Fire Sales, Foreign Entry and Bank Liquidity

Fire Sales, Foreign Entry and Bank Liquidity1
Viral V. Acharya2
Hyun Song Shin3
London Business School and CEPR
Princeton University
Tanju Yorulmazer4
J.E.L. Classi…cation: G21, G28, G32, E58, D61
Keywords: Crises, Systemic risk, Distress,
Liquidation cost, Limited pledgeability
This Draft: May 2007
1 We are grateful to Franklin Allen, Heitor Almeida and Douglas Gale for their suggestions, to
seminar participants at London Business School and University of Durham for their comments, and
to Yili Zhang for excellent research assistance. All errors remain our own.
2 Contact: Department of Finance, London Business School, Regent’s Park, London –NW1 4SA,
England. Tel: +44 (0)20 7262 5050 Fax: +44 (0)20 7724 3317 e–mail: vacharya@london.edu.
Acharya is also a Research A¢ liate of the Centre for Economic Policy Research (CEPR).
3 Contact: Princeton University, Bendheim Center for Finance, 26 Prospect Avenue, Princeton,
NJ 08540-5296, US. Tel: +1 609 258 4467, Fax: +1 609 258 0771, E–mail: hsshin@princeton.edu.
4 E–mail: ty232@nyu.edu.

---

Fire Sales, Foreign Entry and Bank Liquidity
Abstract
Bank liquidity is a crucial determinant of the severity of banking crises. In this paper,
we consider the e¤ect of …re sales and foreign entry on banks’ex ante choice of liquid asset
holdings, and the ex post resolution of crises.
In a setting with limited pledgeability of
risky cash ‡ows and di¤erential expertise between banks and outsiders in employing banking
assets, the market for assets clears only at …re-sale prices following the onset of a crisis –
and outsiders may enter the market if prices fall su¢ ciently low. While …re sales make it
attractive for banks to hold liquid assets, foreign entry reduces this incentive. We exhibit
international evidence on foreign entry following crises and on banks’ex ante liquidity choice
that are consistent with the predictions of the model. Our framework allows us to address
the key welfare question as to when there is too much or too little liquidity on bank balance
sheets relative to the socially optimal level.
J.E.L. Classi…cation: G21, G28, G32, E58, D61
Keywords: Crises, Systemic risk, Distress, Liquidation cost, Limited pledgeability
1

---

1
Introduction
In the aftermath of a major …nancial crisis, one of the most pressing tasks for a country is
to restructure its banking sector which emerges saddled with large non-performing claims
against distressed borrowers - claims that are backed by collateral assets whose prices have
fallen to a fraction of their levels before the crisis. After the …res have been put out, there
follows a protracted period of banking sector resolution. The crises in Scandinavia (Sweden
in particular) in 1992, Mexico in 1994-5, Thailand, Korea and Indonesia in 1997-8, and Turkey
in 2001, are all instances that highlight the two-stage nature of …nancial crises, where the
initial “acute” stage is followed by the longer-term “chronic” stage of bank restructuring.
The common theme that runs throughout both the academic research and the o¢ cial
pronouncements is the di¢ culty of …nding ready buyers of distressed assets in the midst of
a crisis.
During times of distress, when a bank needs to restructure its balance sheet, the
potential buyers of its assets are other banks that have also been severely a¤ected by the
crisis. Indeed, these potential buyers are also likely to be experiencing similar problems, and
may not have enough equity capital or debt capacity to purchase these assets. This theme
is a familiar one from corporate …nance (see Williamson, 1988, and Shleifer and Vishny,
1992), but leads to especially acute problems in banking given the relative opacity of bank
balance-sheets and the high sensitivity of banking assets to major macroeconomic shocks.
Allen and Gale (1998) has shown in the context of a banking model in the spirit of Diamond
and Dybvig (1983) how “cash-in-the-market”(or …re-sale) pricing leads to distressed pricing
in some states wherein asset prices fall below their fundamental value. Anticipation of such
crises determines the ex ante portfolio choice of the banks. Allen and Gale’s analysis is part
of a recent body of work that has explored the propagation of crises through the asset side
of banks’balance sheets, where marking to market in an environment of price falls transmits
distress across banks. We will review this literature shortly.
Our paper builds on two important themes from this recent literature. The …rst is the ex
post issue of the resolution of a banking crisis. Surviving banks in the crisis-stricken economy
may not have the resources to take over failing banks’assets. Banking assets therefore fall
in price so as to clear the market. However, if the crisis is severe enough, assets may end up
in the hands of foreign buyers (both banks and non-banks). Often, such outside buyers are
short-term holders who re-package or securitize the assets for selling on to portfolio investors.
However, outsiders may be unable to realize the full value of the assets for the familiar reason
that bank assets (loans in particular) derive much of their value from the monitoring and
collection e¤orts of loan o¢ cers who can in‡uence the actions of the debtors. Hence, when
distressed assets end up in the hands of outsiders, we may expect deadweight costs from
ine¢ cient allocation of assets.
2

---

The second theme is the ex ante choice of portfolio in anticipation of such crisis events
since surviving banks stand to make windfall pro…ts if they can purchase distressed assets at
low prices, as well as to take over the depositor base of a failed competitor bank.
Even if
crises arrive infrequently, the potential gains from acquisitions at …re sales could be large, and
we would expect banks’choice of their portfolio to take account of such states.
The most
important ingredient in banks’ portfolio choice is its holding of liquid assets, the objective
being to hold enough liquid assets so that in the event that the bank survives the crisis, it
will have resources to take advantage of the low purchase price of distressed assets.
Our objective in this paper is to follow up on both themes - the ex post resolution and
the ex ante portfolio choice. Our approach is primarily theoretical but we also provide some
empirical evidence.
First, we examine in a formal model the implications of endogenous
price e¤ects, …re sales and foreign entry.
We examine the portfolio choice of banks with
the goal of ascertaining both the equilibrium level of liquid asset holdings of banks, and the
socially optimal level of liquidity. We show that banks’equilibrium holding of liquid assets
is decreasing in the pledgeability of risky cash ‡ows and the health of the economy.
The
pledgeability of risky cash ‡ows also turns out to be the critical determinant of whether banks
hold too little liquidity relative to the socially optimal level.
When pledgeability is high,
banks hold less liquidity than is socially optimal due to risk-shifting incentives; otherwise,
banks may hold even more liquidity than is socially optimal in order to capitalize on …re
sales.
Second, we present some suggestive evidence in support of our theory based on the asset
sales observed in the banking and …nancial institutions sector in Asian countries following
the 1997 Asian …nancial crisis. In particular, we show that foreign banks are more active
players during crises resolution than otherwise: they acquire more assets and also take on
greater controlling stakes (greater than 50%). Consistent with our assumption that surviving
banks of the banking sector are also liquidity-strapped during crises whereas foreign banks
are not, we show that foreign banks acquire larger but weaker targets in the aftermath of
crises compared to other years, whereas the pattern is reversed for acquisitions by domestic
banks.
We also explore the theme of the relative expertise of domestic banks versus foreign
banks in the banking business. We document the following “di¤erence-of-di¤erence” e¤ect.
When we compare the acquisitions across industry sectors (compared to within-industry ac-
quisitions), foreign acquirers are involved with larger and better performing targets than
are domestic acquirers.
In other words, foreign acquirers are more heavily represented in
cross-sector takeovers of larger banks.
We believe this to be evidence of the lack of pur-
chasing power of the natural purchasers of distressed assets - namely, the surviving domestic
competitors to the failed …rms.
3

---

Third, we present descriptive cross-country evidence on the asset liquidity of banks across
countries. This evidence suggests that banks’ choice of liquidity does seem to vary along
dimensions that we would expect to be correlated with di¢ culty in raising external …nance
and severity of …nancial distress. We show that banks hold more liquid assets in those coun-
tries that have (i) less developed accounting standards; (ii) lower total market capitalization
relative to GDP; and, (iii) lower liquidity in stock markets.
Finally, we discuss how our model’s implications on management of liquidity by banks
over the business cycle square up with existing evidence and the recently documented facts
concerning leverage targeting by banks and its e¤ects on ampli…cation of the business cycle.
Section 2 presents the related literature. Sections 3 and 4 set up the benchmark model
without outsiders and characterize the e¤ect of …re sales on bank liquidity. Section 5 considers
liquidity-endowed outsiders and the e¤ect of their entry on bank liquidity. Section 6 provides
descriptive empirical evidence on …re sales, foreign entry and liquid asset holdings of banks.
Section 7 concludes. All proofs not in the main text are in the Appendix.
2
Related literature
Our paper is motivated in part by the policy-related literature on bank restructuring and
…nancial crises. The o¢ cial documents from the IMF (2003) and the Basel Committee on
Banking Supervision (2002) arose from an extensive consultation process among the leading
industrialized countries following the series of …nancial crises in the late 1990s. The prefaces
to these o¢ cial documents make it clear that the two reports were coordinated attempts to
provide advice on “best practice” on bank resolution, distilling insights from the experience
gained from tackling banking crises in the 1980s and 1990s. The literature on banking sec-
tor resolution is vast, but shares the key themes examined in our paper (see Acharya and
Yorulmazer, 2005, for a summary).
More broadly, our paper has links to the recent literature on the role of foreign direct
investment (FDI) ‡ows in the aftermath of …nancial crises. Aguiar and Gopinath (2004) have
recently documented evidence that the high FDI ‡ows into the crisis-stricken countries of the
1997 Asian …nancial crisis had many of the features of …re-sales: median o¤er price to book
ratios were substantially lower for the purchase of cash-strapped …rms, especially in 1998 when
national players had low liquidity, resulting in a boost in mergers and acquisitions involving
foreign players. Their paper provides a systematic empirical counterpart to the hypothesis
raised by Krugman (1998) that the investment ‡ows into Asia following the crisis in 1997 and
Mexico following the crisis in 1995 were suggestive of foreign …rms taking advantage of low
4

---

prices of real assets.1 Although we do not address explicitly the role of “foreign” outsiders
in what follows, our model has important implications for their role following a widespread
…nancial crisis. In particular, the welfare implications of our model on the issue of domestic
outsider involvement in the resolution of banking problems are closely related to the issue of
foreign entry, as we will detail below.
From a more narrow modeling perspective, the relationship between liquidity and asset
prices has been used in the literature to examine a number of interesting issues such as
…nancial market runs (Bernardo and Welch, 2004, and Morris and Shin, 2004), strategic
lending and trading (Donaldson, 1992 and Brunnermeier and Pedersen, 2005), contagion
through asset prices (Diamond and Rajan, 2001, Gorton and Huang, 2004, Schnabel and
Shin, 2004, Allen and Gale, 2005, and Cifuentes, Ferrucci and Shin, 2005), and optimal
failure resolution (Acharya and Yorulmazer, 2007, 2005). While liquidity can a¤ect asset
prices, most of the literature cited above treats the level of liquidity (of banks) as exogenous.
In this paper, we concentrate on the e¤ect of liquidity on (endogenously derived) …re sales
during systemic crises, and, in turn, their e¤ect on equilibrium liquidity. On this score, our
paper is more in the spirit of recent papers by Allen and Gale (2004) and Gorton and Huang
(2004) who investigate how liquidity is endogenously determined. Allen and Gale (2004),
for example, build a model where bank runs result in …re-sale liquidation of banking assets.
Investors endogenously choose the level of the liquid asset, which they use to purchase banking
assets. Since on average the liquid asset has a lower return than the risky asset, investors have
to be compensated for holding liquid assets, which can only be possible if they can purchase
the risky asset at a discount leading to cash-in-the-market pricing of the risky asset.
Our model highlights the limited pledgeability arising from moral hazard (in the manner
of Holmstrom and Tirole (1998)) as an important determinant in precluding the ready in‡ow
of capital to a crisis-stricken banking sector.
In this sense, we address the key question
of why the high shadow value of capital may co-exist with ready funds outside the system
seeking investment opportunities. A similar theme is explored in an asset pricing context by
Krishnamurthy and He (2006). Equipped with our richer framework, we can then address
the key welfare issues connected with foreign entry and resolution, and the welfare issues
consequences of liquidity choice at the ex ante stage.
Since liquid assets usually have lower returns than illiquid assets, banks may rationally
choose to rely on an interbank market or a lender of last resort (LOLR). Bhattacharya and
1 Krugman’s article provides interesting headlines from newspapers that talk about foreign entry due to
…re-sale prices in crisis-stricken countries: “Korean companies are looking ripe to foreign buyers” (New York
Times, Dec 27), “Some U.S. companies see …re sale in South Korea”(Los Angeles Times, Jan 25), “Some
companies jump into Asia’s …re sale with both feet” (Chicago Tribune, Jan 18), “While some count their
losses in Asia, Coca-Cola’s chairman sees opportunity” (Wall Street Journal, Feb 6). In news related to the
banking sector, Seoul Bank and Korea First Bank were under consideration for auction to foreign bidders.
5

---

Gale (1987) build a model of the interbank market where individual banks that are subject
to liquidity shocks coinsure each other against these shocks through a borrowing-lending
mechanism. However, in this model, the composition of liquid and illiquid assets in each
bank’s portfolio and the liquidity shocks are private information. Hence, banks have an
incentive to under-invest in liquid assets and free-ride on the common pool of liquidity in the
interbank market. Repullo (2005) shows that the existence of LOLR results in banks holding
a lower level of the liquid asset as they rely on the LOLR for liquidity.2 While we do not
consider inter-bank lending in this paper, we discuss in Section 7 extensions of our model and
the implications of regulatory closure policies on bank liquidity.
3
Benchmark model
Before presenting the formal model, we …rst give an informal description of the building
blocks, and the key assumptions. We consider a setting with a large number of banks.
Banks solve a portfolio choice problem as to how much to invest in risky assets, which are
assumed to have diminishing returns to scale, and how much to park in the safe asset as
liquid reserves. This portfolio choice problem acquires an inter-temporal dimension given the
limited pledgeability of risky cash ‡ows and the bene…t from holding liquidity in states where
banks can pro…t from asset purchases. Speci…cally, while banks have a preference for the risky
asset due to its “option” value in the traditional risk-shifting sense, there is a counteracting
preference for the safe asset due to its greater liquidity relative to the risky asset. Banks’
choice of liquidity trades o¤ the expected returns from the two kinds of assets (adjusted for
option value) taking account of this need for inter-temporal transfers of liquidity. The socially
optimal level of liquid asset holdings in banks’portfolio maximizes the value of banks as a
whole (that is, without any risk-shifting problem). Throughout our analysis, we assume that
deposits are insured by the regulator. To start with, there is no cost of providing insurance
to depositors, in which case the assumption of insured deposits does not play a key role in
determination of liquidity choices of banks. We introduce outsiders in Section 5 and discuss
in Section 7 the implications of costly deposit insurance.
The time line of the formal model is outlined in Figure 1. We consider an economy where
time is indexed by t, and where t 2 0; 1;1;2 . In our model, there are banks, bank owners,
2
depositors and a regulator, who serves to allocate resources in an e¢ cient manner following
a crisis.
2 Gonzalez-Eiras (2003), using Argentine data, tests this argument. He investigates the episode in December
1996 when the Central Bank of Argentina signed an agreement to have access to contingent credit lines with
a group of international banks that enhanced its ability to act as a LOLR. He shows that banks have relied on
the enhanced ability of the Central Bank of Argentina for liquidity and this has resulted in an approximately
6.7 % reduction in banks’liquid asset holdings.
6

---

In particular, we assume that there is a continuum of banks with measure 1, where each
bank has access to its own depositor base. The depositor base of a bank is itself a continuum
of depositors of measure 1. Bank owners as well as depositors are risk-neutral, so that banks
aim to maximize the sum of expected pro…t over time.
Depositors receive a unit endowment at t = 0 and at t = 1. Depositors have access to a
reservation investment opportunity that gives them a utility of 1 per unit of investment. At
dates t = 0 and t = 1, depositors choose to invest their good in this reservation opportunity
or in their bank. Deposits take the form of a simple debt contract with maturity of one period
and the promised deposit rate is not contingent on bank’s investment decisions or its realized
returns.
Banks collect one unit of deposits from depositors and make investments to maximize the
sum of expected pro…ts at t = 1 and t = 2. There is no discounting. In particular, banks
choose a portfolio by investing l units in a safe asset and the remaining (1
l) units in a risky
asset, which is to be thought of as a portfolio of loans to …rms in the corporate sector.
The payo¤ of the bank from its loan is
Rt
with prob
t
eRt,where eRtistherandomvariable:
eRt=
:
(1)
0
with prob 1
t
Rt can be viewed as the notional value of the loan.
The realization of eRt is independent
across banks, so that by the law of large numbers, precisely
t of the banks have positive
payo¤. Moreover, the returns are independent over time.
However, there is aggregate uncertainty in that
t is itself random.
Hence, there is
uncertainty over the proportion of banks that receive positive payo¤.
In what follows, we
will denote by E ( t) the expected realization of t.
We assume that the risky technology eR0 has diminishing returns to scale, that is, the
return R0 is decreasing in (1
l). In order to get a closed form solution, we use a setup
similar to Holmstrom and Tirole (2001) and let
(1
l)
R0(l) = b
(2)
2
Hence, R0 takes values between b
1 and b; and dR0 = 1 > 0. For simplicity we assume
2
dl
2
that eR1 isaconstantreturnstoscaletechnologywithR1>1. Thishelpsusconcentrateon
the e¤ect of choice of liquid asset only in the …rst period and simpli…es the analysis without
a¤ecting our results.
At the intermediate date t = 1 ; the outcome of the …rst-period investments in the risky
2
asset becomes public information, though banks can collect these returns fully only at t = 1.
7

---

The safe asset is completely liquid and pays one unit at any date for each unit invested.
The risky asset is however not completely liquid due to a moral-hazard problem at the bank
level. From date t = 1=2 to date t = 1, if the bank does not exert e¤ort, then when the
return is high, it cannot generate Rt but only Rt
and its owners enjoy a non-pecuniary
bene…t of B 2 (0; ): For the bank owners to exert e¤ort, appropriate incentives have to be
provided by giving bank owners a minimum share of the bank’s pro…ts. We denote this share
as . If rt is the cost of borrowing deposits, then the incentive-compatibility constraint is:
t (Rt
rt) > t ((Rt
)
rt) + B
(IC)
Using this constraint, we can show that bank owners need a minimum share of
= B=
to
monitor these loans properly.3 Therefore, the bank can raise at most a fraction
= 1
of its future income in the capital market if it is required to exert e¤ort to monitor loans.4
We assume that at t = 0, the entire share of the bank pro…ts belongs to the bank owners,
and therefore, moral hazard is not a concern at the beginning.
We assume that deposits are fully insured in the …rst period. Note that the second period
is the last period in our model and there is no further investment opportunity. As a result,
our analysis is not a¤ected by whether deposits are insured for the second investment or not.
Finally, we make technical assumptions (A1)–(A4) which are contained in the Appendix.
We refer to these at a few relevant points of our analysis.
If a bank’s return from the …rst-period investment is high, then the bank operates one
more period and makes the second-period investment. For a bank to continue operating for
another period, it needs to pay its old depositors r0: But, by our assumption (A2), a failed
bank can not generate the necessary funds to avoid default. Thus, if the return is low, then
the bank is in default and the deposit insurance provider puts up the bank for sale at t = 1 .
2
When banks with the high return from the …rst period investment want to acquire failed
banks’assets, they use the liquid asset in their portfolio and/or try to raise funds from the
capital market against their future return. However, because of moral hazard, banks cannot
fully pledge their future income, but only a fraction
of it.
Depending on the …rst period returns, a proportion k of banks fail. Since banks are
3 See Hart and Moore (1994) and Holmstrom and Tirole (1998) for models with similar incentive-
compatibility constraints.
4 Bank-level moral hazard in our model can be addressed by greater ownership of banks by insiders. Caprio
et al (2005) document that banks in general are not widely held (a widely-held bank is one that has no legal
entity owning 10 percent or more of the voting rights), similar to the …ndings of La Porta et al (1999) for
corporations in general. This observation is stronger for countries with weaker shareholder protection laws.
They also …nd that greater inside ownership enhances bank valuation in such countries. Overall, these …ndings
are consistent with the key assumptions of our model since weaker protection laws should imply a greater risk
of cash-‡ow appropriation by insiders, and, in turn, lead to greater inside ownership of banks in equilibrium.
8

---

identical at t = 0, we denote the possible states at t = 1 with k.
4
Analysis
We analyze the model proceeding backwards from the second period to the …rst period.
The surviving banks operate for another period at t = 1. The probability of having the
high return for each bank is equal to
1 for all banks.
As this is the last period, there is
no further investment opportunity and no asset sales take place in this period. Since the
risky asset has a higher expected return than the safe asset and there is no asset purchase
opportunity, banks invest all their funds in the risky asset at t = 1. The expected payo¤ to
the bank from its second-period investment is thus
1[R1
r1] = p.
Next, we investigate the sale of failed banks’assets and the resulting asset prices.
4.1
Asset sales and liquidation values
In examining the purchase of failed banks’ assets, several interesting issues arise. First,
surviving banks may compete with each other if there are enough resources with them to
acquire all failed banks’assets. Second, unless the game for asset acquisition is speci…ed with
reasonable restrictions, an abundance of equilibria arises. To keep the analysis tractable and
at the same time reasonable, we make the following assumptions:
(i) The regulator pools all failed banks’assets and auctions these assets to the surviving
banks.
(ii) Denoting the surviving banks as i 2 [0;(1 k], each surviving bank submits a schedule
yi(p) for the amount of assets they are willing to purchase as a function of the price p at which
a unit of the banking asset (inclusive of associated deposits) is being auctioned.
(iii) The regulator cannot price-discriminate in the auction.
(iv) The regulator determines the auction price p so as to maximize the output of the
banking sector, but subject to the natural constraint that portions allocated to surviving
banks add up at most to the number of failed banks, that is,
Z 1 kyi(p)di k:
(3)
0
(v) We focus on the symmetric outcome where all surviving banks submit the same sched-
ule, that is, yi(p) = y(p) for all i 2 [0;1].
First, we derive the demand schedule for surviving banks. Note that a surviving bank
9

---