Full text of Review (Federal Reserve Bank of St. Louis) : November/December 2011, Vol. 93, No. 6

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

International Liquidity Provision During the
Financial Crisis: A View from Switzerland
Raphael Auer and Sébastien Kraenzlin
The authors document the provision of liquidity in Swiss francs (CHF) by the Swiss National
Bank (SNB) to banks located outside Switzerland during the recent financial crisis. What makes
the Swiss case special is the size of this liquidity provision—at times, 80 percent of all short-term
CHF liquidity provided by the SNB—and the measures adopted to distribute this liquidity. In addition to making CHF available to other central banks via swap facilities, the SNB also allows banks
outside Switzerland to directly participate in its repurchase agreement transactions. Although
this policy was adopted for reasons predating the 2007-09 financial crisis, it proved tremendously
helpful during the crisis by providing the European banking system direct access to the primary
funding facility for CHF. (JEL E41, E52, F33, F34)
Federal Reserve Bank of St Louis Review, November/December 2011, 93(6), pp. 409-17.

I

n the years leading up to 2007, banks
across the globe dramatically increased
their balance sheet exposure to foreign
currencies. This led to increased trading
between banks with a need to refinance in the
foreign currency and domestic banks with
deposits and consequently sufficient funds to
lend in that currency (i.e., extensive cross-border
trading). With the onset of the financial crisis
and the successive drying-up of the repurchase
agreement (repo) market and especially the unsecured interbank money market (see Guggenheim,
Kraenzlin, and Schumacher, 2011), the private
sector no longer provided this liquidity, thus
requiring a coordinated action by the world’s
major central banks.
In particular, the provision of dollar liquidity
to non-U.S. banks by the Federal Reserve garnered
ample attention in the global financial press (for

a discussion, see, for example, Goldberg, Kennedy,
and Miu, forthcoming). Much less noticed was
the Swiss National Bank’s (SNB) large-scale provision of Swiss franc (CHF) liquidity to the banking system throughout the European Union and
beyond.
In this article, we document the CHF liquidity
provision by the SNB to banks located outside
Switzerland. What makes the Swiss case special
is not only the size of the liquidity provision to
banks outside Switzerland (at times, 80 percent
of all short-term CHF liquidity provided by the
SNB), but also the measures adopted to distribute
this liquidity.
In addition to providing CHF to other central
banks via swap facilities, the SNB allows foreign
banks to directly participate in its repo transactions. Although this policy was adopted for reasons predating the recent financial crisis, it proved

Raphael Auer is deputy head of the International Trade and Capital Flows Unit at the Swiss National Bank and a research associate at the
Liechtenstein Institute on Self-Determination at Princeton University. Sébastien Kraenzlin is a senior economist at the Money Market and
Foreign Exchange Division of the Swiss National Bank. The authors thank Jean-Pierre Danthine, Andreas M. Fischer, William T. Gavin,
Andreas Ittner, Dewet Moser, Thomas Moser, Daniel L. Thornton, and another anonymous referee for comments and suggestions.

© 2011, The Federal Reserve Bank of St. Louis. The views expressed in this article are those of the author(s) and do not necessarily reflect the
views of the Federal Reserve System, the Board of Governors, the regional Federal Reserve Banks, or the Swiss National Bank. Articles may
be reprinted, reproduced, published, distributed, displayed, and transmitted in their entirety if copyright notice, author name(s), and full
citation are included. Abstracts, synopses, and other derivative works may be made only with prior written permission of the Federal Reserve
Bank of St. Louis.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

409

Auer and Kraenzlin

tremendously helpful during the crisis when it
gave the European banking system direct access
to the primary funding facility for CHF.
Finally, the Swiss case is exceptional because
from March 2009 to June 2010, faced with deflation risks and zero interest rates, the SNB intervened in the foreign exchange market as part of
its unconventional policies. The resulting largescale inflow of CHF to the financial sector flooded
the international banking system with CHF liquidity. As a consequence, demand for the liquidity
provided by the SNB’s open market operations
virtually ceased to exist. We believe that, although
liquidity provision itself was not an objective of
the foreign exchange interventions, the transactions may have contributed to stabilizing the
European banking system.

THE ORIGINS: SWISS FRANC
LOANS IN AUSTRIA AND CENTRAL
AND EASTERN EUROPE
Because of the traditionally low interest
rates in Switzerland and the low-exchange-rate
volatility observed since the introduction of the
euro (EUR), many households and firms across
Central and Eastern Europe (CEE) relied on CHFdenominated loans as a source of cheap funding.
The resulting aggregate exposure was substantial:
By early 2009, households and non-bankingsector firms in CEE economies had accumulated
the equivalent of CHF 120 billion worth of debt
denominated in Swiss currency. In Austria, primarily because of its geographic proximity to
Switzerland, total exposure was then over CHF
80 billion. Non-banks in the other countries of
the euro zone also relied on such loans. In total,
the exposure of non-Swiss European banks
amounted to about CHF 400 billion in late 2008.1
1

Why lenders that issued these loans were not more concerned
with the embedded default risk might seem puzzling. However, a
microeconomic study of loan issuance to private households in
Austria (see Beer, Ongena, and Peter, 2008) finds that banks did
screen potential borrowers and awarded CHF-denominated loans
only to the more solvent clients (also see Auer et al., 2009). In addition, many such loans contained provisions giving the banks the
right to coerce conversion of a loan to the local currency if the
exchange rate exceeded a certain trigger level. The combination
of these two features probably explains why lenders were not
concerned with default risk when issuing these loans.

410

N OV E M B E R / D E C E M B E R

2011

The size of the exposure has raised many
concerns about the financial stability of the banking sector, given the possibility of continued CHF
strength or even appreciation. (Most notable are
concerns expressed by Krugman, 2009.) Since few
of the debtors have any CHF income, such an
appreciation could cause large-scale default and
the resulting loan losses could strain the banking
sector in these economies.2 However, a second
financial stability concern related to the CHF
loans has received surprisingly little attention—
with the notable exception of Pann, Seliger, and
Übeleis (2010)—namely, the resulting funding
and liquidity risk faced by non-Swiss banks.

SYSTEMIC SWISS FRANC
SHORTAGES DURING THE CRISIS
CHF-denominated loans obtained by nonbanks outside Switzerland are typically granted
by non-Swiss banks that, in turn, finance themselves by borrowing from financial institutions
in Switzerland. As in all bank business, these nonSwiss banks provide long-term loans yet finance
themselves on a short-term basis. Their ability to
roll over maturing CHF positions became stressed
when the interbank money market progressively
dried up following the onset of the financial crisis
in August 2007, particularly after the collapse
of Lehman Brothers in September 2008 (see
Guggenheim, Kraenzlin, and Schumacher, 2011).
In international currency markets, any bank
can potentially obtain financing in any foreign
currency either by going directly to the interbank
money market or by obtaining funds from its central bank and swapping the received funds into
the desired foreign currency. In principle, these
two methods should ensure the rate at which a
currency is funded is the same.
During the recent financial crisis, however,
interbank money markets temporarily faltered.
2

The aggregate exposure of CEE and Austria to low-interest-rate
currencies had already caused losses of around $60 billion (U.S.
dollars) for these nations in 2008-09 alone; see Auer and
Wehrmüller (2009). Note that empirical studies by Beer, Ongena,
and Peter (2008); Brown, Ongena, and Yesin (2011); Brown, Peter,
and Wehrmüller (2009); and Puhr, Schwaiger, and Sigmund (2009)
show that the debtors tend to be creditworthy, thus suggesting that
such concerns are less relevant than the sheer magnitude of the
aggregate exposure suggests.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Auer and Kraenzlin

Figure 1
Difference Between Unsecured and Secured Overnight Interbank Rate for CHF Funds and
EUR Funds
Basis Points
400
CHF
EUR
Oct 15, 2008
Oct 20, 2008

350
300
250
200
150
100
50
0
–50
–100
Aug 13

Sep 2

Sep 22

Oct 12

Nov 1

Nov 21

Dec 11

NOTE: The figure shows the evolution of the difference between the unsecured and secured overnight interbank rate for Swiss franc
(CHF) funds and euro (EUR) funds from August to November 2008 (5-day moving average). The two horizontal lines correspond to the
announcement (October 15, 2008) and the actual start (October 20, 2008) of EUR/CHF swap auctions by the European Central Bank,
Magyar Nemzeti Bank (the central bank of Hungary), and the National Bank of Poland.

For example, Figure 1 documents the strains in
the CHF money market beginning in October 2008.
The figure plots the difference between the unsecured and secured overnight interbank rate for
both CHF funds and EUR funds. While these two
spreads are historically rather low and co-move
closely, the spread on CHF rose steeply during
October 2008, reaching values well over 300 basis
points. Consequently, the movement in the CHF
money market is a result of an increase in liquidity and not credit risk premia since the latter
would be reflected in both currencies.
The CHF-specific spike in the cost of obtaining unsecured funds was caused by a combination
of the need by banks outside Switzerland to continuously roll over maturing interbank loans and
the shrinking supply for these funds. Most Swiss
banks and a considerable number of non-Swiss
banks have access to the Swiss repo system—the
prevailing secured money market in Swiss francs.
In a calm market environment, these banks would
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

have immediately exploited this profit opportunity
and provided unsecured funds to banks without
access to the Swiss repo system.
However, against the backdrop of the global
financial crisis and the fear of counterparty default
risk, this situation did not occur and the spread
between secured and unsecured CHF funds
remained elevated for several trading days. Without access to the Swiss repo system, even banks
with ample collateral could not obtain secured
funding from the SNB or the secured interbank
market. In Switzerland, only a negligible amount
of repo transactions are traded outside the Swiss
repo system (i.e., over the counter).
The lower cross-border trading could have
posed a substantial danger to the stability of the
financial sector at large. If banks across the euro
zone and CEE were unable to obtain CHF in the
money market, then non-Swiss banks, in turn,
could try to reduce their exposure by liquidating
CHF loans they had made to their clients. Given
N OV E M B E R /D E C E M B E R

2011

411

Auer and Kraenzlin

the banking tensions at the time, this move would
have driven many debtors into default and could
have started a disorderly winding-down of CHF
loans, with increasing default rates implying the
need for additional loan-loss provisions, thereby
increasing pressure to liquidate CHF exposure.
This vicious cycle could have had dire consequences for the banking system and the real
economy.

PHASE I: INTER-CENTRAL BANK
SWAP FACILITIES
The drying-up of liquidity distribution in
foreign currency posed a problem more challenging than the breakdown of the domestic interbank
money market: No central bank, on its own, can
provide a large amount of liquidity in a foreign
currency in a timely manner.3 First, the European
Central Bank (ECB) and the central banks in CEE
obviously cannot create CHF liquidity without
issuing their own debt securities in the respective
currency. Second, the SNB can create CHF liquidity but cannot supply this liquidity to banks lacking access to the Swiss repo system or banks with
insufficient SNB-eligible collateral, which in 2007
was the case for most banks involved in CHFdenominated lending in CEE.4
To overcome this market friction, the SNB
jointly announced with the ECB and subsequently
with the Narodowy Bank Polski (the National
Bank of Poland) and the Magyar Nemzeti Bank
(the central bank of the Republic of Hungary) that
all these central banks would directly distribute
CHF-denominated funds to their counterparties.5
Since (i) nearly all banks that require funding of
some CHF exposure are registered with one of
3

The SNB also issued its own debt certificates in U.S. dollars (SNB
USD bills). The U.S. dollars were subsequently used to finance
the SNB’s loan to its stabilization fund. The outstanding volume
peaked at $20 billion USD. A central bank can thus obtain foreign
currency and subsequently provide liquidity to its counterparties.
Depending on the urgency and extent of lending, however, it may
prove difficult.

4

In general, the establishment of access to the Swiss repo system takes
several months. Hence, banks with CHF exposure but no access
to the Swiss repo system could not establish access quickly
enough.

412

N OV E M B E R / D E C E M B E R

2011

these four central banks and (ii) the conditions
for these funds are similar across these countries,
in effect the private sector instantly gained access
to the primary source of CHF: the SNB.6
As Figure 1 clearly shows, on the value date
of the first swap transaction, the CHF tensions
in the unsecured money market ceased once the
CHF auctions were implemented by the partner
central banks. In other words, the swap transactions were effective in reducing the liquidity
premium and in returning the interest rate spread
for Swiss francs to a level similar to that for euro
funds.
Figure 2 documents the extent to which banks
located in the European Union used the EUR/CHF
swap facility. With the introduction of the facility,
demand for CHF in the euro zone jumped to
around CHF 40 billion and stayed there for about
6 months. Thereafter, demand for Swiss francs
under the EUR/CHF swap facility leveled off and
ceased after January 2010.

PHASE II: ENHANCING FOREIGN
BANKS’ DIRECT ACCESS TO THE
REPO SYSTEM
Figure 1 demonstrates that the EUR/CHF swap
was a functioning measure to address short-run
liquidity mismatches. Since swaps are just a
means to distribute liquidity more effectively,
they involve no direct costs, but they do still have
limits. First, their maximum volume is agreed
upon in advance, so they are not as flexible as
measures controlled by only one central bank.
Second, the swap agreement itself to some extent
fractionalizes the market for CHF liquidity since
5

Although the central bank swap agreements are bilateral, it is
sometimes the case that funding is “recycled” to other countries.
This is an especially relevant channel for international CHF liquidity provision as it is likely that banks in the euro area, particularly
Austrian banks, forwarded CHF funds to their subsidiaries across
CEE. Thus, the CHF funding was indirectly available to more
countries.

6

The Hungarian central bank offered slightly different conditions
than the other central banks. See Goldberg, Kennedy, and Miu
(2011), Auer and Kraenzlin (2009), and Aizenman and Pasricha
(2010) for a discussion of various swap line agreements around
the globe.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Auer and Kraenzlin

Figure 2
Volume of CHF Loans to Euro Zone Banks via the EUR/CHF Swap Facility
CHF Billion
45
40
35
30
25
20
15
10
5
0
Aug 08

Dec 08

May 09

the total supply of CHF is split across different
selling platforms.7
Finally, swap agreements also involve some
loss of control over monetary policy because, in
essence, the monetary base is partly controlled by
a foreign central bank. The main worry of policymakers is that such swap agreements could create
inflationary pressure because opening new means
to distribute liquidity can increase the total supply of money. For example, the maximum amount
of a swap is agreed upon months in advance.
Since the receiving central bank may auction off
the maximum amount (but is not obliged to do
so), uncertainty in the growth of the money supply is increased.8 Given these shortcomings, the
SNB, the ECB, and the euro zone member central
7

A further potential worry is that these agreements could entail a
larger counterparty default risk. This is not the case. First, there is
no risk involved for the central bank distributing the funds since
the receiving central banks guarantee these transactions. Second,
there is also no effect on counterparty default risk for the receiving
central bank since it transacts with its regular counterparties against
the regular collateral basket.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Sep 09

Feb 10

Jul 10

banks, as well as all other affected central banks,
advised banks with major exposure to the CHF
to seek access to the SNB’s repo system.9
The SNB is also legally empowered to provide liquidity to banks outside Switzerland.10
The original intent in allowing foreign banks to
access the Swiss repo system was to (i) reduce
8

Two main reasons explain why the loss of control of the monetary
base is rather contained. First, the central bank that originates the
funds can sterilize the effect on the monetary base by issuing its
own debt certificates or providing liquidity, thereby absorbing open
market operations. While ceding some control over monetary policy
to other central banks is of little concern in the current low-inflation
environment, such concerns will definitely become a first-order
political topic once inflationary pressures resume and central banks
must refocus on their core task of maintaining price stability.

9

In particular, the Austrian financial authorities (the Austrian
National Bank and the Austrian Financial Market Authority
[Finanzmarktaufsicht]) have assumed a key role in persuading
commercial banks in Austria to seek access to the Swiss repo
system.

10

The repo system used by the SNB is also the same system in which
the majority of interbank CHF repo transactions are conducted.
Hence, even banks without access to the SNB could use the interbank repo market for refinancing purposes.

N OV E M B E R /D E C E M B E R

2011

413

Auer and Kraenzlin

Figure 3
Use of the SNB Repo System by Banks Within and Outside Switzerland
No. of Banks

Total Outstanding SNB Repo Volume (CHF Billion)
75

60

60

48
Banks Outside Switzerland
Banks in Switzerland
No. of Foreign Repo Participants

45

30

24

15

12

0
Aug 08

0
Dec 08

May 09

the dependence on the few large Swiss financial
institutions, (ii) improve the general liquidity in
the banking system, and (iii) thereby facilitate
the steering of a longer-term money market rate—
namely, the 3-month CHF London Interbank
overnight rate (LIBOR).
This pre-crisis policy also proved useful in
addressing cross-border liquidity shortages during
the financial turmoil. The solid black line in
Figure 3 plots the evolution of the number of
banks in the Swiss repo system located outside
Switzerland (right axis). As of mid-November
2010, 59 such banks had established access to
the Eurex Repo electronic trading platform, a
necessary condition to participate in the SNB’s
repo auctions. Of these 59 banks, 23 were located
in Austria, 16 in Germany, and 6 in the United
Kingdom.11
Figure 3 also documents the volume of CHF
liquidity obtained directly from the SNB by foreign banks, which temporarily exceeded CHF 60
billion. The comparatively small volume of CHF
414

36

N OV E M B E R / D E C E M B E R

2011

Sep 09

Feb 10

liquidity obtained by Swiss banks makes clear
just how sizable the foreign demand was: For
most of 2009 and early in 2010, well over 70 percent of the liquidity demand was from outside
Switzerland. With the money obtained indirectly
through EUR/CHF swaps, short-term CHF liquidity held by non-Swiss Banks rose to 90 percent.
Figure 4 highlights the importance of direct
access to the SNB repo system for banks located
outside Switzerland (see also Kraenzlin and von
Scarpatetti, 2011). This figure presents an area
diagram (stacked) showing total provision of CHF
liquidity to banks located outside Switzerland.
The figure shows the volume supplied within the
Swiss repo system (light blue) and the volume
supplied via EUR/CHF swaps (dark blue).
11

The SNB also accepts securities denominated in foreign currency.
High credit standards and a highly efficient risk management procedure imply that the SNB does not apply haircuts. Banks located
outside Switzerland thus can deliver non-CHF-denominated securities in SNB repo transactions. It is unclear to what extent this
possibility has contributed to the high use of the SNB repo facility
by this group of banks observed during 2009 and early 2010.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Auer and Kraenzlin

Figure 4
Total Liquidity Provision to Banks Outside Switzerland
CHF Billion
100
EUR/CHF Swap Volume
SNB Repo Volume

90
80
70
60
50
40
30
20
10
0
Aug 08

Dec 08

May 09

PHASE III: EXCHANGE RATE
INTERVENTIONS
Figures 3 and 4 document not only the extent
of the CHF shortage during 2008 and 2009, but
also that this demand decreased substantially
starting in mid-2009; demand vanished completely in mid-2010. Although it is tempting to
attribute this to a resurgence of activity in the
interbank money market, this is not fully the
case. Rather, starting in March 2009, the SNB
intervened in the foreign exchange market, eventually building up a foreign reserve position worth
over CHF 200 billion, compared with a pre-2009
level of less than CHF 50 billion.
While the exchange rate interventions were
part of the SNB’s unconventional measures to
avert deflation risks in Switzerland, an unintended
side effect of the interventions was the resolution
of the international CHF liquidity shortage: The
supply of the additional CHF 150 billion is available to the banking system on a permanent basis
and, consequently, the majority of banks are
awash with CHF liquidity.
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Sep 09

Feb 10

Figure 5 puts the extent of the liquidity provision via the exchange rate interventions in perspective. The figure shows the combined total of
CHF liquidity supplied to banks located within
and outside Switzerland. The three key elements
of this supply are the Swiss repo system (dark gray
area), the EUR/CHF swaps (light gray area), and
the SNB exchange rate interventions (black area).
Figure 5 documents that the exchange rate
interventions were so sizable that they in effect
created enough liquidity that demand for liquidity
via repo and swap transactions ceased altogether.
In fact, the SNB currently absorbs liquidity to
implement monetary policy. This is done, on the
one hand, through weekly issuance of the SNB’s
own money market bills (SNB bills) and, on the
other hand, through daily one-week repo auctions
(see also Anderson, Gascon, and Liu, 2010, and
SNB, 2011).
The exchange rate interventions thus proved
helpful from a financial stability perspective.
Using loans denominated in a low-interest-rate
currency such as the CHF is essentially a carry
N OV E M B E R /D E C E M B E R

2011

415

Auer and Kraenzlin

Figure 5
Total Supply of CHF Liquidity
CHF Billion
200
180
160

EUR/CHF Swap
Total Repo Volume
Increase in Foreign Reserves Compared with March 2009

140
120
100
80
60
40
20
0
Apr 07 Jul 07

Oct 07 Jan 08

Apr 08 Jul 08

Oct 08 Jan 09

Apr 09 Jul 09

Oct 09 Jan 10

Apr 10 Jul 10

Oct 10

SOURCE: Data for reserve levels are from the SNB’s Monthly Statistical Bulletin.

trade strategy. Such strategies are always subject
to the danger of a disorderly winding-down of
positions: If the losses stemming from an appreciation of the CHF become too large such that
counterparty default risks surface, carry traders
can no longer refinance their positions and must
liquidate them; this, in turn, causes a further
appreciation of the CHF. The combination of swap
facilities, enhanced direct access to the primary
source of CHF liquidity, and exchange rate interventions was instrumental in ensuring that, to
date, such sizable disruptive winding-downs have
not taken place and are unlikely in the near future.

416

N OV E M B E R / D E C E M B E R

2011

CONCLUSION
Small frictions in how the private sector distributes liquidity internationally can have large
effects on the interest rate paid. The rapid, coordinated, and large policy response by central
banks across Europe may have avoided a disorderly winding-down of the carry trade positions
built up by European households and firms in the
years leading up to the recent financial crisis.
International liquidity mismatches involving
Swiss francs are currently of little concern, which
may be an unintended side effect of the liquidity
injection via SNB interventions in the foreign
exchange market. The establishment of access
to the Swiss repo system by banks outside
Switzerland also contributed to this relatively
calm environment. As of this writing, the private
sector thus has won time to reduce its CHF exposure in an orderly way.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Auer and Kraenzlin

REFERENCES
Aizenman, Joshua and Pasricha, Gurnain Kaur. “Selective Swap Arrangements and the Global Financial Crisis:
Analysis and Interpretation.” International Review of Economics and Finance, June 2010, 19(3), pp. 353-65.
Anderson Richard G.; Gascon, Charles S. and Liu, Yang. “Doubling Your Monetary Base and Surviving: Some
International Experience.” Federal Reserve Bank of St. Louis Review, November/December 2010, 92(6),
pp. 481-505; http://research.stlouisfed.org/publications/review/article/8509.
Auer, Raphael A.; Brown, Martin; Fischer, Andreas M. and Peter, Marcel. “Will the Crisis Wipe Out Small Carry
Traders in Central and Eastern Europe?” VoxEU.org, January 29, 2009; www.voxeu.com/index.php?q=node/2916.
Auer, Raphael A. and Kraenzlin, Sébastien P. “The Effectiveness of Central Bank Swap Agreement as a CrisisFighting Tool.” VoxEU.org, October 14, 2009; www.voxeu.org/index.php?q=node/4084.
Auer, Raphael and Wehrmüller, Simon. “How Big Are the Eastern European Losses?” VoxEU.org, April 20, 2009;
www.voxeu.org/index.php?q=node/3467.
Beer, Christian; Ongena, Steven R.G. and Peter, Marcel. “Borrowing in Foreign Currency: Austrian Households
as Carry Traders.” Journal of Banking and Finance, September 2010, 34(9), pp. 2198-211.
Brown, Martin; Ongena, Steven R.G. and Yesin, Pinar. “Foreign Currency Borrowing by Small Firms in the
Transition Economies.” Journal of Financial Intermediation, July 2011, 20(3), pp. 285-302.
Brown, Martin; Peter, Marcel and Wehrmüller, Simon. “Swiss Bank Lending in Europe.” Policy paper, Swiss
National Bank, February 2009; www.snb.ch/n/mmr/reference/sem_2008_09_22_background/source.
Goldberg, Linda S.; Kennedy, Craig and Miu, Jason. “Central Bank Dollar Swap Lines and Overseas Dollar
Funding Costs.” Federal Reserve Bank of New York Economic Policy Review, May 2011;
www.newyorkfed.org/research/epr/11v17n1/1105gold.pdf.
Guggenheim, Basil; Kraenzlin, Sébastien P. and Schumacher, Silvio. “Exploring an Uncharted Market: Evidence
on the Unsecured Swiss Franc Money Market.” Working Paper No. 2011-5, Swiss National Bank, 2011;
www.snb.ch/n/mmr/reference/working_paper_2011_05/source/working_paper_2011_05.n.pdf.
Kraenzlin, Sébastien P. and von Scarpatetti, Benedikt. “Liaisons and Scheming: The Network of the Swiss Franc
Repo Market.” Swiss National Bank working paper (forthcoming).
Krugman, Paul R. “The Conscience of a Liberal: Austria.” New York Times, April 15 2009;
http://krugman.blogs.nytimes.com/2009/04/15/austria/.
Pann, Johannes; Seliger, Reinhardt and Übeleis, Julia. “Foreign Currency Lending in Central, Eastern and
Southeastern Europe: The Case of Austrian Banks.” Financial Stability Report 20. Austrian National Bank,
December 2010, pp. 56-76; www.oenb.at/en/img/fsr_20_gesamt_tcm16-214505.pdf.
Puhr, Claus; Schwaiger, Markus S. and Sigmund, Michael. “Direct Cross-Border Lending by Austrian Banks to
Eastern Europe.” Financial Stability Report 17. Austrian National Bank, June 2009, pp. 102-22;
www.oenb.at/en/img/fsr_17_gesamt_tcm16-140492.pdf.
Swiss National Bank. Swiss National Bank: 103rd Annual Report 2010. Zurich: SNB, 2011;
www.snb.ch/en/mmr/reference/annrep_2010_komplett/source.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

417

418

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Banking Industry Consolidation
and Market Structure:
Impact of the Financial Crisis and Recession
David C. Wheelock
The number of U.S. commercial banks and savings institutions declined by 12 percent between
December 31, 2006, and December 31, 2010, continuing a consolidation trend begun in the mid1980s. Banking industry consolidation has been marked by sharply higher shares of deposits held
by the largest banks—the 10 largest banks now hold nearly 50 percent of total U.S. deposits. However, antitrust policy is predicated on the assumption that banking markets are local in nature, and
enforcement has focused on preventing bank mergers from increasing the concentration of local
banking markets. The author finds little change over time in the average concentration of local
banking markets or the average number of dominant banks in them, even during the recent financial crisis and recession when numerous bank failures and several large bank mergers occurred.
Concentration did not increase substantially, on average, in markets where mergers occurred among
banks when both the acquiring and acquired banks had existing local offices, though rural markets
generally saw larger increases in concentration from such mergers than did urban markets. Although
the structures of local banking markets, on average, have changed little since the mid-1980s, deposit
concentration has continued to increase at the level of U.S. Census regions. As technology evolves
and the costs of obtaining banking services from distant providers fall further, local market characteristics may become less relevant for analysis of competition in banking. (JEL G21, G28, L41)
Federal Reserve Bank of St. Louis Review, November/December 2011, 93(6), pp. 419-38.

T

he recent financial crisis and recession
produced a sharp increase in the number of commercial bank and savings
institution failures in the United States.
Mergers of non-failed commercial banks and savings institutions (hereafter “banks”) eliminated
still more banks, and in total, the number of U.S.
banks fell by 12 percent between December 31,
2006, and December 31, 2010.1 Over the same
period, the share of total U.S. deposits held by
the 10 largest commercial banks rose from 44 to
49 percent, continuing a trend that began in the

1

The Federal Deposit Insurance Corporation (FDIC) often resolves
bank failures by arranging mergers of failed institutions with other
banks. These are referred to as “assisted” mergers. Mergers that
do not involve failed institutions are referred to as “unassisted”
mergers. During 2007-10, 270 commercial banks and 54 savings
institutions, representing 4 percent of commercial banks and savings institutions in operation at the end of 2006, failed; unassisted
mergers absorbed another 893 commercial banks and 109 savings
institutions. These data refer to FDIC-insured commercial banks
and savings institutions located in U.S. states and the District of
Columbia and were obtained from Historical Statistics on Banking,
Tables CB02 and SI02 (http://www2.fdic.gov/hsob/index.asp).

David C. Wheelock is a vice president and deputy director of research at the Federal Reserve Bank of St. Louis. The author thanks Subhayu
Bandyopadhyay, Alton Gilbert, and Adam Zaretsky for comments on a previous draft of this article. David A. Lopez provided research
assistance.
© 2011, The Federal Reserve Bank of St. Louis. The views expressed in this article are those of the author(s) and do not necessarily reflect the
views of the Federal Reserve System, the Board of Governors, or the regional Federal Reserve Banks. Articles may be reprinted, reproduced,
published, distributed, displayed, and transmitted in their entirety if copyright notice, author name(s), and full citation are included. Abstracts,
synopses, and other derivative works may be made only with prior written permission of the Federal Reserve Bank of St. Louis.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

419

Wheelock

early 1990s toward greater concentration of total
U.S. deposits among the largest banks.2
Federal law prohibits any bank from obtaining more than 10 percent of total U.S. deposits
or more than 30 percent of a single state’s total
deposits by acquiring other non-failed banks, and
some states have imposed even lower deposit
share limits.3 Further, antitrust enforcement prevents mergers of non-failed banks that would
significantly increase the concentration of local
banking markets. However, antitrust policy does
not (i) prevent acquisitions of failed banks that
increase local market concentration or (ii) attempt
to limit increases in concentration that do not
result from mergers. Nonetheless, during the
1990s, local urban banking markets generally
did not become significantly more concentrated,
despite increases in the deposit shares of the
largest U.S. and regional banks (Amel, 1996, and
Dick, 2006).
Banking industry consolidation has since
continued, spurred in part by the recent financial crisis and recession. This article examines
changes since 1999 in the concentration of U.S.
banking markets, defined both at the local level
(metropolitan statistical areas [MSAs] and nonMSA rural counties) and at the Census-region
level. It examines whether the characteristics of
urban and regional banking markets observed
during the 1990s continued over the subsequent
decade. The article focuses in particular on the
years 2006-10 to gauge whether trends in banking
market structures continued during the financial
crisis and recession. The resolution of failed banks
during 2007-10 did not increase the concentration
of most local banking markets (Wheelock, 2011).
However, unassisted mergers accounted for more
of the decline in the number of U.S. banks during
2007-10 than did bank failures, and therefore
2

3

In both 1984 and 1993, the 10 largest banks held 15 percent of
total U.S. bank deposits. However, by 1999, the 10 largest banks
held 28 percent of total U.S. bank deposits. These data are for
December 31 of the year indicated for U.S. commercial banks
located in the 50 states and the District of Columbia.
Caps on deposit shares were imposed by the Riegle-Neal Interstate
Banking and Branching Efficiency Act of 1994. Adequately capitalized banks may exceed the caps by acquiring failing or FDICassisted banks. Banks may also exceed the caps through internally
generated growth. See Spong (2000) for additional details.

420

N OV E M B E R / D E C E M B E R

2011

potentially had a larger impact on the structures
of banking markets.
Following the approach of Dick (2006), this
article uses both the Herfindahl-Hirschman index
(HHI) and the number of dominant firms in a
market—that is, the minimum number of banks
that, combined, hold at least 50 percent of a market’s total deposits—to measure market concentration. However, unlike Dick (2006), this article
examines trends in the concentration of rural
banking markets as well as MSAs, and it includes
both commercial banks and savings institutions
in the analysis of market concentration (for comparison, the article also reports results for commercial banks only).4 Further, the article investigates
the impact of unassisted mergers on banking
market concentration during 2007-10. The results
show that, in general, local banking markets did
not become significantly more concentrated during 2006-10 but, as Dick (2006) finds for the 1990s,
concentration increased markedly at the level of
U.S. Census regions.
The next section investigates trends in bank
deposit concentration for both local banking
markets (MSAs and rural counties) and Census
regions. The following section examines trends in
the number of dominant banks, again at the levels
of local banking markets and Census regions.
Subsequently, the article examines the impact of
unassisted mergers during 2007-10 on the concentration of deposits for MSAs and rural counties.
The final section provides study conclusions.

BANKING CONCENTRATION:
LOCAL AND REGIONAL PATTERNS
The recent decline in the number of U.S.
banks has continued a trend dating back to the
mid-1980s (Figure 1). Hundreds of banks failed in
the late 1980s and early 1990s. Many more were
absorbed through unassisted mergers, spurred
by the relaxation of legal restrictions on bank
4

Regulators consider the presence of savings institutions when
evaluating the implications of proposed bank mergers on market
competition but down-weight the shares of deposits held by savings institutions by one-half in formal analysis of market concentration. See Gilbert and Zaretsky (2003) for analysis of the methods
and assumptions used by regulators in evaluating banking market
competition.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

Figure 1
Number of U.S. Commercial Banks and Savings Institutions (1984-2010)
Number of Institutions
16,000
Commercial Banks
14,000

Savings Institutions

12,000
10,000
8,000
6,000
4,000
2,000
0
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

branching by many states and the federal government (Amel, 1996).5 The number of U.S. commercial banks reached a post-World War II peak
of 14,495 banks in 1984. By the end of 2010, the
number had fallen to 6,532. Similarly, the number
of Federal Deposit Insurance Corporation (FDIC)insured savings institutions fell from 3,566 to
1,128 over the same period (the number of savings
institutions peaked at 3,740 in 1986).
Despite an increase in the share of total U.S.
deposits held by the very largest banks, the concentration of deposits among banks in local markets changed little, on average, from the mid-1980s
through the 1990s (Amel, 1996, and Dick, 2006).
Furthermore, the advent of interstate bank branching in 1997 had little immediate impact on either
5

The Riegle-Neal Interstate Banking and Branching Efficiency Act
of 1994 permitted interstate branching beginning in 1997 but gave
states the option to restrict de novo branching by banks headquartered in other states. The Dodd-Frank Wall Street Reform and
Consumer Protection Act of 2010 (Section 613) substantially
removed remaining restrictions on interstate branching by eliminating this option.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

local banking market concentration or state-level
measures of banking market competition (Dick,
2006, and Yildirim and Mohanty, 2010).6
Bank regulators use Department of Justice
(DOJ) guidelines for market concentration to evaluate the competitive effects of proposed bank
mergers and acquisitions. Proposed transactions
that would substantially increase market concentration are subject to more scrutiny and are more
likely to be rejected on antitrust grounds than
transactions that would not increase concentration
significantly. Regulators use data on deposits held
by individual bank branch offices, which banks are
required to report on June 30 of each year, to measure the concentration of local banking markets.7
6

Yildirim and Mohanty (2010) find that state banking markets could
be characterized as monopolistically competitive both before and
after deregulation; however, they also find that the level of competition declined in 30 states after deregulation, increased in 10
states, and did not change significantly in 10 others.

7

Summary of Deposits data are available from the FDIC
(http://ww2.fdic.gov/sod/index.asp).

N OV E M B E R /D E C E M B E R

2011

421

Wheelock

Ordinarily, proposed mergers are not challenged
on competitive grounds unless they would
result in a post-merger HHI value of more than
1800 points and an increase in the index of more
than 200 points in the relevant banking market.8
A premise of antitrust enforcement is that
banking markets are local in nature, and regulators
calculate pro forma HHI values for local banking
markets (typically MSAs or non-MSA rural counties) to evaluate the competitive implications of
proposed bank mergers.9 In the past, legal restrictions on branching and high transportation and
communications costs made it difficult and costly
for the public to obtain services from geographically distant banks. Further, many studies found
that deposit interest rates were lower, and loan
interest rates were higher, in more concentrated
local banking markets, suggesting that concentration was an important determinant of the competitiveness of banking markets.10 However, branching
deregulation, along with advances in informationprocessing and communications technologies,
have reduced the cost of obtaining financial
services from distant banks and raise the question
whether larger geographic areas, such as states,
Census regions, or even the nation as a whole,
are more relevant for evaluating banking competition. Nonetheless, studies find that (i) households and small businesses, to a substantial degree,
continue to obtain their financial services from
8

The HHI is calculated as the sum of the squared market shares of
each firm competing in a market—that is, HHI = Σi market share 2i ,
where there are i = 1, …, n firms in the market and market sharei
is the percentage of market output (deposits in the present context)
produced by the ith firm. Guidelines for the use of the HHI in antitrust enforcement are established by the DOJ
(www.justice.gov/atr/public/guidelines/6472.htm).

9

Regulators have defined some U.S. banking markets over larger
geographic areas, such as multiple counties, and occasionally
they redefine markets based on changes in commuting patterns,
trade areas, transportation networks, and so forth. Current definitions for all U.S. banking markets are available from the Federal
Reserve Bank of St. Louis (http://cassidi.stlouisfed.org/).

10

The relationship between concentration and competition is potentially ambiguous. For example, if barriers to entry and exit are
sufficiently low, then even a monopolist will not earn excess profits
in the long run because other firms will enter and drive down the
market price if the incumbent firm sets its price above marginal
cost (Baumol, Panzar, and Willig, 1988). See Berger et al. (2004) for
further discussion of the relationship between market concentration and competition and a review of recent research on the determinants and effects of concentration and competition in banking.

422

N OV E M B E R / D E C E M B E R

2011

banks located in their communities11 and (ii) the
structure of local banking markets continues to
affect the level of competition within those markets. For example, Hannan and Prager (2004) find
that banks that operate in a single MSA or nonMSA county market offer lower deposit interest
rates when those markets are more concentrated.
However, the study also finds that the relationship
between local concentration and deposit interest
rates is weaker in markets where the share of
banks operating in more than one market is higher.
Still, the authors conclude that market structure
continues to influence the competitive behavior
of banks operating in local markets.
Dick (2006) investigates whether the level of
bank concentration changed significantly between
1993 and 1999 across MSAs and Census regions
to assess the impact on banking market concentration of the removal of most restrictions on interstate branching in 1997. She finds that the mean
and median HHI values for MSAs declined slightly
between 1993 and 1999, whereas HHI values
increased for all nine Census regions, with the
percentage increases ranging from 17 percent in
the Pacific region to 421 percent in the South
Atlantic region.

Local Market Concentration
The patterns that Dick (2006) observes for
1993-99 continued in later years. Table 1 reports
summary information about the distribution of
HHI values across MSAs in 1999, 2006, and 2010.
The values reported in Panel A of Table 1 are based
on total deposits data for commercial banks only,
as in Dick (2006), whereas those reported in
Panel B are based on data for both commercial
banks and savings institutions.12 The information
in Panel A shows that both mean and median HHI
values declined by more than 100 points between
11

See Gilbert and Zaretsky (2003) for references to these studies.

12

All subsequent tables in this article are divided similarly: Information reported in Panel A is based on data for commercial banks
only, whereas information reported in Panel B is based on data for
both commercial banks and savings institutions. Bank regulators
usually weight the deposits of savings institutions by 0.5 in calculating HHI values to measure the concentration of banking markets.
However, this article assigns them full weight but also presents
results based on data that exclude savings institution deposits
altogether.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

Table 1
Descriptive Statistics for the HHI (MSAs)
Bank type

1999

2006

2010

361

361

366

Panel A (commercial banks)
No. of markets
Minimum

516

515

539

Maximum

8006

8346

6666

1911.8

1760.7

1703.8

1375

1255

1188

Mean
Q1
Median

1746

1588

1459

Q3

2198

1971

1902

905.92

867.02

860.39

361

361

366

Standard deviation
Panel B (commercial banks and savings institutions)
No. of markets
Minimum

374

408

488

Maximum

5726

8145

7247

1530.6

1527.1

1535.8

1139

1089

1089

Mean
Q1
Median

1439

1346

1318

Q3

1750

1705

1689

626.25

835.85

841.60

Standard deviation

NOTE: Q1 is the first quartile of the distribution of the data; Q3 is the third quartile. Five cities were defined as MSAs between June
30, 2006, and June 30, 2010: Lake Havasu (AZ) and Palm Coast (FL) were designated as MSAs in December 2006; Cape Girardeau
(MO), Manhattan (KS), and Mankato-North Mankato (MN) were designated as MSAs in November 2008.

1999 and 2006 for commercial banks, whereas
the information in Panel B shows that median
HHI values declined by 93 points and mean HHI
values declined by 4 points for commercial banks
and savings institutions.13 Thus, for commercial
banks, the decline in mean and median HHI values between 1993 and 1999 at the MSA level
noted by Dick (2006) continued through 2006.
Further, these trends also continued during 200710, despite the financial crisis and recession and
resulting wave of bank failures and mergers.
Table 2 reports similar information for nonMSA (i.e., “rural”) banking markets. Rural banking markets generally are more concentrated than
urban markets. For example, the median HHI
value for non-MSA counties in 2010 was 3195
(based on data for commercial banks only),
whereas the median HHI value for MSAs was
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

13

Many commercial banks and savings institutions are controlled
by bank (or thrift) holding companies, which may have a controlling interest in more than one bank in a given market. Bank regulators and the DOJ consider common control of multiple banks in
a market when evaluating proposed bank mergers. However, in
this article no adjustment is made for common control of multiple
banks in a market in calculating measures of market concentration,
which seems consistent with Dick’s (2006) approach. Although
failing to adjust for common ownership would tend to lead to
understatement of the HHI, on average, holding companies have
increasingly tended to merge their multiple bank subsidiaries into
a single bank, which lessens this bias in more recent years and,
more importantly, would tend to upwardly bias the unadjusted
changes in HHI over time. Hence, on average, increases in unadjusted HHI likely overstate the extent to which concentration has
increased. Since the observed increases in unadjusted HHI in local
banking markets have been small, on average, the average increase
in concentration taking account of common control of multiple
banks in a market would likely be even smaller.

N OV E M B E R /D E C E M B E R

2011

423

Wheelock

Table 2
Descriptive Statistics for the HHI (Non-MSA Rural Counties)
Bank type

1999

2006

2010

2023

2024

2015

Panel A (commercial banks)
No. of markets
Minimum

830

891

839

Maximum

10000

10000

10000

Mean

4032.6

3821.5

3791.6

2405

2268

2243

Median

3399

3199

3195

Q3

5054

4831

4740

2274.11

2187.70

2171.10

2027

2026

2017

Q1

Standard deviation
Panel B (commercial banks and savings institutions)
No. of markets
Minimum

739

735

704

Maximum

10000

10000

10000

Mean

3684.3

3587.8

3594.9

2143

2073

2126

Median

3010

2955

2965

Q3

4558

4392

4405

2236.58

2153.05

2144.41

Q1

Standard deviation

NOTE: Q1 is the first quartile of the distribution of the data; Q3 is the third quartile.

1459. However, as with MSA markets, mean and
median HHI values for rural markets declined
between 1999 and 2010. Thus, in mid-2010, the
mean and median concentrations of both MSA
and rural banking markets were substantially
lower than in 1999 (and in 1993 for MSA markets) even though there were far fewer banks and
savings institutions in the United States in 2010
than in either 1993 or 1999.

Regional Concentration
That the substantial reduction in the number
of banks in the United States from the 1990s
through 2010 did not increase the average concentration of local banking markets is consistent
with the active enforcement of antitrust policy
by bank regulators and the DOJ, whose officials
generally deny bank merger applications that
would substantially increase the concentration
424

N OV E M B E R / D E C E M B E R

2011

of local banking markets. However, antitrust
policy is not applied in banking over larger geographic areas, such as Census regions (though, as
noted previously, federal law prohibits individual banks from holding more than 10 percent of
total U.S. bank deposits, or 30 percent of a state’s
total deposits, if that level of deposits is obtained
through acquisitions of non-failed banks). Dick
(2006) finds that HHI values increased substantially between 1993 and 1999 for all nine U.S.
Census regions.
Table 3 reports HHI values for U.S. Census
regions for 1999, 2006, and 2010.14 HHI values
vary widely across U.S. Census regions. For 2010,
HHI values range from 341 for the East South
14

Dick’s (2006) data exclude savings institutions and rural market
deposits. By contrast, the information reported in Table 3 is based
on data that include both MSA and rural deposits. However, HHI
values and trends are not qualitatively different from those reported
in Table 3 if rural deposits are excluded from the analysis.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

Table 3
HHI Values (Census Regions)
Census region

1999

2006

2010

1419

1194

1377

Panel A (commercial banks)
New England
Middle Atlantic

577

914

997

East North Central

135

284

381

West North Central

167

712

554

South Atlantic

589

845

639

East South Central

233

297

341

West South Central

285

501

508

Mountain
Pacific

370

645

796

1295

1155

1183

Panel B (commercial banks and savings institutions)
New England

571

539

669

Middle Atlantic

371

618

718

East North Central

101

226

315

West North Central

145

631

492

South Atlantic

473

600

517

East South Central

215

277

319

West South Central

229

416

446

Mountain

310

723

650

Pacific

784

781

1069

NOTE: U.S. Census regions include the following states: New England (CT, ME, MA, NH, RI, VT); Middle Atlantic (NJ, NY, PA); East North
Central (IN, IL, MI, OH, WI); West North Central (IA, KS, MN, MO, NE, ND, SD); South Atlantic (DE, DC, FL, GA, MD, NC, SC, VA, WV);
East South Central (AL, KY, MS, TN); West South Central (AR, LA, OK, TX); Mountain (AZ, CO, ID, NM, MT, UT, NV, WY); and Pacific
(AK, CA, HI, OR, WA).

Central region to 1377 for the New England region
(Panel A). However, regional HHI values increased
between 1999 and 2010 in each region except the
New England and the Pacific regions, with the
largest increases occurring between 1999 and
2006. When savings institutions are included in
the analysis (Panel B), the interregional range of
HHI values was narrower. In addition, HHI values
rose between 1999 and 2010 in all regions. Thus,
regardless of whether savings institutions are
included in the analysis, HHI values increased
in most, if not all, regions, indicating increased
concentration at the regional level. Further, in
most regions, a higher percentage of the increase
in HHI values occurred during 1999-2006 than
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

during 2006-10. Thus, the financial crisis and
recession did not generally cause a substantial
increase in banking concentration, as reflected
in HHI values, at either local or regional levels.

DOMINANT AND FRINGE FIRMS
In addition to changes in market concentration, Dick (2006) also investigates changes over
time in the number of “dominant” and “fringe”
banks in urban and regional banking markets.
She defines dominant banks as the smallest set
of banks that jointly hold at least half of a market’s
total deposits. All other banks in a market are
fringe banks. Similarly, regionally dominant banks
N OV E M B E R /D E C E M B E R

2011

425

Wheelock

Table 4
Descriptive Statistics for the Number of Dominant Banks (MSAs)
Bank type

1999

2006

2010

361

361

366

Panel A (commercial banks)
No. of markets
Minimum

1

1

1

Maximum

8

7

7

2.7

2.9

3.0

Q1

Mean

2

2

2

Median

3

3

3

Q3

3

3

3

0.91

1.03

1.03

361

361

366

Standard deviation
Panel B (commercial banks and savings institutions)
No. of markets
Minimum

1

1

1

Maximum

11

9

7

Mean

3.2

3.3

3.2

Q1

3

3

3

Median

3

3

3

Q3

4

4

4

1.08

1.14

1.07

Standard deviation

NOTE: Q1 is the first quartile of the distribution of the data; Q3 is the third quartile. Five cities were defined as MSAs between June
30, 2006, and June 30, 2010: Lake Havasu (AZ) and Palm Coast (FL) were designated as MSAs in December 2006; Cape Girardeau (MO),
Manhattan (KS), and Mankato-North Mankato (MN) were designated as MSAs in November 2008.

are those that jointly hold at least half of a region’s
total deposits. Dick (2006) finds that most urban
markets had two or three dominant banks in both
1993 and 1999. Further, the average number of
fringe banks fell slightly (from 19 banks to 18
banks), but the median number of fringe banks
was 11 banks in both years.
Table 4 reports summary statistics on the
number of dominant banks across MSAs for 1999,
2006, and 2010.15 The mean and median number
of dominant banks, based on data for only commercial banks or for both commercial banks and
savings institutions, changed little between 1999
and 2010. The ranges also varied little across time.
15

As in calculating the HHI, this article makes no adjustments for
cases in which a single owner has a controlling interest in more
than one bank in a given market in calculating the number of
dominant banks in that market (see footnote 13).

426

N OV E M B E R / D E C E M B E R

2011

Table 5 shows the frequency distribution of the
number of dominant banks for each year. In 1999,
15 (of 361) MSAs had only one dominant bank
(Panel A). That number had increased slightly
by 2010, when 23 (of 366) MSAs had only one
dominant bank. However, the number of MSAs
with four or more dominant banks also increased
over time, from 48 (of 361) in 1999 to 86 (of 366)
in 2010.
As shown in Panel B of Table 4, the mean
number of dominant banks in MSA markets is
slightly larger if savings institutions are included
in the analysis, but the median remains at three
banks from 1999 to 2010 and the mean and median
numbers changed little between 1999 and 2010.
Furthermore, the number of markets with four or
more dominant banks increased from 107 (of 361)
in 1999 to 124 (of 366) in 2010 (see Table 5). Hence,
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

Table 5
Distribution of the Number of Dominant Banks (MSAs)
2006

1999

2010

Frequency

%

Frequency

%

Frequency

%

15

4.2

19

5.3

23

6.3

Panel A (dominant commercial banks)
1
2

157

43.5

117

32.4

88

24.0

3

141

39.1

149

41.3

169

46.2

4

35

9.7

53

14.7

62

16.9

5

9

2.5

17

4.7

16

4.4

6

3

0.8

3

0.8

7

1.9

7

—

—

3

0.8

1

0.3

1

0.3

—

—

—

—

8
Total MSAs

361

361

366

Panel B (dominant commercial banks and savings institutions)
1

7

1.9

15

4.2

15

4.1

2

83

23.0

66

18.3

69

18.9

3

164

45.4

153

42.4

158

43.2

4

76

21.1

83

23.0

84

23.0

5

22

6.1

34

9.4

30

8.2

6

5

1.4

6

1.7

9

2.5

7

3

0.8

2

0.6

1

0.3

8

—

—

1

0.3

—

—

9

—

—

1

0.3

—

—

10

—

—

—

—

—

—

1

0.3

—

—

—

—

11
Total MSAs

361

361

366

NOTE: Five cities were defined as MSAs between June 30, 2006, and June 30, 2010: Lake Havasu (AZ) and Palm Coast (FL) were designated
as MSAs in December 2006; Cape Girardeau (MO), Manhattan (KS), and Mankato-North Mankato (MN) were designated as MSAs in
November 2008.

the results indicate that the decline in the number of banks in the United States since 1999 has
not caused the number of dominant banks in
most MSA banking markets to fall.
Rural (non-MSA) banking markets tend to be
more concentrated than urban banking markets.
Furthermore, Wheelock (2011) finds that acquisitions of failed banks by in-market competitors
resulted in substantial increases in concentration
in some rural banking markets during 2007-10
but no significant increases in any large urban
markets. Table 6 reports information on the numF E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

ber of dominant banks in rural markets in 1999,
2006, and 2010. The mean and median numbers
of dominant banks in rural markets are smaller
than those of MSA markets, reflecting the tendency toward greater deposit concentration of
rural banking markets. However, as with MSAs,
the distributions of dominant banks in rural
markets changed little between 1999 and 2010
(Table 7). Thus, as reflected in both HHI values
and the distributions of dominant banks, and
regardless of whether savings institutions are
included in the analysis, the market structure of
N OV E M B E R /D E C E M B E R

2011

427

Wheelock

Table 6
Descriptive Statistics for the Number of Dominant Banks (Non-MSA Rural Counties)
1999

2006

2010

Panel A (commercial banks)
No. of markets
Minimum
Maximum
Mean
Q1
Median
Q3
Standard deviation

2023
1
5
1.7
1
2
2
0.69

2024
1
5
1.8
1
2
2
0.73

2015
1
5
1.8
1
2
2
0.74

Panel B (commercial banks and savings institutions)
No. of markets
Minimum
Maximum
Mean
Q1
Median
Q3
Standard deviation

2027
1
6
1.9
1
2
2
0.77

2026
1
5
1.9
1
2
2
0.80

2017
1
6
1.9
1
2
2
0.78

NOTE: Q1 is the first quartile of the distribution of the data; Q3 is the third quartile.

Table 7
Distribution of the Number of Dominant Banks (Non-MSA Rural Counties)
1999
Frequency

2006

2010

%

Frequency

%

Frequency

%

Panel A (dominant commercial banks)
1
2
3
4
5
Total rural markets

835
962
202
22
2
2023

41.28
47.55
9.99
1.09
0.10

780
960
247
34
3
2024

38.54
47.43
12.20
1.68
0.15

788
947
242
35
3
2015

39.11
47.00
12.01
1.74
0.15

Panel B (dominant commercial banks and savings institutions)
1
2
3
4
5
6
Total rural markets

699
979
298
46
3
2
2027

34.48
48.30
14.70
2.27
0.15
0.10

694
927
354
40
11
—
2026

34.25
45.76
17.47
1.97
0.54
—

705
951
306
49
5
1
2017

34.95
47.15
15.17
2.43
0.25
0.05

428

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

Table 8
Number of Regional Dominant Banks by Census Region
Census region

1999

2006

2010

Panel A (commercial banks)
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific

2
7
25
58
8
14
16
9
3

3
5
13
20
4
13
8
6
3

3
5
9
11
5
15
8
5
3

Panel B (commercial banks and savings institutions)
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific

9
11
35
64
13
16
20
12
5

8
7
17
27
7
15
11
5
5

6
6
12
15
7
16
9
6
4

most local U.S. banking markets did not change
substantially between 1999 and 2010, despite
continued consolidation of the banking industry
as a whole.16

Regionally Dominant Banks
Dick (2006) finds that the number of regionally dominant banks declined by an average of
55 percent across Census regions from 1993 to
1999. Table 8 reports on the number of regionally
dominant banks in each Census region for 1999,
2006, and 2010. As shown in the table, the pattern
identified by Dick (2006) continued over the subsequent decade in most regions, especially when
savings institutions are included in the analysis
(Panel B).17 The decline in the number of regionally dominant banks was especially pronounced
16

Similarly, the average number of fringe banks did not change substantially over time in either MSA or rural banking markets.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

17

Dick (2006) does not include savings institutions in her analysis.
She apparently also includes only banks located in MSAs in her
analysis of regionally dominant banks. The exclusion of rural banks
and bank branches has a larger impact on the calculation of the
number of regionally dominant banks in regions with higher percentages of deposits held outside MSAs. The West North Central
region had the highest percentage of bank deposits held outside
MSAs in 1999 at 38 percent. If rural banks are excluded from the
analysis, the West North Central region had 13, rather than 58,
regionally dominant banks in 1999 (the omission of savings institutions has a much smaller impact on the number of regionally
dominant banks). By contrast, in regions with a high percentage
of deposits held in MSAs, the omission of rural deposits has a much
smaller impact on the number of regionally dominant banks. The
Pacific region had the smallest percentage of deposits held outside
MSAs in 1999. If rural banks are omitted from the analysis, the
number of regionally dominant banks remains three.

N OV E M B E R /D E C E M B E R

2011

429

Wheelock

Table 9
MSA and Regionally Dominant Banks by Census Region
Mean No. of
MSA-dominant banks
Census region

Mean No. of
MSA- and regionally
dominant banks in MSAs

1999

2006

2010

1999

2006

2010

1.8

1.8

1.9

0.8

1.1

1.3

Panel A (commercial banks)
New England
Middle Atlantic

2.6

2.5

2.5

0.9

0.6

0.8

East North Central

2.6

2.9

3.0

1.2

1.3

1.5

West North Central

3.2

3.6

3.6

2.1

1.9

1.6

South Atlantic

2.8

2.9

3.0

1.4

1.5

1.8

East South Central

3.0

3.4

3.2

1.9

2.1

2.2

West South Central

2.9

3.0

3.0

1.8

1.4

1.5

Mountain

2.4

2.7

2.6

1.4

1.6

1.2

Pacific

2.4

2.6

3.0

1.2

1.4

1.7

Panel B (commercial banks and savings institutions)
New England

2.8

3.1

2.9

1.2

1.5

1.6

Middle Atlantic

3.3

3.4

3.2

1.4

0.9

0.9

East North Central

3.2

3.1

3.2

1.6

1.7

1.7

West North Central

3.6

3.8

3.9

2.4

2.1

1.8

South Atlantic

3.1

3.2

3.3

1.6

1.6

1.8

East South Central

3.2

3.5

3.4

2.0

2.3

2.4

West South Central

3.2

3.1

3.1

1.9

1.5

1.4

Mountain

2.8

2.9

2.8

1.7

1.4

1.3

Pacific

3.2

3.3

3.1

1.6

2.0

1.8

in the West North Central region, where the number of regionally dominant banks fell from 58 in
1999 to 20 in 2006, and to just 11 in 2010 (Panel A).
That is, in 1999, the largest 58 banks together held
50 percent of the West North Central region’s
deposits, but in 2006 the largest 20 banks held
50 percent of the region’s deposits, and in 2010
the largest 11 banks held 50 percent of the region’s
deposits. The number of regionally dominant
banks also fell substantially between 1999 and
2010 in the East North Central, South Atlantic,
and West South Central regions. Many states in
these four regions had prohibited or severely
restricted branching within their borders and
were among the last states to loosen their branching laws before the Riegle-Neal Interstate Banking
430

N OV E M B E R / D E C E M B E R

2011

and Branching Efficiency Act was enacted in
1994. Consequently, those states tended to have
large numbers of small- and medium-sized banks
and experienced more consolidation of their
banking systems during the 1990s than did many
states in the New England, Middle Atlantic, and
Pacific regions, which had long been more open to
statewide branching and were among the first to
enter into regional interstate banking compacts.18
Notably, even in regions with more regionally
dominant banks in 1999, the decline in the number of regionally dominant banks during 199918

Kroszner and Strahan (1999) and Garrett, Wagner, and Wheelock
(2005) investigate the determinants of the timing of state deregulation of branching and interstate banking laws in the 1970s, 1980s,
and 1990s.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

2006 generally was much larger than during
2006-10. For example, the number of regionally
dominant banks in the West South Central region
fell from 16 to 8 between 1999 and 2006 but was
still 8 in 2010 (based on data for commercial
banks only). Hence, the financial crisis and recession of 2007-09 apparently did not add momentum
to the ongoing trend toward greater concentration
of a region’s bank deposits in fewer banks.
Table 9 provides further information about
patterns of regionally dominant banks across
regions and over time. The table reports regional
averages for MSAs in 1999, 2006, and 2010 on (i)
the number of MSA-dominant banks and (ii) the
number of banks that are dominant in both the
MSA and its region.19 As Dick (2006) finds for
1993 and 1999, the average number of dominant
banks varies more across regions than it does
across time within regions. For example, based
on data for commercial banks only, MSAs in the
New England region had an average of 1.8 dominant banks in 1999 and 2006 and 1.9 dominant
banks in 2010, whereas MSAs in the West North
Central region had an average of 3.2, 3.6, and
3.6 dominant banks in 1999, 2006, and 2010,
respectively.
Dick (2006) also finds that the mean number
of banks dominant in both an MSA and its region
increased between 1993 and 1999. However, as
shown in Table 9, that trend did not continue
past 1999. The mean number of banks dominant
at both the MSA and regional levels changed little
between 1999 and 2010 in most regions, regardless of whether savings institutions are included
in the analysis. Again, there was more variation
across regions than over time. MSAs in the New
England and Middle Atlantic regions tended to
have the smallest numbers of banks that were
dominant in both the MSA and its region, whereas
19

Banks that are dominant both within an MSA and within the region
in which the MSA is located are banks that are (i) among the group
of banks holding at least 50 percent of the deposits of the MSA and
(ii) among the group of banks holding at least 50 percent of the
deposits of the region. For example, in 2010, the largest five banks
in the St. Louis MSA held just over 50 percent of the MSA’s deposits.
Four of those banks were among the largest 11 banks that together
held just over 50 percent of the deposits of the West North Central
Census region. Hence, there were four banks in the St. Louis MSA
that were dominant in both the MSA and Census region.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

the East South Central region generally had the
highest average number of such banks.

IMPACT OF MERGERS ON
MARKET CONCENTRATION
(2007-10)
The recent financial crisis and recession led
to a wave of bank failures and mergers that contributed to the ongoing consolidation of the U.S.
banking industry. As shown previously, the average concentration of local banking markets did
not increase during 2006-10. This section examines the impact on specific banking markets of
unassisted bank mergers during those years.
Wheelock (2011) finds that acquisitions of failed
banks by in-market competitors (i.e., banks that
already had branches in the markets served by
the failed bank) during 2007-10 did not substantially increase concentration in most local banking markets. However, such acquisitions had a
substantial impact in a few, mostly rural, banking
markets. This section examines the impact on
market concentration of acquisitions of non-failed
banks by in-market competitors during those years.
Several large unassisted mergers involving
banks operating in the same local markets occurred
during 2007-10. Table 10 lists the 10 largest
unassisted bank mergers during 2007-10, ranked
by the total deposits held by the acquired institution as of the most recent June 30 before the merger.
For example, Wachovia Bank NA, which merged
with Wells Fargo Bank NA in March 2010, held
$394 billion of deposits on June 30, 2009.
Although Wachovia Bank NA merged with
Wells Fargo Bank NA in March 2010, the Board of
Governors of the Federal Reserve System approved
the application of Wells Fargo & Company to
acquire Wachovia Corporation and its subsidiaries,
including Wachovia Bank NA, on October 12,
2008. Wachovia Bank and Wells Fargo Bank had
offices in common in several banking markets in
Arizona, California, Colorado, Nevada, and Texas.
In evaluating the competitive implications of an
acquisition of Wachovia Corporation by Wells
Fargo, the Board of Governors used deposit and
market share data for June 30, 2007 (adjusted to
N OV E M B E R /D E C E M B E R

2011

431

Wheelock

432
N OV E M B E R / D E C E M B E R

Table 10

2011

Ten Largest Unassisted Bank Mergers (2007-10)

Acquired bank

Acquiring bank

Date of holding company
acquisition approval

Date of merger

Total deposits of
acquired bank
($ thousands)

Total deposits of
acquiring bank
($ thousands)
325,417,000

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wachovia Bank, NA

Wells Fargo Bank, NA

October 12, 2008

March 20, 2010

394,189,000

National City Bank

PNC Bank, NA

December 15, 2008

November 7, 2009

101,141,375

84,171,396

Wachovia Mortgage, FSB

Wachovia Bank, NA

September 29, 2006

October 12, 2007

73,243,232

314,850,000

Countrywide Bank, FSB

Bank of America, NA

June 5, 2008

April 27, 2009

63,336,672

642,252,215

Fifth Third Bank*

Fifth Third Bank

March 12, 2001

September 30, 2009

41,454,606

31,948,335

Commerce Bank, NA

TD Bank, NA

March 13, 2008

June 1, 2008

40,126,588

28,092,910

North Fork Bank

Capital One, NA

November 8, 2006

August 1, 2007

38,059,484

20,567,194

LaSalle Bank, NA

Bank of America, NA

September 14, 2007

October 17, 2008

29,594,901

642,252,215

Merrill Lynch Bank & Trust Co., FSB Bank of America, NA

November 26, 2008

November 2, 2009

28,965,596

817,989,321

LaSalle Bank Midwest, NA

September 14, 2007

October 17, 2008

25,011,471

642,252,215

Bank of America, NA

NOTE: *Fifth Third Bank (Grand Rapids) and Fifth Third Bank (Cincinnati) were both subsidiaries of the Fifth Third Financial Corporation when they merged under the charter
of Fifth Third Bank (Cincinnati) in 2009. Hence, Fifth Third Bank (Cincinnati) is listed as the acquirer and Fifth Third Bank (Grand Rapids) as the acquired bank.

Wheelock

reflect mergers and acquisitions through October 3,
2008).20 Wachovia Bank had only small shares
of most MSA banking markets in states where
Wells Fargo Bank operated, and consequently
the proposed acquisition would have had little
impact on concentration in most markets. However, Wachovia Bank and Wells Fargo Bank both
had significant market shares in a few small MSA
and rural banking markets. For example, they had
the two largest market shares in the Santa Cruz,
California, banking market with 27 percent
(Wachovia) and 19 percent (Wells Fargo) shares.
However, in their application to acquire Wachovia,
Wells Fargo proposed to divest one of Wachovia
Bank’s branches in the Santa Cruz market to an
out-of market depository institution. Further, in
evaluating the competitive implications of the
proposed merger, the Board of Governors noted
the presence of several other banks and credit
unions with significant market shares in the
Santa Cruz market, as well as the recent entry into
the market of two other depository institutions.
Hence, the Board determined that the merger
would not adversely harm competition in the
Santa Cruz market.21 The Board of Governors
made similar determinations about the few other
banking markets where both Wachovia Bank and
Wells Fargo had relatively large market shares.
Although both Wachovia Bank NA and Wells
Fargo Bank NA were controlled by Wells Fargo &
Company when the banks were formally merged
in March 2010, an indication of the impact of the
merger on concentration in the Santa Cruz and
other markets is obtained by comparing deposits
and market share data for June 30, 2009, and
June 30, 2010. Based on data for June 30, 2009,
the HHI value for the Santa Cruz MSA was 1295,
indicating that the market was moderately concentrated by DOJ guidelines. Had Wachovia Bank
20

The statement by the Board of Governors regarding the application
by Wells Fargo & Company to acquire Wachovia Corporation and
Wachovia’s subsidiary banks and non-banking companies is available on the Board’s website (www.federalreserve.gov/newsevents/
press/orders/orders20081021a1.pdf).

21

Credit unions are not required to report branch-level deposits data
and, hence, ordinarily they are excluded from calculation of market concentration measures, such as the HHI. However, the Board
may consider the presence of credit unions in a market when
evaluating applications for bank mergers.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

and Wells Fargo Bank been merged as of that date,
and assuming no other differences in the market,
the HHI value would have been 1832. Hence, a
simple pro forma analysis suggests that the merger
would produce a larger increase in market concentration than permitted by DOJ guidelines.
However, the divestiture of one of Wachovia
Bank’s branches in the Santa Cruz market reduced
the impact of the merger on market concentration.
In the event, the HHI value for the Santa Cruz
market rose by only 202 points, from 1295 to
1497, between June 30, 2009, and June 30, 2010,
and hence the market remained only moderately
concentrated.22
The acquisition of National City Bank by PNC
Bank NA in November 2009 was the secondlargest merger in terms of total deposits of the
acquired bank during 2007-10. The Board of
Governors approved the application of The PNC
Financial Services Group, Inc. (the parent company of PNC Bank NA) to acquire National City
Corporation (the parent of National City Bank) on
December 15, 2008. The Board relied on deposit
and market share data as of June 30, 2008 (adjusted
to reflect mergers and acquisitions through
November 4, 2008) to evaluate the competitive
implications of the acquisition on individual
banking markets.
National City Bank and PNC Bank competed
directly in 10 banking markets in Florida,
Kentucky, Ohio, and Pennsylvania, and both had
substantial shares of the Erie and Pittsburgh,
Pennsylvania, markets. In its merger application,
PNC proposed to divest several National City
Bank branches in both the Pittsburgh and Erie
markets. In addition, the Board of Governors determined that a substantial portion of the deposits
held by PNC Bank in Pittsburgh were deposits of
customers located outside the Pittsburgh market,
including various municipalities and governments, and escrow accounts for mortgages and
other transactions outside the market. Consequently, the Board determined that, in effect,
PNC had a lower effective share of the Pittsburgh
22

The change in HHI from one year to the next reflects all transactions that occurred in a market during the year, not just the merger
of Wachovia Bank and Wells Fargo Bank, as well as fluctuations in
market shares associated with other deposit inflows and outflows.

N OV E M B E R /D E C E M B E R

2011

433

Wheelock

Table 11
Descriptive Statistics for the HHI in Overlapping Markets for Bank Mergers (2007-10)
Non-MSA rural counties

MSAs
HHI before*

HHI after†

Difference‡

HHI before*

HHI after†

Difference‡

Panel A (Jan. 1, 2007–June 30, 2007)
No. of markets

51

51

51

34

34

34

Minimum

408.48

384.15

–223.25

1121.30

1092.81

–2163.01

Maximum

4625.70

4402.45

635.33

6147.01

4265.75

1727.43

Mean

1296.11

1327.00

30.89

2134.62

2358.65

224.03

910.27

891.98

–49.77

1579.61

1760.57

46.96

Q1
Median

1155.46

1114.97

–10.25

1934.33

2204.36

127.08

Q3

1495.06

1486.63

65.40

2538.31

2774.86

321.55

659.00

705.13

151.71

935.73

899.10

598.69

116

116

116

47

47

47

Standard deviation
Panel B (July 1, 2007–June 30, 2008)
No. of markets
Minimum

384.15

445.02

–244.30

977.23

1007.49

–272.30

Maximum

6559.93

6504.27

1945.82

4590.82

5691.28

2145.23

Mean

1312.70

1432.69

119.99

2124.46

2351.43

226.97

938.47

957.36

–0.06

1334.08

1493.10

30.26

Q1
Median

1165.13

1220.19

48.21

1799.53

1823.53

111.92

Q3

1450.68

1606.12

140.10

2773.81

2859.65

227.27

753.37

855.27

298.28

968.02

1123.69

430.44

66

66

66

31

31

31

Standard deviation
Panel C (July 1, 2008–June 30, 2009)
No. of markets
Minimum

445.02

439.99

–867.48

1113.50

1092.08

–156.84

Maximum

5631.98

5768.96

4684.56

4078.06

7956.74

3878.67

Mean

1398.88

1474.41

75.53

1971.28

2309.60

338.32

889.74

922.43

–65.89

1414.71

1493.24

2.68

Median

1189.65

1190.35

4.73

1843.44

2074.01

148.49

Q3

1489.83

1522.02

49.27

2254.19

2550.92

374.04

933.90

1050.60

611.76

723.49

1331.83

721.43

78

78

78

21

21

21

Q1

Standard deviation
Panel D (July 1, 2009–June 30, 2010)
No. of markets
Minimum

439.99

488.17

–2289.39

943.52

992.73

–421.80

Maximum

6802.87

7246.76

1551.94

3909.51

7410.51

3501.00

Mean

1462.61

1555.52

92.91

2100.92

2540.99

440.06

939.94

1033.07

4.67

1429.57

1542.64

81.51

Q1
Median

1230.18

1296.63

103.94

1892.44

1896.29

174.93

Q3

1441.67

1615.89

225.36

2333.82

2963.18

470.81

Standard deviation

1016.31

1074.24

397.97

840.29

1516.96

816.02

434

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

Table 11, cont’d
Descriptive Statistics for the HHI in Overlapping Markets for Bank Mergers (2007-10)
Non-MSA rural counties

MSAs
HHI before*

HHI after†

Difference‡

HHI before*

HHI after†

Difference‡

Panel E (Jan. 1, 2007–June 30, 2010)
No. of markets

311

311

311

133

133

133

Minimum

384.15

384.15

–2289.39

943.52

992.73

–2163.01

Maximum

6802.87

7246.76

4684.56

6147.01

7956.74

3878.67

Mean

1365.87

1455.02

89.15

2087.64

2373.46

285.82

925.60

955.38

–25.91

1429.57

1542.64

38.49

Median

1177.98

1247.80

31.20

1837.30

2062.67

134.45

Q3

1476.66

1600.65

146.01

2531.96

2719.63

301.14

851.16

935.48

394.31

880.71

1183.03

616.44

Q1

Standard deviation

NOTE: The summary statistics exclude all market overlaps for mergers that occurred after June 30, 2010. *“HHI before” corresponds to
the HHI value on the June 30 before the merger date; †“HHI after” corresponds to the HHI value on the June 30 after the merger date;
‡“Difference” corresponds to the change between the HHI value from the June 30 before the merger date and the HHI value on the
June 30 after the merger date.

banking market than suggested by the Summary
of Deposits data used in calculating market HHI
values.23 Furthermore, the Board noted that a
large number of banks (57) would remain in the
Pittsburgh market after the merger of PNC and
National City, and that 6 banking organizations
had entered the Pittsburgh market during the
previous 4 years. Similarly, considering the proposed branch divestitures and the presence of
several other competitors, including four community credit unions, the Board determined that
competition in the Erie market would not be
adversely affected by PNC’s acquisition of
National City Bank.24
Between January 1, 2007, and June 30, 2010,
unassisted bank mergers occurred in 311 MSAs
and 133 rural counties where both merger part23

Many banking organizations book the deposits of out-of-market
customers at their headquarters location, which distorts market
share and HHI values as measures of local market concentration,
and is one reason why the Board of Governors and DOJ consider
other indicators of market competition in addition to HHI values
when evaluating bank merger applications.

24

The order approving the merger of acquisition of National City
Corporation by PNC Financial Services Group, Inc. is available on
the Board of Governor’s website (http://www.federalreserve.gov/
newsevents/press/orders/orders20081215a1.pdf).

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

ners had existing offices. Table 11 presents summary information about changes in HHI values
from the most recent June 30 before a merger to
the June 30 immediately following the merger in
those markets.25 Thus, Panel A of the table reports
summary statistics for HHI values on June 30,
2006, and June 30, 2007, and the difference in HHI
values between those dates, for markets where
mergers occurred between January 1, 2007, and
June 30, 2007. Panel B reports summary statistics
for HHI values on June 30, 2007, and June 30,
2008, and the difference in HHI values between
those dates, for markets where mergers occurred
between July 1, 2007, and June 30, 2008. Panels C
and D report similar information for markets
where mergers occurred between July 1, 2008,
and June 30, 2009, and between July 1, 2009,
and June 30, 2010, respectively. Panel E presents
summary statistics for all markets where mergers
occurred between January 1, 2007, and June 30,
2010.
The Board of Governors approved many of the
bank mergers that occurred between January 1,
25

Branch-level deposits data for June 30, 2011, are not yet available
to calculate changes in HHI values in markets in which mergers
occurred between July 1 and December 31, 2010.

N OV E M B E R /D E C E M B E R

2011

435

Wheelock

2007, and June 30, 2007, in 2006 or before. In
doing so, the Board would have relied on deposits
and market share information from before June 30,
2006, when evaluating the competitive implications of those mergers. Nonetheless, comparison
of HHI values based on data for June 30, 2006, and
June 30, 2007, shows how mergers that occurred
between January 1, 2007, and June 30, 2007,
affected market concentration, regardless of when
those mergers were approved. Similarly, comparison of HHI values based on data for June 30, 2007,
and June 30, 2008, shows the impact of mergers
that occurred between July 1, 2007, and June 30,
2008, on market concentration, regardless of when
the Board of Governors approved those mergers.
Hence, the data underlying the summary information reported in Table 11 include mergers that
were consummated during the period indicated,
regardless of when the mergers were approved.
As reported in Panel E, for MSAs, for the
entire period January 1, 2007, through June 30,
2010, the mean and median changes in HHI values over the 12-month periods during which one
or more bank mergers occurred were 89 and 31
points, respectively. For individual years, the
mean (median) changes range from 31 points (–10
points) to 120 points (104 points). The range of
changes in HHI values was very wide, from –2289
points to 4685 points across all MSAs where one
or more unassisted mergers of banks occurred
between January 1, 2007, and June 30, 2010. Of
course, mergers are just one cause of changes in
HHI values from one year to the next. Other reasons for changes in HHI values include bank failures, de novo entry, reassignment of deposits
among a bank’s branches, and other changes in
the distribution of deposits across banks not
associated with mergers.
For rural counties over the entire period, the
mean and median changes in HHI values were
286 and 134 points, respectively. For individual
years, the mean (median) changes range from
224 (112) points to 440 (175) points. Hence, HHI
values tended to increase more in rural counties
where mergers occurred than in MSAs. In general,
rural banking markets are more concentrated than
urban markets. Among rural counties where
mergers occurred, the mean HHI value before a
436

N OV E M B E R / D E C E M B E R

2011

merger was 2088 points, compared with 1366
points in MSAs. Many rural banking markets span
more than one county, however, and as noted
previously, banking regulators may consider the
presence of credit unions and other factors that
are not reflected in HHI values when evaluating
the competitive effects of proposed bank mergers.
Nonetheless, it appears that, on average, unassisted bank mergers during 2007-10 had a larger
impact on concentration in rural counties than
in MSA banking markets; Wheelock (2011) finds
a similar result for mergers involving failed banks.

CONCLUSION
The number of U.S. commercial banks and
savings institutions declined by 1,011, or about
12 percent, between December 31, 2006, and
December 31, 2010. Unassisted mergers of nonfailed banks eliminated 1,002 banks during this
period, whereas failures eliminated 324 banks
(the chartering of new banks, voluntary liquidations, and other changes resulted in a net addition
of 315 banks). The consolidation of the banking
industry during 2007-10 continued a trend begun
in the mid-1980s. Advances in informationprocessing and other technologies and the resulting economies of scale have encouraged growth
in the size of banks, which deregulation of bank
branching, first by states and later by the federal
government, has facilitated.26
Banking industry consolidation has been
marked by sharply higher shares of U.S. bank
deposits held by the largest banks, as well as
increased concentration of deposits measured
at the level of U.S. Census regions. This article
extends prior research on the structure of U.S.
banking markets by investigating changes in
deposit concentration at both the local and
regional levels. It shows that trends toward
increased concentration at the regional level in
the 1990s continued through 2010. However, concentration of local banking markets has changed
26

Berger (2003) discusses the implications of technological progress
for the banking industry, whereas Hughes, Mester, and Moon (2001)
and Wheelock and Wilson (forthcoming) report evidence of significant economies of scale in banking.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Wheelock

little over time, even during the recent financial
crisis and recession when several large bank
mergers occurred. Further, the average number
of banks holding at least 50 percent of deposits
in a region declined over time in most U.S. Census
regions, but the number holding at least 50 percent of deposits in local banking markets remained
fairly constant over time. Antitrust policy is predicated on the assumption that banking markets
are local in nature, and enforcement has helped
keep local banking markets from becoming significantly more concentrated.
The article also examines the effects on local
market concentration of mergers of banks operating in the same markets. Two of the largest mergers during 2007-10 were the merger of Wachovia
Bank with Wells Fargo Bank and the merger of
National City Bank with PNC Bank. In approving
these applications, the Board of Governors of the
Federal Reserve System noted plans to divest
local branch offices and other mitigating circumstances that offset pro forma analysis of market
concentration levels based on the HerfindahlHirschman Index. Further, the article finds that
deposit concentration did not increase to the
extent predicted by simple pro forma analysis in
markets where these mergers had raised the most
serious concerns about their competitive effects.

Finally, the article finds that deposit concentration did not increase substantially, on average,
in local banking markets where any unassisted
mergers occurred during 2007-10, though rural
counties generally saw larger average increases
in concentration than urban markets.
Changes in regulation and technology have
reduced the cost of obtaining banking services
from distant banks. However, many consumers
continue to rely exclusively on local banks for
financial services and evidence suggests that the
pricing of banking services continues to reflect,
at least in part, the structure of local banking markets. The recent financial crisis and recession did
not alter the trend toward industry consolidation
or change patterns of concentration at either the
local or regional levels. Antitrust enforcement
has ensured that the structures of local banking
markets have not changed significantly as a result
of unassisted mergers and acquisitions, even as
the industry as a whole has consolidated and total
U.S. deposits have become increasingly concentrated among the very largest banks. As technology
evolves and the costs of obtaining banking services from distant providers fall further, however,
local market characteristics may become less relevant for analysis of competition in banking.

REFERENCES
Amel, Dean F. “Trends in the Structure of Federally Insured Depository Institutions, 1984-94.” Federal Reserve
Bulletin, January 1996, 82(1), pp. 1-15.
Baumol, William J.; Panzar, John C. and Willig, Robert D. Contestable Markets and the Theory of Industry
Structure. Revised Edition. San Diego: Harcourt Brace Jovanovich, 1988.
Berger, Allen N. “The Economic Effects of Technological Progress: Evidence from the Banking Industry.”
Journal of Money, Credit, and Banking, April 2003, 35(2), pp. 141-76.
Berger, Allen N.; Demirgüç-Kunt, Asli; Levine, Ross and Haubrich, Joseph G. “Bank Concentration and
Competition: An Evolution in the Making.” Journal of Money, Credit, and Banking, June 2004, 36(3 Part 2),
pp. 433-51.
Dick, Astrid A. “Nationwide Branching and Its Impact on Market Structure, Quality, and Bank Performance.”
Journal of Business, March 2006, 79(2), pp. 567-92.
Garrett, Thomas A.; Wagner, Gary A. and Wheelock, David C. “A Spatial Analysis of State Banking Regulation.”
Papers in Regional Science, November 2005, 84(4), pp. 575-95.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

437

Wheelock

Gilbert, R. Alton and Zaretsky, Adam M. “Banking Antitrust: Are the Assumptions Still Valid?” Federal Reserve
Bank of St. Louis Review, November/December 2003, 85(6) pp. 29-52;
http://research.stlouisfed.org/publications/review/03/11/gilbert.pdf.
Hannan, Timothy H. and Prager, Robin A. “The Competitive Implications of Multimarket Bank Branching.”
Journal of Banking and Finance, August 2004, 28(8), pp. 1889-914.
Hughes, Joseph P.; Mester, Loretta J. and Moon, C.-G. “Are Scale Economies in Banking Elusive or Illusive?
Evidence Obtained by Incorporating Capital Structure and Risk-Taking into Models of Bank Production.”
Journal of Banking and Finance, December 2001, 25(12), pp. 2169-208.
Kroszner, Randall S. and Strahan, Philip E. “What Drives Deregulation? Economics and Politics of the Relaxation
of Bank Branching Restrictions.” Quarterly Journal of Economics, November 1999, 114(4), pp. 1437-67.
Spong, Kenneth. Banking Regulation: Its Purpose, Implementation and Effects. Fifth Edition. Kansas City, MO:
Federal Reserve Bank of Kansas City, 2000; www.financialpolicy.org/regulation/spong.pdf.
Wheelock, David C. “Have Acquisitions of Failed Banks Increased the Concentration of U.S. Banking Markets?“
Federal Reserve Bank of St. Louis Review, May/June 2011, 93(3), pp. 155-68;
http://research.stlouisfed.org/publications/review/11/05/155-168Wheelock.pdf.
Wheelock, David C. and Wilson, Paul W. “Do Large Banks Have Lower Costs? New Estimates of Returns to
Scale for U.S. Banks.” Journal of Money, Credit, and Banking, forthcoming;
http://web.econ.ohio-state.edu/jmcb/jmcb/09567/09567.pdf.
Yildirim, H. Semih and Mohanty, Sunil K. “Geographic Deregulation and Competition in the U.S. Banking
Industry.” Financial Markets, Institutions & Instruments, 2010, 19(2), pp. 63-94.

438

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

The Effectiveness of Unconventional Monetary
Policy: The Term Auction Facility
Daniel L. Thornton
This paper investigates the effectiveness of one of the Federal Reserve’s unconventional monetary
policy tools, the term auction facility (TAF). At issue is whether the TAF reduced the spread
between the London interbank offered rate (LIBOR) rates and equivalent-term Treasury rates by
reducing the liquidity premium embedded in LIBOR rates. This paper suggests that rather than
reducing the liquidity premium in LIBOR rates, the announcement of the TAF increased the risk
premium in financial and other bond rates because market participants interpreted the announcement by the Fed and other central banks as a sign that the financial crisis was worse than previously thought. Evidence is presented that supports this hypothesis. (JEL E52, E58, G14)
Federal Reserve Bank of St. Louis Review, November/December 2011, 93(6), pp. 439-53.

BACKGROUND

T

he Federal Reserve’s actions in the wake of
the financial crisis have spurred research
into the effectiveness of unconventional
monetary policy. One unconventional policy
that has received considerable attention is the
term auction facility (TAF). At issue is whether
the TAF reduced the spread between the London
interbank offered rate (LIBOR) rates and equivalent-term Treasury or overnight indexed swap
(OIS) rates. The Fed introduced the TAF based
on the belief that the increase in the spreads
between term LIBOR rates and equivalent-term
Treasury or OIS rates at the onset of the financial
crisis was due to an increase in the liquidity
premium in the interbank market. In announcing the TAF the Fed noted that, by allowing the
Federal Reserve to inject term funds through a
broader range of counterparties and against a
broader range of collateral than traditional open
market operations, this facility could help promote the efficient dissemination of liquidity
when the unsecured interbank markets are under

stress.1 In testimony before Congress on January
17, 2008, Chairman Bernanke (2008) indicated
that the goal of the TAF was to reduce the incentive for banks to hoard cash and increase their
willingness to provide credit to households
and firms. That is, the Fed believed banks were
hoarding liquidity. Consequently, the increase
in the LIBOR spreads was a result of an increase
in a liquidity premium that banks were requiring
to lend in the interbank market. Christensen,
Lopez, and Rudebusch (2009, p. 2; hereafter CLR)
summarize the intended effectiveness of the TAF:
In theory, the provision of central bank liquidity could lower the liquidity premium on interbank debt through a variety of channels. On
the supply side, banks that have a greater assurance of meeting their own unforeseen liquidity
needs over time should be more willing to
extend term loans to other banks. In addition,
creditors should also be more willing to provide funding to banks that have easy and
dependable access to funds, since there is a
1

Board of Governors of the Federal Reserve System (2007).

Daniel L. Thornton is a vice president and economic adviser at the Federal Reserve Bank of St. Louis.

© 2011, The Federal Reserve Bank of St. Louis. The views expressed in this article are those of the author(s) and do not necessarily reflect
the views of the Federal Reserve System, the Board of Governors, or the regional Federal Reserve Banks. Articles may be reprinted, reproduced,
published, distributed, displayed, and transmitted in their entirety if copyright notice, author name(s), and full citation are included. Abstracts,
synopses, and other derivative works may be made only with prior written permission of the Federal Reserve Bank of St. Louis.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

439

Thornton

Figure 1
Daily Spread Between the 3-Month LIBOR and T-Bill Rates (January 2, 2007–December 31, 2009)
Percentage Points
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1/2/07 4/2/07 7/2/07 10/2/07 1/2/08 4/2/08 7/2/08 10/2/08 1/2/09 4/2/09 7/2/09 10/2/09

greater reassurance of timely repayment. On
the demand side, with a central bank liquidity
backstop, banks should be less inclined to borrow from other banks to satisfy any precautionary demand for liquid funds because their
future idiosyncratic demands for liquidity over
time can be met via the backstop.

To understand the issue, it is useful to consider Figure 1, which shows the daily spread
between the 3-month LIBOR and Treasury bill
rates from January 2, 2007, through December 31,
2009. The three vertical lines denote the dates of
three important events: August 9, 2007, when BNP
Paribas, France’s largest bank, halted redemption
on three investment funds (the financial crisis is
assumed to begin on this date); December 12,
2007, when the Fed announced the TAF; and
September 15, 2008, when Lehman Brothers filed
for Chapter 11 bankruptcy protection.2 The spread
began increasing in March 2007, on news of problems with subprime loans in the mortgage market,
to a peak of 80 basis points in late June 2007 and
then declined. The spread increased dramatically
2

For a complete time line of events during the financial crisis, see
http://timeline.stlouisfed.org/index.cfm?p=timeline#.

440

N OV E M B E R / D E C E M B E R

2011

at the outset of the financial crisis to a peak of
about 240 basis points, declined again, and
increased again to a peak of nearly 220 basis
points on December 12, 2007; it then declined
dramatically following the TAF announcement
to a cyclical low of about 80 basis points in midJanuary 2008. From January 17, 2008, through
September 14, 2008, the spread averaged 112
basis points. The spread increased dramatically
again on Lehman’s announcement, to a peak of
452 basis points on October 10, 2008, and then
declined, eventually reaching pre-financial-crisis
levels in the latter half of 2009.
The Fed argued that the dramatic increase in
spreads in August 2007 reflected an increase in
banks’ liquidity premium—that is, banks were
demanding a higher rate on interbank lending
because of an increased demand for liquidity.
Taylor and Williams (2008a,b, 2009) and others
have argued that the increase in the interbank rate
spreads was due to an increase in the risk premium rather than an increase in a liquidity premium. If the increase in the LIBOR/T-bill spreads
was the consequence of an increase in the credit
risk premium, the TAF would have no effect on it.
Hence, this is a key question: Was the increase in
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

the LIBOR/T-bill spreads at the outset of the financial crisis due to an increase in a liquidity premium or an increase in the credit risk premium.3
LIBOR spreads can reflect both liquidity and
credit risk premiums. To identify each type of
premium, it is necessary to compare rates and rate
spreads for banks that are without significant
liquidity constraints with comparable rates and
rate spreads for market participants that are liquidity constrained. Most investigations of the efficacy
of the TAF have relied on an event-study methodology (e.g., Taylor and Williams, 2008a,b, 2009;
McAndrews, Sarkar, and Wang, 2008; and Wu,
2008), which has yielded mixed results. Recently,
CLR have presented evidence from a six-factor
term structure model that indicates that the
announcement effect of the TAF had a very large
effect on the LIBOR rate. Specifically, CLR conduct a counterfactual experiment and find that the
announcement of the TAF reduced the liquidity
premium in the 3-month LIBOR rate by 82 basis
points relative to what the spread would have
been otherwise.
This paper adds to the existing literature in
three ways. First, and importantly, I consider
the behavior of the LIBOR/T-bill spreads both
before and after the TAF announcement because,
if nearly all of the change in the LIBOR/T-bill
spreads before the TAF can be accounted for by
changes in risk spreads, it is difficult to see how
the TAF could have generated a large reduction
in the liquidity premium. (If the liquidity premium did not increase significantly at the outside
of the financial crisis, the announcement of the
TAF could not have reduced it dramatically.)
Second, I show that CLR’s conclusion depends
critically on the marked increase in the spreads
between AA-rated financial bond rates and
equivalent-maturity LIBOR rates immediately
following the TAF announcement. I offer an alternative hypothesis for the marked increase in the
financial bond/ LIBOR rate spreads and present
3

See Krishnamurthy (2010) for a discussion of how an increase in
credit risk can cause an increase in the demand for liquid assets—
that is, assets that can be converted to cash quickly with no appreciable market risk. However, this effect is endemic to the market
and is not unique to banks. Consequently, as I will show, it is
reflected in risk spreads generally.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

a variety of evidence supporting this hypothesis.
Finally, I show that nearly all of the behavior of
the LIBOR/T-bill spreads both before and after
the TAF announcement is accounted for by the
risk premium and that when the risk premium is
accounted for, the TAF has at most a modest effect
on the LIBOR/T-bill spreads.
The remainder of the paper is as follows.
The next section briefly reviews the event-study
empirical analyses of the effect of the TAF. The
third section presents CLR’s affine-term-structuremodel approach for analyzing the effect of the
TAF. The section shows that CLR’s announcement
effect depends critically on the marked increase
in the spread between rates on (i) highly rated
corporate financial bond rates and (ii) equivalentmaturity LIBOR rates immediately following the
announcement of the TAF. The fourth section
offers an alternative hypothesis for the marked
increase in this spread and presents evidence
consistent with this hypothesis. An empirical
analysis of the effect of the TAF on the LIBOR/
T-bill spreads is presented in the fifth section.
The final section offers conclusions.

Event-Study Investigations of the
Effects of the TAF
Taylor and Williams (2008a) were the first to
investigate whether the TAF had a significant
effect on the LIBOR rate. They investigated the
effect of the TAF by regressing the 1- and 3-month
spreads between the LIBOR and OIS rates on various measures of counterparty risk and dummy
variables for TAF bid submission dates. In all
cases considered, the coefficient on the measure
of counterpart risk was positive and statistically
significant, indicating that some of the increase
in the spread was accounted for by risk premiums.
The coefficients on the TAF dummy variable were
also positive, but not statistically significant.
Based on their economic and empirical analyses,
Taylor and Williams concluded that increased
counterparty risk between banks contributed to
the rise in spreads and find no empirical evidence
that the TAF has reduced spreads.4
4

Taylor and Williams (2008a, title page).

N OV E M B E R /D E C E M B E R

2011

441

Thornton

McAndrews, Sarkar, and Wang (2008) investigate the effect of the TAF on the LIBOR/OIS
spreads using a regression methodology similar
to that of Taylor and Williams (2008a). However,
they suggest that Taylor and Williams’s use of the
level of the spread in their regressions is valid
only under the assumption that the liquidity risk
premium falls on a day with a TAF event but
reverts to the previous level immediately after
the TAF event.5 Using the change in the spread
as the dependent variable and dummy variables
for all of the various auction announcements and
operations, they find that the TAF significantly
reduced the size of the LIBOR/OIS spreads.
Wu (2008) suggests that the methodology used
by Taylor and Williams (2008a) and McAndrews,
Sarkar, and Wang (2008) is problematic because
they (i) assume that the TAF had no effect on the
spreads other than on event days associated with
it, (ii) do not control for systematic counterparty
risk among major financial institutions, and (iii)
fail to separate the effects of lowering the counterparty risk premiums from those relieving liquidity concerns.6
Wu’s (2008) approach to analyzing the effectiveness of the TAF differs from the two previous
approaches in three respects. First, rather than
using a TAF dummy variable for specific event
days only, Wu uses a TAF dummy variable that
is zero for all days prior to the TAF announcement on December 12, 2007, and 1 thereafter. Wu
(2008) argues that because TAF lending was for
maturities of 28 days or longer, one would expect
that such loans would be able to relieve the financial strains for the duration of the loans—and not
simply affect the spreads on specific event days.
Wu also included alternative measures of stock
and bond market volatility and the eurodollar
rate volatility as well as a mortgage default risk
factor in his regression equations.7 In contrast to
the findings of Taylor and Williams (2008a), Wu
finds that the TAF has, on average, reduced the
1-month LIBOR/OIS spread by at least 31 basis
5

McAndrews, Sarkar, and Wang (2008, p. 10).

6

Wu (2008, p. 3).

7

The mortgage risk factor is the first principal component for credit
default swap rates for three mortgage companies.

442

N OV E M B E R / D E C E M B E R

2011

points and the 3-month LIBOR/OIS spreads by at
least 44 basis points. He also regresses his TAF
dummy variable on two measures of systematic
risk and, consistent with Taylor and Williams’s
analysis, finds that the coefficient is positive and
statistically significant, suggesting that the TAF
has not been able to reduce the counterparty
default risk premiums.8
A later paper by Taylor and Williams (2008b)
responds to criticism by McAndrews, Sarkar, and
Wang (2008) and Wu (2008) and others regarding
their earlier (2008a) work. First, they show that
the spreads between the LIBOR/OIS rates were
very similar to the spread between the LIBOR rate
and the repo rate on government securities, arguing that the LIBOR/repo spread is a very good
measure of interbank risk because it is the difference in rates between secured and unsecured
lending between banks at the same maturity.9
The close correspondence between these rates
suggests that the LIBOR/OIS spreads primarily
reflects credit risk and not liquidity risk.
The authors also suggest that one could discriminate between liquidity risk and counterparty
risk by comparing the behavior of rates paid to
others who lend to banks but are not liquidity
constrained, such as the rates paid on certificates
of deposit (CDs). Term CDs and term LIBOR loans
are alternative ways that banks finance their
shorter-term lending. Because purchasers of CDs
are not liquidity constrained, there is no reason
for CD rates to increase because of liquidity concerns. However, because these instruments are
uninsured, CD rates will rise when market participants believe that lending to banks is more risky.
Consequently, the TAF should have no effect on
any liquidity premium embedded in CD rates.
Taylor and Williams (2008b) note that CD rates
have tracked LIBOR rates of comparable maturities very closely, suggesting that liquidity risk is
not a significant separate factor driving term lending rates.10 They also perform additional regression analysis altering the timing of how the TAF
8

Wu (2008, p. 2).

9

Taylor and Williams (2008b, p. 6).

10

Taylor and Williams (2008b, p. 10).

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

might affect interest rates and using CD rates
based on a broader set of banks; they also conduct
regression analysis with the spreads between the
CD, term federal funds, and eurodollar rates and
the OIS rate (the dependent variable). They find
no evidence of a significant effect of the TAF in
any of these regressions.
Taylor and Williams (2008b) find that the
results using Wu’s (2008) TAF dummy variable
were fragile. Specifically, the coefficient was large
and statistically significant over one sample, but
not when the sample was extended.11 They also
investigate the effectiveness of the TAF using the
outstanding TAF loan balance. The estimated
coefficients were sometimes negative, but seldom
statistically significant.
Finally, the authors find that the results using
the first difference of the spread rather than the
level of the spread depended critically on the
timing of the variable in the regression and on
the particular TAF events considered. Noting that
the relationship between LIBOR/OIS spreads
and various measures of counterparty risk are
robust, they conclude that, while other researchers
have found significant TAF effects by altering
the specification of the empirical equation they
originally proposed, these results are sensitive to
small changes in specification, measures of the
spread, or measures of risk.12

THE EFFECTIVENESS OF THE TAF:
RESULTS FROM A SIX-FACTOR
TERM STRUCTURE MODEL
CLR use a very different approach, noting
that the McAndrews, Sarkar, and Wang (2008)
and Wu (2008) conclusions about the effectiveness of the TAF using regression analyses of
Taylor and Williams (2008a,b) are sensitive to
only small differences in the specifications of
their regression equations.13 Specifically, they
analyze the effectiveness of the TAF by estimat11

Also see Taylor and Williams (2009), which reflects work from
their two 2008 papers.

12

Taylor and Williams (2008b, p. 20).

13

Christensen, Lopez, and Rudebusch (2009, p. 4).

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

ing a six-factor arbitrage-free term structure model
based on a Nelson and Siegel (1987) yield curve.
There are three Nelson-Siegel factors for Treasury
yields, two Nelson-Siegel factors for bank bond
yields, and a single LIBOR factor. They estimate
the model using weekly data over the sample
period January 6, 1995, to July 25, 2008. They
note that their LIBOR factor changed significantly
immediately following the announcement of the
TAF (December 14, 2007), as did parameters of
their model that involve the LIBOR factor. They
then conduct a counterfactual experiment to
quantify the effect of the change in the model’s
behavior for the 3-month LIBOR rate. Specifically,
they fix the mean of the LIBOR factor at its preannouncement level and leave the other factors
unchanged. Their counterfactual experiment
suggests that the 3-month LIBOR rate would have
averaged about 80 basis points higher without
the TAF.14
Given the sensitivity of the regression
approaches to the specification of the equations
and other issues, CLR’s counterfactual result constitutes the most compelling evidence that the
TAF had a significant effect of reducing the LIBOR
spreads. Consequently, it is important that this
evidence be analyzed carefully. Particularly important is that CLR’s counterfactual result depends
critically on their LIBOR factor, which is based
on the spreads between the 3-, 6-, and 12-month
LIBOR rates and rates on AA-rated corporate
financial bonds with the same maturities. Their
factor differs little from the first principal component obtained from these spreads. Given that
CLR assume that the LIBOR is independent of the
other five factors, this result is not surprising.
Figure 2 shows CLR’s factor and the first
principal component of the three rate spreads.
The vertical line denotes December 14, 2007 (the
week of the TAF announcement). The two factors
are very similar. Most important is the fact that
both decline markedly immediately following
the announcement of the TAF. The marked decline
in the LIBOR factor is a consequence of the AArated corporate financial bond rates declining
relatively less than equivalent-term LIBOR rates
14

Christensen, Lopez, and Rudebusch (2009, p. 29).

N OV E M B E R /D E C E M B E R

2011

443

Thornton

Figure 2
CLR LIBOR Factor and the First Principal Component
Percentage Points
1.5
0.059

1.0

0.057

0.5
0

0.055

–0.5
0.053

–1.0

0.051

–1.5
–2.0
–2.5
–3.0

0.049
First Principal Component (left)

0.047

CLR LIBOR Factor (right)

0.045

–3.5
1/6/95 1/6/96 1/6/97 1/6/98 1/6/99 1/6/00 1/6/01 1/6/02 1/6/03 1/6/04 1/6/05 1/6/06 1/6/07 1/6/08

Figure 3
3-Month AA-Rated Corporate Financial Bond and LIBOR Rates and Their Spread
Percentage Points

Percentage Points
7.0

1.8
6.0
5.0

1.3

4.0
0.8

3.0
2.0

0.3

1.0
–0.2

1/
5/
07
2/
5/
0
3/ 7
5/
07
4/
5/
07
5/
5/
07
6/
5/
07
7/
5/
07
8/
5/
07
9/
5/
0
10 7
/5
/0
11 7
/5
/0
12 7
/5
/0
7
1/
5/
08
2/
5/
08
3/
5/
08
4/
5/
08
5/
5/
08
6/
5/
08
7/
5/
08

0.0

LIBOR (left)

444

N OV E M B E R / D E C E M B E R

AA Corporate Financial (right)

2011

Spread (right)

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

Figure 4
3-Month CD and LIBOR Rates and Their Spread
Percentage Points
7.0

Percentage Points
0.15

6.0

0.10

5.0

0.05

4.0

0.00

3.0

–0.05
–0.10

2.0
LIBOR (left)
1.0

CD (right)

–0.15

Spread (right)
0.0
1/5/07

–0.20
4/5/07

7/5/07

10/5/07

immediately following the TAF announcement.
This is illustrated in Figure 3, which shows the
3-month AA-rated corporate financial bond rate,
the 3-month LIBOR rate, and their spread weekly
from January 5, 2007, through July 25, 2008. The
first vertical line denotes the week of the onset of
the financial crisis; the second denotes the week
of the TAF announcement. Both rates fell on the
TAF announcement, but the LIBOR rate declined
more than AA-rated corporate financial bond
rates, so the spread increased.15
Because this marked and very persistent
increase in the spread of AA-rated corporate financial bond rates over LIBOR rates is responsible
for CLR’s counterfactual result, it is important to
understand why highly rated corporate financial
bond rates increased relative to LIBOR rates following the TAF announcement. CLR suggest that
this decline in LIBOR rates relative to financial
bond rates is due to a marked reduction in the
15

The behavior of the 6- and 12-month spreads is very similar to
that of the 3-month spread. Indeed, the first principal component
of these three spreads accounts for 84 percent of the variance of
the three spreads.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

1/5/08

4/5/08

7/5/08

liquidity premium that banks required to lend in
the interbank market. Specifically, CLR suggest
that the bank bond rates are derived from debt
obligations issued to a broad class of investors
that overwhelmingly consists of nonbank institutions. While these two classes of lenders most
likely attach similar probabilities and prices to
credit risk, they likely have different tolerances for
liquidity problems.16 That is, the spread widened
because of a marked decline in the liquidity premium in the LIBOR rates relative to AA-rated
corporate financial bond rates.
There are two reasons to be skeptical of CLR’s
interpretation. First, if the sharp increase in the
spread of AA-rated corporate financial bond rates
over LIBOR rates were due to a decline in the
liquidity premium required by banks, the same
logic would imply that this spread should have
declined markedly at the onset of the financial
crisis because the liquidity premium required by
banks would have increased relative to that of the
16

Christensen, Lopez, and Rudebusch (2009, pp. 26-27).

N OV E M B E R /D E C E M B E R

2011

445

Thornton

financial bond rate. However, this did not occur.
Indeed, Figure 3 shows that, rather than decreasing at the onset of the financial crisis, as CLR’s
interpretation would suggest, the spread increased;
however, it declined subsequently. On average
between the onset of the financial crisis and the
TAF announcement, the spread changed little
from the level for the weeks prior to the beginning of the financial crisis. In short, the spread
increased rather than decreased, contrary to the
logic of CLR’s hypothesis.
Second, CLR’s interpretation suggests that
there should have been a comparable increase in
the spread between the 3-month CD and LIBOR
rates. CDs represent loans to banks by a broad
class of investors that overwhelmingly consists
of nonbank institutions and are a major source of
funds for bank lending. Lenders in the CD market
are not liquidity constrained and did not acquire
liquidity through the TAF. Consequently, we
should expect to see a marked decline in the
LIBOR rate relative to the CD rate following the
TAF announcement. Figure 4 shows the 3-month
CD and LIBOR rates and CD/LIBOR spread weekly
for the period January 5, 2007, through July 25,
2008. As before, the first and second vertical lines
denote the week of the onset of the financial crisis
and the TAF announcement, respectively. The
3-month CD and LIBOR rates are nearly identical
before and after the onset of the financial crisis
and before and after the TAF announcement. The
variability of the CD/LIBOR spread increased with
the onset of the financial crisis, but there was
virtually no change in the average spread, which
was –4 basis points before the financial crisis and
–1 basis point after the TAF announcement.

AN ALTERNATIVE HYPOTHESIS
FOR THE BEHAVIOR OF THE
CORPORATE FINANCIAL BOND/
LIBOR SPREAD
This section offers an alternative hypothesis
for this marked change in behavior of the AA
corporate financial/LIBOR spread following the
446

N OV E M B E R / D E C E M B E R

2011

TAF announcement, which I call the increasedrisk-premium hypothesis (IRPH).17 Specifically,
it is possible that the market participants interpreted the Fed’s announcement of the TAF as an
indicator that the financial crisis was more serious
than previously thought. The IRPH seems particularly credible given that the Bank of England,
the Swiss National Bank, the Bank of Canada, and
the European Central Bank announced measures
designed to address elevated pressures in shortterm funding markets on the same day. If market
participants believed these announcements signaled that the financial crisis was worse than
previously thought, the TAF and other announcements could have caused a reassessment of the
credit risk of financial firms, increasing the spread
between corporate financial bond rates and LIBOR
rates.

Evidence of the IRPH: The Behavior of
Risk Spreads
The IRPH is supported by the fact that spreads
between corporate financial and non-financial
bond rates and the LIBOR rate increased following the TAF announcement. Figure 5 shows the
spreads between 3-month (i) AA-rated corporate
financial, (ii) AA-rated corporate industrial, and
(iii) BBB-rated corporate industrial weekly bond
rates and the 3-month LIBOR rate weekly over
the period January 5, 2007, through July 25, 2008.
The vertical lines denote the onset of the financial
crisis and the TAF announcement, respectively.
The spreads initially declined with the onset of
the financial crisis and the industrial spreads
declined prior to the TAF announcement, while
the AA-rated corporate financial bond spread
remained relatively stable at about 50 basis points.
All three spreads increased following the TAF
announcement. Moreover, all three spreads
increased by similar amounts between the week
17

There were reports that the LIBOR rate (which is obtained from
surveys) was understating the rate that banks were actually paying in the interbank market during the financial crisis (e.g.,
Mollenkamp and Whitehouse, 2008). Kuo, Skeie, and Vickery
(2010) provide evidence supporting these claims. However, their
estimates of the degree of understatement during this period is
not large enough to account for CLR’s findings.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

Figure 5
Spreads Between the 3-Month AA-Rated Corporate Financial and AA-Rated and BBB-Rated
Industrial Bond Rates and the 3-Month LIBOR Rate
Percentage Points
2.0
1.5
1.0
0.5
0.0
–0.5
–1.0
1/5/07

4/5/07

7/5/07

BBB Industrial/LIBOR

10/5/07
AA Industrial/LIBOR

of the announcement and the week of January 25,
2008: The AA corporate financial/LIBOR, AA
industrial/LIBOR, and BBB industrial/LIBOR
spreads increased by 102, 123, and 106 basis
points, respectively. The similarity in the behavior
of the spreads before and after the TAF announcement strongly supports the IRPH. As noted above,
if the announcement was taken as an indicator
that the financial crisis was worse than previously
thought, credit risk premiums would have
increased, which they did. Indeed, not only did
all of these corporate bond rates rise relative to
the LIBOR rate, but the spread between BBB and
AA industrial corporate bonds—a commonly used
measure of credit risk—also increased dramatically, from 33 basis points prior to the TAF
announcement to a peak of 165 basis point in
early June 2008. This establishes the possibility
that the marked increase in the AA corporate
financial/LIBOR spread, which accounts for CLR’s
counterfactual result, is due to an increase in the
risk premium rather than to a decrease in a liquidity premium, as they hypothesize.
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

1/5/08

4/5/08

7/5/08

AA Corporate Financial/LIBOR

Evidence of the IRPH: The Behavior of
Corporate Financial and Bank Bond
Spreads
The IPRH is also consistent with the relative
behavior of corporate financial and bank bond
rates. Figure 6 shows the spread between 3-month
AA-rated corporate financial and AA-rated bank
bond rates. The data are weekly over the period
January 5, 2007, through July 25, 2008. The vertical lines denote the onset of the financial crisis
and the TAF announcement, respectively. The
spread averaged a few basis points in early 2007
and rose on news of subprime mortgage problems.
The spread increased further following the onset
of the financial crisis, averaging about 10 basis
points before the financial crisis and 49 basis
points from the onset of the financial crisis to
the week prior to the TAF announcement. The
spread increased further following the announcement before declining in June 2008. The behavior
of the AA corporate financial/AA bank bond
spread is consistent with the IRPH for two reaN OV E M B E R /D E C E M B E R

2011

447

Thornton

Figure 6
3-Month AA Corporate Financial/AA Bank Bond Spread
Percentage Points
2.0

1.5

1.0

0.5

0

–0.5

–1.0
1/5/07

4/5/07

7/5/07

10/5/07

sons. First, the implicit guarantee to bank
investors associated with “too big to fail” was
initially thought not to apply to non-bank financial corporations, at least before the Bear Sterns
bailout. Second, financial corporations had greater
exposure to mortgage-backed securities (MBS)
than did banks generally.18 For both of these
reasons, it is reasonable to expect that corporate
financial bond rates would rise relative to bank
bond rates.

Evidence of the IRPH: CLR’s LIBOR
Factor and Risk Spreads
The analysis above strongly suggests that
CLR’s LIBOR factor reflects a marked change in
the risk premium rather than a marked change in
a liquidity premium, as they hypothesize. To
see how much of the variation in CLR’s LIBOR
factor can be accounted for by risk premiums,
18

Of the $4.4 trillion of agency and GSE-backed securities held by
financial institutions in the second quarter of 2007, only $1.1
trillion was held by banks.

448

N OV E M B E R / D E C E M B E R

2011

1/5/08

4/5/08

7/5/08

the LIBOR factor is regressed on risk premiums
reflected in the spreads between BBB-rated and
AA-rated corporate bank and industrial bond
rates. The spreads are for maturities of 3, 6, and
12 months—the same maturities that CLR used
to obtain their LIBOR factor. The sample period
begins with the availability of AA-rated bank
bond rate data, March 17, 2000. These six risk
premiums account for 44 percent of the weekly
variation in CLR’s LIBOR factor over the sample
period March 10, 2000, through July 25, 2008.
To see whether these risk premiums account
for more or less of the variation during periods
when the LIBOR factor is relatively more variable
(especially following the announcement of the
TAF), the regression equation is estimated using
a rolling window of 60 weeks. Figure 7 presents
–
the rolling window regression estimates of R 2 over
the sample period. The data are plotted on the last
week in the sample. The vertical line denotes the
first sample to include post-TAF-announcement
data. The estimates show that the risk premiums
account for relatively more of the variation in
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

Figure 7
60-Week Rolling Estimates of Adjusted R 2 of the CLR LIBOR Factor on Corporate Risk Spreads
Percent
90
80
70
60
50
40
30
20
10

4/
27
7/ /01
27
10 /0
/2 1
7
1/ /01
27
4/ /02
27
7/ /02
27
10 /0
/2 2
7
1/ /02
27
4/ /03
27
7/ /03
27
10 /0
/2 3
7
1/ /03
27
4/ /04
27
7/ /04
27
10 /0
/2 4
7
1/ /04
27
4 / /05
2
7/ 7/0
27 5
10 /0
/2 5
7
1/ /05
27
4/ /06
27
7/ /06
2
10 7/0
/2 6
1/ 7/0
27 6
4/ /0 7
27
7/ /07
2
10 7/0
/2 7
7
1/ /07
27
4/ /08
27
/0
8

0

CLR’s LIBOR factor when it is particularly variable
(see Figure 2). For example, between 2001 and
2003, risk premiums account for over 80 percent
of the variation for a period of a year or longer.
Importantly, for the issue of whether CLR’s counterfactual results are evidence of the success of the
TAF in reducing liquidity premiums, the estimate
–
of R 2 increases dramatically when post-TAFannouncement data are included in the sample.
–
The estimate of R 2 peaks at 82 percent for the
60-week period ending April 4, 2008.
It may also be the case that the sharp increase
in the spread of LIBOR rates over equivalentmaturity Treasury rates was at least partly due to
an increase in the risk premium associated with
bank lending. To investigate this possibility, the
3-month LIBOR/T-bill spread was regressed on
the same six risk premiums over the same sample
period. The risk premiums account for 50 percent
of the variation in the LIBOR/T-bill spread over
the entire sample period. Figure 8, which plots
–
the 60-week rolling estimate of R 2 for a regression
of the LIBOR/T-bill spread on the six risk premiums, shows that after declining to essentially zero,
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

–
the estimate of R 2 increased dramatically following the onset of the financial crisis (the first vertical line). It continued to increase to a peak of
nearly 70 percent following the TAF announcement (the second vertical line).

EXPLAINING THE BEHAVIOR OF
THE LIBOR/T-BILL SPREADS
The analysis in the previous section suggests
that CLR’s LIBOR factor is largely accounted for
by risk premiums and does not present strong
support for the effectiveness of the TAF. However,
the evidence using weekly data suggests that the
TAF may have been effective in reducing the
LIBOR/T-bill spread.
This issue is investigated more thoroughly in
this section using daily data using corporate
bond/T-bill spreads not previous used in the literature. The corporate bond/T-bill spreads are for
corporate bank, industrial, and retail bonds. These
spreads are denoted BT3, IT3, and RT3, respectively. The CD/T-bill and LIBOR/T-bill spreads
N OV E M B E R /D E C E M B E R

2011

449

Thornton

Figure 8
60-Week Rolling Estimates of Adjusted R 2 of the 3-Month LIBOR/T-Bill Spread on
Corporate/T-Bill Risk Spreads
Percent
80
70
60
50
40
30
20
10
0
–10
4/
2
6/ 7/0
27 1
8 /
10/27 01
/
12 /27 01
/2 /01
2/ 7/
2 0
4/ 7/01
2
6/ 7/02
2
8/ 7/ 2
10 27 02
/
12/27 02
/2 /02
2/ 7/0
2
4/ 7/ 2
2 0
6/ 7/03
27 3
8 /
10 /27 03
/
12/27 03
/2 /03
2/ 7/0
2
4/ 7/0 3
2
8/ 7/04
2
6/ 7/04
10 27 4
/
12/27 04
/2 /0
2/ 7/04
2
4/ 7/04
2
6/ 7/05
2
8 7/ 5
10/27 05
/
/
12 27 05
/2 /05
2/ 7/0
4/27/ 5
2 0
6/ 7/06
2
8/ 7/06
10 27 6
/
12 /27 06
/2 /0
2/ 7/0 6
2
4/ 7/0 6
2
6/ 7/07
2
8/ 7/07
10 27 7
/
/
12 27 07
/2 /07
2/ 7/0
2
4/ 7/07
2
6/ 7/08
27 8
/0
8

–20

are denoted CDT3 and LT3, respectively. The
effect of the TAF is investigated further by estimating the equation
LT3t = α + βbBT3t + βiIT3t + βrRT3t + δ DUMVEC + ⑀t ,

where DUMVEC is a vector of dummy variables
that reflect important TAF dates used in the previous event-study literature and ⑀t is an i.i.d. error
term. To make the results comparable to the previous event studies, different sets of dummy
variables identical to those used by Taylor and
Williams (2008ab), McAndrews, Sarkar, and Wang
(2008), and Wu (2008) are used. There are six
dummy variables. The first five are those used
by McAndrews, Sarkar, and Wang (2008): The
dates of international announcements related to
the TAF (ANI ), domestic TAF announcements
(AND), dates when the conditions of the announcement were set (CON), when the auction took place
(AUC ), and when banks were notified (NOT ).19
The sixth dummy variable is that used by Wu

(2008), denoted Wu, which is zero before
December 12, 2007, and 1 thereafter. The sample
period is March 10, 2000, through April 30, 2008.20
The results are presented in Table 1. The
p-values are based on HAC standard errors. The
results in the first two columns use McAndrews,
Sarkar, and Wang’s (2008) dummy variables. The
results indicate that LT3 is significantly related
to each of the corporate bond spreads; the coefficient on each bond spread is positive and highly
statistically significant. Moreover, the sum of the
coefficients is 0.92 and the hypothesis that the
sum of the coefficients is 1 is not rejected at the
5 percent significance level. The estimates of the
coefficients on TAF dummy variables provide no
evidence that the TAF had any significant effect
on the LIBOR/T-bill spread: The coefficients on
the ANI and AND dummy variables are positive,
but not statistically significant. The coefficients
on TAF operation dummy variables are negative,
20

19

These dates can be found in McAndrews, Sarkar, and Wang (2008,
Table 1, p. 20).

450

N OV E M B E R / D E C E M B E R

2011

The sample ends on April 30, 2008, to make the TAF sample period
similar to that used by McAndrews, Sarkar, and Wang (2008) and
Wu (2008).

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

Table 1
The Effect of the TAF on the 3-Month LIBOR/T-Bill Spread
Estimate

p-Value

Estimate

p-Value

Estimate

p-Value

–0.040

0.111

–0.040

0.116

–0.089

0.415

0.000

0.414

0.000

0.547

βi

0.275

0.007

0.275

0.007

βr

0.226

0.001

0.225

Constant

βb

Estimate

p-Value

0.001

0.060

0.000

0.000

–0.014

0.229

0.266

0.002

0.007

0.431

0.001

0.230

0.000

0.058

0.000

ANI

0.331

0.168

—

—

—

—

—

—

AND

0.115

0.541

—

—

—

—

—

—

ANI+AND

—

—

0.224

0.241

0.281

0.131

0.042

0.082

CON

–0.005

0.968

–0.004

0.975

0.096

0.377

0.016

0.350

AUC

–0.168

0.160

–0.167

0.160

–0.048

0.639

–0.006

0.591

NOT

–0.214

0.121

–0.213

0.108

–0.087

0.375

–0.016

0.136

Wu

—

—

—

—

–0.340

0.012

–0.031

0.058

CDT3
–
R2

—

—

—

—

—

—

0.928

0.000

0.764

—

0.764

—

0.778

—

0.995

—

SE

0.172

—

0.172

—

0.166

—

0.026

—

NOTE: SE, standard error.

but not statistically significant. The results in the
next two columns show that the conclusion does
not change when the ANI and AND are combined.
There is some evidence that the TAF has been
effective in reducing the LIBOR/T-bill spread
when Wu’s dummy variable is included. The
estimate of the coefficient on Wu is negative and
statistically significant, but the coefficient estimate, 34 basis points, is 10 basis points smaller
than Wu’s estimate. Moreover, consistent with
the findings of Taylor and Williams, the coefficient on Wu tends to decline and becomes statistically insignificant as the length of the sample
increases. It is also the case that evidence of the
effectiveness of the TAF all but disappears when
CDT3 is included as a regressor: The estimate is
negative and statistically significant at slightly
higher than the 5 percent significance level, but
the magnitude of the effect is only 3 basis points.

CONCLUSION
This paper reviews the previous literature on
the effectiveness of the TAF in reducing the spread
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

between equivalent-maturity LIBOR and T-bill
rates and further investigates the effectiveness of
the TAF using weekly and daily data. The previous literature using event-study methodologies
finds mixed results. The most compelling evidence for the effectiveness of the TAF comes from
CLR’s (2009) six-factor term structure model.
Performing a counterfactual analysis based on a
marked change in the LIBOR factor of their model,
CLR indicated that the 3-month LIBOR/T-bill
spread would have been 82 basis points higher
were it not for the TAF. Noting that CLR’s LIBOR
factor is based on the spreads between AA-rate
financial corporate bond rates and LIBOR rates,
I show that these spreads are highly correlated
with risk spreads, especially during the post-TAFannouncement period.
I offer an alternative hypothesis for the behavior of the spread between AA-rated financial corporate bond rates and LIBOR rates following the
announcement of the TAF. Specifically, I hypothesize that market participants revised up their
expectations of the seriousness of the financial
crisis in the wake of the TAF announcement and
N OV E M B E R /D E C E M B E R

2011

451

Thornton

the announcements of other central banks. I
present evidence from a variety of risk spreads
that is consistent with this alternative hypothesis,
including the fact that over 80 percent of CLR’s
LIBOR factor is accounted for by risk spreads
during this period. This suggests that much of
the effect of the TAF that CLR report is actually
due to an increase in the risk premium on financial bonds rather than a reduction in the liquidity
premium embedded in LIBOR rates. Moreover,
this evidence is consistent with the fact that there
was no significant decline in the spread between
the AA-rated corporate financial bond rates and

452

N OV E M B E R / D E C E M B E R

2011

the LIBOR at the outset of the financial crisis:
If there was no significant increase in banks’
liquidity premium, it is difficult to understand
how the TAF could have reduced it.
I also show that the majority of the 3-month
LIBOR/T-bill spread before and after the TAF
announcement can be accounted for by the
spreads between financial and nonfinancial corporate bond rates. Further analysis using daily
data indicates that controlling for these risk premiums, TAF appears to have had little or no effect
on the 3-month LIBOR/T-bill spread.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Thornton

REFERENCES
Bernanke, Ben S. “The Economic Outlook.” Testimony before the Committee on the Budget, U.S. House of
Representatives, January 17, 2008.
Board of Governors of the Federal Reserve System. Press Release. December 12, 2007;
www.federalreserve.gov/newsevents/press/monetary/20071212a.htm.
Christensen, Jens H.E.; Lopez, Jose A. and Rudebusch, Glenn D. “Do Central Bank Liquidity Facilities Affect
Interbank Lending Rates?” Working Paper 2009-13, Federal Reserve Bank of San Francisco, June 2009;
www.frbsf.org/publications/economics/papers/2009/wp09-13bk.pdf.
Krishnamurthy, Arvind. “How Debt Markets Have Malfunctioned in the Crisis.” Journal of Economic
Perspectives, Winter 2010, 24(1), pp. 3-28.
Kuo, Dennis; Skeie, David and Vickery, James. “How Well Did LIBOR Measure Bank Wholesale Funding Rates
During the Crisis?” Unpublished manuscript, Federal Reserve Bank of New York, July 30, 2010.
McAndrews, James; Sarkar, Asani and Wang, Zhenyu. “The Effect of the Term Auction Facility on the London
Inter-Bank Offered Rate.” Staff Report No. 335, Federal Reserve Bank of New York, July 2008;
www.newyorkfed.org/research/staff_reports/sr335.pdf.
Mollenkamp, Carrick and Whitehouse, Mark. “Study Casts Doubt on Key Lending Rate: WSJ Analysis Suggests
Banks May Have Reported Flawed Interest Data for LIBOR.” Wall Street Journal, May 29, 2008, p. A1.
Nelson, Charles R. and Siegel, Andrew F. “Parsimonious Modeling of Yield Curves.” Journal of Business, 1987,
60(4), pp. 473-89.
Taylor, John B. and Williams, James C. “A Black Swan in the Money Market.” Working Paper Series 2008-04,
Federal Reserve Bank of San Francisco, April 2008a;
www.frbsf.org/publications/economics/papers/2008/wp08-04bk.pdf.
Taylor, John B. and Williams, James C. “Further Results on a Black Swan in the Money Market.” SIEPR
Discussion Paper No. 07-046, Stanford Institute for Economic Policy Research, May 2008b;
www-siepr.stanford.edu/papers/pdf/07-46.pdf.
Taylor, John B. and Williams, James C. “A Black Swan in the Money Market.” American Economic Journal:
Macroeconomics, 2009, 1(1), pp. 58-83.
Wu, Tao. “On the Effectiveness of the Federal Reserve’s New Liquidity Facilities.” Working Paper No. 2008-08,
Federal Reserve Bank of Dallas, May 2008; http://dallasfed.org/research/papers/2008/wp0808.pdf.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

453

454

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

A Federal Reserve System Conference on
Research in Applied Microeconomics
Thomas A. Garrett
This article summarizes some of the papers presented at the System Applied Microeconomics
Conference organized and hosted by the Federal Reserve Bank of St. Louis on May 5-6, 2011. This
annual conference brings together economists from the Federal Reserve District Banks across the
Federal Reserve System and the Federal Reserve Board to present their latest economic research.
Federal Reserve Bank of St. Louis Review, November/December 2011, 93(6), pp. 455-62.

T

he Federal Reserve Bank of St. Louis
hosted the annual System Applied
Microeconomics Conference on
May 5-6, 2011. The papers presented
at the conference, some of which are summarized
in this article, showcased research in the areas
of public policy, education and human capital,
labor markets, and housing and consumer finance
during the Great Recession.1

ANALYSIS OF PUBLIC POLICY
The first group of papers focused on public
policy issues. In “Assessing the Evidence on
Neighborhood Effects from Moving to Opportunity,” Aliprantis provides a new framework and
a robust instrument to estimate neighborhood
effects using data from the Moving to Opportunity
for Fair Housing (MTO) program. The MTO program, a 10-year research project, combines tenantbased rental subsidies with housing counseling
1

View the conference agenda at http://research.stlouisfed.org/
conferences/appliedmicro/agenda.html. Not all papers presented
at the conference are included herein because authors were given
the option to not include their papers in these proceedings.

to help poor families move from poor urban areas
to less-poor neighborhoods. Aliprantis’s framework improves on earlier methods for studying
the effects of housing mobility programs by distinguishing between program and neighborhood
effect.
In “The Spending and Debt Response to
Minimum Wage Hikes,” Aaronson, Agarwal, and
French explore how minimum wage increases
influence spending and debt accumulation by
minimum wage earners. They find that (i) both
consumer spending and debt accumulation
increase after a minimum wage increase and (ii)
most spending induced by such an increase goes
toward financing durables. The authors’ empirical
findings are consistent with an augmented bufferstock model in which households are collateral
constrained.
Given the recent decline in aid from state to
local governments, in “Designing Formulas for
Distributing State Aid Reductions,” Zhao and
Coyne develop a new formula for allocation of
such funds. Their formula improves on other
methods because it is based on the underlying
fiscal health of local governments rather than
commonly used ad hoc measures.

Thomas A. Garrett is an assistant vice president and economist at the Federal Reserve Bank of St. Louis.

© 2011, The Federal Reserve Bank of St. Louis. The views expressed in this article are those of the author(s) and do not necessarily reflect
the views of the Federal Reserve System, the Board of Governors, or the regional Federal Reserve Banks. Articles may be reprinted, reproduced,
published, distributed, displayed, and transmitted in their entirety if copyright notice, author name(s), and full citation are included. Abstracts,
synopses, and other derivative works may be made only with prior written permission of the Federal Reserve Bank of St. Louis.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

455

Garrett

HUMAN CAPITAL
A second group of papers focused on the
topic of human capital. In “The Role of Schools
in the Production of Achievement,” Canon states
that previous research has been unable to simultaneously consider the three types of inputs
believed to influence student skills: ability, family inputs, and school inputs. Canon uses parents’
savings for higher education as a measure of
ability in an empirical framework that corrects
for the endogeneity of the three inputs. Unlike
earlier studies that do not address the simultaneity among the three inputs, Canon presents evidence that school inputs are important for the
formation of student skills when controlling for
the ability to learn.
In “Economic Literacy and Inflation Expectations: Evidence from a Laboratory Experiment,”
Burke and Manz present new experimental evidence on heterogeneity in the formation of inflation expectations. They conduct a laboratory
experiment in which subjects complete a set of
inflation-forecasting exercises in a simulated
economic environment. They find that the subjects’ demographic characteristics play a small
role in the variation of their inflation expectations,
but economic literacy plays a large role in explaining the accuracy of inflation forecasts.
In “Financial Literacy and Mortgage Equity
Withdrawals,” Duca and Kumar examine whether
an individual’s financial literacy influences the
decision to make or not make mortgage equity
withdrawals. Their results indicate that the financially literate are 3 to 5 percentage points less
likely to withdraw housing equity but that this
result does not apply to home equity lines of
credit.

selection, they find that smokers do face a wage
penalty. However, this penalty is not a result of
smokers having lower productivity but rather of
the set of personal characteristics smokers bring
to the workplace. The authors also find that individual smoking behavior is responsive to the earnings penalty associated with smoking.

LESSONS FROM THE GREAT
RECESSION AND HOUSING CRISIS
A fourth group of papers, of which one is
summarized here, examined housing and consumer finance during the Great Recession. In
“Financing Constraints and Unemployment:
Evidence from the Great Recession,” DuyganBump, Levkov, and Montoriol-Garriga examine
the link between small-business lending and
unemployment during the Great Recession. They
argue that if a reduction in lending to small businesses influences unemployment, then unemployment should increase more in smaller firms
with greater dependence on bank financing. The
authors find that individuals working in sectors
with high external financial dependence, of which
a large portion is small businesses, were more
likely to become unemployed. However, they find
no difference in the likelihood of unemployment
for workers in small and large firms in sectors
with low external financial dependence.

CONTRIBUTIONS
The following section provides more-detailed
summaries of selected conference papers.

LABOR MARKET ISSUES

“Assessing the Evidence on
Neighborhood Effects from Moving
to Opportunity”

A third group of papers, of which one is summarized here, focused on labor market issues. In
“When Does the Labor Market Consider You a
Smoker and Do You Care?” Armour, Hotchkiss,
and Pitts explore the wage differential between
smokers and nonsmokers. Using a switching
regression framework with unknown sample

Building on recent developments in the program evaluation literature, Aliprantis defines
several treatment-effect parameters and estimates
and interprets some of these parameters using data
from the MTO program. The evaluation framework
makes a clear distinction between (i) program
effects from intent-to-treat and treatment-on-the-

456

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Garrett

treated parameters and (ii) neighborhood effects
from local-average-treatment-effect parameters.
This distinction helps to clarify that, although
the results from MTO are informative about the
design of housing mobility programs, they are
only informative about a small subset of neighborhood effects. Furthermore, examination of the
local average treatment effect identifying assumptions helps to illustrate the limitations of searching for exogenous variation in one causal variable
while abstracting from all others.
Aliprantis presents empirical tests for instrument strength that show that MTO induced large
changes in neighborhood poverty rates but remarkably little variation in many of the other neighborhood characteristics believed to influence
outcomes. He argues that this reinterpretation of
the MTO data stresses the importance of understanding heterogeneity in response to treatment
and suggests two important conclusions. First,
if alternative housing mobility programs were
designed to induce moves to neighborhoods with
characteristics other than low poverty, it is entirely
feasible that such programs might induce larger
effects than those realized with MTO. Second,
local-average-treatment-effect estimates appear to
reconcile the evidence from MTO with prevailing
theories of neighborhood effects.

“The Spending and Debt Response to
Minimum Wage Hikes”
Aaronson, Agarwal, and French provide new
evidence on the spending and debt responses of
consumers to an exogenous income change, in particular an increase in the minimum wage among
households with an adult minimum wage worker.
They present four key empirical findings based
on a variety of large survey and administrative
datasets. First, a $1 per hour minimum wage hike
increases total household spending by approximately $700 per quarter in the near term. This
exceeds the roughly $250 per quarter increase in
household income resulting from the hike. This
pattern is corroborated by independent data showing that debt rises substantially after a minimum
wage increase. Second, the majority of this additional spending goes toward durable goods, espeF E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

cially vehicles. Consequently, the spending
response is concentrated among a small number
of households. Third, total spending increases
within one quarter after a minimum wage hike,
although legislation of the increase typically
passes 6 to 18 months before the hike. Finally,
high levels of durables spending and debt accumulation persist for several quarters after a minimum wage hike.
Aaronson, Agarwal, and French argue that
these results are hard to explain using two canonical models: the permanent income model and the
buffer-stock model with no borrowing. If households were spreading an income gain over their
lifetime, as in the permanent income hypothesis,
the short-run spending increase should be much
smaller than what is observed in the data. The
authors show that augmenting the permanent
income model to account for durables raises the
predicted short-term spending response. However,
it is still an order of magnitude smaller than what
the empirical estimates imply. Moreover, a bufferstock model in which households cannot borrow
against durable goods generates a spending
response of less than $200 per quarter and fails
to explain why some minimum wage households
increase their debt after a minimum wage hike.
The authors further consider an augmented
buffer-stock model in which households are
collateral constrained; that is, they can borrow
against part, but not all, of the value of their
durable goods. If households face collateral constraints, small income increases can generate
small down payments, which in turn can be used
for large durable goods purchases. For example,
with a 20 percent down payment, each additional
dollar of income can purchase $5 of durable goods.
Such a model does much better matching the
facts, including for the magnitude, composition,
and timing of spending and debt. An augmented
buffer-stock model that allows for the cost of
adjusting durables (e.g., the time it takes to shop
for a new car or the trade-in value of a vehicle)
also replicates the skewness of the spending
responses shown in the data. The authors’ results
provide direct microeconomic evidence of the
quantitative importance of collateral constraints,
a factor increasingly used to understand the
N OV E M B E R /D E C E M B E R

2011

457

Garrett

dynamics of consumer durables, housing, and
entrepreneurship.

“Designing Formulas for Distributing
State Aid Reductions”
Zhao and Coyne present a new framework
for distributing state aid to local governments.
Given the ongoing state and local fiscal crises
across the nation, state aid to local governments
has become an increasingly important and contentious budgetary issue. States tend to disproportionately and quickly cut local aid during a fiscal
crisis. Indeed, the Congressional Budget Office
reports that 22 states reduced aid to local governments in fiscal year 2010 and that 20 states propose additional cuts in fiscal year 2011. Some
states cut aid on an ad hoc basis, while others
cut aid across the board, with every community
receiving the same percent or per capita dollar cut.
Such approaches are widely considered unfair:
Critics say ad hoc approaches are not based on
economic rationale and lack transparency in the
decisionmaking process and that across-the-board
cuts are more burdensome for poorer communities that tend to rely more heavily on state aid.
To address these concerns, Zhao and Coyne
develop a framework that reduces aid distribution
based on underlying local fiscal health. They use
a local fiscal gap measure to indicate fiscal health
outside the direct control of local officials. Based
on this framework, state government would cut
less aid from communities with larger fiscal gaps
and less existing aid. The framework also accommodates aid increases, thus providing policymakers with a single tool for aid revision. The
authors use Massachusetts data on unrestricted
municipal aid to conduct simulations and explore
policy implications.
Unlike the Zhao-Coyne framework, the current literature on state aid distribution focuses
on gap-based formulas that preserve existing aid
distributions, or hold them harmless, and distribute only aid increases. Therefore, those formulas
are incompatible with aid-reduction scenarios.
The Zhao-Coyne analysis is useful for policymaking and discussion. First, it provides a more
rational and fair framework for cutting local aid
458

N OV E M B E R / D E C E M B E R

2011

than commonly used ad hoc or across-the-board
methods. Adopting their gap-based framework
for aid reduction can help reduce the burden for
communities with the worst underlying fiscal
health and, therefore, advance fiscal equalization
across cities and towns. Second, the framework
helps the transition from non-gap-based to gapbased aid distribution even in years of aid cuts:
States can accelerate the reform process (without
waiting for aid increases) to implement a gapbased formula with a hold-harmless clause. Third,
the research is practical and timely because many
states are making or planning to make additional
local aid cuts. The framework can be used for
distributing school or non-school aid and is potentially applicable to all states.

“The Role of Schools in the Production
of Achievement”
The literature on sources of inequality finds
that “pre-market” factors (i.e., skills individuals
acquire before entering the labor market) explain
most income inequality across individuals and
between groups of individuals. But what explains
differences in pre-market factors? Three types of
inputs are believed to determine these factors:
ability, family inputs, and school inputs. Therefore, to answer the question it is crucial to understand first the relative importance of each input.
A growing literature in economics tries to
provide an answer to the question by studying
children’s scores on performance tests. The literature on the production of achievement has not
been able to provide an estimation that simultaneously accounts for the three factors at the
student level. Canon intends to fill this gap by
providing an estimation of the production function of achievement where both investment types
(families and schools) are considered in a framework in which the inputs are allowed to be correlated with the unobserved term: ability to learn.
Canon does so by applying an algorithm, which
accommodates endogeneity problems in the
choice of inputs for the production of achievement, to a very suitable dataset for this problem—
the National Education Longitudinal Study of
1988 (from the Institute of Education Sciences,
U.S. Department of Education).
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Garrett

This dataset provides information on both
home and school inputs at the student level as
well as parents’ savings for their children’s postsecondary education. Canon uses the savings data
to control for the unobserved component (i.e., the
ability to learn) in the production of skills. This
allows recovery of the parameters of interest in the
production function of achievement: the effect
of period-by-period investment and the impact
of the achievement acquired in previous periods.
What makes the savings for postsecondary education measure informative is that parents decide
to save at the same time they choose the family
and school inputs that will affect the observed
test score (the current outcome). However, those
savings will not affect the current outcome but
instead will affect future labor market outcomes
through the choice to attend college.
Canon’s estimates for the role of family inputs
are in line with previous findings: They foster
students’ achievement and these inputs are more
crucial at some times than others. However, her
estimates of school inputs show that, contrary to
what has been found in the existing literature,
they are important for the formation of students’
skills. Moreover, school inputs seem to be as
important as family inputs if late remediation
policies are considered. Additionally, Canon also
finds evidence that savings for postsecondary
education are a good proxy for students’ unobserved ability to learn.

“Economic Literacy and Inflation
Expectations: Evidence from a
Laboratory Experiment”
Burke and Manz present new experimental
evidence on heterogeneity in the formation of
inflation expectations and relate the variation to
economic literacy and demographics. They conduct a laboratory experiment in which subjects
complete a set of inflation-forecasting exercises
in a simulated economic environment. Subjects
complete (i) a questionnaire that measures economic and financial literacy and (ii) two simulation exercises that require them to provide
forecasts of near-term and medium-term inflation
for the U.S. economy. In the first exercise, subF E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

jects select information from a menu of information sources with varying degrees of economic
relevance; in the other, they receive preselected,
uniform information sets. These separate exercises
identify two important sources of heterogeneity
across individuals with respect to expectations
formation: (i) differences in the types of information used in forming inflation expectations (i.e.,
information is selected by the subjects) and (ii) differences in the use of identical information when
forming inflation expectations (i.e., information
is chosen for the subjects).
Burke and Manz find that heterogeneity in
information selection significantly increases the
variability of forecasting performance compared
with the case in which information is homogeneous. This finding suggests that models of the
inflation-expectations process vary across subjects, and this variability may contribute to realworld disagreement (and aggregate biases) in
inflation expectations.
Compared with previous studies of survey
data, the authors find that fewer demographic
and socioeconomic factors are associated with
variation in inflation expectations. For example,
they do not observe robust gender differences in
inflation expectations. In some cases, apparent
demographic variation in expectations—such as
between African Americans and whites—is
explained by variation in economic literacy. More
important, economic literacy contributes significantly to the accuracy of inflation forecasts and
is associated with a reduced tendency to overestimate inflation in particular. The impact of literacy is nonlinear, however: Very poor performance
in the bottom quartile of the distribution drives
much of the variation, and the marginal impact
of literacy becomes negligible above the 75th or
80th percentile. The contribution of economic
literacy is not reducible to overall educational
attainment or socioeconomic status and instead
reflects a combination of economic knowledge
and general numeracy.
Burke and Manz find that economic literacy
contributes to the accuracy of forecasts through
both sources of heterogeneity described above.
First, more-literate subjects choose better (morerelevant) information sources in making their
N OV E M B E R /D E C E M B E R

2011

459

Garrett

forecasts. For example, compared with lessliterate individuals, they are more likely to view
data on aggregate inflation and less likely to rely
exclusively on information about price changes
for specific goods such as milk and oil. Second,
more-literate subjects make better use of given
data in the exercises involving preselected information. The results suggest that modest educational interventions targeted at individuals with
very low economic literacy levels could lead to
significant improvements in inflation forecasting.
For example, directing subjects to more-relevant
information, such as recent data on aggregate
inflation rather than specific price changes, may
reduce forecasting errors considerably.

“Financial Literacy and Mortgage
Equity Withdrawals”
Duca and Kumar assert that mortgage equity
withdrawals (MEWs) have been linked to the
U.K. consumption boom of the late 1980s and
the U.S. consumption boom of the early 2000s.
MEWs have been linked to an increased sensitivity of consumption to housing wealth and liquidity constraints, consistent with permanent income
models incorporating credit constraints, which
imply that housing wealth influences consumption by providing collateral for loans to otherwise
credit-constrained families.
However, the recent mortgage bust suggests
that many households were unaware of the risks
they took with MEWs, which is consistent with
evidence that many are financially illiterate.
Using data from the Health and Retirement Study
(sponsored by the National Institute on Aging),
earlier research documents that many families
incorrectly answered questions about compound
interest, money illusion, and portfolio diversification. Incorrect answers have been linked to suboptimal saving for retirements, overborrowing,
and low stock market participation. Previous
research indicates that state-mandated high
school financial education and employer-based
financial literacy programs led to higher saving.
In addition, there is evidence that many homeowners do not choose the lowest-cost home mortgage option because they may be confused by
460

N OV E M B E R / D E C E M B E R

2011

terms in mortgage contracts. Despite a role for
increased MEWs during the recent boom and
bust in consumption, potential linkages between
financial literacy and MEWs are yet unexplored.
Duca and Kumar address this gap by examining whether financial literacy is linked to MEWs.
Using three different measures of financial literacy—compound interest, money illusion, and
portfolio diversification—they find that individuals’ knowledge of portfolio diversification has
the most significant impact on their propensity
to make MEWs. The authors’ results indicate
that the financially literate are 3 to 5 percentage
points less likely to withdraw housing equity by
increasing mortgage debt; however, this result
does not apply to their tapping home equity lines
of credit. The authors also find that the propensity
to make MEWs rises with house price appreciation and incentives to lower mortgage interest
rates. In line with earlier research, the findings
of Duca and Kumar indicate that legal conditions
across states are correlated with MEW behavior.
Given the recent evidence in the literature that
MEWs were correlated with mortgage delinquencies as the housing crisis deepened, the DucaKumar findings suggest that financial education
programs might be effective in preventing mortgage defaults.

“When Does the Labor Market Consider
You a Smoker and Do You Care?”
Armour, Hotchkiss, and Pitts investigate
when the labor market considers an individual a
smoker, how being a smoker affects earnings, and
how the expected earnings differential affects an
individual’s propensity to smoke.2 They use a
switching regression with an unknown sampleselection framework that allows for the data to
indicate when the labor market begins treating
workers differently based on whether they smoke.
Their analysis is performed using data from the
annual Current Population Survey Tobacco Use
Supplement (from the Bureau of Labor Statistics)
for 2000 to 2007.
2

The authors’ paper and this summary do not represent the views
of the Centers for Disease Control and Prevention.

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Garrett

Their analysis indicates that the labor market
classifies an individual who smokes at all as a
smoker. This suggests that even the most casual
smokers experience wage penalties for smoking.
Furthermore, the characteristics of smokers
reduce the wage penalty, indicating that the wage
differential is not a result of smokers bringing
less-productive endowments to the labor market.
Instead, the differential is from employers placing
less value on the endowments brought by smokers
than the same endowments brought by nonsmokers (the coefficient effect is positive), indicating that employers treat smokers differently
than nonsmokers.
The analysis also indicates that individual
smoking behavior responds to the labor market
penalty associated with smoking. The implication
is that policies designed to increase the labor
market penalty of smoking could be a powerful
tool in reducing smoking. The authors’ results
remain robust when they compare different types
of current nonsmokers with current smokers and
estimate the model on a subsample of married
people only.

“Financing Constraints and
Unemployment: Evidence from the
Great Recession”
Duygan-Bump, Levkov, and MontoriolGarriga assert that lending to small businesses in
the United States has fallen dramatically since
the onset of the Great Recession. According to
the most recent data, small-business loans made
by commercial banks declined over $40 billion
between the second quarter of 2008 and the second quarter of 2010. Similarly, the responses to
the Federal Reserve’s Senior Loan Officer Opinion
Survey on Bank Lending Practices indicate that
banks have significantly tightened credit standards
on commercial and industrial loans to small firms
in 13 consecutive quarters (2007:Q1–2010:Q1).
The decline in small-business lending has
received much attention from policymakers,
especially because of its potential link to the high
unemployment rate. Indeed, almost 80 percent
of all firms in the United States have fewer than
nine employees, and small firms employ roughly
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

50 percent of all Americans. Unlike larger firms,
which have broader access to capital markets,
small businesses are highly dependent on bank
financing. Accordingly, smaller firms are likely
to have been disproportionately affected when
banks restricted credit following shocks to their
balance sheets.
The authors investigate the link between
small-business lending and unemployment
during the Great Recession. They argue that if
reductions in small-business lending affect
unemployment, then unemployment would be
expected to increase more in smaller firms, but
only among firms that depend on bank financing.
They test their hypothesis by exploiting variation
across firm size and external financial dependence. Specifically, they combine information
on workers’ firm size and unemployment status
from the Current Population Survey (from the
Bureau of Labor Statistics) with firms’ financial
information from Compustat and the Survey of
Small Business Finances (from the Board of
Governors of the Federal Reserve System). They
then estimate changes in the unemployment rate
during the recent financial crisis by firm size and
across industrial sectors with different degrees
of financial need.
Duygan-Bump, Levkov, and MontoriolGarriga find that during the Great Recession
workers in sectors with high external financial
dependence were more likely to become unemployed, especially workers in smaller firms. By
contrast, the authors do not find significant differences in unemployment propensity between
workers in small and large firms in sectors with
low external financial dependence. They estimate
that during the financial crisis the likelihood of
unemployment among workers in small, financially constrained firms increased by 0.55 percentage points relative to other workers. One
explanation of this finding is that the financial
crisis propagated to the real economy through a
reduction in bank credit.
An alternative explanation of their finding is
that changes in unemployment are not driven by
changes in the supply of credit, but rather reflect
a disproportionate reduction in the demand for
goods and services produced by finance-dependent
N OV E M B E R /D E C E M B E R

2011

461

Garrett

sectors. To address this possible explanation, the
authors repeat their analyses using data around
the 2001 recession. They exploit the fact that the
2001 recession did not originate from stressed
bank balance sheets. The resulting estimates for
that recession show almost identical changes in
the unemployment rate among small and large
firms with different degrees of financial dependence. Comparison of the 2007-09 and 2001 recessions indicates that changes in the supply of credit
contributed to changes in the rate of unemployment during the recent financial crisis.

This paper has important implications for
policy intervention. The authors suggest that
policies aimed at making credit available to small
businesses, such as the recent $30 billion Small
Business Jobs Act or the loans guaranteed by the
Small Business Administration, would help stabilize the labor markets and economic activity in
the United States.

REFERENCES
Aaronson, Daniel; Agarwal, Sumit and French, Eric. “The Spending and Debt Response to Minimum Wage
Hikes.” Working Paper No. 2007-23, Federal Reserve Bank of Chicago, February 2011;
www.chicagofed.org/webpages/publications/working_papers/2007/wp_23.cfm.
Aliprantis, Dionissi. “Assessing the Evidence on Neighborhood Effects from Moving to Opportunity.” Federal
Reserve Bank of Cleveland, Working Paper No. 11-01, January 2011;
www.clevelandfed.org/research/workpaper/2011/wp1101.pdf.
Armour, Brian S.; Hotchkiss, Julie L. and Pitts, M. Melinda. “When Does the Labor Market Consider You a
Smoker and Do You Care?” Unpublished manuscript, Federal Reserve Bank of Atlanta, 2011.
Burke, Mary A. and Manz, Michael. “Economic Literacy and Inflation Expectations: Evidence From a
Laboratory Experiment.” Unpublished manuscript, Federal Reserve Bank of Boston, 2011.
Canon, Maria E. “The Role of Schools in the Production of Achievement.” Working Paper No. 2010-042A,
Federal Reserve Bank of St. Louis, October 2010; http://research.stlouisfed.org/wp/more/2010-042.
Duca, John V. and Kumar, Anil. “Financial Literacy and Mortgage Equity Withdrawals.” Unpublished manuscript,
Federal Reserve Bank of Dallas, December 2010.
Duygan-Bump, Burcu; Levkov, Alexey and Montoriol-Garriga, Judit. “Financing Constraints and Unemployment:
Evidence from the Great Recession.” Working Paper No. QAU10-6, Federal Reserve Bank of Boston, October
2011; www.bos.frb.org/bankinfo/qau/wp/2010/qau1006.htm.
Zhao, Bo and Coyne, David. “Designing Formulas for Distributing State Aid Reductions.” New England Public
Policy Center Working Paper No. 11-2, Federal Reserve Bank of Boston, 2011;
www.bos.frb.org/economic/neppc/wp/2011/neppcwp112.htm.

462

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

Federal Reserve Bank of St. Louis Review,
Annual Index, 2011
JANUARY/FEBRUARY
Thomas A. Garrett and Russell M. Rhine

“Economic Freedom and Employment Growth in U.S. States”
Craig P. Aubuchon, Juan C. Conesa, and Carlos Garriga

“A Primer on Social Security Systems and Reforms”
Parantap Basu and William T. Gavin

“What Explains the Growth in Commodity Derivatives?”
Chanont Banternghansa and Michael W. McCracken

“Real-Time Forecast Averaging with ALFRED”

MARCH/APRIL
Xin Wang and Yi Wen

“Can Rising Housing Prices Explain China’s High Household Saving Rate?”
Subhayu Bandyopadhyay and Suryadipta Roy

“Political Economy Determinants of Non-agricultural Trade Policy”
Silvio Contessi and Johanna L. Francis

“TARP Beneficiaries and Their Lending Patterns During the Financial Crisis”
Rajdeep Sengupta and Mara Faccio

“Corporate Response to Distress: Evidence from the Asian Financial Crisis”

MAY/JUNE
David C. Wheelock

“Have Acquisitions of Failed Banks Increased the Concentration of U.S. Banking Markets?”
Rubén Hernández-Murillo, Leslie S. Ott, Michael T. Owyang, and Denise Whalen

“Patterns of Interstate Migration in the United States from the Survey of Income and Program
Participation”
Yi Wen and Huabin Wu

“Dynamics of Externalities: A Second-Order Perspective”
Kristie M. Engemann, Rubén Hernández-Murillo, and Michael T. Owyang

“Regional Aggregation in Forecasting: An Application to the Federal Reserve’s Eighth District”
F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W

N OV E M B E R /D E C E M B E R

2011

463

JULY/AUGUST
James Bullard

“Measuring Inflation: The Core Is Rotten”
Axel A. Weber

“Challenges for Monetary Policy in the European Monetary Union”
Julie L. Hotchkiss and Menbere Shiferaw

“Decomposing the Education Wage Gap: Everything but the Kitchen Sink”
Rodolfo Manuelli and Adrian Peralta-Alva

“‘Frictions in Financial and Labor Markets’: A Summary of the 35th Annual Economic Policy Conference”

SEPTEMBER/OCTOBER
Christopher J. Waller

“Independence + Accountability: Why the Fed Is a Well-Designed Central Bank”
Christopher J. Neely

“A Foreign Exchange Intervention in an Era of Restraint”
Richard G. Anderson and Barry E. Jones

“A Comprehensive Revision of the U.S. Monetary Services (Divisia) Indexes”
Christopher J. Neely

“A Survey of Announcement Effects on Foreign Exchange Volatility and Jumps”

NOVEMBER/DECEMBER
Raphael Auer and Sébastien Kraenzlin

“International Liquidity Provision During the Financial Crisis: A View from Switzerland”
David C. Wheelock

“Banking Industry Consolidation and Market Structure: Impact of the Financial Crisis and Recession”
Daniel L. Thornton

“The Effectiveness of Unconventional Monetary Policy: The Term Auction Facility”
Thomas A. Garrett

“A Federal Reserve System Conference on Research in Applied Microeconomics”

464

N OV E M B E R / D E C E M B E R

2011

F E D E R A L R E S E R V E B A N K O F S T. LO U I S R E V I E W