Journal of Economic Literature 2014, 52(1), 197-210

203

Myerson: Rethinking the Principles of Bank Regulation

To understand why this role of monitor 

and lender of last resort has been filled by 

a publicly controlled central bank, it may 

be useful to review the history of America’s 

banking crisis in 1907. This was the last such 

crisis before the Federal Reserve System was 

created to serve as America’s central bank. 

When a series of bank failures threatened to 

cause a loss of public confidence in the entire 

banking system, somebody had to do the 

hard work of providing credible information 

to the market about which banks were still 

solvent. In America in 1907, the monitor and 

lender of last resort was J. P. Morgan himself. 

Working with his expert staff, sometimes 

late into the night at his home in New York, 

Morgan reviewed the accounts of threatened 

banks and led a consortium of investors to 

supply liquidity to the banks that he found 

to be healthy. With his leadership, the extent 

of the panic was contained (see Bruner and 

Carr 2007). In the same period, however, 

J. P. Morgan was often accused of using his 

great financial influence to create monopo-

listic cartels. Whether such allegations were 

true or not, his central power over channels 

of credit throughout the economy certainly 

had potential to be used for such purposes, 

against the interest of the consuming public. 

Thus, the role of monitor and lender of last 

resort should be recognized as a vital natu-

ral monopoly, maintaining costly expertise to 

provide public information, but with monop-

oly power that should be publicly controlled. 

4.  Clear Accounting Is of the Essence

The key question of bank regulation is 

how to define the amounts of equity capi-

tal that banks should be required to have. 

Since the 1980s, central banks around the 

world have formulated their regulatory 

rules under the coordinating guidance of an 

international committee on banking super-

vision based in Basel, Switzerland. Admati 

and Hellwig are deeply critical of the 

 general framework that this Basel process 

has generated. Martin Hellwig was actually 

a professor in Basel for several years while 

international central bankers were meeting 

there to develop their principles of bank 

regulation, but the Basel Committee on 

Banking Supervision made very little use of 

outside professional advice from any social 

scientists in this period (see Goodhart 2011, 

p. 572). Before 2008, the process of defin-

ing standards of bank regulation also did 

not get much attention from most econo-

mists (including the author of this review), 

but the banking industry was always paying 

attention. Admati and Hellwig’s critique 

should raise concerns that the whole Basel-

endorsed framework of bank regulation 

may need fundamental reconstruction.

First and foremost, we should recognize 

that the fundamental problem of finance is 

about people having different information. 

Millions of people entrust their savings to 

bankers and other financial intermediaries 

who have better information about where 

the money should go than the average inves-

tor has. The whole system depends on moni-

toring to maintain investors’ trust in their 

financial intermediaries. Thus, measurement 

is of the essence here.

In the financial report of any private cor-

poration, the values of all assets that the 

corporation owns are listed on the left side 

of the balance sheet, and the values of debts 

owed by the corporation are listed as liabili-

ties on the right side. The estimated book 

value of the capital or equity that belongs 

to the owners of the corporation is then 

computed to be the total value of all assets 

minus the total value of all debt obliga-

tions, and it is listed also on the right side 

as the last liability, so that the balance sheet 

balances. The accounting profession has 

defined standards for estimating the values 

of all these assets and liabilities, but ulti-

mately the value of capital that is computed 

from these accounting estimates must be 

---

Journal of Economic Literature, Vol. LII (March 2014)

204

recognized as only an estimate of the value 

of the owners’ equity.

3

For any publicly traded corporation, the 

market also provides an estimate of this 

equity value, which can be computed by mul-

tiplying the current price per share of stock in 

the corporation by the total number of shares 

outstanding. With the passions of a dynamic 

stock market, one cannot say that the mar-

ket’s values are necessarily more accurate 

than the accountants’ values, but one may 

suggest that a low value of either should be 

cause for concern by bank regulators.

Given our best measures of equity, we 

must then face the central question of how to 

determine how much equity is enough. From 

the Modigliani–Miller argument, we may 

infer that the total cost of credit from a bank 

should be relatively insensitive to its fraction 

of equity financing within some broad range 

where its chance of bankruptcy is small, but 

effective regulation requires that a bright 

regulatory line must be drawn somewhere.

Of course, equity requirements can 

only be defined in proportion to other ele-

ments of the bank’s corporate balance sheet. 

Thus, formulas are generally constructed to 

express equity requirements as a fraction of 

the bank’s assets. However, there is some-

thing fundamentally misleading about put-

ting the question in these terms, because a 

bank does not need equity for its assets; it 

needs equity for its liabilities, to make them 

credible to its creditors. In particular, banks 

3 

Accounting for regulatory purposes might appropri-

ately evaluate an asset differently when the asset that has 

been assigned to a creditor as collateral for a loan. The 

purpose of offering an asset as collateral in a secured loan 

or repurchase agreement is to achieve better terms on the 

loan by giving its creditors a claim on the asset ahead of 

depositors, in the event of default. Regulators whose job 

is to protect depositors might reasonably take the value of 

the secured loan as the creditors’ estimate of the expected 

value of the asset in the event of the bank’s failure.  

Applying this standard to the valuation of pledged assets 

in regulatory accounting would exclude from regulatory 

capital any benefits that the bank might expect from such 

transactions that put other creditors ahead of depositors.

need capital regulation where other corpo-

rations do not because banks issue deposits, 

which are debts on the liability side of their 

balance sheets. Of course, any formula that 

specifies required equity for a bank as some 

fraction X of its total assets is equivalent to 

a formula that specifies required equity as a 

fraction X

/(1−X) of its total debt liabilities. 

Under either approach, however, the equity 

requirement would depend on a calculation 

either of total assets or of total debt liabilities.

Unfortunately, some accounting conven-

tions may allow closely linked assets and 

liabilities to be netted out, so that their gross 

amounts do not appear on either side of the 

balance sheet. Consider, for example, two 

banks that enter into an interest rate swap that 

is equivalent to each bank selling the other a 

five-year bond that has a value of $1 million 

now, but with one bond paying a fixed inter-

est rate, and the other bond paying an inter-

est rate that will vary with short-term interest 

rates in the future. Banks may use such swaps 

to transfer interest-rate risks among them-

selves at a mutually agreeable price. But how 

should these swaps be listed on the balance 

sheet? A conservative accounting standard 

might list each bank as having added $1 mil-

lion to both its assets and its debt liabilities 

by the transaction. Under such accounting, 

the swap would have no effect on either 

bank’s equity value, but it would increase 

each bank’s required equity by the factor X 

(if we think in terms of assets) or X

/(1−X) (if 

we think in terms of debt liabilities). On the 

other hand, some might argue instead that 

the assets and liabilities that each bank has 

gotten in this swap are so similar (both are 

five-year bonds with the same counterparty) 

that they should simply be cancelled out on 

the bank’s balance sheet, so that the swap 

would have no effect on the required equity. 

But such swaps do entail potential liabilities 

that should not be ignored.

In fact, different accounting standards 

make significantly different judgments on 

---