PROJECT REPORT ON

NARROW BANKING
THE PARTIAL FULLFILLMENT OF THE DEGREE
AWARDED AT

B.COM (BANKING AND INSURANCE)

SEMESTER V

SUBMITED TO
UNIVERSITY OF MUMBAI
FOR THE ACADEMIC YEAR 2016-17
SUBMITTED BY
NAME: AKSHAY MANJREKAR
ROLL NO. 51

VIVA COLLEGE OF ARTS, COMMERCE AND SCIENCE

VIRAR (WEST)
401303

ACKNOWLEDGEMENT

I Akshay Manjrekar of VIVA college pursuing my B.COM- (Banking and Insurance),

would like to pay the credits, for the all those who helped in the making this project.

The 1st in accomplishment of this project is our principal Dr. R. D. Bhagat, VicePrincipal professor Prajakta Paranjape, course Co-ordinate Professor Roshni Nagar and Guide
Professor VASANTHI MISS and Teaching and non-teaching staff of VIVA
College.

I would also like to thank all my college friends those who influenced my project in order
To achieve the desired result.

DECLARATION

I hereby declare that the project NARROW BANKING is an original work prepair by me and is
being submitted to UNIVERCITY OF MUMBAI in partial fulfilment of the degree that B.com
(Banking and Insurance) for academy year 2016-17

Place

Name

Date:

Signature

INDEX

Chapter

Particulars
INTRODUCTION

Narrow banking
Narrow banks in history
Narrow banks reform proposal
Analysis of Narrow banks reforms
Evaluation of Narrow banks reforms
Modern Narrow banking
Potential consequences of narrow banking

Page no

Narrow banking versus universal banking

Narrow banking substitues deposit insurance
Policy discussion
conclusion

1. INTRODUCTION
Narrow banks have existed for hundreds of years and continue to operate today. Many
prominent economists and policy makers believe that narrow banks should play a greater role
in a reformed financial system. This review discusses the history of narrow banks, reform
proposals involving narrow banks, and theory and empirical evidence regarding whether
narrow banks should play a more prominent role in the financial system.
A narrow bank is a financial institution that issues demandable liabilities and invests
in assets that have little or no nominal interest rate and credit risk. The following financial
intermediaries are examples of narrow bank.

100% Reserve Bank (RB): Assets are high-powered money in the form of currency or
central bank reserves. Liabilities are noninterest-bearing, demandable deposits issued
in an amount equal to or less than the reserves.

Treasury Money Market Mutual Fund (TMMMF): Assets are Treasury bills or shortterm investments collateralized by Treasury bills (i.e., repurchase agreements).
Liabilities are demandable equity shares having a proportional claim on the assets.

Prime Money Market Mutual Fund (PMMMF): Assets are Treasury bills and shortterm Federal agency securities, short-term bank certificates of deposits, bankers
acceptances, highly rated commercial paper, and repurchase agreements backed by
low-risk collateral. Liabilities are demandable equity shares having a proportional

claim on the asset

Collateralized Demand Deposit Bank (CDDB): Assets include low-credit- and
interest-rate-risk money market instruments. Liabilities are demandable deposits that
have a secured claim on the money market instruments and are issued in an amount
equal to or less than the money market instruments.

Utility Bank (UB): Similar to a CDDB but collateral can include retail loans to
consumers and small businesses in addition to money market instruments.

In the above sequence of examples, asset portfolios are increasingly less restrictive in terms of
their credit- and interest-rate risks. Another difference in these examples is the composition of
liabilities. Some narrow banks issue only equity shares (TMMMF and PMMMF). Such a
liability structure limits the banks assets to cash and marketable securities for which a net
asset value (NAV) could be readily computed to buy and redeem equity shares. An alternative
liability structure (CDDB and UB) is to issue both demandable deposits and nonredeemable
equity capital. Under this structure, the bank is permitted to set the interest rate on deposits, as
opposed to having it determined by the returns on the banks assets, as in the case of the allequity narrow bank. Gorton & Pennacchi (1993) note the freedom to set deposit rates allows
banks to exercise greater market power and also to invest in some nonmarketable assets
because a NAV is not required to redeem deposits.

Narrow banking
Narrow banking is often described as a new Glass-Steagall. The Glass-Steagall
Act, passed in the US in 1933, required the separation of investment and commercial
banking. Its most famous consequence was the division of the House of Morgan into a
commercial bank, J P Morgan, and an investment bank, Morgan Stanley. The sentiment
behind the expression a new Glass- Steagall the separation of utility from casino
banking remains entirely appropriate, but that specific measure would not now achieve
the desired purpose.
The Glass-Steagall Act had been outdated by the increased scope and complexity
of financial markets well before its final repeal in 1999. Although few of the activities
which have led to the failures of predominantly retail banks in 2007-8 would fall within the
scope of retail banking as traditionally conceived, many of them could properly be
classified as commercial rather than investment banking. It is in proprietary trading that
conglomerate banks have made catastrophic losses, but these activities differ only in scale
and motivation, not inherent nature, from the treasury operations which a commercial
bank might properly undertake in its normal day to day business.
The description below is an illustration of how narrow banking might be defined, intended to
give sufficient detail to permit discussion of the specifics of how the proposal might be
implemented effectively and its advantages and disadvantages.

Narrow banking implies the creation of banking institutions focussed on the traditional
functions that the financial system offers to the non-financial economy
- payments systems (national and international), for institutions of all sizes
- deposit taking, from individuals and small and medium-sized enterprises.
Only narrow banks specialising in these activities could describe themselves as banks.
Only narrow banks could take deposits from the general public (deposits of less than a
minimum amount, say 50,000). Only narrow banks could access the principal payments
systems (CHAPS or BACS), or qualify for deposit protection.
Narrow banks might (but need not) engage in consumer lending, lend on
mortgage, and lend to businesses, but would not enjoy a monopoly of these functions.
Narrow banks could be subsidiaries of other companies (including financial holding
companies) and initially generally would be. Thus Barclays Bank might be the narrow
banking subsidiary of the Barclays Financial Group. Over time, it is likely that while
some narrow banks would be members of a group of financial companies, some would
be members of other groups, especially retailing groups. Others still would be stand
alone institutions.
In a free market, narrow banking would have emerged spontaneously and
immediately. As a result of recent history, depositors would strongly favour conservative,
transparent institutions which eschewed complex financial instruments and demonstrated
comprehensible balance sheets and organisational structures. The reason this outcome has
not emerged is that government intervention has distorted the market. All savers enjoy
equal deposit protection, however risky the activities of the institution with which they
save. Those who save with large financial conglomerates enjoy the further reassurance
that comes from frequent reiteration of the slogan that these businesses are
too big too fail. The outcome of market forces has been suppressed, and the
natural outcome of market forces narrow banking - should be imposed by
regulation.
Narrow banks would be regulated, not supervised. The regulator would monitor
compliance with the rules governing narrow banks, but would not review the business
strategies of banks or take a view on whether they are well run businesses. If the regulator
doubted the ability of a narrow bank to meet its obligations, the preliminary steps of the
resolution procedure would begin. There are several ways in which the activities of narrow
banks might be constrained: explicit restriction on the range of activities of a narrow bank,
special rules governing creditor priority in liquidation, and reserve requirements.

In other regulated industries, the best approach to definition of the regulatory

regime has been the prescription of a licence. Legislation would define the obligations of
the regulator. The primary obligations would be the protection of depositors and
regulation of the prices and terms of access to payments systems. The obligations of
licensees the narrow bank and the managers of payment systems managers would be
defined in their licences.
The licence might define the regulated activities of the narrow bank deposit
taking and access to the payment system. There would be a class of activities which
would be prohibited particularly acting as issuer of securities and securities trading for
purposes other than the facilitation of narrow banking objectives. The debate between
principles and rules, which has been extensively discussed in financial services, is the
product of a
regime based on supervision, not licensing, and would disappear.
Some licence obligations would be of a general nature protection of the
interests of depositors and others would be specific the terms of access to the payment
system. The licence would be capable of amendment, but not easily utility legislation
provides for amendment by agreement or, failing that, reference to the Competition
Commission. The constitution of the regulatory authority, and its composition, would look
more like that of OFWAT than the existing FSA. Not only would the board and staff be
drawn from a wider range of backgrounds, but the specific expertise that would be valued
in board members would be regulatory rather than financial services expertise. The same
would be true, though to a lesser degree, of the senior executives of the agency. A
substantial category of activities would neither be explicitly permitted nor prohibited. But
if these activities grew to be the principal part of its business the company would cease to
be eligible for a narrow banking licence.
An objective is to prohibit narrow banks from engaging in proprietary trading. But I have
already noted that trading is hard to define, because distinguishable only by its scale from
the necessary treasury functions of any bank. This does not mean that legal restriction is
wholly ineffective. If the scope of a narrow banks wholesale market activity became
inappropriately large, a warning

to the board of the licensed institution concerned would normally have the desired
effect.
If these restrictions on the activities of retail banks had been in place before
2007, they would have limited the trading losses incurred by retail banks. These losses
might still have been large, and there might in addition have been substantial losses from
bad lending. While this form of narrow banking would provide greater protection for
depositors (and the taxpayer who stands behind the depositors) than currently exists, that
protection would still be inadequate to protect the taxpayer interest. Such restriction
would not provide sufficient protection for depositors, or have prevented the failure of
Northern Rock.
An additional measure would give retail depositors (and, in their shoes, the
deposit protection scheme) priority over general creditors in a liquidation. That
requirement has many attractions.
It almost certainly would have prevented the Northern Rock failure: if the retail deposit
base had enjoyed priority, and since the pool of securitised mortgages was not available to
creditors, the assets to which Northern Rocks unsecured creditors would have had
recourse would have been very limited. In these circumstances, it is unlikely that
wholesale markets would have extended credit to support that companys rapid expansion
(provided, of course, those markets did not gain reassurance from the belief that Northern
Rock was too
big to fail).
In this way, the measure uses market forces rather than supervisory judgment to enforce
appropriate business discipline. Northern Rock would have been obliged to maintain
substantial liquid assets to support its retail deposits, but a diversified, well capitalised bank
like HSBC would have been able to operate much as before.

NARROW BANKS IN HISTORY

Prior to the twentieth century, British and American commercial banks lent almost exclusively
for short maturities. Primarily, loans financed working capital and provided trade credit for
borrowers who were expected to obtain cash for repayment in the near future. The real bills
doctrine, also called the commercial loan theory of liquidity or credit whose origins can be
traced to John Law and Adam Smith, was the guiding theory of
bank operations during the nineteenth century. Per this doctrine, the main function of banks is
to create a sound currency by issuing notes or deposits that are backed by short- term selfliquidating loans. Loans often were bills of exchange which financed trade and were
collateralized by the goods in transit. Promissory notes were another common loan backed by a
borrowers and any cosigners or guarantors personal wealth. Both types of loans had short
maturities, typically averaging between 65 and 80 days. Bodenhorn (2000) confirms this range
of loan maturities from the records of three US banks during the 1850s and early 1860s.
Scottish banking statements of the real bills doctrine also required that banks lend at a maturity
of no more than 60 days.
Following the US Civil War, many banks with insufficient loan demand and surplus funds
invested in commercial paper, which were promissory notes issued by corporations and often
sold through commercial-paper dealers (Greef 1938). With the establishment of the Federal
Reserve System in 1913, commercial paper became especially desired because it was eligible
collateral for borrowing from the Feds Discount Window. According to Foulke (1931), prior
to the 1930s, banks and trust companies held more than 99% of commercial paper. Banks
were attracted to its high credit quality and short maturity. In contrast, banks today hold very
little commercial paper. The Federal Reserve Flow of Funds data on ownership of open
market paper estimates that, as of the fourth quarter of 2011, commercial banks, savings
institutions, and credit unions each held less than $1 billion, while money market mutual
funds (MMMFs) held $355.1 billion.

One credit service of banks that is ubiquitous today but was completely absent
from banks in the nineteenth and early-twentieth centuries was the loan commitment. In
recent years, more than 70% of business lending was from loan-commitment
drawdowns. Prior to the 1930s, banks often had long-term relationships with particular
borrowers: Banks would lend repeatedly for short terms to the same borrower. Using the
records of the Black River Bank, an apparently typical bank located in Watertown, New
York, Bodenhorn (2003) found that in 1855 the bank made 2674 loans to 978 different
borrowers, with an average loan maturity of 68.2 days. Only 3.5% of these loans were
renegotiated and extended at maturity. During the financial panic period of 1857--1858,
the banks number of borrowers declined by nearly 75%, though higher quality and
longer-relationship borrowers were more likely to continue receiving loans.
However, early banks made virtually no formal loan commitments. According to
Summers (1975), longer-term loans, term loan commitments, and lines of credit first
appeared in the 1930s. He states, Early usage of revolving credits was very limited, their
number being estimated as only 5 percent of the number of term loans outstanding in
1941. There appears to have been resistance on the part of banks to enter revolving credit
arrangements, presumably due to uncertainties involved with credit usage.
Although restricting bank assets to short-maturity, high-credit-quality loans and
investments was considered prudent banking practice, for the most part it was voluntary
and not followed by all banks. An exception where such a practice was mandated by law
occurred with the passage of the Louisiana Banking Act of 1842. The impetus for this
legislation was the economic and financial crisis that started in 1837 and led to a
3

suspension of convertibility by New Orleans banks. As detailed in Sumner (1896) and

Hammond (1957), this act required a bank to hold specie (gold) and bills of exchange and
promissory notes maturing in 90 days or less in amounts at least equal to its deposits and
notes issued. Moreover, the ratio of specie to the total of deposits plus notes had to be at
least one-third. The bank could make loans with maturities greater than 90 days, such as
mortgages, and hold real estate and other fixed assets but they must be funded with equity
capital, not deposits or notes. Thus, equity capital must exceed all the banks assets that

do not qualify to back the banks deposits and notes.

The system established by the Louisiana Act of 1842 continued until the Civil War.
Hammond (1957, p.683) states, The available evidence is that the system operated with
distinguished successAlthough the banks of New Orleans were well known throughout
the country for their strength and integrity, the law governing them was not generally
emulated. Sumner (1896, pp. 387, 389) is more enthusiastic, calling the act the most
remarkable law to regulate banks, which was produced in this period, in any StateIt
obviously proceeded from very mature study of the principles and practice of banking,
and may justly be regarded as one of the most ingenious and intelligent acts in the
history of legislation about banking. Probably it could not have been passed except at
just such a crisis in banking affairs.
In summary, prior to the early-twentieth century, many US banks functioned
similarly to narrow banks by holding primarily short-maturity assets to match their shortmaturity liabilities. Despite the several episodes of banking panics, it may be argued that
panics occurred primarily at banks that deviated from the narrow-banking ideal.
However, the typical structure of these early banks contrasts with the modern view of
banks, according to which the received wisdom is that [t]he principal function of a bank
is that of maturity transformation---coming from the fact that lenders prefer deposits to be
of a shorter maturity than borrowers, who typically require loans for longer periods
(Noeth & Sengupta 2011, p.8). Indeed, maturity transformation is the starting assumption
of many theories of banking, such as that provided by Diamond & Dybvig (1983). Yet,
historically, maturity transformation was often considered a violation of prudent banking.
The financial panic of 1907 led to greater government interventions in banking,
thereby creating fundamental changes in banking operations. One reaction to the panic
was the creation in 1910 of the US Postal Savings System, which operated as a type of
narrow bank (Jessup & Bochnak 1992). The system took individuals deposits and
invested them in Treasury securities or deposits at local banks that pledged Treasury
securities as collateral (c.f., TMMMF). The systems design contained flaws, however,
because savings interest rates and rates paid to banks on deposits were legislatively

fixed.A more important response to the 1907 panic was the establishment in 1913 of a
government lender of last resort and central bank in the form of the Federal Reserve
System. Access to the Feds Discount Window made it less costly for banks to hold longer6

term and more illiquid loans. Indeed, Friedman & Schwartz (1963) argue that the Feds
existence changed banks behavior in ways that led to more bank failures during
the early 1930s. Banks shifted to higher credit-risk loans and felt less need to lend to each
other during times of stress because that was now considered the Feds responsibility
(which the Fed failed to perform adequately).

NARROW BANK REFORM PROPOSALS

As with other proposed bank reforms, recommendations for narrow banks appear most
frequently following major financial crises. With the exception of the Louisiana Banking
Act of 1842, and possibly the U.S. Postal Savings System, proposals involving narrow
banks have not been implemented. To fully evaluate proposed narrow bank reforms, one
must consider the implications in terms of financial services that traditional banks
provided but that narrow banks would not offer. In most instances, different financial
institutions would provide such services.
An often-cited narrow bank reform was the 1933 proposal by University of
7

Chicago economists known as the Chicago Plan (Phillips 1996). Its main feature of
required that banks hold reserves (vault cash plus deposits at the Fed) equal to the amount
their demand deposits. With demand deposits secured by reserves, they would be default
free, and deposit insurance would be unnecessary. Furthermore, the M1 money supply
(currency in circulation plus demand deposits) would not be prone to contractions from
bank runs as occurred in the 1930s. Other than this 100% reserve requirement for
demandable (and checkable) accounts, a bank could continue the same lending activities
that it previously employed but loans and other assets must be funded with uninsured
savings accounts, time deposits, or equity capital. Milton Friedman (1960) also advocated
for 100% reserve banking with the proviso that interest could be paid on reserves. The
Chicago Plan can be viewed as an RB and CDDB type of narrow bank. Likely, it would
have had little impact on banks operations prior to FDIC insurance except that demand
deposits could not fund loans and investments.
Several related narrow-banking proposals surfaced in response to the failures of
thousands of US thrift institutions and commercial banks during the 1980s and early
8

1990s. Many bank failures were due to 1930s reforms that proved to be ill adapted to the
economic environment of the 1970s and 1980s. The Banking Act of 1933 allowed the
Fed to set ceilings on banks deposit interest rates, which it did as part of Regulation Q.
For four decades, these ceilings acted as a price-fixing mechanism that reduced deposit
market competition and provided market power (and charter value) to banks. But with
high inflation and market interest rates following the 1970s breakdown of the Bretton
Woods fixed-exchange rate system, Regulation Q ceilings became a hindrance to banks

as disintermediation made it difficult for banks to fund long-maturity loans, such as

fixed-rate mortgages. The beneficiaries of banks regulatory straitjackets were MMMFs,
a form of narrow bank. The success and viability of MMMFs were among the reasons
motivating bank reforms.
While Regulation Q ceilings were being phased out during the 1980s, banks also
faced increased competition for borrowers. Advances in telecommunications and
computer technology reduced information costs and gave formerly bank-dependent
borrowers access to funding via public securities markets. Greater competition reduced
banks franchise values and likely raised their incentives to take excessive risks. Models
by Merton (1978) and Marcus (1984) show that, when banks have insured deposits and
their charter value declines, they will switch from a strategy of reducing asset volatility
and leverage to one that raises them. Whether motivated by this moral hazard or a view
that other financial services were more profitable and could diversify risks, most large
commercial banks wanted Glass-Steagall underwriting restrictions lifted so that they
could enter investment banking. Several hoped to become financial conglomerates
offering one-stop financial servicing.
Litans (1987) proposal would not affect individual banks that provide only
traditional lending and deposit taking. However, it would require nontraditional banking
services, such as underwriting and proprietary trading, to be financed with uninsured,
non-deposit liabilities. Acknowledging that social welfare may be enhanced when
individual firms offer a broad range of financial services, his proposal allows firms to
combine deposit-taking and nonbanking financial services but only by forming a
financial holding company (FHC). An FHC could issue deposits only from a narrow9

bank subsidiary. The FHCs lending and investment banking services must occur in a
separate subsidiary funded only by uninsured liabilities.
Tobin (1987) proposes narrow banks called deposited currency accounts (DCAs). One
way they could be offered to investors is directly from the Federal Reserve but marketed
at private banks or postal offices. The accounts would pay interest sufficiently below the
Treasury bill rate. Alternatively, private banks could offer them with the funds invested in
reserves or Treasury bills of no more than three months maturity. In addition to DCAs,
commercial banks eligible for deposit insurance could hold only loans and

investments of short duration and limited credit risk. All other more risky activities would
need to take place in investment banks funded by uninsured debt and equity. Thus,
Tobins plan envisions a system that includes both a TMMMF (DCA) and a UB
(commercial bank).
Bryans (1988) plan is similar to Litans (1987) in which risk-free depository
institutions must hold portfolios with no significant credit risk (such as government
obligations or securities rated AA or better) and no significant interest-rate risk (such as
floating-rate risk or short-term bonds and notes). These institutions would carry deposit
insurance at low premium rates, and only they would have access to the payments
system. Higher-risk loans, investments, and investment banking activities must be carried
out in separate firms or subsidiaries funded with uninsured liabilities. Bryan suggests that
migration from the present system of risky banks could be accomplished by imposing
increasingly higher regulatory capital requirements for risky assets and activities. This
would encourage the depositories to migrate all risk off their balance sheets either by
placing them with commercial finance affiliates or by securitizing them.
Pierce (1991) offers a similar plan, calling his narrow bank a monetary service
company (MSC). It would operate as a CDDB, though all of its assets would need to have
low interest rate and credit risks similar to those of PMMMFs. The Federal Reserve
would supervise MSCs and insure their deposits. MSCs could operate as stand-alone
institutions or be part of a financial or nonfinancial holding company. All other financial
services would be performed by financial service companies (FSCs) funded by uninsured
liabilities and subject to much less regulation, though FSCs may have access to the Feds
Discount Window on an emergency basis. If FSCs were to issue high-credit quality,

short-term debt, it could be purchased by a nonaffiliated MSC. Pierce recommends

transitioning to this new system by gradually reducing de facto and de jure deposit
insurance, as well as raising capital requirements, for banks that do not convert to an
MSC structure.
Pollock (1993) believes the aforementioned narrow-banking proposals are going
in exactly the right direction, but tend needlessly to complicate the issue by proposing
separate narrow bank corporations and dubious holding company relationships. Pointing
to the Louisiana Bank of Act 1842, he advocates the simple collateralized money
approach of the CDDB. Demand deposits would be collateralized by high-quality
marketable assets under a regulation similar to SEC Rule 15c3-3 (customer protection
rule) that requires broker-dealers to hold at all times sufficient liquid assets as collateral
for their customers accounts. All other nondemandable bank liabilities would be
uninsured. Other than requiring banks to report daily their demand deposits and
collateral, regulation can be greatly reduced.

Merton & Bodie (1993, p.5) advocate the same CDDB, making collateral be
equal to 100% of transactions deposits and that collateral should be restricted to US
Treasury bills or their equivalent. Their proposal would not prohibit financial firms that
issue transactions accounts from engaging in other financial activities: Thus, our
proposal does not eliminate any opportunities for economies of scope or scale from onestop shopping for customers of financial services. Merton and Bodie deem government
insurance of savings deposits to be unnecessary because there are many available defaultfree instruments; one example is TMMMFs. They do not advocate forbidding PMMMFs

from offering check-writing, but they do think it is undesirable for these funds to
maintain a fixed NAV because investors might wrongly think they are risk free.
The 2008--2009 financial crisis generated new narrow-banking proposals.
Mervyn King, Governor of the Bank of England, supported separating the functions of
banks into those that were utility in nature from those that involved risky financial
10

activities. Financial functions are considered a utility if their uninterrupted supply is

essential and/or their efficient provision entails a natural monopoly. Provision of
payments settled by a central (bank) clearinghouse would be considered a utility. Only
the utility functions of banks would qualify for deposit insurance and be regulated. He
did not specify how narrow the activities of the UB would be, but at a minimum, UBs
could not use government-insured deposits to fund proprietary trading (the Volcker rule).
At the other extreme (see Kay 2009), UBs may be quite narrow such that they are
permitted to hold only government securities. However, a middle-ground proposal would
allow UBs to lend to consumers and small businesses, because retail loans to these
traditionally bank-dependent borrowers could constitute a utility function. Similarly, in
2011, the U.K. Independent Commission on Banking required large U.K. banks to ringfence retail and payments-related banking operations in a separate subsidiary. Activities
related to security trading and dealing would be prohibited and the subsidiarys wholesale
funding would be limited.
Kotlikoff (2010) proposes a reform that foresees a greater role for mutual funds.
Narrow banks called cash mutual funds would operate as an RB and provide payments
services. Risky lending would ultimately be financed by mutual fund investors in the
following manner: After receiving a loan application, a private bank sends it to a

government agency called the Federal Financial Authority (FFA), which verifies the
applicants creditworthiness and assigns a numerical rating to the loan. A public auction
of such loans is then held; these loans would be bought by mutual funds who issue equity
shares to investors. Presumably, unless there is a liquid market for the underlying loans,
these mutual funds would be closed-end, and liquidity would be provided to the mutual
fund investors by having their shares trade on a secondary (stock) market. Thus, by
securitizing loans and funding them with equity shares, financial institutions avoid the
direct costs of bankruptcy (as when banks finance loans that lose value).
Another recent proposal by Ricks (2011) envisions CDDBs that would be eligible
for deposit insurance and pay risk-based insurance premiums. Different from other
proposals, under this plan, the Fed would set adjustable restrictions on the risk of
CDDBs portfolios. The Fed may start by allowing the CDDBs to hold only the safest of
securities (cash reserves and Treasury bills), but later it could permit assets of increasing
risk, such as consumer and business loans, until the Fed deems there is an adequate
supply of investments to support the payments system. Moreover, if the Fed allowed
CDDBs to hold riskier portfolios, it would simultaneously raise capital requirements and
adjust risk-based deposit insurance premia. This plan would also forbid other (nonCDDB) financial firms from issuing short-term debt (deposit-like) claims, thereby
preventing them from performing maturity transformation. Thus, via regulation, Ricks
(2011) restricts deposit-like claims to CDDBs and reduces the chance that runs would
occur in the shadow banking sector.

ANALYSIS OF NARROW BANK REFORMS

Narrow bank reforms have been evaluated using different theoretical frameworks. This
section gives a critical review of several of them.
Model by Diamond and Dybvig
One branch of analysis considers narrow banking in the context of the model by Diamond
& Dybvig (1983) (hereafter, DD). DD is a model of maturity transformation. A mutual
(depositor-owned) bank is assumed to issue demand deposits that can be withdrawn after
either one or two periods, at the discretion of the individual depositor. However, the
investment that the bank makes with depositors funds pays a high return only if it is not
liquidated after one period but is held for two periods. Specifically, liquidating the
investment after one period produces a per-dollar return of
investment for two periods produces a per-dollar return of
However, the demand-deposit contract is beneficial only if the late-consuming individual
chooses not to withdraw his deposit after one period. Such an action is privately optimal if
the late consumer believes other late consumers will not withdraw. This is the no bank
run equilibrium. But there also exists a bank run equilibrium where the late consumer
finds it privately optimal to withdraw after one period if she believes other late consumers
also will withdraw. This is due to the banks sequential servicing where the first
depositors to withdraw are paid their promised deposit interest
in full until the bank fully liquidates its investment, at which time remaining depositors
receive nothing. In this case, the expected return to all depositors is only R1 because the
bank ends up liquidating all of its investment and failing.
Thus, one important insight of DD is that maturity transformation carries with it the
possibility of runs. In my view, this is a robust result that holds even for nonbank
institutions. When an institution funds a long-term investment by issuing short-term debt,
a run by the short-term debt holders is possible if selling the long-term investment incurs
liquidation or fire-sale costs or if the institution cannot access a lender of last resort.

Such a run equilibrium obtains if short-term debt holders believe other investors will not
rollover their debt. Thus, runs can be a self-filling phenomenon when investors lose
confidence.
DD further show that a deposit insurance scheme that is financed by a tax on all
individuals after the first period can eliminate the bank run equilibrium, leaving only the
good no-bank run equilibrium. Hence, with deposit insurance, the deposit contract that
provides optimal insurance against early consumption risk can be implemented. The
conclusion from DD is that insured-bank demand deposits provide optimal risk sharing
for depositors when maturity transformation is inevitable.
Jacklin (1987) analyzes the DD model in more depth and derives several new
results. First, under DDs assumptions, an optimal liquidity insurance contract can be
implemented by a firm, such as a mutual fund, that buys the investment and pays
dividends each period. All consumers would buy the firms shares at the initial date.
Those who turn out to be early consumers obtain their first-period dividends and sell (or
redeem to the fund) their shares at the market price. Late consumers use their first-period
dividends to buy shares at the market price. Notably, this equity share contract provides
the same optimal risk sharing as the demand-deposit contract but is immune from the
11

run equilibrium. In this sense, an equity-share contract, perhaps taking the form of a
mutual fund, seems preferable because it does not require government intervention to
insure deposits. The equity-share contract differs from the demand-deposit contract
because consumers trade dividends for shares. Trading can occur in a secondary market
(the case of a closed-end mutual fund) or internally (the case of an open-end mutual
fund).

Second, Jacklin (1987) further examines the DD model but with more general
consumer preferences. In this extended setting, differences emerge between demand
deposits and equity-share contracts in terms of their liquidity insurance. For some
parameter values, both the demand deposits and the equity-share contracts achieve
optimal risk sharing; that is, they both provide the best liquidity insurance. However, for
other parameter values, only the demand-deposit contract achieves optimal liquidity
insurance. Thus, one may conclude that the demand-deposit contract has superior risk
sharing.
Focusing on this second result, Freixas & Rochet (2008, p. 223) use the DDJacklin framework to discuss how narrow banking may compare to the demand-deposit
contract. However, they note that there are several interpretations of what constitutes a
narrow bank. One extreme assumes that a narrow bank must have sufficient liquidity
after one period to meet all possible withdrawals, even if there were a run. This implies
that the entire investment must be liquidated after the first period, giving all consumers a
Return of R1 , the same expected return as in the bank run equilibrium and worse than even
autarky. This is Wallaces (1996) interpretation of a narrow bank and leads to his
conclusion that imposing narrow banking would destroy the banking system because
individuals would prefer autarky. A second interpretation of a narrow bank is to assume
that it liquates only what is needed to meet withdrawals by early consumers and liquates
the remainder only if a bank run emerges. This leads to the autarkic equilibrium. Finally,
a third interpretation of a narrow bank assumes that it is a dividend-paying mutual fund;
that is, the equity share contract. On the basis of Jacklins second result, it can be argued
that, even in this case, risk sharing is weakly dominated by the optimal demand-deposit
contract. Thus, insured demand deposits may be preferred to narrow banking.
However, a third result from Jacklin (1987) casts much doubt on the validity of
this conclusion. He shows that the liquidity insurance benefits of a demand-deposit
contract are extremely fragile to a change of assumptions that is arguably most realistic
for a modern financial system. The optimal demand-deposit contract is viable only if
trading after the first period is ruled out. To see this, suppose we begin from the insured
demand-deposit (no bank run) equilibrium. Then, at the initial date, a deviant firm
forms, holding the investment and issuing a two-period, zero-coupon bond that pays a

R
return of at maturity. Also suppose that there is a secondary market for trading thisbond
after the first period. Then Jacklin (1987) shows that an individual would be privately
better off by initially investing in this bond rather than putting his savings in a demand
deposit at the bank: The market price of the bond after one period exceeds the consumption
available from withdrawing demand deposits, whereas the bonds return after two periods
exceeds consumption available from holding the demand deposit for two periods. Hence,
when a simple bond market opens, demand deposits that provide optimal liquidity
insurance are no longer viable.
Thus, in a financial system that includes securities trading, it is highly
questionable whether demand deposits can really provide the liquidity insurance
envisioned by DD. von Thadden (1998, 2002) comes to this basic conclusion after
analyzing this issue in depth using a richer model. Consequently, the argument that a
maturity-mismatched, demand-deposit-issuing bank dominates a narrow bank on
liquidity-insurance grounds seems unconvincing.
4.2. Model by Kashyap, Rajan, and Stein
Another theory that has implications for narrow bank reforms is by provided by Kashyap,
Rajan & Stein (2002) (hereafter, KRS). Their model predicts that it is efficient for banks
to provide liquidity simultaneously to borrowing firms in the form of loan commitments
and to savers in the form of demand deposits. The reason is that loan commitments (or
lines of credit) and demand deposits are similar cash-management services. By providing
them together, a bank diversifies cash inflows and outflows, thereby conserving the liquid
assets it needs to hold to support both types of transactions. If holding liquid assets is
socially costly, then banks would minimize this cost by providing these services together.
One implication of this theory, which KRS show is supported by empirical
evidence, is that banks with relatively high proportions of transactions deposits tend to
have high proportions of loan commitments. Another implication is that the synergistic
benefit of combining loan commitments with deposits is greatest the lower the correlation
is between deposit withdrawals and commitment drawdowns. Gatev & Strahan (2006)
provide evidence on this implication by analyzing bank behavior during episodes of
financial market illiquidity, where illiquidity is measured by the commercial paper--Treasury bill spread. Using data from bank balance sheets and market interest rates from

1988 to 2002, they provide a number of convincing tests in support of the condition that
both loans and deposits tend to respond positively to an illiquidity shock.
The KRS theory and the empirical work that supports it cast doubt on the efficacy
of narrow-banking reforms. If banks are broken up into separate lending (finance
company) and deposit-taking (narrow bank) subsidiaries, then this synergy between loan
commitments and transactions deposits would be lost. As KRS (p. 34) state, if there is a
real synergy, a forced switch to narrow banking could lead to large inefficiencies.
Pennacchi (2006) re-examines the KRS theory and its empirical evidence by
asking whether its synergy is derived either from the natural lending and deposit-taking
structure of a bank or from a government safety net in the form of FDIC insurance. First,
it is argued that the KRS theory implicitly incorporates deposit insurance. Their model
assumes that a financial intermediarys cost of nondeposit debt includes an adverseselection premium that rises with the amount of debt issued to meet loan-commitment
drawdowns. However, this adverse-selection premium does not affect bank deposits. It
would seem that this asymmetric treatment of debt and deposits must be justified by an
14

implicit assumption that deposits are government insured but nondeposit debt is not.

This assumption leads to a bias toward funding loan-commitment drawdowns with

deposits rather than nondeposit debt.
Second, in the era before FDIC deposit insurance, the synergistic positive
correlation of bank loans and deposits was absent during market illiquidity shocks. Using
data on US commercial banks from 1920 to 1933 and from 1988 to 2004, Pennacchi
(2006) conducts tests similar to those run by Gatev & Strahan (2006). For both periods,
the data come from the Federal Reserves sample of weekly reporting banks. These
banks tend mainly to be the largest ones. The tests confirm the result by Gatev and
Strahan that illiquidity shocks during the 1988--2004 period led to simultaneous increases

in banks loans and deposits, particularly in large time deposits. However, repeating the
tests for the pre-FDIC 1920--1933 era shows that an illiquidity shock led to a decline in
banks loans and an insignificant reduction in deposits.
The difference in results between the two periods may not be surprising once one
accounts for deposit insurance. Post-FDIC, depositors viewed banks as a safe haven, so
that when there was stress in financial markets (an illiquidity shock), investors moved
their funds into banks (particularly institutional investors in the form of large time
deposits). Banks could then use this deposit inflow to satisfy loan-commitment
drawdowns, explaining the simultaneous increase in loan growth.
In contrast, pre-FDIC, there was no flight to quality by investors into bank
deposits because investors viewed deposits as default risky. Banks realized this and did
not make loan commitments. At times of market stress, banks decreased their lending,
most likely because they also perceived a decline in borrowers creditworthiness.
Pennacchi (2006) presents an additional test that asks whether uninsured MMMFs
perform similarly to pre-FDIC uninsured banks or to post-FDIC insured banks. This
question is potentially important for narrow-banking reforms because it indicates how
stable narrow banks may be during an illiquidity shock and whether they could continue
to channel credit to borrowers, say, via their purchases of commercial paper. Using data
on MMMF flows from 1975 to 2004, this test shows that, when illiquidity shocks
occurred, funds flowed into both retail and institutional MMMFs. Apparently during this
period, investors viewed MMMFs similar to banks in the sense that they were safe
havens during times of stress. This behavior is consistent with the findings by Gorton &
Pennacchi (1993), who examine MMMF flows from 1986 to 1991 when there were 11

different commercial-paper defaults. They find no significant declines in MMMF assets

following these defaults.
In related work, Miles (2001) examines inflows into large banks, small banks, and
MMMFs in response to a monetary tightening as proxied by a rise in the federal funds rate.
From tests performed using IMF data from 1974 to 1999, he concludes that investors
perceived MMMFs as safer than commercial banks and certainly less risky than smaller
depository institutions. Moreover, he finds that, during a monetary tightening, MMMFs
increased their purchases of commercial paper. This result is consistent with Kashyap, Stein
& Wilcox (1993), who find that firms shift from bank loans to commercial paper during a
monetary tightening.
Investors view of MMMFs as a safe haven changed dramatically following the
September 15, 2008 bankruptcy by Lehman Brothers. The management of the Reserve
Primary (money market mutual) Fund, which held a large amount of Lehmans commercial
paper, chose not to support the funds $1 NAV, leading to a breaking of the buck.

15

Wermers (2010) shows that the main outflows from PMMMFs were by institutional, rather
than retail, investors, and institutional investors tended to reinvest their funds in TMMMFs
within the same mutual fund complex. Thus, it appears that investors still considered
TMMMFs, but not MMMFs that held privately issued securities, to be safe. The outflow of
funds from PMMMFs during the 2008 crisis led the
SEC to impose greater safeguards on the credit quality, maturity, and liquidity of MMMF
assets (SEC Release No. IC-29132). Also, there are proposals to require a floating NAV for
MMMFs (for example, see Squam Lake Group 2011).
Although 2008 marked a change in investors views on the safety of PMMMFs, it is
important to note that during this especially severe crisis period investors also changed their
attitude with regard to commercial-bank deposits. Cornett et al. (2011) show that, following
the Lehman Brothers bankruptcy, banks lost large (uninsured) wholesale time deposits but
gained small (insured) retail time deposits. This is counter to the prior evidence in Gatev &
Strahan (2006) and Pennacchi (2006), which found that banks gained large time deposits
when there was a liquidity shock.
Thus, during the especially severe 2008 crisis, investors behaved similarly with
respect to banks and MMMFs. They withdrew funds from investments that were not fully
backed by the full faith and credit of the federal government (uninsured large time deposits
and PMMMFs) to ones that were (insured retail deposits and TMMMFs). The

US government reacted by expanding guarantees against default to both banks and

MMMFs. Notably, after these new guarantees were put in place in October 2008, 396 insured
banks and thrifts failed over the subsequent three years at a cost to the FDIC of over $64 billion.

16

In comparison, there were zero claims under the US Treasurys Temporary Guarantee Program for
Money Market Funds, which ended after one year. Since the demise of the Reserve Primary Fund,
no other MMMF has reduced its $1
NAV.

17

Also related to the KRS model, Mester, Nakamura & Renault (2007) point to another synergy between
lending and deposit taking that relates to credit information that a lender gains by monitoring a
borrowers transaction account activity. They show that information on a firms checking account
activity can help a lender to monitor the firms accounts receivables that collateralize its loan. However,
if such information is important,
it can be shared if narrow banks that provide checking accounts and finance companies that make
loans are different subsidiaries of the same holding company or if narrow banks take the form of a
CDDB.

EVALUATION OF NARROW BANKING

The concerns expressed by some on the risks raised by interbank overdrafts (see for instance Pierce,
1991; Spong, 1993; and Phillips, 1995) are overstated. In a narrow banking regime, the exchange of
fully covered deposits through the payment system would make payment settlement by netting safer
than under fractional reserve banking, since all the interbank intraday lending implicit in the netting
process would be fully collateralized (by construction): banks could not create mutual debit/credit
positions that exceeded their deposit cover, since no uncovered deposits would be allowed in the
system. Adopting a gross settlement rule would therefore add only protection against the risk of some
participants deliberately refusing to settle their outstanding debit balances, when required, or against
the (even more remote) risk of some participants losing access to the means
necessary to settle interbank payments.
As a result, capital requirements for narrow banks could be reduced substantially, the
potential recourse to the taxpayer for depositor protection would become infrequent, and the
inequitable too-large-to-fail bailout clause would be removed by making the failure The
benefits of narrow banking seem straightforward and immediately evident. First, by locking bank
assets in high-quality instruments, narrow-banking regulation would minimize banks' liquidity and
credit risks. Second, as narrow banks would be precluded from supplying loans and collateralize
deposits with high-quality assets, the confidence in
the value of their claims used to make payments could not be weakened by changes in the
value of loans. Third, with payment system access restricted to narrow banks, payments would be
fully secure since payment system participants would be protected against liquidity, credit, and
settlement risksof large narrow banks less likely. There would thus be a much lesser need for
subjecting narrow banks to special regulation and supervision (Bruni, 1995; Thomas, 2000).
Also, since
narrow banks would be protected from nonbank activities, a broader range of activities and a
wider ownership structure might be possible for their nonbank affiliates than under current
banking regulations in many countries (Spong, 1993).
Other important benefits are associated with narrow banking. Short of a (socially
costly) deposit insurance mechanism, the availability of narrow-bank deposits would
eliminate any discrimination between well-informed and uninformed depositors, and it
would instead leave the investors free to choose among alternative asset risk/return
configurations on the basis on non-subsidized terms, while it would protect all depositors,
reduce moral hazard, and prevent expensive bank runs. As well, the delegation of lending
decisions to uninsured and market-disciplined institutions would halt the inequitable practice,
induced by deposit insurance, of granting equal access to funds to both weak and sound
lenders.
Furthermore, a narrow-banking regime would afford greater shock resiliency to the
whole financial system: a failure of the market to elicit enough sound behavior from
nonbanks would still leave the economy's monetary sector unexposed to shocks. While the
market would/should eventually punish (ex post) the untoward behavior from individual
institutions or investors, both money and the payment system would be unaffected by such
behavior.
On the developmental side, narrow banking is expected to spur the relevant financial
system structural changes already underway in the advanced economies. Commercial banks
having to switch to narrow-banking regulation could be expected to transfer their credit
exposures to existing or newly-established finance companies, which typically operate with
higher capital ratios and fund themselves with relatively larger volumes of long-term debt.
Banks would remove loans from the portfolio of prospective narrow banks through

securitization, packaging similar types of credit into new securities and selling them to a host
of institutional investors. Furthermore, as commercial banks would progressively move out
of the long-term stretch of the market, insurance companies, pension funds, and non
financial companies interested to assume bank-like functions would fill the gap and expand
their lending activity.
As to the viability of the narrow banking model, its advocates cite the successful
experience of the US money market mutual funds industry. The increasing demand for
mutual funds products shows the potential attractiveness of narrow-bank deposits and
transaction services.
Not the least important, at times of disorderly market conditions the industry has
proven capable to weather depositor runs (McCulloch, 1986; Kareken 1985, 1986; Phillips,
1995).
But the downsides to narrow banking do not appear to be any less substantial than the
advantages. Some of them in fact challenge the benefits just discussed. This section

assesses narrow banking against contemporary theories of banking and shows that narrow
banking regulations would dissipate the benefits associated with the specialness of
conventional banks. This section also evaluates the potential consequences of narrow
banking on finance and the economy, and estimates the effects of narrow banking on the cost
and availability of credit in the economy.
A. Narrow banking vs. banking: Insights from theory16
Strong reservations to narrow banking emerge from contrasting its notion with recent
theories of banking. In this subsection, the narrow-banking concept is evaluated against
contemporary theories of banking as a mechanism for: liquidity generation, optimal contract
design, efficient joint-production of deposit-taking and lending, and money creation.
Banks as liquidity generators
An important strand of research, following Diamond and Dybvig (1983), stress
the role of banks as insurers against liquidity shocks. In a setting where all individuals
are initially identical but learn only subsequently to have different intertemporal
consumption preferences, banks are shown to generate liquidity to help individuals who
discover to be
'"'patient" consumers to satisfy their needs. They do so by transforming illiquid assets into
liquid deposits. This is possible because the averaging out of the withdrawal demands from
a large number of depositors allows banks to stabilize their deposit base and transfer deposit
ownership without liquidating the assets. From this angle, the social benefit of banking
derives from an improvement in risk-sharing, i.e., the increased flexibility of those who have
an urgent need to withdraw their funds before the assets mature (Diamond and Dybvig,
1986).
In fact, the benefit of banking cannot be fully appreciated if the asset and the
liability side of the bank balance sheet are not considered connectedly. The benefit derives
from the banks using their stable deposit base to finance production technologies that
increase output over time. The output effect enables banks to provide a pattern of returns to
depositors which is superior to what they could obtain by holding the illiquid assets and/or a
perfectly (non interest bearing) asset such as cash.17

In such light, banks generate liquidity to depositors and simultaneously assure

patient money to producing enterprises in a way that would not be feasible without the
special intermediation of banks (a point hinted at by Gorton and Pennacchi, 1990, but
surprisingly not emphasized by Diamond and Dybvig themselves).
This crucial link between liquidity and production is explicitly recognized in
Diamond and Rajan (1998, 1999), where banks are regarded as superior devices to tie human
16

The approach here used expands on Kobayakawa and Nakamura (2000).

17

In other words, if depositors hold the illiquid assets, they may have to forego
immediate consumption needs. If they hold the perfectly liquid asset, they forego higher
future consumption possibilities capital with real (illiquid) assets, and where the sequential
service constraint ordering the way in which banks service withdrawal demands (up to when
they become illiquid) work as an incentive for bankers to behave prudently. As Wallace
(1996) elegantly demonstrates, narrow banking would break that link, thereby eliminating
the social benefit of banking.
Merton and Bodie (1993) argue that asset and liquidity transformation services are
performed whenever a collection of assets is repackaged and the resulting liabilities created
have a smaller bid-ask spread than the original assets. They conclude that, in order to
generate liquidity, banks need not invest in highly illiquid assets with Large bid-ask spreads.
Instead, they support the use of "next-nearest" asset for transformation, underpinning

narrow banking, to support safe demand deposits.

The point remains, however, that production requires patient money and involves
risks, while agents with money may not be as patient and risk-inclined to lend it to firms:
banks do provide a mechanism to reconcile both sets of preferences by generating liquidity.
Narrow banks are designed precisely not to do so.
Banks as providers of optimal contractsfor transforming liquidity and maturity
Without specially designed contracts, uninformed parties would be reluctant to
provide credit for fear of being exploited by better-informed parties. Depositors would not
lend their funds to banks, and banks would not lend their funds to firms. But while bank
loan agreements incorporate features that permit banks to protect the bank's interests,
depositors are typically less informed that bank managers and face much higher costs of
protection from acts against their interests. Demandable debt (i.e., the demand deposit
contract) is for the depositors a particularly potent and economical means of avoiding bank
exploitation. By requiring banks to make good on deposits at fixed nominal value on
demand and with low transaction costs, depositors can easily withdraw their investment in a
poorly managed bank and thus exert discipline on bank behavior (Calomiris and Kahn, 1991;
Diamond and Rajan
1998, 1999).
This property of deposits allows banks to expand the volume of resources that they
can make available to their borrowers for illiquid investments. Although alternative forms
of liabilities can be used to finance illiquid investment, they would not be as widely accepted
as deposits by uninformed investors and would thereby reduce the funds for such
investments and raise their cost. Equity is more costly than debt in a setting of asymmetric
information about the value of a bank's assets, since uninformed equity holders are reluctant
to agree to the issuance of new equity against unknown new assets and cause new issues to
depress

share values (Myers and Majluf, 1984). On the other hand, in an environment of asymmetric
information long-term debt is more costly than demandable debt to depositors as control on
banks would involve longer lags and higher transaction costs. The use of demand deposits,
therefore, occurs precisely because bank assets are of uncertain value and because
depositors are less informed than bankers (Pozdena, 1991).
Banks as efficientjoint-producers of deposit-taking and lending
The provision of lending and deposit-taking services is more efficient if processed
by the same institution (Kashyap, Rajan, and Stein, 1999). Deposits, like loan
commitments, provide liquidity on demand. Since liquidity commitments need to be
supported by abuffer stock of cash and safe securities, banks can economize on such stock
by combining the two types of services (provided that deposit and loan withdrawals are not
strongly mutually correlated). Banks can thus hold a smaller buffer than would be required
by two intermediaries offering the same services separately. Such economies allow banks to
offer liquidity services to their customers at lower prices than other intermediaries.
Saidenberg
and Strahan (1999) corroborate this theory with evidence showing that banks maintain a
very important role as reliable suppliers of liquidity to the corporate sector, especially at
times of securities market distress. Since at such time investors revert to bank deposits,
banks do not have to run down their buffer stock of liquid assets to provide liquidity to
borrower.
By separating deposit and lending services, narrow banking would suppress the
synergetic effect and generate inefficiency in the supply of liquidity to the private sector,
with the level of inefficiency varying with the type of asset portfolio allowed to narrow
banks.
Banks as creators of money
Even when banks are modeled as generators of liquidity, still they act as
intermediaries who ultimately draw (accumulated) real resources from one side of the
market, against the issue of liquid claims, and transfer them to the other side of the
market, against the issue of longer-term claims. Intermediaries presuppose the existence
of real resources, or of claims thereof, which can be deposited with them in exchange for
promissory notes. Such promissory notes are therefore claims on some output already
produced and accumulated in the economy.
But banks and bank money do more than that. As deposits represent debt claims on
banks that the public accepts as money, banks can mobilize production factors by lending to
producers newly issued debt claims. The new money is to be backed by the output coming
out of the new production financed with it, and this can only happen ex post, once the
output is produced and sold. Ultimately, the money is as good as the banks are able to select
good borrowers.
Banks stand at the inception of the production cycle, as the theory of the monetary
circuit holds (Bossone 2001a, 2001b), and conveniently create money and allocate it to the
firms that need it to carry out their activity. This process is evident in the hypothetical case
of an economy with only one bank and only one type of money, where all agents hold their
deposit accounts with the same bank. Aside from inflationary and risk considerations, the
bank would be able to create all the deposits it so wished (to the extent that depositors
would continue to accept them), since the re-depositing automaticalll following payments
would prevent the bank from running into illiquidity or insolvency.1 In a multi-bank

environment, the money-creation power of banks needs to be supported by interbank (non

reserve) credit arrangements (such as overdraft and netting facilities) to supplement the redepositing mechanism. The cost of creating money thus depends on the (explicit and
implicit) cost of interbank credit and the cost of holding higher-powered reserves against
liabilities.
In fact, even in this special case, the depositors' potential demand for convertibility would
require the bank to hold a fraction of its liabilities in the form of some higher-powered
claim.
Narrow banking would suppress the money-creation feature of ban.king. As a result,
credit to the economy would become scarcer and more costly. To see why, let's analyze how
money would be supplied under narrow banking. If the nonbanks fund their assets with

short-term non-deposit debt, overall lending to the economy can be maintained only if the
investors are willing to replace bank deposits with nonbank debt in their portfolio. But, all
else equal, this would require a higher remuneration of the nonbank debt, which would make
lending costlier and reduce the liquidity in the system (since, by regulation, the nonbank
debt cannot be used as money).
Alternatively, the nonbanks could borrow or purchase money from the central bank,
against collateral or in exchange for securities, and on-Lendit to the business sector. But the
nonbanks' cost of lending would be larger than for conventional banks since the latter can
fund their loans by creating deposits. Still, the central bank could lend uncollateralized
reserves to the nonbanks, but this would come at a risk for the central bank. In this case,
either the central bank would be capable to determine the optimal interest rate risk-premium
for each borrower or it would misprice risk, with system-wide distortions. In the end,
assuming away money-substitution phenomena, narrow banking bring the whole money
creation process back to the central bank, therefore maximizing the central bank's control of
the money supply, but it would so at a considerable efficiency loss to the economy.19
Finally, since it relies heavily on reserve money, a narrow-banking regime is
vulnerable to a serious potential failure: in the event of a net overall reserve shortage, the
nonbanks would need eligible paper to raise reserves, but they might not be able to buy or to
borrow the paper precisely because they don't have enough reserves! To be sure, the class of
eligible securities could be broadened as to allow wider access to reserve money, but in most
cases the units in the economy with surplus holdings of eligible securities would have to
either lend securities directly to the deficit units, or use their securities to raise cash from the
central bank and lend it to the deficit units, in both cases bearing the related credit risk.
Once more, there is a natural need for somebody in the system to get the power to create
liquidity (money or securities, as necessary) at a risk.20
The importance of the money-creation feature of banking survives to today's
financial market transformation and to the banks' lesser involvement in direct lending to
production. Aside from the continuing relevance of bank lending to small and medium sized
businesses, it is undoubtedly the case that in the advanced economies nonbank quasi-money
and financing products are taking increasing business shares away from banks, while
nonbanks manage to offer products which allow investors and consumers to economize on
less remunerative bank deposits. Yet, money transactions take place by deposit transfers
across bank accounts, and the acceptance by the public of nonbank products owes to these
products' convertibility into bank deposits. This presupposes the existence of banks and

For a very recent and insightful theoretical analysis of the (in)efficiencies associated with
private and public money regimes, see Azariadis, Bullard and Smith (2001).
Note that the liquidity creation in this example rests on the regulatory fiat broadening the
class of papers eligible for conversion into reserve money.

their readiness to supply deposits to refinance such products when necessary.21

Narrow banking would hamper the development of the nonbank financial sector,
contrary to what its advocates assert.
Today, banks increasingly specialize in retail services or in wholesale
businesses. Through both channels, they continue to use lending and loan commitments
to supply the economy with the money needed to effect transactions. Narrow banking,
at least in its more conservative versions, would abort those channels at a major loss to
the society.
In conclusion, as one judges from theory, forcing a syncronization of maturity
between bank assets and liabilities, and eliminating potential bank difficulties by
removing conventional banks, would dissipate the significant benefits associated with
conventional banking systems, which stem from issuing demandable deposits to
finance other than government liabilities.

Modern narrow banking

Narrow banking proposals resurfaced in the U.S. in the 1980s, when tumultuous
financial innovation and financial crisis episodes called for a reassessment of the extant
banking regulatory regime. Before the issue became the object of more extensive policy
analysis, some highly-reputed scholars voiced their su~port for 100% reserve banking, but
without giving much emphasis to institutional aspects. Various proposals have been
9

See, for instance, Black (1985), Tobin (1985), and Kareken (1985). Tobin (1987)

elaborates on his earlier proposal, moving away from narrow banking strictly and proposing
a redefinition of commercial banking that preserves the link between deposit money and
commercial lending. This idea is close to Bryan's (l99l)"core banking" model, whereby
the scope of banking is narrowed down to a core of activities where banks have a
demonstrated comparative advantage: issuance of checking, savings, and money market
deposit accounts; provision of payment, trust, and custody services; and loans to individuals,
small businesses, and medium-sized companies. A core bank, on the other hand, would not
lend to large corporations and developing countries; it would not engage in high-leveraged
transactions and in large commercial real estate projects; it would not undertake global
money market activities of large money center banks or large regional banks; and it would
not underwrite securities. According to Bryan, a core bank is a "safe place to keep your
formulated since. (Recently, even the World Bank (2001) has suggested that narrow
banking could be used in some countries as a response to crises). Proposals differ in terms
of restrictions to be placed on bank asset portfolios and feature different institutional
designs.
As regards the asset-portfolio dimension, proposals differ in the degree of restriction
placed on the types of asset that narrow banks should be permitted to hold. Proposals vary
from introducing a 100% reserve requirement that bound banks to fully back transaction
accounts with marketable short-term Treasury debt (Tobin, 1985; Kareken, 1986; Spong,
1991; Mishkin, 1999; Thomas, 2000), to requiring banks to invest (fully insured) deposits
only in high-grade securities including government paper or government-guaranteed securities
of various maturity (Litan, 1987; Herring and Litan, 1995), to allowing banks the use of
insured checkable deposits for short-term lending to consumers and businesses (against
securitized instruments), including securities issued by non-affiliated financial services
companies (Pierce, 1991).
Some proponents argue that if the demand for highly liquid, riskless transaction deposits exceeds
the supply of government paper, then the class of collateral assets should be broadened to include
a well diversified portfolio of traded short-term, high-grade corporate debt (Merton and Bodie,
1993; Spong, 1993). Such broadening, however, is controversial as it obviously reintroduces
default risk into narrow banking (Litan, cit.).10

Some advocates of the more restrictive versions of narrow banking also propose that narrow
banks marked to market the debt held in their portfolio frequently (Kareken, 1986; Mishkin,
1999). This provision would force them to adjust the value of their liabilities to that of their
assets, much as mutual funds do. As a result, while the nominal value of the transaction account
balances outstanding would not be guaranteed, the taxpayers would not
be called upon to rescue insolvent institutions.
The narrow banking proposals vary also with respect to institutional factors. Noting that
technology and bookkeeping techniques allow banks wanting to offer risky balances to evade such
regulation by switching funds across different types of accounts at high speed and low costs,
Kareken (1986) suggests that banks offering transaction balances be prevented
money", and the core banking functions he refers to are those where banks reportedly make
the overwhelming bulk of their profits.
Litan discusses the consequences of broadening the class and term-structure of the securities
available to narrow banks for investments, and the regulatory actions that would have to be
associated with those changes.
As Kareken points out, there is a clear tradeoff here between a contract of certain value but
carrying the risk of not being honored and a contract of uncertain value but which is going to
be honored with certainty. Goodhart (1988) evaluates (and supports) the idea of introducing
such type of mutual-funds money that highly diversified (financial or non-financial) firms
offering both insured depository services and lending services be transformed into
financial holding companies that engage in different activities through separate
subsidiaries.

In particular, depository functions would be carried out by subsidiaries

operating under narrow-banking restrictions, while other subsidiaries would have to be

used to extend loans funded by uninsured liabilities and equity. Only narrow banks
would have access to the payment system, and nonbank corporations would not be
allowed to hold accounts at their affiliated narrow banks.
Functional segregation (which, by the way, neither Litan nor Pierce envision for smaller

banksj':' would prevent financial holding companies from using the resources of their
depositories to bail out risky nonbank affiliates or to finance connected lending. It would
also serve to limit the high concentration of economic power in the hands of a few large
financial organizations.

Functional segregation has been advocated by market specialists,

as well, in that securitization and competition allow banks to be unbundled into their
component functions, so that each can be performed by whichever players are most
capable of delivering the best service at the best price. from issuing other types of services,
12

although he remains agnostic as to whether tills should be accomplished by splitting

banks into distinct subsidiaries or by allowing two separate industries to emerge (one

made up of narrow banks and the other, broader and more inclusive, made up of lending
companies).
Others see the functional segregation of banking from nonbanking activities as the
optimal regulatory response to the risks associated with financial deregulation and
innovation, and with the increasing power of banks. Litan and Pierce, in the two studies
cited above, propose
Mishkin (1999) proposes regulation that would permit banks to issue both insured
narrow-bank deposits collateralized by high-grade securities and uninsured deposits to
finance private-sector lending. This solution would preserve banks' economies of scope
between deposit issuance and bank lending, which would otherwise be lost under
restrictive functional segregation rules. Furthermore, this solution would leave to the
banks the decision to combine narrow banking and conventional banking activities, based
only on profit considerations.
Through sweep accounts, banks can handle large volumes of transaction balances
throughout each business day but record only a small amount at the end of the day. While
recorded balances would be insured, the unrecorded ones would be unprotected by
insurance and bear full risk to their holders.
The exemption of smaller banks recognizes that narrow banks need some minimum scale
to operate efficiently. The exemption would also help to counteract any contraction of
credit supply to small businesses, without imperiling systemic stability. As importantly,
the exemption would facilitate the access to funds from small organizations that lack direct
access to major credit markets (Spong, 1993).
As Mishkin notes, eventual reduction of insured deposits would be consistent with no
loss of efficiency since the deposits that disappear would have been issued in larger amounts
only of the because of the deposit insurance subsidy.
Recent theoretical contributions have revisited the issue of narrow banking from
the standpoint of informational efficiency. Peters and Thakor (1995) show that it is not
feasible for a single bank to provide optimal incentives for depositors to monitor asset
allocation from bank managers, and for bank managers to monitor borrowers. They thus
prescribe to
separate banks along functional lines whereby some narrow-bank types of institutions
offer
insured deposits against assets whose value are invariant to bank monitoring, while
nonbanks issue uninsured liabilities to fund riskier assets.

In a different context, Craine (1995) studies an environment with two types of

intermediaries, one investing in non-traded private-information

assets and the other

holding traded public-information

assets, and calls "banks" those intermediaries that

choose to issue insured transaction accounts. Craine shows that, when regulation
restricts the issue of transaction accounts to private-information intermediaries, an
inefficient equilibrium is achieved where these intermediaries have an incentive to overinvest in risky portfolios and earn rents on insured accounts. He further shows that,
when regulation is relaxed and all intermediaries are given access to bank status (i.e.,
they are authorized to issue insured transaction accounts), a voluntary efficient
separating equilibrium ensues whereby all public information intermediaries issue
insured accounts and do not extract rents from them, and all private-information
intermediaries fund assets with private sources.

Potential consequences of narrowbanking

Aside from the theoretical considerations above, there are important, more
operational issues concerning the impact of narrow banking on finance and the real
economy, which need discussing. These include:
The supply of high-quality assets
All narrow-banking proposals must confront the question of whether there is in
the economy enough of the instruments eligible to be used as collateral for transaction

deposits. If narrow banks were required to hold government paper only, the supply of
such paper would depend on the government's debt management strategy. To the extent
that a country runs large debt-financed deficits, the stock of public securities
outstanding might be enough to serve the economy's monetary needs.22 But this need
not be the case, especially when the country decides to cut its public deficits and reduce
its stock of debt (see Schinasi, Kramer and Smith, 2001, and International Monetary
Fund, 2001, Ch. IV).
More importantly, tying the provision of monetary services to public debt
management and assigning it to the government might not be a good policy. Similarly,
tying public debt management to monetary and payment system objectives does not look
sensible either. (By construction, such a link would become the strongest once the
flexibility afforded
As confidence in the nonbank products grows, nothing prevents the public from
accepting nonbank quasi-monies as money at some point. The issuing nonbanks would
then no longer need to rely on bank deposits to effect transactions. At that point, they
would have an incentive to start lending their own-produced money, precisely like banks
do. Add to this that, as the system shifted to the narrow-banking mode, the public might
want to economize on transaction deposits and invest elsewhere their wealth - say in
uninsured short-term paper - so that the overall stock of transaction deposits would
shrink, thus making collateralization easier.
by conventional banking as a creator of private money would be suppressed by
narrow banking, and all the stock of money available to the public would have to be
fully backed with government paper).
The alternative is to extend the admissible narrow-bank portfolios to a broader
class of assets, including private-sector securities. This, obviously, would reflect back on
the creditworthiness of the narrow banks and on the costs of monitoring more diversified
portfolios. In particular, if narrow banks were permitted to hold high-grade corporate
bonds, one would have to question whether the quality of such bonds would still be the
same in the absence of bank commitments to provide contingent liquidity to corporations
when needed.

Narrow banks and the safety nets (I)

Can narrow banks do without safety nets? Insuring narrow-bank deposits is
usually predicated for the only purpose of protecting depositors against the residual risk
of bank fraud (Mishkin, 1999). For narrow banks to be able to do without safety nets,
they must obviously be perceived by the public as being financially viable and fully
safe. I shall deal with the viability issue further below. On the safety issue, narrow
banks are clearly as good as their assets. Now, even under regulation requiring narrow
banks to hold only short-term government paper, full safety cannot be achieved in the
absence of a credible commitment from the issuing banks to convert on request into
cash, at par, all deposit holdings. To the extent that the narrow-bank collateral is not
accepted in the economy as money, there is always a chance that depositors would rush
to their banks if they perceived their collateral to
be losing value or to become illiquid, and if they fear that other depositors might do the
same at the same time.
Perceptions of less than full safety may become significant when fluctuations
in the market value of government paper are marked, and when sovereign risk is not
23
negligible.

Developing countries may suffer from both these problems. Ghosh and Saggar (1998)
discuss the significant market and capital loss risks that narrow banks would likely
confront in developing economies. In particular, they point to the many historical cases
of governments and central banks deliberately inflicting capital losses on public debt
holders through inflation, repudiation, and outright manipulation of yields. Since a large
exposure to a single borrower (i.e., the government) might not be advisable in some
countries, an alternative could be to permit narrow banks to hold foreign assets
(although this would introduce the foreign exchange risk dimension).
Narrow banks would likely take huge capital losses in the process of massive
stock selling, even if they could liquidate the collateral in a large, deep, and welldeveloped secondary market. They might still be unable to face withdrawal demands
unless they held enough extra capital.

The cost of restructuring

A practical concern on narrow banking regards the cost of breaking longstanding multifunction banks into specialized, legally, and physically separate
corporations (Benston and Kaufman, 1993). New structures would have to be built, or
old structures redesigned; employees would have to be reassigned to each organization;
and information costs to and for customers might be significant.
As a least-cost alternative, some propose that the institutions that wished to offer
depository services (even in addition to other nonbank riskier services) be simply
required to keep a 100% reserve in government securities against their transaction deposit
liabilities.26
The proponents of this option believe that such constraint would be enough to
protect transaction deposits from the other activities of the depository institution, and
would not eliminate bank scale or scope economies from "one-stop" consumer
shopping.
Although appealing, this solution must be proven to be able to raise a tight
enough "firewall" between the monetary and the non-monetary financial activities. This
is deemed necessary to prevent that narrow-bank assets were used to bail out the nonmonetary
financial activities in the event of some liquidity or solvency problems. For this
reason, narrow-banking proposals usually recommend that narrow-bank assets be
segregated and
27
unreachable.

The viability of narrow banks

How attractive is the narrow-banking business? Based on the experience of
money market mutual funds, Spong (1993) argues that, once freed from major regulatory
burdens, narrow banks should be able to offer depositors a return competitive with other
low-risk investment alternatives. Narrow banks would earn interest income from their
assets and non interest income from the fees charged on transaction services. Also,
because of their

25

As Calomiris (1999) notes, politically, the absence of de jure protection on bank

liabilities outside the narrow bank does not imply the absence of de facto protection by
the
government. Absent the government's credible commitment to not intervene to
prop up banks during a crisis, narrow banking may only end up substituting ad hoc
bailouts for explicit insurance coverage.
26

See Benston and others (1989), Merton and Bodie (1993), and Benston and Kaufman
(1993).
27

See the discussion on this point in Phillips (1995) minimal capital needs, narrow banks
could operate on low margins and still be able to earn satisfactory returns on equity.

From the opposite stance, Ely ( 1991) sees the restrictions in the range of
investment activity to result in a reduction of narrow banks' size and income. This
tendency, he contends, would be accentuated by the small bank exemption clause (see
the proposals by Pierce and Litan above), which would prompt the formation of many
small banks, especially in urban areas.28 In fact, the income losses associated with the
smaller scale might tum out
to be significant in light of the recent findings on scale economies in the banking
industry.29
Moreover, in countries where banks are not free to close branches as they see
fit, a reduction in activity would face would-be narrow ban.kswith increasing unit
operating expenses, possibly leading them to depress the interest rate paid on
deposits, and causing outflows of funds toward nonbanks.
The reduction in size anticipated by Ely is consistent with country evidence
indicating a small demand-deposit base in relation to overall banking activity. Using
1995 survey data on India, Ghosh and Saggar (1998) conclude that the imposition of
narrow banking would squeeze the country's banking sector. They also argue that
requiring banks to invest only in government paper may lower their income
substantially, even after adjusting for the gains from lower non-performing Loans, and
note that such gains could be easily
wiped out by a single interest rate shock similar to those experienced in India in the late
1990s.
Separating lending from deposit-taking activities may also cause
intermediaries to miss the efficiency gains from the joint-production of the two
services (see subsection A).
Finally, in the case of developing economies with large price volatility in the
government securities markets, it is less than certain that the after-risk rate of return
on narrow-bank portfolios would be competitive with the return on well diversified
and well managed conventional bank portfolios.
Private-sector credit availability
Opponents of narrow banking argue that not enough credit would flow to the
private sector, if commercial banks were to turn into narrow banks. Credit would
become relatively scarcer and costlier, most notably for smaller (firm and consumer)

borrowers, since non insured financial companies would be motivated to invest in

larger enterprises.
28

Considering the high standardization of the narrow-banking products and the need for
achieving economies of scale, one could make the case that the narrow-banking business
is a natural monopoly. In this case, narrow banking services could be provided directly
by the central bank, or by one single banking entity with a large geographical presence
across the country. The Postal Office could be an example of such an entity.
29

For an extensive review of the literature on scale economies in banking

and the nonbanking financial sector, see Bessone, Honohan, and Long
(2001).With the exception of Wallace (1996) and the discussion in Section
IV.A, this issue has usually been debated outside of well-defined theoretical
frameworks. The following are the typical arguments used by narrow-banking
proponents to dismiss the concern on short credit supply.
Proponents of narrow banking trust that the incentives prompted by narrow
banking would further the ongoing transformation of the financial system away from
traditional banking and into nonbanking intermediation. They hold that experience so
far shows that nonbanks in the advanced economies attract more and more business
away from banks. Thus, new entry into the market of finance companies, investment
banks, and institutional investors, as well as the increasing use of financing instruments
alternative to deposits (e.g., securitization, equities, and junks bonds), it is asserted,
should supplement the supply of credit as needed.

On the other hand, the exemption of small banks from narrow-banking

regulation would help protect the flow of credit supply to the small business sector.
However, if the cost of credit did increase overall, this would be the (explicit) price that
society would have to pay to achieve higher financial stability in a system that does not
subsidize risk taking through public guarantees (Burnham, 1991). Finally, if credit to
small borrowers remains a concern, then some form of explicit public support
mechanism would be preferable than the continued use of distortive indirect subsidy
schemes such as deposit insurance.

Narrow versus Universal Banking

Traditionally in India, short-term credit for working capital requirements were
provided by banks, while financial institutions advanced long-term finance for industrial
development. In recent times, there has been a gradual blurring of distinction between these two
sorts of entities. On the demand side, with the drying up of concessional sources of finance,
financial institutions have to perforce raise resources from the market at competitive rates. On the
other hand, with each of them making a foray into the others traditional domain of operations,
the competition for supply of funds has also intensified. This has compelled both types of
entities to fine-tune their (interest) rate strategies with even provision for sub-PLR rates. Such
moves represent the first step towards universal banking practices in the Indian financial system.
In economies affected by severe banking crises, it is only natural to ask whether it
would serve the interest of financial stability to restrict the scope of banks activities and

whether alternative institutions could provide the financial services that households and firms
demand. As has been argued, the costs and benefits of alternative financial structures should
take into account the fact that the main goal of financial sector is to have a system
characterised by high performance and stability. In systems with weak regulatory and
supervisory framework, restricting bank portfolios may lead to fewer bank failures, but not
necessarily result in greater overall financial stability-the deficiencies in the regulatory
framework may merely cause risks to be shifted from banks to non-banks. On the other hand,
universal banks allowed to carry more risky assets may be subject to larger losses, but the
diversification of their activities may lend them greater stability. However, universal banks are
often closely interconnected with the rest of the financial system, so that the potential systemic
effect of a bank failure could be greater than under narrow banking regime.
definition of safe assets that narrow banks may hold is determined by the
authorities, the temptation remains for policymakers to make safe synonymous with
Government or Government-related, which, in turn, increases the incentives of
policymakers to finance fiscal deficits with Government paper held by banks.

Narrow Banking Substitute Deposit

Insurance
Deposit insurance found its rationale in the seminal contribution of Diamond
and Dybvig (1983). However, there has been a greater recognition of the moral
hazard problem of deposit insurance in recent times. It has been suggested that
banking failures make bailouts a political necessity, which even reasonably
independent central banks may find difficult to avoid. Any bailout can only be
undertaken at the taxpayers cost as bailout invites transfer of resources from poorly
organised taxpayers to better lobbying depositors10. While deposit insurance, in
itself, reduces pressure for other forms of bailouts, it also creates moral hazard in
weaker prudence attached to lending activities of the bank. Adverse selection
effects in the form of more risky investment behaviour compound the problem.
Narrow banking proposal has sometimes been advocated on the ground that it would
limit the compulsions for deposit protection. Since narrow banking is based on
defining a class of deposits which are backed by assets that are sufficiently liquid

and safe to cover any bank run, deposit insurance no longer remains necessary to
meet the bunching of deposit withdrawals.
It seems likely that absence of deposit insurance will increase pressures for
better disclosure norms, enforce tighter supervision and regulation and eventually
compel the bank management to adopt more sound banking practices. Yet, it is not
clear whether this in itself can preclude banking failures. Banking failures are more
likely to arise in absence of deposit insurance. Notwithstanding the moral hazard
argument, limited deposit insurance may still be the best way to avoid bank runs as
it can avoid a shift to a bad equilibrium b shifting some of the incentive of monitoring
to shareholders (i.e., making the system incentive-compatible). Narrow banking
can obviate deposit insurance need to the degree it can be seen as a credible
alternative to avoiding the need for bailouts to arise in the first place. But, if narrow
banking scheme limits the size of the banking business of these weak banks to a
degree that it calls into question the viability of the intermediation process, it can
hasten the shift to a bad equilibrium. This can put increased pressure for
interventions and bailouts by the authorities. Substitution of deposit insurance by
narrow banking, which tends to lack credibility, would raise the return on deposits,
which the narrow banks have to offer to compensate for the expected risk to their
deposit in a deregulated environment. This could dampen the gains of reduced
interest outgo that narrow banking affords. As such, narrow banking proposals might
turn out to be more costly and more distortionary than deposit insurance. It would be
more costly as the bailout size and the probability of occurrence of such an event
would be larger in the absence of deposit insurance. It will also be more distortionary
as costs of deposit insurance, at least in part, will be borne by the failed institution.

POLICY DISCUSSION AND CONCLUSION

From the above analysis it seems fair to conclude that narrowing the scope of
banking would, at best, bear less than certain benefits in terms of greater financial
stability, while it would exact some heavy costs in terms of efficiency and credit
availability.
Narrow banking would severe the link between liquidity, money, credit and
economic activity, which banking has a natural incentive to establish efficiently (under
stable macroeconomic conditions). By suppressing bank money as an instrument to
finance
lending to the private sector, narrow banking would create serious market
incompleteness.
The consequent economic losses might generate incentives for other financial firms to
fill in the gap left by undertaking conventional banking activities. While this would
forswear the very purpose of narrow banking, it would eventually replicate the world
before narrow banking, engendering the same risks that were supposed to be removed
in the first place through limiting the scope of banks.

The economic costs of narrow banking could be particularly significant for

developing countries, where the need is vital for an efficient banking system both as an
engine of economic growth and a support for the development of a strong nonbank
financial sector. Also, in many developing countries propositions to move to narrow
banking should be resisted given the absence of a well-developed secondary market for
government securities, a highly volatile environment for securities prices, the existence
of sovereign risk,
and a non-credible government commitment not to insure deposits or financial
instruments of
large public use.
Definitely better alternatives to narrow banking would be Bryan's core banking
model (footnote 9, Section III), or those regulatory regimes that separate commercial
(short term) banking from investment (longer-term) banking activities, even within the
same financial holding companies.
There is some favor to the idea that narrow banking could be used in some
countries as a response to crises (World Bank, 2001). In particular, weak banks could
be required to operate as narrow banks with a view to fixing their balance sheets.
Whereas selective intervention on individual banks could be justified, policymakers
should be aware that the banks required to operate as narrow banks would rapidly
dissipate their banking knowledge capital.l'
While mandatory narrow-banking regulations should be rejected, nothing should stand

in the way of individual institutions wanting to offer narrow-banking services to their

customers on a voluntary basis, or to create narrow-bank subsidiaries that would be
segregated from other businesses within the same bank holding companies.
An efficient, free-choice regulatory solution would be that prospected by
Mishkin (1999) and discussed in Section III.B. While not suppressing the risks
inherent in conventional banking, such solution would retain the money-creation power
of conventional banks, avail risk-averse investors of a full risk-proof money instrument,
and let the financial institutions and their customers free to opt for conventional and/or
narrow banking instruments based on their own convenience.
There are two additional (not mutually exclusive) alternatives that would
improve the incentives to prudence for both banks and depositors while preserving
conventional banking. Banks could issue uninsured deposits bearing an option clause
whereby, in the event of liquidity problems, a bank could suspend deposit convertibility
up to a predetermined time interval while it liquidates its assets in an orderly fashion. In
the meantime, the bank's deposits would continue to circulate in the payment system.
To induce depositors to accept such provision, the bank would commit to pay an interest
penalty in the event it invoked the option or to pay a premium on the deposit interest
rate.32
On the other hand, banks could issue uninsured subordinated debt, as proposed
by Keehn (1989), Wall (1989), and recently by Calomiris (1997, 1999). The
subordinated debt would produce a signal to the market and the banking supervisors on
the relative risk-ness of the issuing banks. By inducing debt-holders to use information
optimally, it would strengthen the incentive to prudence inherent in the banks'
sequential service constraint and in the associated risk of runs.33

A few years ago, while visiting officially one major emerging economy where a long
record of macroeconomic instability and disproportionately high interest rates on
government securities had driven most banks to disengage from lending to the private
sector and to invest in government paper, I was told by major local bankers that a
transition to a stable environment and to less distorted financial prices would find most
banks unprepared to
return to their old banking business and to make profits by selecting good risks.
In the early nineteenth century Scottish banks adopted this solution for their notes
(England, 1991).
Evanoff and Wall (2001) show that, while some measures such as capital ratios have
no predictive power and, therefore, are not good candidates to trigger prompt corrective
actions from supervisors, subordinated-debt yield spreads (on a riskless rate) perform
better in this regard. See also the recent supportive evidence by Hancock and Kwast
(2001).