Description
Monetary economics is a branch of economics that historically prefigured and remains integrally linked to macroeconomics.[1] Monetary economics provides a framework for analyzing money in its functions as a medium of exchange, store of value, and unit of account.
Federal Reserve Bank of New York
Staff Reports
Financial Intermediaries and Monetary Economics
Tobias Adrian
Hyun Song Shin
Staff Report no. 398
October 2009
Revised May 2010
This paper presents preliminary findings and is being distributed to economists
and other interested readers solely to stimulate discussion and elicit comments.
The views expressed in the paper are those of the authors and are not necessarily
reflective of views at the Federal Reserve Bank of New York or the Federal
Reserve System. Any errors or omissions are the responsibility of the authors.
Financial Intermediaries and Monetary Economics
Tobias Adrian and Hyun Song Shin
Federal Reserve Bank of New York Staff Reports, no. 398
October 2009; revised May 2010
JEL classification: E00, E02, G28
Abstract
We reconsider the role of financial intermediaries in monetary economics. We explore
the hypothesis that financial intermediaries drive the business cycle by way of their role
in determining the price of risk. In this framework, balance sheet quantities emerge as a
key indicator of risk appetite and hence of the “risk-taking channel” of monetary policy.
We document evidence that the balance sheets of financial intermediaries reflect the
transmission of monetary policy through capital market conditions. Our findings suggest
that the traditional focus on the money stock for the conduct of monetary policy may
have more modern counterparts, and we suggest the importance of tracking balance sheet
quantities for the conduct of monetary policy.
Key words: financial intermediation, monetary policy, risk-taking channel
Adrian: Federal Reserve Bank of New York (e-mail: [email protected]). Shin: Princeton
University (e-mail: [email protected]). This paper is a preliminary version of a chapter
prepared for the Handbook of Monetary Economics. The views expressed in this paper are
those of the authors and do not necessarily reflect the position of the Federal Reserve Bank
of New York or the Federal Reserve System.
1. Introduction
In conventional models of monetary economics commonly commonly used in cen-
tral banks, the banking sector has not played a prominent role. The primary
friction in such models is the price stickiness of goods and services. Financial
intermediaries do not play a role, save as a passive player that the central bank
uses as a channel to implement monetary policy.
However, ?nancial intermediaries have been at the center of the global ?nancial
crisis that erupted in 2007. They have borne a large share of the credit losses from
securitized subprime mortgages, even though securitization was intended to parcel
out and disperse credit risk to investors who were better able to absorb losses.
Credit losses and the associated ?nancial distress have ?gured prominently in the
commentary on the downturn in real economic activity that followed. These
recent events suggest that ?nancial intermediaries may be worthy of separate
study in order to ascertain their role in economic ?uctuations.
The purpose of this chapter in the Handbook of Monetary Economics is to
reconsider the role of ?nancial intermediaries in monetary economics. In ad-
dressing the issue of ?nancial factors in macroeconomics, we join a spate of recent
research that has attempted to incorporate a ?nancial sector in a New Keynesian
DSGE model. Curdia and Woodford (2009) and Gertler and Karadi (2009) are
recent examples. However, rather than phrasing the question as how ?nancial
“frictions” a?ect the real economy, we focus on the ?nancial intermediary sector
itself. We explore the hypothesis that the ?nancial intermediary sector, far from
being passive, is instead the engine that drives the boom-bust cycle. To explore
this hypothesis, we propose a framework for study with a view to addressing the
following pair of questions. What are the channels through which ?nancial in-
termediaries exert an in?uence on the real economy (if at all), and what are the
1
implications for monetary policy?
Banks and other ?nancial intermediaries borrow in order to lend. Since the
loans o?ered by banks tend to be of longer maturity than the liabilities that fund
those loans, the term spread is indicative of the marginal pro?tability of an extra
dollar of loans on intermediaries’ balance sheets. The net interest margin (NIM)
of the bank is the di?erence between the total interest income on the asset side
of its balance sheet and the interest expense on the liabilities side of its balance
sheet. Whereas the term spread indicates the pro?tability of the marginal loan
that is added to the balance sheet, the net interest margin is an average concept
that applies to the stock of all loans and liabilities on the balance sheet.
The net interest margin determines the pro?tability of bank lending and in-
creases the present value of bank income, thereby boosting the forward-looking
measures of bank capital. Such a boost in bank capital increases the capacity
of the bank to increase lending in the sense that the marginal loan that was not
made before the boost in bank capital now becomes feasible under the greater
risk-bearing capacity of the bank. As banks expand their balance sheets, the
market price of risk falls.
In this framework, ?nancial intermediaries drive the ?nancial cycle through
their in?uence on the determination of the price of risk. Quantity variables -
particularly the components of ?nancial intermediary balance sheets - emerge as
important economic indicators due to their role in re?ecting the risk capacity
of banking sector and hence on the marginal real project that receives funding.
In this way, the banking sector plays a key role in determining the level of real
activity. Ironically, our ?ndings have some points of contact with the older
theme in monetary economics of keeping track of the money stock at a time
when it has fallen out of favor among monetary economists.
1
The common
1
See Friedman (1988) for an overview of the role of monetary aggregates in macroeconomic
2
theme between our framework and the older literature is that the money stock
is a balance sheet aggregate of the ?nancial sector. Our approach suggests that
broader balance sheet aggregates such as total assets and leverage are the relevant
?nancial intermediary variables to incorporate into macroeconomic analysis.
When we examine balance sheet measures that re?ect the underlying funding
conditions in capital markets, we ?nd that the appropriate balance sheet quantities
are of institutions that are marking to market their balance sheets. In this regard,
?uctuations in shadow bank and broker-dealer assets are more informative than
movements in commercial bank assets. However, as commercial banks begin
to mark more items of their balance sheets to market, commercial bank balance
sheet variables are likely to become more important variables for studying the
transmission mechanism.
Our ?ndings have important implications for the conduct of monetary pol-
icy. According to the perspective outlined here, ?uctuations in the supply of
credit arise from the interactions between bank risk-taking and the market risk
premium. The cost of leverage of market-based intermediaries is determined by
two main variables — risk and risk-taking capacity. The expected pro?tability of
intermediaries is proxied by spreads such as the term spread and various credit
spreads. Variations in the policy target determine short-term interest rates, and
have a direct impact on the pro?tability of intermediaries. For these reasons,
short-term interest rates matter directly for monetary policy.
The e?ect of keeping policy rates low in the aftermath of the ?nancial crisis of
2008 has illustrated again the potency of low policy interest rates in raising the
pro?tability of banks and thereby recapitalizing the banking system from their
dangerous low levels. When considering the debates in early 2009 about the
necessity (or inevitability) of capital injections into the U.S banking system, the
?uctuations in the United States.
3
turnaround in the capital levels of the U.S. banking sector has been worthy of
note.
Empirically, there is (for the United States) a near perfect negative one-to-
one relationship between 4-quarter changes of the Fed Funds target and 4-quarter
changes of the term spread de?ned as the 10-year/3-month term Treasury spread
(Figure 1.1 uses data from 1987q1 to 2008q3). Thus, shifts in the policy rate
translate directly into shifts in the slope of the yield curve. Since the term spread
a?ects the pro?tability of the marginal loan and the future net interest margin
(NIM) of the bank, the short rate signals future risk-taking capacity of the banking
sector. In this way, variations in the target rate a?ect real activity because they
change the risk-taking capacity of ?nancial intermediaries, thus shifting market
risk premiums and the supply of credit. Borio and Zhu (2008) have coined
the term “risk-taking channel” of monetary policy to describe this set of e?ects
working through the risk appetite of ?nancial intermediaries.
This perspective on the importance of the short rate as a price variable is
in contrast to current monetary thinking, where short-term rates matter only to
the extent that they determine long term interest rates, which are seen as being
risk-adjusted expectations of future short rates. Current models of monetary
economics used at central banks emphasize the importance of managing market
expectations. By charting a path for future short rates and communicating this
path clearly to the market, the central bank can in?uence long rates and thereby
in?uence mortgage rates, corporate lending rates, and other prices that a?ect
consumption and investment. This “expectations channel”, which is explained
in Bernanke (2004), Svensson (2004), and Woodford (2003, 2005), has become an
important consideration for monetary policy. In his book on central banking,
Alan Blinder (1998, p.70) phrases the claim in a particularly clear way.
“central banks generally control only the overnight interest rate,
4
Figure 1.1: The term spread and the Federal Funds rate.
an interest rate that is relevant to virtually no economically inter-
esting transactions. Monetary policy has important macroeconomic
e?ects only to the extent that it moves ?nancial market prices that
really matter - like long-term interest rates, stock market values and
exchange rates.”
In contrast, our results suggest that short-term rates may be important in their
own right. Short rates matter because they largely determine the term spread,
which in turn determine the net interest margin and the forward-looking capital
of the banking sector. Continued low short rates imply a steep yield curve for
some time, higher net interest margin in the future, and hence higher risk-taking
capacity of the banking sector. Conversely, higher short rates imply lower future
net interest margin and a decline in the risk-taking capacity of the banking sector.
In particular, an inverted yield curve is a sign of diminished risk-taking capacity,
5
and by extension of lower real activity.
There is empirical support for the risk-taking channel of monetary policy. We
?nd that the growth in shadow bank balance sheets and broker-dealer balance
sheets help to explain future real activity. However, we also ?nd that ?uc-
tuations in the balance sheet size of shadow banks and security broker-dealers
appear to signal shifts in future real activity better than the ?uctuations of the
larger commercial banking sector. Thus, one lesson from our empirical analysis
is that there are important distinctions between di?erent categories of ?nancial
intermediaries. In fact, the evolutions of shadow bank and broker-dealer assets
have time signatures that are markedly di?erent from those of commercial banks.
Our results point to key di?erences between banking, as traditionally conceived,
and the market-based banking system that has become increasingly in?uential in
charting the course of economic events.
Having established the importance of ?nancial intermediary balance sheets in
signaling future real activity, we go on to examine the determinants of balance
sheet growth. We ?nd that short-term interest rates are important. Indeed, the
level of the Fed Funds target is a key variable: a lowering of short-term rates are
conducive to expanding balance sheets. In addition, a steeper yield curve, larger
credit spreads, and lower measures of ?nancial market volatility are conducive to
expanding balance sheets. In particular, an inverted yield curve is a harbinger of
a slowdown in balance sheet growth, shedding light on the empirical feature that
an inverted yield curve forecasts recessions. The Fed Funds target determines
other relevant short term interest rates, such as repo rates and interbank lending
rates through arbitrage in the money market. As such, we may expect the Fed
Funds rate to be pivotal in setting short-term interest rates more generally.
These ?ndings re?ect the economics of ?nancial intermediation, since the busi-
ness of banking is to borrow short and lend long. For an o?-balance sheet vehicle
6
such as a conduit or SIV (structured investment vehicle) that ?nances holdings of
mortgage assets by issuing commercial paper, a di?erence of a quarter or half per-
cent in the funding cost may make all the di?erence between a pro?table venture
and a loss-making one. This is because the conduit or SIV, like most ?nancial
intermediaries, is simultaneously both a creditor and a debtor — it borrows in
order to lend.
The outline of this chapter is as follows. We begin with a simple equilibrium
model where ?nancial intermediaries are the main engine for the determination
of the price of risk in the economy. We then present empirical results on the real
impact of shadow bank and broker-dealer balance sheet changes, and on the role
of short-term interest rates in the determination of balance sheet changes. We
also consider the role of the central bank as the lender of last resort (LOLR) in
light of our ?ndings. We conclude by drawing some lessons for monetary policy.
2. Financial Intermediaries and the Price of Risk
To motivate the study of ?nancial intermediaries and how they determine the
price of risk, we begin with a stylized model set in a one period asset market.
2
The general equilibrium model below is deliberately stark. It has two features
that deserve emphasis.
First, there is no default in the model. The debt that appears in the model
is risk-free. However, as we will see, the ampli?cation of the ?nancial cycle is
present. Geanakoplos (2009) has highlighted how risk-free debt may still give
rise to powerful spillover e?ects through ?uctuations in leverage and the pricing
of risk. The model also incorporates insights from Shleifer and Vishny (1997),
who demonstrate that ?nancial constraints can lead to ?uctuations of risk premia
2
A similar model appeared in Shin (2009).
7
even if arbitrageurs are risk neutral.
3
Adrian and Shin (2007) exhibit empirical
evidence that bears on the ?uctuations in the pricing of risk from the balance
sheets of ?nancial intermediaries.
Second, in the example, there is no lending and borrowing between ?nancial
intermediaries themselves. So, any e?ect we see in the model cannot be attributed
to what we may call the “domino model” of systemic risk, where systemic risk
propagates through the ?nancial system via a chain of defaults of ?nancial inter-
mediaries.
4
This is not to say, of course, that interlocking claims do not matter.
However, the benchmark case serves the purpose of showing that chains of default
are not necessary for ?uctuations in the price of risk.
To anticipate the punchline from the simple model, we show that aggregate
capital of the ?nancial intermediary sector stands in a one-to-one relation with
the price of risk and the availability of funding that ?ows to real projects. The
larger is the aggregate intermediary sector capital, the lower is the price of risk,
and the easier is credit.
2.1. Model
Today is date 0. A risky security is traded today in anticipation of its realized
payo? in the next period (date 1). The payo? of the risky security is known at
date 1. When viewed from date 0, the risky security’s payo? is a random variable
˜ n, with expected value ¡ 0. The uncertainty surrounding the risky security’s
payo? takes a particularly simple form. The random variable ˜ n is uniformly
distributed over the interval:
[¡ ?.. ¡ + .]
3
Shleifer and Vishny (2009) present a theory of unstable banking that is closely related to
our model.
4
See Adrian and Shin (2008c) for an argument for why the “domino model” is inappropriate
for understanding the crisis of 2007 -9.
8
The mean and variance of ˜ n is given by
1 ( ˜ n) = ¡
o
2
=
.
2
3
There is also a risk-free security, which we call “cash”, that pays an interest rate
of i. Let j denote the price of the risky security. For an investor with equity c
who holds ¸ units of the risky security, the payo? of the portfolio is the random
variable:
\ ? ˜ n¸ + (1 + i) (c ?j¸) (2.1)
= ( ˜ n ?(1 + i) j)
| {z }
risky excess return
¸ + (1 + i) c
| {z }
risk-free ROE
(2.2)
There are two groups of investors - passive investors and active investors. The
passive investors can be thought of as non-leveraged investors such as households,
pension funds and mutual funds, while the active investors can be interpreted as
leveraged institutions such as banks and securities ?rms who manage their bal-
ance sheets actively. The risky securities can be interpreted as loans granted to
ultimate borrowers or securities issued by the borrowers, but where there is a risk
that the borrowers do not fully repay the loan. Figure 2.1 depicts the relation-
ships. Under this interpretation, the market value of the risky securities can be
thought of as the marked-to-market value of the loans granted to the ultimate
borrowers. The passive investors’ holding of the risky security can then be inter-
preted as the credit that is granted directly by the household sector (through the
holding of corporate bonds, for example), while the holding of the risky securities
by the active investors can be given the interpretation of intermediated ?nance
where the active investors are banks that borrow from the households in order to
lend to the ultimate borrowers.
9
Banks
(Active
Investors)
Households
(Passive
Investors)
end-user
borrowers
Intermediated
Credit
Debt
Claims
Directly granted credit
Figure 2.1: Intermediated and Directly Granted Credit
We assume that the passive investors have mean-variance preferences over the
payo? from the portfolio. They aim to maximize
l = 1 (\) ?
1
2t
o
2
W
(2.3)
where t 0 is a constant called the investor’s “risk tolerance” and o
2
W
is the
variance of \. In terms of the decision variable ¸, the passive investor’s objective
function can be written as
l (¸) = (¡,j ?(1 + i))
| {z }
Expected Excess Return
j¸ + (1 + i) c ?
1
6t
¸
2
.
2
(2.4)
The optimal holding of the risky security satis?es the ?rst order condition:
(¡ ?(1 + i) j) ?
1
3t
¸.
2
= 0
The price must be below the expected payo? for the risk-averse investor to hold
any of the risky security. The optimal risky security holding of the passive investor
(denoted by ¸
1
) is given by
¸
1
=
?
?
?
?
?
3t
.
2
(¡ ?(1 + i) j) if ¡ j (1 + i)
0 otherwise
(2.5)
10
These linear demands can be summed to give the aggregate demand. If t
i
is the
risk tolerance of the ith investor and t =
P
i
t
i
, then (2.5) gives the aggregate
demand of the passive investor sector as a whole.
Now turn to the portfolio decision of the active (leveraged) investors. These
active investors are risk-neutral but face a Value-at-Risk (VaR) constraint, as is
commonly the case for banks and other leveraged institutions.
5
The general VaR
constraint is that the capital cushion be large enough that the default probability
is kept below some benchmark level. Consider the special case where that bench-
mark level is zero. This is an extreme assumption which we adopt for the purpose
of simplifying the model. By setting the Value-at-Risk constraint to allow no de-
fault by the bank, we can treat bank liabilities as a perfect substitute for cash. It
would be possible to allow for a less stringent Value-at-Risk constraint that allows
possible default by the bank, but then the modeling has to make allowance for
the bank’s liabilities being risky debt and being priced accordingly. However, the
key qualitative features of the model would be una?ected. Thus, in what follows,
we will adopt the stringent version of the VaR constraint where the bank holds
enough capital to meet the worst case loss, and where the bank’s liabilities are
risk-free.
Denote by VaR the Value-at-Risk of the leveraged investor. The constraint is
that the investor’s capital (equity) c be large enough to cover this Value-at-Risk.
The optimization problem for an active investor is:
max
j
1 (\) subject to VaR ? c (2.6)
If the price is too high (i.e. when j ¡, (1 + i) so that the price exceeds
the discounted expected payo?) the investor holds no risky securities. When
j < ¡, (1 + i), then 1 (\) is strictly increasing in ¸, and so the Value-at-Risk
5
A microfoundation for the VaR constraint is provided by Adrian and Shin (2008b).
11
constraint binds. The optimal holding of the risky security can be obtained by
solving VaR = c. To solve this equation, write out the balance sheet of the
leveraged investor as
Assets Liabilities
securities, j¸
equity, c
debt, j¸ ?c
For each unit of the risky security, the minimum payo? is ¡ ? .. Thus, the
worst case loss is (j (1 + i) ?(¡ ?.)) ¸. In order for the bank to have enough
equity to cover the worst case loss, we require:
(j (1 + i) ?(¡ ?.)) ¸ ? c (2.7)
This inequality also holds in the aggregate. The left hand side of (2.7) is the
Value-at-Risk (the worst possible loss), which must be met by the equity bu?er
c. Since the constraint binds, the optimal holding of the risky securities for the
leveraged investor is
¸ =
c
j (1 + i) ?(¡ ?.)
(2.8)
So the demand from the bank for the risky asset depends positively on the expected
excess return to the risky asset ¡?(1 + i) j, and positively on the amount of equity
that the bank is endowed with c.
Since (2.8) is linear in c, the aggregate demand of the leveraged sector has the
same form as (2.8) when c is the aggregate capital of the leveraged sector as a
whole.
Replacing the constraint (2.8) in the amount of debt j¸ ?c allows us to write
the new balance sheet as follows:
12
Assets Liabilities
securities, j¸
equity, c
debt,
¡ ?.
1 + i
¸
(2.9)
where the debt
q?:
1+i
¸ was constructed by substituting c = ?((¡,j ?(1 + i)) j ?.) ¸
into j¸ ? c. We assume that ¡ . so as to ensure that the payo? of the risky
security is non-negative. The bank’s leverage is the ratio of total assets to equity,
which can be written as:
leverage =
j¸
c
=
j
j (1 + i) ?(¡ ?.)
(2.10)
Denoting by ¸ the holding of the risky securities by the active investors and
by ¸
1
the holding by the passive investors, the market clearing condition is
¸ + ¸
1
= o (2.11)
where o is the total endowment of the risky securities. Figure 2.2 illustrates the
equilibrium for a ?xed value of aggregate capital c. For the passive investors, their
demand is linear, with the intercept at ¡, (1 + i). The demand of the leveraged
sector can be read o? from (2.8). The solution is fully determined as a function
of c. In a dynamic model, c can be treated as the state variable (see Danielsson,
et al. (2009)).
Now consider a possible scenario involving an improvement in the fundamen-
tals of the risky security where the expected payo? of the risky securities rises
from ¡ to ¡
0
. In our banking interpretation of the model, an improvement in the
expected payo? should be seen as an increase in the marked-to-market value of
bank assets. For now, we simply treat the increase in ¡ as an exogenous shock.
Figure 2.3 illustrates the scenario. The improvement in the fundamentals of the
risky security pushes up the demand curves for both the passive and active in-
vestors, as illustrated in Figure 2.3. However, there is an ampli?ed response from
13
p
0
S
demand of
passive investors
demand of
VaR-constrained
investors
i
q
? 1 i
q
? 1
Figure 2.2: Market Clearing Price
the leveraged investors as a result of marked-to-market gains on their balance
sheets.
From (2.9), denote by c
0
the new equity level of the leveraged investors that
incorporates the capital gain when the price rises to j
0
. The initial amount of
debt was
q?:
1+i
¸. Since the new asset value is j
0
¸, the new equity level c
0
is
c
0
= (j
0
(1 + i) ?(¡ ?.)) ¸ (2.12)
Figure 2.4 breaks out the steps in the balance sheet expansion. The initial balance
sheet is on the left, where the total asset value is j¸. The middle balance sheet
shows the e?ect of an improvement in fundamentals that comes from an increase
in ¡, but before any adjustment in the risky security holding. There is an increase
in the value of the securities without any change in the debt value, since the debt
was already risk-free to begin with. So, the increase in asset value ?ows through
entirely to an increase in equity. Equation (2.12) expresses the new value of
equity c
0
in the middle balance sheet in Figure 2.4.
The increase in equity relaxes the value at risk constraint, and the leveraged
14
p
0
S
? ? i q ? 1 /
' p
? ? i q ? 1 / '
Figure 2.3: Ampli?ed response to improvement in fundamentals ¡
Initial
balance sheet
After q shock
Final
balance sheet
debt
equity
assets
increase
in equity
equity
assets
debt
assets
increase in
value of
securities
equity
debt
new
borrowing
new
purchase of
securities
Figure 2.4: Balance sheet expansion from ¡ shock
15
sector can increase its holding of risky securities. The new holding ¸
0
is larger,
and is enough to make the VaR constraint bind at the higher equity level, with a
higher fundamental value ¡
0
. That is,
c
0
= (j
0
(1 + i) ?(¡
0
?.)) ¸
0
(2.13)
After the ¡ shock, the investor’s balance sheet has strengthened, in that capital
has increased without any change in debt value. There has been an erosion of
leverage, leading to spare capacity on the balance sheet in the sense that equity
is now larger than is necessary to meet the Value-at-Risk. In order to utilize the
slack in balance sheet capacity, the investor takes on additional debt to purchase
additional risky securities. The demand response is upward-sloping. The new
holding of securities is now ¸
0
, and the total asset value is j
0
¸
0
. Equation (2.13)
expresses the new value of equity c
0
in terms of the new higher holding ¸
0
in the
right hand side balance sheet in Figure 2.4. From (2.12) and (2.13), we can write
the new holding ¸
0
of the risky security as
¸
0
= ¸
µ
1 +
¡
0
?¡
j
0
(1 + i) ?¡
0
+ .
¶
(2.14)
From the demand of passive investors (2.5) and market clearing,
(1 + i) j
0
?¡
0
=
.
2
3t
(¸
0
?o)
Substituting into (2.14),
¸
0
= ¸
Ã
1 +
¡
0
?¡
:
2
3t
(¸
0
?o) + .
!
(2.15)
This de?nes a quadratic equation in ¸
0
. The solution is where the right hand
side of (2.15) cuts the 45 degree line. The leveraged sector ampli?es booms and
busts if ¸
0
? ¸ has the same sign as ¡
0
? ¡. Then, any shift in fundamentals
16
gets ampli?ed by the portfolio decisions of the leveraged sector. The condition
for ampli?cation is that the denominator in the second term of (2.15) is positive.
But this condition is guaranteed from (2.14) and the fact that j
0
q
0
?:
1+i
(i.e. that
the price of the risky security is higher than its worst possible realized discounted
payo?).
Note also that the size of the ampli?cation is increasing when fundamental risk
is small, seen from the fact that ¸
0
?¸ is large when . is small. Recall that . is
the fundamental risk. When . is small, the associated Value-at-Risk is also small,
allowing the leveraged sector to maintain high leverage. The higher is the leverage,
the greater is the marked-to-market capital gains and losses. Ampli?cation is large
when the leveraged sector itself is large relative to the total economy. Finally,
note that the ampli?cation is more likely when the passive sector’s risk tolerance
t is high.
2.2. Pricing of Risk
We now explore the ?uctuations in risk pricing in our model. The risk premium in
our model is the excess expected return on the risky security, which can be written
in terms of the ratio of the discounted expected payo? on the risky security and
its price:
Risk premium =
¡
j (i + 1)
?1 (2.16)
Rather than working with the risk premium directly, it turns out to be more
convenient to work with a monotonic transformation of the risk premium de?ned
as
: ? 1 ?
j (1 + i)
¡
(2.17)
The “:” stands for “premium”. The variable : is a monotonic transformation
of the risk premium that varies from zero (when the risk premium is zero) to 1
17
(when the risk premium is in?nite).
The market-clearing condition for the risky security is ¸ + ¸
1
= o, which can
be written as
c
. ?¡:
+
3t
.
2
¡: = o (2.18)
Our primary interest is in the relationship between total equity c and the risk
premium :. Here, c has the interpretation of the total capital of the banking
sector, and hence its risk-taking capacity. In our model, the total lending of the
banking sector bears a very simple relationship to its total capital c, since the
holding of the risky security by the active investors (the banks) is c, (. ?¡
.
We impose the restriction that the active investors have a strictly positive total
holding of the risky security, or equivalently that the passive sector’s holding is
strictly smaller than the total endowment o. From (2.5) this restriction can be
written as
¡: <
.
2
3t
o (2.19)
By de?ning 1 (c. as below, we can write the market-clearing condition as:
1 (c. ? c +
3t
.
2
¡: (. ?¡ ?o (. ?¡ = 0 (2.20)
We then have
J1
J:
= ¡
?
?
?
3t
.
2
(. ?¡
| {z }
¹
+ o ?
3t
.
2
¡:
| {z }
1
?
?
?
(2.21)
Both ¹ and 1 are positive. ¹ is positive since the holding of the risky
security by the active sector is c, (. ?¡, and so . ? ¡: 0 in order that the
active investors hold positive holdings of the risky security. Another way to view
this condition is to note that the market price of the risky security cannot be
18
lower than the lowest possible realization of the payo? of the risky asset, so that
j
¡ ?.
1 + i
(2.22)
which can be written as (1 + i) j ?(¡ ?.) = . ?¡: 0. The second term (term
1) inside the big brackets is positive from our condition (2.19) that the passive
investors do not hold the entire supply. Since J1,Jc = 1, we have
d:
dc
= ?
J1,Jc
J1,J:
< 0 (2.23)
In other words, the market risk premium is decreasing in the total equity c
of the banking sector. As stated above, we view c as the risk-taking capacity of
the banking sector. Any shock that increases the capital bu?er of the banking
sector will lower the risk premium. We therefore have the following empirical
hypothesis.
Empirical Hypothesis 1. Risk premiums fall when the equity of the banking
system increases.
This empirical hypothesis is key to our discussion on the role of short-term
interest rates on risk-taking capacity of the banking sector, through the slope of
the yield curve and hence the greater pro?tability of bank lending. We return to
this issue shortly.
2.3. Shadow Value of Bank Capital
Another window on the risk premium in the economy is through the Lagrange
multiplier associated with the constrained optimization problem of the banks,
which is to maximize the expected payo? from the portfolio 1 (\) subject to the
Value-at-Risk constraint. The Lagrange multiplier is the rate of increase of the
objective function with respect to a relaxation of the constraint, and hence can
19
be interpreted as the shadow value of bank capital. Denoting by ` the Lagrange
multiplier, we have
` =
d1 (\)
dc
=
J1 (\)
J¸
J¸
Jc
=
¡ ?(1 + i) j
. ?(¡ ?(1 + i) j)
(2.24)
where we have obtained the expression for d1 (\) ,d¸ from (2.1) and d¸,dc is
obtained from (2.8), which gives the optimal portfolio decision of the leveraged
investor. Using our : notation, we can re-write (2.24) as
` =
¡:
. ?¡:
(2.25)
We see from (2.25) that as the risk premium : becomes compressed, the Lagrange
multiplier ` declines. The implication is that the marginal increase of a dollar’s
worth of new capital for the leveraged investor is generating less expected payo?.
As the risk premium : goes to zero, so does the Lagrange multiplier, implying
that the return to a dollar’s worth of capital goes to zero.
Furthermore, the shadow value of bank capital can be written as:
` =
. (o ?¸)
3t ?. (o ?¸)
(2.26)
The shadow value of bank capital is decreasing in the size of the leveraged
sector, given by ¸. Moreover, since there is a one-to-one relationship between
` and the risk premium :, we can also conclude that market risk premiums fall
when the size of the intermediary sector increases.
Empirical Hypothesis 2. Risk premiums fall when the size of the banking
sector increases.
3. Changing Nature of Financial Intermediation
In preparation for our empirical investigations, we brie?y review the structure
of ?nancial intermediation in the United States. In particular, we highlight the
20
increasing importance of market-based ?nancial intermediaries and the shadow
banking system.
3.1. Shadow Banking System and Security Broker-Dealers
As recently as the early 1980s, traditional banks were the dominant ?nancial
intermediaries. In subsequent years, however, they were quickly overtaken by
market-based ?nancial institutions. Figure 3.1 plots the size of di?erent types
of ?nancial intermediaries for the United States starting in 1985. We see that
market-based ?nancial intermediaries, such as security broker dealers and ABS
issuers, have become important components of the intermediary sector. The
series labeled “shadow banks” aggregates ABS issuers, ?nance companies, and
funding corporations.
0
2
0
0
0
4
0
0
0
6
0
0
0
8
0
0
0
1
0
0
0
0
B
i
l
l
i
o
n
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Security Broker-Dealers ABS issuers
Shadow Banks Commercial Banks
Note: Shadow banksareABS issuers, financecompanies, and fundingcorporation
Source: Boardof Governors of theFederal Reserve
Figure 3.1: Total Assets of Commercial Banks, Shadow Banks, and Broker-
Dealers.
In 1985, shadow banks were a tiny fraction of the commercial bank sector, but
had caught up with the commercial bank sector by the eve of the crisis. The
21
increased importance of the market-based banking system has been mirrored by
the growth of the broker-dealers, who have traditionally played market-making
and underwriting roles in securities markets. However, their importance in the
supply of credit has increased dramatically in recent years with the growth of
securitization and the changing nature of the ?nancial system toward one based
on the capital market, rather than one based on the traditional role of the bank
as intermediating between depositors and borrowers. Although total assets of
the broker-dealer sector are smaller than total assets of the commercial banking
sector, our results suggest that broker-dealers provide a better barometer of the
funding conditions in the economy, capturing overall capital market conditions.
Perhaps the most important development in this regard has been the changing
nature of housing ?nance in the US. The stock of home mortgages in the US
is now dominated by the holdings of market-based institutions, rather than by
traditional bank balance sheets. Broker-dealer balance sheets provide a timely
window on this world.
The growth of market-based ?nancial intermediaries is also re?ected in the
aggregates on the liabilities side of the balance sheet. Figure 3.2 shows the
relative size of the M1 money stock together with the outstanding stock of repos
of the primary dealers - the set of banks that bid at US Treasury security auctions,
and for whom data are readily available due to their reporting obligations to the
Federal Reserve. We also note the rapid growth of ?nancial commercial paper as
a funding vehicle for ?nancial intermediaries.
Figure 3.3 charts the relative size of M2 (bank deposits plus money market fund
balances) compared to the sum of primary dealer repos and ?nancial commercial
paper outstanding. As recently as the 1990s, the M2 stock was many times larger
than the stock of repos and commercial paper. However, by the eve of the crisis,
the gap had narrowed considerably, and M2 was only some 25% larger than the
22
0
1
0
0
0
2
0
0
0
3
0
0
0
4
0
0
0
B
i
l
l
i
o
n
1990q1 1995q1 2000q1 2005q1 2010q1
Money Stock M1 Primary Dealer Repo
Financial Commercial Paper Outstanding
Sources: Boardof Governorsof theFederal Reserveand Federal ReserveBank of New York
Figure 3.2: Liquid funding of ?nancial institutions: Money (M1), Primary Dealer
Repo, and Commercial Paper.
stock of repos and ?nancial commercial paper. However, with the eruption of the
crisis, the gap has opened up again.
Not only have the market-based intermediaries seen the most rapid growth
in the run-up to the ?nancial crisis, they were also the institutions that saw the
sharpest pull-back in the crisis itself. Figure 3.4 shows the comparative growth
rate of the total assets of commercial banks (in red) and the shadow banks (in
blue), while Figure 3.5 shows the growth of commercial paper relative to shadow
bank asset growth. We see that, whereas the commercial banks have increased
the size of their balance sheet during the crisis, the shadow banks have contracted
substantially. Traditionally, banks have played the role of a bu?er against ?uctu-
ations in capital market conditions, and we see that they have continued their role
through the current crisis. As such, looking only at aggregate commercial bank
lending may give an overly rosy picture of the state of ?nancial intermediation.
23
0
2
0
0
0
4
0
0
0
6
0
0
0
8
0
0
0
B
i
l
l
i
o
n
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Money Stock M2
Primary Dealer Repo plus Commercial Paper Outstanding
Sources: Boardof Governorsof theFederal Reserveand Federal ReserveBank of New York
Figure 3.3: Short Term Funding: M2 versus Commercial Paper + Primary Dealer
Repo.
Figure 3.6 shows that the broker-dealer sector of the economy has contracted
in step with the contraction in primary dealer repos, suggesting the sensitivity
of the broker-dealer sector to overall capital market conditions. Therefore, in
empirical studies of ?nancial intermediary behavior, it is be important to bear in
mind the distinctions between commercial banks and market-based intermediaries
such as broker dealers. Market-based intermediaries who fund themselves through
short-term borrowing such as commercial paper or repurchase agreements will be
sensitively a?ected by capital market conditions. But for a commercial bank, its
large balance sheet masks the e?ects operating at the margin. Also, commercial
banks provide relationship-based lending through credit lines. Broker-dealers, in
contrast, give a much purer signal of marginal funding conditions, as their balance
sheets consist almost exclusively of short-term market borrowing and are not as
constrained by relationship-based lending.
24
0
5
1
0
1
5
C
o
m
m
e
r
c
i
a
l
B
a
n
k
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
-
1
0
0
1
0
2
0
3
0
4
0
S
h
a
d
o
w
B
a
n
k
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Shadow Bank Asset Growth (Annual %)
Commercial Bank Asset Growth (Annual %)
Note: Shadow banksareABS issuers, financecompanies, and fundingcorporation
Source: Boardof Governors of theFederal Reserve
Figure 3.4: Total Asset Growth of Shadow Banks and of Commercial Banks.
3.2. Haircuts and Value at Risk
The Value at Risk constraint that is at the heart of the ampli?cation mechanism in
the model of Section 2 characterizes market-based ?nancial intermediaries such as
security broker-dealers and shadow banks. The active balance sheet management
of ?nancial institutions is documented in Adrian and Shin (2007), who show that
investment banks exhibit ”procyclical leverage”: i.e., increases in balance sheet
size are associated with increases in leverage. In contrast, the balance sheet
behavior of commercial banks is consistent with leverage targeting: for commercial
banks, leverage growth is uncorrelated with the growth of balance sheet size.
One useful perspective on the matter is to consider the implicit maximum lever-
age that is permitted in collateralized borrowing transactions such as repurchase
agreements (repos). Repos are the primary source of funding for market-based
?nancial institutions, as well as being a marginal source of funding for traditional
banks. In a repurchase agreement, the borrower sells a security today for a price
25
-
4
0
-
2
0
0
2
0
4
0
C
o
m
m
e
r
c
i
a
l
P
a
p
e
r
O
u
t
s
t
a
n
d
i
n
g
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
-
1
0
0
1
0
2
0
S
h
a
d
o
w
B
a
n
k
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Shadow Bank Asset Growth (Annual %)
Commercial Paper Outstanding Growth (Annual %)
Sources: Boardof Governorsof theFederal Reserve
Figure 3.5: Marginal Funding of Shadow Banks is Commercial Paper.
below the current market price on the understanding that it will buy it back in
the future at a pre-agreed price. The di?erence between the current market price
of the security and the price at which it is sold is called the “haircut” in the repo,
and ?uctuates together with market conditions. The ?uctuations in the haircut
largely determine the degree of funding available to a leveraged institution. The
reason is that the haircut determines the maximum permissible leverage achieved
by the borrower. If the haircut is 2%, the borrower can borrow 98 dollars for 100
dollars worth of securities pledged. Then, to hold 100 dollars worth of securities,
the borrower must come up with 2 dollars of equity. Thus, if the repo haircut is
2%, the maximum permissible leverage (ratio of assets to equity) is 50.
Suppose that the borrower leverages up the maximum permitted level. The
borrower thus has a highly leveraged balance sheet with leverage of 50. If at this
time, a shock to the ?nancial system raises the market haircut, then the borrower
faces a predicament. Suppose that the haircut rises to 4%. Then, the permitted
26
-
4
0
-
2
0
0
2
0
4
0
P
r
i
m
a
r
y
D
e
a
l
e
r
R
e
p
o
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
-
4
0
-
2
0
0
2
0
4
0
S
e
c
u
r
i
t
y
B
r
o
k
e
r
-
D
e
a
l
e
r
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
1992q1 1996q3 2001q1 2005q3 2010q1
Security Broker-Dealer Asset Growth (Annual %)
Primary Dealer Repo Growth (Annual %)
Sources: Boardof Governorsof theFederal Reserveand Federal ReserveBank of New York
Figure 3.6: Marginal Funding of Broker-Dealers is Repo.
leverage halves to 25, from 50. The borrower then faces a hard choice. Either
it must raise new equity so that its equity doubles from its previous level, or it
must sell half its assets, or some combination of both. However, asset disposals
have spillover e?ects that exacerbate the distress for others. The “margin spiral”
described by Brunnermeier and Pedersen (2009) models this type of phenomenon.
Considerations of repo haircuts suggest that measured risks will play a piv-
otal role in the determination of leverage. Adrian and Shin (2008b) present a
contracting model which yields this outcome as a central prediction, and present
empirical evidence consistent with the prediction. Adrian and Shin (2008b) ?nd
that measures of Value-at-Risk (VaR) that are computed from the time series of
daily equity returns explains shifts in total assets, leverage, and key components
of the liabilities side of the balance sheet, such as the stock of repos.
In the benchmark case where losses are exponentially distributed, the con-
tracting model of Adrian and Shin (2008b) yields the widely used Value-at-Risk
27
(VaR) rule which stipulates that exposures be adjusted continuously so that eq-
uity exactly matches total Value-at-Risk. Among other things, the Value-at-Risk
rule implies that exposures are adjusted continuously so that the probability of
default is kept constant - at the level given by the VaR threshold. Given the
ubiquitous use of the VaR rule both by private sector ?nancial institutions and
by regulators, this microfoundation of the VaR concept gives a basis for further
study.
To be sure, showing that the VaR rule is the outcome of a contracting model
says little about the desirability of the widespread adoption of such practices from
the point of view of economic e?ciency. Indeed, there are strong arguments to
suggest that risk management tools such as Value-at-Risk cause spillover e?ects to
other ?nancial institutions that are detrimental to overall e?ciency. For instance,
the prudent reduction in exposures by the creditors to Lehman Brothers is a run
from the point of view of Lehman Brothers itself. The spillover e?ects are a
natural consequence of the contracts being bilateral arrangements. They do not
take account of the spillover e?ects across more than one step in the ?nancial
network.
3.3. Relative Size of the Financial Sector
The rapid growth of the market-based intermediaries masks the double-counting
involved when adding up balance sheet quantities across individual institutions.
Before going further, we therefore note some accounting relationships that can
help us to think about the extent of the double-counting.
Let c
i
be the total assets of bank i, and r
i
the total debt of bank i (i.e., total
liabilities minus equity). The total size of the banking sector in gross terms can
be written as the sum of all bank assets, given by
P
a
i=1
c
i
. A closely related
measure would be the aggregate value of all bank debt, given by
P
a
i=1
r
i
.
28
De?ne leverage `
i
as the ratio of total assets to equity of bank i:
`
i
=
c
i
c
i
?r
i
. (3.1)
Then, solving for r
i
and using the notation o
i
= 1 ?
1
A
?
, we have
r
i
= o
i
Ã
¸
i
+
X
)
r
)
:
)i
!
= o
i
¸
i
+
£
r
1
· · · r
a
¤
?
?
?
o
i
:
1i
.
.
.
o
i
:
ai
?
?
?
(3.2)
Let r =
£
r
1
· · · r
a
¤
, ¸ =
£
¸
1
· · · ¸
a
¤
, and de?ne the diagonal matrix ?
as follows.
? =
?
?
?
o
1
.
.
.
o
a
?
?
?
(3.3)
Then we can write (3.2) in vector form as:
r = ¸?+ r??
Solving for r,
r = ¸?(1 ???)
?1
= ¸?
¡
1 +??+ (??)
2
+ (??)
3
+ · · ·
¢
(3.4)
The matrix ?? is given by
?? =
?
?
?
?
?
0 o
2
:
12
· · · o
a
:
1a
o
1
:
21
0 o
a
:
2a
.
.
.
.
.
.
.
.
.
o
1
:
a1
o
2
:
a2
· · · 0
?
?
?
?
?
(3.5)
The in?nite series in (3.4) converges since the rows of ??sum to a number strictly
less than 1, so that the inverse (1 ???)
?1
is well-de?ned.
29
Equation (3.4) suggests what to look for when gauging the extent of double-
counting of lending to ultimate borrowers that results from the heavy use of fund-
ing raised from other ?nancial intermediaries. The comparison is between ¸ (the
pro?le of lending to the ultimate borrowers in the economy) and r (the pro?le of
debt values across all banks, which gives a gross measure of balance sheet size).
The factor that relates the two is the matrix:
?
¡
1 +??+ (??)
2
+ (??)
3
+ · · ·
¢
This matrix has a ?nite norm, since the in?nite series 1+??+(??)
2
+(??)
3
+· · ·
converges to (1 ???)
?1
. However, for a ?nancial system where leverage is high,
and to the extent to which banks are interwoven tightly, the norm can grow
without bound. This is because as leverage becomes large, o
i
? 1 and, hence,
? tends to the identity matrix. Moreover, as the degree of interconnectedness
between banks becomes large, the norm of the matrix ? converges to 1, since
each row of ? will sum to a number that converges to 1. In the limit, as ? ?1
and k?k ?1, the norm of the matrix ?
¡
1 +??+ (??)
2
+ (??)
3
+ · · ·
¢
grows
without bound.
Consequently, the size of the ?nancial intermediation sector relative to the
size of the economy can vary hugely over the ?nancial cycle. We illustrate this
phenomenon in Figures 3.7 and 3.8, which show the growth of four sectors in the
United States from 1954. The four sectors are (i) the non-?nancial corporate
sector, (ii) the household sector, (iii) the commercial banking sector, and (iv) the
security broker-dealer sector. The data are taken from the Federal Reserve’s Flow
of Funds accounts. The series are normalized so that the size in 1954Q1 is set
equal to 1.
Over this time period, three of the four sectors had grown to roughly 80 times
their 1954 size by the end of 2009. These trends, however, have been dwarfed by
30
0
100
200
300
400
500
600
700
800
900
1
9
5
4
Q
1
1
9
5
6
Q
3
1
9
5
9
Q
1
1
9
6
1
Q
3
1
9
6
4
Q
1
1
9
6
6
Q
3
1
9
6
9
Q
1
1
9
7
1
Q
3
1
9
7
4
Q
1
1
9
7
6
Q
3
1
9
7
9
Q
1
1
9
8
1
Q
3
1
9
8
4
Q
1
1
9
8
6
Q
3
1
9
8
9
Q
1
1
9
9
1
Q
3
1
9
9
4
Q
1
1
9
9
6
Q
3
1
9
9
9
Q
1
2
0
0
1
Q
3
2
0
0
4
Q
1
2
0
0
6
Q
3
Non-financial
corporate
Households
Security Broker
Dealers
Commercial
Banks
Figure 3.7: Growth of Four US Sectors (1954Q1 =1)
1980Q1
1
10
100
1000
1
9
5
4
Q
1
1
9
5
7
Q
1
1
9
6
0
Q
1
1
9
6
3
Q
1
1
9
6
6
Q
1
1
9
6
9
Q
1
1
9
7
2
Q
1
1
9
7
5
Q
1
1
9
7
8
Q
1
1
9
8
1
Q
1
1
9
8
4
Q
1
1
9
8
7
Q
1
1
9
9
0
Q
1
1
9
9
3
Q
1
1
9
9
6
Q
1
1
9
9
9
Q
1
2
0
0
2
Q
1
2
0
0
5
Q
1
2
0
0
8
Q
1
Non-financial
corporate
Households
Security Broker
Dealers
Commercial
Banks
Figure 3.8: Growth of Four US Sectors (1954Q1 = 1) (in log scale)
31
that of the broker-dealer sector, which had grown to around 800 times its 1954
level at the height of the boom, before collapsing in the recent crisis. Figure 3.8
is the same chart, but in log scale. The greater detail a?orded by the chart in log
scale reveals that the securities sector kept pace with the rest of the economy up
to about 1980, but then started a growth spurt that outstripped the other sectors.
On the eve of the crisis, the size of the securities sector was roughly ten times
that of the other sectors in the economy.
4. Empirical Relevance of Financial Intermediary Balance
Sheets
Our discussion thus far suggests that ?nancial intermediaries deserve independent
study in models of monetary economics due to their impact on ?nancial condi-
tions. Asset prices are in?uenced by the tightness of balance sheet constraints of
?nancial intermediaries. In this section, we examine empirically whether ?nan-
cial intermediaries’ impact on ?nancial conditions can feed through to a?ect real
economic outcomes. The analysis follows Adrian and Shin (2008d) and Adrian,
Moench and Shin (2009).
We label the looseness of balance sheet constraints “risk appetite” following
Adrian, Moench, and Shin (2009, 2010) and Adrian, Etula, and Shin (2009). Risk
appetite refers to the shadow value of capital of the (leveraged) intermediary sector
in the model of section 2. This shadow value of capital indicates the additional
pro?t that the banking sector may earn by having one dollar of extra bank capital.
The looser is the capital constraint, the lower is the Lagrange multiplier, and hence
the higher is the risk appetite.
The terminology of “risk appetite” is intended to highlight the apparent change
in preferences of the banking sector. We say “apparent” change in preferences,
since the ?uctuations in risk appetite are due to the constraints faced by the
32
banks rather than their preferences as such. However, to an outside observer, the
?uctuations in risk appetite would have the outward signs of ?uctuations in risk
preferences of the investor.
Adrian, Moench, and Shin (2009) estimate the risk appetite of ?nancial inter-
mediaries for a ”macro risk premium”. The risk premium measures the hurdle
rate of return for new projects that are ?nanced in the economy, and hence re?ects
the ease of credit conditions, and correponds to the risk premium of the model
in section 2. The macro risk premium is estimated from the yield spreads of
?xed income securities. In particular, the macro risk premium is estimated as a
linear combination of spreads that is tracking GDP growth most closely. In doing
so, we allow both term spreads of the Treasury yield curve and credit spreads to
enter. Both term spreads and credit spreads are measures of hurdle rates — the
additional yields on longer-dated or riskier bonds that induce market investors to
fund additional investment or consumption.
To estimate the macro risk premium, Adrian, Moench, and Shin (2009) con-
temporaneously regress GDP growth on a wide variety of Treasury and credit
spreads. They then use the seven constant maturity yields published in the H.15
release of the Federal Reserve Board and compute spreads relative to the Fed
Funds target, and corporate bond spreads for credit ratings AAA, AA, A, BBB,
BB, and B from Standard & Poors in excess of the 10-year constant maturity
Treasury yield. The analysis of Adrian, Moench, and Shin (2009) starts in the
?rst quarter of 1985, and ends in the fourth quarter of 2009.
An estimate of the macro risk premium is obtained as the ?tted value of a linear
regression of GDP growth on these corporate and Treasury bond spreads. Hence,
empirically, the macro risk premium is a weighted average of spreads where the
weights are given by the regression coe?cients. The weights can be interpreted
33
-
1
0
-
5
0
5
1
0
Q
u
a
r
t
e
r
l
y
G
D
P
g
r
o
w
t
h
1
1
.
5
2
2
.
5
3
M
a
c
r
o
R
i
s
k
P
r
e
m
i
u
m
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
Macro Risk Premium
Quarterly GDP growth
Figure 4.1: The Macro Risk Premium and GDP Growth
as portfolio weights of a portfolio that is tracking GDP growth. Conceptually,
the macro risk premium represents the analogue of the risk premium term :
discussed in Section 2 above. The estimated macro risk premium, together with
GDP growth, are plotted in Figure 4.1. For ease of interpretation, the macro risk
premium is rotated using an a?ne transformation so as to match the average level
and the volatility of the AA credit spread. The plot shows the strong negative
correlation between the macro risk premium and GDP growth.
We now turn to our measure of the looseness of ?nancial intermediary capital
constraints, which we have called “risk appetite” as shorthand. As sketched in
Section 2 , the willingness of banks to lend will be positively associated with
the size of intermediary balance sheets. The scenario outlined in Section 2 is
that ?nancial intermediaries manage their balance sheets actively by employing a
Value-at-Risk constraint when choosing the size and composition of their portfolio.
Intuitively, the relationship between the macro risk premium, ?nancial inter-
mediary balance sheets, and real activity goes as follows. When ?nancial interme-
34
diaries have ample balance sheet capacity in terms of higher capital, their balance
sheet constraints are loose, risk premia are compressed, the supply of credit is
plentiful, which in turn leads to lower threshold rates of return for real projects,
and hence higher GDP growth. E?ective risk aversion is low, and real growth is
high. Conversely, when ?nancial intermediary funding conditions worsen, their
risk appetite declines, leading to lower real growth.
The macro risk premium measure can be interpreted as a portfolio of Treasury
and corporate bond yields where the portfolio weights are chosen so as to maximize
the contemporaneous correlation with real GDP growth. In a similar vein, we
obtain a measure of intermediary risk appetite by ?nding the linear combination
of one-year lagged balance sheet variables that best predict the (negative) one-
year change of the macro risk premium. To a good approximation, the negative
change re?ects the return on the macro risk premium. A priori, it is not clear
which institutions are the most important ones in determining risk premia for
the di?erent asset classes. We therefore build on our related work in Adrian,
Moench, and Shin (2010) in which we use subset selection methods to identify
the best predictors for excess returns on di?erent asset classes among a large
number of potential explanatory balance sheet proxies for various types of ?nancial
institutions. Adrian, Moench, and Shin (2010) document that annual leverage
growth of security brokers and dealers is a strong predictor of excess returns on
equity and corporate bond portfolios and that quarterly shadow bank asset growth
is a strong forecasting variable for excess returns on corporate and Treasury bonds.
Adrian, Moench, and Shin (2009) therefore restrict the set of right-hand side
variables to these two types of institutions, complemented by commercial banks
so as to highlight the di?erential impact that balance sheets of market based
intermediaries have with respect to traditional banks. For each of these three
types of institutions, one-year lagged asset growth and the growth of net worth
35
are included as potential variables. Adrian, Moench, and Shin (2009) weight the
growth rates of assets and net worth by the relative size of total assets of each
intermediary in order to capture the trends of assets under management across
di?erent institutions.
Risk appetite is estimated by regressing the (negative) change of the macro
risk premium over one year on the balance sheet measures of the security broker-
dealers, the shadow banks, and the commercial banks. These negative changes
capture returns to the risk premia. When the negative change of a risk premium
is positive, prices today increase, leading to lower expected returns. The measure
of risk appetite is constructed as the ?tted value of the regressions of (the negative
changes to the macro risk premium) on the intermediary balance sheet variables.
The risk appetite measure is displayed together with the macro risk premium in
Figure 4.2. The plot shows that risk appetite is highly negatively correlated with
changes to the macro risk premium. Higher risk appetite leads to balance sheet
expansions, which are associated with increases in asset prices and hence declines
in spreads. Movements in risk appetite are thus strongly negatively correlated
with the macro risk premium.
We can also investigate the importance of ?nancial intermediary balance sheets
for macroeconomic activity by relating intermediary balance sheets directly to
GDP growth. Relative to commercial banks, broker-dealer and shadow bank
balance sheets potentially hold more information related to underlying ?nancial
conditions, as they are a signal of the marginal availability of credit. At the
margin, all ?nancial intermediaries (including commercial banks) have to borrow
in capital markets (for instance via commercial paper or repos). For commercial
banks, however, their large balance sheets mask the e?ects operating at the mar-
gin. Broker-dealers or shadow banks, in contrast, give a purer signal of marginal
36
-
.
8
-
.
6
-
.
4
-
.
2
0
.
2
I
n
t
e
r
m
e
d
i
a
r
y
R
i
s
k
A
p
p
e
t
i
t
e
1
1
.
5
2
2
.
5
3
M
a
c
r
o
R
i
s
k
P
r
e
m
i
u
m
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
Macro Risk Premium
Intermediary Risk Appetite
Figure 4.2: Macro Risk Premium and Intermediary Risk Appetite
funding conditions, as (i) their liabilities are short term and (ii) their balance
sheets are closer to being fully marked to market.
In addition, broker-dealers originate and serve as market makers for securitized
products, whose availability determines the credit supply for consumers and non-
?nancial ?rms (e.g., for mortgages, car loans, student loans, etc.). Consequently,
broker-dealers are important variables not only because they are the marginal
suppliers of credit, but also because their balance sheets re?ect the ?nancing
constraints of the market-based ?nancial system.
To the extent that balance sheet dynamics a?ect the supply of credit, they
have the potential to a?ect real economic variables. To examine this empirically,
we estimate macroeconomic forecasting regressions. In Table 4.1, we report the
results of regressions of the quarterly growth rate of GDP components on lagged
macroeconomic and ?nancial variables. In addition, we add the lagged growth
rate of total assets and market equity of security broker-dealers. By adding lags
of additional ?nancial variables (equity market return, equity market volatility,
37
term spread, credit spread), we o?set balance sheet movements that are purely
due to a price e?ect. By adding the lagged macroeconomic variables, we control
for balance sheet movements due to past macroeconomic conditions. In Table
4.1, (and all subsequent tables), * denotes statistical signi?cance at the 10%, **
signi?cance at the 5% level, and *** at the 1% level. All our empirical analysis
is done using quarterly data from 1986Q1 to 2009Q2. Variable de?nitions are
given in the data appendix at the end of this chapter.
-.2
0
.2
.4
0 5 10 15
ImpulseResponse Function of GDP Growth to Shadow Bank Asset Growth Shock
Quarters
Figure 4.3: The ?gure plots the impulse response GDP growth from a shock to
shadow bank total asset growth. The impulse response is estimated from a vector
autoregression with gdp growth, pce in?ation, shadow bank asset growth, credit
spread, vix, the term spread, and the Federal Fund target rate as variables, and
that ordering is used to produce the impulse response functions using a Cholesky
decomposition. The time span is 1986Q1 to 2009Q1.
The growth rate of security broker-dealer total assets has strongest signi?cance
for the growth rate of future housing investment, and weak signi?cance for total
GDP growth (Tables 4.1 and 4.2, columns 1). Our interpretation of this ?nding is
that the mechanisms that determine the liquidity and leverage of broker-dealers
38
Table 4.1: Impact of Balance Sheets on GDP. This table reports regressions
of quarterly GDP growth on the total asset and equity growth of broker-dealers,
shadow banks, and commercial banks for 1986Q1 to 2009Q2. *** denotes sig-
ni?cance at the 1% level, ** denotes signi?cance at the 5% level, and * denotes
signi?cance at the 10% level. Signi?cance is computed from robust standard er-
rors.
(1) (2) (3)
Quarterly Quarterly Quarterly
GDP GDP GDP
Growth Growth Growth
Broker-Dealer Asset Growth (lag) 0.03*
Broker-Dealer Equity Growth (lag) 0.18
Shadow Banks Asset Growth (lag) 0.21***
Shadow Banks Equity Growth (lag) 0.71**
Commercial Bank Asset Growth (lag) 0.02
Commercial Bank Equity Growth (lag) -0.12
GDP Growth (lag) 0.03 -0.18 0.09
PCE In?ation (lag) -1.01** -1.00** -1.16***
VIX (lag) 0.01 -0.03 -0.02
Credit Spread (lag) -1.37* -1.81** -1.01
Term spread (lag) 0.75** 1.18*** 0.75*
Fed Funds (lag) 0.40 0.19 0.49*
Constant 4.67*** 4.94*** 4.44**
Observations 93 93 93
1
2
0.288 0.409 0.263
39
a?ect the supply of credit, which, in turn, a?ect investment and consumption.
The total assets and total equity of shadow banks have signi?cant forecasting
power for total GDP growth, re?ecting their increased role in the total supply
of credit for the US economy. Commercial banks, on the other hand, have no
forecasting power for GDP, and forecast housing growth with the wrong sign.
Adrian, Moench, and Shin (2010) systematically investigate the forecasting power
of all ?nancial intermediaries in the U.S. Flow of Funds, and con?rm that broker-
dealers and shadow banks forecast real activity.
-.1
0
.1
.2
.3
0 5 10 15
ImpulseResponseFunction of Residential Investment Growthto Dealer Asset Growth
Quarters
Figure 4.4: The ?gure plots the impulse response of residential investment growth
from a shock to broker-dealer total asset growth. The impulse response is esti-
mated from a vector autoregression with residential investment growth, pce in-
?ation, broker-dealer asset growth, credit spread, vix, the term spread, and the
Federal Fund target rate as variables, and that ordering is used to produce the
impulse response functions using a Cholesky decomposition. The time span is
1986Q1 to 2009Q1.
The forecasting power of dealer assets for housing investment is graphically
illustrated in Figure 4.4. The impulse response function is computed from a ?rst
40
Table 4.2: Impact of Balance Sheets on Housing Investment. This table
reports regressions of quarterly residential investment growth on the total asset
and equity growth of broker-dealers, shadow banks, and commercial banks for
1986Q1 to 2009Q2. *** denotes signi?cance at the 1% level, ** denotes signi?-
cance at the 5% level, and * denotes signi?cance at the 10% level. Signi?cance is
computed from robust standard errors.
(1) (2) (3)
Quarterly Quarterly Quarterly
Housing Housing Housing
Growth Growth Growth
Broker-Dealer Asset Growth (lag) 0.09***
Broker-Dealer Equity Growth (lag) 0.10
Shadow Banks Asset Growth (lag) -0.00
Shadow Banks Equity Growth (lag) 0.14
Commercial Bank Asset Growth (lag) -0.44*
Commercial Bank Equity Growth (lag) 0.23
Housing Growth (lag) 0.89*** 0.94*** 0.96***
PCE In?ation (lag) -0.30 -0.11 -0.09
VIX (lag) 0.11 0.01 0.03
Credit Spread (lag) -0.92 -0.49 0.01
Term spread (lag) 1.11** 0.60 -0.07
Fed Funds (lag) -0.06 -0.04 -0.27
Constant -2.53 0.13 3.76
Observations 93 93 93
1
2
0.902 0.881 0.888
41
order vector autoregression that includes all variables of Table 4.1, Column (1).
The plot shows that the response of housing investment to a positive shock in
broker-dealer assets growth is positive, large, and persistent.
The di?erences between the interactions of market based intermediaries and
commercial banks with the macroeconomic aggregates are further highlighted in
column (iii) of Table 4.1, where we see that commercial bank assets and equity
do less well than shadow bank or security-broker-dealer variables as forecasting
variables. Our interpretation of these ?ndings is that commercial bank balance
sheets are less informative than broker-dealer balance sheets as they (largely) did
not mark their balance sheets to market over the time span in our regressions.
In addition, in Table 4.2, we ?nd that growth in commercial bank total assets
precedes declines in housing investment. This result is primarily due to the
fact that commercial banks o?er credit line pre-commitments that tend to be
drawn in times of crisis. In fact, in Figure 3.4, we saw that commercial bank
total assets grew at the onset of the recent ?nancial crisis, as structured credit
was reintermediated onto commercial bank balance sheets. The ?nding that
commercial bank assets do not predict future real growth is also consistent with
Bernanke and Lown (1991), who use a cross-sectional approach to show that
credit losses in the late 80’s and early 90’s did not have a signi?cant impact on
real economic growth across states. See Kashyap and Stein (1994) for an overview
of the debate on whether there was a “credit crunch” in the recession in the early
1990s.
In the same vein, Ashcraft (2006) ?nds small e?ects of variations in commercial
bank loans on real activity when using accounting based loan data. However,
Ashcraft (2005) ?nds large and persistent e?ects of commercial bank closures on
real output (using FDIC induced failures as instruments). Morgan and Lown
(2006) show that the senior loan o?cer survey provides signi?cant explanatory
42
power for real activity — again, a variable that (i) is more likely to re?ect underlying
credit supply conditions and (ii) is not based on accounting data.
The credit supply channel sketched so far di?ers from the ?nancial ampli?ca-
tion mechanisms of Bernanke and Gertler (1989), and Kiyotaki and Moore (1997,
2005). These papers focus on ampli?cation due to ?nancing frictions in the bor-
rowing sector, while we focus on ampli?cation due to ?nancing frictions in the
lending sector. We return to a more thorough review of the literature in a later
section.
5. Central Bank as Lender of Last Resort
The classical role of the central bank as the lender of last resort (LOLR) is framed
in terms of meeting panics that a?ect solvent, but illiquid, banks. In the simplest
case, bank runs arise when depositors fail to achieve coordination in a situation
with multiple equilibria. For example, in Bryant (1980) and Diamond and Dybvig
(1983), an individual depositor runs for fear that others will run, leaving no assets
in place for those who do not run.
However, in the ?nancial crisis of 2007-2009, the withdrawal of credit was not
restricted to one of even a subset of institutions. Instead, entire market sectors
were targeted. Figure 5.1 plots the new issuance of asset backed securities (ABS)
over a three month interval preceding the measurement date, and clearly illustrates
the generalized contraction of credit. If there was a run driven by a coordination
failure, it was a simultaneous run from all institutions in the ?nancial system.
Albeit, the extent to which each institution su?ered from the run depended on its
particular vulnerability. In the model outlined in Section 2, it is the interaction
between measured risks and the risk-bearing capacity of banks that determines
overall lending. Financial institutions that rely on value at risk cut back lending
when risk constraints become more binding (i.e., when the Lagrange multiplier
43
0
50
100
150
200
250
300
350
M
a
r
-
0
0
S
e
p
-
0
0
M
a
r
-
0
1
S
e
p
-
0
1
M
a
r
-
0
2
S
e
p
-
0
2
M
a
r
-
0
3
S
e
p
-
0
3
M
a
r
-
0
4
S
e
p
-
0
4
M
a
r
-
0
5
S
e
p
-
0
5
M
a
r
-
0
6
S
e
p
-
0
6
M
a
r
-
0
7
S
e
p
-
0
7
M
a
r
-
0
8
S
e
p
-
0
8
$
B
i
l
l
i
o
n
s
Other
Non-U.S. Residential
Mortgages
Student Loans
Credit Cards
Autos
Commercial Real
Estate
Home Equity
(Subprime)
Figure 5.1: New Issuance of ABS in Previous Three Months
associated with the constraint increases). From the point of view of an individual
bank, this prudent cutting of exposures by creditors will closely resemble a ”run.”
In this sense, the runs on Northern Rock, Bear Stearns, and Lehman Brothers
may be better seen as the tightening of constraints on the creditors of these banks,
rather than as a coordination failure among them.
Of course, we should not draw too ?ne a distinction between the coordination
view of bank runs on the one hand, and the “leverage constraints” view on the
other. Coordination (or lack thereof) will clearly exacerbate the severity of any
run when a bank has many creditors. The point is, rather, that an explanation
of a run on the system needs to appeal to more than just coordination failures.
For example, this means that explanations of the runs on Bear Stearns or Lehman
Brothers, should make reference to market-wide factors, as well as to the particular
characteristics of those ?rms and their creditors. This is one more instance of
the general maxim that, in a modern market-based ?nancial system, banking and
capital market conditions cannot be viewed in isolation.
44
Mar-09
Sep-08
Oct-08
Nov-08
0%
2%
4%
6%
8%
10%
12%
D
e
c
-
8
1
D
e
c
-
8
3
D
e
c
-
8
5
D
e
c
-
8
7
D
e
c
-
8
9
D
e
c
-
9
1
D
e
c
-
9
3
D
e
c
-
9
5
D
e
c
-
9
7
D
e
c
-
9
9
D
e
c
-
0
1
D
e
c
-
0
3
D
e
c
-
0
5
D
e
c
-
0
7
Figure 5.2: Cash as Proportion of US Commercial Bank Assets (Source: Federal
Reserve, H8 database).
To the extent that the credit crunch can be seen as the consequence of a
collapse of balance sheet capacity in the ?nancial intermediary sector, we can
interpret the policy response by central banks as an attempt to restore this lost
capacity by lending directly into the market. The Federal Reserve has been one
of the most aggressive central banks in this context, e?ectively interposing its own
balance sheet between the banking sector and the ultimate borrowers. The Fed
has taken in deposits from the banking sector (through increased reserves) and
then lent out the proceeds to ultimate borrowers through the holding of securities
(Treasuries, mortgage backed securities, commercial paper and other private sector
liabilities), as well as through currency swap lines to foreign central banks. One
indication of the dramatic increases in the Fed’s balance sheet can be seen in the
sharp increase in the holding of cash by US commercial banks, as seen from Figure
5.2. The increased cash holdings re?ect the sharp increase in reserves held at the
Fed — a liability of the Fed to the commercial banks.
In this way, central bank liquidity facilities have countered the shrinking of in-
45
termediary balance sheets and have become a key plank of policy, especially after
short-term interest rates were pushed close to their zero bound. The management
of the increased Federal Reserve balance sheet has been facilitated by the intro-
duction of interest on reserves on October 1, 2008, which e?ectively separates the
management of balance sheet size from that of the Federal Funds interest rate (see
Keister and McAndrews (2009) for a discussion of the interest on reserve payment
on the Federal Reserve’s balance sheet management).
The Federal Reserve has also put in place various other lender of last resort
programs in order to cushion the strains on balance sheets, and to thereby target
the unusually wide spreads in a variety of credit markets. Liquidity facilities
have been aimed at the repo market (TSLF and PDCF), the commercial paper
market (CPFF and AMLF), and ABS markets (TALF). In addition, the Federal
Reserve has conducted outright purchases of Treasury and agency securities, and
has provided dollar liquidity in the FX futures markets (FX Swap lines). The
common motivating element in these policies has been to try and alleviate the
strains associated with the shrinking balance sheets of intermediaries by substi-
tuting the central bank’s own balance sheet. The spirit of these policies di?ers
from that of classic monetary policy in that they are explicitly aimed at replacing
the collapse of private sector balance sheet capacity. Since the deleveraging of
?nancial intermediary balance sheets is associated with a widening of risk premia,
the e?ectiveness of balance sheet policies can be judged by the level of risk premia
in various ?nancial markets. In practice, the degree to which risk premia are
associated with the expansions and contractions of intermediary balance sheets
are important indicators for the risk appetite of the ?nancial sector, which, in
turn, a?ect credit supply and real activity. Adrian, Moench and Shin (2010)
use this insight to decompose the risk premia of several asset classes into var-
ious components, including those associated with the risk appetite of ?nancial
46
intermediaries.
One instance of the Fed’s liquidity facilities can be seen in Figure 5.3 (taken
from Adrian, Marchioni, and Kimbrough (2010), which charts total outstanding
commercial paper alongside net Federal Reserve commercial paper holdings in the
”Commercial Paper Funding Facility” (CPFF).
6
Following the Lehman Brothers
bankruptcy in September 2008, the outstanding amount of commercial paper be-
gan to fall precipitously, as can be seen by the sharp downward shift in the shaded
blue area. With the creation of the CPFF in October 2008, the Fed’s holdings of
commercial paper in the CPFF began to increase rapidly, as shown by the green
area in Figure 5.3. The Fed’s holdings can be seen to replace virtually dollar-
for-dollar the decline in the outstanding amount of commercial paper. In this
respect, the Fed’s balance sheet was being used to directly replace the decline in
balance sheet capacity of the ?nancial intermediary sector. An important feature
of the CPFF is that, as the facility was intended to be only a temporary liquidity
backstop, it was designed to become more unattractive as market conditions begin
to normalize. Accordingly, while the red line in Figure 5.3 shows that the Federal
Reserve held as much as 20% of all outstanding commercial paper at the height
of the crisis, CPFF holdings have since fallen steadily as market functioning has
improved.
As another example, Figure 5.1 illustrated how new issuance of asset-backed
securities (ABS) had collapsed by the end of 2008. In response, the Fed instituted
the Term Asset-Backed Loan Facility (TALF), whereby the central bank provides
secured loans to new AAA-rated ABS at a low haircut to private sector investors.
TALF was designed speci?cally to revitalize the ABS market, and its e?ectiveness
can be gauged by Figure 5.4, which shows the e?ect on new issuance of ABS
before and after the introduction of TALF. The light colored bars on the right
6
See Adrian, Marchioni, and Kimbrough (2010) for a detailed description of the CPFF.
47
0%
10%
20%
30%
40%
50%
0
400
800
1200
1600
2000
Aug-08 Dec-08 Apr-09 Aug-09
Billions of Dollars Percent
Source: Federal Reserve Board of Governors
Total Outstanding
in Market
Total Commercial Paper Outstanding
Total Outstanding
in CPFF
CPFF as %
of Market
CPFF Launch
Figure 5.3: The Federal Reserve’s Commercial Paper Funding Facility
0
5
10
15
20
25
30
J -08 F-08 M-08 A-08 M-08 J -08 J -08 A-08 S-08 O-08 N-08 D-08 J -09 F-09 M-09 A-09 M-09
$Bln
TALF-Financed Not TALF-Financed
ABS Include autos, credit cards, student loans, SBA, equipment, floorplans, and fleet leases
Figure 5.4: New ABS Issuance under TALF
48
show that much of the recent issuance of ABS has been TALF-?nanced, and that
TALF-backed issuance dwarfs the issuance of standard issues.
The balance sheet expansion of the Federal Reserve in response to the ?nan-
cial crisis of 2007-2009 has refocused the monetary policy debate on the role of
quantities in the monetary policy transmission mechanism. The crisis force-
fully demonstrated that the collapse of balance sheet capacity in the ?nancial
sector can have powerful adverse a?ects on the real economy, and, accordingly,
the traditional role of the central bank as the lender of last resort has undergone
far-reaching innovations.
6. Role of Short-Term Interest Rates
Having established that increases in broker-dealer and shadow bank balance sheets
signal increases in real activity, we now investigate the determinants of balance
sheet growth itself. Broker-dealers, shadow banks, and commercial banks fund
themselves with short-term debt. Broker-dealers are primarily funded in the repo
market (see Figure 3.6); shadow banks are primarily funded in the commercial
paper market (see Figure 3.5); and the majority of commercial banks’ short-term
funding is through money (i.e., checking and savings deposits). In the case of
broker-dealers, part of the repo funding is directly passed on to other leveraged
institutions, such as hedge funds, in the form of reverse repos, while another part
is invested in longer-term, less liquid securities. Shadow banks, on the other hand,
tend to fund holdings of ABS and MBS directly. Commercial banks primarily
hold non-tradable loans.
Because the majority of the liability side of ?nancial institutions comes from
short-term borrowing arrangements, their cost of borrowing is tightly linked to
short-term interest rates, such as the Federal Funds target rate. As broker-dealers
and shadow banks hold longer term assets, proxies for their expected returns are
49
spreads–particularly term spreads–which capture the maturity transformation
of ?nancial institutions. The leverage of these intermediaries is constrained by
risk: in more volatile markets, leverage is more risky, margins and haircuts are
higher, and credit supply tends to be more constrained. We saw in Section 2 how
Value-at-Risk determined balance sheet size, risk premia, and credit supply.
Much of these balance sheet adjustments occur at high frequencies. Though
the total assets used in the previous regressions are available only at a quarterly
frequency, on the liability side of the balance sheet, there are weekly data available
on outstanding repo, outstanding commercial paper, and total money. We use
repo data that are collected for the primary dealer universe by the Federal Reserve
Bank of New York. Outstanding commercial paper is collected by the depository
trust corporation (DTC), and is published at a weekly frequency by the Federal
Reserve Board. The broad money measure M2 is also made available by the
Federal Reserve Board.
We ?nd that increases in the Fed Funds target rate are generally associated
with a slower growth rate of short-term liabilities. In Table 6.1, we show regres-
sions of growth rates of repo, repo + commercial paper, and M2 on changes of the
Fed Funds target as well as on other asset prices (and lags of the left hand side
variables). The three types of regressions correspond to the funding of the three
main ?nancial institutions: broker-dealers, shadow banks, and commercial banks.
In each case, increases in the Fed Funds target are associated with declines in the
short term funding liabilities.
Financial market volatility, as measured by the VIX index of implied equity
volatility, relates negatively to security repo growth and repo+cp growth. As
higher volatility is associated with higher haircuts and tighter capital constraints,
both induce tighter constraints on dealer leverage (columns (1) and (2)). For
M2, we ?nd that higher VIX is associated with larger money growth, which we
50
Table 6.1: Determinants of Balance Sheet Growth. This table reports re-
gressions of repo growth, repo + commercial paper growth, and M2 growth on
their own lags, and asset price variables. The data frequency is weekly from Oc-
tober 3, 1990 to February 3, 2010. Changes refer to 1-week changes, and lags to 1
week lags. *** denotes signi?cance at the 1% level, ** denotes signi?cance at the
5% level, and * denotes signi?cance at the 10% level. Signi?cance is computed
from robust standard errors.
(1) (2) (3)
Repo Repo+CP Growth M2 Growth
(weekly growth) (weekly growth) (weekly growth)
Fed Funds (1 week change) -0.630*** -0.355*** -0.054***
Equity Return (1 week) -0.022* -0.013* 0.001**
VIX (1 week change) -0.052 -0.027 0.001
Treasury spread (1 week change) 0.703 0.291 0.151**
Credit spread (1 week change) 0.311 0.031 0.337**
Repo Growth (1 week lag) -0.134*** -0.075*** -0.001
CP Growth (1 week lag) 0.022 0.028 -0.020
M2 Growth (1 week lag) 0.515 0.063 -0.016
Constant 0.136* 0.105** 0.050***
Observations 990 990 989
1
2
0.042 0.032 0.121
interpret as ?ight to quality: in times of crisis, households and non-?nancial cor-
porations tend to reallocate short-term savings to commercial banks (see Gatev,
Schuermann, and Strahan, 2009).
Increases in the term spread are associated with higher repo growth. This
?nding is consistent with the notion that ?nancial intermediaries fund themselves
with short-term debt, but lend out longer term, so that a higher term spread
increases the carry between assets and liabilities and is associated with larger
balance sheets.
51
6.1. The Risk-Taking Channel of Monetary Policy
Current models in monetary economics emphasize the importance of managing
market expectations. By charting a path for future short rates and communicat-
ing this path clearly to the market, the central bank can in?uence long rates and
thereby in?uence mortgage rates, corporate lending rates and other prices that
a?ect consumption and investment. In constrast, our ?ndings point to the short-
term interest rate as an important price variable in its own right. Empirically,
we have seen that the Fed Funds rate is an important explanatory variable for
the growth of balance sheet aggregates. Our model suggests that increasing bank
capital increases the risk-taking capacity of the banking system, which in turn
leads to a lower equilibrium risk premium, and an increase the supply of credit
by lowering the hurdle rate at which projects are ?nanced.
Banks and other ?nancial intermediaries borrow in order to lend. Since the
loans o?ered by banks tend to be of longer maturity than the liabilities that fund
those loans, the term spread is indicative of the marginal pro?tability of an extra
dollar of loans on intermediaries’ balance sheets. The net interest margin (NIM)
of the bank is the di?erence between the total interest income on the asset side
of its balance sheet and the interest expense on the liabilities side of its balance
sheet. Whereas the term spread indicates the pro?tability of the marginal loan
that is added to the balance sheet, the net interest margin is an average concept
that applies to the stock of all loans and liabilities on the balance sheet.
The net interest margin determines the pro?tability of bank lending and in-
creases the present value of bank income, thereby boosting the forward-looking
measures of bank capital. Such a boost in bank capital increases the capacity
of the bank to increase lending in the sense that the marginal loan that was not
made before the boost in bank capital now becomes feasible under the greater
risk-bearing capacity of the bank. As banks expand their balance sheets, the
52
Initial
balance sheet
Fall in Fed Funds
Final
balance sheet
debt
equity
assets
increase
in equity
equity
assets
debt
assets
increasein
valueof
assets
equity
debt
new
borrowing
new
lending
Figure 6.1: The impact of a decline in the Fed Funds rate on intermediary balance
sheets.
market price of risk falls.
The logic of the argument is illustrated in Figure 6.1 for the case of monetary
easing. A decline in the Fed Funds target leads to an increase in asset prices and,
thus, an increase in the net worth of ?nancial institutions. In response, levered
?nancial institutions expand their assets by taking on additional leverage. Thus,
the impact of changes in short term interest rates is ampli?ed via the balance
sheet management of ?nancial institutions.
In order for the argument to go through a key assumption is that the term
spread is determined in large part by the short-term interest rate. On this score,
the evidence is supportive. Figure (1.1) in the introductory section to this chap-
ter highlights the striking fact that there is a near perfect negative one-to-one
relationship between 4-quarter changes of the Fed Funds target and 4-quarter
changes of the term spread (the plot uses data from 1987q1 to 2008q3). In this
way, variations in the target rate has a one-for-one relationship with the slope
of the yield curve. Since the term spread leads the net interest margin (NIM),
shifts in the short rate a?ect real activity because they change the pro?tability of
53
?nancial intermediaries, thus shifting the supply of credit.
The connection between ?nancial intermediary balance sheet management,
the slope of the yield curve, and real economic activity in the United States was
recently examined by Adrian, Estrella and Shin (2010), who conducted a vector
autoregression study of the relationship between the following variables: quarterly
GDP growth as a measure of real activity, the 10-year/3-month term Treasury
spread, the net interest margin (NIM) of large commercial banks from their Y-9C
?lings, the quarterly asset growth of shadow banks, the 3-month Treasury yield
as a measure of the short term interest rate, and the quarterly change in the
Chicago Board Options Exchange Volatility Index (VIX) as a measure of risk.
The balance sheet aggregate of the intermediary sector is taken to be the total
assets of shadow banks de?ned as the sum of total assets of asset-backed securities
issuers (ABS), ?nance companies, and funding corporations (each component is
pulled from the Federal Reserve’s Flow of Funds). The VAR includes one lag of
each of the variables, as suggested by the Bayesian Information Criterion, and
is estimated over the period from 1990Q3 to 2008Q3, where the starting date is
determined by availability of the VIX data.
Adrian, Estrella and Shin (2010) present empirical results that are consistent
with the following logic. An increase in the term spread tends to increase net
interest margin. This is fairly mechanical as the term spread directly impacts net
interest margin for newly originated loans funded with shorter-term liabilities.
Higher net interest margin – a major source of pro?ts for ?nancial intermedi-
aries – leads to an increase in total assets of ?nancial intermediaries: as lending
becomes more pro?table, the supply of credit is expanded and intermediaries’ bal-
ance sheets grow. Larger asset growth of intermediaries, in turn, predicts higher
GDP growth, which we interpret as a shift in the supply of credit curve. Since the
VAR includes only one lag of each variable, the signi?cance levels of the coe?-
54
cients may also be interpreted as a set of Granger causality tests. These tests are
consistent with our hypothesis of a causal chain that runs from the term spread
to net interest margin to lending volume and ?nally to real growth.
Adrian, Estrella and Shin (2010) also conduct impulse response studies so as
to verify the main strands in the narrative. In the impulse response studies, a
positive shock to the term spread leads to statistically signi?cant increases in net
interest margin over a considerable horizon. The shape of these responses is also
consistent with the fact that average net interest margin tends to trail marginal
changes in the term spread, as argued before. Moreover, a positive shock to net
interest margin tends to increased lending by the shadow banking sector. Finally,
a shock to asset growth in shadow banking has a quick and signi?cant e?ect on
real economic growth.
The evidence is supportive of the “risk-taking channel” of monetary policy.
Variations in short term interest rates lead real economic outcomes through their
impact on the slope of the yield curve. Our interpretation of this evidence is an
economic mechanism that operates via the balance sheet management of ?nan-
cial intermediaries, who borrow short and lend long. Tighter policy leads to a
compression of net interest margin and causes intermediaries to reduce lending.
The ?atter the term spread at the end of the tightening cycle, the greater the
subsequent reduction in lending activity. This has a direct e?ect on the supply of
credit to the real economy.
6.2. Two Case Studies
Two recent empirical papers throw further light on the channel of monetary policy
that works through changes in the market value of existing loans. Jimenez,
Ongena, Peydro and Saurina (2008) examine a large database of European loans
through the detailed information contained in the loan register and show that a
55
lower short-term interest rate lowers the hazard rate of default on existing loans.
In addition, they show that the hazard rate of default for new loans increases after
the cut in short-term rates.
The fact that the riskiness of existing loans decline may be due to a fall in
the interest burden of the borrower. The increased credit quality of the assets
will give rise to an increase in the pro?tability of the lending, and lead to greater
lending capacity, as outlined already. However, it is the second ?nding which is
more telling. The fact that the riskiness of new lending increases suggests that
the new loans are of lower quality, suggesting that the hurdle rate for lending
has fallen. Such a combination of (i) greater lending capacity and (ii) erosion of
lending standards is consistent with the risk-taking channel of monetary policy.
The same combination of (i) a lowering of a hazard rate of default on existing
loans and (ii) an increase in the hazard rate of default on new loans is also observed
in Ioannidou, Ongena and Peydro (2009). In this study, the authors examine the
e?ect of shifts of the US Fed Funds rate on the quality of bank loans in Bolivia,
which had a banking system which was close to being dollarized. To the extent
that the US Fed Funds rate is determined independently of the events in Bolivia,
the authors regard the e?ect of short-term interest rate changes as being a quasi-
natural experiment of the e?ect of short-term interest rate movements on bank
asset quality. As with the paper by Jimenez et al. (2008), the Bolivian study
reveals the same combination whereby a cut in the US Fed Funds rate leads to an
improvement in the quality of existing assets, but new assets are of a lower quality.
Paravisini (2008) provides estimates of the impact of bank funding constraints on
the supply of bank credit using an instrumental variable approach, and Khwaja
and Mian (2008) provide estimates of a bank funding shocks for an emerging mar-
ket. Freixas (2009) provides an overview of the monetary transmission literature
in the context of the 2007-2009 global ?nancial crisis.
56
This combination of results on existing and new loans suggest that the risk-
taking channel is a potentially fruitful avenue for further study. The model in
section 2 provides some of the conceptual background that may be necessary to
understand the results.
6.3. Related Literature
In order to highlight what we view as the speci?c contribution of the “risk-taking
channel” of monetary policy, it is important to give an account of the points of
contact between our approach and the existing literature in monetary economics
and corporate ?nance. The discussion can be organized along a number of di-
mensions, but one classi?cation that is useful is to distinguish those papers that
have emphasized the borrower’s balance sheet (and the demand for credit) from
those emphasizing the lender’s balance sheet (and the supply of credit).
Bernanke and Gertler (1989) is a classic example of the former - an explanation
based on the borrower’s balance sheet. Following the earlier work of Bernanke
(1983) who argued for the importance of borrower balance sheet distress during
the Great Depression, the Bernanke and Gertler model focuses on the agency prob-
lems entailed by the asymmetry of information between a non-?nancial corporate
borrower and the ?nancial market as a whole.
In the presence of asymmetric information between the borrower and lender,
ine?ciencies in the optimal contract manifest themselves in the form of deadweight
costs and, in particular, in the spread between the cost of internal funds and that
of external funds. The size of the deadweight cost is a function of the net worth
of the borrower - i.e., the borrower’s “skin in the game”. Moreover, Bernanke
and Gertler argue that the borrower’s net worth is procyclical, and so the funding
spread between the internal and external funds should be countercyclical.
An alternative approach that emphasizes the borrower’s balance sheet is the
57
work on credit cycles by Kiyotaki and Moore (1997, 2005)), who examine the dy-
namic ampli?cation of credit constraints. In their approach, Kiyotaki and Moore
assume a collateral constraint where the size of the loan that can be obtained
by a borrower depends on the current market price of the collateral that can be
pledged to the lender. Under such an assumption, the size of the borrower’s
balance sheet can depend positively on the market price of the asset - in other
words, the demand reaction of the borrower can be upward-sloping. When the
price of an asset increases, the borrower’s funding ability increases, thereby gen-
erating larger balance sheets. When the greater demand for the asset pushes up
the price of the asset, there is the possibility of ampli?ed responses where asset
price increases fuels further investment and aggregate activity, which raises prices
further.
The common thread between the work of Bernanke and Gertler (1989) and
Kiyotaki and Moore (1997, 2005) is that the focus is on the borrower’s balance
sheet, and the ?uctuations in the creditworthiness of the borrower. The supply
of lending is determined in the market as a whole, without a separate role for the
banking sector as such.
However, to the extent that the borrower in the Bernanke and Gertler (1989)
model can be re-interpreted as a bank, the model can be reoriented in terms of the
agency problems in the banking sector. The “double-decker” moral hazard model
in Holmstr¨ om and Tirole (1997) is a good illustration of such a re-interpretation
where the banking sector enters the model as a borrower subject to borrowing
constraints from its lenders. In the Holmstr¨ om and Tirole (1997) model, there
are two tiers in the agency problem. At the bottom tier, there is a moral haz-
ard problem between a non-?nancial borrower who needs funding to undertake a
project and a bank who supplies the funding. The moral hazard problem entails
an optimal contract where the incentive constraint stipulates that the borrower
58
has enough of a stake in the project that the good action is taken, rather than
the ine?cient action that yields private bene?t. The importance of borrower net
worth, or “skin in the game,” is a theme that Holmstr¨ om and Tirole (1997) shares
with Bernanke and Gertler (1989). However, the innovation in Holmstr¨ om and
Tirole (1997) is that there is a second tier to the agency problem in which the
bank (the lender) itself is subject to a moral hazard problem, so that there is a
constraint on the minimum equity capital that the bank itself must hold at all
times. In this respect, the minimum capital requirement of banks emerges as an
endogenous feature of an agency problem where banks must raise funding from
depositors and other suppliers of funds.
By re-orienting the agency problem so that the focus is on the bank (as bor-
rower) and the ?nancial market (as lenders), the earlier results of Bernanke and
Gertler and Kiyotaki and Moore can be transferred to the context of bank distress
and bank lending. The recent paper by Gertler and Kiyotaki (2009, this volume)
is a good example of such a re-orientation. The agency relationship between
the bank and the ?nancial market lender is a moral hazard problem due to the
possibility that the bank can steal some portion of the project outcome, so that
the bank is required to keep a minimum net worth in place at all times. The
bank must then keep a minimum amount of “skin in the game”, which translates
to a minimum capital ratio that the bank must maintain. When credit losses or a
fall in the price of assets depletes the capital of the bank, the incentive constraint
binds, entailing the withdrawal of lending by the bank.
Adrian and Shin (2008b) take up a similar theme of the binding incentive
constraint of the bank, where the agency problem comes in the form of a risk-
shifting problem where the bank may take the riskier asset when a lower risk asset
may be value-enhancing for the pair as a whole. The bank’s market-determined
minimum capital requirement arises from the need for the bank (the borrower) to
59
keep su?cient stake in the payo?s from the total balance sheet of the bank. Since
the agency problem manifests itself as a risk-shifting problem, Adrian and Shin
(2008b) can address how second-moment incentives can enter the problem, and
how the value at risk (VaR) constraint can emerge as an outcome of the optimal
contracting problem.
Brunnermeier and Sannikov (2009) take the moral hazard theme one step
further by embedding the moral hazard problem in a dynamic, continuous time
contracting environment. In this richer framework, Brunnermeier and Sannikov
examine two separate incentive constraints. One is the familiar one where the
borrower needs to keep su?cient “skin in the game”, and results in a minimum
capital ratio requirement set by the market. The second is a constraint in the
spirit of a value at risk (VaR) constraint which makes the debt instantaneously
risk-free. In order to accommodate a role for both types of constraints, Brunner-
meier and Sannikov incorporate the innovative feature that two types of equity
play an essential role. First, there is equity that carries control rights. This ?rst
type of equity is the stake of the controlling party. Second, there is equity that is
loss-absorbing, but which does not carry control rights. This is the type of equity
that is typi?ed by passive investors in hedge funds whose stake can be returned
by the controlling investor in the hedge fund. The interaction between these two
types of equity is a distinctive feature of the Brunnermeier and Sannikov model,
as well as the dynamic contracting framework, which gives the model considerable
richness and complexity. The role of equity as a bu?er has an a?nity with the
work on the e?ect of regulatory capital on lending as examined by Van den Heuvel
(2002).
So far, we have described the set of papers that have as their starting point
some type of agency problem between a borrower and a lender. However, there
is another strand of the literature from monetary economics that has emphasized
60
the institutional features surrounding the commercial banking industry, especially
for the United States.
The notable example is Bernanke and Blinder (1988), which proposed a simple
model of the supply of credit by banks that emphasizes the binding nature of
the reserve requirement of banks. The constraint itself is not motivated with
further microfoundations; instead, the institution of reserve requirements is taken
as given. The reserve requirement stipulates that, for a given amount of deposit
funding used by a bank, some minimum amount must be kept on deposit at the
Federal Reserve as cash assets of the bank. This is a constraint that links the
two sides of the balance sheet, and the assumption is that such a constraint binds
all the time.
The idea that reserve requirements bind all the time has been dealt a severe
blow by the experience of the aftermath of the recent ?nancial crisis in which
commercial banks in the United States held close to one trillion dollars of excess
reserves on their balance sheet. Excess reserves have also been a common feature
in other parts of the world after the ?nancial crisis. Nevertheless, until recently,
the assumption of binding reserve requirements has been an important feature of
the academic literature in banking.
Building on the initial short paper by Bernanke and Blinder (1988), their
follow-up paper (Bernanke and Blinder (1992)) is an in-depth empirical investiga-
tion of the monetary transmission mechanism. In particular, the focus is on how
the Fed Funds rate works through the ?nancial system to in?uence real activity.
The key section (Section 4) of the paper highlights the important empirical role of
the Fed Funds rate in in?uencing the future loan supply of banks. In particular,
Bernanke and Blinder (1992) show that an increase in the Fed Funds rate leads
to an eventual slow-down of bank lending at a time horizon that is similar to
the impact of the Fed Funds rate on unemployment. In addition, Bernanke and
61
Blinder (1992) show that an initial increase in the Fed Funds rate is met with a
rapid adjustment in the Bank’s portfolio in which the holding of securities ?rst
falls, and then is slowly rebuilt.
Bernanke and Blinder (1992) interpret these ?ndings as showing that the Fed
Funds rate a?ects the supply of bank lending directly through the portfolio con-
straints of the bank itself. In particular, when the Fed Funds rate is raised, the
deposit funding of the bank is squeezed, which puts pressure on the asset side
of the bank’s balance sheet to contract. Since loans are long-term contractual
arrangements, bank lending is initially slow to adjust and all the short-term ad-
justment is made via the holding of securities. Over time, the holding of securities
is built up, but the bank’s loan portfolio adjusts slowly to its new (lower) opti-
mum. In this way, a higher Fed Funds rate is seen to a?ect bank lending through
the squeeze in the deposit funding of the bank, which eventually feeds into the
decrease in bank lending.
The bank lending channel examined by Bernanke and Blinder (1992) has close
a?nities with the risk-taking channel of monetary policy proposed in our chapter.
The common theme is that the Fed Funds rate has a direct impact on credit
supply. However, the di?erences are also apparent. In Bernanke and Blinder
(1992), the mechanism that links the Fed Funds rate with the supply of bank
lending is the binding reserve requirement of the commercial banks. This poses
two challenges in the context of the recent crisis. First, reserve requirements
have not been binding in the aftermath of the crisis. Second, the Bernanke and
Blinder (1992) account focuses on the commercial banking sector, since this is the
sector for which there is a reserve requirement. However, as we have described
in some detail above, the credit crunch in the recent ?nancial crisis originated in
the shadow banking system and the market-based ?nancial intermediaries that
serve it, rather than in the traditional commercial banking sector. Indeed, the
62
commercial banking sector had seen increased balance sheets until the summer of
2009.
Nevertheless, the Bernanke and Blinder (1992) paper stands as a milestone in
the literature on the relationship between monetary policy and the banking sys-
tem. The conjectures that they proposed in their original paper were con?rmed
in a careful cross-sectional empirical study by Kashyap and Stein (2000), who
used a large dataset of banks in the United States to examine portfolio changes in
response to monetary policy shifts. They investigated the conjecture in Bernanke
and Blinder (1992) that banks with less liquid balance sheets (i.e., with fewer secu-
rities and more loans) were subject to greater downward impact in their banking
lending as a result of monetary tightening. They ?nd strong evidence that less
liquid banks are, indeed, subject to greater loan contraction due to monetary
tightening. The results are driven in particular by small banks, which form the
bulk of their sample.
Relative to the literature surveyed above, the risk-taking channel of monetary
policy proposed in this chapter has some distinctive features. First, in contrast
to Bernanke and Gertler (1989) and other approaches that emphasize the demand
for credit and the non-?nancial borrowers’ balance sheet, the risk-taking channel
emphasizes the role of the supply of credit by the ?nancial intermediary sector.
In this respect, the risk-taking channel has greater a?nity with approaches that
emphasize the supply of credit and the constraints that bind on the lenders’ side.
However, the distinctive feature of the risk-taking channel is the role played by
the price of risk and by the market-determined risk premium. The supply of
credit is determined by the threshold value of the risk-premium charged by the
market. In the model sketched in this chapter, the asset choice decision of the
banks is determined by an underlying risk-management problem where banks are
subject to a value at risk (VaR) constraint. In this chapter, we have left this
63
constraint without a further microfoundation. However, such a search would
lead naturally to a meeting point with the agency literature that emphasizes the
constraint imposed on the borrowers by the market as a whole. Adrian and Shin
(2008b) is an example of just such a setting.
7. Concluding Remarks
We conclude with some implications of our ?ndings for the conduct of mone-
tary policy. Our emphasis on the role of balance sheet aggregates of ?nancial
intermediaries leads to policy prescriptions that bear a super?cial similarity to
an older tradition in monetary economics that emphasized the money stock as a
pivotal quantity in monetary policy. The older monetarist tradition emphasized
the stock of money because of the supposed direct link between the money stock
and real expenditures through the portfolio adjustment of individual consumers
who rebalance their portfolios consisting of money and real goods. Monetary
aggregates had fallen from favor in the conduct of monetary policy mainly as a
backlash against the older monetarist line (see Friedman (1988)).
In this chapter, we have focused on balance sheet aggregates of ?nancial inter-
mediaries, but the rationale is quite di?erent from the older monetarist literature.
Our approach has been to emphasize the role of intermediary balance sheets as
a determinant for the risk appetite ruling in the economy, and how monetary
policy can a?ect the growth of intermediary balance sheets. Although our ratio-
nale for looking at balance sheets di?ers from the older monetarist literature, our
discussion nevertheless suggests that there is a case for rehabilitating some role
for balance sheet quantities in the conduct of monetary policy. By in?uencing
the rate of growth of intermediary balance sheets, the monetary authorities can
impact real decisions that depend on the price of risk ruling in the economy. Real
64
decisions that are sensitive to ?nancial conditions, such as residential investment,
will be particularly susceptible to shifts in the price of risk.
To the extent that monetary policy decisions ripple through to the real econ-
omy via the ?nancial system, our discussion also highlights the importance of
tracking the institutional underpinnings of the ?nancial system itself. The insta-
bility of money demand functions that undermines the practical use of monetary
aggregates in the older monetarist-style analysis is closely related to the emergence
of the market-based ?nancial system. As a result of those structural changes, not
all balance sheet quantities will be equally useful. The money stock is a measure
of the liabilities of deposit-taking banks, and so may have been useful before the
advent of the market-based ?nancial system. However, the money stock will be
of less use in a ?nancial system such as that in the United States. More useful
may be measures of collateralized borrowing, such as the weekly series on repos
of primary dealers. The model presented in the paper shows that balance sheet
quantities of ?nancial intermediaries are closely tied to risk premia and the supply
of credit, which, in turn, makes them useful in analyzing the ?nancial conditions
that determine the supply of credit. Adrian, Moench and Shin (2010) present
an empirical analysis that uses balance sheet quantities from a broad range of
?nancial intermediaries in order to gauge ?nancial conditions.
Finally, our results highlight the channel through which monetary policy and
policies toward ?nancial stability are linked. When the ?nancial system as a whole
holds long-term, illiquid assets ?nanced by short-term liabilities, any tensions
resulting from a sharp pullback in leverage will show up somewhere in the system.
Even if some institutions can adjust down their balance sheets ?exibly, there will
be some who cannot. These pinch points will be those institutions that are highly
leveraged, but who hold long-term illiquid assets ?nanced with short-term debt.
When the short-term funding runs away, they will face a liquidity crisis. The
65
traditional lender of last resort tools (such as the discount window), as well as
the recent liquidity provision innovations, are tools that mitigate the severity of
the tightening of balance sheet constraints. However, experience has shown time
and again that the most potent tool in relieving aggregate ?nancing constraints
is a lower target rate. Past periods of ?nancial stress such as the 1998 crisis were
met by reductions in the target rate aimed at insulating the real economy from
?nancial sector shocks. Our ?ndings suggest that, in conducting monetary policy,
the potential for ?nancial sector distress should be explicitly taken into account
in a forward-looking manner.
66
References
Adrian, Tobias, Arturo Estrella, and Hyun Song Shin (2010) “Monetary Cycles,
Financial Cycles, and the Business Cycle,” Federal Reserve Bank of New York
Sta? Report 421.
Adrian, Tobias, Dina Marchioni, and Karin Kimbrough (2010) “The Federal Re-
serve’s Commercial Paper Funding Facility,” Federal Reserve Bank of New York
Sta? Reports 423.
Adrian, Tobias, Erkko Etula and Hyun Song Shin (2009) “Risk Appetite and
Exchange Rates,” Federal Reserve Bank of New York Sta? Report 361.
Adrian, Tobias, Emanuel Moench and Hyun Song Shin (2010) “Financial Inter-
mediation, Asset Prices, and Macroeconomic Dynamics,” Federal Reserve Bank
of New York Sta? Report 422.
Adrian, Tobias, Emanuel Moench and Hyun Song Shin (2009) “Macro Risk Pre-
mium and Intermediary Balance Sheet Quantities,” Federal Reserve Bank of New
York Sta? Report 428.
Adrian, Tobias and Hyun Song Shin (2007) “Liquidity and Leverage,” Journal
of Financial Intermediation, forthcoming. Available as Federal Reserve Bank of
New York Sta? Reports 328.
Adrian, Tobias and Hyun Song Shin (2008a) “Liquidity, Monetary Policy, and
Financial Cycles,” Federal Reserve Bank of New York Current Issues in Economics
and Finance, Volume 14 (1).
Adrian, Tobias and Hyun Song Shin (2008b) “Financial Intermediary Leverage
and Value at Risk,” Federal Reserve Bank of New York Sta? Reports, 338.
Adrian, Tobias and Hyun Song Shin (2008c) “Liquidity and Financial Contagion,”
Banque de France Financial Stability Review, February 2008
Adrian, Tobias and Hyun Song Shin (2008d) “Financial Intermediaries, Finan-
cial Stability, and Monetary Policy,” Federal Reserve Bank of Kansas City 2008
Jackson Hole Economic Symposium Proceedings.
67
Adrian, Tobias and Hyun Song Shin (2009) “The Changing Nature of Financial
Intermediation and the Financial Crisis of 2007 - 2009,” forthcoming in the Annual
Review of Economics.
Ashcraft, Adam (2005) “Are Banks Really Special? New Evidence from the FDIC-
Induced Failure of Healthy Banks,” American Economic Review 95, pp. 1712-
1730.
Ashcraft, Adam (2006) “New Evidence on the Lending Channel,” Journal of
Money, Credit, and Banking 38, pp. 751-776.
Bernanke, Ben (1983) “Nonmonetary Aspects of the Financial Crisis in the Prop-
agation of the Great Depression” American Economic Review, 73, 257-76
Bernanke, Ben (2004) “The Logic of Monetary Policy,” remarks before the Na-
tional Economists Club, December 2, 2004.
www.federalreserve.gov/boarddocs/speeches/2004/20041202/default.htm
Bernanke, B. and A. Blinder (1988) “Credit, Money and Aggregate Demand”
American Economic Review, 78, 435-39.
Bernanke, B. and A. Blinder (1992) “The Federal Funds Rate and the Channel
of Monetary Transmission” American Economic Review 82, pp. 901 - 921.
Bernanke, Ben and Mark Gertler (1989) “Agency Costs, Net Worth, and Business
Fluctuations,” American Economic Review 79, pp. 14 - 31.
Bernanke, Ben and Mark Gertler (1995) “Inside the Black Box: The Credit Chan-
nel of Monetary Policy Transmission,” Journal of Economic Perspectives 9, pp.
27-48.
Bernanke, Ben and Mark Gertler (1999) “Monetary Policy and Asset Volatility,”
Federal Reserve Bank of Kansas City Economic Review 84, pp. 17-52.
Bernanke, Ben and Cara Lown (1991) “The Credit Crunch,” Brookings Papers on
Economic Activity 2, pp. 205-247.
Blinder, Alan (1998) Central Banking in Theory and Practice, MIT Press, Cam-
bridge.
68
Borio, Claudio and Haibin Zhu (2008) “Capital Regulation, Risk-taking and Mon-
etary Policy: A Missing Link in the Transmission Mechanism?” Bank for Inter-
national Settlements Working Paper 268.
Brunnermeier, Markus and Yuliy Sannikov (2009) “A Macroeconomic Model with
a Financial Sector,” unpublished working paper, Princeton University
Curdia, Vasco and Michael Woodford (2009) “Credit Spreads and Optimal Mon-
etary Policy,” Federal Reserve Bank of New York Sta? Reports 385.
Danielsson, Jon, Hyun Song Shin and Jean—Pierre Zigrand (2009) “Risk Appetite
and Endogenous Risk”, unpublished working paper, London School of Economics
and Princeton University.
Estrella, Arturo, and Gikas Hardouvelis (1989) “The Term Structure as a Pre-
dictor of Real Economic Activity.” Federal Reserve Bank of New York Research
Paper, May.
Estrella, Arturo, and Gikas Hardouvelis. (1991) “The Term Structure as a Pre-
dictor of Real Economic Activity.” Journal of Finance, 46: 555-576.
Estrella, Arturo, Anthony P. Rodrigues, and Sebastian Schich (2003) “How Stable
is the Predictive Power of the Yield Curve? Evidence from Germany and the
United States.” Review of Economics and Statistics, 85: 629-644.
Freixas, Xavier (2009) ”Monetary Policy in a Systemic Crisis,” unpublished work-
ing paper, University Pompeu Fabra.
Friedman, Benjamin (1988) “Monetary Policy Without Quantity Variables,” Amer-
ican Economic Review 78, 440-45.
Gatev, Evan, Til Schuermann, and Philip Strahan (2009) ”Managing Bank Liq-
uidity Risk: How Deposit-Loan Synergies Vary with Market Conditions,” Review
of Financial Studies 22(3), pp. 995-1020.
Geanakoplos, John (2009) “The Leverage Cycle” in 2009 NBER Macroeconomics
Annual 24, edited by Daron Acemoglu, Kenneth Rogo?, and Michael Woodford.
University of Chicago Press.
Gertler, Mark and Peter Karadi (2009) “A Model of Unconventional Monetary
Policy,” unpublished working paper, New York University.
69
Gertler, Mark and Nobuhiro Kiyotaki (2009) “Financial Intermediation and Credit
Policy in Business Cycle Analysis” working paper, NYU and Princeton University,
forthcoming in the Handbook of Monetary Economics.
Harvey, Campbell R. (1989) “Forecasts of Economic Growth from the Bond and
Stock Markets.” Financial Analysts Journal, September/October: 38-45.
Holmstr¨ om Bengt and Jean Tirole (1997) “Financial Intermediation, Loanable
Funds, and the Real Sector,” Quarterly Journal of Economics, 112, 663-692.
Ioannidou, Vasso, Ongena, Steven and Jose Luis Peydro (2008) “Monetary Policy,
Risk-Taking and Pricing: Evidence from a Quasi-Natural Experiment,” unpub-
lished working paper, ECB.
Jimenez, Gabriel, Ongena, Steven, Peydr´ o, Jose Luis and Jesus Saurina (2008)
“Hazardous Times for Monetary Policy: What do 23 Million Bank Loans Say
About the E?ects of Monetary Policy on Credit Risk?” unpublished working pa-
per, ECB and Bank of Spain.
Kashyap, Anil and Jeremy Stein (1994) “Monetary Policy and Bank Lending,” in
N. Gregory Mankiw (ed.) Monetary Policy, University of Chicago Press.
Kashyap, Anil and Jeremy Stein (2000) “What Do a Million Observations on
Banks Say about the Transmission of Monetary Policy?” American Economic
Review 90, pp. 407-428.
Keister, Todd, and James McAndrews (2009) “Why Are Banks Holding So Many
Excess Reserves?” Federal Reserve Bank of New York Sta? Report 380.
http://www.ny.frb.org/research/sta? reports/sr380.html
Lown, Cara, and Don Morgan (2006) “The Credit Cycle and the Business Cycle:
New Findings Using the Loan O?cer Opinion Survey,” Journal of Money, Credit,
and Banking 38, pp. 1575-1597.
Khwaja, Asim Ijaz and Atif Mian (2008) “Tracing The Impact of Bank Liquidity
Shocks.” American Economic Review 98, pp. 1413—1442.
Kiyotaki, Nobuhiro, and John Moore (1997) “Credit Cycles,” Journal of Political
Economy 105, pp. 211-248.
70
Kiyotaki, Nobuhiro, and John Moore (2005) “Liquidity and Asset Prices,” Inter-
national Economic Review 46, pp. 317-349.
Paravisini, Daniel (2008) “Local Bank Financial Constraints and Firm Access to
External Finance,” Journal of Finance 63, pp. 2161 - 2193.
Shin, Hyun Song (2009) Risk and Liquidity, 2008 Clarendon Lectures in Finance,
forthcoming from Oxford University Press.
Sims, Christopher (1980) ”Macroeconomics and Reality,” Econometrica 48, 1-48
Shleifer, Andrei, and Robert Vishny (1997) ”The Limits of Arbitrage,” Journal
of Finance 52(1), pp. 35-55.
Shleifer, Andrei, and Robert Vishny (2009) ”Unstable Banking,” Journal of Fi-
nancial Economics, forthcoming.
Stock, James, and Mark Watson. (1989) “New Indexes of Coincident and Lead-
ing Indicators.” In NBER Macroeconomic Annual 4, ed. Olivier Blanchard and
Stanley Fischer, 351-394.
Stock, James, and Mark Watson. (1993) “A Procedure for Predicting Recessions
with Leading Indicators: Econometric Issues and Recent Performance.” In Busi-
ness Cycles, Indicators, and Forecasting, ed. James Stock and Mark Watson.
Chicago, IL: University of Chicago Press.
Svensson, Lars (2004) “Challenges for Monetary Policy,” paper for the Bellagio
Group Meeting at the National Bank of Belgium, January 2004.
www.princeton.edu/˜svensson
Taylor, John (1993) “Discretion Versus Policy Rules in Practice,” Carnegie-Rochester
Series on Public Policy 39, pp. 195-214.
Van den Heuvel, S. (2002) “The Bank Capital Channel of Monetary Policy”,
working paper, Wharton School, University of Pennsylvania.
Woodford, Michael (2003) Interest and Prices: Foundations of a Theory of Mon-
etary Policy, Princeton University Press.
Woodford, Michael (2005) “Central Bank Communication and Policy E?ective-
ness,” Proceedings of the Federal Reserve Bank of Kansas City Economic Sympo-
sium at Jackson Hole 2005.
http://www.kc.frb.org/publicat/sympos/2005/sym05prg.htm
71
Appendix: Data Sources
Figure 1.1. The 10-year and 3-month Treasury constant maturity yields as well
as the e?ective Fed Funds rate are from the Federal Reserve Board’s H.15 release.
Figures 3.1—3.6. Figures 3.1, 3.4, 3.5, 3.6 use total assets of security broker-
dealers, ABS issuers, shadow banks (the sum of ABS issuers, ?nance companies,
funding corporations), and nationally chartered commercial banks from the Flow
of Funds published by the Federal Reserve Board. In Figures 3.2 and 3.3, the
money stock measure M1 and M2. Total outstanding and ?nancial commercial
paper used in Figures 3.2, 3.3, and 3.5 are from the Federal Reserve Board. Pri-
mary dealer repo in Figures 3.2, 3.3, and 3.6 is from the Federal Reserve Bank of
New York.
Figures 3.7 and 3.8. The ?gures use total ?nancial assets from the Federal
Reserve Board’s Flow of Funds.
Figures 4.1 and 4.2. The ?gures are based on computations by Adrian, Moench,
and Shin (2009). The macro risk premium as the predicted part of a regression
of real GDP growth on constant maturity Treasury yield spreads and corporate
bond spreads. The risk appetite variable is obtained by regressing (negative)
changes of the macro risk premium on lagged balance sheet variables of security
broker-dealers, shadow banks, and commercial banks.
Tables 4.1 and 4.2: Impact of Balance Sheets on GDP and Residential
Investment. The tables report regressions of GDP and residential investment
growth on the total asset growth of broker-dealers, shadow banks, and commercial
banks for 1986Q1 to 2009Q2. Lags are one quarter lags; growth rates are an-
nual. Total assets are from the Federal Reserve Board’s Flow of Funds. Shadow
banks include ABS issuers, funding corporations, and ?nance companies. Gross
domestic product (GDP) and residential investment is from the Bureau of Eco-
nomic Analysis (BEA). PCE in?ation is the personal consumption expenditures
de?ator excluding food and energy as reported by BEA. The equity return is the
one quarter return of Standard & Poor’s S&P500 index. The VIX is CBOE’s
72
implied volatility index (the VXO from 1986-1989, and the VIX from 1990 on-
wards). The term spread is the di?erence between the 10-year constant maturity
Treasury yield and the 3-month Treasury bill rate, both are from the Federal Re-
serve Board. The credit spread is the di?erence between Moody’s Baa spread
and the 10-year Treasury rate, both are from the Federal Reserve Board.
Table 6.1: Determinants of Balance Sheet Growth The table reports re-
gressions of repo growth, repo + commercial paper growth, and M2 growth on
their own lags, and asset price variables. The data frequency is weekly from
October 3, 1990 to February 3, 2010. Changes refer to 1-week changes, and lags
to 1-week lags. Fed Funds denotes the Federal Funds Target as reported by the
Federal Reserve Board. The equity return is the 1-week return of Standard &
Poor’s S&P500. The VIX is CBOE’s implied volatility index for the S&P500.
The term spread is the di?erence between the 10-year constant maturity Treasury
yield and the 3-month Treasury bill rate, both from the Federal Reserve Board.
The credit spread is the di?erence between Moody’s Baa spread and the 10-year
Treasury rate. Commercial paper growth is the 1-week growth rate of total com-
mercial paper outstanding reported by the Federal Reserve Board. Repo growth
is the 1-week growth rate of primary dealer repo, from the Federal Reserve Bank
of New York. M2 growth is the 1-week growth of the money measure M2 from
the Federal Reserve Board.
73
doc_677767475.pdf
Monetary economics is a branch of economics that historically prefigured and remains integrally linked to macroeconomics.[1] Monetary economics provides a framework for analyzing money in its functions as a medium of exchange, store of value, and unit of account.
Federal Reserve Bank of New York
Staff Reports
Financial Intermediaries and Monetary Economics
Tobias Adrian
Hyun Song Shin
Staff Report no. 398
October 2009
Revised May 2010
This paper presents preliminary findings and is being distributed to economists
and other interested readers solely to stimulate discussion and elicit comments.
The views expressed in the paper are those of the authors and are not necessarily
reflective of views at the Federal Reserve Bank of New York or the Federal
Reserve System. Any errors or omissions are the responsibility of the authors.
Financial Intermediaries and Monetary Economics
Tobias Adrian and Hyun Song Shin
Federal Reserve Bank of New York Staff Reports, no. 398
October 2009; revised May 2010
JEL classification: E00, E02, G28
Abstract
We reconsider the role of financial intermediaries in monetary economics. We explore
the hypothesis that financial intermediaries drive the business cycle by way of their role
in determining the price of risk. In this framework, balance sheet quantities emerge as a
key indicator of risk appetite and hence of the “risk-taking channel” of monetary policy.
We document evidence that the balance sheets of financial intermediaries reflect the
transmission of monetary policy through capital market conditions. Our findings suggest
that the traditional focus on the money stock for the conduct of monetary policy may
have more modern counterparts, and we suggest the importance of tracking balance sheet
quantities for the conduct of monetary policy.
Key words: financial intermediation, monetary policy, risk-taking channel
Adrian: Federal Reserve Bank of New York (e-mail: [email protected]). Shin: Princeton
University (e-mail: [email protected]). This paper is a preliminary version of a chapter
prepared for the Handbook of Monetary Economics. The views expressed in this paper are
those of the authors and do not necessarily reflect the position of the Federal Reserve Bank
of New York or the Federal Reserve System.
1. Introduction
In conventional models of monetary economics commonly commonly used in cen-
tral banks, the banking sector has not played a prominent role. The primary
friction in such models is the price stickiness of goods and services. Financial
intermediaries do not play a role, save as a passive player that the central bank
uses as a channel to implement monetary policy.
However, ?nancial intermediaries have been at the center of the global ?nancial
crisis that erupted in 2007. They have borne a large share of the credit losses from
securitized subprime mortgages, even though securitization was intended to parcel
out and disperse credit risk to investors who were better able to absorb losses.
Credit losses and the associated ?nancial distress have ?gured prominently in the
commentary on the downturn in real economic activity that followed. These
recent events suggest that ?nancial intermediaries may be worthy of separate
study in order to ascertain their role in economic ?uctuations.
The purpose of this chapter in the Handbook of Monetary Economics is to
reconsider the role of ?nancial intermediaries in monetary economics. In ad-
dressing the issue of ?nancial factors in macroeconomics, we join a spate of recent
research that has attempted to incorporate a ?nancial sector in a New Keynesian
DSGE model. Curdia and Woodford (2009) and Gertler and Karadi (2009) are
recent examples. However, rather than phrasing the question as how ?nancial
“frictions” a?ect the real economy, we focus on the ?nancial intermediary sector
itself. We explore the hypothesis that the ?nancial intermediary sector, far from
being passive, is instead the engine that drives the boom-bust cycle. To explore
this hypothesis, we propose a framework for study with a view to addressing the
following pair of questions. What are the channels through which ?nancial in-
termediaries exert an in?uence on the real economy (if at all), and what are the
1
implications for monetary policy?
Banks and other ?nancial intermediaries borrow in order to lend. Since the
loans o?ered by banks tend to be of longer maturity than the liabilities that fund
those loans, the term spread is indicative of the marginal pro?tability of an extra
dollar of loans on intermediaries’ balance sheets. The net interest margin (NIM)
of the bank is the di?erence between the total interest income on the asset side
of its balance sheet and the interest expense on the liabilities side of its balance
sheet. Whereas the term spread indicates the pro?tability of the marginal loan
that is added to the balance sheet, the net interest margin is an average concept
that applies to the stock of all loans and liabilities on the balance sheet.
The net interest margin determines the pro?tability of bank lending and in-
creases the present value of bank income, thereby boosting the forward-looking
measures of bank capital. Such a boost in bank capital increases the capacity
of the bank to increase lending in the sense that the marginal loan that was not
made before the boost in bank capital now becomes feasible under the greater
risk-bearing capacity of the bank. As banks expand their balance sheets, the
market price of risk falls.
In this framework, ?nancial intermediaries drive the ?nancial cycle through
their in?uence on the determination of the price of risk. Quantity variables -
particularly the components of ?nancial intermediary balance sheets - emerge as
important economic indicators due to their role in re?ecting the risk capacity
of banking sector and hence on the marginal real project that receives funding.
In this way, the banking sector plays a key role in determining the level of real
activity. Ironically, our ?ndings have some points of contact with the older
theme in monetary economics of keeping track of the money stock at a time
when it has fallen out of favor among monetary economists.
1
The common
1
See Friedman (1988) for an overview of the role of monetary aggregates in macroeconomic
2
theme between our framework and the older literature is that the money stock
is a balance sheet aggregate of the ?nancial sector. Our approach suggests that
broader balance sheet aggregates such as total assets and leverage are the relevant
?nancial intermediary variables to incorporate into macroeconomic analysis.
When we examine balance sheet measures that re?ect the underlying funding
conditions in capital markets, we ?nd that the appropriate balance sheet quantities
are of institutions that are marking to market their balance sheets. In this regard,
?uctuations in shadow bank and broker-dealer assets are more informative than
movements in commercial bank assets. However, as commercial banks begin
to mark more items of their balance sheets to market, commercial bank balance
sheet variables are likely to become more important variables for studying the
transmission mechanism.
Our ?ndings have important implications for the conduct of monetary pol-
icy. According to the perspective outlined here, ?uctuations in the supply of
credit arise from the interactions between bank risk-taking and the market risk
premium. The cost of leverage of market-based intermediaries is determined by
two main variables — risk and risk-taking capacity. The expected pro?tability of
intermediaries is proxied by spreads such as the term spread and various credit
spreads. Variations in the policy target determine short-term interest rates, and
have a direct impact on the pro?tability of intermediaries. For these reasons,
short-term interest rates matter directly for monetary policy.
The e?ect of keeping policy rates low in the aftermath of the ?nancial crisis of
2008 has illustrated again the potency of low policy interest rates in raising the
pro?tability of banks and thereby recapitalizing the banking system from their
dangerous low levels. When considering the debates in early 2009 about the
necessity (or inevitability) of capital injections into the U.S banking system, the
?uctuations in the United States.
3
turnaround in the capital levels of the U.S. banking sector has been worthy of
note.
Empirically, there is (for the United States) a near perfect negative one-to-
one relationship between 4-quarter changes of the Fed Funds target and 4-quarter
changes of the term spread de?ned as the 10-year/3-month term Treasury spread
(Figure 1.1 uses data from 1987q1 to 2008q3). Thus, shifts in the policy rate
translate directly into shifts in the slope of the yield curve. Since the term spread
a?ects the pro?tability of the marginal loan and the future net interest margin
(NIM) of the bank, the short rate signals future risk-taking capacity of the banking
sector. In this way, variations in the target rate a?ect real activity because they
change the risk-taking capacity of ?nancial intermediaries, thus shifting market
risk premiums and the supply of credit. Borio and Zhu (2008) have coined
the term “risk-taking channel” of monetary policy to describe this set of e?ects
working through the risk appetite of ?nancial intermediaries.
This perspective on the importance of the short rate as a price variable is
in contrast to current monetary thinking, where short-term rates matter only to
the extent that they determine long term interest rates, which are seen as being
risk-adjusted expectations of future short rates. Current models of monetary
economics used at central banks emphasize the importance of managing market
expectations. By charting a path for future short rates and communicating this
path clearly to the market, the central bank can in?uence long rates and thereby
in?uence mortgage rates, corporate lending rates, and other prices that a?ect
consumption and investment. This “expectations channel”, which is explained
in Bernanke (2004), Svensson (2004), and Woodford (2003, 2005), has become an
important consideration for monetary policy. In his book on central banking,
Alan Blinder (1998, p.70) phrases the claim in a particularly clear way.
“central banks generally control only the overnight interest rate,
4
Figure 1.1: The term spread and the Federal Funds rate.
an interest rate that is relevant to virtually no economically inter-
esting transactions. Monetary policy has important macroeconomic
e?ects only to the extent that it moves ?nancial market prices that
really matter - like long-term interest rates, stock market values and
exchange rates.”
In contrast, our results suggest that short-term rates may be important in their
own right. Short rates matter because they largely determine the term spread,
which in turn determine the net interest margin and the forward-looking capital
of the banking sector. Continued low short rates imply a steep yield curve for
some time, higher net interest margin in the future, and hence higher risk-taking
capacity of the banking sector. Conversely, higher short rates imply lower future
net interest margin and a decline in the risk-taking capacity of the banking sector.
In particular, an inverted yield curve is a sign of diminished risk-taking capacity,
5
and by extension of lower real activity.
There is empirical support for the risk-taking channel of monetary policy. We
?nd that the growth in shadow bank balance sheets and broker-dealer balance
sheets help to explain future real activity. However, we also ?nd that ?uc-
tuations in the balance sheet size of shadow banks and security broker-dealers
appear to signal shifts in future real activity better than the ?uctuations of the
larger commercial banking sector. Thus, one lesson from our empirical analysis
is that there are important distinctions between di?erent categories of ?nancial
intermediaries. In fact, the evolutions of shadow bank and broker-dealer assets
have time signatures that are markedly di?erent from those of commercial banks.
Our results point to key di?erences between banking, as traditionally conceived,
and the market-based banking system that has become increasingly in?uential in
charting the course of economic events.
Having established the importance of ?nancial intermediary balance sheets in
signaling future real activity, we go on to examine the determinants of balance
sheet growth. We ?nd that short-term interest rates are important. Indeed, the
level of the Fed Funds target is a key variable: a lowering of short-term rates are
conducive to expanding balance sheets. In addition, a steeper yield curve, larger
credit spreads, and lower measures of ?nancial market volatility are conducive to
expanding balance sheets. In particular, an inverted yield curve is a harbinger of
a slowdown in balance sheet growth, shedding light on the empirical feature that
an inverted yield curve forecasts recessions. The Fed Funds target determines
other relevant short term interest rates, such as repo rates and interbank lending
rates through arbitrage in the money market. As such, we may expect the Fed
Funds rate to be pivotal in setting short-term interest rates more generally.
These ?ndings re?ect the economics of ?nancial intermediation, since the busi-
ness of banking is to borrow short and lend long. For an o?-balance sheet vehicle
6
such as a conduit or SIV (structured investment vehicle) that ?nances holdings of
mortgage assets by issuing commercial paper, a di?erence of a quarter or half per-
cent in the funding cost may make all the di?erence between a pro?table venture
and a loss-making one. This is because the conduit or SIV, like most ?nancial
intermediaries, is simultaneously both a creditor and a debtor — it borrows in
order to lend.
The outline of this chapter is as follows. We begin with a simple equilibrium
model where ?nancial intermediaries are the main engine for the determination
of the price of risk in the economy. We then present empirical results on the real
impact of shadow bank and broker-dealer balance sheet changes, and on the role
of short-term interest rates in the determination of balance sheet changes. We
also consider the role of the central bank as the lender of last resort (LOLR) in
light of our ?ndings. We conclude by drawing some lessons for monetary policy.
2. Financial Intermediaries and the Price of Risk
To motivate the study of ?nancial intermediaries and how they determine the
price of risk, we begin with a stylized model set in a one period asset market.
2
The general equilibrium model below is deliberately stark. It has two features
that deserve emphasis.
First, there is no default in the model. The debt that appears in the model
is risk-free. However, as we will see, the ampli?cation of the ?nancial cycle is
present. Geanakoplos (2009) has highlighted how risk-free debt may still give
rise to powerful spillover e?ects through ?uctuations in leverage and the pricing
of risk. The model also incorporates insights from Shleifer and Vishny (1997),
who demonstrate that ?nancial constraints can lead to ?uctuations of risk premia
2
A similar model appeared in Shin (2009).
7
even if arbitrageurs are risk neutral.
3
Adrian and Shin (2007) exhibit empirical
evidence that bears on the ?uctuations in the pricing of risk from the balance
sheets of ?nancial intermediaries.
Second, in the example, there is no lending and borrowing between ?nancial
intermediaries themselves. So, any e?ect we see in the model cannot be attributed
to what we may call the “domino model” of systemic risk, where systemic risk
propagates through the ?nancial system via a chain of defaults of ?nancial inter-
mediaries.
4
This is not to say, of course, that interlocking claims do not matter.
However, the benchmark case serves the purpose of showing that chains of default
are not necessary for ?uctuations in the price of risk.
To anticipate the punchline from the simple model, we show that aggregate
capital of the ?nancial intermediary sector stands in a one-to-one relation with
the price of risk and the availability of funding that ?ows to real projects. The
larger is the aggregate intermediary sector capital, the lower is the price of risk,
and the easier is credit.
2.1. Model
Today is date 0. A risky security is traded today in anticipation of its realized
payo? in the next period (date 1). The payo? of the risky security is known at
date 1. When viewed from date 0, the risky security’s payo? is a random variable
˜ n, with expected value ¡ 0. The uncertainty surrounding the risky security’s
payo? takes a particularly simple form. The random variable ˜ n is uniformly
distributed over the interval:
[¡ ?.. ¡ + .]
3
Shleifer and Vishny (2009) present a theory of unstable banking that is closely related to
our model.
4
See Adrian and Shin (2008c) for an argument for why the “domino model” is inappropriate
for understanding the crisis of 2007 -9.
8
The mean and variance of ˜ n is given by
1 ( ˜ n) = ¡
o
2
=
.
2
3
There is also a risk-free security, which we call “cash”, that pays an interest rate
of i. Let j denote the price of the risky security. For an investor with equity c
who holds ¸ units of the risky security, the payo? of the portfolio is the random
variable:
\ ? ˜ n¸ + (1 + i) (c ?j¸) (2.1)
= ( ˜ n ?(1 + i) j)
| {z }
risky excess return
¸ + (1 + i) c
| {z }
risk-free ROE
(2.2)
There are two groups of investors - passive investors and active investors. The
passive investors can be thought of as non-leveraged investors such as households,
pension funds and mutual funds, while the active investors can be interpreted as
leveraged institutions such as banks and securities ?rms who manage their bal-
ance sheets actively. The risky securities can be interpreted as loans granted to
ultimate borrowers or securities issued by the borrowers, but where there is a risk
that the borrowers do not fully repay the loan. Figure 2.1 depicts the relation-
ships. Under this interpretation, the market value of the risky securities can be
thought of as the marked-to-market value of the loans granted to the ultimate
borrowers. The passive investors’ holding of the risky security can then be inter-
preted as the credit that is granted directly by the household sector (through the
holding of corporate bonds, for example), while the holding of the risky securities
by the active investors can be given the interpretation of intermediated ?nance
where the active investors are banks that borrow from the households in order to
lend to the ultimate borrowers.
9
Banks
(Active
Investors)
Households
(Passive
Investors)
end-user
borrowers
Intermediated
Credit
Debt
Claims
Directly granted credit
Figure 2.1: Intermediated and Directly Granted Credit
We assume that the passive investors have mean-variance preferences over the
payo? from the portfolio. They aim to maximize
l = 1 (\) ?
1
2t
o
2
W
(2.3)
where t 0 is a constant called the investor’s “risk tolerance” and o
2
W
is the
variance of \. In terms of the decision variable ¸, the passive investor’s objective
function can be written as
l (¸) = (¡,j ?(1 + i))
| {z }
Expected Excess Return
j¸ + (1 + i) c ?
1
6t
¸
2
.
2
(2.4)
The optimal holding of the risky security satis?es the ?rst order condition:
(¡ ?(1 + i) j) ?
1
3t
¸.
2
= 0
The price must be below the expected payo? for the risk-averse investor to hold
any of the risky security. The optimal risky security holding of the passive investor
(denoted by ¸
1
) is given by
¸
1
=
?
?
?
?
?
3t
.
2
(¡ ?(1 + i) j) if ¡ j (1 + i)
0 otherwise
(2.5)
10
These linear demands can be summed to give the aggregate demand. If t
i
is the
risk tolerance of the ith investor and t =
P
i
t
i
, then (2.5) gives the aggregate
demand of the passive investor sector as a whole.
Now turn to the portfolio decision of the active (leveraged) investors. These
active investors are risk-neutral but face a Value-at-Risk (VaR) constraint, as is
commonly the case for banks and other leveraged institutions.
5
The general VaR
constraint is that the capital cushion be large enough that the default probability
is kept below some benchmark level. Consider the special case where that bench-
mark level is zero. This is an extreme assumption which we adopt for the purpose
of simplifying the model. By setting the Value-at-Risk constraint to allow no de-
fault by the bank, we can treat bank liabilities as a perfect substitute for cash. It
would be possible to allow for a less stringent Value-at-Risk constraint that allows
possible default by the bank, but then the modeling has to make allowance for
the bank’s liabilities being risky debt and being priced accordingly. However, the
key qualitative features of the model would be una?ected. Thus, in what follows,
we will adopt the stringent version of the VaR constraint where the bank holds
enough capital to meet the worst case loss, and where the bank’s liabilities are
risk-free.
Denote by VaR the Value-at-Risk of the leveraged investor. The constraint is
that the investor’s capital (equity) c be large enough to cover this Value-at-Risk.
The optimization problem for an active investor is:
max
j
1 (\) subject to VaR ? c (2.6)
If the price is too high (i.e. when j ¡, (1 + i) so that the price exceeds
the discounted expected payo?) the investor holds no risky securities. When
j < ¡, (1 + i), then 1 (\) is strictly increasing in ¸, and so the Value-at-Risk
5
A microfoundation for the VaR constraint is provided by Adrian and Shin (2008b).
11
constraint binds. The optimal holding of the risky security can be obtained by
solving VaR = c. To solve this equation, write out the balance sheet of the
leveraged investor as
Assets Liabilities
securities, j¸
equity, c
debt, j¸ ?c
For each unit of the risky security, the minimum payo? is ¡ ? .. Thus, the
worst case loss is (j (1 + i) ?(¡ ?.)) ¸. In order for the bank to have enough
equity to cover the worst case loss, we require:
(j (1 + i) ?(¡ ?.)) ¸ ? c (2.7)
This inequality also holds in the aggregate. The left hand side of (2.7) is the
Value-at-Risk (the worst possible loss), which must be met by the equity bu?er
c. Since the constraint binds, the optimal holding of the risky securities for the
leveraged investor is
¸ =
c
j (1 + i) ?(¡ ?.)
(2.8)
So the demand from the bank for the risky asset depends positively on the expected
excess return to the risky asset ¡?(1 + i) j, and positively on the amount of equity
that the bank is endowed with c.
Since (2.8) is linear in c, the aggregate demand of the leveraged sector has the
same form as (2.8) when c is the aggregate capital of the leveraged sector as a
whole.
Replacing the constraint (2.8) in the amount of debt j¸ ?c allows us to write
the new balance sheet as follows:
12
Assets Liabilities
securities, j¸
equity, c
debt,
¡ ?.
1 + i
¸
(2.9)
where the debt
q?:
1+i
¸ was constructed by substituting c = ?((¡,j ?(1 + i)) j ?.) ¸
into j¸ ? c. We assume that ¡ . so as to ensure that the payo? of the risky
security is non-negative. The bank’s leverage is the ratio of total assets to equity,
which can be written as:
leverage =
j¸
c
=
j
j (1 + i) ?(¡ ?.)
(2.10)
Denoting by ¸ the holding of the risky securities by the active investors and
by ¸
1
the holding by the passive investors, the market clearing condition is
¸ + ¸
1
= o (2.11)
where o is the total endowment of the risky securities. Figure 2.2 illustrates the
equilibrium for a ?xed value of aggregate capital c. For the passive investors, their
demand is linear, with the intercept at ¡, (1 + i). The demand of the leveraged
sector can be read o? from (2.8). The solution is fully determined as a function
of c. In a dynamic model, c can be treated as the state variable (see Danielsson,
et al. (2009)).
Now consider a possible scenario involving an improvement in the fundamen-
tals of the risky security where the expected payo? of the risky securities rises
from ¡ to ¡
0
. In our banking interpretation of the model, an improvement in the
expected payo? should be seen as an increase in the marked-to-market value of
bank assets. For now, we simply treat the increase in ¡ as an exogenous shock.
Figure 2.3 illustrates the scenario. The improvement in the fundamentals of the
risky security pushes up the demand curves for both the passive and active in-
vestors, as illustrated in Figure 2.3. However, there is an ampli?ed response from
13
p
0
S
demand of
passive investors
demand of
VaR-constrained
investors
i
q
? 1 i
q
? 1
Figure 2.2: Market Clearing Price
the leveraged investors as a result of marked-to-market gains on their balance
sheets.
From (2.9), denote by c
0
the new equity level of the leveraged investors that
incorporates the capital gain when the price rises to j
0
. The initial amount of
debt was
q?:
1+i
¸. Since the new asset value is j
0
¸, the new equity level c
0
is
c
0
= (j
0
(1 + i) ?(¡ ?.)) ¸ (2.12)
Figure 2.4 breaks out the steps in the balance sheet expansion. The initial balance
sheet is on the left, where the total asset value is j¸. The middle balance sheet
shows the e?ect of an improvement in fundamentals that comes from an increase
in ¡, but before any adjustment in the risky security holding. There is an increase
in the value of the securities without any change in the debt value, since the debt
was already risk-free to begin with. So, the increase in asset value ?ows through
entirely to an increase in equity. Equation (2.12) expresses the new value of
equity c
0
in the middle balance sheet in Figure 2.4.
The increase in equity relaxes the value at risk constraint, and the leveraged
14
p
0
S
? ? i q ? 1 /
' p
? ? i q ? 1 / '
Figure 2.3: Ampli?ed response to improvement in fundamentals ¡
Initial
balance sheet
After q shock
Final
balance sheet
debt
equity
assets
increase
in equity
equity
assets
debt
assets
increase in
value of
securities
equity
debt
new
borrowing
new
purchase of
securities
Figure 2.4: Balance sheet expansion from ¡ shock
15
sector can increase its holding of risky securities. The new holding ¸
0
is larger,
and is enough to make the VaR constraint bind at the higher equity level, with a
higher fundamental value ¡
0
. That is,
c
0
= (j
0
(1 + i) ?(¡
0
?.)) ¸
0
(2.13)
After the ¡ shock, the investor’s balance sheet has strengthened, in that capital
has increased without any change in debt value. There has been an erosion of
leverage, leading to spare capacity on the balance sheet in the sense that equity
is now larger than is necessary to meet the Value-at-Risk. In order to utilize the
slack in balance sheet capacity, the investor takes on additional debt to purchase
additional risky securities. The demand response is upward-sloping. The new
holding of securities is now ¸
0
, and the total asset value is j
0
¸
0
. Equation (2.13)
expresses the new value of equity c
0
in terms of the new higher holding ¸
0
in the
right hand side balance sheet in Figure 2.4. From (2.12) and (2.13), we can write
the new holding ¸
0
of the risky security as
¸
0
= ¸
µ
1 +
¡
0
?¡
j
0
(1 + i) ?¡
0
+ .
¶
(2.14)
From the demand of passive investors (2.5) and market clearing,
(1 + i) j
0
?¡
0
=
.
2
3t
(¸
0
?o)
Substituting into (2.14),
¸
0
= ¸
Ã
1 +
¡
0
?¡
:
2
3t
(¸
0
?o) + .
!
(2.15)
This de?nes a quadratic equation in ¸
0
. The solution is where the right hand
side of (2.15) cuts the 45 degree line. The leveraged sector ampli?es booms and
busts if ¸
0
? ¸ has the same sign as ¡
0
? ¡. Then, any shift in fundamentals
16
gets ampli?ed by the portfolio decisions of the leveraged sector. The condition
for ampli?cation is that the denominator in the second term of (2.15) is positive.
But this condition is guaranteed from (2.14) and the fact that j
0
q
0
?:
1+i
(i.e. that
the price of the risky security is higher than its worst possible realized discounted
payo?).
Note also that the size of the ampli?cation is increasing when fundamental risk
is small, seen from the fact that ¸
0
?¸ is large when . is small. Recall that . is
the fundamental risk. When . is small, the associated Value-at-Risk is also small,
allowing the leveraged sector to maintain high leverage. The higher is the leverage,
the greater is the marked-to-market capital gains and losses. Ampli?cation is large
when the leveraged sector itself is large relative to the total economy. Finally,
note that the ampli?cation is more likely when the passive sector’s risk tolerance
t is high.
2.2. Pricing of Risk
We now explore the ?uctuations in risk pricing in our model. The risk premium in
our model is the excess expected return on the risky security, which can be written
in terms of the ratio of the discounted expected payo? on the risky security and
its price:
Risk premium =
¡
j (i + 1)
?1 (2.16)
Rather than working with the risk premium directly, it turns out to be more
convenient to work with a monotonic transformation of the risk premium de?ned
as
: ? 1 ?
j (1 + i)
¡
(2.17)
The “:” stands for “premium”. The variable : is a monotonic transformation
of the risk premium that varies from zero (when the risk premium is zero) to 1
17
(when the risk premium is in?nite).
The market-clearing condition for the risky security is ¸ + ¸
1
= o, which can
be written as
c
. ?¡:
+
3t
.
2
¡: = o (2.18)
Our primary interest is in the relationship between total equity c and the risk
premium :. Here, c has the interpretation of the total capital of the banking
sector, and hence its risk-taking capacity. In our model, the total lending of the
banking sector bears a very simple relationship to its total capital c, since the
holding of the risky security by the active investors (the banks) is c, (. ?¡
.We impose the restriction that the active investors have a strictly positive total
holding of the risky security, or equivalently that the passive sector’s holding is
strictly smaller than the total endowment o. From (2.5) this restriction can be
written as
¡: <
.
2
3t
o (2.19)
By de?ning 1 (c.
as below, we can write the market-clearing condition as:1 (c.
? c +3t
.
2
¡: (. ?¡
?o (. ?¡ = 0 (2.20)We then have
J1
J:
= ¡
?
?
?
3t
.
2
(. ?¡
| {z }
¹
+ o ?
3t
.
2
¡:
| {z }
1
?
?
?
(2.21)
Both ¹ and 1 are positive. ¹ is positive since the holding of the risky
security by the active sector is c, (. ?¡
, and so . ? ¡: 0 in order that theactive investors hold positive holdings of the risky security. Another way to view
this condition is to note that the market price of the risky security cannot be
18
lower than the lowest possible realization of the payo? of the risky asset, so that
j
¡ ?.
1 + i
(2.22)
which can be written as (1 + i) j ?(¡ ?.) = . ?¡: 0. The second term (term
1) inside the big brackets is positive from our condition (2.19) that the passive
investors do not hold the entire supply. Since J1,Jc = 1, we have
d:
dc
= ?
J1,Jc
J1,J:
< 0 (2.23)
In other words, the market risk premium is decreasing in the total equity c
of the banking sector. As stated above, we view c as the risk-taking capacity of
the banking sector. Any shock that increases the capital bu?er of the banking
sector will lower the risk premium. We therefore have the following empirical
hypothesis.
Empirical Hypothesis 1. Risk premiums fall when the equity of the banking
system increases.
This empirical hypothesis is key to our discussion on the role of short-term
interest rates on risk-taking capacity of the banking sector, through the slope of
the yield curve and hence the greater pro?tability of bank lending. We return to
this issue shortly.
2.3. Shadow Value of Bank Capital
Another window on the risk premium in the economy is through the Lagrange
multiplier associated with the constrained optimization problem of the banks,
which is to maximize the expected payo? from the portfolio 1 (\) subject to the
Value-at-Risk constraint. The Lagrange multiplier is the rate of increase of the
objective function with respect to a relaxation of the constraint, and hence can
19
be interpreted as the shadow value of bank capital. Denoting by ` the Lagrange
multiplier, we have
` =
d1 (\)
dc
=
J1 (\)
J¸
J¸
Jc
=
¡ ?(1 + i) j
. ?(¡ ?(1 + i) j)
(2.24)
where we have obtained the expression for d1 (\) ,d¸ from (2.1) and d¸,dc is
obtained from (2.8), which gives the optimal portfolio decision of the leveraged
investor. Using our : notation, we can re-write (2.24) as
` =
¡:
. ?¡:
(2.25)
We see from (2.25) that as the risk premium : becomes compressed, the Lagrange
multiplier ` declines. The implication is that the marginal increase of a dollar’s
worth of new capital for the leveraged investor is generating less expected payo?.
As the risk premium : goes to zero, so does the Lagrange multiplier, implying
that the return to a dollar’s worth of capital goes to zero.
Furthermore, the shadow value of bank capital can be written as:
` =
. (o ?¸)
3t ?. (o ?¸)
(2.26)
The shadow value of bank capital is decreasing in the size of the leveraged
sector, given by ¸. Moreover, since there is a one-to-one relationship between
` and the risk premium :, we can also conclude that market risk premiums fall
when the size of the intermediary sector increases.
Empirical Hypothesis 2. Risk premiums fall when the size of the banking
sector increases.
3. Changing Nature of Financial Intermediation
In preparation for our empirical investigations, we brie?y review the structure
of ?nancial intermediation in the United States. In particular, we highlight the
20
increasing importance of market-based ?nancial intermediaries and the shadow
banking system.
3.1. Shadow Banking System and Security Broker-Dealers
As recently as the early 1980s, traditional banks were the dominant ?nancial
intermediaries. In subsequent years, however, they were quickly overtaken by
market-based ?nancial institutions. Figure 3.1 plots the size of di?erent types
of ?nancial intermediaries for the United States starting in 1985. We see that
market-based ?nancial intermediaries, such as security broker dealers and ABS
issuers, have become important components of the intermediary sector. The
series labeled “shadow banks” aggregates ABS issuers, ?nance companies, and
funding corporations.
0
2
0
0
0
4
0
0
0
6
0
0
0
8
0
0
0
1
0
0
0
0
B
i
l
l
i
o
n
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Security Broker-Dealers ABS issuers
Shadow Banks Commercial Banks
Note: Shadow banksareABS issuers, financecompanies, and fundingcorporation
Source: Boardof Governors of theFederal Reserve
Figure 3.1: Total Assets of Commercial Banks, Shadow Banks, and Broker-
Dealers.
In 1985, shadow banks were a tiny fraction of the commercial bank sector, but
had caught up with the commercial bank sector by the eve of the crisis. The
21
increased importance of the market-based banking system has been mirrored by
the growth of the broker-dealers, who have traditionally played market-making
and underwriting roles in securities markets. However, their importance in the
supply of credit has increased dramatically in recent years with the growth of
securitization and the changing nature of the ?nancial system toward one based
on the capital market, rather than one based on the traditional role of the bank
as intermediating between depositors and borrowers. Although total assets of
the broker-dealer sector are smaller than total assets of the commercial banking
sector, our results suggest that broker-dealers provide a better barometer of the
funding conditions in the economy, capturing overall capital market conditions.
Perhaps the most important development in this regard has been the changing
nature of housing ?nance in the US. The stock of home mortgages in the US
is now dominated by the holdings of market-based institutions, rather than by
traditional bank balance sheets. Broker-dealer balance sheets provide a timely
window on this world.
The growth of market-based ?nancial intermediaries is also re?ected in the
aggregates on the liabilities side of the balance sheet. Figure 3.2 shows the
relative size of the M1 money stock together with the outstanding stock of repos
of the primary dealers - the set of banks that bid at US Treasury security auctions,
and for whom data are readily available due to their reporting obligations to the
Federal Reserve. We also note the rapid growth of ?nancial commercial paper as
a funding vehicle for ?nancial intermediaries.
Figure 3.3 charts the relative size of M2 (bank deposits plus money market fund
balances) compared to the sum of primary dealer repos and ?nancial commercial
paper outstanding. As recently as the 1990s, the M2 stock was many times larger
than the stock of repos and commercial paper. However, by the eve of the crisis,
the gap had narrowed considerably, and M2 was only some 25% larger than the
22
0
1
0
0
0
2
0
0
0
3
0
0
0
4
0
0
0
B
i
l
l
i
o
n
1990q1 1995q1 2000q1 2005q1 2010q1
Money Stock M1 Primary Dealer Repo
Financial Commercial Paper Outstanding
Sources: Boardof Governorsof theFederal Reserveand Federal ReserveBank of New York
Figure 3.2: Liquid funding of ?nancial institutions: Money (M1), Primary Dealer
Repo, and Commercial Paper.
stock of repos and ?nancial commercial paper. However, with the eruption of the
crisis, the gap has opened up again.
Not only have the market-based intermediaries seen the most rapid growth
in the run-up to the ?nancial crisis, they were also the institutions that saw the
sharpest pull-back in the crisis itself. Figure 3.4 shows the comparative growth
rate of the total assets of commercial banks (in red) and the shadow banks (in
blue), while Figure 3.5 shows the growth of commercial paper relative to shadow
bank asset growth. We see that, whereas the commercial banks have increased
the size of their balance sheet during the crisis, the shadow banks have contracted
substantially. Traditionally, banks have played the role of a bu?er against ?uctu-
ations in capital market conditions, and we see that they have continued their role
through the current crisis. As such, looking only at aggregate commercial bank
lending may give an overly rosy picture of the state of ?nancial intermediation.
23
0
2
0
0
0
4
0
0
0
6
0
0
0
8
0
0
0
B
i
l
l
i
o
n
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Money Stock M2
Primary Dealer Repo plus Commercial Paper Outstanding
Sources: Boardof Governorsof theFederal Reserveand Federal ReserveBank of New York
Figure 3.3: Short Term Funding: M2 versus Commercial Paper + Primary Dealer
Repo.
Figure 3.6 shows that the broker-dealer sector of the economy has contracted
in step with the contraction in primary dealer repos, suggesting the sensitivity
of the broker-dealer sector to overall capital market conditions. Therefore, in
empirical studies of ?nancial intermediary behavior, it is be important to bear in
mind the distinctions between commercial banks and market-based intermediaries
such as broker dealers. Market-based intermediaries who fund themselves through
short-term borrowing such as commercial paper or repurchase agreements will be
sensitively a?ected by capital market conditions. But for a commercial bank, its
large balance sheet masks the e?ects operating at the margin. Also, commercial
banks provide relationship-based lending through credit lines. Broker-dealers, in
contrast, give a much purer signal of marginal funding conditions, as their balance
sheets consist almost exclusively of short-term market borrowing and are not as
constrained by relationship-based lending.
24
0
5
1
0
1
5
C
o
m
m
e
r
c
i
a
l
B
a
n
k
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
-
1
0
0
1
0
2
0
3
0
4
0
S
h
a
d
o
w
B
a
n
k
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Shadow Bank Asset Growth (Annual %)
Commercial Bank Asset Growth (Annual %)
Note: Shadow banksareABS issuers, financecompanies, and fundingcorporation
Source: Boardof Governors of theFederal Reserve
Figure 3.4: Total Asset Growth of Shadow Banks and of Commercial Banks.
3.2. Haircuts and Value at Risk
The Value at Risk constraint that is at the heart of the ampli?cation mechanism in
the model of Section 2 characterizes market-based ?nancial intermediaries such as
security broker-dealers and shadow banks. The active balance sheet management
of ?nancial institutions is documented in Adrian and Shin (2007), who show that
investment banks exhibit ”procyclical leverage”: i.e., increases in balance sheet
size are associated with increases in leverage. In contrast, the balance sheet
behavior of commercial banks is consistent with leverage targeting: for commercial
banks, leverage growth is uncorrelated with the growth of balance sheet size.
One useful perspective on the matter is to consider the implicit maximum lever-
age that is permitted in collateralized borrowing transactions such as repurchase
agreements (repos). Repos are the primary source of funding for market-based
?nancial institutions, as well as being a marginal source of funding for traditional
banks. In a repurchase agreement, the borrower sells a security today for a price
25
-
4
0
-
2
0
0
2
0
4
0
C
o
m
m
e
r
c
i
a
l
P
a
p
e
r
O
u
t
s
t
a
n
d
i
n
g
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
-
1
0
0
1
0
2
0
S
h
a
d
o
w
B
a
n
k
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
1990q1 1995q1 2000q1 2005q1 2010q1
dateq
Shadow Bank Asset Growth (Annual %)
Commercial Paper Outstanding Growth (Annual %)
Sources: Boardof Governorsof theFederal Reserve
Figure 3.5: Marginal Funding of Shadow Banks is Commercial Paper.
below the current market price on the understanding that it will buy it back in
the future at a pre-agreed price. The di?erence between the current market price
of the security and the price at which it is sold is called the “haircut” in the repo,
and ?uctuates together with market conditions. The ?uctuations in the haircut
largely determine the degree of funding available to a leveraged institution. The
reason is that the haircut determines the maximum permissible leverage achieved
by the borrower. If the haircut is 2%, the borrower can borrow 98 dollars for 100
dollars worth of securities pledged. Then, to hold 100 dollars worth of securities,
the borrower must come up with 2 dollars of equity. Thus, if the repo haircut is
2%, the maximum permissible leverage (ratio of assets to equity) is 50.
Suppose that the borrower leverages up the maximum permitted level. The
borrower thus has a highly leveraged balance sheet with leverage of 50. If at this
time, a shock to the ?nancial system raises the market haircut, then the borrower
faces a predicament. Suppose that the haircut rises to 4%. Then, the permitted
26
-
4
0
-
2
0
0
2
0
4
0
P
r
i
m
a
r
y
D
e
a
l
e
r
R
e
p
o
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
-
4
0
-
2
0
0
2
0
4
0
S
e
c
u
r
i
t
y
B
r
o
k
e
r
-
D
e
a
l
e
r
A
s
s
e
t
G
r
o
w
t
h
(
A
n
n
u
a
l
%
)
1992q1 1996q3 2001q1 2005q3 2010q1
Security Broker-Dealer Asset Growth (Annual %)
Primary Dealer Repo Growth (Annual %)
Sources: Boardof Governorsof theFederal Reserveand Federal ReserveBank of New York
Figure 3.6: Marginal Funding of Broker-Dealers is Repo.
leverage halves to 25, from 50. The borrower then faces a hard choice. Either
it must raise new equity so that its equity doubles from its previous level, or it
must sell half its assets, or some combination of both. However, asset disposals
have spillover e?ects that exacerbate the distress for others. The “margin spiral”
described by Brunnermeier and Pedersen (2009) models this type of phenomenon.
Considerations of repo haircuts suggest that measured risks will play a piv-
otal role in the determination of leverage. Adrian and Shin (2008b) present a
contracting model which yields this outcome as a central prediction, and present
empirical evidence consistent with the prediction. Adrian and Shin (2008b) ?nd
that measures of Value-at-Risk (VaR) that are computed from the time series of
daily equity returns explains shifts in total assets, leverage, and key components
of the liabilities side of the balance sheet, such as the stock of repos.
In the benchmark case where losses are exponentially distributed, the con-
tracting model of Adrian and Shin (2008b) yields the widely used Value-at-Risk
27
(VaR) rule which stipulates that exposures be adjusted continuously so that eq-
uity exactly matches total Value-at-Risk. Among other things, the Value-at-Risk
rule implies that exposures are adjusted continuously so that the probability of
default is kept constant - at the level given by the VaR threshold. Given the
ubiquitous use of the VaR rule both by private sector ?nancial institutions and
by regulators, this microfoundation of the VaR concept gives a basis for further
study.
To be sure, showing that the VaR rule is the outcome of a contracting model
says little about the desirability of the widespread adoption of such practices from
the point of view of economic e?ciency. Indeed, there are strong arguments to
suggest that risk management tools such as Value-at-Risk cause spillover e?ects to
other ?nancial institutions that are detrimental to overall e?ciency. For instance,
the prudent reduction in exposures by the creditors to Lehman Brothers is a run
from the point of view of Lehman Brothers itself. The spillover e?ects are a
natural consequence of the contracts being bilateral arrangements. They do not
take account of the spillover e?ects across more than one step in the ?nancial
network.
3.3. Relative Size of the Financial Sector
The rapid growth of the market-based intermediaries masks the double-counting
involved when adding up balance sheet quantities across individual institutions.
Before going further, we therefore note some accounting relationships that can
help us to think about the extent of the double-counting.
Let c
i
be the total assets of bank i, and r
i
the total debt of bank i (i.e., total
liabilities minus equity). The total size of the banking sector in gross terms can
be written as the sum of all bank assets, given by
P
a
i=1
c
i
. A closely related
measure would be the aggregate value of all bank debt, given by
P
a
i=1
r
i
.
28
De?ne leverage `
i
as the ratio of total assets to equity of bank i:
`
i
=
c
i
c
i
?r
i
. (3.1)
Then, solving for r
i
and using the notation o
i
= 1 ?
1
A
?
, we have
r
i
= o
i
Ã
¸
i
+
X
)
r
)
:
)i
!
= o
i
¸
i
+
£
r
1
· · · r
a
¤
?
?
?
o
i
:
1i
.
.
.
o
i
:
ai
?
?
?
(3.2)
Let r =
£
r
1
· · · r
a
¤
, ¸ =
£
¸
1
· · · ¸
a
¤
, and de?ne the diagonal matrix ?
as follows.
? =
?
?
?
o
1
.
.
.
o
a
?
?
?
(3.3)
Then we can write (3.2) in vector form as:
r = ¸?+ r??
Solving for r,
r = ¸?(1 ???)
?1
= ¸?
¡
1 +??+ (??)
2
+ (??)
3
+ · · ·
¢
(3.4)
The matrix ?? is given by
?? =
?
?
?
?
?
0 o
2
:
12
· · · o
a
:
1a
o
1
:
21
0 o
a
:
2a
.
.
.
.
.
.
.
.
.
o
1
:
a1
o
2
:
a2
· · · 0
?
?
?
?
?
(3.5)
The in?nite series in (3.4) converges since the rows of ??sum to a number strictly
less than 1, so that the inverse (1 ???)
?1
is well-de?ned.
29
Equation (3.4) suggests what to look for when gauging the extent of double-
counting of lending to ultimate borrowers that results from the heavy use of fund-
ing raised from other ?nancial intermediaries. The comparison is between ¸ (the
pro?le of lending to the ultimate borrowers in the economy) and r (the pro?le of
debt values across all banks, which gives a gross measure of balance sheet size).
The factor that relates the two is the matrix:
?
¡
1 +??+ (??)
2
+ (??)
3
+ · · ·
¢
This matrix has a ?nite norm, since the in?nite series 1+??+(??)
2
+(??)
3
+· · ·
converges to (1 ???)
?1
. However, for a ?nancial system where leverage is high,
and to the extent to which banks are interwoven tightly, the norm can grow
without bound. This is because as leverage becomes large, o
i
? 1 and, hence,
? tends to the identity matrix. Moreover, as the degree of interconnectedness
between banks becomes large, the norm of the matrix ? converges to 1, since
each row of ? will sum to a number that converges to 1. In the limit, as ? ?1
and k?k ?1, the norm of the matrix ?
¡
1 +??+ (??)
2
+ (??)
3
+ · · ·
¢
grows
without bound.
Consequently, the size of the ?nancial intermediation sector relative to the
size of the economy can vary hugely over the ?nancial cycle. We illustrate this
phenomenon in Figures 3.7 and 3.8, which show the growth of four sectors in the
United States from 1954. The four sectors are (i) the non-?nancial corporate
sector, (ii) the household sector, (iii) the commercial banking sector, and (iv) the
security broker-dealer sector. The data are taken from the Federal Reserve’s Flow
of Funds accounts. The series are normalized so that the size in 1954Q1 is set
equal to 1.
Over this time period, three of the four sectors had grown to roughly 80 times
their 1954 size by the end of 2009. These trends, however, have been dwarfed by
30
0
100
200
300
400
500
600
700
800
900
1
9
5
4
Q
1
1
9
5
6
Q
3
1
9
5
9
Q
1
1
9
6
1
Q
3
1
9
6
4
Q
1
1
9
6
6
Q
3
1
9
6
9
Q
1
1
9
7
1
Q
3
1
9
7
4
Q
1
1
9
7
6
Q
3
1
9
7
9
Q
1
1
9
8
1
Q
3
1
9
8
4
Q
1
1
9
8
6
Q
3
1
9
8
9
Q
1
1
9
9
1
Q
3
1
9
9
4
Q
1
1
9
9
6
Q
3
1
9
9
9
Q
1
2
0
0
1
Q
3
2
0
0
4
Q
1
2
0
0
6
Q
3
Non-financial
corporate
Households
Security Broker
Dealers
Commercial
Banks
Figure 3.7: Growth of Four US Sectors (1954Q1 =1)
1980Q1
1
10
100
1000
1
9
5
4
Q
1
1
9
5
7
Q
1
1
9
6
0
Q
1
1
9
6
3
Q
1
1
9
6
6
Q
1
1
9
6
9
Q
1
1
9
7
2
Q
1
1
9
7
5
Q
1
1
9
7
8
Q
1
1
9
8
1
Q
1
1
9
8
4
Q
1
1
9
8
7
Q
1
1
9
9
0
Q
1
1
9
9
3
Q
1
1
9
9
6
Q
1
1
9
9
9
Q
1
2
0
0
2
Q
1
2
0
0
5
Q
1
2
0
0
8
Q
1
Non-financial
corporate
Households
Security Broker
Dealers
Commercial
Banks
Figure 3.8: Growth of Four US Sectors (1954Q1 = 1) (in log scale)
31
that of the broker-dealer sector, which had grown to around 800 times its 1954
level at the height of the boom, before collapsing in the recent crisis. Figure 3.8
is the same chart, but in log scale. The greater detail a?orded by the chart in log
scale reveals that the securities sector kept pace with the rest of the economy up
to about 1980, but then started a growth spurt that outstripped the other sectors.
On the eve of the crisis, the size of the securities sector was roughly ten times
that of the other sectors in the economy.
4. Empirical Relevance of Financial Intermediary Balance
Sheets
Our discussion thus far suggests that ?nancial intermediaries deserve independent
study in models of monetary economics due to their impact on ?nancial condi-
tions. Asset prices are in?uenced by the tightness of balance sheet constraints of
?nancial intermediaries. In this section, we examine empirically whether ?nan-
cial intermediaries’ impact on ?nancial conditions can feed through to a?ect real
economic outcomes. The analysis follows Adrian and Shin (2008d) and Adrian,
Moench and Shin (2009).
We label the looseness of balance sheet constraints “risk appetite” following
Adrian, Moench, and Shin (2009, 2010) and Adrian, Etula, and Shin (2009). Risk
appetite refers to the shadow value of capital of the (leveraged) intermediary sector
in the model of section 2. This shadow value of capital indicates the additional
pro?t that the banking sector may earn by having one dollar of extra bank capital.
The looser is the capital constraint, the lower is the Lagrange multiplier, and hence
the higher is the risk appetite.
The terminology of “risk appetite” is intended to highlight the apparent change
in preferences of the banking sector. We say “apparent” change in preferences,
since the ?uctuations in risk appetite are due to the constraints faced by the
32
banks rather than their preferences as such. However, to an outside observer, the
?uctuations in risk appetite would have the outward signs of ?uctuations in risk
preferences of the investor.
Adrian, Moench, and Shin (2009) estimate the risk appetite of ?nancial inter-
mediaries for a ”macro risk premium”. The risk premium measures the hurdle
rate of return for new projects that are ?nanced in the economy, and hence re?ects
the ease of credit conditions, and correponds to the risk premium of the model
in section 2. The macro risk premium is estimated from the yield spreads of
?xed income securities. In particular, the macro risk premium is estimated as a
linear combination of spreads that is tracking GDP growth most closely. In doing
so, we allow both term spreads of the Treasury yield curve and credit spreads to
enter. Both term spreads and credit spreads are measures of hurdle rates — the
additional yields on longer-dated or riskier bonds that induce market investors to
fund additional investment or consumption.
To estimate the macro risk premium, Adrian, Moench, and Shin (2009) con-
temporaneously regress GDP growth on a wide variety of Treasury and credit
spreads. They then use the seven constant maturity yields published in the H.15
release of the Federal Reserve Board and compute spreads relative to the Fed
Funds target, and corporate bond spreads for credit ratings AAA, AA, A, BBB,
BB, and B from Standard & Poors in excess of the 10-year constant maturity
Treasury yield. The analysis of Adrian, Moench, and Shin (2009) starts in the
?rst quarter of 1985, and ends in the fourth quarter of 2009.
An estimate of the macro risk premium is obtained as the ?tted value of a linear
regression of GDP growth on these corporate and Treasury bond spreads. Hence,
empirically, the macro risk premium is a weighted average of spreads where the
weights are given by the regression coe?cients. The weights can be interpreted
33
-
1
0
-
5
0
5
1
0
Q
u
a
r
t
e
r
l
y
G
D
P
g
r
o
w
t
h
1
1
.
5
2
2
.
5
3
M
a
c
r
o
R
i
s
k
P
r
e
m
i
u
m
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
Macro Risk Premium
Quarterly GDP growth
Figure 4.1: The Macro Risk Premium and GDP Growth
as portfolio weights of a portfolio that is tracking GDP growth. Conceptually,
the macro risk premium represents the analogue of the risk premium term :
discussed in Section 2 above. The estimated macro risk premium, together with
GDP growth, are plotted in Figure 4.1. For ease of interpretation, the macro risk
premium is rotated using an a?ne transformation so as to match the average level
and the volatility of the AA credit spread. The plot shows the strong negative
correlation between the macro risk premium and GDP growth.
We now turn to our measure of the looseness of ?nancial intermediary capital
constraints, which we have called “risk appetite” as shorthand. As sketched in
Section 2 , the willingness of banks to lend will be positively associated with
the size of intermediary balance sheets. The scenario outlined in Section 2 is
that ?nancial intermediaries manage their balance sheets actively by employing a
Value-at-Risk constraint when choosing the size and composition of their portfolio.
Intuitively, the relationship between the macro risk premium, ?nancial inter-
mediary balance sheets, and real activity goes as follows. When ?nancial interme-
34
diaries have ample balance sheet capacity in terms of higher capital, their balance
sheet constraints are loose, risk premia are compressed, the supply of credit is
plentiful, which in turn leads to lower threshold rates of return for real projects,
and hence higher GDP growth. E?ective risk aversion is low, and real growth is
high. Conversely, when ?nancial intermediary funding conditions worsen, their
risk appetite declines, leading to lower real growth.
The macro risk premium measure can be interpreted as a portfolio of Treasury
and corporate bond yields where the portfolio weights are chosen so as to maximize
the contemporaneous correlation with real GDP growth. In a similar vein, we
obtain a measure of intermediary risk appetite by ?nding the linear combination
of one-year lagged balance sheet variables that best predict the (negative) one-
year change of the macro risk premium. To a good approximation, the negative
change re?ects the return on the macro risk premium. A priori, it is not clear
which institutions are the most important ones in determining risk premia for
the di?erent asset classes. We therefore build on our related work in Adrian,
Moench, and Shin (2010) in which we use subset selection methods to identify
the best predictors for excess returns on di?erent asset classes among a large
number of potential explanatory balance sheet proxies for various types of ?nancial
institutions. Adrian, Moench, and Shin (2010) document that annual leverage
growth of security brokers and dealers is a strong predictor of excess returns on
equity and corporate bond portfolios and that quarterly shadow bank asset growth
is a strong forecasting variable for excess returns on corporate and Treasury bonds.
Adrian, Moench, and Shin (2009) therefore restrict the set of right-hand side
variables to these two types of institutions, complemented by commercial banks
so as to highlight the di?erential impact that balance sheets of market based
intermediaries have with respect to traditional banks. For each of these three
types of institutions, one-year lagged asset growth and the growth of net worth
35
are included as potential variables. Adrian, Moench, and Shin (2009) weight the
growth rates of assets and net worth by the relative size of total assets of each
intermediary in order to capture the trends of assets under management across
di?erent institutions.
Risk appetite is estimated by regressing the (negative) change of the macro
risk premium over one year on the balance sheet measures of the security broker-
dealers, the shadow banks, and the commercial banks. These negative changes
capture returns to the risk premia. When the negative change of a risk premium
is positive, prices today increase, leading to lower expected returns. The measure
of risk appetite is constructed as the ?tted value of the regressions of (the negative
changes to the macro risk premium) on the intermediary balance sheet variables.
The risk appetite measure is displayed together with the macro risk premium in
Figure 4.2. The plot shows that risk appetite is highly negatively correlated with
changes to the macro risk premium. Higher risk appetite leads to balance sheet
expansions, which are associated with increases in asset prices and hence declines
in spreads. Movements in risk appetite are thus strongly negatively correlated
with the macro risk premium.
We can also investigate the importance of ?nancial intermediary balance sheets
for macroeconomic activity by relating intermediary balance sheets directly to
GDP growth. Relative to commercial banks, broker-dealer and shadow bank
balance sheets potentially hold more information related to underlying ?nancial
conditions, as they are a signal of the marginal availability of credit. At the
margin, all ?nancial intermediaries (including commercial banks) have to borrow
in capital markets (for instance via commercial paper or repos). For commercial
banks, however, their large balance sheets mask the e?ects operating at the mar-
gin. Broker-dealers or shadow banks, in contrast, give a purer signal of marginal
36
-
.
8
-
.
6
-
.
4
-
.
2
0
.
2
I
n
t
e
r
m
e
d
i
a
r
y
R
i
s
k
A
p
p
e
t
i
t
e
1
1
.
5
2
2
.
5
3
M
a
c
r
o
R
i
s
k
P
r
e
m
i
u
m
1985q1 1990q1 1995q1 2000q1 2005q1 2010q1
Macro Risk Premium
Intermediary Risk Appetite
Figure 4.2: Macro Risk Premium and Intermediary Risk Appetite
funding conditions, as (i) their liabilities are short term and (ii) their balance
sheets are closer to being fully marked to market.
In addition, broker-dealers originate and serve as market makers for securitized
products, whose availability determines the credit supply for consumers and non-
?nancial ?rms (e.g., for mortgages, car loans, student loans, etc.). Consequently,
broker-dealers are important variables not only because they are the marginal
suppliers of credit, but also because their balance sheets re?ect the ?nancing
constraints of the market-based ?nancial system.
To the extent that balance sheet dynamics a?ect the supply of credit, they
have the potential to a?ect real economic variables. To examine this empirically,
we estimate macroeconomic forecasting regressions. In Table 4.1, we report the
results of regressions of the quarterly growth rate of GDP components on lagged
macroeconomic and ?nancial variables. In addition, we add the lagged growth
rate of total assets and market equity of security broker-dealers. By adding lags
of additional ?nancial variables (equity market return, equity market volatility,
37
term spread, credit spread), we o?set balance sheet movements that are purely
due to a price e?ect. By adding the lagged macroeconomic variables, we control
for balance sheet movements due to past macroeconomic conditions. In Table
4.1, (and all subsequent tables), * denotes statistical signi?cance at the 10%, **
signi?cance at the 5% level, and *** at the 1% level. All our empirical analysis
is done using quarterly data from 1986Q1 to 2009Q2. Variable de?nitions are
given in the data appendix at the end of this chapter.
-.2
0
.2
.4
0 5 10 15
ImpulseResponse Function of GDP Growth to Shadow Bank Asset Growth Shock
Quarters
Figure 4.3: The ?gure plots the impulse response GDP growth from a shock to
shadow bank total asset growth. The impulse response is estimated from a vector
autoregression with gdp growth, pce in?ation, shadow bank asset growth, credit
spread, vix, the term spread, and the Federal Fund target rate as variables, and
that ordering is used to produce the impulse response functions using a Cholesky
decomposition. The time span is 1986Q1 to 2009Q1.
The growth rate of security broker-dealer total assets has strongest signi?cance
for the growth rate of future housing investment, and weak signi?cance for total
GDP growth (Tables 4.1 and 4.2, columns 1). Our interpretation of this ?nding is
that the mechanisms that determine the liquidity and leverage of broker-dealers
38
Table 4.1: Impact of Balance Sheets on GDP. This table reports regressions
of quarterly GDP growth on the total asset and equity growth of broker-dealers,
shadow banks, and commercial banks for 1986Q1 to 2009Q2. *** denotes sig-
ni?cance at the 1% level, ** denotes signi?cance at the 5% level, and * denotes
signi?cance at the 10% level. Signi?cance is computed from robust standard er-
rors.
(1) (2) (3)
Quarterly Quarterly Quarterly
GDP GDP GDP
Growth Growth Growth
Broker-Dealer Asset Growth (lag) 0.03*
Broker-Dealer Equity Growth (lag) 0.18
Shadow Banks Asset Growth (lag) 0.21***
Shadow Banks Equity Growth (lag) 0.71**
Commercial Bank Asset Growth (lag) 0.02
Commercial Bank Equity Growth (lag) -0.12
GDP Growth (lag) 0.03 -0.18 0.09
PCE In?ation (lag) -1.01** -1.00** -1.16***
VIX (lag) 0.01 -0.03 -0.02
Credit Spread (lag) -1.37* -1.81** -1.01
Term spread (lag) 0.75** 1.18*** 0.75*
Fed Funds (lag) 0.40 0.19 0.49*
Constant 4.67*** 4.94*** 4.44**
Observations 93 93 93
1
2
0.288 0.409 0.263
39
a?ect the supply of credit, which, in turn, a?ect investment and consumption.
The total assets and total equity of shadow banks have signi?cant forecasting
power for total GDP growth, re?ecting their increased role in the total supply
of credit for the US economy. Commercial banks, on the other hand, have no
forecasting power for GDP, and forecast housing growth with the wrong sign.
Adrian, Moench, and Shin (2010) systematically investigate the forecasting power
of all ?nancial intermediaries in the U.S. Flow of Funds, and con?rm that broker-
dealers and shadow banks forecast real activity.
-.1
0
.1
.2
.3
0 5 10 15
ImpulseResponseFunction of Residential Investment Growthto Dealer Asset Growth
Quarters
Figure 4.4: The ?gure plots the impulse response of residential investment growth
from a shock to broker-dealer total asset growth. The impulse response is esti-
mated from a vector autoregression with residential investment growth, pce in-
?ation, broker-dealer asset growth, credit spread, vix, the term spread, and the
Federal Fund target rate as variables, and that ordering is used to produce the
impulse response functions using a Cholesky decomposition. The time span is
1986Q1 to 2009Q1.
The forecasting power of dealer assets for housing investment is graphically
illustrated in Figure 4.4. The impulse response function is computed from a ?rst
40
Table 4.2: Impact of Balance Sheets on Housing Investment. This table
reports regressions of quarterly residential investment growth on the total asset
and equity growth of broker-dealers, shadow banks, and commercial banks for
1986Q1 to 2009Q2. *** denotes signi?cance at the 1% level, ** denotes signi?-
cance at the 5% level, and * denotes signi?cance at the 10% level. Signi?cance is
computed from robust standard errors.
(1) (2) (3)
Quarterly Quarterly Quarterly
Housing Housing Housing
Growth Growth Growth
Broker-Dealer Asset Growth (lag) 0.09***
Broker-Dealer Equity Growth (lag) 0.10
Shadow Banks Asset Growth (lag) -0.00
Shadow Banks Equity Growth (lag) 0.14
Commercial Bank Asset Growth (lag) -0.44*
Commercial Bank Equity Growth (lag) 0.23
Housing Growth (lag) 0.89*** 0.94*** 0.96***
PCE In?ation (lag) -0.30 -0.11 -0.09
VIX (lag) 0.11 0.01 0.03
Credit Spread (lag) -0.92 -0.49 0.01
Term spread (lag) 1.11** 0.60 -0.07
Fed Funds (lag) -0.06 -0.04 -0.27
Constant -2.53 0.13 3.76
Observations 93 93 93
1
2
0.902 0.881 0.888
41
order vector autoregression that includes all variables of Table 4.1, Column (1).
The plot shows that the response of housing investment to a positive shock in
broker-dealer assets growth is positive, large, and persistent.
The di?erences between the interactions of market based intermediaries and
commercial banks with the macroeconomic aggregates are further highlighted in
column (iii) of Table 4.1, where we see that commercial bank assets and equity
do less well than shadow bank or security-broker-dealer variables as forecasting
variables. Our interpretation of these ?ndings is that commercial bank balance
sheets are less informative than broker-dealer balance sheets as they (largely) did
not mark their balance sheets to market over the time span in our regressions.
In addition, in Table 4.2, we ?nd that growth in commercial bank total assets
precedes declines in housing investment. This result is primarily due to the
fact that commercial banks o?er credit line pre-commitments that tend to be
drawn in times of crisis. In fact, in Figure 3.4, we saw that commercial bank
total assets grew at the onset of the recent ?nancial crisis, as structured credit
was reintermediated onto commercial bank balance sheets. The ?nding that
commercial bank assets do not predict future real growth is also consistent with
Bernanke and Lown (1991), who use a cross-sectional approach to show that
credit losses in the late 80’s and early 90’s did not have a signi?cant impact on
real economic growth across states. See Kashyap and Stein (1994) for an overview
of the debate on whether there was a “credit crunch” in the recession in the early
1990s.
In the same vein, Ashcraft (2006) ?nds small e?ects of variations in commercial
bank loans on real activity when using accounting based loan data. However,
Ashcraft (2005) ?nds large and persistent e?ects of commercial bank closures on
real output (using FDIC induced failures as instruments). Morgan and Lown
(2006) show that the senior loan o?cer survey provides signi?cant explanatory
42
power for real activity — again, a variable that (i) is more likely to re?ect underlying
credit supply conditions and (ii) is not based on accounting data.
The credit supply channel sketched so far di?ers from the ?nancial ampli?ca-
tion mechanisms of Bernanke and Gertler (1989), and Kiyotaki and Moore (1997,
2005). These papers focus on ampli?cation due to ?nancing frictions in the bor-
rowing sector, while we focus on ampli?cation due to ?nancing frictions in the
lending sector. We return to a more thorough review of the literature in a later
section.
5. Central Bank as Lender of Last Resort
The classical role of the central bank as the lender of last resort (LOLR) is framed
in terms of meeting panics that a?ect solvent, but illiquid, banks. In the simplest
case, bank runs arise when depositors fail to achieve coordination in a situation
with multiple equilibria. For example, in Bryant (1980) and Diamond and Dybvig
(1983), an individual depositor runs for fear that others will run, leaving no assets
in place for those who do not run.
However, in the ?nancial crisis of 2007-2009, the withdrawal of credit was not
restricted to one of even a subset of institutions. Instead, entire market sectors
were targeted. Figure 5.1 plots the new issuance of asset backed securities (ABS)
over a three month interval preceding the measurement date, and clearly illustrates
the generalized contraction of credit. If there was a run driven by a coordination
failure, it was a simultaneous run from all institutions in the ?nancial system.
Albeit, the extent to which each institution su?ered from the run depended on its
particular vulnerability. In the model outlined in Section 2, it is the interaction
between measured risks and the risk-bearing capacity of banks that determines
overall lending. Financial institutions that rely on value at risk cut back lending
when risk constraints become more binding (i.e., when the Lagrange multiplier
43
0
50
100
150
200
250
300
350
M
a
r
-
0
0
S
e
p
-
0
0
M
a
r
-
0
1
S
e
p
-
0
1
M
a
r
-
0
2
S
e
p
-
0
2
M
a
r
-
0
3
S
e
p
-
0
3
M
a
r
-
0
4
S
e
p
-
0
4
M
a
r
-
0
5
S
e
p
-
0
5
M
a
r
-
0
6
S
e
p
-
0
6
M
a
r
-
0
7
S
e
p
-
0
7
M
a
r
-
0
8
S
e
p
-
0
8
$
B
i
l
l
i
o
n
s
Other
Non-U.S. Residential
Mortgages
Student Loans
Credit Cards
Autos
Commercial Real
Estate
Home Equity
(Subprime)
Figure 5.1: New Issuance of ABS in Previous Three Months
associated with the constraint increases). From the point of view of an individual
bank, this prudent cutting of exposures by creditors will closely resemble a ”run.”
In this sense, the runs on Northern Rock, Bear Stearns, and Lehman Brothers
may be better seen as the tightening of constraints on the creditors of these banks,
rather than as a coordination failure among them.
Of course, we should not draw too ?ne a distinction between the coordination
view of bank runs on the one hand, and the “leverage constraints” view on the
other. Coordination (or lack thereof) will clearly exacerbate the severity of any
run when a bank has many creditors. The point is, rather, that an explanation
of a run on the system needs to appeal to more than just coordination failures.
For example, this means that explanations of the runs on Bear Stearns or Lehman
Brothers, should make reference to market-wide factors, as well as to the particular
characteristics of those ?rms and their creditors. This is one more instance of
the general maxim that, in a modern market-based ?nancial system, banking and
capital market conditions cannot be viewed in isolation.
44
Mar-09
Sep-08
Oct-08
Nov-08
0%
2%
4%
6%
8%
10%
12%
D
e
c
-
8
1
D
e
c
-
8
3
D
e
c
-
8
5
D
e
c
-
8
7
D
e
c
-
8
9
D
e
c
-
9
1
D
e
c
-
9
3
D
e
c
-
9
5
D
e
c
-
9
7
D
e
c
-
9
9
D
e
c
-
0
1
D
e
c
-
0
3
D
e
c
-
0
5
D
e
c
-
0
7
Figure 5.2: Cash as Proportion of US Commercial Bank Assets (Source: Federal
Reserve, H8 database).
To the extent that the credit crunch can be seen as the consequence of a
collapse of balance sheet capacity in the ?nancial intermediary sector, we can
interpret the policy response by central banks as an attempt to restore this lost
capacity by lending directly into the market. The Federal Reserve has been one
of the most aggressive central banks in this context, e?ectively interposing its own
balance sheet between the banking sector and the ultimate borrowers. The Fed
has taken in deposits from the banking sector (through increased reserves) and
then lent out the proceeds to ultimate borrowers through the holding of securities
(Treasuries, mortgage backed securities, commercial paper and other private sector
liabilities), as well as through currency swap lines to foreign central banks. One
indication of the dramatic increases in the Fed’s balance sheet can be seen in the
sharp increase in the holding of cash by US commercial banks, as seen from Figure
5.2. The increased cash holdings re?ect the sharp increase in reserves held at the
Fed — a liability of the Fed to the commercial banks.
In this way, central bank liquidity facilities have countered the shrinking of in-
45
termediary balance sheets and have become a key plank of policy, especially after
short-term interest rates were pushed close to their zero bound. The management
of the increased Federal Reserve balance sheet has been facilitated by the intro-
duction of interest on reserves on October 1, 2008, which e?ectively separates the
management of balance sheet size from that of the Federal Funds interest rate (see
Keister and McAndrews (2009) for a discussion of the interest on reserve payment
on the Federal Reserve’s balance sheet management).
The Federal Reserve has also put in place various other lender of last resort
programs in order to cushion the strains on balance sheets, and to thereby target
the unusually wide spreads in a variety of credit markets. Liquidity facilities
have been aimed at the repo market (TSLF and PDCF), the commercial paper
market (CPFF and AMLF), and ABS markets (TALF). In addition, the Federal
Reserve has conducted outright purchases of Treasury and agency securities, and
has provided dollar liquidity in the FX futures markets (FX Swap lines). The
common motivating element in these policies has been to try and alleviate the
strains associated with the shrinking balance sheets of intermediaries by substi-
tuting the central bank’s own balance sheet. The spirit of these policies di?ers
from that of classic monetary policy in that they are explicitly aimed at replacing
the collapse of private sector balance sheet capacity. Since the deleveraging of
?nancial intermediary balance sheets is associated with a widening of risk premia,
the e?ectiveness of balance sheet policies can be judged by the level of risk premia
in various ?nancial markets. In practice, the degree to which risk premia are
associated with the expansions and contractions of intermediary balance sheets
are important indicators for the risk appetite of the ?nancial sector, which, in
turn, a?ect credit supply and real activity. Adrian, Moench and Shin (2010)
use this insight to decompose the risk premia of several asset classes into var-
ious components, including those associated with the risk appetite of ?nancial
46
intermediaries.
One instance of the Fed’s liquidity facilities can be seen in Figure 5.3 (taken
from Adrian, Marchioni, and Kimbrough (2010), which charts total outstanding
commercial paper alongside net Federal Reserve commercial paper holdings in the
”Commercial Paper Funding Facility” (CPFF).
6
Following the Lehman Brothers
bankruptcy in September 2008, the outstanding amount of commercial paper be-
gan to fall precipitously, as can be seen by the sharp downward shift in the shaded
blue area. With the creation of the CPFF in October 2008, the Fed’s holdings of
commercial paper in the CPFF began to increase rapidly, as shown by the green
area in Figure 5.3. The Fed’s holdings can be seen to replace virtually dollar-
for-dollar the decline in the outstanding amount of commercial paper. In this
respect, the Fed’s balance sheet was being used to directly replace the decline in
balance sheet capacity of the ?nancial intermediary sector. An important feature
of the CPFF is that, as the facility was intended to be only a temporary liquidity
backstop, it was designed to become more unattractive as market conditions begin
to normalize. Accordingly, while the red line in Figure 5.3 shows that the Federal
Reserve held as much as 20% of all outstanding commercial paper at the height
of the crisis, CPFF holdings have since fallen steadily as market functioning has
improved.
As another example, Figure 5.1 illustrated how new issuance of asset-backed
securities (ABS) had collapsed by the end of 2008. In response, the Fed instituted
the Term Asset-Backed Loan Facility (TALF), whereby the central bank provides
secured loans to new AAA-rated ABS at a low haircut to private sector investors.
TALF was designed speci?cally to revitalize the ABS market, and its e?ectiveness
can be gauged by Figure 5.4, which shows the e?ect on new issuance of ABS
before and after the introduction of TALF. The light colored bars on the right
6
See Adrian, Marchioni, and Kimbrough (2010) for a detailed description of the CPFF.
47
0%
10%
20%
30%
40%
50%
0
400
800
1200
1600
2000
Aug-08 Dec-08 Apr-09 Aug-09
Billions of Dollars Percent
Source: Federal Reserve Board of Governors
Total Outstanding
in Market
Total Commercial Paper Outstanding
Total Outstanding
in CPFF
CPFF as %
of Market
CPFF Launch
Figure 5.3: The Federal Reserve’s Commercial Paper Funding Facility
0
5
10
15
20
25
30
J -08 F-08 M-08 A-08 M-08 J -08 J -08 A-08 S-08 O-08 N-08 D-08 J -09 F-09 M-09 A-09 M-09
$Bln
TALF-Financed Not TALF-Financed
ABS Include autos, credit cards, student loans, SBA, equipment, floorplans, and fleet leases
Figure 5.4: New ABS Issuance under TALF
48
show that much of the recent issuance of ABS has been TALF-?nanced, and that
TALF-backed issuance dwarfs the issuance of standard issues.
The balance sheet expansion of the Federal Reserve in response to the ?nan-
cial crisis of 2007-2009 has refocused the monetary policy debate on the role of
quantities in the monetary policy transmission mechanism. The crisis force-
fully demonstrated that the collapse of balance sheet capacity in the ?nancial
sector can have powerful adverse a?ects on the real economy, and, accordingly,
the traditional role of the central bank as the lender of last resort has undergone
far-reaching innovations.
6. Role of Short-Term Interest Rates
Having established that increases in broker-dealer and shadow bank balance sheets
signal increases in real activity, we now investigate the determinants of balance
sheet growth itself. Broker-dealers, shadow banks, and commercial banks fund
themselves with short-term debt. Broker-dealers are primarily funded in the repo
market (see Figure 3.6); shadow banks are primarily funded in the commercial
paper market (see Figure 3.5); and the majority of commercial banks’ short-term
funding is through money (i.e., checking and savings deposits). In the case of
broker-dealers, part of the repo funding is directly passed on to other leveraged
institutions, such as hedge funds, in the form of reverse repos, while another part
is invested in longer-term, less liquid securities. Shadow banks, on the other hand,
tend to fund holdings of ABS and MBS directly. Commercial banks primarily
hold non-tradable loans.
Because the majority of the liability side of ?nancial institutions comes from
short-term borrowing arrangements, their cost of borrowing is tightly linked to
short-term interest rates, such as the Federal Funds target rate. As broker-dealers
and shadow banks hold longer term assets, proxies for their expected returns are
49
spreads–particularly term spreads–which capture the maturity transformation
of ?nancial institutions. The leverage of these intermediaries is constrained by
risk: in more volatile markets, leverage is more risky, margins and haircuts are
higher, and credit supply tends to be more constrained. We saw in Section 2 how
Value-at-Risk determined balance sheet size, risk premia, and credit supply.
Much of these balance sheet adjustments occur at high frequencies. Though
the total assets used in the previous regressions are available only at a quarterly
frequency, on the liability side of the balance sheet, there are weekly data available
on outstanding repo, outstanding commercial paper, and total money. We use
repo data that are collected for the primary dealer universe by the Federal Reserve
Bank of New York. Outstanding commercial paper is collected by the depository
trust corporation (DTC), and is published at a weekly frequency by the Federal
Reserve Board. The broad money measure M2 is also made available by the
Federal Reserve Board.
We ?nd that increases in the Fed Funds target rate are generally associated
with a slower growth rate of short-term liabilities. In Table 6.1, we show regres-
sions of growth rates of repo, repo + commercial paper, and M2 on changes of the
Fed Funds target as well as on other asset prices (and lags of the left hand side
variables). The three types of regressions correspond to the funding of the three
main ?nancial institutions: broker-dealers, shadow banks, and commercial banks.
In each case, increases in the Fed Funds target are associated with declines in the
short term funding liabilities.
Financial market volatility, as measured by the VIX index of implied equity
volatility, relates negatively to security repo growth and repo+cp growth. As
higher volatility is associated with higher haircuts and tighter capital constraints,
both induce tighter constraints on dealer leverage (columns (1) and (2)). For
M2, we ?nd that higher VIX is associated with larger money growth, which we
50
Table 6.1: Determinants of Balance Sheet Growth. This table reports re-
gressions of repo growth, repo + commercial paper growth, and M2 growth on
their own lags, and asset price variables. The data frequency is weekly from Oc-
tober 3, 1990 to February 3, 2010. Changes refer to 1-week changes, and lags to 1
week lags. *** denotes signi?cance at the 1% level, ** denotes signi?cance at the
5% level, and * denotes signi?cance at the 10% level. Signi?cance is computed
from robust standard errors.
(1) (2) (3)
Repo Repo+CP Growth M2 Growth
(weekly growth) (weekly growth) (weekly growth)
Fed Funds (1 week change) -0.630*** -0.355*** -0.054***
Equity Return (1 week) -0.022* -0.013* 0.001**
VIX (1 week change) -0.052 -0.027 0.001
Treasury spread (1 week change) 0.703 0.291 0.151**
Credit spread (1 week change) 0.311 0.031 0.337**
Repo Growth (1 week lag) -0.134*** -0.075*** -0.001
CP Growth (1 week lag) 0.022 0.028 -0.020
M2 Growth (1 week lag) 0.515 0.063 -0.016
Constant 0.136* 0.105** 0.050***
Observations 990 990 989
1
2
0.042 0.032 0.121
interpret as ?ight to quality: in times of crisis, households and non-?nancial cor-
porations tend to reallocate short-term savings to commercial banks (see Gatev,
Schuermann, and Strahan, 2009).
Increases in the term spread are associated with higher repo growth. This
?nding is consistent with the notion that ?nancial intermediaries fund themselves
with short-term debt, but lend out longer term, so that a higher term spread
increases the carry between assets and liabilities and is associated with larger
balance sheets.
51
6.1. The Risk-Taking Channel of Monetary Policy
Current models in monetary economics emphasize the importance of managing
market expectations. By charting a path for future short rates and communicat-
ing this path clearly to the market, the central bank can in?uence long rates and
thereby in?uence mortgage rates, corporate lending rates and other prices that
a?ect consumption and investment. In constrast, our ?ndings point to the short-
term interest rate as an important price variable in its own right. Empirically,
we have seen that the Fed Funds rate is an important explanatory variable for
the growth of balance sheet aggregates. Our model suggests that increasing bank
capital increases the risk-taking capacity of the banking system, which in turn
leads to a lower equilibrium risk premium, and an increase the supply of credit
by lowering the hurdle rate at which projects are ?nanced.
Banks and other ?nancial intermediaries borrow in order to lend. Since the
loans o?ered by banks tend to be of longer maturity than the liabilities that fund
those loans, the term spread is indicative of the marginal pro?tability of an extra
dollar of loans on intermediaries’ balance sheets. The net interest margin (NIM)
of the bank is the di?erence between the total interest income on the asset side
of its balance sheet and the interest expense on the liabilities side of its balance
sheet. Whereas the term spread indicates the pro?tability of the marginal loan
that is added to the balance sheet, the net interest margin is an average concept
that applies to the stock of all loans and liabilities on the balance sheet.
The net interest margin determines the pro?tability of bank lending and in-
creases the present value of bank income, thereby boosting the forward-looking
measures of bank capital. Such a boost in bank capital increases the capacity
of the bank to increase lending in the sense that the marginal loan that was not
made before the boost in bank capital now becomes feasible under the greater
risk-bearing capacity of the bank. As banks expand their balance sheets, the
52
Initial
balance sheet
Fall in Fed Funds
Final
balance sheet
debt
equity
assets
increase
in equity
equity
assets
debt
assets
increasein
valueof
assets
equity
debt
new
borrowing
new
lending
Figure 6.1: The impact of a decline in the Fed Funds rate on intermediary balance
sheets.
market price of risk falls.
The logic of the argument is illustrated in Figure 6.1 for the case of monetary
easing. A decline in the Fed Funds target leads to an increase in asset prices and,
thus, an increase in the net worth of ?nancial institutions. In response, levered
?nancial institutions expand their assets by taking on additional leverage. Thus,
the impact of changes in short term interest rates is ampli?ed via the balance
sheet management of ?nancial institutions.
In order for the argument to go through a key assumption is that the term
spread is determined in large part by the short-term interest rate. On this score,
the evidence is supportive. Figure (1.1) in the introductory section to this chap-
ter highlights the striking fact that there is a near perfect negative one-to-one
relationship between 4-quarter changes of the Fed Funds target and 4-quarter
changes of the term spread (the plot uses data from 1987q1 to 2008q3). In this
way, variations in the target rate has a one-for-one relationship with the slope
of the yield curve. Since the term spread leads the net interest margin (NIM),
shifts in the short rate a?ect real activity because they change the pro?tability of
53
?nancial intermediaries, thus shifting the supply of credit.
The connection between ?nancial intermediary balance sheet management,
the slope of the yield curve, and real economic activity in the United States was
recently examined by Adrian, Estrella and Shin (2010), who conducted a vector
autoregression study of the relationship between the following variables: quarterly
GDP growth as a measure of real activity, the 10-year/3-month term Treasury
spread, the net interest margin (NIM) of large commercial banks from their Y-9C
?lings, the quarterly asset growth of shadow banks, the 3-month Treasury yield
as a measure of the short term interest rate, and the quarterly change in the
Chicago Board Options Exchange Volatility Index (VIX) as a measure of risk.
The balance sheet aggregate of the intermediary sector is taken to be the total
assets of shadow banks de?ned as the sum of total assets of asset-backed securities
issuers (ABS), ?nance companies, and funding corporations (each component is
pulled from the Federal Reserve’s Flow of Funds). The VAR includes one lag of
each of the variables, as suggested by the Bayesian Information Criterion, and
is estimated over the period from 1990Q3 to 2008Q3, where the starting date is
determined by availability of the VIX data.
Adrian, Estrella and Shin (2010) present empirical results that are consistent
with the following logic. An increase in the term spread tends to increase net
interest margin. This is fairly mechanical as the term spread directly impacts net
interest margin for newly originated loans funded with shorter-term liabilities.
Higher net interest margin – a major source of pro?ts for ?nancial intermedi-
aries – leads to an increase in total assets of ?nancial intermediaries: as lending
becomes more pro?table, the supply of credit is expanded and intermediaries’ bal-
ance sheets grow. Larger asset growth of intermediaries, in turn, predicts higher
GDP growth, which we interpret as a shift in the supply of credit curve. Since the
VAR includes only one lag of each variable, the signi?cance levels of the coe?-
54
cients may also be interpreted as a set of Granger causality tests. These tests are
consistent with our hypothesis of a causal chain that runs from the term spread
to net interest margin to lending volume and ?nally to real growth.
Adrian, Estrella and Shin (2010) also conduct impulse response studies so as
to verify the main strands in the narrative. In the impulse response studies, a
positive shock to the term spread leads to statistically signi?cant increases in net
interest margin over a considerable horizon. The shape of these responses is also
consistent with the fact that average net interest margin tends to trail marginal
changes in the term spread, as argued before. Moreover, a positive shock to net
interest margin tends to increased lending by the shadow banking sector. Finally,
a shock to asset growth in shadow banking has a quick and signi?cant e?ect on
real economic growth.
The evidence is supportive of the “risk-taking channel” of monetary policy.
Variations in short term interest rates lead real economic outcomes through their
impact on the slope of the yield curve. Our interpretation of this evidence is an
economic mechanism that operates via the balance sheet management of ?nan-
cial intermediaries, who borrow short and lend long. Tighter policy leads to a
compression of net interest margin and causes intermediaries to reduce lending.
The ?atter the term spread at the end of the tightening cycle, the greater the
subsequent reduction in lending activity. This has a direct e?ect on the supply of
credit to the real economy.
6.2. Two Case Studies
Two recent empirical papers throw further light on the channel of monetary policy
that works through changes in the market value of existing loans. Jimenez,
Ongena, Peydro and Saurina (2008) examine a large database of European loans
through the detailed information contained in the loan register and show that a
55
lower short-term interest rate lowers the hazard rate of default on existing loans.
In addition, they show that the hazard rate of default for new loans increases after
the cut in short-term rates.
The fact that the riskiness of existing loans decline may be due to a fall in
the interest burden of the borrower. The increased credit quality of the assets
will give rise to an increase in the pro?tability of the lending, and lead to greater
lending capacity, as outlined already. However, it is the second ?nding which is
more telling. The fact that the riskiness of new lending increases suggests that
the new loans are of lower quality, suggesting that the hurdle rate for lending
has fallen. Such a combination of (i) greater lending capacity and (ii) erosion of
lending standards is consistent with the risk-taking channel of monetary policy.
The same combination of (i) a lowering of a hazard rate of default on existing
loans and (ii) an increase in the hazard rate of default on new loans is also observed
in Ioannidou, Ongena and Peydro (2009). In this study, the authors examine the
e?ect of shifts of the US Fed Funds rate on the quality of bank loans in Bolivia,
which had a banking system which was close to being dollarized. To the extent
that the US Fed Funds rate is determined independently of the events in Bolivia,
the authors regard the e?ect of short-term interest rate changes as being a quasi-
natural experiment of the e?ect of short-term interest rate movements on bank
asset quality. As with the paper by Jimenez et al. (2008), the Bolivian study
reveals the same combination whereby a cut in the US Fed Funds rate leads to an
improvement in the quality of existing assets, but new assets are of a lower quality.
Paravisini (2008) provides estimates of the impact of bank funding constraints on
the supply of bank credit using an instrumental variable approach, and Khwaja
and Mian (2008) provide estimates of a bank funding shocks for an emerging mar-
ket. Freixas (2009) provides an overview of the monetary transmission literature
in the context of the 2007-2009 global ?nancial crisis.
56
This combination of results on existing and new loans suggest that the risk-
taking channel is a potentially fruitful avenue for further study. The model in
section 2 provides some of the conceptual background that may be necessary to
understand the results.
6.3. Related Literature
In order to highlight what we view as the speci?c contribution of the “risk-taking
channel” of monetary policy, it is important to give an account of the points of
contact between our approach and the existing literature in monetary economics
and corporate ?nance. The discussion can be organized along a number of di-
mensions, but one classi?cation that is useful is to distinguish those papers that
have emphasized the borrower’s balance sheet (and the demand for credit) from
those emphasizing the lender’s balance sheet (and the supply of credit).
Bernanke and Gertler (1989) is a classic example of the former - an explanation
based on the borrower’s balance sheet. Following the earlier work of Bernanke
(1983) who argued for the importance of borrower balance sheet distress during
the Great Depression, the Bernanke and Gertler model focuses on the agency prob-
lems entailed by the asymmetry of information between a non-?nancial corporate
borrower and the ?nancial market as a whole.
In the presence of asymmetric information between the borrower and lender,
ine?ciencies in the optimal contract manifest themselves in the form of deadweight
costs and, in particular, in the spread between the cost of internal funds and that
of external funds. The size of the deadweight cost is a function of the net worth
of the borrower - i.e., the borrower’s “skin in the game”. Moreover, Bernanke
and Gertler argue that the borrower’s net worth is procyclical, and so the funding
spread between the internal and external funds should be countercyclical.
An alternative approach that emphasizes the borrower’s balance sheet is the
57
work on credit cycles by Kiyotaki and Moore (1997, 2005)), who examine the dy-
namic ampli?cation of credit constraints. In their approach, Kiyotaki and Moore
assume a collateral constraint where the size of the loan that can be obtained
by a borrower depends on the current market price of the collateral that can be
pledged to the lender. Under such an assumption, the size of the borrower’s
balance sheet can depend positively on the market price of the asset - in other
words, the demand reaction of the borrower can be upward-sloping. When the
price of an asset increases, the borrower’s funding ability increases, thereby gen-
erating larger balance sheets. When the greater demand for the asset pushes up
the price of the asset, there is the possibility of ampli?ed responses where asset
price increases fuels further investment and aggregate activity, which raises prices
further.
The common thread between the work of Bernanke and Gertler (1989) and
Kiyotaki and Moore (1997, 2005) is that the focus is on the borrower’s balance
sheet, and the ?uctuations in the creditworthiness of the borrower. The supply
of lending is determined in the market as a whole, without a separate role for the
banking sector as such.
However, to the extent that the borrower in the Bernanke and Gertler (1989)
model can be re-interpreted as a bank, the model can be reoriented in terms of the
agency problems in the banking sector. The “double-decker” moral hazard model
in Holmstr¨ om and Tirole (1997) is a good illustration of such a re-interpretation
where the banking sector enters the model as a borrower subject to borrowing
constraints from its lenders. In the Holmstr¨ om and Tirole (1997) model, there
are two tiers in the agency problem. At the bottom tier, there is a moral haz-
ard problem between a non-?nancial borrower who needs funding to undertake a
project and a bank who supplies the funding. The moral hazard problem entails
an optimal contract where the incentive constraint stipulates that the borrower
58
has enough of a stake in the project that the good action is taken, rather than
the ine?cient action that yields private bene?t. The importance of borrower net
worth, or “skin in the game,” is a theme that Holmstr¨ om and Tirole (1997) shares
with Bernanke and Gertler (1989). However, the innovation in Holmstr¨ om and
Tirole (1997) is that there is a second tier to the agency problem in which the
bank (the lender) itself is subject to a moral hazard problem, so that there is a
constraint on the minimum equity capital that the bank itself must hold at all
times. In this respect, the minimum capital requirement of banks emerges as an
endogenous feature of an agency problem where banks must raise funding from
depositors and other suppliers of funds.
By re-orienting the agency problem so that the focus is on the bank (as bor-
rower) and the ?nancial market (as lenders), the earlier results of Bernanke and
Gertler and Kiyotaki and Moore can be transferred to the context of bank distress
and bank lending. The recent paper by Gertler and Kiyotaki (2009, this volume)
is a good example of such a re-orientation. The agency relationship between
the bank and the ?nancial market lender is a moral hazard problem due to the
possibility that the bank can steal some portion of the project outcome, so that
the bank is required to keep a minimum net worth in place at all times. The
bank must then keep a minimum amount of “skin in the game”, which translates
to a minimum capital ratio that the bank must maintain. When credit losses or a
fall in the price of assets depletes the capital of the bank, the incentive constraint
binds, entailing the withdrawal of lending by the bank.
Adrian and Shin (2008b) take up a similar theme of the binding incentive
constraint of the bank, where the agency problem comes in the form of a risk-
shifting problem where the bank may take the riskier asset when a lower risk asset
may be value-enhancing for the pair as a whole. The bank’s market-determined
minimum capital requirement arises from the need for the bank (the borrower) to
59
keep su?cient stake in the payo?s from the total balance sheet of the bank. Since
the agency problem manifests itself as a risk-shifting problem, Adrian and Shin
(2008b) can address how second-moment incentives can enter the problem, and
how the value at risk (VaR) constraint can emerge as an outcome of the optimal
contracting problem.
Brunnermeier and Sannikov (2009) take the moral hazard theme one step
further by embedding the moral hazard problem in a dynamic, continuous time
contracting environment. In this richer framework, Brunnermeier and Sannikov
examine two separate incentive constraints. One is the familiar one where the
borrower needs to keep su?cient “skin in the game”, and results in a minimum
capital ratio requirement set by the market. The second is a constraint in the
spirit of a value at risk (VaR) constraint which makes the debt instantaneously
risk-free. In order to accommodate a role for both types of constraints, Brunner-
meier and Sannikov incorporate the innovative feature that two types of equity
play an essential role. First, there is equity that carries control rights. This ?rst
type of equity is the stake of the controlling party. Second, there is equity that is
loss-absorbing, but which does not carry control rights. This is the type of equity
that is typi?ed by passive investors in hedge funds whose stake can be returned
by the controlling investor in the hedge fund. The interaction between these two
types of equity is a distinctive feature of the Brunnermeier and Sannikov model,
as well as the dynamic contracting framework, which gives the model considerable
richness and complexity. The role of equity as a bu?er has an a?nity with the
work on the e?ect of regulatory capital on lending as examined by Van den Heuvel
(2002).
So far, we have described the set of papers that have as their starting point
some type of agency problem between a borrower and a lender. However, there
is another strand of the literature from monetary economics that has emphasized
60
the institutional features surrounding the commercial banking industry, especially
for the United States.
The notable example is Bernanke and Blinder (1988), which proposed a simple
model of the supply of credit by banks that emphasizes the binding nature of
the reserve requirement of banks. The constraint itself is not motivated with
further microfoundations; instead, the institution of reserve requirements is taken
as given. The reserve requirement stipulates that, for a given amount of deposit
funding used by a bank, some minimum amount must be kept on deposit at the
Federal Reserve as cash assets of the bank. This is a constraint that links the
two sides of the balance sheet, and the assumption is that such a constraint binds
all the time.
The idea that reserve requirements bind all the time has been dealt a severe
blow by the experience of the aftermath of the recent ?nancial crisis in which
commercial banks in the United States held close to one trillion dollars of excess
reserves on their balance sheet. Excess reserves have also been a common feature
in other parts of the world after the ?nancial crisis. Nevertheless, until recently,
the assumption of binding reserve requirements has been an important feature of
the academic literature in banking.
Building on the initial short paper by Bernanke and Blinder (1988), their
follow-up paper (Bernanke and Blinder (1992)) is an in-depth empirical investiga-
tion of the monetary transmission mechanism. In particular, the focus is on how
the Fed Funds rate works through the ?nancial system to in?uence real activity.
The key section (Section 4) of the paper highlights the important empirical role of
the Fed Funds rate in in?uencing the future loan supply of banks. In particular,
Bernanke and Blinder (1992) show that an increase in the Fed Funds rate leads
to an eventual slow-down of bank lending at a time horizon that is similar to
the impact of the Fed Funds rate on unemployment. In addition, Bernanke and
61
Blinder (1992) show that an initial increase in the Fed Funds rate is met with a
rapid adjustment in the Bank’s portfolio in which the holding of securities ?rst
falls, and then is slowly rebuilt.
Bernanke and Blinder (1992) interpret these ?ndings as showing that the Fed
Funds rate a?ects the supply of bank lending directly through the portfolio con-
straints of the bank itself. In particular, when the Fed Funds rate is raised, the
deposit funding of the bank is squeezed, which puts pressure on the asset side
of the bank’s balance sheet to contract. Since loans are long-term contractual
arrangements, bank lending is initially slow to adjust and all the short-term ad-
justment is made via the holding of securities. Over time, the holding of securities
is built up, but the bank’s loan portfolio adjusts slowly to its new (lower) opti-
mum. In this way, a higher Fed Funds rate is seen to a?ect bank lending through
the squeeze in the deposit funding of the bank, which eventually feeds into the
decrease in bank lending.
The bank lending channel examined by Bernanke and Blinder (1992) has close
a?nities with the risk-taking channel of monetary policy proposed in our chapter.
The common theme is that the Fed Funds rate has a direct impact on credit
supply. However, the di?erences are also apparent. In Bernanke and Blinder
(1992), the mechanism that links the Fed Funds rate with the supply of bank
lending is the binding reserve requirement of the commercial banks. This poses
two challenges in the context of the recent crisis. First, reserve requirements
have not been binding in the aftermath of the crisis. Second, the Bernanke and
Blinder (1992) account focuses on the commercial banking sector, since this is the
sector for which there is a reserve requirement. However, as we have described
in some detail above, the credit crunch in the recent ?nancial crisis originated in
the shadow banking system and the market-based ?nancial intermediaries that
serve it, rather than in the traditional commercial banking sector. Indeed, the
62
commercial banking sector had seen increased balance sheets until the summer of
2009.
Nevertheless, the Bernanke and Blinder (1992) paper stands as a milestone in
the literature on the relationship between monetary policy and the banking sys-
tem. The conjectures that they proposed in their original paper were con?rmed
in a careful cross-sectional empirical study by Kashyap and Stein (2000), who
used a large dataset of banks in the United States to examine portfolio changes in
response to monetary policy shifts. They investigated the conjecture in Bernanke
and Blinder (1992) that banks with less liquid balance sheets (i.e., with fewer secu-
rities and more loans) were subject to greater downward impact in their banking
lending as a result of monetary tightening. They ?nd strong evidence that less
liquid banks are, indeed, subject to greater loan contraction due to monetary
tightening. The results are driven in particular by small banks, which form the
bulk of their sample.
Relative to the literature surveyed above, the risk-taking channel of monetary
policy proposed in this chapter has some distinctive features. First, in contrast
to Bernanke and Gertler (1989) and other approaches that emphasize the demand
for credit and the non-?nancial borrowers’ balance sheet, the risk-taking channel
emphasizes the role of the supply of credit by the ?nancial intermediary sector.
In this respect, the risk-taking channel has greater a?nity with approaches that
emphasize the supply of credit and the constraints that bind on the lenders’ side.
However, the distinctive feature of the risk-taking channel is the role played by
the price of risk and by the market-determined risk premium. The supply of
credit is determined by the threshold value of the risk-premium charged by the
market. In the model sketched in this chapter, the asset choice decision of the
banks is determined by an underlying risk-management problem where banks are
subject to a value at risk (VaR) constraint. In this chapter, we have left this
63
constraint without a further microfoundation. However, such a search would
lead naturally to a meeting point with the agency literature that emphasizes the
constraint imposed on the borrowers by the market as a whole. Adrian and Shin
(2008b) is an example of just such a setting.
7. Concluding Remarks
We conclude with some implications of our ?ndings for the conduct of mone-
tary policy. Our emphasis on the role of balance sheet aggregates of ?nancial
intermediaries leads to policy prescriptions that bear a super?cial similarity to
an older tradition in monetary economics that emphasized the money stock as a
pivotal quantity in monetary policy. The older monetarist tradition emphasized
the stock of money because of the supposed direct link between the money stock
and real expenditures through the portfolio adjustment of individual consumers
who rebalance their portfolios consisting of money and real goods. Monetary
aggregates had fallen from favor in the conduct of monetary policy mainly as a
backlash against the older monetarist line (see Friedman (1988)).
In this chapter, we have focused on balance sheet aggregates of ?nancial inter-
mediaries, but the rationale is quite di?erent from the older monetarist literature.
Our approach has been to emphasize the role of intermediary balance sheets as
a determinant for the risk appetite ruling in the economy, and how monetary
policy can a?ect the growth of intermediary balance sheets. Although our ratio-
nale for looking at balance sheets di?ers from the older monetarist literature, our
discussion nevertheless suggests that there is a case for rehabilitating some role
for balance sheet quantities in the conduct of monetary policy. By in?uencing
the rate of growth of intermediary balance sheets, the monetary authorities can
impact real decisions that depend on the price of risk ruling in the economy. Real
64
decisions that are sensitive to ?nancial conditions, such as residential investment,
will be particularly susceptible to shifts in the price of risk.
To the extent that monetary policy decisions ripple through to the real econ-
omy via the ?nancial system, our discussion also highlights the importance of
tracking the institutional underpinnings of the ?nancial system itself. The insta-
bility of money demand functions that undermines the practical use of monetary
aggregates in the older monetarist-style analysis is closely related to the emergence
of the market-based ?nancial system. As a result of those structural changes, not
all balance sheet quantities will be equally useful. The money stock is a measure
of the liabilities of deposit-taking banks, and so may have been useful before the
advent of the market-based ?nancial system. However, the money stock will be
of less use in a ?nancial system such as that in the United States. More useful
may be measures of collateralized borrowing, such as the weekly series on repos
of primary dealers. The model presented in the paper shows that balance sheet
quantities of ?nancial intermediaries are closely tied to risk premia and the supply
of credit, which, in turn, makes them useful in analyzing the ?nancial conditions
that determine the supply of credit. Adrian, Moench and Shin (2010) present
an empirical analysis that uses balance sheet quantities from a broad range of
?nancial intermediaries in order to gauge ?nancial conditions.
Finally, our results highlight the channel through which monetary policy and
policies toward ?nancial stability are linked. When the ?nancial system as a whole
holds long-term, illiquid assets ?nanced by short-term liabilities, any tensions
resulting from a sharp pullback in leverage will show up somewhere in the system.
Even if some institutions can adjust down their balance sheets ?exibly, there will
be some who cannot. These pinch points will be those institutions that are highly
leveraged, but who hold long-term illiquid assets ?nanced with short-term debt.
When the short-term funding runs away, they will face a liquidity crisis. The
65
traditional lender of last resort tools (such as the discount window), as well as
the recent liquidity provision innovations, are tools that mitigate the severity of
the tightening of balance sheet constraints. However, experience has shown time
and again that the most potent tool in relieving aggregate ?nancing constraints
is a lower target rate. Past periods of ?nancial stress such as the 1998 crisis were
met by reductions in the target rate aimed at insulating the real economy from
?nancial sector shocks. Our ?ndings suggest that, in conducting monetary policy,
the potential for ?nancial sector distress should be explicitly taken into account
in a forward-looking manner.
66
References
Adrian, Tobias, Arturo Estrella, and Hyun Song Shin (2010) “Monetary Cycles,
Financial Cycles, and the Business Cycle,” Federal Reserve Bank of New York
Sta? Report 421.
Adrian, Tobias, Dina Marchioni, and Karin Kimbrough (2010) “The Federal Re-
serve’s Commercial Paper Funding Facility,” Federal Reserve Bank of New York
Sta? Reports 423.
Adrian, Tobias, Erkko Etula and Hyun Song Shin (2009) “Risk Appetite and
Exchange Rates,” Federal Reserve Bank of New York Sta? Report 361.
Adrian, Tobias, Emanuel Moench and Hyun Song Shin (2010) “Financial Inter-
mediation, Asset Prices, and Macroeconomic Dynamics,” Federal Reserve Bank
of New York Sta? Report 422.
Adrian, Tobias, Emanuel Moench and Hyun Song Shin (2009) “Macro Risk Pre-
mium and Intermediary Balance Sheet Quantities,” Federal Reserve Bank of New
York Sta? Report 428.
Adrian, Tobias and Hyun Song Shin (2007) “Liquidity and Leverage,” Journal
of Financial Intermediation, forthcoming. Available as Federal Reserve Bank of
New York Sta? Reports 328.
Adrian, Tobias and Hyun Song Shin (2008a) “Liquidity, Monetary Policy, and
Financial Cycles,” Federal Reserve Bank of New York Current Issues in Economics
and Finance, Volume 14 (1).
Adrian, Tobias and Hyun Song Shin (2008b) “Financial Intermediary Leverage
and Value at Risk,” Federal Reserve Bank of New York Sta? Reports, 338.
Adrian, Tobias and Hyun Song Shin (2008c) “Liquidity and Financial Contagion,”
Banque de France Financial Stability Review, February 2008
Adrian, Tobias and Hyun Song Shin (2008d) “Financial Intermediaries, Finan-
cial Stability, and Monetary Policy,” Federal Reserve Bank of Kansas City 2008
Jackson Hole Economic Symposium Proceedings.
67
Adrian, Tobias and Hyun Song Shin (2009) “The Changing Nature of Financial
Intermediation and the Financial Crisis of 2007 - 2009,” forthcoming in the Annual
Review of Economics.
Ashcraft, Adam (2005) “Are Banks Really Special? New Evidence from the FDIC-
Induced Failure of Healthy Banks,” American Economic Review 95, pp. 1712-
1730.
Ashcraft, Adam (2006) “New Evidence on the Lending Channel,” Journal of
Money, Credit, and Banking 38, pp. 751-776.
Bernanke, Ben (1983) “Nonmonetary Aspects of the Financial Crisis in the Prop-
agation of the Great Depression” American Economic Review, 73, 257-76
Bernanke, Ben (2004) “The Logic of Monetary Policy,” remarks before the Na-
tional Economists Club, December 2, 2004.
www.federalreserve.gov/boarddocs/speeches/2004/20041202/default.htm
Bernanke, B. and A. Blinder (1988) “Credit, Money and Aggregate Demand”
American Economic Review, 78, 435-39.
Bernanke, B. and A. Blinder (1992) “The Federal Funds Rate and the Channel
of Monetary Transmission” American Economic Review 82, pp. 901 - 921.
Bernanke, Ben and Mark Gertler (1989) “Agency Costs, Net Worth, and Business
Fluctuations,” American Economic Review 79, pp. 14 - 31.
Bernanke, Ben and Mark Gertler (1995) “Inside the Black Box: The Credit Chan-
nel of Monetary Policy Transmission,” Journal of Economic Perspectives 9, pp.
27-48.
Bernanke, Ben and Mark Gertler (1999) “Monetary Policy and Asset Volatility,”
Federal Reserve Bank of Kansas City Economic Review 84, pp. 17-52.
Bernanke, Ben and Cara Lown (1991) “The Credit Crunch,” Brookings Papers on
Economic Activity 2, pp. 205-247.
Blinder, Alan (1998) Central Banking in Theory and Practice, MIT Press, Cam-
bridge.
68
Borio, Claudio and Haibin Zhu (2008) “Capital Regulation, Risk-taking and Mon-
etary Policy: A Missing Link in the Transmission Mechanism?” Bank for Inter-
national Settlements Working Paper 268.
Brunnermeier, Markus and Yuliy Sannikov (2009) “A Macroeconomic Model with
a Financial Sector,” unpublished working paper, Princeton University
Curdia, Vasco and Michael Woodford (2009) “Credit Spreads and Optimal Mon-
etary Policy,” Federal Reserve Bank of New York Sta? Reports 385.
Danielsson, Jon, Hyun Song Shin and Jean—Pierre Zigrand (2009) “Risk Appetite
and Endogenous Risk”, unpublished working paper, London School of Economics
and Princeton University.
Estrella, Arturo, and Gikas Hardouvelis (1989) “The Term Structure as a Pre-
dictor of Real Economic Activity.” Federal Reserve Bank of New York Research
Paper, May.
Estrella, Arturo, and Gikas Hardouvelis. (1991) “The Term Structure as a Pre-
dictor of Real Economic Activity.” Journal of Finance, 46: 555-576.
Estrella, Arturo, Anthony P. Rodrigues, and Sebastian Schich (2003) “How Stable
is the Predictive Power of the Yield Curve? Evidence from Germany and the
United States.” Review of Economics and Statistics, 85: 629-644.
Freixas, Xavier (2009) ”Monetary Policy in a Systemic Crisis,” unpublished work-
ing paper, University Pompeu Fabra.
Friedman, Benjamin (1988) “Monetary Policy Without Quantity Variables,” Amer-
ican Economic Review 78, 440-45.
Gatev, Evan, Til Schuermann, and Philip Strahan (2009) ”Managing Bank Liq-
uidity Risk: How Deposit-Loan Synergies Vary with Market Conditions,” Review
of Financial Studies 22(3), pp. 995-1020.
Geanakoplos, John (2009) “The Leverage Cycle” in 2009 NBER Macroeconomics
Annual 24, edited by Daron Acemoglu, Kenneth Rogo?, and Michael Woodford.
University of Chicago Press.
Gertler, Mark and Peter Karadi (2009) “A Model of Unconventional Monetary
Policy,” unpublished working paper, New York University.
69
Gertler, Mark and Nobuhiro Kiyotaki (2009) “Financial Intermediation and Credit
Policy in Business Cycle Analysis” working paper, NYU and Princeton University,
forthcoming in the Handbook of Monetary Economics.
Harvey, Campbell R. (1989) “Forecasts of Economic Growth from the Bond and
Stock Markets.” Financial Analysts Journal, September/October: 38-45.
Holmstr¨ om Bengt and Jean Tirole (1997) “Financial Intermediation, Loanable
Funds, and the Real Sector,” Quarterly Journal of Economics, 112, 663-692.
Ioannidou, Vasso, Ongena, Steven and Jose Luis Peydro (2008) “Monetary Policy,
Risk-Taking and Pricing: Evidence from a Quasi-Natural Experiment,” unpub-
lished working paper, ECB.
Jimenez, Gabriel, Ongena, Steven, Peydr´ o, Jose Luis and Jesus Saurina (2008)
“Hazardous Times for Monetary Policy: What do 23 Million Bank Loans Say
About the E?ects of Monetary Policy on Credit Risk?” unpublished working pa-
per, ECB and Bank of Spain.
Kashyap, Anil and Jeremy Stein (1994) “Monetary Policy and Bank Lending,” in
N. Gregory Mankiw (ed.) Monetary Policy, University of Chicago Press.
Kashyap, Anil and Jeremy Stein (2000) “What Do a Million Observations on
Banks Say about the Transmission of Monetary Policy?” American Economic
Review 90, pp. 407-428.
Keister, Todd, and James McAndrews (2009) “Why Are Banks Holding So Many
Excess Reserves?” Federal Reserve Bank of New York Sta? Report 380.
http://www.ny.frb.org/research/sta? reports/sr380.html
Lown, Cara, and Don Morgan (2006) “The Credit Cycle and the Business Cycle:
New Findings Using the Loan O?cer Opinion Survey,” Journal of Money, Credit,
and Banking 38, pp. 1575-1597.
Khwaja, Asim Ijaz and Atif Mian (2008) “Tracing The Impact of Bank Liquidity
Shocks.” American Economic Review 98, pp. 1413—1442.
Kiyotaki, Nobuhiro, and John Moore (1997) “Credit Cycles,” Journal of Political
Economy 105, pp. 211-248.
70
Kiyotaki, Nobuhiro, and John Moore (2005) “Liquidity and Asset Prices,” Inter-
national Economic Review 46, pp. 317-349.
Paravisini, Daniel (2008) “Local Bank Financial Constraints and Firm Access to
External Finance,” Journal of Finance 63, pp. 2161 - 2193.
Shin, Hyun Song (2009) Risk and Liquidity, 2008 Clarendon Lectures in Finance,
forthcoming from Oxford University Press.
Sims, Christopher (1980) ”Macroeconomics and Reality,” Econometrica 48, 1-48
Shleifer, Andrei, and Robert Vishny (1997) ”The Limits of Arbitrage,” Journal
of Finance 52(1), pp. 35-55.
Shleifer, Andrei, and Robert Vishny (2009) ”Unstable Banking,” Journal of Fi-
nancial Economics, forthcoming.
Stock, James, and Mark Watson. (1989) “New Indexes of Coincident and Lead-
ing Indicators.” In NBER Macroeconomic Annual 4, ed. Olivier Blanchard and
Stanley Fischer, 351-394.
Stock, James, and Mark Watson. (1993) “A Procedure for Predicting Recessions
with Leading Indicators: Econometric Issues and Recent Performance.” In Busi-
ness Cycles, Indicators, and Forecasting, ed. James Stock and Mark Watson.
Chicago, IL: University of Chicago Press.
Svensson, Lars (2004) “Challenges for Monetary Policy,” paper for the Bellagio
Group Meeting at the National Bank of Belgium, January 2004.
www.princeton.edu/˜svensson
Taylor, John (1993) “Discretion Versus Policy Rules in Practice,” Carnegie-Rochester
Series on Public Policy 39, pp. 195-214.
Van den Heuvel, S. (2002) “The Bank Capital Channel of Monetary Policy”,
working paper, Wharton School, University of Pennsylvania.
Woodford, Michael (2003) Interest and Prices: Foundations of a Theory of Mon-
etary Policy, Princeton University Press.
Woodford, Michael (2005) “Central Bank Communication and Policy E?ective-
ness,” Proceedings of the Federal Reserve Bank of Kansas City Economic Sympo-
sium at Jackson Hole 2005.
http://www.kc.frb.org/publicat/sympos/2005/sym05prg.htm
71
Appendix: Data Sources
Figure 1.1. The 10-year and 3-month Treasury constant maturity yields as well
as the e?ective Fed Funds rate are from the Federal Reserve Board’s H.15 release.
Figures 3.1—3.6. Figures 3.1, 3.4, 3.5, 3.6 use total assets of security broker-
dealers, ABS issuers, shadow banks (the sum of ABS issuers, ?nance companies,
funding corporations), and nationally chartered commercial banks from the Flow
of Funds published by the Federal Reserve Board. In Figures 3.2 and 3.3, the
money stock measure M1 and M2. Total outstanding and ?nancial commercial
paper used in Figures 3.2, 3.3, and 3.5 are from the Federal Reserve Board. Pri-
mary dealer repo in Figures 3.2, 3.3, and 3.6 is from the Federal Reserve Bank of
New York.
Figures 3.7 and 3.8. The ?gures use total ?nancial assets from the Federal
Reserve Board’s Flow of Funds.
Figures 4.1 and 4.2. The ?gures are based on computations by Adrian, Moench,
and Shin (2009). The macro risk premium as the predicted part of a regression
of real GDP growth on constant maturity Treasury yield spreads and corporate
bond spreads. The risk appetite variable is obtained by regressing (negative)
changes of the macro risk premium on lagged balance sheet variables of security
broker-dealers, shadow banks, and commercial banks.
Tables 4.1 and 4.2: Impact of Balance Sheets on GDP and Residential
Investment. The tables report regressions of GDP and residential investment
growth on the total asset growth of broker-dealers, shadow banks, and commercial
banks for 1986Q1 to 2009Q2. Lags are one quarter lags; growth rates are an-
nual. Total assets are from the Federal Reserve Board’s Flow of Funds. Shadow
banks include ABS issuers, funding corporations, and ?nance companies. Gross
domestic product (GDP) and residential investment is from the Bureau of Eco-
nomic Analysis (BEA). PCE in?ation is the personal consumption expenditures
de?ator excluding food and energy as reported by BEA. The equity return is the
one quarter return of Standard & Poor’s S&P500 index. The VIX is CBOE’s
72
implied volatility index (the VXO from 1986-1989, and the VIX from 1990 on-
wards). The term spread is the di?erence between the 10-year constant maturity
Treasury yield and the 3-month Treasury bill rate, both are from the Federal Re-
serve Board. The credit spread is the di?erence between Moody’s Baa spread
and the 10-year Treasury rate, both are from the Federal Reserve Board.
Table 6.1: Determinants of Balance Sheet Growth The table reports re-
gressions of repo growth, repo + commercial paper growth, and M2 growth on
their own lags, and asset price variables. The data frequency is weekly from
October 3, 1990 to February 3, 2010. Changes refer to 1-week changes, and lags
to 1-week lags. Fed Funds denotes the Federal Funds Target as reported by the
Federal Reserve Board. The equity return is the 1-week return of Standard &
Poor’s S&P500. The VIX is CBOE’s implied volatility index for the S&P500.
The term spread is the di?erence between the 10-year constant maturity Treasury
yield and the 3-month Treasury bill rate, both from the Federal Reserve Board.
The credit spread is the di?erence between Moody’s Baa spread and the 10-year
Treasury rate. Commercial paper growth is the 1-week growth rate of total com-
mercial paper outstanding reported by the Federal Reserve Board. Repo growth
is the 1-week growth rate of primary dealer repo, from the Federal Reserve Bank
of New York. M2 growth is the 1-week growth of the money measure M2 from
the Federal Reserve Board.
73
doc_677767475.pdf