wp 2010 00 monetary policy risk taking

8/6/2019 Wp 2010 00 Monetary Policy Risk Taking

1/41

MONETARY POLICY

AND RISK TAKING

IGNAZIO ANGELONI*, ESTER FAIA** ANDMARCO LO DUCA

Highlights

The financial and economic crisis has taught central banks

a lesson. They were previously primarily concerned withmaintaining price stability, but the crisis has producedevidence to show that monetary policy can contribute to thedevelopment (or the mitigation) of systemic risk in the

financial sector. Monetary policy therefore needs to takefinancial stability implications into account, and themacroeconomic implications of bank supervision andregulation must also be considered.

It is therefore important to understand the linkages between

financial risk, monetary policy and the business cycle. Thispaper considers the thinking to date on risk taking inmonetary policy, presents comparative evidence for the

euro area and the United States, and interprets the evidenceusing a macro dynamic stochastic general equilibriummodel.

This approach produces three preliminary conclusions:central banks' monetary policy approaches affect, with timelags, the propensity of financial markets and banks toassume risk; it is possible to take account of these effectsthrough macroeconomic models that embody optimising

agents with limited information and include explicitlybanking and financial sectors; and there is some indicationthat the modelling done in the paper can be used as thebasis for predicting the relationship between monetarypolicy and the effect of risk on macroeconomic variables,though this area needs considerable further research.

* Visiting Scholar at Bruegel and advisor to the ExecutiveBoard of the European Central Bank.

** Goethe University Frankfurt, Kiel IfW and CEPREMAP. European Central Bank.

BRUEGELW

ORKINGPAPER2010/00


2/41

Monetary Policy and Risk Taking

Ignazio AngeloniEuropean Central Bank and BRUEGEL

Ester FaiaGoethe University Frankfurt, Kiel IfW and CEPREMAP

Marco Lo DucaEuropean Central Bank

February 15, 2010

Abstract

We combine data evidence and a new DSGE model with banks to examine the links betweenmonetary policy, financial risk and the business cycle. The model includes banks (modelled asin Diamond and Rajan, JF 2000 and JPE 2001) and a financial accelerator (Bernanke et al.,1999 Handbook). A monetary expansion increases the propensity of banks to assume risks. Inturn, financial risks affect economic activity and prices. This "risk-taking" channel of monetarytransmission, absent in pure financial accelerator models, operates via the leverage decisions ofbanks. The model results match certain features of the data, as emerged in recent panel datastudies and in our own time series estimates for the US and the euro area.

Keywords: monetary policy, bank behavior, leverage, financial accelerator.

The authors are the sole responsible for the views expressed in this paper.

1


3/41

1 Introduction

Central banking can be unpredictable at times. Think of what happened recently. Until 2007,

central banks were operating within a well established paradigm, composed of three fundamental

elements. The first call it single focus stipulated that in conducting monetary policy central

banks should aim at maintaining price stability, i.e. a low and stable rate of change in the price of

a basket of consumer goods and services. The specific time horizon for achieving this objective was

disputed, but it is safe to say that most believed it should be rather short; the prevailing practice,

inflation targeting, prescribed that central banks should steer their own inflation forecasts, which (as

for all economic variables, for that matter) are notoriously unreliable at long horizons. The second

tenet was that central banks should be independent, i.e. not influenced in their policy decisions

by other actors, whether governments, businesses, trade unions, or other. The last element was

a sort of assignment principle: central banks should not be "distracted" by concerns for other

policy domains, nor should other policy actors share responsibility for price stability. For two

reasons: first because multiple or shared objectives give rise to confusion on where responsibilities

really belong; second, because potential failures in attaining other goals may destroy the credibility

of central banks in achieving their primary objective, price stability. This argument was used in

several contexts, but perhaps most forcefully to suggest that central banks should not be made

responsible for banking and financial supervision 1. In the years of the "great moderation" the

three legs of this paradigmatical tripod seemed so stable and solid that monetary policy was often

referred to as a "science" (for example Giavazzi and Mishkin, [27]).

In just two years, the financial crisis has inferted a powerful blow to many of these earlier

certainties. Revisited in light of the turmoil, the experience of the previous decade suggests that

the transmission of monetary policy may be more complex than was earlier assumed. Its effects may

extend much beyond inflation and aggregate demand at short-medium horizons, to encompass the

risk-taking propensity of economic agents, with second-round effects at much longer and unknown

lags. Naturally, the existence of a "risk-taking channel" channel of monetary policy for which

evidence has recently been provided, surveyed in this paper would put the single focus tenet into

question. But also the assignment argument would somehow be affected; if monetary policy can

1 See the classic curvey of Goodhart and Schoenmaker [29].

2


4/41

contribute to the formation (or the mitigation) of systemic risks in the financial sector, and if the

latter in turn feed back on macroeconomic performance with unknown lags, it is hard to escape

the conclusion that monetary policy needs to keep financial stability implications into account, and

that the macroeconomic implications of bank supervision and regulation need to be considered as

well. Completing the triangle, though nobody has seriously questioned the merits of central banks

independence yet, recent vibrations in certain political quarters the US Congress, for example

reveal a temptation to attach to the assignment of new responsibilities to central banks in the area

offinancial stability a tighter scrutiny over their decisions2.

In this complex picture, a key priority is to understand the linkages between financial risk,

monetary policy and the business cycle. Our plan in this paper is the following. First, we briefly

review the recent arguments and evidence on the risk taking channel of monetary policy. Second, we

present some comparative time series evidence on this channel in the US and the euro area. Third,

we introduce a new macro-DSGE model with banks to interpret this evidence. The model combines

a banking sector based on the theory of Diamond and Rajan (see Angeloni and Faia [6]) with a

standard financial accelerator (Bernanke, Gertler and Gilchrist [9]). In the model, firms use both

internal funds and external finance (bank loans and corporate bonds) to finance their investment.

Thisfl

exible structure allows to examine the impact of monetary policy and other shocks on thelending and the funding behavior of banks, hence explicitly modelling the leverage of banks and

firms. Our empirical results support the idea, corroborated also by other recent empirical evidence,

that monetary policy influences risk-taking in the financial sector, and that financial risk in turn

significantly influences output and price dynamics at longer lags. The model results, presented in

comparison with the two simpler benchmarks (the model by Angeloni and Faia [6] and the classic

financial accelerator model) broadly confirm certain salient features of the data.

The paper is organised as follows. In section 2 we start by describing some recently published

empirical evidence, based mainly on micro-panel data. In section 3 we present our new time-series

evidence on the transmission of monetary policy on financial risk in the US and the euro area. In

section 4 we present our macro model. In section 5 we use the model to describe the impact of

a few selected shocks, including specifically the effects of monetary policy on risk and the effect

2 Kashyap and Mishkin [31].

3


5/41

of risk on the rest of the economy. In doing so, we compare its properties of the model with two

simpler alternative benchmarks: the model in Angeloni and Faia [6], based on the Diamond-Rajan

theory, and the classic financial accelerator of Bernanke et al. Finally, section 6 concludes.

2 Survey of recent empirical evidence

The surge of interest for the implications of monetary policy on financial risks after the 2007-2009

crisis contrasts sharply with the virtual absence of any reference to risk in the earlier literature

on the monetary policy transmission mechanism. The classic 1995 survey by Mishkin, Taylor

and others in the Journal of Economic Perspectives [33] does not mention risk except as a factor

capable of reinforcing the strength of the financial accelerator. In the multi-country empirical

study of monetary transmission in the euro area conducted by Eurosystem central banks, dated

2003 (Angeloni, Kashyap and Mojon [7]), extensive use is made of indicators of bank risk in the

econometric estimates of the lending channel, but only to quantify certain features of the banking

sector that may affect the strength of the transmission, not because monetary policy may itself

influence those characteristics or for their financial stability implications.

In a different context, however, other papers have highlighted the potential importance of the

link between monetary policy and financial risks well before the onset of the financial crisis. In 2002

Borio and Lowe [17], described how asset market bubbles, leading to financial risk and instability,

can develop out of a mix of benign macroeconomic conditions including high growth, low inflation,

low interest rates and accommodative monetary policy. Their seminal contribution was followed

by a host of publications by the Bank for International Settlements calling for the adoption of a

"macroprudential approach" to financial stability including, notably, a response of monetary policy

to asset prices. Earlier on, Allen and Gale [3] had provided a theoretical underpinning for these

ideas by showing how leveraged positions in asset markets create moral hazard: leveraged investorscan back-stop losses by defaulting, and this makes asset prices deviate from fundamentals. The

link with monetary policy, clarified later in Allen and Gale [4], consists is in the fact that credit

developments in the economy are (at least partly) under the control of monetary policy.

In 2005, Rajan [35] analysed how the incentives structures in the financial system may induce

managers of banks and insurance companies to assume more risk under persistently low interest

4


6/41

rates. In a low interest rate environment, portfolio managers compensated on the basis of nominal

market returns have an incentive to search for higher yields by taking on more risk. Risk built

up under monetary accommoation turns into instability when monetary tightness returns, in the

form of confidence crises and "sudden stops" of credit. Two implications for central banks: first,

when a high risk level is already entrenched in the financial sector, abrupt policy tightening should

be avoided; second, and more important, monetary policy should preemptively avoid prolonged

periods of excessively low interest rates.

To help the empirical analysis, and also because their policy implications differ, its useful to

distinguish between two different channels through which the risk-taking mechanism can operate.

The first is via changes in the degree of riskiness of the intermediarys asset side. In presence

of low and persistent interest rates levels, asset managers of banks and other investment pools

may have the incentive to shift the composition of their investments towards a riskier mix, for the

reasons explained by Rajan. A second way in which more risk can be acquired is via the degree of

leverage and the maturity composition of the balance sheet (or off-balance sheet structures implicitly

guaranteed by the mother institution). In other words, a riskier balance sheet configuration can

be realised, even if the asset mix remains the same, by borrowing in money markets. Risk-taking

is stronger, other things equal, the shorter is the maturity of borrowing. This "leverage channel"operates specifically when short term rates are low and the yield curve upward sloping, an effect

emphasised by Adrian and Shin [1]. While the two channels (asset mix vs. leverage-maturity

changes) are conceptually distinct, it may be difficult to distinguish them since they tend to move

together. All statistical and anecdotal information we have suggests that financial institutions on

both sides of the Atlantic (banks, conduits and SIVs, investment funds, insurance companies, etc.)

became riskier, in the pre-crisis years, due to a mix of riskier investments and more fragile balance

sheet structures.

The empirical evidence on these transmission mechanisms is limited. Two strands can be

distinguished. A first one tries to identify effective leading indicators of financial crises. It has

been noted that, in a variety of different national contexts and historical periods, financial crises

tend to be preceded by a recurrent set or economic developments (Reinhardt and Rogoff, [37]). In

particular, using time series using data for 18 OECD countries between 1970 and 2007, Alessi and

5


7/41


8/41

determines less strict credit "standards" by banks may or may not have implications for risk.

Optimizing banks receiving more liquidity from the central bank and facing lower opportunity

costs will naturally move down the expected loan return schedule, in fact typically lowering their

lending rates. Though they probably interpret it as softening "credit standards", this does not

necessarily increase lending risk. Even if some more risky borrowers happen to be financed, this

may still be efficient and remain within acceptable safety bounds. So, a positive answer to the

question above does not necessarily imply that a monetary expansion has an undesirable impact on

bank risk. Conversely, even if the answer is in the negative, this does not mean that bank risk may

not be increasing, perhaps exessively, in other ways. Asset quality is only one of the ways financial

intermediaries use to take on more risk; leverage and maturity mismatch are other, probably more

important channels.

Another recent paper (Altunbas et al. [5]) uses a more comprehensive sample and a different

measure of bank risk. They consider over 600 listed European banks, in 16 countries, for which

Moodys KMV has computed expected default frequencies (EDF). EDFs are a widely used measure

of bank risk, shown to have predictive power in many cases. EDFs are obtained translating, with a

model, several market and balance sheet indicators into a single measure, a time-varying probability

of default at a specifi

c time horizon. The authors make this the dependent variable in a panelregression, that includes a variety of explanatory factors macroeconomic variables, market data,

other bank characteristics as well as monetary policy. The results suggest that an increase of

short term rates decreases bank risk on impact but increases it over time. This is interpreted as

the combination of two effects: in the short run, an increase in rates lowers the risk of outstanding

loans; over time, the higher level of rates induces hazardous lending behavior, leading to more risk.

A similar conclusion is reached by Jimenez et al. [30] in their analysis of a large sample of Spanish

banks, using more detailed credit register data.

The estimates of Altunbas et al. [5] have the advantage of having direct implications for

bank risk, but, given the very general nature of the risk measure employed, they do not allow to

distinbguish among different transmission channels. To do this, we now turn to our time series

evidence.

7


9/41

3 Some new time series evidence

Table 1 reports the results of Granger causality tests to investigate the relationship between mon-

etary policy (measured as de-trended nominal policy rate) and different indicators capturing bank

leverage and balance sheet risk. The column Equation specifies the dependent variable, column

Excluded shows the variable whose exclusion is being tested; the p-value of the test is reported

in the last column. Our data definitions are given in the Appendix. In the US (Panel A), we find

strong evidence of causality from monetary policy to the different measures of leverage and balance

sheet risk, while causality in the other direction is not detected. For the euro area, causality from

monetary policy to leverage and balance sheet risk is also detected, but the evidence is weaker than

for the US. The results for the euro area show that causality runs also in the other direction, i.e.

from leverage and risk to monetary policy.

We then extend the analysis and we estimate VAR models where we also incorporate macro

variables. We adopt the approach of Bloom[13], [14] and extend it in two directions: first, to

examine the effect of monetary policy on financial risk alongside with the effect of risk on the

business cycle; second, to consider both the euro area and the US.

To help identify the different transmission channels, we consider three indicators of risk in our

baseline model. The first (henceforth RISK1) is a stock market volatility index, constructed as a

binary variable changing value when the stock market index change is beyond a given threshold

(see Appendix for the description of the variables). This is the variable preferred by Bloom [13]

for uncertainty/corporate risk; he also shows that it is highly correlated with several other proxies

of the variability or dispersion of corporate returns. Our other two variables are based on bank

balance sheets. RISK2 is the de-trended ratio of consumer credit and household mortgages to

total bank assets. We interpret this as a measure of the (change of) bank assets specific riskiness,

based on the fact that personal loans were the most risky component of the bank loan portfolios,

particularly in the years preceding the latest financial crisis.

Finally, RISK3 is a measure of bank leverage, calculated as the 12 month difference of the ratio

total bank assets to total deposits. We decided to use this measure instead of the more customary

ratio of assets to capital for two reasons: first because data on capital are often not reliable and

not easily comparable across countries; second, more importantly, because regular deposits have

8


10/41


11/41

Finally, chart 3 shows the effect of a monetary policy shock on the three measures of risk. A

monetary policy restriction tends to have a negative effect on risk, but its strength, profile and

significance depends on the risk measure used and differs in the US relative to the euro area. The

reduction of leverage (RISK3) is significant only in the US. The effect on bank asset risk (RISK2) is

significant, delayed and persistent in both areas. Finally, the effects on corporate risk / uncertainty

are insignificant4.

We finally did a series of robustness checks to assess the stability of our results. We changed the

definition and the measurement of our variables measuring overall risk and uncertainty (RISK1),

balance sheet risk (RISK2) and leverage (RISK3). The alternative variables used are those listed

in the Appendix tables that do not enter in the baseline specification. For the US the results

are stable, only the proxy for overall balance sheet uncertainty turns insignificant in a few of the

specifications. The impact of monetary policy on leverage is very robust. For the euro area, the

results are less stable and the impact on monetary policy on overall balance sheet risk vanishes in

several specifications. Finally, we ran the estimates on quarterly data and used alternative proxies

of uncertainty calculated by Bloom for the US. The effects on leverage and balance sheet risk in

the US are robust.

4 A macroeconomic model with banks

The real sector of the model consists in a conventional DSGE model with nominal rigidities. The

economy is populated by workers (that are also bank depositors and bank capitalists), entrepreneurs

and bank managers. Entrepreneurs launch projects that require an initial investment, financed

partly by bank loans and partly by internal funds (net worth of the entrepreneur). The bank

macro variables to a shock to uncertainty is at best weakly significant. One reason for the lack of significance could bethat the series for the euro area are too short. We therefore exclude from the model the variables RISK2 and RISK3

that are available only since the end of the 90s and we re-estimated the model with data since 1990 using only withthe macro variables and RISK1 (or the implied volatility). In this specification, RISK1 is still only weakly significantwhile when the implied volatility is used, it has a strong negative and significant initial impact on inflation, a slightand significant negative impact on industrial production and no impact on employment.

4 The option implied volatility that we use as measure of overall uncertainty and corporate risk as suggested byBloom, can be decomposed into a component that reflects actual expected stock market volatility (uncertainty) and aresidual (the so-called variance premium) that reflects risk aversion and other non-linear pricing effects (see Bekaert,Hoerova and Scheicher [11]). In a related paper, Bekaert Hoerova and Lo Duca [10] investigate the impact of monetarypolicy on the two components of implied volatility using SVAR. They find that the monetary policy shock for theUS has a positive impact on risk aversion while it does not affect the component related to uncertainty

10


12/41

raises funds from depositors and bank capitalists. Bank capitalists claims the residual value after

depositors are paid out. The bank capital structure is determined by the bank manager, that

acts on behalf of depositors and capitalists by maximizing their overall return. The bank lends to

the entrepreneur the funds needed, in addition to internal funds, to finance capital investment. If

the return on the investment project (a stochastic variable) is insufficient to reimburse the bank

loan, inclusive of its contractual return, the firm fails and the project is appropriated by the bank.

Likewise, if the return to the bank is insufficient to pay out depositors, there is a run on the bank, in

which case the bank capital holders get zero while depositors get the market value of the liquidated

loan.

4.1 Households

There is a continuum of identical households who consume, save and work. Households include real

sector "workers" and bankers. Following Gertler and Karadi [25] we assume that in every period a

fraction of household members are bankers and a fraction (1 ) workers. Bankers have a finite

tenure in their job; with probability they remain bankers every period, otherwise they become

workers. A corresponding fraction of workers become bankers every period, so that the share of

bankers remains constant over time. This finite survival scheme is needed to avoid that bankers

accumulate enough wealth to ease up the liquidity constraint; we will return on this point later.

Workers earn wages and return them to the household; similarly bankers earn a fee from their

services, that is returned to the household. Consumption and investment decisions are made by

the household, pooling all available resources.

Households maximize the following discounted sum of utilities:

0

X=0

( ) (1)

where denotes aggregate consumption and denotes labour hours. The households receive

at the beginning of time a real labour income As households can work both in the industrial

and in the banking sector, we consider labour income as inclusive of the fees workers receive as

managers of the banks. Those fees are determined in the next section.

Households save and invest in bank deposits and bank capital. Deposits, pay a gross

11


13/41

nominal return one period later. Alternatively, and without qualitative change in the analysis,

we could assume the existence of another asset, say government bonds; in this case deposits can be

assumed to be held for their transaction or liquidity services, that justify a wedge between the bond

rate and the deposit rate. Finally, households are the owners of both the monopolistic competitive

sector and the banking sector. Because of this they are entitled to receive from the monopolistic

sector nominal profits for an amount, , and from any banker who ceases activity nominal profits

of an amount, . The budget constraint reads as follows:

+ + +1 + + + (2)

Note that the return from, and the investment in, bank capital do not appear in equation

2. The reason is that we have assumed, as explained later, that all returns on bank capital are

reinvested every period.

Households choose the set of processes { }=0 and assets {+1 +1}

=0 taking as given

the set of processes { }

=0 and the initial wealth 0 0 so as to maximize 1 subject to 2.

The following optimality conditions hold:

= (3)

= {+1} (4)

Equation 3 gives the optimal choice for labour supply. Note that, since labor income includes

the bankers fee, the supply of labor determined in 3 depends also on this fee5. Equation 4 gives

the Euler condition with respect to deposits and government bonds. Optimality requires that the

first order conditions and No-Ponzi game conditions are simultaneously satisfied.

5 In principle this fee is endogenous and depends (as will be seen later in the paper) on the bank capital structure.In practice, given the small ratio of bankers relative to workers, this component is negligible and we neglect itsendogeneity for simplicity.

12


14/41

4.2 Banks

Our bankfi

nances itself through uninsured deposits6

and bank capital, and uses funds to lend torisky entrepreneurs. Lending and fund-raising are administered by separate compartments of the

bank, a credit department and a funding department, linked together by an internal contract. The

lending department receives funds from the funding department and uses them to enter into credit

contracts with entrepreneurs.

4.2.1 Lending

The credit side is modelled as a standard financial accelerator (Bernanke et al. [9]).

In synthesis (for a complete exposition of the model, see Faia and Monacelli [24]), the invest-

ment projects financed by the bank are characterized by an expected return equal to +1+1,

where +1, and +1 are respectively the nominal return on capital, the price of capital and

real capital (individual subscripts are omitted for brevity). Project outcomes are subject to idio-

syncratic uncertainty, taking the form of a multiplicative random shock applied to productivity.

The shock is assumed to have a uniform density function () with mean equal to one and is

defined between zero and two; hence (1 1 + ); () = 12 ; () = 1. Therefore, the

monetary return of an investment project financed at is given by +1+1+1.

The contract between the firm and the lending department stipulates ex-ante a fixed gross

return for the bank +1and a financing amount +1 = +1 +1, where +1 is

the entrepreneurs net worth. +1 includes a fixed proportion 1 of own bank funds, +1,

while the rest is raised by issuing corporate bonds. Ex-post there are two cases. If the firm doesnt

default, it pays back the debt, inclusive of the contractual return +1, and the entrepreneur retains

any surplus. If not, the firm goes bankrupt and the lending department liquidates the project. In

this case, the lender has to bear a monitoring cost equal to a fraction of the project value.

Formally, the contract defines a default threshold such that bankrupcy of the firm occurs only if

, where is defined by

6 Our arguments apply equally to other leveraged entities, issuing uninsured short-maturity debt such as, forexample, asset-backed commercial paper.

13


15/41

+1 +1+1

Y

+1+1

(5)

where Y+1 is the unitary monetary return from holding a unit of capital.

For each euro lent, the expected return for the lender is

(+1) 1

2

Z+10

+1 + +1

Z2+1

!

and the expected return to the firm is the complement, 1 (+1). The cost of monitoring is

(+1)

2Z+10

+1

and the net unit return accruing to the lender is (+1)(+1).

The firms investment is financed by a mix of internal and external funds. The participation

constraint for the lender thus takes the form:

(1 ) +

(+1 +1) Y

+1+1((+1)(+1) (6)

The contract betwen the lender and the firm specifies a pair {+1,+1} which solves the

maximum problem:

(1 (+1))Y+1+1 (7)

subject to 6. The first order conditions are

0(+1) = (

0( (+1)) (8)

+1(1 ) +

(1 (+1

)) +

((

+1

) (

+1

)) =

(9)

where is the Lagrange multiplier. For 0 6 must hold with equality.

The external finance premium can be written as

+1

(1 ) + = (+1)

1

+1

+1

(10)

14


16/41

where (+1) = [(+1)(+1)]1. Note that the premium increases with the leverage

ratio +1+1 .

The aggregate net worth for the firm, in expected terms, accumulates over time according to

+1 = (1 (+1))Y (11)

where is a survival probability for the firm. Rearranging and using 6 one can write this as

+1 = 1

(1 ) +

+

(1 ) (12)

where =

() 1

1 is the expected monitoring cost per unit of loan.

4.2.2 Funding

The liability side of the bank is modeled as in Angeloni and Faia [6]. In summary (see [6] for

details), banks receive finance from two classes of agents: depositors and capitalists. Bank loans in

the balance sheet are equal to the sum of deposits () and bank capital, ():

= + (13)

The funding department determines the capital structure (shares of deposits and capital) on

behalf of depositors and bank capitalists. Following Diamond and Rajan ([21], [22], we assume the

banker maximizes the combined expected return of depositors and capitalists, in exchange for a

fee.

The timing is as follows. At time , the banker decides the optimal capital structure, expressed

by the ratio of deposits to the total cost of the project, , collects the funds, and transfers the funds

to the lending department. At time + 1, the projects outcome is revealed, the contractual return

is transferred from the lending to the funding department, as discussed below, and payments to

depositors and capitalists are made. A new round of projects starts.

The credit and the funding departments are tied by a contract that stipulates that the former

pays to the latter a return on the funds received , equal to its expected return on assets. More-

over, we assume that the return on assets for the funding department is subject to an idiosynchratic

shock with a uniform distribution defined in the space {; } (again, individual subscripts are

15


17/41

omitted for simplicity; the linearity of the problem allows easy aggregation). We can think of

as a liquidity shock, due for example to an unexpected withdrawal of deposits. Each period the

realized return + is split between the depositors, the bank capitalists and the funding de-

partment itself, that earns a fee for its liability management service. If the cash flow accruing to

the funding department is insufficient to reimburse the depositors (including the non-contingent

return on deposits, , there is a run on the bank, which forces an early liquidation of the loan by

the credit department.

We assume the bank is a relationship lender: by lending, the credit department acquires a

specialized non-sellable knowledge of the characteristics of the project that determines an advantage

in extracting value from it before the project is concluded, relative to other agents. Let the ratio

of the value for the outsider (liquidation value) to the value for the bank be 0 1. Even if

the value is extracted by the bank, hence without loss of relationship knowledge, a bank run may

entail a specific cost 1 0; when a run occurs, the liquidation value of the project extracted

by the bank loses a constant fraction .

Consider the payoffs to each of our players (the depositor; the bank capitalist; the bank as

liability manager). There are three cases.

Case A: Run for sure. The return is too low to pay depositors; + . Payoff

s in caseof run are distributed as follows. Capitalists receive the leftover after depositors are served, so they

get zero in this case. Depositors alone (without bank) would get only a fraction (1 )( + )

of the projects outcome; the remainder (1)(1 )( + ) is split in half between depositors

and the bank. Therefore, depositors get

(1 + )(1 )( + )

2(14)

and the bank

(1 )(1 )( + )

2(15)

Case B: Run only without the bank. The return is high enough to allow depositors to be served if

the projects value is extracted by the bank, but not otherwise; i.e (+) (+).

In this case, the capitalists alone cannot avoid the run, but with the bank they can. So depositors

16


18/41

are paid in full, , and the remainder is split in half between the banker and the capitalists,

each getting+

2 . Total payment to outsiders is++

2 .

Case C: No run for sure. The return is high enough to allow all depositors to be served, with

or without the banks participation. This happens if ( + ). Depositors get .

However, unlike in the previous case, now the capitalists have a higher bargaining power because

could decide to liquidate the project alone and pay the depositors in full, getting (+);

this is thus a lower threshold for them. The banker can extract ( + ), and again we

assume that the capitalist and the bank split this extra return in half. Therefore, the bank gets:

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

2

=( )(1 + )

2

This is less than what the capitalist gets. Total payment to outsiders is:

(1 + )( + )

2

The expected value of total payments to outsiders as follows:

12

Z

(1 + )(1 )( + )2

+ 12

Z

( + ) + 2

+ (16)

+1

2

Z

(1 + )( + )

2

The three terms express the payoffs to outsiders in the three cases described above, in order.

The bankers problem is to maximise expected total payments to outsiders by choosing the suitable

value of .

It can be shown (see [6] for details) that the value of that maximises equation 16 is comprised

in the interval +

+

. In this zone, the third integral in the equation vanishes and

the expression reduces to

1

2

Z

(1 + )(1 )( + )

2 +

1

2

Z

( + ) + 2

(17)

17


19/41

Differentiating and solving for yields the following equilibrium condition

= 1

+ 2 + (1 + )

(18)

Since the second derivative is negative, this is the optimal value of .

The optimal deposit ratio depends positively on , and , and negatively on . An

increase of reduces deposits because it increases the probability of run. Moreover, an increase

in raises the marginal return in the no-run case (the third effect just mentioned), while it does

not affect the other two effects, hence it raises . An increase in reduces the cost in the run case

(first effect), while not affecting the others, so it raises . The effect of is more tricky. At first

sight it would seem that an increase in the dispersion of the project outcomes, moving the extreme

values of the distribution both upwards and downwards, should be symmetric and have no effect.

But this is not the case. When increases, the probability of each given project outcome 12 falls.

Hence the expected loss stemming from the change in the relative probabilities (sum of the first

two effects) falls, but the marginal gain in the no-run case (third term) does not, because the upper

limit increases. The marginal effect is 2 , because depositors get the full return, but half is lost by

the capitalist to the banker. Hence, the increase of has on a positive effect, as .

The probability of a run occurring can be written as:

=1

2

Z

=1

2

1

(19)

we can interpret as a measure of bank riskiness. For plausible values of and , ranges

between 0.3 and 1. See Angeloni and Faia [6] for details.

Aggregate deposits are given by

=

+

2 + (1 + )(20)

and the banks optimal capital is:

= (1 1

+

2 + (1 + )) (21)

18


20/41

The above expressions suggest that contractionary monetary policy, raising , increases the

optimal amount of bank capital on impact. The cyclical behavior of the bank capital structure is

more complex, depending on several counterbalancing factors. Equation 21 can be interpreted as

a demand for bank capital given the volume of loans and the interest rate structure (, ).

As to the supply, we assume that bank capital accumulates in the form of undistributed dividends.

After remunerating depositors and paying the competitive fee to the banker, a return accrues to the

bank capitalist as retained earning (including any reinvested dividends). Bank capital accumulates

from retained earnings as follows:

+1 = [ + +1+1+1] (22)

where +1 is the unitary return to the capitalist. The parameter is a decay rate, inclusive

of both the bank survival rate (Gertler and Karadi [25]) and bank capital depreciation. +1 can

be derived from equation 17 as follows:

+1 =1

2

Z+1+1+1

(+1 + +1)+1+12

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

2

8(23)

Note that this expression considers only the no-run state because if a run occurs the capitalist

receives no return. The accumulation of bank capital obtained substituting 23 into 22:

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

2

8+1+1] (24)

4.3 Producers

Each firm has monopolistic power in the production of its own variety and therefore has leverage

in setting the price. In changing prices it faces a quadratic cost equal to 2 (()

1() )2 where

is the steady state inflation rate and where the parameter measures the degree of nominal price

rigidity. The higher the more sluggish is the adjustment of nominal prices. In the particular case

of = 0 prices are flexible. Each firm assembles labour (supplied by the workers) and (finished)

entrepreneurial capital to operate a constant return to scale production function for the variety

of the intermediate good:

19


21/41

() = (() ()) (25)

Each monopolistic firm chooses a sequence {() () ()} taking nominal wage rates

and the rental rate of capital as given, in order to maximize expected discounted nominal

profits:

0{X=0

0[()() (() + ())

2

()

1()

2]} (26)

subject to the constraint () (), where 0 is the households stochastic discount

factor.

Lets denote by {}

=0 the sequence of Lagrange multipliers on the above demand constraint,

and by ()

the relative price of variety The first order conditions of the above problem

read:

()= (27)

()= (28)

0 = ((1 ) +

1

1+ (29)

+

+1

+1

+1

+12

where is the marginal product of labour, the marginal product of capital and =

1

is the gross aggregate inflation rate (its steady state value, , is equal to 1). Notice that all

firms employ an identical capital/labour ratio in equilibrium, so individual prices are all equal in

equilibrium. The Lagrange multiplier plays the role of the real marginal cost of production.

In a symmetric equilibrium = 1 This allows to rewrite equation 29 in the following form:

( ) = {+1(+1 )+1} + (30)

+()

(

1

)

20


22/41

The above equation is a non-linear forward looking New-Keynesian Phillips curve, in which

deviations of the real marginal cost from its desired steady state value are the driving force of

inflation.7

4.3.1 Capital Producers

A competitive sector of capital producers combine investment (expressed in the same composite as

the final good, hence with price ) and existing capital stock to produce new capital goods. This

activity entails physical adjustment costs. The corresponding CRS production function is ( )

so that capital accumulation obeys:

+1 = (1 ) + (

) (31)

where () is increasing and convex.

Define as the re-sell price of the capital good. Capital producers maximize profits (

)

implying the following first order condition:

0(

) = (32)

The gross (nominal) return from holding one unit of capital between and + 1 is composed of

the rental rate plus the re-sell price of capital (net of depreciation and physical adjustment costs):

+ ((1 ) 0(

)

+ (

)) (33)

The gross (real) return to entrepreneurs from holding a unit of capital between and + 1 is

equalized in equilibrium to the gross (real) return that entrepreneurs return to banks for their loan

services, +1:

+1

+1

+1

(34)

7 Woodford [42].

21


23/41

4.4 Goods Market Clearing

Equilibrium in thefi

nal good market requires that the production of thefi

nal good equals thesum of private consumption by households and entrepreneurs, investment, public spending, and

the resource costs that originate from the adjustment of prices:

= + + +

2( )

2 (35)

In the above equation, is government consumption of the final good which evolves exogenously

and is assumed to be financed by lump sum taxes.

4.5 Monetary Policy

We assume that monetary policy is conducted by means of an interest rate reaction function of this

form:

ln

1 + 1 +

= (1 )

ln

+ ln

+ ln

+ ln

(36)

+ ln1 + 1

1 + All variables are deviations from the target or steady state (symbols without time subscript).

Note that the reaction function includes two alternative terms that express leaning-against-the-wind

behavior, respectively a reaction to asset prices () or to the deposit ratio (). Our approach

will consist in finding policy specifications { } that maximize household welfare8

We solve the model by computing a second order approximation of the policy functions around the

non-stochastic steady state.

4.6 Parameter values

Household preferences and production. The time unit is the quarter. The utility function of house-

holds is ( ) =1 11 + log(1) with = 2 as in most reasl business cycle literature.

We set set equal to 3, chosen in such a way to generate a steady-state level of employment

8 See ofr instance Kim and Kim 2003, Kollmann [?], [?] Schmitt-Grohe and Uribe [38], [39], [40], Faia and Monacelli[24], Faia [23].

22


24/41


25/41

deviation to the standard deviation of a uniform distribution. The result, 0.5, is our benchmark.

To this we apply a shock as described in the next section.

One way to interpret is to see it as the ratio of two present values of the project, the first at the

interest rate applied to firms external finance, the second discounted at the bank internal finance

rate (the money market rate). A benchmark estimate can be obtained by taking the historical ratio

between the money market rate and the lending rate. In the US over the last 20 years, based on

30-year mortgage loans, this ratio has been around 3 percent. This leads to a value of around

0.6. In the empirical analyses we have chosen a steady state value of 0.5, to which we apply a shock

as described in the next section. Finally we parametrize the survival rate of banks at 0.97.

Shocks. Total factor productivity is assumed to evolve as:

= 1 exp(

) (37)

where the steady-state value is normalized to unity (which in turn implies = 1) and

where is an i.i.d. shock with standard deviation In line with the real business cycle literature,

we set = 095 and = 0008. Log-government consumption is assumed to evolve according to

the following process:

ln(

) = ln(

1

) +

where is the steady-state share of government consumption (set in such a way that

= 025)

and is an i.i.d. shock with standard deviation We follow the empirical evidence for the U.S.

in Perotti [?] and set = 00074 and = 09

We introduce a monetary policy shock as as an additive disturbance to the interest rate

set through the monetary policy rule. The monetary policy shock is assumed to have zero or

moderate persistence, depending on different cases examined. Following empirical evidence for US

and Europe, the standard deviations of the shocks is set to 0006.

5 The transmission of shocks in the model

Charts 4 to 7 show the effects in the model of four types of shocks: a positive productivity shock;

an expansionary fiscal shock; a restrictive monetary policy shock; and a risk shock. The first two

24


26/41

mainly serve the purpose of illustrating the properties of the model; the other two are of more

direct relevance for our investigation.

In each of the charts, the results from our model (AFBGG model), represented in green dashed

lines, are compared with those of two other models: the one of Angeloni and Faia [6] (AF model;

solid blue line) and the pure financial accelerator (BGG model; dash&dot red line).

The productivity shock (chart 4) produces in all three models the usual pattern of declining

inflation and rising output in the short run, rapidly returning to the steady state. Relative to

the other models, AFBGG displays a stronger amplification of output on impact, followed by

undershooting before returning to baseline. Similarly, AFBGG displays much more volatility of

investment and capital in the short run. Employment displays a hump-shaped pattern, declining

first and rising back. In contrast with the AF model (fourth row of plots) the deposit ratio declines

on impact, and this determines a sharp decline in bank riskiness. This pattern reflects the decline

in firms leverage and the associated decline in bank lending (fifth row of plots in the chart). Note

that, in contrast to BGG, this shock produces a slight increase in the external finance premium.

A persistent fiscal expansion (chart 5) increases output and inflation, as expected, in all models,

but the size and timing of the effects are quite different. Again, AFBGG amplifies the effect on

output and capital, as a consequence of the interaction between thefi

nancial accelerator and thebehavior of banks. Bank riskiness rises as a result of the increase in the deposit ratio and the

interest rate. The chain of effects under a fiscal shock tends to be opposite to that observed in the

case of a productivity shock.

Chart 6 shows the effect of a persistent monetary policy restriction. Inflation and output

decline, as in the data, but the model response does not exhibit the usual hump-shape that observed

in VAR model responses. Note also that the impact of the shock on banks is sharply different in

the AFBGG model relative to the AF model. In the first, an increase in the deposit ratio and (after

a lag) in firm leverage is accompanied by an increase in risk. On the contrary, in the AF model

deposit decline relative to loans, hence reducing bank leverage. This generates a decrease in bank

riskiness, that matches the results of our VAR estimates (see chart 3).

Finally, chart 7 shows the results of an increase in market risk, modelled as a positive shock in

the external finance premium accompanied by a simultaneous decline in the value of the parameter

25


27/41


28/41


29/41

[10] Bekaert, G., M. Hoerova and M. Lo Duca, 2009, "Risk, uncertainty and monetary policy",

mimeo, ECB, Frankfurt.

[11] Bekaert, G., M. Hoerova and Scheicher, 2009, "Uncertainty and Risk Attitude: Evidence from

Equity Markets", ECB Working Paper No. 1037.

[12] Bils M. and P. Klenow (2004), "Some Evidence on the Importance of Sticky Prices", Journal

of Political Economy, October.

[13] Bloom, Nicholas 2009, "The Impact of Uncertainty Shocks," Econometrica, May.

[14] Bloom, Nicholas, Max Floteotto and Nir Jiaimovich, 2009. "Really Uncertain Business Cycles".Mimeo.

[15] Blum, J. and M. Hellwig, 1995. "The Macroeconomic Implications of Capital Requirements

Adequacy for Banks". European Economic Review, vol. 39.

[16] Borio, Claudio and H. Zhu (2008), Capital regulation, risk-taking and monetary policy: a

missing link in the transmission mechanism?, BIS Working Papers No. 268.

[17] Borio, Claudio and Philip Lowe (2002), "Asset prices, financial and monetary stability: ex-ploring the nexus", BIS Working Papers No 114.b

[18] Brunnermeier, M., A, Crockett, C. Goodhart, M. Hellwig, A. Persaud and H. Shin, 2009, The

Fundamental Principles of Financial Regulation", Geneva Report on the World Economy 11,

CEPR and ICMB.

[19] Carlstrom, C. and T. Fuerst (1997), Agency Costs, Net Worth and Business Fluctuations: A

Computable General Equilibrium Analysis, American Economic Review, 87, 893-910.

[20] Cecchetti, S.G., H. Genberg, J. Lipsky and S. Whadwani, 2000, Asset Prices and Central

Bank Policy, Geneva Report on the World Economy 2, CEPR and ICMB.

[21] Diamond, Douglas W. and Rajan, Raghuram G. 2000. "A Theory of Bank Capital", Journal

of Finance, vol. LV. No. 6.

28


30/41

[22] Douglas W. Diamond & Raghuram G. Rajan, 2001. "Liquidity Risk, Liquidity Creation and

Financial Fragility: A Theory of Banking," Journal of Political Economy.

[23] Faia, Ester, 2008 "Optimal Monetary Policy Rules with Labor Market Frictions", Journal of

Economic Dynamics and Control.

[24] Faia, Ester, Monacelli, Tommaso, 2007. "Optimal interest rate rules, asset prices, and credit

frictions", Journal of Economic Dynamics and Control.

[25] Gertler, Mark and Peter Karadi, 2009. "A model of Unconventional Monetary Policy". NYU.

[26] Gertler, Mark and Nobuhiro Kiyotaki, 2009, "Financial Intermediation and Credit Policy inBusiness Cycle Analysis", mimeo.

[27] Giavazzi, F. and F.S. Mishkin (2006), An Evolution of Swedish Monetary Policy between

1995 and 2005, Riksdagstryckeriet, Stockholm.

[28] Goodhart, Charles and Boris Hofmann (2008), "House Prices, Money, Credit and the Macro-

economy", ECB Working paper Series n. 802.

[29] Goodhart, Charles and Dirk Schoenmaker (1995), "Should the Functions of Monetary Policyand Banking Supervision be Separated?, Oxford Economic Papers, 47, 539560

[30] Jimnez, J., S. Ongena, J.-L. Peydr-Alcalde and J. Taurina (2007), Hazardous Times for

Monetary Policy: What do Twenty-Three Million Bank Loans Say About the Effects of Mon-

etary Policy on Credit Risk?, CEPR Discussion Paper 6514.

[31] Kashyap, A. and F. Mishkin (2009), "The Fed is Already Transparent", the Wall Street

Journal, 10 November.

[32] Maddaloni A., and J.L. Peydr Alcalde, 2009, Bank risk-taking, Securitiza-

tion and Monetary Policy; Evidence from the bank Lending Survey, ECB,

mimeo. Downloadable at: http://reserved03.timeandmind.net/carefin.unibocconi.it/

download/21_settembre_Intesa_Sanpaolo/peydro-maddaloni.pdf

[33] Mishkin, Rick and John Taylor, 2005), "...... " Journal of Economic Perspectives

29


31/41

[34] Morris, S. and H.S. Shin, 2008, "Financial Regulation in a System Context." Brookings Papers

of Economic Activity.

[35] Rajan, R. (2005), Has Financial Development Made the World Riskier?, NBER Working

Paper 11728.

[36] Sbordone A. (2002), "Prices and Unit Labor Costs: A New Test of Price Stickiness", Journal

of Monetary Economics Vol. 49 (2).

[37] Reinhardt, Carmen and Ken Rogoff, 2009, "Is the 2007 U.S. sub-prime financial crisis so

different? An international comparison", NBER Working Paper 13761.

[38] Schmitt-Grohe S. and M. Uribe, 2003. "Optimal, Simple and Implementable Monetary and

Fiscal Rules." Mimeo Duke University.

[39] Schmitt-Grohe, Stephanie and Uribe, Martin, 2004a. "Solving Dynamic General Equilibrium

Models Using a Second-Order Approximation to the Policy Function" Journal of Economic

Dynamics and Control 28, 645-858.

[40] Schmitt-Grohe S. and M. Uribe, 2004b, "Optimal Operational Monetary Policy in Christiano-

Eichenbaum-Evans Model of the U.S. Business Cycle." Mimeo Duke University.

[41] Taylor, J.B. (2009), The Financial Crisis and the Policy Responses: An Empirical Analysis

of What Went Wrong, NBER Working Paper Series No. 14631.

[42] Woodford, Michael, 2003 Interest and Prices.

30


32/41

Table 1. Granger Causality Tests

Panel A. United States

Equation Excluded Lags p-Value

monetary policy leverage I 25 0.321

leverage I monetary policy 25 0.002 ***

monetary policy leverage II 17 0.591

leverage II monetary policy 17 0.058 *

monetary policy balance sheet risk I 14 0.672

balance sheet risk I monetary policy 14 0.133

monetary policy balance sheet risk II 25 0.900

balance sheet risk II monetary policy 25 0.005 ***

monetary policy balance sheet risk III 14 0.540

balance sheet risk III monetary policy 14 0.030 **

monetary policy balance sheet risk IV 25 0.855

balance sheet risk IV monetary policy 25 0.009 ***

Panel B. Euro Area

Equation Excluded Lags p-Value

monetary policy leverage I 13 0.001 ***

leverage I monetary policy 13 0.954

monetary policy leverage II 5 0.002 ***

leverage II monetary policy 5 0.016 **

monetary policy balance sheet risk I 18 0.000 ***

balance sheet risk I monetary policy 18 0.233

monetary policy balance sheet risk II 19 0.000 ***

balance sheet risk II monetary policy 19 0.000 ***

monetary policy balance sheet risk III 3 0.090 *

balance sheet risk III monetary policy 3 0.042 **

monetary policy balance sheet risk IV 13 0.316

balance sheet risk IV monetary policy 13 0.113

Notes: The table reports the results of granger causality tests of bivariate VAR models, where theendogenous variables are monetary policy (de-trended effective federal fund rate) and different

indicators capturing leverage and balance sheet risk of banks (see table X for the description of the

variables). Column Equation specifies the VAR equation; column Excluded reports the variable

excluded from the equation; column Lags specifies the number of lags used in the estimation of the

VAR model, lags are selected by looking at Akaike information criterion. Column p-value specifiesthe p-value of the granger causality test, rejection at 1% (5%) confidence level is marked with *** (**).


33/41

Chart 1: Impulse Responses to a monetary policy shock (macro variables)

US inflation EA inflation

-0.0040

-0.0030

-0.0020

-0.0010

0.0000

0.0010

0.0020

0.0030

0.0040

0.0050

0 4 8 12 16 20 24 28 32 36

-0.0020

-0.0015

-0.0010

-0.0005

0.0000

0.0005

0.0010

0.0015

0 4 8 12 16 20 24 28 32 36

US industrial production EA industrial production

-0.0070

-0.0060

-0.0050

-0.0040

-0.0030

-0.0020

-0.0010

0.0000

0.0010

0.0020

0 4 8 12 16 20 24 28 32 36

-0.0050

-0.0040

-0.0030

-0.0020

-0.0010

0.0000

0.0010

0.0020

0.0030

0 4 8 12 16 20 24 28 32 36

US employment EA unemployment

-0.0040

-0.0035

-0.0030

-0.0025

-0.0020

-0.0015

-0.0010

-0.0005

0.0000

0.0005

0.0010

0 4 8 12 16 20 24 28 32 36

-0.0010

-0.0008

-0.0006

-0.0004

-0.0002

0.0000

0.0002

0.0004

0.0006

0 4 8 12 16 20 24 28 32 36

Notes: for the calculation of impulse responses we use the same estimation order of thevariables as in Bloom (2009), with the difference that we insert the bank balance sheet

variables measuring risk and leverage between the macro and the financial block. Therefore,

the variables in the estimation order are industrial production, employment, inflation, bank

balance sheet risk, bank leverage, monetary policy rate, the stock market volatility indicator

and the stock market index. As pointed out by Bloom, including the stock-market levels as thelast variable in the VAR ensures the impact of stock market levels is already controlled for

when looking at the impact of volatility shocks. The ordering used here is based on the

assumption that shocks instantaneously influence the stock market (levels and volatility), thenthe interest rate. Subsequently bank balance sheet variables adjusts, followed by macro

adjustments (first the price index and then quantities). As it is not clear which variable amongleverage and balance sheet risk should be placed first, we estimate different VAR models

inverting the order of these two variables. The results are not affected. For the US, the model

is estimated with monthly data from January 1974 to June 2008. For the euro area the sample

covers the period March 1991 to December 2008.


34/41


35/41


36/41


37/41

0 5 10 15 200

2

4

6

8x 10

-3 Inflation

AF

AFBGGBGG

0 5 10 15 20-0.1

0

0.1

0.2Output

0 5 10 15 200

0.1

0.2

0.3

0.4Employment

0 5 10 15 20-0.03

-0.02

-0.01

0Capital

0 5 10 15 20-0.1

-0.05

0

0.05

0.1Tobin Q

0 5 10 15 200

0.01

0.02

0.03Interest rate

0 5 10 15 20-0.05

0

0.05

0.1

0.15 Deposit ratio (specific AF)

0 5 10 15 20-0.2

0

0.2

0.4

0.6Bank riskiness (specific AF)

0 5 10 15 200

1

2

3

4x 10

-4External finance premium (specific BGG)

0 5 10 15 20-0.05

0

0.05

0.1

0.15Firm leverage ratio (specific BGG)

CHART 5: EXPANSIONARY FISCAL SHOCK


38/41

0 5 10 15 20-0.2

-0.1

0

0.1

0.2Inflation

AF

AFBGGBGG

0 5 10 15 20-4

-2

0

2Output

0 5 10 15 20-6

-4

-2

0

2Employment

0 5 10 15 20-1

-0.5

0Capital

0 5 10 15 20-1.5

-1

-0.5

0

0.5Tobin Q

0 5 10 15 200

0.2

0.4

0.6

0.8Interest rate

0 5 10 15 20-1

-0.5

0

0.5

1 Deposit ratio (specific AF)

0 5 10 15 20-2

-1

0

1

2Bank riskiness (specific AF)

0 5 10 15 200

0.005

0.01

0.015External finance premium (specific BGG)

0 5 10 15 20-2

-1

0

1Firm leverage ratio (specific BGG)

CHART 6: RESTRICTIVE MONETARY POLICY SHOCK


39/41

0 5 10 15 20-0.2

-0.1

0

0.1

0.2Inflation

AF

AFBGG

BGG

0 5 10 15 20-10

-5

0

5Output

0 5 10 15 20-15

-10

-5

0

5Employment

0 5 10 15 20-1.5

-1

-0.5

0Capital

0 5 10 15 20-3

-2

-1

0

1Tobin Q

0 5 10 15 20-1

-0.5

0

0.5Interest rate

0 5 10 15 20-2

0

2

4 Deposit ratio (specific AF)

0 5 10 15 20-10

0

10

20Bank riskiness (specific AF)

0 5 10 15 200

0.005

0.01

0.015

0.02External finance premium (specific BGG)

0 5 10 15 20-4

-2

0

2

4Firm leverage ratio (specific BGG)

CHART 7: POSITIVE RISK SHOCK


40/41


41/41

APPENDIX Table A1: Variables Description - euro area

Variable name DescriptionBaseline

Model

Stock marketDe-trended logarithm of the Eurostoxx index. Source:

Authors calculation and DatastreamY

Monetary policy rateDe-trended EONIA rate (de-trended German overnight

rate before 1999). Source: Authors calculation and ECB.Y

InflationAnnualised rate of change of the HCPI over the last 3

months. Source: Authors calculation and ECB.Y

Industrial production

De-trended logarithm of the industrial production index

for manufacturing. Source: Source: Authors calculationand OECD.

Y

Employment De-trended logarithm of total employment1

. Source:Authors calculation and Eurostat.

Y

Uncertainty shock I

(RISK1)

This is the same variable used for the US, see the

description of Uncertainty Shock I for the US.Y

Balance sheet risk I

(RISK2)

De-trended ratio of consumer credit and mortgages over

total assets. The series starts in 1999. Source: Authorscalculation and ECB.

Y

Leverage I

(RISK3)

12 month difference of the ratio total assets to total

deposits. The series starts in 1997. Source: Authors

calculation and ECB

Y

Uncertainty shock II Implied volatility of the stock market index. Source:Datastream

N

Balance sheet risk II

12 month difference of the ratio of consumer credit and

mortgages over total assets. Source: Authors calculation

and ECB

N

Balance sheet risk III

De-trended ratio of consumer credit and mortgages over

total assets. The series start in 1991. Calculated byaggregating data for 6 euro area countries.

2Source: ECB

and authors calculations.

N

Balance sheet risk IV

12 month difference of the ratio consumer credit and

mortgages over total assets. The series start in 1991.

Calculated by aggregating data for 6 euro area countries

(see the description of Balance sheet risk III). Source:

ECB and authors calculations

N

Leverage II De-trended ratio total assets to total deposits. Source:Authors calculation and ECB

N

1

wp 2010 00 monetary policy risk taking

Documents