
the relative price

+1. We can define

+1 that satisfies the solvency constraint with

equality

= ( )
+1

(8)

We refer to and +1 as the threshold interest rate and capital price,

respectively. Hereafter, we denote a variable with an asterisk as the variable

on the threshold. And we note that given ( +1 ), the bank leverage

( ) is uniquely determined once is chosen, and hence we refer to

as leverage hereafter.

2.3 Market clearing condition

Four markets need to clear in the competitive equilibrium: (i)liquidity;
(ii)consumption goods; (iii)capital goods; and (iv)labor.
The liquidity market clearing condition is given by

+1

) = (

+1

+ ) (1 )

(9)

Next, the market clearing condition for consumption goods is

+ (

+1

) = 1, + 2,+1 + 2,+1 + 2,+1

(10)

where is consumption goods produced in period t.

The left-hand side of (10) includes the supply of goods from the liquidated
projects. On the right-hand side of (10), 2,+1, 2,+1 2,+1 are
consumption when generation t-1 is old.
is represented by a standard constant-returns- to-scale Cobb-Douglas
production function:
= F(, ) = 1

where denote the capital stock and human capital, respectively.

Demand for labor and capital satisfies

= , = (1 ) ( )

(11)

= , = ( )

(12)

Accordingly, the second derivatives are denoted by ,, , and , .

The capital goods market clearing condition is

+1

+ ( +1 )
={

(13)

Here the equation suggests that the capital goods supply sharply declines, in
the aftermath of a crisis. Throughout the paper, we use to denote
the wage rate and the marginal product of labor evaluated at +1 = .
Finally, both young and old generations supply a unit of labor in each period.
Therefore, the labor market clearing condition is
= = 2

(14)

Overview of this model is shown below.

Labor market

Households

Liquidity market

Consumption
goods market
Capital goods
market

Banks

Entrepren
eurs

10

2.4 Optimal bank leverage in LF economy

We now consider the banks optimal leverage, which is chosen before the
realization of the liquidity shock. We focus on the laissez-faire (LF) banks in
this subsection.
The banks are competitive at issuing demand deposits, and we assume that
households endowments are scarce in comparison to entrepreneurs projects.
As a result of competition, the banks make a competitive offer of deposits for
households, aiming to maximize the household welfare (Allen and Gale 1998,
2007), while in fact they are maximizing their own profits. Maximizing the
household utility via the deposit offers mean that banks internalize the
liquidity market clearing condition in determining the offer. Through this
internalization, the banks take into account possible changes in the crisis
probability . On the other hand, outside the liquidity market, they take
the capital price and wages as given.
To understand how the banks choice of affects , we take three steps.
First, we define a function as
( ) = +1 1 ( )
(15)
from (8). We emphasize that, in (Dt ), the threshold capital price +1 is
treated as a parameter, reflecting the price-taking behavior of the LF banks.
Second, using (9), we define a function as

( ) =

+1 +

(
)
+1
+ +

(16)

where = ( ) while , +1 +1 are given parameters for the

LF banks. We reemphasize that these prices are not constant parameters but
in fact vary according to (11) and (12). But for the LF banks, when they
determine the optimal , they just take them as given. With = ( )
and other threshold variables, the threshold level of the liquidity shock
= ( ) clears the liquidity market with the LF banks as shown in (16).
The final step is to connect to the crisis probability . The above
defined ( ) means that any changes in always give rise to changes
in for the liquidity market to clear. By the solvency constraint with

11

equality (8), any level of , once chosen, determines the threshold relative

price, . Hence, can be interpreted as the liquidity shock on the

+1
brink of a financial crisis. Namely, when t is strictly greater than , the
banks turn out to be insolvent and a crisis is precipitated. Thus, the crisis
probability has a one-to-one relationship to via the probability
density function f(t ) :
1

= ( )

(17)

In sum, the banks choice of the leverage specifies

+1 and the

threshold relative price determines the threshold level of the liquidity shock
, completing the link between the bank leverage and the crisis probability.
We are now ready to set up the optimization problem for the banks to
determine the size of their leverage. In the problem, banks take into account
the endogenously changing .
Problem LF

In a laissez-faire economy, banks maximize the household expected utility

0 { ( + ) + (1 )[+1 + ( )]} ( ) +
1

[ ( + ) + (1 )]( )

(18)

subject to (9) and (16).

The banks choose their leverage according to the following first-order
condition:
(1 )

[ log ( + ) + (1 ) (

= 0 [ (1

+1
2

(, )) + (1 )

where + +

)] ( )

(, )

] ( )

+1

, + + +1

(19)
denote the

lifetime income of households. More importantly, ( ) is calculated

from (16), taking capital prices and wages as given in line with the
behavior of the LF banks.

12

+1

( ) =

+1

+1

( ) < 0

(20)

Likewise, with a slight abuse of notation, (, ) denotes marginal

changes in R t with respect to Dt , given t . By (9),

+1 + +1 2

(, )

>0

(21)

where is the derivative of with respect to

+1 .

Equation (19) provides an economic interpretation that is in line with broad

intuition. The terms in brackets on the left-hand side of (19) represent the
loss of utility in a crisis compared to the threshold. From (17), the term
outside the brackets indicates that the marginal changes in a crisis
probability with respect to bank leverage. The left-hand side of the equation
consists of the expected loss of utility and the marginal change in the crisis
probability. Simply-put: the left-hand side of (19) is the marginal cost of
increasing .
The right-hand side of (19) consists of the effects of increasing leverage on
the expected households utility through their lifetime income. On the one
hand, the increase in , has an outright positive effect on the households
income: the higher the leverage, the larger the withdrawal, allowing
households to enjoy more consumption. On the other hand, the increase in
leads to a higher interest rate via liquidity shortage, discounting the
households labor income in period t+1 and reducing returns on forgoing
withdrawal until period t+1. Hence, as far as the outright effect on the
lifetime income exceeds the effect on the interest rate, the higher leverage is
beneficial to households. Simply put: the right-hand side of (19) is the
marginal benefit of increasing . By (19), we can lead the optimal bank
leverage in LF economy.

2.5 Optimal bank leverage in SP economy

We now consider the optimal bank leverage of the social planning (SP) banks

13

which choose their leverage as the constrained social planner. We

characterize the allocations for the SP banks as the constrained social
optimum and compare the allocations with those in the LF economy. We now
check the conditions of SP economy. We assume that the SP banks must
make all their decisions before observing t . After realizing t , they are left
with no options. In other words, the SP banks are subject to the constraint
that they can neither control households behaviors nor choose their outright
consumption levels because households can react to any realized value of t .
By the way, we cannot disregard a maturity mismatch and resulting
financial crises. As such, we assume that the SP banks are entities engaged
in a maturity mismatch and pre-commit to payment on their debt regardless
of the states realized following their commitment. The extra ability given to
the SP banks compared to the LF banks is that the former can internalize all
price effects in all markets when they make decisions regarding their
leverage.
The SP banks take into account their changes reflecting the marginal
product. Formally, we replace q t and wt with , and , , respectively.
Nonetheless, the SP banks take the households behaviors as given, as they
cannot make their contract contingent on t . In other words, households
always choose their consumption and withdrawal given Dt pre-committed
by the SP banks.
To solve the problem for the SP banks, we clarify the solvency constraint
with which the SP banks are faced, (

,+1

), where ,+1 = +1

from (12). The newly solvency constraint for the SP banks has different
effects on the threshold because (8) and (15) are now replaced with
= (

)
,+1

( ) = ,+1 1( )

(22)
(23)

, respectively, in the problem for the SP banks where the SP banks can
internalize general equilibrium effects of the factor prices. We summarize
the SP banks problem as follows:
Problem SP

The social planning banks maximize the household expected utility,

14

0 { ( + ) + (1 )[+1 + ( )]}( ) +
1

[ ( + ) + (1 ) ] ( )

(24)

subject to,
(

,+1

,+1

) = (

+ ) (1 ),

(25)

+,
,+1

( )

)
,+1

, + +

(26)

where ,.+1 = .+1.

Note that all the factor prices, including +1 +1 in problem LF, are
replaced with the marginal products in problem SP. More importantly,
because the SP banks factor in all general equilibrium effects, ( ) can

be denoted as
. The solution of problem SP is conceptually
comparable to the solution of the problem LF.
In problem SP, all the factor prices in period t+1 are functions of +1 or +1 .
In this context, the capital goods market clearing condition in problem SP
needs extra attention when we replace the capital price in (13) with the
marginal products. We note that +1 depends only on the market interest
rate R t and this relationship is denoted by a function (R t ). Provided that a
crisis does not take place, +1 evolves according to
+1 = + (

,+1

) ( )

(27)

where < 0 represents the derivative of +1 with respect to .

Our first main result is as follows. With the factor prices replaced by
marginal products in problem SP, the allocations that the SP banks achieve
differ from those achieved by the LF banks because of the extra ability given
to the SP banks. Comparison between problems LF and SP confirms that the
two problems are subject to exactly the same constraints. With identical
constraints, any discrepancy in the first-order conditions generally results in
different allocations across the two problems. To see this, we focus on

15

, both of which are the part of the first-order

conditions in each problem:

+1

( ) =

+1

+1

( ) < 0

(20:see

above)

+1

( ) =

+1

+1

( ) + < 0

(28)

)
2

where (

) [(
) ,+1 + ,+1 ] (29)

,+1

In general, non-zero ensures the difference between the two equilibria,

while in fact comparison of the two first-order conditions reveals further
discrepancies in addition to .
Also, we can introduce marginal changes in the threshold interest rate of the
LF and SP from (15) and (23).
( ) =

( ) =

+1

(30)

<0

1
1 ,+1

+1

+1

<0

(31)

(
)
Above these conditions, we can lead the optimal bank leverage in LF and SP
banks. We show the result that LF banks are overleveraged at the LF
equilibrium by suggesting the graph below2.
This graph is quoted from ,ISFJ
2012 1st.-2nd. Dec.2012

16

We will explain why the LF banks are likely to be overleveraged at the LF

equilibrium in the next section particularly.

3. Comparison to LF and SP economies

3.1 Crisis probability and Marginal Systemic Risk (MSR)
To facilitate assessment of the systemic risk of an economy, we define
marginal systemic risk (MSR) as the marginal increase in the crisis
probability against a unit change in bank leverage at and round the
equilibrium. Specifically, let , be the level of bank leverage chosen in the
LF economy. Then ,

, = (, ) = ( ) (, ) = ,

(32)

Recall that a bank in the LF economy takes other banks decisions as given,
but in fact the crisis probability is affected by the synchronized decisions by

17

the banking sector as a whole. In this regard, ( )

can be

interpreted as the marginal risks perceived by the individual price-taking

banks, which can be contrasted with the true marginal risks,

( ) at the chosen Dt = , . Technically, MSR is applicable for

any level of Dt . We utilize MSRs evaluated at Dt = , because this allows

us to directly compare the MSRs across different problems under the same
allocations and prices.
We stress that if is smaller than , the gap indicates that the
banking sector in the LF economy underestimates the marginal cost of
higher leverage by not taking into account the systemic risks. When we
prove this problem, we can reveal the reason why the LF banks are likely to
be overleveraged at the LF equilibrium. Instead, if , takes a larger
value than , , the gap points to underleverage in the LF economy. In
principle, an undervaluation of , , would provide
the ground for government intervention to rein in excessive leverage of
banks, because such regulatory risk reduction can improve welfare.
Our second main result is that , , , relying on the
concept of the MSR. For the formal proof, our second main result requires
one technical condition:
Condition 1
, (1 ), = ,
(33)

where , , are the elasticity of the liquidity demand and supply

with respect to in the liquidity market, respectively.
Condition 1 ensures that 0 a chosen . This condition is likely to be
satisfied because, in general, the capital ratio takes a substantially
1

smaller than the labor share(e.g., = 3). With the Cobb-Douglass production,
these shares, and 1 , can also be interpreted as the elasticities of wages
and the capital price with respect to +1 , respectively, both of which
translate into demand and supply in the liquidity market. SP banks can
internalize all the factor prices via the production technology. Due to this
internalization by the SP banks, a discrepancy arises. This is
positive under condition1 since we assume that the capital price is more
sensitive to changes in the capital than wages and/or the supply curve is

18

relatively flatter than the demand curve with respect to in the liquidity
market. Then, we are ready to state proposition 1.
Proposition1 Under condition1, , is strictly larger than , .
Proof.
Proposition 1 provides a foundation for understanding why the crisis
probability is higher in the LF economy than in the social optimum. We focus
on the key discrepancy between ( ) ( ). We represent (30)
and (31) below:
( ) =

+1

(30)

<0

+1

( ) =

1 ,+1

+1

(31)

<0

(
)
These equilibria make the magnitude correlation of the marginal changes in
the threshold interest rate between LF and SP economies clear.

+1

1,+1

+1

+1

,+1

>0

(34)

where = (

+1

).

The sign of

= (, ) is ensured to

be negative.
So, we can show the magnitude correlation of MSRs between the LF and SP
economies strictly.
, ,
,+1

= ( ) [ (, ) (, )] =

} ( ) + ] > 0 , > ,

( )

[{

,+1

(35)

Above the inequality, we proved the proposition 1.

Equation (34) indicates that an individual price-taking bank underestimates
the marginal changes in due to the pecuniary externalities. An increase
in leverage reduces , because, in general, highly leveraged banks would
be more likely to default under a lower threshold interest rate. But the

19

perceived reduction differs between LF and SP banks and this gap

creates the wedge in the two MSRs in our model. In this regard, this wedge
reflects the LF banks underestimation of the marginal cost of higher
leverage. As you understand seeing above the graph, LF banks
underestimate the marginal cost in changing a unit of bank leverage. As a
result of the underestimation of the marginal cost, the LF banking system
finds itself insolvent more frequently than expected.

To better understand the gap in the MSRs, we can focus on ,+1 ,

included in (34). This term points to a side effect arising from high leverage:
in general equilibrium, the reduction in increases +1 = ( ). That
is, the lower stimulates capital supply on the threshold and this
increase in +1 triggers the decline in the threshold capital price +1 via
the lower marginal product of capital. With the side effect, the lower capital
price further undermines the banks solvency, compared to the case without
the side effect of increasing the leverage. Because the atomistic banks do not
take into account this side effect, the lower-than-expected capital price and
the undermined banks solvency raise the probability of a financial crisis
compared with the economy with the SP banks. The SP banks estimate the
decline in the threshold capital price and the increase in a crisis probability
via the lower marginal product of capital correctly, so they are likely to hold
the lower leverage than the LF banks leverages.

3.2 Calibration
We provide numerical solutions of the model in this subsection to see the
concrete results of the banks leverage, a crisis probability, and the household
expected utility.
We assume that the liquidity shock follows the beta distribution with a
mean of 0.50and a standard deviation of 0.07. And we collect the other
parameters under the table:

Putting the specific parameters, we can get the benchmark results

summarized under the table:

20

In comparison to the LF and SP economies, it is obvious that the optimal

leverage and a crisis probability in the SP banks are lower than the LF
banks and the expected household utility is higher than the LF banks.
So, we got the ground for government intervention to rein in excessive
leverage of banks, because such regulatory risk reduction can improve
welfare.

4. Capital requirement with Prompt Corrective Action

(PCA)
4.1 Optimal bank leverage in PCA economy
In section 3, we showed the ground for government intervention to rein
excessive leverage of banks. In this section, we introduce the Prompt
Corrective Action (PCA) as the policy intervention for realizing the stability
of the banking system.
In an attempt to decrease the crisis probability, we add a government /
central bank (GC) to the model. In fact, PCA has been up and running as the
primary banking sector regulatory tool, as typically represented by Basel .
The regulation requires banks to hold a minimum level of capital. In this
context, this regulation can be translated into the temporary transfer of the
required bank capital from banks to the GC. If a bank fails to maintain the
required level of capital, it would be taken into receivership by the GC.
Because of the capital requirement, the banks are faced with the constraint,

[ ( +1 ) ] / ( +1 )

(36)

, where points to the required minimum capital ratio. The requirement

21

can equivalently be rewritten as the PCA activation condition,

(1 ) ( +1 )

(37)

Basically, (37) is not a solvency constraint for banks because, even if the
constraint is violated, banks may still hold positive capital and remain

solvent. If a bank fails to hold ( +1 ) of capital, it is taken into

receivership by the GC. Under the PCA, the bank can continue to operate but
it is run by new management, typically appointed by the GC.
The new management appointed by the GC may have inferior skills in
fostering the remaining long-term projects because the new management
consists of less experienced bankers who inherited unfamiliar projects.
Reflecting the inferior skills, the new management of the bank can obtain
from the completed project. If =1.00, the banks ability as a
relationship lender is fully retained while, by contrast, if <1.00, it points
to a loss of human capital in the banks because of the receivership.
The banks assets under the PCA can be expressed as:

( +1 : ) = +1 () + +1 ()

+1

(38)

where we denote ( +1 : 1) = ( +1 ) and +1 =

(

+1 )

Using the new notations, the bona fide solvency constraint of the bank is

(1 ) ( +1 : ) + +1

(39)

where the required bank capital is represented by +1 = ( +1 : )

that the bank takes as given. The solvency constraint remains broadly
unchanged compared to the LF case, except for , because the required
capital is possessed by the new management of the bank. Although the
capital requirement generates the incentives for banks to deleverage, this

22

requirement is much less stringent on the banks solvency because +1 is

left with banks as usable funds for payout.
Next, a threshold value for PCA activation needs to be introduced. Suppose
that the realization t is larger than a certain level of . This condition
defines the interest rate and capital price on the brink of the PCA activation
such that

1
1

+1

If the relative price

(40)

exceeds
+1
+1 , the PCA is activated. In this

case, the banks can remain solvent but are taken into receivership due to
undercapitalization. Then, is written as from (9)

+1 +

(
)

+ + +1

(41)

By contrast, if > , the bona fide solvency constraint (39) is violated

and a crisis is precipitated. This condition reintroduces the interest rate and
capital price on the brink of financial crises:

1
1

+1

: )

If the relative price

+ 1 +1

(42)

+1 exceeds
+1 , the banking system is

precipitated into a crisis, and this is likely to take place for low values of .
Accordingly, we define a function as

1 (
( ) = +1
1

+1
1

: )

(43)

Note that the inverse function theorem assures ( ) = +1

[(1 ) ]

,where (

+1

<0

: ).

Under the PCA, the threshold level of the liquidity shock for banks solvency
takes a form similar to (16) in the LF economy. We define a function as

23

+1 : +

( ) =

(
)
+1
+ +

(44)

where = ( +1 : ) = +1 () (45)

is the liquidity supply under the PCA, which is also a function of and
= ( ) from (43). The liquidity market clearing condition under the
PCA is

( +1 : ) = ( +1 + ) (1 )

(46)

We then state the banks problem under the capital requirement with the
PCA:
Problem PCA

In an economy with the capital requirement with prompt corrective action

(PCA), banks maximize

0 { ( + ) + (1 )[+1 + ( )]}( ) +

1
{ ( + ) + (1 )[+1 + ( )]} ( ) + [ ( +

) + (1 )]( )

(47)

subject to (9), (46), (41), and (44).

4.2 Calibration
We will confirm that whether this policy intervention can reduce the crisis
probability compared with that in LF economies or not. We put the same
specific parameters, and we show the benchmark results under the table:

24

Taking an example of panel A where the required minimum capital ratio is

4%, the probability is 6.32% when =1.00, lower than 6.58% in the LF
economy. The capital requirement per se discourages banks risk-taking
while keeping unchanged the banks resources that are payable to creditors.
This improved resilience of the banking system, however, may not be
achieved under the alternative conditions. As the third column in table (
=0.75) indicates, the crisis probability is 7.30% , which is higher than that in
the LF economy. If the banks under the PCA are run by low-skilled bankers,
risk-reduction effect through deleveraging could be dominated by the perils
of lower solvency owing to the loss of resources.
This table shows that the success of this policy option largely depends on
how efficiently the GC can manage the troubled banks under receivership.

5. Extension of the model PCA

In section 4, it is revealed that the success of this policy option largely
depends on how efficiently the GC can manage the troubled banks under
receivership. If we could hold =1.00, strengthening might be effective
for realizing the stability of the banking system. However, we should the case

25

is changeable affected by other parameters change. So, we internalize

with respect to . we define as
+1

(48)

= 2 +2

steps-down as increasing .
The reason for internalizing instead of is that we consider shifting of
could change abilities of supervisors from the GC.
We reproduce (17) and (44),
1

= ( )

(17)

+1 : +

( ) =

(
)
+1
+ +

(44)

We differentiate (44) with respect to , then we get the equilibrium

() = 2 [ ( + ) ( +1
2

)]

(49)

(A)

(B) ( + ) ( +1
2

)<0

We consider three patterns under the PCA:

(1) If A is positive, then decreases.
(2) If A is negative, but |A|<|B|, then decreases.
(3) If A is negative, but |A|>|B|, then increases.
Recall that under the PCA, the liquidity supply is

= ( +1 : ) = +1 () (45)

This equality shows that the move of

Next, we reproduce (6)

+1 =

+1

(6)

and of () accords.

26

, and we differentiate (6) with respect to ,

+1

1
2

(+1 )

(+1 ) (+1 )

(50)

Now, assuming LF and PCA economies are homogeneous, it is set

= 0.00 to compare with them.
We verify how an increase of affects the crisis probability in this situation.
Before discussing this problem, we posit five hypotheses as follows:
(1)

(2)
(3)

>0

+1
2
2

< 0 for 0 1

+1

<0

<0

(4)
+1|=0 ,
+1|=1 > 0
(5) 0 < < 1
Under these hypotheses, we examine three cases (a)~(c).

(a) is low sufficiently (nearly 0)

+1

<0, but it has a small change, so

< 0 and () > 0, then

decreases.
(b) satisfies 0 < < 1
+1

< 0 but the change is larger than above the case and is smaller than

below the case, so

0 and we lead the optimal required minimum capital

ratio .
(c) is high sufficiently(nearly 1)
+1

< 0 but the change is the largest in these cases, so

< 0 and

() < 0,then increases.

About hypothesis 3, we explain the two situations:
If is low sufficiently, banks can lend goods to the entrepreneurs which
produce the low output since the minimum level of capital is required.
Therefore, the impact to small and medium entrepreneurs will be relatively
minimized even if the GC strength . Conversely, if is high sufficiently,

27

banks hesitate to lend goods to the entrepreneurs except the entrepreneurs

which produce the high output. In this situation, strengthening makes
the bank assess the entrepreneurs more strictly, so the banks decide to
liquidate more projects.
Although we put these specific hypotheses, if we could suppose above these
cases, we could obtain the optimal under 0 < < 1 and at this point,
the crisis probability is the lowest.

6. Conclusion
Comparing with LF economy and SP economy, we showed that LF banks are
likely to get excessive leverages and operate inefficiently. In comparison with
LF economy, SP economy remains lower leverage and crisis probability in
spite of remaining high expected utility of households. So we get the ground
for government intervention to rein in excessive leverage of banks.
Previous study showed that introducing of PCA can make leverage low and
the banking system efficient. However, the success of this policy option
largely depends on how efficiently the GC can managed troubled banks
under receivership. In the situation that is fully low, the crisis
probability may be higher than LF economy even if the GC strengthen .
In this paper, we showed that internalizing with respect to lead the
optimal level of required minimum capital ratio . As a result, we are able
to raise a doubt with respect to strengthening BIS accord in recent years and
insist that we should discuss strengthening BIS accord more deliberately.

7. Further discussion
In the verification in this time, we derived optimal under the particular
condition. But it is possible that a corner solution may be led according to
second order differential of (A) and (B) because slope between (A) and (B)
may be different largely, then optimal may be over 0 1. Hereafter,
we need to set the equilibrium to lead optimal along the real economy
and hope to discuss more strictly this problem according to the setting set
more rigidly.

28

References
[1] Ryo Kato and Takayuki Tsuruga, April,2012 Bank Overleverage and
Macroeconomic Fragility Kyoto University Graduate School of Economics

Research Project Center Discussion Paper Series, Discussion Paper

No.E-12-002
[2] Allen,F., and D.Gale (1998),The Optimal Financial Crises, Journal of
Finance, 53(4),pp.1245-1284
[3] Allen.F., and D.Gale(2007), Understanding Financial Crises, Oxford
University Press.
[4]2013
[5]

,ISFJ 2012
1st-2nd,Dec,2012