International Banking

c h a p t e r 8 ....................................................................................................................................................

THE CORPORATE

STRUCTURE OF

INTERNATIONAL

FINANCIAL

CONGLOMERATES

COMPLEXITY AND ITS

IMPLICATIONS FOR SAFETY

AND SOUNDNESS .....................................................................................................................................................

richard herring jacopo carmassi

1

Introduction

.........................................................................................................................................................................................

International Wnancial conglomerates have become an increasingly important

feature of the Wnancial landscape. Universal banking countries have long integrated

the securities business with traditional commercial banking, but over the last

1 The authors are grateful to the editors of this volume and Robert Eisenbeis for helpful comments

on an earlier draft.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 172 18.6.2009 10:55am

decade the US and Japan, which formerly required strict separation of commercial

banking from the securities business, have permitted banks to combine these two

activities subject to some limitations. Increasingly combinations of banking and

securities business have expanded to include insurance operations as well. Allianz

in Germany, ING and Fortis in the Netherlands, Credit Suisse in Switzerland, and

Citi in the US have all made important cross-sector acquisitions in recent years to

combine banking and insurance activities (although both Credit Suisse and Citi

have subsequently divested some of their insurance acquisitions). Indeed, virtually

all of the large, international Wnancial institutions are to some extent Wnancial

conglomerates combining at least two of the three formerly distinct functions of

banks, securities Wrms, or insurance companies.

This consolidation and conglomeration appears to be motivated by hopes for

cost savings and revenue enhancements from large, lumpy expenditures on infor-

mation technology (Group of Ten, 2001). Economies of scope in production 2 may

be important whenever a signiWcant Wxed cost can be shared across several diVerent

products. In addition to investments in information technology, several other

kinds of Wxed costs may be important—the costs of distribution channels, man-

aging a client relationship, or establishing and maintaining a sound reputation and

brand image. But diseconomies of scope such may also be important (Herring and

Santomero, 1990). In any event it is diYcult to Wnd evidence of signiWcant

economies of scope in the data. Indeed, Laeven and Levine (2007) Wnd evidence

of a diversiWcation discount applied to Wnancial conglomerates.

The trend toward consolidation and conglomeration may also be motivated by the

hope of achieving greater market power. By controlling the full range of substitutes for

a Wnancial product, a Wnancial conglomerate may be able to raise prices above

marginal costs. In order to sustain such market power, the Wnancial conglomerate

would also need to be able to limit entry and enforce mandatory joint product sales.

Of course, antitrust policy is intended to prevent such abuses. Moreover, intensiWed

cross-border competition and technological advances that render all major markets

for Wnancial products highly contestable make it unlikely that any Wnancial conglom-

erate could sustain market power should antitrust policy prove ineVectual. 3

More than thirty countries have restructured and uniWed their regulatory and

supervisory systems to deal with Wnancial conglomerates in a more integrated

2 Economies of scope in consumption may also be important. But they could be exploited by using

the distribution network of one institution to sell packages of Wnancial services produced by other

Wrms and thus cannot explain the formation of institutions such as LCFIs that produce and distribute

several diVerent kinds of Wnancial services (Herring and Santomero, 1990). 3 See Berger, Demsetz, and Strahan, 1999 for a review of the literature on consolidation and market

power. They note that market power is most likely to be of concernwith regard to in-market rather than

market extension mergers and that retail customers are more likely to be adversely aVected than

wholesale customers. Berger (1995) makes a careful distinction between the market power and eYcient

structure hypotheses taking account of both X-eYciency and scale eYciency. He concludes that neither

the market power nor eYcient scale hypotheses ‘are of great importance in explaining bank proWts’.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 173 18.6.2009 10:55am

international financial conglomerates 173

fashion (Herring and Carmassi, 2008). Many of these international Wnancial

conglomerates have achieved a scale of operation and centrality in the functioning

of the international Wnancial system that render them systemically important.

Traditionally, systemic concerns have been the preoccupation of bank regulators,

but these concerns do not diminish when a bank becomes part of a group that

includes insurance and securities activities as well. Although it is possible that

larger, more diversiWed international Wnancial conglomerates will be less likely to

fail, if a failure should occur the spillover eVects on the rest of the Wnancial system

are bound to be greater. Moreover, the heavy involvement of these Wrms in trading

activities around the clock, around the globe means that the authorities would have

very little time to react if one should experience extreme Wnancial distress.

Our central premise is that the complexity of the corporate structures that most

international Wnancial conglomerates have developed is itself a signiWcant source of

systemic risk. In the event of bankruptcy, hundreds of legal entities would need to be

resolved. Since most of these Wrms are managed in an integrated fashion along lines of

business with only minimal regard for legal entities, national borders, or functional

regulatory domains, and with substantial and complex intragroup relationships,

simply mapping an institution’s business activities into its legal entities presents a

formidable challenge. Moreover, these legal entities would be subject to scores of

diVerent national regulatory and bankruptcy procedures, many of which conXict.

The corporate complexity of international Wnancial conglomerates is likely to

impede timely regulatory intervention and disposition. This exacerbates the moral

hazard implicit in the Wnancial safety net and diminishes market discipline on

some of the most systemically important institutions, while at the same time

constraining the ability of the supervisory authorities to substitute regulatory

discipline for market discipline. In eVect, several of these institutions may have

become ‘too complex to fail’.

We will begin with a consideration of the corporate structure that international

Wnancial conglomerates might prefer in the absence of regulatory and tax distortions.

Then we will examine some of the (largely unintended) consequences for corporate

structure of tax and regulatory policies. We will conclude with an analysis of some of

the challenges this corporate complexity poses to an orderly winding down of an

international Wnancial conglomerate. But Wrst we present an overview of the large,

complex Wnancial institutions that we use to illustrate several aspects of the problem.

Large complex financial institutions

.........................................................................................................................................................................................

The regulatory authorities have identiWed sixteen Wnancial conglomerates as

large, complex Wnancial institutions (LCFIs) that are of crucial importance to the

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 174 18.6.2009 10:55am

174 the theory of banking

functioning of the international Wnancial system. 4 LCFIs ‘include the world’s

largest banks, securities houses and other Wnancial intermediaries that carry out

a diverse and complex range of activities in major Wnancial centres’ (Bank of

England, 2007b: 29). These Wrms are key intermediators of risk through their

market-making activities and principal risk-taking, as well as their provision of

liquidity to capital markets. The concept was given empirical content in the Bank

of England’s Financial Stability Review (Bank of England, 2001) and since that time

both the Financial Stability Review and the International Monetary Fund’s Global

Financial Stability Report have tracked developments among this group of Wnancial

institutions. The Bank of England (2007a: 7) has expressed concern about the rising

systemic importance of LCFIs: ‘Given their scale and their pivotal position in most

markets, distress at an LCFI could have a large, unanticipated, impact on other

Wnancial markets participants. This could arise from losses on direct exposures to

an LCFI that failed or from the wider market implications of actions taken by an

LCFI to manage problems.’

Like the Holy Roman Empire, which was not holy, nor Roman, nor an empire,

the term ‘large and complex Wnancial institutions’ is imprecise. It does not include

some of the largest Wnancial institutions, nor some of the institutions that pursue

the most diverse lines of business. Criteria for inclusion in the group require that

an institution achieve a position as one of the ten largest participants in two or

more of the following activities: book runners of international bond issues, book

runners of international equity issues, book runners of global syndicated loans,

notional interest-rate derivatives outstanding, foreign exchange revenue, or world-

wide assets under custody (Hawkesby, Marsch, and Stevens, 2003). LCFIs are

completely dominant in some of these activities. For example, just two LCFIs act

as custodians for around three-quarters of all assets in value terms (Bank of

England, 2007a: 30) and three LCFIs are the dominant intermediaries in the market

for credit derivatives (Bank of England, 2007a: 35).

Table 8.1 displays the sixteen AQ1institutions that are currently classiWed as LCFIs by

the Bank of England (Bank of England, 2007: 29) and the IMF. At year end 2006, all

of these institutions (except Lehman Brothers) ranked among the world’s twenty-

Wve largest banking groups in terms of total assets. Although these institutions

diVer with regard to the diversity of their activities (see column 6, the HHI for

revenues for individual lines of business) and the extent of their international

engagement (see column 4, the percentage of foreign subsidiaries, and column 5,

the percentage of net foreign income), they are all major participants in inter-

national capital markets. LCFIs have had a greater than 70 per cent market share as

4 The term, LCFI, was introduced by a task force of the Financial Stability Forum, the G10

Ministers and Governors and the Basle Committee formed in 2000 to review the issues likely to arise

in winding down an LCFI (Hüpkes, 2005). The Group of Ten Report (2001) on consolidation also

considered a number of problems that might arise as consequence of the growth of large and complex

Wnancial organizations.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 175 18.6.2009 10:55am

international financial conglomerates 175

Ta b le 8 .1 . O ve rv ie w o f la rg e co m p le x fi n a n ci a l in st it u ti o n s

1 2

3 4

5 6

7 8

9 LC

F Is

To ta l a ss et s

(b il li o n s o f $ ,

ye a re n d 2 0 0 6 )1

To ta l

su b si d ia ri es

1 %

o f fo re ig n

su b si d ia ri es

% o f fo re ig n n et

in co m e b ef o re

ta x es

(2 0 0 6 )2

H H I— b u si n es s

li n es

re ve n u es

(2 0 0 6 )3

N u m b er

o f

co u n tr ie s4

S u b si d ia ri es

in O F C s,

n u m b er

5

S u b si d ia ri es

in O F C s, %

5

U B S A G

1 ,9 6 4

4 1 7

9 6 %

6 2 %

2 ,9 0 3

4 1

3 8

9 %

B a rc la ys

P lc

1 ,9 5 7

1 ,0 0 3

4 3 %

4 4 %

2 ,1 7 9

7 3

1 4 5

1 4 %

B N P P a ri b a s

1 ,8 9 7

1 ,1 7 0

6 1 %

5 1 %

1 ,8 4 3

5 8

6 2

5 %

C it i

1 ,8 8 4

2 ,4 3 5

5 0 %

4 4 %

4 ,1 2 2

8 4

3 0 9

1 3 %

H S B C H o ld in g s P lc

1 ,8 6 1

1 ,2 3 4

6 1 %

7 8 %

3 ,9 4 5

4 7

1 6 1

1 3 %

Th e R o ya l B a n k o f S co tl a n d

G ro u p P lc

1 ,7 1 1

1 ,1 6 1

1 1 %

3 4 %

1 ,9 6 6

1 6

7 3

6 %

D e u ts ch e B a n k A G

1 ,4 8 3

1 ,9 5 4

7 7 %

8 0 %

3 ,9 3 1

5 6

3 9 1

2 0 %

B a n k o f A m e ri ca

C o rp o ra ti o n

1 ,4 6 0

1 ,4 0 7

2 8 %

1 2 %

4 ,2 5 6

2 9

1 1 8

8 %

JP M o rg a n C h a se

& C o .

1 ,3 5 2

8 0 4

5 1 %

2 6 %

2 ,0 8 6

3 6

5 4

7 %

A B N A M R O H o ld in g N V �

1 ,3 0 0

6 7 0

6 3 %

7 7 %

1 ,3 8 1

4 3

3 7

6 %

S o ci é té

G én é ra te

1 ,2 6 0

8 4 4

5 6 %

4 6 %

4 ,1 2 8

6 0

6 4

8 %

M o rg a n S ta n le y

1 ,1 2 1

1 ,0 5 2

4 7 %

4 2 %

4 ,4 7 6

4 6

2 0 3

1 9 %

C re d it S u is se

G ro u p

1 ,0 2 9

2 9 0

9 3 %

7 1 %

3 ,8 6 8

3 1

5 3

1 8 %

M e rr il l Ly n ch

& C o ., In c.

8 4 1

2 6 7

6 4 %

3 5 %

4 ,0 8 9

2 5

2 3

9 %

G o ld m a n S a ch s G ro u p , In c.

8 3 8

3 7 1

5 1 %

4 8 %

5 ,3 9 1

2 1

2 9

8 %

Le h m a n B ro th e rs H o ld in g s In c.

5 0 4

4 3 3

4 5 %

3 7 %

7 ,8 0 7

2 0

4 1

9 %

N o te : Y e a r e n d 2 0 0 7 (u n le ss

o th e rw is e sp e ci fi e d ).

� A ft er

th e m o st re ce n t li st o f LC FI s (B a n k o f E n g la n d , 2 0 0 7 b ) w a s p u b li sh e d , a co n so rt iu m

o f th re e b a n ks

(R B S , Fo rt is , a n d S a n ta n d e r) a cq u ir e d A B N A M R O .

S o u rc es : 1 B a n ks co p e . D a ta

o n su b si d ia ri e s re fe r to

m a jo ri ty -o w n e d su b si d ia ri e s fo r w h ic h th e LF C I is th e u lt im a te

o w n e r w it h a m in im u m

co n tr o l p a th

o f 5 0 .0 1 % .

2 a n n u a l re p o rt s fo r e a ch

LC FI . N e t in co m e b e fo re

ta x es

w it h fi ve

e x ce p ti o n s: n e t in co m e a ft e r ta x e s fo r C it i, a n d n e t re ve n u e s fo r B a rc la ys

p lc , B N P P a ri b a s, Le h m a n B ro th e rs

H o ld in g s In c. , M e rr il l Ly n ch

& C o ., In c.

3 O li ve r W ym

a n .T h e H e rf in d a h l- H ir sc h m a n In d e x ra n g e s fr o m 0 to

1 0 ,0 0 0 a n d it is ca lc u la te d o n th e p e rc e n ta g e o f re ve n u e s p e r b u si n es s li n e .H

ig h e r va lu e s in d ic a te a h ig h e r d e g re e

o f sp e ci a li za ti o n . Lo w er

va lu e s im p ly a h ig h e r d e g re e o f d iv er si fi ca ti o n .

4 N u m b e r o f co u n tr ie s in

w h ic h th e LC FI h a s a t le a st o n e m a jo ri ty -o w n e d su b si d ia ry .

5 O ff sh o re

Fi n a n ci a l C e n te rs id en ti fi e d b y th e Fi n a n ci a l S ta b il it y Fo ru m (2 0 0 0 ). W e e x cl u d e S w is s su b si d ia ri e s fo r C re d it S u is se

a n d U B S a n d H o n g K o n g su b si d ia ri e s fo r H S B C . Fo u r

su b si d ia ri e s w e re

a ll o ca te d to

O FC s o n th e b a si s o f lo ca ti o n s d e si g n a te d in

th ei r n a m e s e ve n th o u g h B a n ks co p e d id

n o t sp e ci fy

a h o m e co u n tr y.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 176 18.6.2009 10:55am

176 the theory of banking

lead arrangers and book runners of issues of residential mortgage-backed secur-

ities, leveraged syndicated loans, corporate debt, and asset-backed securities during

the Wrst three quarters of 2007 (Bank of England, 2007b: 38). LCFIs have experi-

enced remarkable growth since the turn of the century, with total assets more than

doubling in size from 2000 to 2006. In 2006, trading assets constituted more than

one-third of the total (Bank of England, 2007a: 9).

LCFIs have also developed a remarkable degree of corporate complexity. In what

follows we focus on the number of majority-owned subsidiaries as an indicator of

corporate complexity. Of course, this is a somewhat arbitrary measure. The Federal

Reserve Board, for example, takes a more expansive view of control in bank holding

companies, establishing a 25 per cent ownership level as the threshold. Moreover, it

is a regrettably superWcial measure of corporate complexity. Unfortunately, the

Bankscope 5 data do not permit us to identify shell corporations or other inconse-

quential subsidiaries. Although it would be useful to supplement this simple

quantitative measure with an indication of each entity’s importance in the overall

Wnancial group, cross-guarantees, and role in the overall business structure, such

information is not publicly available for many subsidiaries. None the less, the

number of majority-owned subsidiaries is an indication of the magnitude of the

legal challenge that would confront the authorities in taking an LCFI through

bankruptcy. All of the LCFIs have several hundred subsidiaries. Eight have more

than 1,000 subsidiaries and one (Citi) has nearly 2,500 subsidiaries.

In the absence of tax and regulatory constraints, how much corporate complex-

ity would LCFIs choose to adopt? The formation of subsidiaries can be costly. In

addition to the start-up costs of obtaining a charter and creating a governance

structure, there are ongoing costs for accounting, Wnancial reporting, and tax

Wlings. None the less, LCFIs have adopted a considerable amount of corporate

complexity even within some countries where they are under no regulatory obli-

gation to do so. Germany, for example, has followed a universal banking model

that permits banking and securities activities to be conducted within a single legal

entity. Only investment funds, building societies, and insurance companies require

the establishment of a separate legal entity. None the less, Deutsche Bank, the

leading German bank, has over 300 fully owned domestic subsidiaries (Bankscope,

Oct. 2007). What are the perceived, compensating beneWts that justify the forma-

tion of corporate subsidiaries?

In the frictionless world of Modigliani and Miller (1958), a Wrm’s choice of

capital structure and, by extension, its corporate structure, cannot aVect its

value. But Wnancial institutions lack any rationale in such a world. As Berger,

et al. (1995: 394) note, most ‘[R]esearch on Wnancial institutions has begun with a

set of assumed imperfections’, which includes asymmetric information and

5 Bankscope is a global database containing information on public and private banks < http://

www.bvdep.com/en/bankscope.html>.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 177 18.6.2009 10:55am

international financial conglomerates 177

transactions costs, costs of Wnancial distress, taxes, and regulation. Each of these

imperfections may inXuence a Wnancial institution’s choice of corporate structure.

Asymmetric information and

transactions costs

.........................................................................................................................................................................................

Asymmetric information problems appear to aZict Wnancial institutions more

seriously than many other kinds of Wrms. Morgan (2002) presents evidence that

Wnancial institutions are inherently more opaque than other Wrms based on

disagreements among bond-rating agencies. Because many Wnancial institutions

specialize in lending to opaque borrowers and their trading positions can be easily

and almost instantaneously changed, they are hard to monitor. Morgan Wnds that

insurance companies may be even more opaque than banks since their primary

assets are privately placed, long-term loans and the indemnity risks they under-

write may be even more uncertain to outsiders than bank liabilities.

Asymmetric information problems arise when one party to a transaction or

relationship has information that the other does not, and it is too costly to write,

monitor, and enforce a contract that would compensate adequately for the imbalance

in information. When the objectives of the parties conXict, Wrms incur agency costs

because of concerns about adverse selection—the fear the better-informed party will

take advantage of the less-informed party by misrepresenting the quality of the

product or service—or moral hazard—the fear that, once the transaction takes

place, one party will covertly shift risk to the other’s disadvantage. Financial Wrms

have devised many diVerent ways of mitigating these costs, including, sometimes, the

creation of separate subsidiaries. Asymmetric information exacerbates conXicts of

interest, which may arise between shareholders and creditors, between shareholders

and managers, and between the Wrm and its customers. We will consider each in turn.

Asymmetric information: shareholders vs. creditors

The fundamental conXict of interest between shareholders and creditors springs

from diVerences in their pay-oV functions. After debt-servicing costs have been

paid, shareholders reap all the upside returns. They participate in the downside

losses, however, only to the extent of their equity stake. In contrast, the upside

return of creditors is limited to the promised return, while they may lose all that they

have lent. Creditors will, thus, generally prefer safer investments than shareholders.

With asymmetric information, creditors will be concerned that shareholders may

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 178 18.6.2009 10:55am

178 the theory of banking

engage in risk shifting after the terms of a loan have been set by substituting riskier

assets for the safer assets. To safeguard against this possibility creditors may charge a

higher premium and attempt to constrain the Wrm in a number of ways, perhaps

even refusing to lend. Kahn and Winton (2004) have shown that the choice of a

corporate structure can ease this problem. By forming a risky subsidiary, the Wrm

provides a commitment that limits its incentive to engage in risk-shifting. Placing

safer assets in a separate subsidiary increases the safe subsidiary’s net returns in bad

states of the world and reduces its incentives to engage in risk shifting. It may also

improve terms on which the safe subsidiary can obtain external Wnancing. Although

the Wrm may still have an incentive to engage in risk shifting in the riskier subsidiary,

Kahn and Winton (2004) argue that this limits the amount of risk-shifting that

can take place within the conglomerate. (For an opposing view, see Merton and

Perold, 1993).

In support of their theory, Kahn and Winton (2004) note the tendency of

commercial banks to form separate subsidiaries for their Wnance companies—for

insurance, companies to form separate subsidiaries for riskier policy lines, and

for investment banks to form separate subsidiaries for their riskier private equity

investments. Their theory also provides a rationale for good-bank/bad-bank

restructurings such as the regulatory restructuring of the Continental Illinois

National Bank and Mellon Bank’s creation of Grant Street Bank in 1988. In fact,

Kahn and Winton (2004: 2532) emphasize that several of the commonly advanced

rationales for ‘bad’ bank structures are not convincing unless the implications for

incentives to engage in asset shifting are taken into account.

Asymmetric information: shareholders vs. managers and

internal agency problems

International Wnancial conglomerates generally have broadly dispersed share-

holders with no one dominant owner. This separation of ownership from man-

agerial control means that shareholders face an asymmetric information problem

vis-à-vis the managers of a Wrm. This is a classic principal–agent problem in which

managers may be tempted to pursue their own objectives, such as empire building

or the enjoyment of lavish corporate perquisites, rather than serving the interests of

shareholders. This may lead to several diVerent kinds of resource misallocations that

diminish share values. Managers may be excessively risk-averse and seek to protect

their entrenched positions by underinvesting in risky, positive net-present-value

projects (Smith and Stulz, 1985). Or managers may take advantage of free cash Xows

to overinvest in value-destroying, negative net-present-value projects (Jensen,

1986). More broadly, managers may shirk.

Senior managers face similar issues with regard to managers lower down the

corporate hierarchy. These internal agency costs include managerial entrenchment,

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 179 18.6.2009 10:55am

international financial conglomerates 179

misallocations of resources, and rent-seeking behavior (Fulghieri and Hodrick,

2005). Although a number of corporate governance mechanisms deal with these

problems, the choice of organizational form can also be used as an instrument to

control the behavior of multiple agents and better align the incentives of owners

and managers.

If a particular line of business has compensation practices or a culture that is

very diVerent from other lines of business in the conglomerate, segregation of that

line of business into a separate entity may facilitate oversight and control. For

example, it has often proven diYcult to manage traders or deal-oriented invest-

ment bankers within the same compensation structure as relationship-oriented

commercial bankers. Corporate separateness provides greater Xexibility to better

align incentives with the interests of shareholders and tailor employment contracts

to prevailing market standards without destroying the culture necessary to make

the business segment successful. As Aron (1991: 506) observes, the normal practice

of tying the compensation of the manager of a business unit to the overall stock

value of the group may not provide eYcient incentives: ‘When a division is part of

a multiproduct corporation, the stock value of the Wrm is a noisy signal of the

market’s evaluation of any one divisional manager’s productivity. Loosely speaking,

the more noise there is in the signal, the costlier it is to properly motivate the

manager.’

Despite massive investments in management information systems, integrated

Wnancial conglomerates may Wnd it diYcult to track and evaluate the performance

of individual lines of business. Informal, internal capital markets sometimes

contribute to the blurring of performance and result in unintended cross subsidies

(Rajan, et al., 2000). 6 A degree of corporate separateness may be introduced to

sharpen strategic focus and improve monitoring. For example, some groups have

established separate units to handle client transaction processing with the intention

of clarifying the performance of other risk-taking units and giving senior managers

better control over costs, pricing, product design, and delivery of transactions

services. 7 This organizational innovation also facilitates benchmarking the trans-

action processing business against publicly traded, stand-alone businesses that

provide similar services.

Occasionally a Wrm may take the additional step of partially spinning-oV a

subsidiary so that it has a separate listing and can be publicly traded. As Habib,

Johnsen, and Naik (1997) observe, this enlists the help of capital markets in

generating information that should improve the quality of investment decision.

6 Holod and Peek (2006), however, provide evidence that internal capital markets in multi-bank

holding companies enhance the eYciency of capital allocation. In particular, internal secondary loan

markets avoid the asymmetric information problems faced by participants in the external secondary

loan market and thus mitigate Wnancial constraints faced by individual subsidiaries. 7 For a description of the formation of PROFITCO at Bankers Trust, which was the Wrst bank to

restructure its processing services in this way, see Guil, 2008.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 180 18.6.2009 10:55am

180 the theory of banking

It may also reduce the uncertainty of uninformed investors regarding the value of

the subsidiary. Both eVects should increase the value of the Wrm.

Firms may achieve some of the incentive beneWts by simply forming a separate

entity even though the spin-oV never actually occurs. Aron (1991: 505) notes that

‘The possibility of a future spinoV induces the divisional manager to act as if he

were being monitored and evaluated by the capital market, even though the capital

market’s evaluation is observed only if a spinoV actually occurs’.

Information asymmetry: Customer

concerns about conflicts of interest

.........................................................................................................................................................................................

ConXicts of interest are ubiquitous even in specialized Wnancial institutions, but, as

Walter (2003: 21) notes, ‘[A] matrix approach to mapping conXicts of interest

demonstrates that the broader the range of clients and products, the more numer-

ous are the potential conXicts of interest and the more diYcult is the task of

keeping them under control—and avoiding even larger franchise losses’. Customers

fear that a Wrm may use its informational advantage to their detriment. Firms

invest substantial resources to reassure clients and potential customers that they

will not be disadvantaged vis-à-vis the Wrm or other clients. Such eVorts include the

erection of ‘Chinese walls’ restricting the Xow of information across lines of

business, the adoption of codes of conduct reinforced with compliance audits,

and disclosures of potential conXicts (for a detailed study on conXicts of interest in

the Wnancial industry, see Walter, 2004).

Sometimes Wrms take the additional step of segregating activities in separate

subsidiaries. For example, investment advisory services may be provided by a

separate entity from underwriter and broker/dealer. Or, management consulting

services may be oVered through a separate entity in a separate location from the

parent to reassure customers that conWdential information would not be used in

lending decisions or to aid other Wrms in which the parent might have an

ownership position. Equally, corporate separateness may provide greater Xexibility

for operating units that would otherwise be constrained by conXict-of-interest

concerns or burdensome reporting requirements. For example, Cox and Curry

(2007: C12) reported that Goldman Sachs moved some of its proprietary trading

desks from its investment bank into a separate, asset-management unit. They

speculate that one of the advantages may be that ‘[T]he stock-arbitrage desk may

Wnd it has more freedom to invest in companies involved in mergers or acquisitions

that were once oV limits because of the investment bank’s activities as the world’s

top M and A adviser’.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 181 18.6.2009 10:55am

international financial conglomerates 181

Krozner and Rajan (1997) found evidence of this behavior in the way in

which US banks organized their investment-banking operations before the 1933

Glass-Steagall Act forced a separation between commercial and investment

banking. During this period, some banks organized their investment-banking

operations as an internal department within the bank, while others formed

separately incorporated aYliates with separate boards of directors. They found

that the market attached a higher risk premium to issues underwritten by

internal departments. Krozner and Rajan (1997: 475) conclude that this is

consistent with ‘investors discounting for the greater likelihood of conXicts of

interest when lending and underwriting are within the same structure’ and

conclude that a separate aYliate structure is ‘an eVective commitment mech-

anism’ to reassure customers that the underwriter will not abuse its informa-

tion advantage.

Costs of financial distress: Protecting

the group from a risky subsidiary

.........................................................................................................................................................................................

Financial distress occurs when a Wnancial institution is expected to have diYculty

in honoring its commitments. Costs of Wnancial distress include not only costs of

bankruptcy, but also the loss in value that may occur as a result of the perception

that bankruptcy may be imminent even though it may ultimately be avoided.

Talented employees may leave, suppliers may demand payment on delivery, rev-

enues from credit-risk sensitive products may decline, and conXicts of interest

between shareholders and creditors may degrade the quality of operating, invest-

ment, and Wnancial decisions. As Berger, et al. (1995: 396) note, ‘Financial distress

should be distinguished from economic distress. The cost of Wnancial distress may

be measured as the additional loss from economic distress for a leverage bank

versus an identical bank that is unleveraged. When asset quality deteriorates, both

banks will experience economic distress, but the leveraged bank experiences a

greater loss of value’.

When costs of Wnancial distress are substantial, Wrms may prefer to segregate

risky activities in separately incorporated subsidiaries even though information

is shared equally between corporate insiders and capital markets. A holding

company structure, in which subsidiaries are separately funded, can limit the

damage to the rest of the group from Wnancial distress in one of its aYliates.

Corporate separateness provides the option of partial liquidation when losses in

one of the subsidiaries would otherwise jeopardize the solvency of the rest of the

group.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 182 18.6.2009 10:55am

182 the theory of banking

Bianco and Nicodano (2002) show that both shareholders of the Wnancial

group and the rest of society are better oV when external debt is raised through

separately incorporated subsidiaries rather than through the holding company

and then downstreamed to the subsidiaries. In either case, gains from co-insurance

could be realized: the holding company may choose to rescue a faltering subsi-

diary with proWts from the rest of the group. But, if funding is primarily from the

holding company, a group-threatening loss that hits a subsidiary will certainly

inXict the costs of Wnancial distress on the rest of the group. In contrast, if

subsidiaries are separately funded in external capital markets, the loss could be

stopped at the subsidiary directly aVected, reducing the costs of Wnancial distress

to the rest of the group. Of course, the providers of debt will charge a higher-risk

premium when they lend to the subsidiary. But, as long as the premium does

not include a substantial, adverse-selection premium, both shareholders and

society should be better oV. (Of course, this depends crucially on the authors’

assumption of full information. If lenders are concerned that they are less-well-

informed about risk, then the Kahn and Winton model discussed above is more

relevant.)

It is sometimes asserted that a Wnancial group could not aVord to walk away

from a faltering subsidiary because it would undermine conWdence in the rest of

the group. For example, Baxter and Sommer (2005: 187) argue that ‘it is unlikely

that limited liability is a strong argument for complex aYliate structures . . . [I]f

limited liability aids an entity within the group, it is only at the expense of other

entities in the group’. And Walter Wriston (1981), former Chairman of the prede-

cessor of Citi, testiWed before Congress that ‘[I]t is inconceivable that any major

bank would walk away from any subsidiary of its holding company’. While it is true

that a loss of reputation may be more costly to Wnancial Wrms than to other, less-

leveraged Wrms, limited liability has option value. In some instances, banks have

walked away from insolvent subsidiaries without notable detrimental impact on

the rest of their business. For example, ING cut loose a failing insurance subsidiary

in London without substantial repercussions (Herring and Schuermann, 2005) and

Bank of Nova Scotia and Crédit Agricole abandoned insolvent subsidiaries in

Argentina (Dermine 2006).

Moreover, banks sometimes appear to isolate riskier activities in separate sub-

sidiaries. Dermine (2006) and Cerutti, Dell’Ariccia, and Martinez-Peria (2005), for

example, have observed that banks tend to prefer to organize as subsidiaries (rather

than branches) in riskier countries. Herring and Santomero (1990) reported that

some banks chose to join clearing and settlement schemes that had open-ended

loss-sharing agreements with separately capitalized subsidiaries in order to limit

potential losses. The panic that swept through Asian securities markets after the

collapse of Barings stemmed, in part, from the fear that a number of institutions

would abandon their subsidiaries if losses should exceed their capital investments

in memberships in some of the exchanges (Herring 2003). But, in other cases—for

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 183 18.6.2009 10:55am

international financial conglomerates 183

example, in dealing with troubled SIVs, Wnancial institutions have provided add-

itional funds to protect their reputations even though they were under no legal

obligation to do so.

In some jurisdictions, moreover, the limited liability option is constrained by

regulation. The Federal Reserve Board has long held that the failure of a parent

bank holding company to act as a source of strength to a troubled banking

subsidiary would be considered ‘an unsafe and unsound banking practice’ (Ash-

craft, 2004). The source-of-strength doctrine is intended to enhance the position of

the bank within a holding company. It implies that during periods of Wnancial

stress, the regulatory authorities should be permitted to use the resources of the

holding company and its subsidiaries to support the bank. In essence, the source-

of-strength doctrine would give the regulatory authorities an option on the assets

of the rest of the holding company to prevent the default of the bank. None the less,

the Fed’s attempt to enforce this doctrine in the Mcorp case was thwarted by the

courts and the Federal Deposit Insurance Corporation settled two cases where the

parent of a failed bank sued the receivership to recover funds and assets that were

downstreamed by the holding company to a faltering bank subsidiary. But, subse-

quently, Congress enacted two laws that enhanced the ability of the regulatory

authorities to force bank holding companies to act as a source of strength in some

circumstance. First, the Financial Institutions Reform, Recovery and Enforcement

Act (FIRREA) of 1989 contained a cross-guarantee provision that permitted the

FDIC to charge oV any expected losses from a failing banking subsidiary to the

capital of non-failing aYliate banks. Second, under the prompt corrective action

section of the Federal Deposit Insurance Corporation Improvement Act (FDICIA)

of 1991, the Federal Reserve Board was given authority to force a parent bank

holding company to guarantee the performance of a troubled aYliate as part of a

capital restoration plan.

Ashcraft (2004) has presented evidence that the ability of the FDIC to claim the

capital in a non-failing banking subsidiary increased the incentives of bank holding

companies to bail out a subsidiary before it fails and diminished the attractiveness

of walking away from a distressed subsidiary. He concludes (Ashcraft, 2004: 19)

that, ‘In contrast to the historical experience before FIRREA, bank holding com-

panies now appear to be a source of strength to their subsidiaries. Distressed

aYliate banks are more likely to receive injections of capital than stand-alone

banks, and recover from distress more quickly’.

In addition, Wnancial groups sometimes voluntarily choose to forego the poten-

tial advantages of limited liability by explicitly guaranteeing the external debt of

some subsidiaries, presumably to achieve more favorable borrowing terms. For

example, Citigroup (2007: 156) provides explicit guarantees for external debt of

four of its wholly owned subsidiaries: Citigroup Global Markets Holdings Inc.,

Citigroup Funding Inc., CitiFinancial Credit Company (CCC), and Associates First

Capital Corporation.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 184 18.6.2009 10:55am

184 the theory of banking

Costs of financial distress: Protecting

a subsidiary from the rest of the group

.........................................................................................................................................................................................

The growth of securitization has led to a proliferation of special purpose vehicles

(SPVs), 8 which are designed to be Wnancially insulated from the rest of

the group. An SPV is a legal entity set up by a corporate sponsor for a speciWc,

limited purpose. It buys pools of assets, usually originated by the sponsor, and

issues debt to be repaid by cash Xows from that pool of assets. It is tightly bound

by a set of contractual obligations that ensure the activities of the entity are

essentially predetermined at the inception of the vehicle. SPVs tend to be thinly

capitalized, lack independent management or employees, and have all adminis-

trative functions performed by a trustee who receives and distributes cash

according to detailed contracts. Most SPVs involved in securitization are organ-

ized as trusts, although they may also be organized as limited-liability com-

panies, limited partnerships, or corporations. For some kinds of transactions

substantial tax beneWts can be achieved if an SPV is domiciled oVshore—usually

in Bermuda, the Cayman Islands, or the British Virgin Islands (Gorton and

Souleles, 2006).

LCFIs (Bank of England, 2007b: 50) ‘have been at the heart of the growth of the

structured credit markets’ and have dominant shares in arranging residential

mortgage-backed and other asset-backed securitizations that rely heavily on

SPVs. It is evident from Table 8.2 that trusts may represent a very substantial

number of subsidiaries for each of the LCFIs. Some of these trusts are SPVs, but

most securitization vehicles are unlikely to be included in our count of majority-

owned subsidiaries because sponsors generally seek to avoid the appearance of

voting control.

SPVs are constructed to be bankruptcy remote. The objective is to reassure

investors in the SPV that their rights to the promised cash Xows will not be

compromised by Wnancial distress or insolvency in the sponsor or its aYliates.

Similarly, the SPV itself is structured so that it cannot be taken through bank-

ruptcy. Typically, any shortfall of cash that would otherwise cause an event of

default will trigger, instead, an early amortization of the pool of assets. The beneWt

of this structure is that it should avoid the deadweight costs of Wnancial distress and

so the debt issued by the SPV should not be subject to a bankruptcy premium. By

separating the control rights over assets from the Wnancing of these assets, the SPV

reduces the costs of Wnancial distress and thus the cost of debt Wnancing (Gorton

and Souleles, 2006).

8 The term ‘special purpose entity’ (SPE) is used more or less interchangeably. For an analysis of

SPVs see also Strahan (Chap. 5 in this volume). For an overview of Structured Investment Vehicles

(SIVs) see Allen and Saunders (Chap. 4 in this volume).

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 185 18.6.2009 10:55am

international financial conglomerates 185

Although the desire to avoid the deadweight costs of Wnancial distress may be the

primary motive for securitizing assets, Tufano (2006) notes that other factors may

also be important. For example, SPVs may be formed to achieve more favorable

accounting treatment for the sponsor, to increase tax eYciency, to avoid regulatory

capital requirements, to tap new pools of capital through changing the risk

characteristics of a pool of assets or to reduce the deadweight costs of information

asymmetry by separating the funding of a more transparent pool of assets from the

rest of the sponsor’s balance sheet.

Protection of the bankruptcy-remote status of SPVs requires that the sponsor

refrain from making any commitment to support the SPV. The concern is that a

legal commitment might undo the bankruptcy-remote structure. If a sponsor

Table 8.2. Breakdown by industry of subsidiaries of large complex financial institutions

LCFIs Banks Insurance companies

Mutual & pension funds/nominees/ trusts/trustees

Other financial

subsidiaries1

Non- financial

subsidiaries2

ABN AMRO Holding NV �

50 7 129 204 280 Bank of America Corporation 32 24 396 282 673 Barclays Plc 49 21 309 239 385 BNP Paribas 88 74 102 433 473 Citi 101 35 706 584 1,009 Credit Suisse Group 31 4 91 63 101 Deutsche Bank AG 54 9 458 526 907 Goldman Sachs Group, Inc. 7 4 48 151 161 HSBC Holdings Plc 85 37 246 381 485 JP Morgan Chase & Co. 38 17 229 145 375 Lehman Brothers Holdings Inc. 9 3 84 210 127 Merrill Lynch & Co. Inc. 16 9 85 89 68 Morgan Stanley 19 22 225 170 616 The Royal Bank of Scotland Group Plc

31 29 168 450 483

Société Générale 81 13 93 270 387 UBS AG 29 2 121 66 199

Total By Industry 720 310 3,490 4,263 6,729 % by industry 5% 2% 22% 27% 43%

Note: Year end 2007.

Sources: Bankscope. Majority-owned subsidiaries. For methodology see footnote AQ2for Table. 8.1. � See footnote for Table 8.1. 1 ‘Other financial subsidiaries’ include private equity subsidiaries. 2 ‘Non-financial subsidiaries’ include all companies that are neither banks nor insurance companies nor financial companies. They can be involved in manufacturing activities but also in trading activities (whole- salers, retailers, brokers, etc.). We have allocated foundations and research institutes to this category as well.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 186 18.6.2009 10:55am

186 the theory of banking

should enter bankruptcy proceeding, the judge might recharacterize the sale of

assets to the SPVas a secured Wnancing, which would bring the assets back onto the

sponsor’s balance sheet. Attempts to minimize this possibility account for a

considerable amount of the complexity of securitization vehicles. For example,

sponsors often employ a two-tiered SPV structure to provide an extra layer of

insulation between the claims of the investors and the sponsor (Gorton and

Souleles, 2006: 558).

The requirements for a true sale are set out in Financial Accounting Standard

No. 140. The sponsor must surrender control of the assets sold to the SPV and

the SPV must be a qualifying SPV (QSPV). QSPVs must be demonstrably

distinct from their sponsors, as evidenced by the fact that the sponsor cannot

unilaterally dissolve the SPV, and at least 10 per cent of the fair value of its

beneWcial interests must be held by unrelated third parties (Gorton and Sou-

leles, 2006: 556). QSPVs need not be consolidated in their sponsor’s Wnancial

statements. Some variable interest entities (SPVs that do not meet the require-

ments for QSPVs) must be consolidated. Other VIEs, in which the sponsor is

unlikely to absorb a majority of the expected losses or receive the majority of

the expected residual returns, need not be consolidated (Soroosh and Ciesielski,

2004). Thus, at best, our measure of corporate complexity is likely to capture

consolidated VIEs. For many of the LCFIs, this is a relatively small fraction of

the total securitization activity. For example, J. P. Morgan Chase and Co.

reports that in 2006 its revenue from qualifying special-purpose entities

(QSPEs) was almost Wfteen times greater than the combined revenues of its

consolidated and signiWcant unconsolidated VIEs (J. P. Morgan Chase and Co.,

2007: 59).

If SPVs are, in fact, bankruptcy-remote, would they complicate the unwinding of

an LCFI? Perhaps not, but Gorton and Souleles (2006) present evidence that

sponsors have supported their SPVs and, based on the pricing of debt issued by

SPVs and the credit rating of the sponsoring institution, conclude that investors

rely on this implicit support. Gorton and Souleles (2006) argue that this implicit

commitment is essential to deal with moral hazard and adverse selection problems

implicit in the asymmetric information between the originator of the assets and

investors in the SPV. None the less, the eVorts by several LCFIs to support their

SIVs and asset-backed commercial paper (ABCP) conduits during the turmoil in

Wnancial markets in the latter half of 2007 appear to have surprised shareholders

and some regulators. In any event, this disconnect between explicit and implicit

contracts complicates any analysis of how the existence of SPVs might aVect the

resolution of an LCFI experiencing extreme Wnancial distress. Moreover, many of

the innovative securitization structures have not been tested in a bankruptcy

proceeding. Although these bankruptcy-remote structures may well turn out to

be ‘bulletproof’, they are likely to complicate the resolution of a faltering LCFI,

none the less.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 187 18.6.2009 10:55am

international financial conglomerates 187

The legacy of mergers and acquisitions

.........................................................................................................................................................................................

Mergers and acquisitions may have a signiWcant impact on the degree of complex-

ity of corporate structure. Relative to a Wrm of equal size that has grown organic-

ally, an acquisitive Wnancial conglomerate is likely to have many more subsidiaries,

if only because it may be costly to close or consolidate them. Most of the LCFIs

have engaged in a remarkable amount of merger activity. LCFIs have engaged in a

large number of mergers, some of them exceptionally large. For example, since

1990, Bank of America, Deutsche Bank, J. P. Morgan Chase, and UBS have

implemented mergers in which the target institution was larger than 10 per cent

of the acquiring Wrm’s total assets (Thomson Securities Data Company). The

acquiring Wrm may choose to retain a considerable amount of corporate separate-

ness in the target Wrm for two reasons. First, it may perceive value in the brand and

hope to retain the reputational capital of the target Wrm. Second, the willingness to

retain the existing corporate structure may facilitate acceptance of the merger. As

Dermine (2006) notes, by committing to keep in place a local structure and staV,

local shareholders and the board of directors of the target may be reassured about

the future of the target Wrm. Also, as we discuss below, host country regulatory

authorities sometimes require that the acquiring bank maintain the target bank as a

separate, locally chartered corporation. Dermine (2006) observes, however, that

this decision to maintain a separate entity is often tactical rather than strategic.

Over time, LCFIs generally decide to build a global brand identity, which may be

inconsistent with the retention of separate subsidiaries bearing legacy names. Based

on his interviews with ING and Nordea, Dermine (2006) found that even though

both Wrms initially left many legacy organizations intact, they were also committed

to building a global brand over time.

J. P. Morgan Chase provides a good example of how mergers may increase

corporate complexity. The current organization is the result of a series of mergers

of very large banks that began in 1991 with the merger of Chemical Bank Corpor-

ation and Manufacturers Hanover Corporation. This merger resulted in a near

doubling of the size of the surviving institution, Chemical Bank, and, in 1996, was

followed by the merger of Chemical Bank with The Chase Manhattan Corpor-

ation. The resulting institution merged with J. P. Morgan and Co. Incorporated in

2000 forming J. P. Morgan Chase and Co. (JPMC). This series of mergers Wnally

culminated in July 2004 with the merger of JPMC and Bank One Corporation

(BOC). At year end 2003 JPMC had 248 wholly owned subsidiaries and BOC had

239. After the merger, at year end 2004, the surviving organization had 360 wholly

owned subsidiaries (SEC Info database). (Note that the data in Tables 8.1 and 8.2

reXect majority-owned subsidiaries and are not directly comparable.) Although

this represents nearly a 30 per cent reduction relative to the combined total

number of subsidiaries of the predecessor institutions, the result of the merger

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 188 18.6.2009 10:55am

188 the theory of banking

was, none the less, a much larger institution of considerably greater corporate

complexity.

The eVorts of JPMC to reduce its corporate complexity are consistent with

evidence presented by Klein and Saidenberg (2005) that bank holding companies

with many bank subsidiaries are valued at a discount relative to similar bank

holding companies with fewer bank subsidiaries. Although this conglomerate

discount has sometimes been attributed to ineYcient internal capital markets,

they Wnd that aYliated banks beneWt from access to internal capital markets by

lending more and holding less capital than comparable unaYliated banks. Since

activity and geographic diversiWcation is broadly similar for their sample of

aYliated and unaYliated banks, they infer that the valuation discount is attribut-

able mainly to greater complexity of organizational structure rather than diversiW-

cation (but Laeven and Levine (2007) adopt a diVerent approach and Wnd a

diversiWcation discount in Wnancial conglomerates; they identify agency problems

and insuYcient economies of scope as probable causes). This Wnding may help

explain why several large banks have attempted to simplify their corporate struc-

tures. Rosengren (2003: 111) presents evidence that from 1993 to 2002 eight large US

bank holding companies reduced their number of subsidiaries relative to the

number of subsidiaries in their predecessor organizations. Also, Citigroup (2007:

97) reported a consolidation project to merge twelve of its US-insured depository

institutions into four. These eVorts notwithstanding, continuing merger activity

undoubtedly adds to corporate complexity.

Tax frictions

.........................................................................................................................................................................................

Taxes can have a major impact on the choice of corporate structure for all Wrms,

especially international Wnancial Wrms, because they tend to have more Xexibility

to shift proWts from one entity to another. Demirgüç-Kunt and Huizinga (2001:

430) observe that ‘[M]ultinational banks, perhaps even more than other multi-

national Wrms, have opportunities for reducing their tax burdens in high-tax

countries by way of intraWrm transfer pricing’. The choice of corporate structure

may be inXuenced by income taxes (and the details of permissible deductions and

credits), capital gains taxes, taxes on interest and dividends, value-added taxes,

withholding taxes, transactions taxes and stamp duties. 9 It is diYcult to generalize

about the inXuence of taxes on corporate structure because tax codes diVer

9 Banks are often subject to a number of implicit taxes as well, which may include the obligation to

hold required reserves at the central bank at less than the market rate of interest or deposit insurance

premiums that exceed the fair value of insurance.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 189 18.6.2009 10:55am

international financial conglomerates 189

markedly across countries, even among the relatively homogeneous members of

the European Union. Moreover, the application of tax laws often depends on

complex interpretations and rulings by the tax authorities.

None the less, tax considerations appear to play a central role in a number of

choices regarding corporate structure, including the location and organizational

form of SPEs for leasing, real-estate holdings, investment management, and private

equity. In the US, speciWc tax code provisions make it advantageous to organize

real-estate mortgage conduits (REMICs), Wnancial asset securitization investment

trusts (FASITs), regulated investment companies (RICs), and real-estate invest-

ment trusts (REITs) (Gorton and Souleles, 2006: 550). In general, SPEs are struc-

tured so that proWts are not taxed in order to avoid double taxation that would

otherwise occur if both the income of the sponsor and distributions from the SPE

are taxed. Tax motives have also led to the creation of trusts for issuance of trust-

preferred securities that are taxed like debt obligations, so that interest payments

are deductible, yet are treated as Tier 1 capital by the bank regulatory authorities.

Citi alone had established nineteen of these subsidiary trusts by the end of 2006

(Citigroup, 2007: 141).

Tax considerations are especially important for internationally active Wnancial

groups. Because home countries often tax groups on their consolidated worldwide

income and, at the same time, most host countries tax locally generated income as

well, cross-border transactions are usually subject to double taxation. Without

some sort of relief, multiple taxes could stiXe cross-border transactions completely.

Governments have devised a number of ways to alleviate double taxation, such

as exempting foreign source income from the computation of taxable income or

negotiating tax treaties to reduce or eliminate withholding taxes among pairs of

countries. Some countries have also negotiated tax-sparing conventions to preserve

tax concessions granted by less-developed countries. These conventions attempt to

preserve the beneWt of host-country tax incentives (such as tax holidays, credits,

deductions, or exemptions) through tax sparing. In the absence of such tax-sparing

arrangements these incentives may be reduced or eliminated by the home country,

particularly when the home country provides recognition for taxes paid to the host

country under the credit system. Tax-sparing treaties generally grant home country

tax credit for taxes that were not actually collected by the home country. The

rationale for such arrangements is that host country tax concessions are econom-

ically equivalent to grants or subsidies. Proponents of such treaties argue that, just

as it would be inappropriate for the home country to insist on repayment by the

parent company for grants or subsidies received by its foreign subsidiaries, it is

inappropriate to recoup the value of tax incentives.

More broadly, when foreign-source income is not exempt from taxation in the

home country, Wrms are often permitted to credit foreign taxes paid against

domestic tax owed. Generally, the foreign tax credit is limited by the amount of

taxes that the Wrm would have paid if the income had been earned at home. Thus,

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 190 18.6.2009 10:55am

190 the theory of banking

Wrms have a strong incentive to reduce the average tax rate on foreign source

income by shifting proWts from relatively high-tax countries to tax havens (per-

missible foreign tax credits may be constrained in other ways as well; see Demirgüç-

Kunt and Huizinga (2001) for restrictions imposed on proWt-shifting by the US).

A crude indication of the extent to which tax issues may have contributed to the

corporate complexity of LCFIs may be seen in the number of entities located in tax

havens. Our list of tax havens is based on the forty-two countries/territories/

jurisdictions classiWed by the Financial Stability Forum as OVshore Financial

Centers (Financial Stability Forum, 2000; and International Monetary Fund,

2000). The list includes countries/territories/jurisdictions which provide low or

zero taxation, moderate or light Wnancial regulation, and/or banking secrecy and

anonymity. Of course, the impact of tax issues on organizational complexity is

much more pervasive and complex than can be represented by a count of the

number of subsidiaries in these centers. None the less, even this number is

substantial for some of the LCFIs (see Fig 8.1). Six of our LCFIs each have more

than 100 subsidiaries located in these booking centers. Moreover, three of the LCFIs

have located nearly 20 per cent of their subsidiaries in tax havens.

Regulatory constraints

.........................................................................................................................................................................................

All of the preceding rationales for corporate separateness—asymmetric informa-

tion problems, insulation against risk, the legacy of mergers and acquisitions, and

taxes—apply to large corporations in general, not just Wnancial groups. But

Wnancial groups are subject to an additional source of constraints that complicates

their corporate structures—regulation. This may help explain, at least in part, why

they have a substantially greater number of subsidiaries than non-Wnancial groups

of comparable size. On average, the sixteen LCFIs have nearly two and a half times

as many majority-owned subsidiaries as the sixteen largest non-Wnancial Wrms

ranked by market capitalization at year end 2007 (Bankscope and Osiris Corp.

data).

Banks are among the most regulated institutions in every country, although

countries diVer with regard to the constraints imposed on banks’ expansion into

other lines of business. Broadly, three diVerent regulatory models can be discerned:

(1) complete integration; (2) parent bank with non-bank operating subsidiaries;

and (3) holding company parent with bank and non-bank aYliates. 10

Universal

banking countries (see Chapter 7 in this volume) tend to follow the Wrst model,

10 Seven of the 16 LCFIs have bank holding companies (source: Bankscope). See Herring and

Santomero, 1990 for a more detailed discussion of these models and their variations.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 191 18.6.2009 10:55am

international financial conglomerates 191

with only minimal corporate separateness imposed for regulatory reasons. For

example, Germany allows the combination of bank and securities businesses in a

single legal entity. The US Comptroller of the Currency (which regulates banks, but

not bank holding companies) has long argued for the second model and has

permitted the national banks, which it supervises, to create subsidiaries to conduct

some non-bank activities. The dominant model in the US, however, is the third.

Moreover, the corporate separateness imposed on bank holding companies and

Wnancial services holding companies is reinforced by restrictions on the Xows of

credit between diVerent functional units and the bank. Sections 23A and 23B of the

Federal Reserve Act limit the amount of credit from banks to their aYliates and

require that such transactions be collateralized and made at market prices. The

Gramm-Leach-Bliley Act, which authorized Wnancial services holding companies,

extended these provisions to credit Xows between banks and their own Wnancial

subsidiaries and, to some extent, to Xows between holding companies and the

Wnancial subsidiaries of banks.

In a survey of 143 countries Barth, Caprio, and Levine (2007) Wnd that of the

majority of 127 countries that permit banks to engage in some securities activities,

Wfty-nine impose some form of corporate separateness on these activities. Of the

eighty-seven countries that permit banks to engage in the insurance business,

eighty-Wve impose some form of corporate separateness. Finally, of the sixty-two

countries that permit banks to engage in the real-estate business, forty-Wve require

some form of corporate separateness.

In countries that have not adopted the single or integrated regulator model,

diVerent functional regulators often require that the activities which they regulate

be conducted in separate legal entities. This not only facilitates oversight, but

makes it easier to ring-fence those activities should it become necessary to inter-

vene. 11 Thus, even without consideration of the complexities introduced by inter-

national expansion, Wnancial conglomerates may be required to adopt a certain

amount of corporate separateness for regulatory purposes.

LCFIs have established subsidiaries in numerous countries (see column 7 in

Table 8.1) and international expansion may require substantial additional corpor-

ate complexity for two reasons. First, host countries that apply some variation of

model three to domestic Wnancial conglomerates generally impose the same

restrictions on foreign Wrms to maintain a level playing Weld. The fact that the

US, the largest market in the world for Wnancial services, applies model three to

domestic and foreign Wrms can account for a signiWcant amount of the complexity

of the corporate structure of LCFIs headquartered outside the US.

Second, even if the host country has not adopted a variation of model three for

domestic Wrms, it may require that foreign-owned Wrms incorporate locally to

11 In some jurisdictions it is possible to ring-fence entities that are not separately incorporated;

for example, the US regulatory authorities can ring-fence a foreign branch.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 192 18.6.2009 10:55am

192 the theory of banking

ensure that the domestic authorities can intervene to protect domestic residents.

New Zealand provides, perhaps, the most extreme example of the second rationale.

More than 85 per cent of the banking system is controlled by foreign-owned banks

and the New Zealand authorities have been uncomfortable accepting the passive

role often associated with host country oversight of resident foreign branches

(Woolford and Orr, 1995). They have insisted that systemically important foreign

entities be organized as subsidiaries. Moreover, they have buttressed this corporate

separateness by additional measures that ensure that a subsidiary could continue

operation without interruption (and without its previous owners) should it be-

come necessary.

Barth, Caprio, and Levine (2007) Wnd that in their sample of 143 countries only

three countries prohibit entry by foreign subsidiaries, but twenty-eight countries

prohibit entry by foreign branches. Moreover, even if foreign branch entry is not

prohibited, host countries often impose stricter regulatory requirements on foreign

branches that make the formation of a separate subsidiary relatively attractive. For

example, of the nineteen Latin American and Central European countries surveyed

by Cerutti, Dell’Ariccia, and Martinez-Peria (2005), seven restrict foreign branches

more heavily than foreign subsidiaries.

Functional and national regulators frequently employ corporate separateness as

a means of regulating, supervising, and monitoring the part of a Wnancial con-

glomerate that falls in their bailiwick. While this may enhance local regulatory

oversight, an unintended consequence may be that international Wnancial con-

glomerates may have signiWcantly more complex corporate structures than domes-

tic Wrms of comparable size.

More broadly, LCFIs often respond to new regulations with a still more corpor-

ate complexity. Kane (1977; 1981; and 1984) has characterized this dynamic as a

regulatory dialectic, in which regulators impose a rule (or implicit tax) and the

regulated Wrms react within their constrained environment to minimize the

implicit tax burden. The regulators in turn react to perception of regulatory

avoidance with still more regulations. Robert Eisenbeis, in correspondence with

the authors, described how the regulatory dialectic evolved under the Bank

Holding Company (BHC) Act:

From the very beginning, Wnancial conglomerates exploited the BHC loopholes to expand

geographically as well as into new activities. Finance companies were acquired to expand

across state lines. Credit card special purpose banks were designed to get around usury

ceilings. SPEs and oV-balance-sheet activities were designed to avoid capital constraints.

Mortgage banking subsidiaries were established to avoid having to pay taxes for doing

business across state lines.

This kind of dynamic has undoubtedly increased the corporate complexity of

LCFIs. In the event of Wnancial distress, however, this complexity could impede

an eVective regulatory response.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 193 18.6.2009 10:55am

international financial conglomerates 193

Implications of corporate complexity

for safety and soundness of the

financial system

.........................................................................................................................................................................................

Despite their corporate complexity, LCFIs tend to be managed in an integrated

fashion along lines of business with only minimal regard for legal entities, national

borders, or functional regulatory authorities. Moreover, there are often substantial

interconnections among the separate entities within the Wnancial group. Baxter and

Sommer (2005) note that, in addition to their shared (although possibly varying)

ownership structure, the entities are likely to be linked by cross-aYliate credit

relationships, cross-aYliate business relationships, and reputational relationships.

What would happen should one of these LCFIs experience extreme Wnancial

distress? Quite apart from the diYculty of disentangling operating subsidiaries that

provide critical services to other aYliates and mapping an integrated Wrm’s activ-

ities into the entities that would need to be taken through a bankruptcy process, the

corporate complexity of such institutions would present signiWcant challenges., the

fundamental problem stems from conXicting approaches to bankruptcy across

regulators, across countries, and, sometimes, even within countries. There are likely

to be disputes over which law and which set of bankruptcy procedures should

apply. Some authorities may attempt to ring-fence the parts of an LCFI within their

reach to satisfy their regulatory objectives without necessarily taking into account

some broader objective such as the preservation of going concern value or Wnancial

stability. At a minimum, authorities will face formidable challenges in coordination

and information sharing across and among jurisdictions. Yet, experience has shown

that in times of stress information-sharing agreements are likely to fray (see

Eisenbeis and Kaufman, 2008 and Herring, 2007, for examples).

Bad news tends to be guarded as long as possible. Managers of a regulated entity

are often reluctant to share bad news with their regulators because they fear they

will lose discretion for dealing with the problem (and, indeed, may lose their jobs).

Similarly, the primary supervisor of the regulated entity is likely to be reluctant to

share bad news with other supervisory authorities out of concern that the leakage

of bad news could precipitate a liquidity crisis or that other supervisory authorities

might take action—or threaten to take action—that would constrain the primary

supervisor’s discretion for dealing with the problem or cause it to take action rather

than forbear. As Baxter, Hansen, and Sommer (2004: 79) note, ‘Once the bank’s

condition degrades, supervisors think less about monitoring and more about

protecting their creditors. This creates a conXict among supervisors’ (see Kane,

1989 for a thorough analysis of the incentives to forbear).

Generally, the primary supervisor will use its discretion to forbear so long as

there is a possibility that the regulated entity’s condition may be self-correcting,

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 194 18.6.2009 10:55am

194 the theory of banking

particularly if the alternative is closure. A closure decision is sure to be challenged

and so supervisors will tend to forbear until losses are so large that there can be no

reasonable doubt that the entity is insolvent. Losses that spill across national

borders, however, will intensify conXicts between home and host authorities and

make it diYcult to achieve a cooperative resolution of an insolvent Wnancial group

(see Eisenbeis and Kaufman, 2006 for an analysis of diVerences in resolution

policies and procedures among member countries of the European Union). Freixas

(2003) has argued that disagreements regarding the causes of losses and metrics for

allocating losses across countries would lead to the underprovision of recapitaliza-

tions of international banks even when the social beneWts of recapitalization exceed

the cost.

Within the relatively homogeneous banking sector, despite thirty years of har-

monization initiatives by the Basle Committee on Banking Supervision,

approaches to bank resolution diVer substantially across countries. For example,

countries diVer with regard to the point at which a weak bank requires resolution.

In many countries, intervention is required when a bank’s net worth (which may be

deWned in a number of diVerent ways) declines to zero, but in the US, which has

adopted a Structured Early Intervention and Resolution policy, action must be

taken when the ratio of tangible equity to total assets is equal to or less than 2 per

cent. In Switzerland, the authorities may intervene even earlier if they perceive a

threat to depositors’ interests. Countries also diVer with regard to what entity

initiates the resolution process. The supervisory authorities? The courts? Or the

bank itself? Barth, Caprio, and Levine (2007) Wnd signiWcant diVerences across the

143 countries they survey. The bank supervisor can legally declare that a bank is

insolvent in sixty-six countries. Courts have this prerogative in ninety-seven

countries and the deposit insurance agency in only in four, while in twenty-six

other countries this function is exercised by other agencies or the bank itself. In

many countries more than one entity can declare insolvency. Clearly cross-border

diVerences in regard to how and when the resolution process is initiated can cause

delays that may be costly in a crisis.

In the event that an entity is declared insolvent, which jurisdiction will oversee

the insolvency? The place where the bank was chartered? Where the management

resides? The principal place of business? The domain of the largest concentration of

assets? Or where the largest concentration of creditors resides? The collapse of

BCCI revealed that each of these questions may have a diVerent answer. Baxter,

Hansen, and Sommer (2004: 61) observe that it is diYcult to devise a good

jurisdictional rule that ‘would be both ex ante predictable (to defeat forum

shopping or subsequent jurisdictional squabbling) and sensible in application (to

discourage name-plate incorporations or prevent unseemly jurisdictional choices)’.

The choice of jurisdiction, however, may have important implications for the

outcome of the insolvency proceedings. Most countries have adopted a universal

approach to insolvency in which one jurisdiction conducts the main insolvency

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 195 18.6.2009 10:55am

international financial conglomerates 195

proceedings and makes the distribution of assets, while other jurisdictions collect

assets to be distributed in the main proceedings. But the US follows a more

territorial approach with regard to US branches of foreign banks and will conduct

its own insolvency proceedings based on local assets and liabilities. Assets are

transferred to the home country only after (and if) all local claims are satisWed.

The choice of jurisdiction will also determine a creditor’s right to set oV claims

on the insolvent bank against amounts that it owes the bank. The BCCI case

revealed striking diVerences across members of the Basle Committee (Basle

Committee, 1992). In the US, the right of set oV can be exercised only with regard

to claims denominated in the same currency at the same branch. Claims denom-

inated in diVerent currencies or at diVerent branches may not be set oV. In

contrast, in the UK, the right to set oV may be exercised even when the claims

are not denominated in the same currency, at the same branch or even at branches

in the same country. And in Luxembourg the right to set oV may not be exercised

after a liquidation order and may be exercised before a liquidation order only when

the claims are Wxed in amount, liquid, and mature.

Similarly, the ability to exercise close-out netting provisions under the Inter-

national Swap Dealers Association (ISDA) Master Contracts may vary from jurisdic-

tion to jurisdiction. In principle, in the event of a default, the non-defaulting

counterparty can close out all existing transactions under the Master Agreement,

which may include many diVerent kinds of derivative contracts with many diVerent

aYliates of the defaulting entity, making them immediately due and payable. The

non-defaulting counterparty can then oVset the amount it owes the defaulting entity

against the amount it is owed to arrive at a net amount. In eVect, close-out netting

permits the non-defaulting counterparty to jump the bankruptcy queue for all but the

net value of its claims. But the ability to apply close-out netting and the extent to

which it may be applied may depend on whether the country in which the insolvency

proceeding is conducted has enacted legislation to ensure that all outstanding trans-

actions under a master netting agreement can be terminated upon the occurrence of

an insolvency and that close-out netting will be respected by the bankruptcy trustee.

The outcome of insolvency proceedings will also depend on the powers and

obligations of the resolution authority, which may diVer from country to country.

For example, does the resolution authority have the authority to impose ‘haircuts’

on the claims of creditors without a lengthy judicial proceeding? Does the reso-

lution authority have the power (and access to the necessary resources) to provide a

capital injection? With regard to banks, is the resolution authority constrained to

choose the least costly resolution method, as in the US? 12 Or is the resolution

12 The US resolution authority can choose a resolution method that is more costly to the FDIC only

if the systemic risk exception is invoked. This requires agreement by two-thirds of the Federal Reserve

Board, two-thirds of the FDIC Board and the Secretary of the Treasury in consultation with the

President that the implementation of the least-costly resolution method would have serious adverse

eVects on economic conditions or Wnancial activity.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 196 18.6.2009 10:55am

196 the theory of banking

authority obliged to give preference to domestic depositors as the law requires in

Australia and the US?

More fundamentally, what is the objective of the supervisory intervention and

the resolution process? Is it to protect the domestic Wnancial services industry? Or

to safeguard the domestic Wnancial system? Or to protect domestic employment?

Or to protect the deposit insurance fund? Or to minimize the Wscal costs of the

insolvency to domestic taxpayers? Or to minimize the spillover costs in all coun-

tries in which the insolvent bank conducts business? Only the last of these alter-

natives is implausible. The priority that supervisors will inevitably place on

domestic objectives in the event of insolvency is the essential source of conXict

between home and host authorities.

Three asymmetries between the home and host country may create additional

problems even if procedures could be harmonized. First is asymmetry of resources.

Although international agreements among sovereigns are, necessarily, based on the

polite Wction that all sovereigns are equal, this is demonstrably not the case.

Supervisory authorities may diVer greatly in terms of human capital—the number

and quality of employees—and Wnancial resources. This means that even if the

fundamental conXicts of interest could be set aside, the home supervisory author-

ity may not be able to rely on the host supervisory authority (or vice versa) simply

because it may lack the capacity to conduct eVective oversight.

Second, asymmetries of Wnancial infrastructure may give rise to discrepancies in

the quality of supervision across countries. Weaknesses in accounting standards

and the quality of external audits may impede the eVorts of supervisors just as

informed, institutional creditors and an aggressive and responsible Wnancial press

may aid them. The legal infrastructure matters as well. IneYcient or corrupt

judicial procedures may undermine even the highest quality supervisory eVorts.

Perhaps the most important conXict, however, arises from asymmetries of

exposures: what are the consequences if the entity should fail? Perspectives may

diVer with regard to whether a speciWc entity jeopardizes Wnancial stability. This

will depend on whether the entity is systemically important in either or both

countries and whether the foreign entity is economically signiWcant within the

parent group.

A number of proposals have been advanced recently to enhance the oversight of

LCFIs and safeguard their solvency. For example, Čihák and Decressin (2007)

propose the creation of a European Banking Charter, to improve and harmonize

supervision of LCFIs with systemic cross-border exposures. Nieto and Schinasi

(2007) focus on decentralization and cooperation issues which arise from the

nature of public good of the European Union Wnancial stability. Garcia and

Nieto (2007) question the eVectiveness of decentralization and voluntary cooper-

ation in safeguarding Wnancial stability in the European Union and support the

enhancement of market discipline and the adoption of prompt-corrective-action

and least-cost resolution. Mayes, Nieto, and Wall (2007) propose a US-style

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 197 18.6.2009 10:55am

international financial conglomerates 197

prompt-corrective-action framework for preventing cross-border banking crises in

the European Union. Hüpkes (2005) advocates adoption of a functional approach

to regulation and supervision. She favors a tighter alignment between legal entities

and the functions they perform so that systemically important functions could be

more easily protected in the event of a crisis either by insulating them from

problems in the rest of the LCFI or detaching them from the LCFI. While these

proposals to enhance supervision have many attractive features, none can be relied

upon to prevent insolvencies. Thus it is also important to consider ways to improve

the resolution of insolvent institutions.

Concluding comments

.........................................................................................................................................................................................

The corporate complexity of LCFIs is likely to defy eYcient resolution in the event

of bankruptcy. It seems doubtful that going-concern value could be protected

adequately and, worse still, the unwind is likely to spill-over to damage other

institutions and market participants if counterparties attempt to liquidate posi-

tions at once, driving down prices and causing problems for other investors with

similar positions. In the absence of workable procedures to unwind the aVairs of a

failing LCFI in an orderly manner, the result is likely to be a chaotic scramble for

assets that could infect other markets and institutions, with potential disruption of

the real economy.

Despite ex ante protestations to the contrary, the authorities are unlikely to risk

such an outcome and so the result is likely to be a bailout that will prop up the

failing group. The continuation of recent trends toward globalization, conglomer-

ation, consolidation, and increasing reliance on trading of over-the-counter (OTC)

derivatives implies that we may be confronted with a growing category of Wrms that

are ‘too complex to fail’. This, of course, has ominous implications for moral

hazard. A market perception that such Wrms will beneWt from oYcial support in

times of stress gives them a competitive advantage completely unrelated to their

ability to add value to the Wnancial system. It dulls the incentives for creditors to

demand disclosure of risky positions and monitor such exposures. Weakened

market discipline will enable such institutions to take larger, riskier positions

without paying appropriately higher-risk premiums to their creditors. The result

may be larger potential insolvencies that require still larger bailouts to forestall

systemic risk.

For market discipline to operate eVectively in constraining risk taking by LCFIs,

the regulatory authorities need a credible procedure to unwind the aVairs of an

LCFI in an orderly manner, without systemic spillovers. SimpliWcation of the

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 198 18.6.2009 10:55am

198 the theory of banking

corporate structure of large complex Wnancial institutions would be a good place to

start. Since regulatory and tax policies have contributed signiWcantly to the prob-

lem, they need to be part of the solution.

Postscript on the bankruptcy of Lehman Brothers

The preceding was written in 2007 before the actual collapse of an LCFI. The

editors asked us to reXect on what we had learned from the costly experiences of

2008 about the implications of corporate complexity for systemic risk. The Wrst

observation to be made is that the list of sixteen LCFIs was not nearly long enough

to reXect the perceptions of regulators when they were confronted with the

prospect of collapse of institutions not on the LCFI list. Bear Stearns may be the

most obvious case. Although Bear Stearns was one of the Wve largest investment

banks in the US, it was less than half the size of the fourth-largest investment bank,

Lehman Brothers (LB). None the less, when Bear Stearns was about to collapse, the

Table 8.3. Corporate structure of lehman brothers

Country majority-owned subsidiaries Lehman brothers holdings Inc

USA 238 United Kingdom 120 Cayman Islands 18 Australia 9 Luxembourg 6 Ireland 5 Netherlands 5 Bermuda 4 France 4 Hong Kong 4 Japan 4 Korea (Republic of) 4 Germany 3 Singapore 2 Thailand 2 Argentina 1 Canada 1 Switzerland 1 India 1 Mauritius 1 Total 433 Number of countries 20

Note: Year end 2007.

Source: Bankscope. Majority-owned subsidiaries.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 199 18.6.2009 10:55am

international financial conglomerates 199

US regulatory authorities subsidized a merger of Bear Stearns with J. P. Morgan

Chase out of concern for the ‘interconnectedness’ of Bear Stearns with the rest of

the Wnancial system. The enormous subsidy to AIG is another case in which

intervention was justiWed on similar grounds. Other countries took similar meas-

ures to support other institutions that were not large enough or complex enough to

make the oYcial list.

There was one signiWcant exception, however, to the general trend of hastily

improvised bailouts. After trying to broker a merger of LB with other, stronger

institutions, the US authorities declined to bail it out and sent the holding

company, Lehman Brothers Holdings International (LBHI) to the bankruptcy

courts for protection under Chapter 11 of the US bankruptcy code, the largest

bankruptcy in US history. Although LB was by far the smallest and one of the least

complex institutions on the list of LCFIs, it was none the less of suYcient systemic

importance that its collapse led to substantial spillovers on global capital markets.

Credit risk spreads rose to record highs, equity prices fell by 4 per cent worldwide

when the bankruptcy was announced and government bond yields declined

sharply as foreign exchange carry trades were unwound.

Lehman’s total reported assets were roughly $700 billion. Table 8.3 shows its

corporate structure at the end of 2007. It included 433 subsidiaries in twenty

countries. This corporate complexity greatly impeded the orderly resolution of

the Wrm, created signiWcant spillovers to other institutions and markets, and led the

Group of 7 Wnance ministers to pledge (Guha, 2008) ‘to do everything in their

power to prevent any more Lehman Brothers-style failures of systemically import-

ant Wnancial institutions’. 13

Understandably, after the US government had subsidized the merger of Bear

Stearns, a much smaller, less-complex investment bank, the market expected that

Lehman Brothers would receive similar treatment. Why then was LB permitted to

fail? The Fed and the Treasury claimed they lacked authority to bail it out. It is also

likely that they wished to limit moral hazard by engaging in a bit of ‘constructive

ambiguity’, a dubious remedy at a time when a consistent policy framework might

have helped stabilize expectations. Moreover, since they had a team of examiners in

LB ever since the collapse of Bear Stearns, they knew much more about the

condition of LB and may have believed they could predict and control the spillover

costs. They may have thought that counterparties and creditors had suYcient

warning about LB’s weakening condition to take precautionary measures. But, of

course, in a complex and integrated Wnancial system, regulatory action or inaction

can have unintended consequence through indirect exposures and linkages that are

apparent only after the fact.

13 Observers said that it came close to a Group of seven-wide temporary implicit guarantee for

many or all of the liabilities of systemically important Wnancial Wrms. See Guha, 2008.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 200 18.6.2009 10:55am

200 the theory of banking

One of the major concerns was that LB was the sixth largest counterparty in

OTC derivatives markets. But, back oYces succeeded in processing billions of

dollars of contracts and the International Swap Dealers Association organized an

auction to determine settlement prices. Because derivatives contracts in which LB

was a counterparty were usually marked to market daily and collateral was adjusted

each evening to reXect changes in market prices, losses were relatively light. Losses

were much greater, however, with regard to credit-default swap contracts written

on LB. Those selling protection on LB are in a similar position to bondholders and

received a similar price. Buyers of $100 of default protection will receive $91.375, a

substantial loss for sellers of protection.

A second major concern was LB’s key role in the Repo market, which totals

roughly $11 trillion and is the short-term, collateralized lending market that banks,

broker/dealers, and hedge funds use to Wnance securities positions. The Fed

attempted to address the risk that the market would seize up by allowing broader

use of the Primary Dealer Credit Facility through expanding the list of eligible

securities. In addition, a group of global banks announced plans to use their own

capital to establish a $70 billion private-sector credit facility for those securities not

eligible for the Fed facility. The Fed also announced an increase in its Treasury

Securities Lending Facility to $200 billion.

What turned out to be more disruptive were the traditional exposures to LB’s

outstanding debt. Among the largest unsecured creditors were the US federal

government’s Pension BeneWt Guaranty Corp., the German government’s deposit-

insurance arm (McCracken, 2008), and money-market mutual funds. The last

proved to be one of the most important channels of contagion. One of the oldest

money-market funds, the Reserve Primary fund, was forced to write oV $785 million

of short and medium-term notes and became the Wrst money-market mutual fund

to ‘break the buck’ in fourteen years. This triggered $184 billion in money-market

mutual fund redemptions and forced fund managers to sell assets into illiquid

markets. This spilled over into commercial paper markets including not only

asset-backed commercial paper, but also non-asset backed commercial paper that

had held up reasonably well and was a key means of Wnancing corporations and

banks. The interbank market seized up entirely with the almost complete collapse of

conWdence in counterparties in money markets. Spreads between the euro-dollar

interbank rate and the comparable US Treasury rate rose to nearly 450 basis points,

more than double the already high spreads that prevailed before the LB bankruptcy.

In addition, failed trades proved particularly disruptive. Prior to LB’s bank-

ruptcy, portfolio managers placed thousands of trades with LB’s broker dealer LBI,

many of which were subsequently transferred for settlement to LBI aYliates

throughout the world. After the bankruptcy, these failed to settle and this has led

to civil proceedings on three continents. The UK administrator said that about

43,000 trading deals were still ‘live’ in the London subsidiaries alone and would

need to be negotiated with each counterparty (Hughes, 2008).

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 201 18.6.2009 10:55am

international financial conglomerates 201

But, the fundamental problem was that LB was managed as an integrated entity

with minimal regard for the legal entities that would need to be taken through the

bankruptcy process. LBHI issued the vast majority of unsecured debt and invested

the funds in most of its regulated and unregulated subsidiaries. This is a common

approach to managing a global corporation, designed to facilitate control over

global operations, while reducing funding, capital, and tax costs. LBHI, in eVect,

served as banker for its aYliates, running a zero-balance cash-management system.

LBHI lent to its operating subsidiaries at the beginning of each day and then swept

the cash back to LBHI at the end of each day. The bankruptcy petition was Wled

before most of the subsidiaries had been funded on 15 September and so most of

the cash was tied up in court proceedings in the US.

Lehman also centralized its information technology so that data for diVerent

products and diVerent subsidiaries were co-mingled. This was an eYcient way of

running the business as a going concern, but presents an enormous challenge in

global bankruptcy proceedings. LB stored data in 26,666 servers, 20,000 of which

contained accumulated emails, Wles, voicemail messages, instant messages, and

recorded calls. The largest data centers were in New York, London, Tokyo, Hong

Kong, and Mumbai. Moreover, LB used approximately 2,700 proprietary, third-

party, and oV-the-shelf programs, each of which interacted with or created trans-

actions data. The bankruptcy administrators must preserve, extract, store, and

analyze data relevant to the entities they are dealing with. This problem was made

more diYcult by the success of the administrators of LBHI in selling two important

entities that were rapidly declining in value because of loss of human capital: its

investment-banking operations and its asset-management business.

Most of the US investment-banking operations—the assets, not the legal

entities—were sold to Barclays. This necessitated bringing a Securities Investor

Protection Corporation (SIPC) proceeding, which put all LBI accounts under the

control of the SIPC Trustee and permitted the broker-dealer to be liquidated.

Nomura bought most of the investment-banking business in Asia and continental

Europe and LB’s asset-management business was sold in a management buyout.

But this meant that the data were owned by Barclays, Nomura, and the now

independent asset-management division and so bankruptcy administrators are

dependent on the new owners for access to data to determine the assets and

liabilities of each legal entity. The administrator of the four London subsidiaries

complained that nine weeks after the bankruptcy, he has yet to receive a conWrma-

tion of the assets owned by these subsidiaries.

The US administrators expressed the optimistic view that they would be able to

complete the resolution within eighteen to twentyu-four months, but the presiding

judge reminded the administrator that the biggest impediments to a timely com-

pletion of the administration are the timetables of the other insolvency Wduciaries

around the world. The administrators in London warned that it may take years for

creditors to get their money back, noting that they were continuing to work on

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 202 18.6.2009 10:55am

202 the theory of banking

Enron, which failed seven years ago, which was about one-tenth the size and

complexity of Lehman (Hughes, 2008).

The conclusion we draw from the LB experiment is not that all systemically

important institutions should be bailed out, but rather than regulators and super-

visors should focus on devising orderly resolution plans that will enable them to

unwind even the largest, most-complex institution with minimal spillover to the

rest of the Wnancial system. A useful Wrst step would be to require that each

institution create and maintain a plan for winding down the institution just as

they now maintain plans for business continuity. The bankruptcy administrator of

LBIH has claimed that the hastily prepared bankruptcy Wling has cost as much as

$75 billion in lost value (McCracken, 2008). If the regulators deem the plan

unworkable, the institution may be required to reduce its complexity or set aside

a higher capital charge. An institution that is ‘too complex to fail’ is simply too

complex and presents too great a threat to the rest of the Wnancial system.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 203 18.6.2009 10:55am

international financial conglomerates 203

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 204 18.6.2009 10:55am

Author Queries

[AQ1] Kindly confirm the citations for Tables 1–3. We have deleted the citations for Table and

Changed figure citations to Table. Please check.

[AQ2] Please confirm the changes made here. Fig 8.1 has been changed Table 8.1.

Berger et al / The Oxford Handbook of Banking 08-Berger-C08 Page Proof page 205 18.6.2009 10:55am