Chen 2009

International Review of Finance, 9:3, 2009: pp.
177–209
DOI: 10.1111/j.1468-2443.2009.01089.x
Banking Reforms for the 21st

Century: A Perfectly Stable Banking
System Based on Financial
Innovationsn
NAI-FU CHEN
University of California, Irvine, Irvine, CA, USA
ABSTRACT
Although bank loans themselves are somewhat illiquid because of private

information, most of their cashflows are not. Recent financial innovations
allow commercial loans to be liquefied via credit derivatives and actual and
synthetic securitizations. The loan originating bank holds the remaining
illiquid equity tranche containing the concentrated credit risk, private
information rent and the ‘excess spread’ that incentivize the bank to
continue to monitor and service the loans. Empirically, we find that the
average size of the equity tranche is about 3% for the representative
commercial loan portfolios in our sample. The liquefaction of bank loans
makes possible a banking system that restricts the guaranteed accounts to be
backed by 100% reserves and the non-guaranteed deposits to be backed by
liquid securitized loan tranches, while retaining the deposit-lending synergy.
Such a system is perfectly safe without deposit insurance and it renders banks
bankruptcy-remote without sacrificing a bank’s traditional role as a financial
intermediary.
I. INTRODUCTION
The fractional reserve banking system prevalent in most countries certainly has
not been a conspicuous success. Over the past 25 years, we have witnessed
numerous banking crises in industrialized countries as well as developing
n
This paper started as joint work with Merton Miller based on our meeting with the Hong
Kong Monetary Authority. This version of the paper was completed after Merton Miller passed
away and the current authorship reflects the wishes of his estate. We thank Jeremy Berkowitz,
Bruce Brittain, Yuk-shee Chan, Ellen Chen, Stijn Claessens, Doug Diamond, Greg Duffee, Eugene
Fama, Jie Gan, Vidhan Goyal, Diane Lam, Marie Lam, Steve Lang, Andrew Lui, Elaine Ng, James
O’Brien, David Stanley, Neal Stoughton, Juichi Takeuchi, Lee Thomas, Powell Thurston, Irene
Tsao, Ram Willner, Patrick Wright, Michael Ye and seminar participants at the Bank of Finland,
HEC, INSEAD, Swedish School of Economics and Business Administration, Hong Kong University
of Science and Technology and UC Irvine for helpful comments, Moody’s and Standard and
Poor’s for generously sharing their data bases, and Daniela Balkanska and Gaiyan Zhang for
research assistance.
r 2009 The Author. Journal compilation r International Review of Finance Ltd. 2009. Published by Blackwell
Publishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
International Review of Finance
countries, from Northern Europe to Latin America to Southeast Asia.1 None of

these, however, were as spectacular as those in the world’s largest two economies
– the United States and Japan. The banking crises of the 1980s and early 1990s in
the United States ultimately cost about US$200 billion to clean up. The direct
cost to clean up the Japanese banking crises easily exceeded one trillion dollars.
The indirect cost of dragging the Japanese economy down for more than a decade
is likely to be many times that. Just as the Japanese economy is finally emerging
from the ‘lost decade,’ many major US and European banks are facing a fresh
financial crisis in 2008 from their investments in risky loans. By the autumn of
2008, the banking systems in the United States and Western Europe have
essentially collapsed. All major banks in the United States and many major banks
in the United Kingdom, Germany, Switzerland, Spain, Belgium, the Netherlands,
Iceland and other parts of Europe are partially or wholly nationalized.
At the root of many of the banking system problems is the common practice
for banks to take ‘riskless’ liquid deposits and turn around and invest the money
in risky illiquid loans. Bank loans are illiquid in part because banks are generally
thought to have private information about the borrower that the market does
not have. Banks play an important role in providing private (versus public)
sources of funding to borrowers who cannot credibly convey or are not willing
to divulge inside information to the public.2 This, unfortunately, induces a
fundamental mismatch of risk and liquidity in banks if riskless liquid bank
deposits are backed by risky illiquid bank loans.
Banking authorities have been trying to solve this problem with a convoluted
system of deposit insurance, regulations, monitoring, surveillance, capital
requirements and bailouts. Despite the heavy intrusions of government into the
private investment process in regulating banks, banking crises continue to
erupt. The moral hazards posed by deposit insurance and anticipated
government bailouts, combined with the high leverage ratios in banking, make
such a banking system highly fragile and disaster prone. Perhaps, it is time to
take a fresh look at this old problem!
During the banking crisis years in the 1930s, there was a proposal by Irving
Fisher and others to separate a bank’s deposit liabilities into two distinct
components: the transaction accounts would be backed by 100% reserves and
the non-guaranteed savings would be backed by risky loans. In such a system,
there is no need for deposit insurance (except for fraud) and there is no need for
the government to intrude into the private loan and investment process. While
this proposal was favored by some bankers to deal with the banking crises then,
many other bankers preferred a system of deposit insurance with government
1 See, e.g., Kaminsky and Reinhart (1999) for a list of countries that experienced banking crises in
recent years. The 2008 crisis is still ongoing.
2 Banks play a special role as a financial intermediary. See, e.g., Kane and Malkiel (1965),
Diamond and Dybvig (1983), Diamond (1984), Santomero (1984), Fama (1985), James (1987),
Calomiris and Kahn (1991), Freixas and Rochet (1997), Boot (2000), James and Smith (2000),
Diamond and Rajan (2001), Kashyap, Rajan and Stein (2002), Dahiya, Puri and Saunders (2003)
and the review paper by Ongena and Smith (2000).
r 2009 The Author

178 Journal compilation r International Review of Finance Ltd. 2009
Banking Reforms for the 21st Century
regulations and bailouts (Calomiris 2000). Fisher’s proposal was ultimately

defeated in the US Congress as Amendments to H.R. 5357 (see Fisher 1936).
Critics of Fisher’s proposal would argue that the illiquid bank loans that
support the non-guaranteed deposits would also make the deposits illiquid. If
few depositors were willing to hold illiquid deposits, banks would lose their role
as a private source of financing. Such an equilibrium could well be less
economically efficient than the fractional reserves system (with deposit
insurance and government interventions) that preserves the liquidity of most
deposits and preserves the traditional role of banks as financial intermediaries.
The main purpose of this paper is to show that recent financial innovations
provide a simple solution to this dilemma. The phenomenal growth of the
credit default swap (CDS) market over the past few years allows highly efficient
transfer of credit risks between financial institutions, institutional investors and
intermediate market instruments. Furthermore, when credit risks are packaged
via actual and synthetic securitizations, the vast majority of the value of a bank
loan can be financed with liquid market instruments, and the loan originating
and monitoring bank would hold the remaining illiquid residual tranche as
incentives. Thus, recent financial engineering can slice out the illiquid portion
containing the private information of a bank loan for the loan-originating bank
to hold, and securitize the liquid portion of a bank loan for the institutional
investors and non-guaranteed depositors.
Empirically, we find that the illiquid portion in representative loan pools is
small and the typical level of capital in the banking system is sufficient to support
this illiquid portion. The recent market innovation of using ‘excess spread’ as a
dynamic incentive mechanism further helps align the interests of the bankers
with the institutional investors and non-guaranteed depositors. Consequently,
banks can retain their traditional role in providing private sources of financing
and fund the loans with liquid non-guaranteed deposits, all without government
interventions and free of deposit insurance. At the general equilibrium level, if it
is our national policy for the government to intervene in the relative supply of
riskfree to risky investments in the economy to smooth out credit shocks, this
can be accomplished with less distortion under a system with 100% reserves than
under a system with fractional reserves and deposit insurance.
An important lesson we learn from the 2008 crisis is that exotic financial
instruments may become difficult to evaluate under crisis conditions. The 2008
financial and credit crises arise because of excessive risk taking by banks in
search of higher profits. When the values of those risky loans fall precipitously,
the monetary authorities have to step in to bail the banks out. But the
complexity of the financial instruments involved has made the valuation and
the bailout more difficult in many cases. Therefore, in this paper, our central
idea is based on those plain vanilla financial instruments and avoids the more
exotic variety.
A banking reform scheme of this scale invariably has many interesting minor
issues that cannot be addressed in a single paper. The main objective of this
paper is to set up the overall framework of such a banking system. In Section II,
r 2009 The Author

Journal compilation r International Review of Finance Ltd. 2009 179
we briefly review the recent banking crises and the history of the 100% reserve
idea. Section III examines the feasibility of a variant of the 100% reserve idea
taking into account of the latest innovations in market technology and credit
derivatives. We illustrate how securitization can slice out the liquid portion of a
loan’s cashflows and transform them into marketable securities, and save the
illiquid economic core to be retained by the originating bank as the ‘skin in the
game.’ In Section IV, we use the databases provided to us by Moody’s and
Standard and Poor’s before the recent easy credit period to estimate how large is
this illiquid portion of representative bank loan pools that has to be supported
by bank capital. Section V considers the sufficiency of the market in linking
liquid bank loan tranches to non-guaranteed deposits and the supply of
government securities for the guaranteed deposits. Section VI discusses the
desirability of the resultant banking system over the prevalent fractional
reserves banking system. Section VII concludes the paper.
II. THE RECENT BANKING CRISES AND THE 100% RESERVE PLAN
Perhaps the best way to appreciate the problems inherent in the current
banking system is through the lessons of the recent banking crises in the largest
two economies in the world: US and Japan. Both countries run a fractional
reserve system supported by deposit insurance laden with moral hazard.
The early 1980s marked the beginning of a new banking era for banks in the
world. Financial innovations, deregulations and international competition
pushed many US banks away from their once sheltered market (including
deposit interest rate ceilings) into risky portfolios. Large banks assumed greater
risk to boost profits, but many of them failed. At the same time, the S&Ls were
hit hard because of the mismatch in interest rates. The banking crisis period in
the United States did not end until the mid 1990s when the US economy was
several years into its longest expansion period.
The S&L debacle ultimately cost about US$160 billion, of which an estimated
US$132 billion was borne by the taxpayers. The cost of Federal Deposit
Insurance Corporation (FDIC) failed-bank resolutions during 1980–1994
was US$36.3 billion (the mutual savings bank crisis, Continental Illinois,
Texas bank crisis, Northeast bank crisis and others).3 With the outbreak of the
3 The historical facts for this section are collected mainly from a book published by FDIC called
‘The banking crises of the 1980s and early 1990s,’ Japan’s Financial Supervisory Authority, and
financial press like the Wall Street Journal and The Economist. There were 2978 bank and S&L
failures between 1978 and 1999 in the United States (source: FDIC). In 1983, it was estimated
that it would take FSLIC US$25 billion to close those insolvent S&Ls but FSLIC had only US$6.3
billion. Thus, FSLIC granted forbearance and encouraged those that were insolvent to continue
operations and hoped that they would grow out of insolvency. The huge moral hazard problem
(with already insolvent institutions) coupled with less than spectacular business environment
for the S&Ls doomed such forbearance and cost taxpayers US$132 billion, an average of about
US$500 per man, woman and child in America.
r 2009 The Author

less-developed-country (LDC) loan crisis in 1983, the problem for the big
money-centered banks posed even greater systemic risks for the survival of the
US banking system, as ‘seven or eight of the largest 10 banks in the
United States might have been deemed insolvent (FDIC)’ if their loans had
to be marked to market without regulatory forbearance. The FDIC review
(footnote 3) pointed out that ‘the seven-year period (1983–89) . . . was devoted
to . . . protect the solvency of the US financial system.’ Fortunately, unlike
the S&Ls, the money-centered banks were able to take advantage of the
healthy world economic growth from 1982 to 1990 to slowly build up their
reserves until the late 1980s to absorb the loss [see ‘‘The banks’ great escape’’ in
The Economist (February 1989)], but not without the additional help from the
World Bank, the government of Japan, the IMF (taxpayers of the member
countries) under the Brady plan and the economic recoveries of the LDC
themselves.
Just as the banking crisis started to fade in the United States in the 1990s, the
economy of Japan entered into a period of slow growth and the banking crisis in
Japan began. Moral hazard in banking was clearly a major factor in the real
estate bubble in Japan during the 1980s when most banks speculated directly
and indirectly (via subsidiaries and other ‘related’ companies) in fueling the real
estate price boom. When it burst, the banks were saddled with a bad loan
problem so huge that it would impair their ability to finance even normal
operations of viable firms. Forbearance was also being practiced in Japan in
dealing with its banking crisis with the help of the Japanese Resolution Trust4
since 1992.
Now, just as Japan is finally growing out of its banking problems, many of
the US and European banks are facing new crises in 2008 from excessive
speculation on risky loans, in particular subprime mortgages. The final cost of
cleaning up the current worldwide systemic banking failure is expected to be
multitrillion.
The current convoluted banking system relies on regulators to monitor the
banks in regard to the riskiness of their investment and loan portfolios. But, as
the FDIC review (see footnote 3) suggests, it is difficult for bank regulators to
restrict banks from those risky investments while they are so profitable, whether
they were LDC loans in the 1980s or the subprime mortgages two decades later.
Perhaps all these crises could have been avoided if banks keep ‘100% reserves’
to back ‘riskless’ deposits and package risky loans to back the non-guarantee
deposits. The risky loans are subject to direct market discipline because
disenchanted investors and depositors can simply pull the plug by not
participating if they do not like the overall risk. This is the basis of Fisher’s
(1936) plan. Fisher ascribed the original 100% reserve idea (to reform the
fractional banking system) to the ‘Chicago Plan of Banking Reform’ that first
appeared in a University of Chicago memorandum by Henry Simons, Aaron
4 Kyodo Saiken Kaitori Kiko, or Cooperative Credit Purchasing Company Ltd., established August
28, 1992.
r 2009 The Author

Bank
A L
Deposits with the Fed Transaction accounts

(access to payment system)
Loans Non-guaranteed deposits
Working capital
Other loans Other debt and equity

Other assets
Figure 1 Irving Fisher’s Proposal.
Directors, Frank Knight, Lloyd Mints, Henry Schultz and others (see, e.g., Hart
1935; Fisher 1936; Simons 1948; Mints 1950).5 Over the past half century, it has
been revised and redefined to address the periodic crises ranging from the
banking disasters in the United States and Japan to the international financial
turmoil that started in the summer of 1997 (see, e.g., Friedman 1960; Tobin
1986; Litan 1987; Miller 1995, 1998; Chen 2001). The following is a brief
outline of the main ideas on the 100% reserves system.
For a typical 100% reserve bank, its balance sheet would look like the one in
Figure 1. In a nutshell, the plan calls for all ‘insured’ transaction accounts to be
backed by deposits with the monetary authority (at interest rates reflecting the
cost of access to the payment system).6 In that sense, banks are required to have
100% reserves. This itself will guarantee the safety of the payment system and
there would be no need for any deposit insurance except for fraud and it would
not be necessary to place any limit on such deposit insurance.7
5 The idea behind the 100% reserve banking has a long history and it is discussed in detail in a
book by Fisher (1936). A similar idea went way back to Ricardo and The Bank Act of 1844
requiring 100% reserves for all Bank of England notes, putting an end to the era commonly
known as the wild cat banking in England. Since then, fractional reserve backing of bank notes
was replaced by fractional backing of bank deposits, which aggravated the economic depression
and precipitated the general banking failure in the 1930s. In the words of Fisher who had to live
through it, the fractional reserve banking system almost wrecked the capitalistic civilization.
History repeats itself again in 2008.
6 See, e.g., Friedman (1960) and Litan (1987) for different variations of the same general concept.
7 Thus, there is no need for regulations beyond what is normal for other types of financial
institutions. Banks may still want to have additional reserves or cash equivalent for their
working capital to initiate loans and for the purposes of clearing and carrying out normal
banking transactions.
r 2009 The Author

Bank
A L
Interest bearing deposits Transaction accounts

(access to payment system)
Safe with the Fed Guaranteed
Assets Accounts
T-bills and their close
Riskfree deposits
equivalents
Tranches of risk class A Deposits of risk class A
Risky Tranches of risk class B Deposits of risk class B Non-

Loans “ “ “ “ “ “ “ “ “ “ “ “ “ “ guaranteed
Accounts
Tranches of risk class K Deposits of risk class K
Working capital
Illiquid residuals and
Other debt and equity
other loans and tranches
Other assets
Figure 2 100% Reserve Banking without Deposit Insurance.
Banks will raise funds to support their loan portfolios by issuing non-
guaranteed securities at rates reflecting each bank’s risk characteristics. A bank
would typically initiate the loan process using its working capital. After the loan
portfolio is created, the bank can keep all or part of it on its book and repackage
the remainder for the non-guaranteed depositors. Thus the items on the
balance sheet below the transaction accounts would resemble those of a typical
investment trust or brokerage firm and could be structured in accordance with
current market demand.
A concern about this system is that the illiquid nature of private loans would
also render non-guaranteed deposits illiquid. If depositors prefer liquid deposits,
would such a system destroy the banking system as we now know it? Could our
current convoluted banking system, despite its periodic catastrophic collapses,
be welfare improving (over a 100% reserve banking system) because illiquid
loans can support liquid deposits under the protection of deposit insurance and
government bailouts? It is impossible to have a conclusive answer to this
question without a large-scale social experiment. Fortunately, with the recent
advances in financial engineering, there is a simple but elegant solution that
gives us the best of both worlds. The key lies in the recent innovations in the
market that allow banks to liquefy consumer, commercial and industrial loans
alike into marketable securities.
Figure 2 illustrates the typical structure of a 100% reserve bank after its loans are
liquefied. Transaction accounts with access to the payment system are 100%
backed by interest bearing deposits with the monetary authority, where the lower
interest rates reflect the cost of access to the payment system. Riskfree savings
deposits are 100% backed by short-term T-bills or their close equivalent.
r 2009 The Author

A government deposit insurance, whose cost is borne by account holders, may

still be desirable as insurance against fraud. Risky private loans are liquefied into
tranches of different risk and liquidity classes and an illiquid residual. These liquid
tranches support the non-guaranteed deposits with the risk–reward–liquidity
characteristics as determined by the market supply-demand. The illiquid residual is
held by the bank against its capital. This is to ensure that the loan originating and
monitoring bank has enough ‘skin in the game’ to align its interests with the
investors of the liquid tranches.8 With such a suitably revised 100% reserve
banking system, a bank can provide perfectly safe transaction accounts with
access to the payment system, guaranteed riskfree deposits backed by government
papers, private sources of financing and liquid non-guaranteed deposits, all
without deposit insurance (except for fraud) and government intrusions into the
private loan and investment process. The reality of such a banking system
depends on the ability of the market to liquefy most bank loans.
III. FEASIBILITY OF A 100% RESERVE BANKING SCHEME WITH

LIQUID NON-GUARANTEED DEPOSITS
There are three key financial innovations in recent years that combine to make
most bank loans liquid. The first one is the growth of the bank loan market. The
second one is the growth of credit derivatives, in particular CDS. The third one
is the growth of collateralized debt obligations (CDO), which include both
collateralized bond obligations (CBO) and collateralized loan obligations (CLO).
It is well known that these financial innovations are very effective
instruments for risk-sharing, but they do not change the risk nature of the
underlying asset. In 2008, we discover painfully that the risk of the US subprime
loans is spread perhaps too efficiently to any bank in pursuit of higher profits.
But, it is worth emphasizing that it is not the financial instruments that create
the underlying risks. The financial instruments are there to spread the risk
efficiently. In this section, we will look at how properly structured financial
innovations (dictated by the market before the easy credit period of 2002–2005)
would discourage excessive risk taking and incentivize banks to act in the
interests of all parties. By forcing the banks to be a meaningful residual claimant
throughout the life of the loans, banks have enough ‘skin in the game’ to be
diligent in their loan originating and monitoring roles.9
Table 1 shows a breakdown of the aggregate loan portfolio of all the FDIC-
insured banks in the United States at the end of 2007. Of the approximate
US$6.6 trillion in loans, the largest three categories are real estate loans (55%),
8 The lack of ‘skin in the game’ for the risky mortgage originators is often blamed as the root of
the 2008 financial crisis.
9 We are indebted to many bankers, asset managers and rating agency analysts of structured
products for information and insightful comments, in particular Diane Lam (S&P), Marie Lam
(Moody’s), Powell Thurston (PIMCO), Irene Tsao (Societe Generale), Ram Willner (Banc of
America) and Patrick Wright (Deutsche Bank).
r 2009 The Author

Table 1 US commercial bank (FDIC) loan portfolio

End of 2006 End of 2007
(in millions) (in millions)
Total 5,981,184 6,626,157

Real estate loans 3,432,254 3,674,508
Loans to depository institutions 128,314 136,181
Commercial and industrial loans 1,139,853 1,369,934
Consumer loans 857,879 959,579
Leases 136,878 128,734
Others 286,006 357,221
Notional amount of derivatives 132,168,588 164,767,281

Credit derivatives 9,019,653 14,412,090
commercial and industrial loans (21%) and consumer loans (14%). Real estate
loans and consumer loans (credit card, auto loans and leases, home equity lines,
student loans, etc.) have been securitized and resold in the secondary market
since the early 1980s. They are easy to liquefy because their loan process usually
follows a standardized procedure with information related to income, wealth,
FICO and other credit measures10 and very little other private information. A
diversified portfolio of these loans has essentially only credit risk that is a
function of the average credit rating, geographic characteristics, economic
conditions and interest rates. For large and small banks alike, after they
originate these loans, they can choose to sell them off easily.
In this study, we focus on the commercial and industrial loans that are
considered to be less liquid because they tend to be more chunky and with
private information. These corporate loans are also the focus of many
theoretical models and empirical research on banking (see review article by
Ongena and Smith 2000). Recent financial innovations, however, allow the
market to slice out the most illiquid portion of the loans to be retained by the
loan originating agent bank and liquefy the rest. The contract design leaves
sufficient incentives for the agent bank to continue to service and monitor the
loans on behalf of the investors. In many aspects, this theory is also well
understood in practice11 and it is similar to the standard corporate finance
paradigm of a manager acting on behalf of the shareholders (see, e.g., Gorton
and Pennacchi 1995; DeMarzo and Duffie 1999). The interesting empirical
10 Failure to follow this time-honored procedure (since the inception of mortgage-backed

securities in the early 1980s) led to the subprime crisis that ultimately brought down the
banking systems in the United States and Europe in 2008. The moral hazard to pursue higher
profits with increasingly risky loans during the easy credit period of 2002–2005 highlighted the
fatal flaws of our convoluted banking systems.
11 It is commonly recognized in the industry that ‘structural mitigants, balanced equity return
profiles and managerial interest-aligning incentives [are] the powerful drivers of CDO
performance’ (see, e.g., S&P: ‘Balancing Debtholder and Equityholder Interests in CDOs,
November 2002’).
r 2009 The Author

Traditional New Financial Intermediaries Final Funding

Financial (Financial innovations Sources
Intermediaries that liquefy bank loans)
Borrowers (Originating,
Requiring monitoring, Pension Funds
Funding servicing, Institutional
repackaging) Loan Market
Large Insurance
Money Companies
CLO
Center
Banks
Consumer
Credit Protection
and Credit
Mutual Funds
Protection
Premium
Default
Payment
Business
Loans Swaps
CDO Non-guaranteed
Deposits
Smaller MBS, CMBS

Banks keep the
Banks ABS
illiquid economic
core containing
information rent
Figure 3 Liquefaction of Bank Loans and Final Funding Sources.
question is what the market actually requires as incentive mechanisms that would
allow bank loans to be liquefied in practice.
Figure 3 provides a road map of how large and small loans from big and small
banks become liquefied in the market. Large loans are often traded in the
institutional loan market. Smaller loans can be bundled together for securitiza-
tion. Big money-centered banks can off-load most of the value of their loans via
many channels. Smaller banks can do it through big banks. We will discuss each
of these in turn and then consider them collectively.
A. Growth of the institutional loan market

Starting in the 1980s, bank loans, which were once thought to be nonmarke-
table, became marketable (see, e.g., Gorton and Pennacchi 1995). The average
size of these marketable syndicated bank loans is not small, with a typical size of
US$200 million to US$1 billion. The liquidity of the loan market has
significantly increased recently due to innovations in information technologies
that allow efficient sharing of information from comprehensive databases (e.g.,
Moody’s, S&P Leveraged Commentary & Data). By the end of 2007, over 90% of
these loans (in the outstanding value) are rated by rating agencies. These
standardized measures of risk contribute to the liquidity of the loan market. With
r 2009 The Author

information so widely available on those loans and their obligors, the

institutional investors (based on their own analyses) believe that they do not
suffer any informational disadvantages relative to the loan originating financial
institutions. They trade these institutional loans in the same way they trade
corporate bonds.12 This is probably the most significant reason why the
institutional loan market is liquid.
B. CDS
According to International Swap Dealers Association Inc., outstanding CDS
reached US$62.2 trillion by the end of 2007 (total wealth of the United States
was about US$55 trillion). A CDS is simply an agreement between two parties to
exchange credit risk on a reference asset (e.g., a bank loan or a corporate bond)
or a basket of them. The main idea behind a CDS is that it will allow a bank to
sell the credit risk of a loan without actually selling the loan.
For example, a bank wants to transfer the credit risk of a loan on its book to
an institutional investor, say, a mutual fund. In this case, the bank buys credit
protection from the mutual fund and the mutual fund sells credit protection to
the bank. The bank will make periodic swap counterparty payments to the
mutual fund (like interest payments). In the event of a credit event related to
the reference loan (failure to pay, bankruptcy, restructuring, repudiation/
moratorium and obligation acceleration: based on International Swap Dealers
Association Inc. credit swap master agreement), with the event independently
verified by third parties or public information, the mutual fund will make a
credit protection payment to the bank. The credit protection payment is equal
to the difference between the notional amount of the defaulted reference
obligations and the loan recovery value determined by a ‘calculation agent’
(usually a bank or a group of banks, verified by third parties) at certain time, say
180 days, after the credit event.
In this example, the transaction is similar to the mutual fund ‘buying’ the
loan from the bank. The mutual fund will be receiving periodic ‘interest
payments’ in the form of swap counterparty payments, but will have to bear the
risk of default. The bank transfers the credit risk of the loan to the mutual fund
investors and typically receives the ‘excess spread’ (the spread between the
interest it receives on the actual loan and the ‘interest’ it pays to the mutual
fund) as incentives for monitoring and servicing the loan. In a typical CDS
involving a basket of loans (see also synthetic CLO below), there is usually a first
loss piece kept by the bank, making the bank the residual claimant, and a
12 We thank PIMCO, one of the world’s largest fixed income management companies and bank
loan buyers, for providing us with details on the stylized facts of the institutional loan market.
Interested readers can also consult ‘A guide to the loan market (S&P).’ When Tyco International
Ltd. turned to its banks for a US$1.5 billion loan in 2003, ‘Morgan Stanley, Bank of America, J.P.
Morgan, Citigroup, Goldman Sachs Group and the Credit Suisse First Boston unit of Credit
Suisse Group – each agreed to back $250 million of the $1.5 billion credit line. The banks likely
won’t retain all of their $250 million pieces; they are in the process of parceling out pieces to
other banks and institutional investors’ (Wall Street Journal 1/13/2003).
r 2009 The Author

Swap Counterparty Payment

(protection premium)
Credit Protection Payments

Reference Swap
Portfolio (reimbursement of losses
after reserves and threshold) Counterparty
of
Loans
Threshold Realized losses

(first loss) reduce reserves
and then
Excess Spread
the threshold
(periodic replenishment) Reserves before CDS
Figure 4 Credit Default Swap (CDS) on a Portfolio of Loans.
reserve account (see Figure 4) funded by the bank to align the bank’s interests
with the swap counterparty.13 The reserve account is periodically replenished from
the excess spread cash flows to keep the bank’s interests continue to be aligned
throughout the life of the CDS even after some defaults.
This type of setup has far-reaching applications in terms of channels for
transferring credit risks. As noted by Jones (2000), ‘. . . a money-center bank or a
securities firm might sell credit protection to regional banks whereby the
guarantor promises to cover all losses above a certain amount against a specified
pool of loans.’ Thus, a CDS is a feasible way for regional banks, not large enough
to directly liquefy their loan portfolio into the market, to unload their credit
risks to larger banks and securities firms, who may then bundle many of these CDS
into a deal large enough to be of interests to institutional investors in the
market. Conversely, smaller banks may sell credit protection to larger banks on
syndicated loans in order to get exposure to the risk and reward of the corporate
clients of larger banks (Bank for International Settlements Annual Report, June
2003). The very low structuring costs fuel the phenomenal growth of the CDS
market and make the transfer of a loan’s credit risk extremely liquid. With the
prevalence of CDS, it is increasingly unclear how much of the credit risk of the
commercial and industrial loans remaining on the bank’s book is still borne by
the banks as ‘banks were net purchasers of credit protection while insurance
companies and financial guaranty insurers were important net sellers’ (BIS,
2003).
C. CDO and CLO

CLO and CBO are simply securitizations of cash flows from loans and bonds.
They are collectively called CDO. In the simplest case of a CLO, it takes a
portfolio of commercial and industrial loans as assets and issues securities with
13 See related incentive issues discussed by Pennacchi (1988) for bank loan sales.
r 2009 The Author

Asset Manager Trustee Obligor Obligor

(Manages issuer’s (Protects investor’s #1 #2
assets) securities interest in
the collateral and Loan
performs other agreements
fiduciary duties)
Bank Bank
#1 #2
Assignment
Portfolio of bank agreements
ABS loans
Investors Issuer
(Buy rated ABS) Special-Purpose Entity Seller/Servicer
Proceeds of (Buys loan assignments and Proceeds of (Assigns a portfolio of
rated issues ABS using the loans rated loans to the issuer of rated
securities as collateral) securities securities)
Interest and Interest and
Principal on Principal on
rated securities rated securities Assignment
Swap agreement agreement
Bank
#3
Swap counterparty
(Provides currency & Assignment
interest rate swap to hedge agreement
against currency & interest
rate-related risk Obligor Agent
#3 Loan Bank
agreement
Sample CLO Transaction

(Source: S&P)
Figure 5 A Typical Collateralized Loan Obligation.
claims against the cash flows. Figure 5 illustrates a typical structure (S&P: Global
CBO/CLO Criteria). The seller/servicer chooses loans from different banks to
form a diversified portfolio and transfer it to the issuer, usually a special-purpose
vehicle (SPV), with an asset manager, a trustee and possibly swap agreements to
hedge against interest rate and currency risks. The bankruptcy-remote SPV sells
asset-backed securities (ABS) backed by most of the cash flows from the loans
and retains an equity position with claims against the residual cash flows.
The main idea is similar to that of mortgage-backed securities (MBS), which
became popular following the financial market liberalization of the late 1970s.
At the time, it was obvious that it would be more profitable for banks to
originate and service the mortgages rather than funding them because banks do
not have any particular advantages in holding something so standardized,
homogeneous and devoid of private information as residential mortgages. By the
r 2009 The Author

early 2000s, the size of MBS alone was roughly the same as the US Treasury
market. The CLO market took off around 1996 following the other ABS market.
The late start of the CLO market was due in part to the uncertainty of how to
structure the securitizations acceptable to the market, knowing that banks may
have private information – the same reason why bank loans are considered
illiquid in the first place. Over the past several years, the CLO market has
endured market tests with exponential growth and the outstanding global
volume already exceeded the one trillion benchmark by 2004.
The main reason for the success of the CLO market is that securitization
transforms something heterogeneous and illiquid (loans with private information) into
something homogeneous and standardized (e.g., an AAA tranche) so that it is
relatively liquid. Since this is the critical link in liquefying banks’ commercial and
industrial loan portfolios, we will illustrate with several examples how CLO
technologies have evolved over time to deal with incentive issues and provide
empirical evidence on the feasibility of liquefying bank loans.
i. Typical Bank CLO and their incentive alignment mechanisms

A simple example. The NationsBank (now part of the Banc of America)
Commercial Master Trust14 provides an example of a simple CLO (Figure 6). In
their series 1997-2, NationsBank had an initial loan pool of more than 1000
loans in 50 industries. The internal/external loan rating distribution was 2.07%
AAA, 0.37% AA, 10.35% A, 45.21% BBB and 42.0% BB.15 The securitization
backed by these loans has four tranches due in 2002. Class A is a US$2 billion
floating-rate asset-backed certificate with an AAA rating, Class B (subordinated
to Class A) is a US$66 million certificate with a rating of A, and Class C
(subordinated to Class B) is a US$66 million certificate with a rating of BBB.
Class D, with a par value of US$66 million, is the residual ‘equity’ tranche
retained by the originating bank, NationsBank, which backs the securitization
with an additional 1% reserve. The credit enhancement to bring Class A to an
AAA rating, in a loan pool with a weighted average rating of BBB , was made
possible by the subordination of Class B, C and D (representing about 9% of the
loans) and the 1% reserve. Since Class A, representing more than 90% of par
14 We thank Greg Duffee for pointing to us this example.

15 This is also roughly the distribution of ratings in many bank CLO and the distribution of the
corporate borrowers at the point of their bank loan originations. Approximately, 80% of bank
loans are to companies with ratings Baa/BBB or Ba/BB (including companies that are not
publicly rated, but would have fallen into this category based on banks’ internal ratings;
interestingly, the loans to those publicly unrated corporate borrowers would have a risk-based
capital requirement at the same level of publicly rated BBB corporate borrowers in Basel II [April
2003]). This distribution is consistent with the findings of Quantitative Impact Study 3 –
Overview of Global Results for Basel II (May 2003) for the performing loan qualities in Group 1
banks (large international banks) and in Group 2 banks (smaller specialized banks). Corporate
borrowers with higher ratings would find raising money directly in the financial markets
cheaper. Companies with ratings lower than BB would find bank financing prohibitively
expensive – they are better off raising fund from junk bonds or private capital. We thank Banc of
America, PIMCO and CreditSpectrum for providing us the relevant information.
r 2009 The Author

Class A
Trustee AAA
$2,000mn (91%)
Class B
A
Principal &
Bank loan interests $65.9mn (3%)
Special Purpose Vehicle
NationsBank
(SPV)
$ NationsBank Notes
CLO Corp. Class C
proceeds BBB
$2,197mn
$65.9mn (3%)
Class D
NR
$65.9mn (3%)
Reserve
$22mn (1%)
NationsBank CLO
(Source: Moody’s, S&P, Fitch)
Figure 6 Example of a Collateralized Loan Obligation (CLO).
value the loans, has a rating of Aaa/AAA/AAA (Moody’s, S&P, Fitch ratings),
there is a market for it among the world’s institutional investors. The credit
rating of Class A note is actually higher than the AA rating of NationsBank
because of the survivability of the security interest in the notes even if
NationsBank becomes insolvent. Class B and C are also of investment grade and
there is reasonable liquidity for them. The credit risk of the original loans is
concentrated in Class D (to be kept by NationsBank) with a par value that is 3%
of the original loan value (with additional support through a 1% reserve).
Through subordination, the CLO creates reasonable liquidity for more
than 96% of the loan value of a portfolio dominated by BBB and BB loans that
are presumably laden with private information. As finance theory would
predict, when there is substantial asymmetric information between the sellers
and buyers of bank loans, it is efficient for the bank that originated the loans to
hold the risk–reward for the private information (the residual tranche). In this
r 2009 The Author

CLO, financial innovations combined with an independent rating agent, who

also serves as the monitoring agent,16 mitigate the problems associated with
private information to the extent that most of the loan value can be sold off to
outside investors. This is the typical structure of a CLO: after securitization of
the original diversified loan portfolio with an average rating of BBB , ‘97% of
the issued securities could receive an investment-grade rating. The remaining
3% of the pool would be held as equity’ (Fitch research report December 1997,
‘Bank Collateralized Loan Obligations: An Overview’).
Synthetic CLO: offloading credit risk without selling loans. The transfer of
credit risks can be more easily done via CDS because of the lower structuring
costs, especially if the loans are from different legal jurisdictions (to gain more
geographic diversification). In Figure 7, the Deutsche Bank’s Globe-R 2000-1 is a
synthetic CLO because Deutsche Bank will keep the loans on its book but use
CDS to transfer the credit risks to eventual investors. The super senior tranche
of h1.722 billion (86% of the deal) is an ‘unfunded CDS’ in the sense that the
super senior tranche is not sold as notes, therefore there is no funding. The
Deutsche Bank portfolio contains mostly spot loans whose interest payments
can be thought of being decomposed into two components: one supporting the
funding and the other supporting the risk (swap counterparty payments on the
CDS). In periods when funding is widely available in the market, banks would
prefer to sell the credit risk component without selling the funding
component.17
Credit lines and synthetic CLO with unfunded CDS. A synthetic CLO
structure with an unfunded CDS is often used when the reference portfolio
contains credit lines that are undrawn. This is an important financial innovation
that takes into account the recent phenomenon that a significant fraction of
money-centered bank loans are in the form of credit lines (see Kashyap et al.
2002 in the context of deposit-funding synergy). For example, CitiStar
(Citibank) 1999-2 is a US$4 billion synthetic CLO that references 233 senior
unsecured facilities in the United States and Canada in which around 80% was
undrawn at offer. Verdi (IntesaBci) 2002 is a h4 billion synthetic CLO in which
89.5% are undrawn revolving credit facilities (Figure 8). When the credit lines
are drawn, the bank would fund them with its working capital or through the
interbank market (others’ working capital).18 As the credit line is already
16 A critical failure of the monitoring and risk assessment functions of the rating agencies
contributed to the subprime crisis in 2008. As the crisis developed, the rating agencies hastily
revised downward the ratings of many mortgage-backed securities based on risky mortgages.
17 See, e.g., Wall Street Journal (December 17, 2003) article entitled ‘Banks itch to lend, but firms sit
tight.’ If funding becomes tight later, banks can always sell the funding of the already
securitized tranches.
18 In our 100% reserve banking, working capital includes cash, deposits with monetary authority,
T-bills or their close equivalents. The LIBOR market, the EURIBOR market and the interbank
capacity are ultimately determined by the money supply controlled by the monetary
authorities [see Fisher (1936) and Friedman (1960) for related issues on monetary policies
corresponding to 100% reserve banking].
r 2009 The Author

GLOBE-R 2000-1 Structure
Deutsche Bank
A.G. and
subsidiaries Premium Unfunded
(Portfolio of loans credit default
Trustee
and guarantees to swap
Principal and interest on class A notes

“AAA” fandbriefe
large international Credit Eur 1.722 Bn (86%)
corporations) protection
Eur 2.003 Bn payments
Class A
note proceeds
Class A+ (N.R.)
Eur 0.001 Bn (0.05%)
Class A (AAA)
Eur 0.06 Bn (3%)
Deutsche
Bank A.G. Realized losses reduce principal
Credit events (Issuer) on all notes sequentially Class B (A)
leading to realized Eur 0.06 Bn (3%)
losses on the
portfolio
Class C (BBB)
Eur 0.04 Bn (2%)
Class B, C, D, and E
notes principal and interest
Class D (BB)
Eur 0.04 Bn (2%)
Class E (N.R.)
Class B, C, D, and E Eur 0.08 Bn (4%)
note proceeds
(Source: S&P)
Figure 7 A Synthetic Collateralized Loan Obligation (CLO).
securitized with a synthetic CLO, when the credit line is drawn, the structure
evolves into a CLO structure where the bank would hold an obligor’s spot loan
(rather than credit line) whose credit risk has already been transferred through a
CDS.
The underlying idea is best illustrated with the help of the CLO of
IntesaBci corporate loan portfolio in Figure 8. The credit risk is transferred via a
CDS to a counterparty (usually a financial institution with an OECD bank, such
r 2009 The Author

Transaction Structure
(Source: Merrill Lynch)
Figure 8 Example of a Collateralized Loan Obligation (CLO) with Undrawn Credit

Facilities.
as Merrill Lynch, for regulatory capital reasons), who will in turn parcel out the
risk to other institutional investors. The commitment fees from the undrawn
credit lines are used to cover the swap counterparty payments on the
‘unfunded’ super senior CDS (h3640 million or 91% of the deal). When the
credit lines are drawn on IntesaBci, the structure will gradually evolve into the
case of Deutsche Bank with spot loans in Figure 7.
‘Excess spread’ as a dynamic incentive alignment mechanism. In many CLO

structures, the bank will use ‘excess spread’ in addition to the ‘first loss’ as an
incentive alignment mechanism to convince investors that the bank will
continue to diligently monitor and service the loans. An example would be the
synthetic CLO of Amstel Corporate Loan Offering (ACLO 2000, the SPV) offered
by ABN AMRO Bank. In the ACLO structure, AMRO bank will use the ‘excess
spread,’ [i.e., the difference between the interests that it receives from the
original loans and the interest (including swap counterparty payments) it pays
r 2009 The Author

to ACLO] to build up a reserve account at an annual rate of 0.5% of the

reference portfolio amount, which is available to ACLO when making credit
protection payments. For each period (quarter), if the reserve account is
exhausted because of credit events, the loss will be allocated to the tranches in
reverse order of their priority: Class F first, and then, Class E, and so on until
Class A and then the super senior. The excess spread is an important dynamic
incentive alignment mechanism that diverts fresh money from the loan-
originating bank into the reserve account, regardless of what the past default
experience was, to keep the loan-originating bank as the residual claimant
throughout the life of the CLO.
It is worth emphasizing here the importance of such an incentive
alignment mechanism in driving the performance of recent CLO (see footnote
11). Diamond and Rajan (2001) discuss the implications of the fragility of our
banking system (demand deposits) in aligning the interests of the bankers with
the depositors in creating liquidity. Here, the excess spread is another practical
alternative devised by the market to solve the same problem without the
fragility of the banking system. With this mechanism, even if all the lower
tranches are in default, the excess spread still puts in fresh money periodically
into the reserve to entice the banker to act in the interests of the investors of the
surviving tranches.
Loans to small- and medium-size enterprises (SME). As expected, a large
diversified portfolio to SME is easy to securitize. Each securitization may contain
thousands, or even tens of thousands, of loans. An example is Geldilux 99-1
offered by HypoVereinsbank. The total loan portfolio is slightly more than h2
billion with 1818 small- and medium-size corporations and private borrowers.
Almost no borrower has a rating from Moody’s or S&P.19 Of the five tranches:
Class A (94%) is Aaa/AAA, Class B (2.6%) is A/A, Class C (1%) is Baa/BBB, Class D
(1.5%) is Ba/BB and Class E (0.9%) is the residual tranche without rating. The
size of the residual tranche is small because SME loans are like consumer loans
in many aspects. This fact is also recognized in the Basel II consultation paper
(April 2003) in that SME loans may be separated from other corporate loans,
and loans to small businesses can be qualified as retail risk exposure (like
consumer loans) warranting a lower risk-based capital requirement. The SME
loans originating from smaller banks can be ‘sold’ (via CDS) to larger banks or
sold directly into ‘arbitrage CDO’ in the same way smaller banks unload their
consumer loans.
Representativeness of the loan portfolio. There are two competing motiva-
tions for banks in selecting which loans to include in the portfolio to be
securitized. Banks would want to get rid of their worst loans, but, over time if
the bank misrepresents the true risk of the underlying loan portfolio, this
‘lemon problem’ will ruin its reputation for future deals and cause its own
19 In rating SME, Moody’s would use a binomial simulation based on the bank’s internal rating
system while S&P would use an actuarial approach based on recent default rates. The ratings
from the two agencies are usually consistent.
r 2009 The Author

securitization market to break down. On the other hand, banks would like to
securitize their best loans in order to reduce their capital requirements as the
best loans would require a smaller residual tranches to be kept, but this would
get the regulators really upset because what would remain on the bank’s book
would be loans of lower quality.
Banks, regulators and investors are all aware of this selection bias problem.
Thus, in many deals, there is voluntary disclosure on the representativeness of
the loans. Indeed, in offering prospectuses and rating agency reports, there are
always comments on the loan quality with explicit statements attesting to
being representative of the loans within certain category (e.g., performing loans
originated from normal banking operations). To appease regulators and
investors alike, some structures go as far as to state that the loans are selected
randomly or that all loans are included within a certain category. Furthermore,
rating agencies, being the outside monitors, often insist in their reports that the
ratings depend on the quality and consistence of the bank’s internal credit
rating and the bank’s continuing effort in monitoring and servicing and
extracting recovery values for defaulted loans. Any deterioration of these would
lead to a rating downgrade that will impact the bank’s ability for future
securitization.
In summary, financial engineering is capable of liquefying almost the
entire loan portfolio of a bank. As we will see in the next section, the remaining
illiquid portion that must be kept by the banks is relatively small.
IV. EMPIRICAL ESTIMATION OF THE SIZE OF THE RESIDUAL

TRANCHE
There are more than 20,000 structured finance deals in the 2003 data bank of
Moody’s and S&P, dating back to the early 1980s. Most of these are residential
and commercial mortgages and consumer loans because securitizations of
commercial loans have been a more recent phenomenon. A sample of recent
CLO of commercial loans from North America, Asia and Europe is given in
Table 2. This sample is based on a data set of CLO reports from S&P for the
purpose of illustrating the typical structures of all the recent CLO before the easy
credit period. These are supplemented by Moody’s and Lehman Brothers reports
for the missing data. Our final sample is based on all the CLO that have complete
data with respect to structure details, the sponsoring bank, ratings and tranche
sizes, especially the residual tranche (or first loss). If there are several offerings
based on the same master trust, only the most recent one with complete data is
included. There is a sample bias in favor of more observations from European
banks because European Banks are still the main source of financing for many
corporations in Europe.20
20 We were also told that American deals become so commonplace that usually only a one-page
deal summary with the ratings is in the database.
r 2009 The Author

Table 2 Bank collateralized loan obligations (CLO) 1997–2002

Issue Tranches with Tranches with non- Residual (%)
investment grade (%) investment grade (%)
ACLO 2001 1-2 99.75 0.00 0.25

ALCO 1 95.50 0.00 4.50
Aurora Funding 94.70 4.30 1.00
Brooklands Euro 94.80 2.20 3.00
CnStrategic 1999-1 97.70 1.30 1.00
CnStrategic 1999-2 97.59 1.41 1.00
Cast 1999-1 96.00 1.00 3.00
Cast 2000-2 95.25 1.55 3.20
CDO 95.20 0.00 4.80
CDO 2 97.60 0.00 2.40
CDO 3 97.85 0.00 2.15
Chase 95.00 0.00 5.00
Clover No. 2-3 94.52 1.31 4.17
Clover No. 4 96.90 0.00 3.10
Cordusio 94.80 3.00 2.20
CORE 1998-1 96.10 2.18 1.72
CORE 1999-1 96.63 2.12 1.25
CORE 1999-2 96.56 1.97 1.47
Credico Funding CBO 97.00 0.00 3.00
CROWN CLO 95.10 0.90 4.00
CYGNUS 97.40 1.90 0.70
Cygnus 2001-1 96.73 0.73 2.53
Eirles Two Ltd. 93.50 2.50 4.00
Fleet CLO 96.50 0.00 3.50
Fondo BBVA-1 94.88 2.44 2.68
Fondo BBVA-2 SME 94.67 1.83 3.50
Fondo PYMECAT-1 98.20 0.00 1.80
FTPYME TDA 95.70 2.40 1.90
Geldilux 02-1 97.90 1.30 0.80
Geldilux 99-1 97.64 1.47 0.89
Geldilux 99-2 97.05 1.75 1.20
Globe R 2000-1 94.01 2.00 4.00
Imperial II CDO 94.50 0.50 5.00
London Wall 2002 1-2 96.70 0.70 2.60
Melrose 2001 1-2 92.15 3.43 4.42
NationsBank CLO I & II 96.01 0.00 3.99
Olan II 98.10 0.00 1.90
Park Mountain Capital 95.60 1.60 2.80
Promise A 94.80 1.10 4.10
Promise I 95.75 1.25 3.00
Promise K 94.51 0.95 4.53
Promise Z 93.90 1.60 4.50
Repon 16 98.10 0.25 1.65
Riviera 1 S.A. 94.79 0.00 5.21
Riviera 2 S.A. 95.79 0.00 4.21
Rose No. 2 96.29 0.00 3.71
Scala 3 97.60 0.00 2.40
SMILE 2001 99.01 0.00 0.99
Sundial 95.50 3.00 1.50
Verdi 97.00 1.00 2.00
r 2009 The Author

Table 2 Continued
Summary Percentage of Percentage of non- Percentage of
statistics investment grade investment grade residual
Average 96.02 1.16 2.82

Median 96.00 1.10 2.80
Maximum 99.01 4.30 5.21
Minimum 92.15 0.00 0.70
Standard deviation 1.45 1.09 1.33
Total issues 50
All the CLO in Table 2 are from major international banks. Roughly half of
these CLO are based on SME loans. With a few exceptions, the size of each deal
is at least US$1 billion. The range is from about US$150 million to about US$12
billion (the exact size depends on the exchange rate as different tranches of the
same deal may be offered in different currencies to different markets). Most of
the deals are in the range of US$1 billion to US$2 billion, which appears to be
the typical size of interest to the market. Although we do not have hundreds of
independent deals, each data point itself is an extensive empirical and
optimization study on its own (no data mining, just billions of dollars at
stake). The patterns across data points are remarkably consistent. Perhaps the
most telling figures from Table 2 are the average sizes of the investment grade,
non-investment grade and the residual tranches (including reserves, if any).
Our sample shows that about 96% of the cash flows from the loans can be sold
as investment grade. The most illiquid portion pertaining to the residual
tranche is only about 3%.
The ‘3% residual’ is a very revealing figure that reflects the size of the ‘market
determined capital’ required to support the securitizations that sell off the rest
of the cash flows. The size of the residual depends on the quality of the
underlying loan pool. Most of the reported weighted average ratings (if
available from the rating agency report) of the underlying loan pools in Table 2
are about Baa3/BBB and the required equity tranche is about 3% for the
liquefaction. If the average loan quality of the pool is BB, a rough estimate
(source: PIMCO, Deutsche Bank) of the size for the required equity tranche is
between 8% and 12%.
We should emphasize that the underlying loans in our sample on Table 2 are
all performing loans from large international banks including Citibank, Chase,
Banc of America, FleetBoston (now part of Banc of America), HSBC, Sumitomo
Bank, Deutsche Bank, ABN AMRO, BNP, IntesaBci and others during a non-crisis
period. The loans they securitize are loans from their normal banking
operations. Most of these large banks have an Aa/AA rating, which is an
indirect reflection of the average quality of their loan portfolios. The rating on
the bank itself is based on its own unsecured senior debt, with liquid deposits
above it in the capital structure. In other words, the existing capital in these banks,
taking into account of the quality of its entire loan portfolio and the liquid deposits
r 2009 The Author

above, can support an Aa/AA rating for its unsecured senior debt.21 For these banks,
it is not surprising to find that the market-determined capital required for
supporting the liquefaction of their performing commercial loan portfolios is
not much more than 3%.22 Of course, this number is dependent on the general
economic conditions and would likely be higher during periods of economic
crisis.
On the other hand, it is interesting to look at cases at the other extreme
where the bank is of marginal rating because of the quality of its loan portfolio
during a banking crisis. Here, the 3% residual may not be a good estimate of the
market-determined capital to support its commercial loans.23 The CLO of
Shinsei Bank provides a meaningful example very different from the banks in
Table 2. Shinsei Bank (not one of those mega banks in Japan) used to be Long
Term Credit Bank of Japan. It was taken over by the Japanese government in
1998 and then re-privatized in 2000. Shinsei Bank’s own rating is at the
minimum investment grade Baa3/BBB (Moody’s/S&P), and it has a series of
CLO based on a master trust on its loan book. In order to rate the CLO tranches,
the rating agencies had to consider the fact that many ‘performing’ loans in
Japan could be doubtful. Furthermore, the recovery value once the loan is
defaulted is assumed to be zero (S&P), which is materially different from the
assumptions about other OECD bank loans (S&P recovery assumption: 50–60%
for senior secured, 25–50% for senior unsecured and 15–28% for subordinated
loans; Moody’s experience: 69.5% for senior secured and 52.1% for senior
unsecured). Despite the doubtful nature of the performing loans and the
extreme assumption of zero recovery value, Shinsei Funding One (03/06/2002;
source: Lehman Brothers), has a Class A tranche (75%) rated Aaa/AAA, a Class B
tranche (10%) rated Baa/BBB and the rest (15%) non-investment grade and
residual. This example suggests that a rough bound of 15% of the capital is
necessary for a solvent but marginal bank to liquefy its commercial loan
portfolio.
While it is clear that CLO tranches are more liquid than typical bank loans,
they are not as liquid as bonds of a single major corporation. A single name AAA
21 The weighted average rating of Western European and US banks is between AA and A1 in
2002 (Financial Times, May 21, 2002).
22 Interestingly enough, if we split our sample into two subsets, one containing only loans to SME
(which presumably private information plays a lesser role in a large diversified portfolio as the
loan process follows rather standardized procedures) and the other containing loans to larger
corporations, the average size of the residual is about the same. The higher average credit
quality of the larger companies offsets the higher amount of private information presumably
more important in the larger loans.
23 The average corporate loan quality in banks (excluding SME) is between BBB and BB (footnote
15). The number of CLO based on defaulted loans, non-performing loans and doubtful
performing loans is increasing, but the total number is still small. For example, Ark CLO 2000-1
is based on a portfolio of distressed and defaulted loans from Fleet Boston, Korea Asset Funding
2000-1 is a CLO based on restructured corporate loans and International Credit Recovery –
Japan One Ltd. is based on non-performing loans. In these cases, summary statistics are less
appropriate when the underlying portfolios are so different.
r 2009 The Author

bond is more liquid than an AAA-securitized tranche because a single name

bond is generally much easier to analyze than a pool of loans. To increase
liquidity of those securitized loan tranches, the CLO market must follow the
path of MBS in the 1980s to be more standardized and therefore more liquid.
One lesson from the current 2008 meltdown is that by mixing too many
different types of assets in the same pool (e.g., prime mortgages, subprime
mortgages, CDS, equity tranches and subordinated tranches from prior
securitizations), it would make the analysis of such a pool extremely difficult.
In contrast, a plain vanilla CLO such as the NationsBank example in Figure 6 is
easy to analyze and therefore its tranches more liquid.
Going forward, the trend toward securitization is irreversible. To make the
end products more liquid, there should be greater transparency, standardization
and regulation in the process. (1) CLO should be classified into distinct types,
with the loan pool supporting each type originating from more homogeneous,
narrowly limited categories of loans. This would render the CLO easier to
analyze and more liquid. (2) As the securitization and CDS markets have grown
to the point of posing systemic risks to the financial system, it is appropriate to
have a central clearing house and relevant systemic regulations in place in order
to build confidence. (3) Finally, as we emphasize repeatedly, loan-originating
and monitoring banks must hold the equity tranche for proper incentives.
V. LINKS TO THE GUARANTEED AND NON-GUARANTEED

DEPOSITS
A. Bank loans financed by small depositors

In the previous sections, we find that most bank loans can be liquefied.
Furthermore, the market that connects bank loans to non-guaranteed deposits
has already existed and is rapidly expanding. Nowadays, when a small depositor
comes to his favorite bank, he can invest in bank-sponsored, non-guaranteed,
non-FDIC-insured money market funds, mutual funds specializing in bank
loans or Treasury securities, just as easily as he would deposit his money in an
FDIC-insured account. A casual browse through the offerings of major mutual
funds reveals that traditional mutual funds are already playing such a financial
intermediary role in bringing bank loans and depositors together in a
significant way without deposit insurance. Thus even a regional bank can both
retain its information advantage in knowing its borrowers (relationship
banking) and have access to the world market of non-guaranteed depositors
supporting its loan portfolio.
Of course, there is nothing special about the abilities of mutual funds in
securitizing bank loans for the non-guaranteed depositors. Banks can easily play
the same role, and even have some natural advantages, in that they can directly
offer these types of non-guaranteed deposits. This is especially true now when
the distinction between commercial banks, investment banks and brokerages is
r 2009 The Author

fast disappearing. Therefore, in a typical 100% reserve bank (Figure 2), deposits
in risk Class A can be supported by a diversified portfolio of Aaa/AAA tranches.
Deposits in risk Cass B can be linked to a bank-sponsored mutual fund. Mutual
funds provide the benefit of diversifying across many banks and across many
investment types (as in a diversified stock fund) to reduce the risk for non-
guaranteed depositors, in addition to their role in monitoring the banks whose
loans they invest in. Furthermore, competition among mutual funds keeps
them vigilant in regard to their investments. Deposits in risk Class C can be
caveat emptor with limited liquidity backed by lower rated tranches, emerging
market loans, exotic loans, subprime mortgages or even the equity tranche of a
CLO. Such depositors would vote with their deposits and diversify across banks
just like a typical investor would in stock mutual funds. Indeed, since financial
engineering allows banks loans to be liquefied, their risks can be efficiently
distributed in forms consistent with the market demand.
i. Community banks
US bank regulators have been suggesting that only the largest 10–20 banks need
to conform to Basel II because their instabilities might induce systemic risk to
the economy. The 70001 small banks in the United States do not pose such a
systemic risk, although they are still subject to general economic risks and thus
remain a potential liability to tax payers. Thus, it might be politically expedient
for the government to continue to support those community banks with
guarantees,24 even though channels for the smaller banks to liquefy their loan
portfolios already exist. In this age of banking consolidations, it may be simpler
for the smaller banks to (i) understand the local market for loan originations
and monitoring and (ii) sell funds and act more like the deposit taking affiliates
of larger banks.
Therefore, the necessary markets to linking bank loans with investors,
including large institutions and small depositors, have already developed and
will continue to develop. The 100% reserve plan would not disrupt the main
roles of banking beyond some minor repackaging that has already existed for
decades and has been fast expanding in response to regulatory and
technological changes. The government role would be greatly reduced under
the 100% reserve plan. This is fundamentally different from the government’s
present role of having to constantly keep a watchful eye on thousands of banks
in their lending and investment processes with the attendant deposit insurance
moral hazard.
B. Sufficient government debt for the 100% reserve

In the 100% reserve plan, an important ingredient is to back the riskfree deposits
with direct government obligations. This would require the government to sell
24 In the same way as with Small Business Administration, Farm Service Agency, etc. even if they
might be a potential burden to the taxpayers.
r 2009 The Author

enough obligations to satisfy the demand of the economy. At the end of 2007,
the total transaction account (demand deposits, checkable deposits, etc.) in
commercial banks was about US$708 billion, the total US government securities
held in commercial banks was about US$950 billion (source: FDIC), the total
government debt outstanding was about US$7.3 trillion and the contingent
liabilities of Social Security were about US$13.6 trillion (source: Federal Reserve
and the US Treasury). It is not clear if the Treasury needs to float more
government debts to support the 100% plan. If indeed it is necessary, there is a
question as to where the proceeds should go.
These concerns parallel the concerns of what to do if the government on-
budget surpluses would pay off the national debt and the government may have
to accumulate private assets. The original 100% reserve plan suggests that the
proceeds from issuing government securities be used to finance deficits, reduce
taxes, abolish federal taxes, or invest in quasi-government obligations,
obligations of states and municipalities, obligations of international organiza-
tions and of foreign governments, acceptances and other commercial paper.
Any reasonable combination of these would work. Another option would be to
privatize Social Security. Using the Chilean model of recognizing the
contingent liabilities by issuing Recognition Bonds, the government would
issue debt and put the proceeds in a professional managed pension fund (like
the California Public Employee Retirement Fund) in order to insulate federal
investment decisions from political pressures.
The fund would be able to invest in a great variety of financial instruments,
even credit risks in bank loans or CDS. By doing this, the government would not
only create liquidity by issuing liquid debt and investing (like institutional
investors) in bank loans, but would also increase the relative supply of riskfree
investment to risky investments in the economy.
i. Some general equilibrium considerations when there is a systemic shock

This brings out an important but rather subtle implication that deposit
insurance has had on our economic equilibrium. With deposit insurance, if
there is a systemic shock to the economy and there is a flight to riskfree
investments, investors may pull their money out of risky investments and
deposit them into insured bank accounts (assuming for this argument that
FDIC is considered to be riskfree). If banks speculate the proceeds in risky
investments, the government is effectively standing ready to invest in the risky
investments (without the upside) through deposit insurance with its attendant
moral hazard.
In a 100% reserve banking system, if the government does not interfere with
the natural market forces in response to a shock, there is no immediate change
in the relative supply of risky versus riskfree investments and the economy’s risk
premium would rise until the market clears. This is the free market equilibrium.
On the other hand, if it is our national policy for the government to intervene
in the market in response to a shock (e.g., LTCM in 1998 and the collapse of the
banking system in 2008), it can commit to do so by standing ready to swap
r 2009 The Author

government securities with risky loans (Bank of England 2008), or lend out
government securities to banks who put up risky loan portfolios as collateral
(US Federal Reserves 2008). In a dire emergency, the government can commit to
stand ready to buy or finance the purchase of the loans and loan tranches at
prevailing market prices (or to sell credit insurance via CDS at market prices).25
Here, the banks do not face any new moral hazard problems as the tranches
would already be owned by the depositors in our 100% reserve banking system.
The government is merely helping the depositors to rebalance their portfolios
from risky investments into riskfree investments in an orderly way, which
would incidentally change the relative supply of risky versus riskfree invest-
ments and smooth out the shocks in the credit market. Thus, if it is our national
policy for the government to assume some temporary credit risks in order to
absorb unnecessary credit market volatility in a crisis,26 this can be
accomplished with less distortion and in a much less convoluted way than
with our current system of deposit insurance.
VI. THE ADVANTAGES OF THE 100% RESERVE PLAN
The original intention of the 100% reserve plan was to prevent monetary
meltdowns such as those of the 1930s. Interested readers should refer to Fisher
(1936), Friedman (1960) and Friedman and Schwartz (1963) for their analyses
on the advantages of the 100% reserve plan in regard to financial system
stability, monetary policies, price level determination and interest rates. Here,
our analysis is limited to comparing the implications for banking systems
between a 100% reserve system and the current fractional banking system with
deposit insurance and capital requirements.
1. Separation of riskfree and risky investments: In a 100% reserve system,
there is no illusion on what is riskfree and what is risky. The transaction
accounts are backed by deposits with the Fed and guaranteed riskfree
investment accounts are backed by T-bills. Non-guaranteed deposits are
backed by risky loans and their liquefied tranches. For the brave souls
who deposit in tranches backed by risky loans, they knowingly accept
the risks and rewards and vote with their money (just like investors in
the stock market). The 100% reserve system is so simple and
transparent that the responsibility for risky investments would fall
25 Indeed, by the end of 2008, the US Treasury is empowered by the US Congress to purchase loans
and loan tranches from the banks, and finance the purchase by banks of commercial paper that
backs money market funds.
26 Central banks and monetary authorities have also routinely added liquidity to the market by
open market operations and direct deposits (usually via short notice auctions) into the banking
system whenever the credit conditions in the market require (e.g., Federal Reserve 2007–2008).
This can be much better accomplished with a 100% reserve system, as the central bank and
monetary authority do not have to counteract the slack and the positive feedback effect in the
fractional reserve system with capital requirements.
r 2009 The Author

squarely on the shoulders of risk-takers, rather than the government.

The payment system is safe with the 100% reserves and it does not have
any unintended externalities.
2. Moral hazard: At the dawn of federal deposit insurance, Fisher felt that
the accompanying moral hazard that it introduced, together with greed,
would be the root of future banking crises and it could not be solved
by mere government regulations and monitoring. In the last 25 years,
the spectacular banking crises in the United States, Japan and around
the world have confirmed Fisher’s conjecture.27 The FDIC review
(see footnote 3) had a long discussion on why these problems cannot
be easily solved by regulations and risk-based capital requirements and
insurance.28 Under the 100% reserve plan, there would be no need for the
deposit insurance that creates the moral hazard problems. The incentive
problems that would face a 100% reserve bank would be no more than
the incentive problems faced by a typical corporation with an agent
manager and different liability classes, and much less than the acute
moral hazard under deposit insurance, compounded by high leverage
ratios, that banks face in the present system.
3. Market discipline: The FDIC report also lamented the difficulty for
regulators to persuade the banks to adopt more responsibilities in their
investment behavior (e.g., in the LDC loans case) in times when banks
and their loan portfolios are doing well. The depositors were largely
unconcerned as they were protected by deposit insurance and there was
little market discipline. In the 100% reserve banking system, if the
banks cannot justify their risk profiles, the non-guaranteed depositors
and the institutional investors can vote with their money by staying
away from buying those risky tranches. Banks would learn very quickly
that they cannot liquefy those risky loans when there is no market
demand.
4. Contagious runs: When a large bank fails (e.g., Continental Illinois in
the 1980s, Bank of New England in the 1990s, Washington Mutual in
2008), there is always a fear of contagious runs on otherwise viable
banks. For example, in 1984, Continental Illinois suffered a high-speed
27 In October 2008, many of the industrialized countries in the world raised the limits of their
banks deposit insurance to stem bank runs. Ireland, Germany, Iceland, Australia, New Zealand,
Austria, Denmark and others moved to guarantee 100% of their bank deposits.
28 For example, they found several risk characteristics common to the majority of the failed banks,
but such characteristics would ‘flag a much larger number of banks that did not fail.’ The latter
group of banks could well have extracted more profits from the same assumed risks because of
superior managerial skills. It is hard to imagine that standard government one-size-fits-all
regulations on ‘risk’-based penalties can solve the moral hazard problems when the
measurement of risk itself is fraught with errors from the omissions of relevant but non-
quantifiable characteristics such as managerial skills in risk management and relationship
banking (see also Mingo 2000 and Jones 2000). It is obvious that government regulations can
never hope to match the market balancing of interests through optimal contracting and
monitoring in our variant of the 100% reserve banking system.
r 2009 The Author

electronic bank run and sustained enormous withdrawals of foreign

deposits through electronic transfers. The first move of FDIC was to
protect both insured and uninsured depositors29 to contain the damage
(at the cost of aggravating the moral hazard problem even further) and
prevent possibly other bank runs. In the LDC loans case, the US
regulators chose stability over market discipline30 to prevent the US
financial system from a monetary meltdown like those in the early
1930s. On September 25, 2008, FDIC seized Washington Mutual, the
largest bank failure with US$307 billion in assets after depositors
withdrew US$16.7 billion in the previous 10 days, and sold it to JP
Morgan in the same evening to prevent any bank runs the next day.
What has started as a subprime crisis in the United States would soon
engulf banks as far away as Iceland where the government nationalized
all the major Icelandic banks in October 2008, even though the
Icelandic banks do not invest in subprime loans. Such catastrophic
externalities would never happen to a 100% reserve system.
5. Capital requirements, credit crunch and positive feedback effect: Our current
system of backing insured deposits with capital requirements produces
an unfortunate positive feedback effect that occasionally freezes up
credit.31 This is especially true for the Japanese banking crisis in the
1990s and the world’s banking crisis in today. The loan overhang and
its impact on the capital requirements in banks have been hampering
the banks’ traditional role in lubricating the economy. It was a major
contributing factor in the Japanese economic stagnation and its
destabilizing effect on the East Asian economies that led to the 1997–
1998 financial crises (see Miller 1998). In contrast, under the 100%
reserve system, a bank’s capital is compartmentalized (as in a
submarine) into thousands of units, each supporting the residual
tranche of a separate securitization. The bank would not sink even if
the capital supporting a particular CLO is driven to zero (in response to
major credit events like LDC loans or subprime mortgages). If there are
profitable opportunities, banks will continue to lend and securitize.
This is because the maximum that a bank could lose would be limited
to the residual tranche that it keeps as incentives. Thus, unless there is
fraud, banks can continue to function and they are bankruptcy-remote.
29 Federal Deposit Insurance Corporation Improvement Act of 1991 made it more difficult to
protect uninsured depositors in resolving bank failures.
30 ‘US bank regulators, given the choice between creating panic in the banking system or going
easy . . . had chosen the latter course. It would appear that the regulators made the right choice’
[L. William Seidman (former chairman of FDIC) in Full Faith and Credit (1993)].
31 The US ‘credit crunch’ of the early 1990s was well reported. See, e.g., The Economist ‘Crunch by
Credit,’ (November 1990), The Economist (November 1991), Wall Street Journal (November 4,
1992), New York Times (February 23, 1993). The 2008 credit crunch prompted governments
worldwide to guarantee interbank lending and money market funds, inject capital directly into
banks and buy up commercial paper and bad loans.
r 2009 The Author

A. The role of banks in a 100% reserve system

Recent banking literature focuses on the special role that banks play as a
financial intermediary. Obviously, banks can play their traditional financial
intermediary role by bringing together borrowers and investors with less
restriction imposed under a 100% reserve system than a fractional reserve
system with deposit insurance and government regulations. If banks can
capture their private information rent in their relationship banking under the
existing system, they can likewise do so after their loans are liquefied into
tranches to match any of a variety of market demands and incentive schemes.
These are private arrangements arising from optimal contracting and not the
incidental by-products of our clumsy banking system.
Thus, a bank within the 100% reserve banking system has four roles to play:
1. It provides a perfectly safe payment system including an electronic
medium for exchange. The deposits in transaction accounts are
matched against deposits with the monetary authority.
2. It accepts guaranteed riskfree deposits and invests them in treasury bills
or their close equivalents. Such deposits would be guaranteed by
government agencies like FDIC and there is no need to place any limit
on the size of the accounts. Banks are not allowed to take government-
guaranteed riskfree deposits and speculate the proceeds in risky private loans.
3. It originates and structures private loans and repackages them for large
and small non-guaranteed depositors and earns fees in servicing and
monitoring them.
4. It holds concentrated risk positions, funded by the shareholders of the
bank, to take risk that it cannot credibly sell off and capture the
potential reward and the private information rent.
In this 100% reserve system, the cashflows from the bank loan portfolio are
decomposed and structured to match the demands of those who have the
natural economic rationale for the ownership. The homogeneous liquid senior
tranches, in which the banks have no comparative advantages in funding them,
are sold to those who ultimately want to own them. The more speculative
investors would take the junior tranches. Banks, as the claimants of the residual
cashflows in the ‘excess spread’ and the equity tranche, are incentivized and
rewarded for their natural financial intermediary role – relationship banking,
private information, monitoring and servicing.
VII. CONCLUSIONS
Centuries ago, banking started with a 100% reserve concept of depositing gold
and other valuables for safekeeping with goldsmiths and transferred through
paper evidence called ‘bank money.’ All these began to change when certain
goldsmiths decided to start a side business of issuing bank money (loaning out
r 2009 The Author

gold) in exchange for loan repayment promises. It led to the collapse of the
Bank of Amsterdam two centuries ago, the wild cat banking era in England
more than a century ago, the financial meltdown in the 1930s and the crisis in
the 1980s and 1990s in the United States, the banking crisis in Japan and the
collapse of the banking system in the United States and Europe today. The
mixing of risky loans and safe storage of value has given depositors the
unrealistic expectation that their money is safe with the banks, even though
banks turn around and speculate their money in risky loans.
It would seem that a simple cure for this phenomenon would be to separate
these two functions of a bank into two different departments, while keeping
them within the same bank to preserve synergy. The depositor who does not
want to risk his money can put his money in a transaction account backed by
100% reserves or a guaranteed riskfree savings account backed by T-bills. A
depositor who wants to take some risk can ask the agent bank to invest on his
behalf in risky loans. Instead, the United States opted for a clumsy system of
deposit insurance, regulations, surveillance, and capital requirements – with
unsuspecting taxpayers standing by as occasional unintended participants.
Economists of future generations, looking back on the 20th century, will marvel
at how we ever came up with such a convoluted system.
The evidence over the past 25 years in the United States, Japan and the world
clearly shows that the current banking system is not much of a success.
Fundamental banking reform is necessary to make our system more robust and
stable. Recent financial innovations allow liquefaction of almost all of a bank’s
loan portfolio. These liquefied tranches can be funded by institutional investors
and small non-guaranteed depositors. The bank then holds the residual that
contains a concentrated tranche of risk and reward for its private information
and its monitoring efforts. Consequently, our variant of the Fisher (1936) 100%
reserve narrow banking scheme incentivizes banks to continue their traditional
role in originating and monitoring private loans while simultaneously
providing a simple and robust solution that eliminates banking crises and
makes banks bankruptcy-remote.
Nai-fu Chen
Merage School of Business
University of California
Irvine
CA 92697
USA
nchen@uci.edu
REFERENCES
Boot, A. (2000), ‘Relationship Banking: What Do We Know?’, Journal of Financial

Intermediation, 9, 7–25.
r 2009 The Author

Calomiris, C. (2000), U.S. Bank Deregulation in Historical Perspective. Cambridge, UK:

Cambridge University Press.
Calomiris, C. W., and C. M. Kahn (1991), ‘The Role of Demandable Debt in Structuring
Optimal Banking Arrangements’, American Economic Review, 81, 497–513.
Chen, N.-f. (2001), ‘The Hong Kong Currency Board during the 1997–98 crisis:
Problems and Solutions’, International Review of Finance, 2, 99–112.
Dahiya, S., M. Puri, and A. Saunders (2003), ‘Bank Borrowers and Loan Sales: New
Evidence on the Uniqueness of Bank Loans’, Journal of Business, 76, 563–82.
DeMarzo, P., and D. Duffie (1999), ‘A Liquidity-Based Model of Security Design’,
Econometrica, 67, 65–99.
Diamond, D. (1984), ‘Financial Intermediation and Delegated Monitoring’, Review of
Economic Studies, 51, 393–414.
Diamond, D., and P. Dybvig (1983), ‘Bank Runs, Deposit Insurance and Liquidity’,
Journal of Political Economy, 91, 401–19.
Diamond, D., and R. G. Rajan (2001), ‘Liquidity Risk, Liquidity Creation and Financial
Fragility: A Theory of Banking’, Journal of Political Economy, 109, 287–327.
Fama, E. F. (1985), ‘What’s Different about Banks?’, Journal of Monetary Economics, 15,
5–29.
Fisher, I. (1936), 100% Money. New York: Adelphi Company.
Freixas, X., and J.-C. Rochet (1997), Microeconomics of Banking. Cambridge, Ma: The
MIT Press.
Friedman, M. (1960), A Program for Monetary Stability. New York: Fordham University
Press.
Friedman, M., and A. Schwartz (1963), A Monetary History of the United States, 1867–
1960. Princeton, NJ: Princeton University Press.
Gorton, G., and G. Pennacchi (1995), ‘Bank and Loan Sales: Marketing Nonmarketable
Assets’, Journal of Monetary Economics, 35, 389–411.
Hart, A. G. (1935), ‘The ‘‘Chicago Plan’’ of Banking Reform’, Review of Economic Studies,
2, 104–16.
James, C. (1987), ‘Some Evidence on the Uniqueness of Bank Loans’, Journal of
Financial Economics, 19, 217–35.
James, C., and D. Smith (2000), ‘Are Banks Still Special? New Evidence on their Role in
the Capital-Raising Process’, Journal of Applied Corporate Finance, 13, 52–63.
Jones, D. (2000), ‘Emerging Problems with the Basel Capital Accord: Regulatory
Capital Arbitrage and Related Issues’, Journal of Banking and Finance, 24, 35–58.
Kaminsky, G., and C. Reinhart (1999), ‘The Twin Crises: The Causes of Banking and
Balance-of-Payment Problems’, American Economic Review, 89, 473–500.
Kane, E. J., and B. G. Malkiel (1965), ‘Bank Portfolio Allocation, Deposit Variability,
and the Availability Doctrine’, Quarterly Journal of Economics, 79, 257–61.
Kashyap, A., R. Rajan, and J. Stein (2002), ‘Banks as Liquidity Providers: An
Explanation for the Co-Existence of Lending and Deposit-Taking’, Journal of
Finance, 57, 33–73.
Litan, R. (1987), What should Banks Do? Washington, DC: The Brookings Institution.
Miller, M. (1995), ‘Do the M&M Propositions Apply to Banks?’, Journal of Banking and
Finance, 19, 483–9.
Miller, M. (1998), ‘The Current Southeast Asia Financial Crisis’, Pacific-Basin Finance
Journal, 6, 225–33.
Mingo, J. J. (2000), ‘Policy Implications of the Federal Reserve Study of Credit Risk
Models at Major US Banking Institutions’, Journal of Banking and Finance, 24, 15–33.
Mints, L. W. (1950), Monetary Policy for a Competitive Society. New York: McGraw-Hill.
Ongena, S., and D. C. Smith (2000), ‘Bank Relationships: A Review’, in S. A. Zenios
and P. Harker (eds), Performance of Financial Institutions, 221–58, Cambridge, UK:
Cambridge University Press.
r 2009 The Author

Pennacchi, G. (1988), ‘Loan Sales and the Cost of Bank Capital’, Journal of Finance, 43,
375–96.
Santomero, A. M. (1984), ‘Modeling the Banking Firm’, Journal of Money, Credit and
Banking, 16, 576–602.
Simons, H. C. (1948), Economic Policy for a Free Society. Chicago: University of Chicago
Press.
Tobin, J. (1986), ‘Financial Innovation and Deregulation in Perspective’, in Financial
Innovation and Monetary Policy: Asia and the West Proceedings of the Second
International Conference Held by the Institute for Monetary and Economic Studies of the
Bank of Japan, Y. Suzuki and H. Yomo (eds), 19–29. Tokyo: University of Tokyo Press.
Distributed by Columbia University Press, New York.
r 2009 The Author


Chen 2009

Uploaded by

Copyright:

Available Formats

Chen 2009

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chen 2009

Uploaded by

Copyright:

Available Formats

International Review of Finance, 9:3, 2009: pp.

Banking Reforms for the 21st

Although bank loans themselves are somewhat illiquid because of private

countries, from Northern Europe to Latin America to Southeast Asia.1 None of

r 2009 The Author

regulations and bailouts (Calomiris 2000). Fisher’s proposal was ultimately

r 2009 The Author

r 2009 The Author

r 2009 The Author

Deposits with the Fed Transaction accounts

Loans Non-guaranteed deposits

Other loans Other debt and equity

Figure 1 Irving Fisher’s Proposal.

r 2009 The Author

Interest bearing deposits Transaction accounts

Tranches of risk class A Deposits of risk class A

Risky Tranches of risk class B Deposits of risk class B Non-

Figure 2 100% Reserve Banking without Deposit Insurance.

r 2009 The Author

A government deposit insurance, whose cost is borne by account holders, may

III. FEASIBILITY OF A 100% RESERVE BANKING SCHEME WITH

r 2009 The Author

Table 1 US commercial bank (FDIC) loan portfolio

Total 5,981,184 6,626,157

Notional amount of derivatives 132,168,588 164,767,281

10 Failure to follow this time-honored procedure (since the inception of mortgage-backed

r 2009 The Author

Traditional New Financial Intermediaries Final Funding

Smaller MBS, CMBS

Figure 3 Liquefaction of Bank Loans and Final Funding Sources.

A. Growth of the institutional loan market

r 2009 The Author

information so widely available on those loans and their obligors, the

r 2009 The Author

Swap Counterparty Payment

Credit Protection Payments

Threshold Realized losses

Figure 4 Credit Default Swap (CDS) on a Portfolio of Loans.

C. CDO and CLO

r 2009 The Author

Asset Manager Trustee Obligor Obligor

Sample CLO Transaction

Figure 5 A Typical Collateralized Loan Obligation.

r 2009 The Author

i. Typical Bank CLO and their incentive alignment mechanisms

14 We thank Greg Duffee for pointing to us this example.

r 2009 The Author

(Source: Moody’s, S&P, Fitch)

Figure 6 Example of a Collateralized Loan Obligation (CLO).

r 2009 The Author

CLO, financial innovations combined with an independent rating agent, who

r 2009 The Author

GLOBE-R 2000-1 Structure

Principal and interest on class A notes

Figure 7 A Synthetic Collateralized Loan Obligation (CLO).

r 2009 The Author

(Source: Merrill Lynch)

Figure 8 Example of a Collateralized Loan Obligation (CLO) with Undrawn Credit

‘Excess spread’ as a dynamic incentive alignment mechanism. In many CLO