View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

John Maynard Keynes
Milton Friedman


ohn Maynard Keynes (1883–1946) is the latest in a line of great British
economists who had a profound influence on the discipline of economics.
By common consent, the line starts with Adam Smith (1723–1790), whose
Wealth of Nations (1776) is generally regarded as the founding document of
modern economics. It continues with David Ricardo (1772–1823), whose Principles of Political Economy (1817) dominated classical economics for much
of the nineteenth century, and, incidentally, provided Karl Marx with one of
his central concepts: the labor theory of value. John Stuart Mill’s (1806–1873)
Principles of Political Economy, published in the same year, 1848, as the Communist Manifesto by Marx and Engels, became the standard textbook in the
English-speaking world—and beyond—for decades. William Stanley Jevons’s
(1835–1882) Theory of Political Economy (1871) inaugurated the “marginal
revolution,” which replaced, or supplemented, emphasis on cost of production
(supply) as determining value with emphasis on utility (demand). He resolved
the classic diamond-water paradox—diamonds are a luxury, water a necessity,
yet diamonds command a higher price than water—by showing that “marginal
utility”—the utility gained from having one more unit of something—not “total
utility” plays the key role in determining price. Alfred Marshall (1842–1924),
Keynes’s own teacher, guide, and patron, dominated economics in the Englishspeaking world from the publication of the first edition of his classic, Principles
of Economics (1890), to the 1930s.

The Federal Reserve Bank of Richmond is indebted to Professor Friedman for his kind
permission to publish this article in its original English version. The article first appeared
in German translation in the volume of commentaries accompanying the facsimile edition
of John Maynard Keynes’s the General Theory of Employment, Interest and Money (1936)
published in 1989 by Verlag Wirtschaft und Finanzen GmbH, D¨ sseldorf, as part of its series
Klassiker der National¨ konomie. The generous permission of that publisher and its principal,
Mr. Michael Tochtermann, are gratefully acknowledged.

Federal Reserve Bank of Richmond Economic Quarterly Volume 83/2 Spring 1997



Federal Reserve Bank of Richmond Economic Quarterly

Keynes clearly belongs in this line. In listing “the” classic of each of these
great economists, historians will cite the General Theory as Keynes’s pathbreaking contribution. Yet, in my opinion, Keynes would belong in this line
even if the General Theory had never been published. Indeed, I am one of a
small minority of professional economists who regard his Tract on Monetary
Reform (1923), not the General Theory, as his best book in economics. Even
after sixty-five years, it is not only well worth reading but continues to have a
major influence on economic policy.

From 1908 to his death in 1946, Keynes was an active Fellow of King’s College,
Cambridge, influencing successive generations of students. For many years, he
was also Bursar of King’s College, and is credited with making it one of the
wealthiest of the Cambridge colleges. From 1911 to 1944, he was the editor
or joint-editor of the Economic Journal, at the time the leading professional
economic periodical in the English-speaking world. Simultaneously, he was
also Secretary of the Royal Economic Society.
Despite his lifelong commitment to economics, the earliest work he
completed—though not the earliest to be published—was in mathematics not
economics—A Treatise on Probability—essentially completed by 1911, but
first published in 1921. It is a mark of Keynes’s range, creative originality, and
insight that much recent work in statistics has returned to the themes of the
Treatise on Probability. In economics, his first major publication was Indian
Currency and Finance (1913), a product of his service in the India Office of
the British government from 1906 to 1908.
Monetary Reform (1923) was followed in 1930 by the two-volume Treatise
on Money, much of which remains of value, though Keynes himself came to
regard its theoretical analysis as simply a step on the road to the General Theory,
the last of his major works. These major works were supplemented by numerous
articles, reviews, and biographical essays on some of his predecessors. 1
Keynes’s interest and influence were by no means limited to the confines
of the academy. For decades he exerted a major influence on public affairs and
played an active role in the world of business. His Economic Consequences
of the Peace (1919), based on his activities as an adviser to the British Treasury during the negotiation of the Versailles Peace Treaty, had a major impact
on public opinion and public policy, not only in Britain but throughout the
world, and not only immediately. It was translated into many languages, became a worldwide best-seller, and first established Keynes as a major public
1 The biographical essays on economists are gathered together in his Essays in Biography
(1933), along with similar essays on politicians and others.

M. Friedman: John Maynard Keynes


figure. It influenced the reaction of both victors and vanquished to the Versailles
Peace Treaty. Indeed, in a book, The Carthaginian Peace; or, the Economic
Consequences of Mr. Keynes, published more than two decades later (1946),
Etienne Mantoux pays the Economic Consequences a backhanded compliment
by arguing that Keynes’s debunking of the peacemakers was the source of all
subsequent evil, including World War II.
From 1919 on, Keynes remained active in public matters, publishing a
steady stream of articles on current affairs in nonprofessional journals and newspapers, advising and participating in the deliberations of the Liberal party, serving as chairman of the Nation and Athenaeum when it was acquired by a group
of Liberals in 1922, and later as director of the combined New Statesman and
Nation, leading journals of opinion for which he wrote frequently. He brought
together many of his most significant pieces on public affairs in Essays in Persuasion (1931). He served on government commissions, notably the Macmillan
Commission, and advised and consulted with successive governmental ministers. He was chairman of the National Mutual Insurance Company and director
of several other insurance companies. His interests were truly catholic: E. A. G.
Robinson, who was co-editor of the Economic Journal with Keynes for some
years and succeeded him as editor, begins an Encyclopaedia Britannica article
on Keynes by describing him as “1st Baron . . . , British economist who revolutionized economic theories, critic and architect of national economic policies,
political essayist, successful financier, bibliophile and patron of the arts.” His
interest in one particular art, ballet, was both cause and effect of his marriage in
1925 to Lydia Lopokova, a famous Russian ballerina. He established and largely
financed the Cambridge Arts Theater and was a trustee of the National Gallery.
From 1919 to World War II, Keynes’s connection with government was
primarily as an influential outsider. From 1940 on, he served in government
in a variety of capacities concerned with the economic conduct of the war
and postwar reconstruction. He was the chief British representative at Bretton
Woods in 1944, where he was a major architect of the plans for the International
Monetary Fund and the World Bank for Reconstruction and Development. He
was the chief negotiator of the large U.S. loan to Britain in 1945. On his return
to Britain, he played an important role in persuading the British Parliament to
adopt the Bretton Woods agreement. He died shortly thereafter, on April 21,

To return to the General Theory: its influence on both economic thinking and
economic practice was profound. The “Keynesian revolution” was far more
than a figure of speech. From shortly after the publication of the book in 1936
to at least the 1960s, the majority of professional economists, and certainly


Federal Reserve Bank of Richmond Economic Quarterly

the most prominent, termed themselves “Keynesians.” Those who called themselves non- or anti-Keynesians were a beleaguered minority, supplemented, it
must be said, by some important writers on economics who were not members of the professional guild.2 Governments around the world hastened to
adopt “Keynesian policies,” though many an economist—both Keynesians and
anti-Keynesians—regarded some of the policies, particularly when they led to
inflation, as at best “bastard Keynesianism.”3
As of this writing (1988), the status and influence of the book has changed.
It continues to have a major influence on economic thinking and economic policy, and will long continue to do so, but for very different reasons and in a very
different way than it did initially. The catalyst for the change was the inflation
and stagflation of the 1970s. As Robert Lucas wrote in 1981, “Proponents
of a class of models which promised 31/2 to 41/2 percent unemployment to a
society willing to tolerate annual inflation rates of 4 to 5 percent have some
explaining to do after a decade such as we have gone through [i.e., the 1970s,
when inflation rose to 16 percent and unemployment to 8 percent in the United
States, and to 30 percent and 6 percent in the U.K. Inflation rose as high as 25
percent in Japan and 7 percent in Germany, though unemployment remained
relatively low]. A forecast error of this magnitude and central importance to
policy has consequences, as well it should.”
The predictions to which Lucas refers were based on the so-called Phillips
curve which linked inflation inversely to unemployment—allegedly, the higher
the rate of inflation, the lower the level of unemployment. The curve was
asserted by many Keynesians to be stable over time and to specify a menu of
combinations of inflation and unemployment, any of which was attainable by
the appropriate monetary and fiscal policy. Lucas went on to note that “in the
late 1960s Milton Friedman (1968) and Edmund Phelps (1968) had argued . . .
that these predicted Phillips curve trade-offs were spurious.” They emphasized
the importance of distinguishing between anticipated and unanticipated inflation in interpreting the Phillips curve, and Friedman introduced the concept of
a “natural rate of employment” to which the economy would tend as economic
actors adjusted their anticipations.
“The central forecast to which [Friedman’s and Phelps’s] reasoning led,”
Lucas continued, “was a conditional one, to the effect that a high-inflation
decade should not have less unemployment on average than a low-inflation
decade. We got the high-inflation decade, and with it as clear-cut an

2 In the U.S., the most important was doubtless Henry Hazlitt, The Failure of the New
Economics: An Analysis of the Keynesian Fallacies (Princeton, N.J.: Van Nostrand, 1959).
3 The phrase was coined by Joan Robinson, one of the earliest and most dedicated members
of Keynes’s inner circle, in her review of Harry Johnson’s Money, Trade and Economic Growth
(1962), Economic Journal, vol. 72 (September 1962), p. 690. However, she used it to refer to the
theories of some of Keynes’s followers, rather than to policies.

M. Friedman: John Maynard Keynes


experimental discrimination as macro-economics is ever likely to see, and
Friedman and Phelps were right.”4
The 1980s have been no kinder to the earlier Keynesian models. In the
U.S., inflation was brought down drastically, accompanied by a temporary
increase in unemployment to a peak of nearly 11 percent—a short-term reaction to unanticipated disinflation along Phillips curve lines. But then, from
1983 on, unemployment fell concurrently with further declines in inflation,
reaching 6 percent by the end of 1987 when inflation was about 4 percent—a
flat contradiction of the asserted negative relation between unemployment and
inflation embodied in the Phillips curve. In the U.K., too, an initial decline in
inflation was accompanied by a sharp rise in unemployment, which was very
much slower to decline but has more recently begun to do so. In Germany,
inflation has come down since the early 1980s; unemployment rose initially,
as in the U.S. and the U.K., but, in contrast to them, continued to rise after
inflation had settled down, and has remained high. Japan, which was the first
of the major countries to cut sharply the rate of inflation, has succeeded in
keeping inflation low with little change in its recorded unemployment rate.
All in all, this experience is hardly consistent with a stable trade-off between
inflation and unemployment.
Experience led to disillusionment with initial Keynesianism on the part not
only of professional economists but also of policymakers. The most dramatic
evidence came from James Callaghan, when he was the Labour prime minister
of the U.K.—the party and the country that had gone farthest in embracing and
adopting Keynesian policies. Said Callaghan in 1976, “We used to think that
you could just spend your way out of a recession and increase employment
by cutting taxes and boosting government spending. I tell you, in all candour,
that that option no longer exists; and that insofar as it ever did exist, it only
worked by injecting bigger doses of inflation into the economy followed by
higher levels of unemployment as the next step. That is the history of the past
twenty years.”
Despite the widespread rejection of some of the key propositions that
constituted the “Keynesian revolution,” the book continues to have a major
impact on economic thinking. Some indication of its influence is given by the
continuing citations to the book in the professional literature. Data from one
citation index, which covers a wide range of economic journals, are available
for sixteen years, 1972 to 1987. In all, there were 1,558 citations to the General
Theory, or an average of nearly 100 a year. Of the total, 729 occurred in the
first eight years, 829 in the second eight, so there is no sign that interest in the
book is declining. However, the character of the book’s influence has changed.

4 Robert E. Lucas, Jr., “Tobin and Monetarism: A Review Article,” Journal of Economic
Literature, vol. 19 (June 1981), p. 560.


Federal Reserve Bank of Richmond Economic Quarterly

Some years ago, I remarked to a journalist from Time magazine, “We are all
Keynesians now; no one is any longer a Keynesian.” In regrettable journalist
fashion, Time quoted the first half of what I still believe to be the truth, omitting
the second half. We all use Keynesian terminology; we all use many of the
analytical details of the General Theory; we all accept at least a large part of the
changed agenda for analysis and research that the General Theory introduced.
However, no one accepts the basic substantive conclusions of the book, no one
regards its implicit separation of nominal from real magnitudes as possible or
desirable, even as an analytical first approximation, or its analytical core as
providing a true “general theory.”
As one, no doubt somewhat idiosyncratic, view of the book, I quote from
a reply that I wrote some years ago to criticisms of my work mostly from a
“Keynesian” point of view:
“One reward from writing this reply has been the necessity of rereading earlier
work, in particular [Keynes’s] . . . General Theory. The General Theory is
a great book, at once more naive and more profound than the ‘Keynesian
economics’ that Leijonhufvud contrasts with the ‘economics of Keynes.’ . . .5
“I believe that Keynes’s theory is the right kind of theory in its simplicity,
its concentration on a few key magnitudes, its potential fruitfulness. I have
been led to reject it, not on these grounds, but because I believe that it has
been contradicted by evidence: its predictions have not been confirmed by
experience. This failure suggests that it has not isolated what are ‘really’ the
key factors in short-run economic change.
“The General Theory is profound in the wide range of problems to which
Keynes applies his hypothesis, in the interpretations of the operation of modern economies and, particularly, of capital markets that are strewn throughout
the book, and in the shrewd and incisive comments on the theories of his
predecessors. These clothe the bare bones of his theory with an economic
understanding that is the true mark of his greatness.
“Rereading the General Theory has . . . reminded me what a great economist Keynes was and how much more I sympathize with his approach and
aims than with those of many of his followers.”6

As its title indicates, the General Theory is almost pure abstract theory. There
is only passing reference to applied economics, statistical magnitudes, or economic policy. Yet, like all of Keynes’s writings on economics, it was inspired
by a major contemporary problem and written in the hope and expectation
5 Axel Leijonhufvud, On Keynesian Economics and the Economics of Keynes (London:
Oxford University Press, 1968).
6 Milton Friedman’s Monetary Framework: A Debate with His Critics, ed. by Robert J.
Gordon (Chicago: University of Chicago Press, 1974), pp. 133–34.

M. Friedman: John Maynard Keynes


of providing a solution. The book was written during the worldwide Great
Depression following 1929, when idle men, idle machines, and unmet demand
coexisted on a large scale for years on end and produced widespread poverty,
misery, and deprivation. For Britain, it followed a near-decade of economic
stagnation, high unemployment, and long-term dependence of many families
on a government dole. The key problem of the time was how to explain the
apparent paradox, and, more urgently, how to resolve it.
Ups and downs in economic activity involving occasional periods of widespread unemployment had long occurred and had engaged the attention of
numerous economists under the rubric of “business fluctuations,” or “business
cycles.” Various theories had been offered to explain them. Most earlier theories
implicitly accepted the proposition that a private-enterprise capitalist system
contained self-correcting forces that would keep disturbances temporary. By
corrective adjustments to changes in circumstances, the system, it was believed,
would tend toward full employment of both men and machines—save only for
fractional and transitory unemployment implicit in a dynamic economy. However, the long duration and magnitude of the unemployment during the Great
Depression and the prior years in Britain did not seem to fit this pattern. Could
these be interpreted as simply a temporary, if long-lasting, disturbance? Or did
they indicate a defect in the supposed self-adjusting forces at work, so that
the economy could get stuck for long periods of time at a position of high
unemployment—a position that might have just as much reason to be regarded
as an “equilibrium” as a position of full employment?
Such a possibility had frequently been asserted by socialist and other critics of a capitalist system, whom the mainline professional economists had
regarded as “crackpots.” Keynes took the possibility seriously and proceeded
to construct an hypothesis that he believed demonstrated the possibility—indeed
the frequent reality—that, without government intervention, a private-enterprise
capitalist system using a non-commodity money would tend toward a position
characterized by a high level of involuntary unemployment of persons who
would willingly be employed at the current wage rate but could not find jobs.
The classical remedy for idle men, according to Keynes, was a decline in
the real wage rate, which would reduce the number of persons seeking jobs
and increase the number of persons employers wanted to hire. The classical
remedy for idle machines was a reduction in the cost to enterprises of using
and producing such machines, and that was expected to occur via a reduction
in the real interest rate.
In the 1920s and 1930s in Britain, these classical remedies seemed either
inoperative or ineffective. Keynes set himself the task of explaining why, of
constructing an alternative theory that would both explain what was happening
and justify alternative policies—such as the large public works programs he
had been recommending since the mid-1920s.


Federal Reserve Bank of Richmond Economic Quarterly

In one sense, his approach was strictly Marshallian: in terms of demand
and supply. However, whereas Marshall dealt with specific commodities and
“partial equilibrium,” Keynes proposed to deal with what he called “aggregate
demand” and the “aggregate supply function,” and with general not partial
Where he deviated from Marshall was in the key variables that he regarded
as producing equilibrium between demand and supply and in the process of adjustment to a change in demand or supply. In Marshallian analysis, the key role
was played by prices, which reacted quickly to any change in circumstances. Let
there be a sudden increase in demand, in the sense of a demand function relating the quantity demanded to price. In Marshall’s view, the immediate reaction
would be on prices, which would rise to choke off the quantity demanded to the
prior level plus whatever additional quantities might be made available from inventories. The rise in prices during the “market period” would give producers an
incentive to increase output in the “short run” by using existing plant and equipment more intensively, and, if the increased demand persisted, in the longer run
by adding to plant and equipment. In short, prices adjusted rapidly, quantities
slowly, and changes in prices played the major role in producing equilibrium.
To Keynes, it seemed clear that this process had been inoperative or ineffective with respect to the economy as a whole. Nominal wage rates had indeed
declined, but so had nominal prices, so that real wages had hardly moved,
and may indeed have increased. He concluded that movements in prices and
interest rates could not be counted on. Accordingly, he reversed Marshall’s
presumptions: prices of labor and capital, at least “real wages” and “real interest rates,” are very slow to adjust; quantities, which is to say consumption,
investment, and their sum, total output, are highly flexible and adjust rapidly.
Changes in output (aggregate supply), not in prices, play the major role in
producing equilibrium. Accordingly, as a first approximation—though one he
never really relaxed—he took prices as given by forces outside his analysis.
As a first approximation, also, he abstracted from both government spending
and international trade, but these could readily be integrated into the analysis
without affecting its substance.
Keynes defined aggregate demand and aggregate supply in terms of
employment, in line with his view that he was developing a “theory of employment.” However, both Keynes and his followers tended to replace employment by output and to express aggregate demand and aggregate supply
in terms of the value of output demanded by the public and supplied by
Aggregate demand, in these terms, is the sum of expenditures on consumption goods and expenditures on investment goods. Keynes regarded expenditures on consumption as depending on income, introducing one of his
key concepts: the propensity to consume, or, in his words, “the functional
relationship . . . between . . . a given level of income in terms of wage-

M. Friedman: John Maynard Keynes


units, and . . . the expenditure on consumption out of that level of income.”
A “fundamental psychological law,” which plays a key role in the Keynesian
system, is that “men are disposed . . . to increase their consumption as their
income increases, but not by as much as the increase in their income”—i.e.,
the “marginal propensity to consume” is less than unity.7
Keynes defined investment as “the current addition to the value of the capital equipment which has resulted from the productive activity of the period.”
He regarded investment as depending on the “marginal efficiency of capital,”
the second of his key concepts, which he defined as “that rate of discount which
would make the present value of the series of annuities given by the returns
expected from the capital-asset during its life just equal to its supply price,”
i.e., “the cost of producing” one more unit of the asset. Like the propensity to
consume, the marginal efficiency of capital is a function or schedule relating
the amount of investment to the interest rate, since entrepreneurs would have
an incentive to add to investment so long as the yield exceeded the interest rate
at which they could borrow the funds to finance the investment.8
The interest rate, in turn, he regarded as determined by “liquidity preference,” the third of his key concepts. “An individual’s liquidity-preference
is given by a schedule of the amounts of his resources, valued in terms of
money or of wage-units, which he will wish to retain in the form of money
in different sets of circumstances.” He regarded the amount of their assets that
individuals would want to hold in the form of money as depending on both
income and the interest rate—income because that would affect the amount
held for “transactions- and precautionary-motives,” the interest rate, because
that would affect the amount held “to satisfy the speculative-motive.” 9
If, as Keynes did, we let Y be income, identical with the value of output,
C be consumption, I be investment, L liquidity preference, M the quantity of
money, and r the interest rate, then aggregate demand is given by
Y = C(Y) + I(r),


M = L(Y, r).


and the demand for money by
In line with his implicit assumption about the relative speed of adjustment
of prices and output, Keynes regarded supply as essentially passive, expanding
or contracting as demand expanded or contracted, subject only to the proviso
that employment is less than “full,” which he defined as the point at which an
increase in aggregate demand would call forth no additional workers willing to

7 The General Theory of Employment Interest and Money (London: Macmillan, 1936), pp.
90, 96, and 114.
8 Ibid., pp. 62 and 135.
9 Ibid., pp. 166 and 199.


Federal Reserve Bank of Richmond Economic Quarterly

work at the wage offered. This leads him to regard aggregate supply as given
simply by aggregate demand, or
YS = YD ,
and the level of aggregate supply and demand as affecting not a price but solely
If we regard the interest rate as fixed, along with other prices, then equations
(1) and (3) define the famous Keynesian “multiplier” (attributed by Keynes to
Richard Kahn). For a simple version, assume that the consumption function is
C = a + bY,
with b, of course, less than one. Substituting (4) in (1) and solving for Y, we
a + I(r)
[a + I(r)].

The multiplier is 1/(1 − b), which, given that b is between zero and unity,
is necessarily greater than unity. The multiplicand, (a + I), came to be termed
“autonomous” spending, i.e., spending not dependent on the level of income. In
addition, once government was introduced into the analysis, autonomous spending was regarded as including not only autonomous consumption spending (a)
and investment (I ) but also government spending.
Equations (1) and (3) define also the equally famous “Keynesian cross,”
which has been reproduced in literally hundreds of textbooks in the past half
century and is reproduced here in Figure 1.
The graph makes clear the key importance of the “fundamental psychological law” that the marginal propensity to consume is less than unity. If it were
unity, the YD line would parallel the YS line and there would be either no or
an infinite number of equilibrium positions, according as the two parallel lines
were distinct or identical. If it exceeded unity, the YD line would slope more
steeply than the YS line, and any point of intersection would be an unstable
equilibrium position. Because it slopes less steeply, the intersection at YO is
a stable equilibrium. If output were temporarily higher than YO , employers
would be making losses, since the aggregate supply price would exceed aggregate demand, and would seek to contract output. Conversely, if output were
temporarily lower than YO , employers would be making profits and would seek
to expand.
If, for whatever reason, investment were to increase from IO to IO , the YD
line would shift to YD and the new equilibrium would shift to YO . At YF , the
point of full employment, the process would end, and “the crude quantity theory
of money,” which is the particular object of Keynes’s scorn and derision—no
doubt because of his long earlier adherence to it—“is fully satisfied.”10

Ibid., p. 289.

M. Friedman: John Maynard Keynes


Figure 1 The Keynesian Cross

Y S = YD


Y D = C + IO
Y D = C + IO





Marvelously simple. A key that apparently unlocks the mystery of
long-continued unemployment: inadequate autonomous spending or too low a
propensity to consume. Increase either, or both, being careful simply not to go
too far, and full employment could be attained. What a wonderful prescription:
for consumers, spend more out of your income, and your income will rise;
for governments, spend more, and aggregate income will rise by a multiple of
your additional spending; tax less, and consumers will spend more with the
same result. Though Keynes himself, and even more, his disciples, produced
much more sophisticated and subtle versions of the theory, this simple version
contains the essence of its great appeal to non-economists and especially governments. Here was one of the most famous and respected economists in the
world informing governments that the way to full employment was paved with
higher spending and lower taxes. What more attractive advice could politicians
wish for? Long regarded public vices turned into public virtues!
Marvelously simple, yes. But also simply marvelous. How could a position
such as YO in Figure 1 be regarded as a long-term equilibrium—as was implied
in the claim that the theory was “general”? At that point, men and machines


Federal Reserve Bank of Richmond Economic Quarterly

are idle. Would not the excess supply of men and machines exert downward
pressures on the prices of both? Yes, said Keynes, but, if effective, that would
be accompanied by lower money prices of output that would cancel the lower
money wages and money cost of capital, so that real wages and the real cost
of capital would be unaffected—which is why Keynes expressed all aggregate
magnitudes in “wage-units.” Hence, said Keynes, flexible wages and prices
would do no good. Far better to operate directly on spending.
Of course, Keynes recognized that changes in prices, interest rates, and
quantity of money did have effects that provided alternative avenues of escape
from the so-called “underemployment equilibrium.” At best, it was a transitory
equilibrium position, the existence of which would set in motion self-corrective
forces. But Keynes tended to rule out these alternative avenues of escape as of
no practical significance because of his empirical judgment that prices, wages,
and interest rates were highly sluggish. Indeed, some commentators on Keynes
maintain that he deliberately overstated his case in order to shock the economics
profession into paying attention—a tactic that is common to every innovator,
whether it be of an idea or a product.
Only one alternative avenue of adjustment is explicitly present in equations (1) and (2)—via the interest rate and the quantity of money. This avenue,
analyzed at some length in the General Theory, and found wanting to produce,
by itself, a full employment equilibrium, also was rapidly incorporated in an
alternative, more sophisticated graphical representation of the Keynesian system developed almost simultaneously by John Hicks and Roy Harrod.11 Figure
2 presents Hicks’s IS-LM version, which very quickly became the orthodox
In this diagram, the vertical axis is the interest rate. The horizontal axis is
income expressed in wage-units, so that it is also output and employment. The
IS curve traces equation (5), i.e., it shows the combinations of interest rate and
output that would satisfy equation (1): the higher the interest rate, the lower
investment and hence income, and conversely, which is why the IS curve has
a negative slope. Put differently, it shows the combinations of interest rate and
output at which the amount some people wish to invest is equal to the amount
other people wish to save, which is what explains the S in IS. But note that the
accommodation of saving to investment is produced not by the direct effect of
the interest rate on saving, but by the effect of the level of income on saving,
via the propensity to consume.
The LM curve traces equation (2) for a fixed quantity of money. Here, the
higher the interest rate, the lower the quantity of money that the public would
want to hold for a given income, and hence the higher income must be in order
11 John R. Hicks, “Mr. Keynes and the ‘Classics’: A Suggested Interpretation,” Econometrica, vol. 5 (April 1937), pp. 147–59; Roy F. Harrod, “Mr. Keynes and Traditional Theory,”
Econometrica, vol. 5 (January 1937), pp. 74–86.

M. Friedman: John Maynard Keynes


Figure 2 The IS-LM Diagram



(LM) (M=MO)





(IS) (I=IO)

for the actual quantity of money to be willingly held. Hence the positive slope
of the LM curve.
The intersection of the IS and LM curve at YO is the counterpart of the
intersection of the aggregate demand and supply curves in Figure 1 at YO . Similarly, the IS curve is the counterpart of the YO curve in Figure 1, reflecting a
higher level of investment. It is the IS curve moved to the right by the change
in income assumed to be produced by the increase in investment—the change
in investment times the investment multiplier.
What is new in Figure 2 are the LM curves. Each LM curve is for a specific
quantity of money: the LM curve for M = MO , the (LM) curve for M = MO ,
which is larger than MO . For the community to hold the larger quantity of
money willingly, either the interest rate must be lower for a given income or
income higher for a given interest rate, which is why the (LM) curve is to the
right of the LM curve.
The IS curve in the diagram embodies a possible Keynesian escape from
underemployment via increases in investment (or, more generally, autonomous
spending including government spending). Let autonomous spending be high
enough so that the IS curve intersects the LM curve at point F, and full employment would be attained with the initial quantity of money.
The LM curve offers an alternative escape via the quantity of money. Let
the quantity of money be large enough so that the LM curve intersects the


Federal Reserve Bank of Richmond Economic Quarterly

Figure 3 An Extreme IS-LM Diagram with Perfectly Elastic
Liquidity Preference and Inelastic Investment

IS curve at point F , and full employment would be attained with the initial
marginal efficiency of capital schedule.
Keynes and his followers rejected this possibility as highly unrealistic,
largely on the alleged empirical grounds that (1) private autonomous expenditures were little affected by changes in the interest rate while (2) there was
a floor to the interest rate at which the community would be willing to hold
assets other than money, so that, in the neighborhood of this floor, the quantity
of money the community would be willing to hold would be highly sensitive
to the interest rate: in short, a low elasticity of investment, but a high elasticity
of liquidity preference, with respect to the interest rate.
Figure 3 shows an extreme version of these assumptions: perfectly inelastic
investment and perfectly elastic liquidity preference. We are back to the Keynesian cross of Figure 1. No changes in the quantity of money can produce a full
employment equilibrium. This LM curve depicts a “liquidity trap,” of which
Keynes wrote, “whilst the limiting case might become practically important in
future, I know of no example of it hitherto. Indeed, owing to the unwillingness
of most monetary authorities to deal boldly in debts of long term, there has

M. Friedman: John Maynard Keynes


Figure 4 A Less-Extreme Liquidity Trap

not been much opportunity for a test.”12 Of course, it is not necessary to go to
this extreme to generate Keynesian unemployment equilibria, and Keynes and
his followers did not, though some of the more enthusiastic of his disciples
came very close during the high tide of the Keynesian revolution. It is only
necessary to suppose a highly inelastic IS curve, and a highly elastic LM curve,
as in Figure 4. In this version, a negative interest rate would be required for a
full employment equilibrium. The Keynesians ruled out this possibility by the
assumption of given prices.13
The avenue of adjustment that is not explicitly allowed for in either equations (1) and (2) or in the more sophisticated IS-LM diagram is the level of
prices and wages. As already noted, a Keynesian position of underemployment
equilibrium means downward pressure on wages and prices. Keynes explicitly

The General Theory, p. 207.
The interest rate that is relevant to investment is the “real” interest rate, i.e., the nominal
rate of interest less the rate of inflation, and the “real” interest rate has often been negative.


Federal Reserve Bank of Richmond Economic Quarterly

recognized that a change in real wages would affect employment by altering
both the supply and the demand for labor.14 However, he ruled out that avenue
of escape on the grounds that prices and wages would tend to change pari-passu
leaving real wages largely unchanged—not a bad empirical approximation for
the kind of major disturbances, such as the Great Depression, whose origin and
cure Keynes was seeking. Keynes discussed two other effects of changes in the
level of prices and wages. The first is on the real quantity of money, and thence
the rate of interest. A lower level of prices is equivalent to a higher quantity
of money, and like an increase in the quantity of money would shift the LM
curve to the right. The second is the effect of a lower rate of interest on the
consumption function, an effect that has come to be called the Keynes effect.
The lower the interest rate, the higher the capital value of a given stream of
income—such as rent on a piece of land, or coupons on a bond. Hence, a lower
interest rate increases the wealth of the community. The higher the wealth, the
less pressure to add to wealth via savings, and hence the higher is likely to be
the average and marginal propensity to consume at any income.
Though Keynes recognized the existence of these avenues of adjustment,
he largely dismissed them on empirical grounds. Sluggishness of price movements had pride of place, but inelasticity of investment and elasticity of liquidity
preference with respect to the interest rate and inelasticity of consumption with
respect to wealth were also important.
A third effect of a pari-passu change in prices and wages, which came to
be known as the “Pigou” effect, was not discussed explicitly by Keynes. The
lower the price level, the higher the real value of the fixed quantity of money. In
principle, there is no limit to the real value of a fixed nominal quantity of money,
and hence no limit to the wealth of a community, and accordingly, no limit to
the extent to which the IS curve could be shifted to the right by the reduction in
the incentive to save.15 There is much dispute about the empirical importance
of this effect. I personally regard it as minor. However, on the purely abstract
theoretical level of the General Theory, it conclusively demonstrates that there
is no such flaw in the price system as Keynes professed to demonstrate. His
position of underemployment equilibrium, whatever else it might be, was not
a long-run equilibrium position that set in motion no effective forces tending
toward full employment.
What difference does this abstract analysis make? Is it not simply arguing
about how many angels can dance on the point of a pin? The answer is that it
destroys Keynes’s most striking and radical claim made in the first paragraph
of the General Theory: that what he called the “classical economics,” and, in

See, for example, ibid., p. 289.
For a fuller theoretical analysis of (a) the possibility of a negative equilibrium interest
rate, and (b) the Keynes and Pigou effects, see Milton Friedman, Price Theory (Chicago: Aldine
Publishing Co., 1976), pp. 313–21.

M. Friedman: John Maynard Keynes


particular, the quantity theory of money, were fundamentally fallacious, “that
the postulates of the classical theory are applicable to a special case only and
not to the general case, the situation which it assumes being a limiting part
of the possible positions of equilibrium. Moreover, the characteristics of the
special case assumed by the classical theory happen not to be those of the
economic society in which we actually live, with the result that its teaching is
misleading and disastrous if we attempt to apply it to the facts of experience.” 16
If this extreme claim is wrong, Keynes’s theory becomes not a theory of
“equilibrium” but at best a theory of disequilibrium, readily encompassed in
the earlier orthodoxy. Conventional wisdom prior to the General Theory had
always recognized that fluctuations existed, and that periods of widespread
unemployment did occur from time to time. But it regarded these as responses
to changes in circumstances, plus rigidities in prices, wages, and other variables
that impeded rapid adjustment to the new circumstances. And, indeed, conventional economic wisdom has by now come to regard the Keynesian theory as
a theory of disequilibrium, which provides a useful way to analyze the process
of adjustment to changes in circumstances in a world of relatively rigid prices
and wages. It should be added that there does remain a significant number of
respected economists who continue to regard Keynes’s contribution as providing a truly general theory fully justifying his initial claims, and continue to
regard him as having demolished the so-called classical theory.17
There remains the twin questions of why Keynes, who described himself
in the preface to the German edition as having been “a priest of” the English
classical quantity theory tradition, regarded it as incompetent to explain the
persistence of high unemployment in the 1920s and 1930s, and of how those
of us who disagree with him reconcile that remarkable phenomenon with the
earlier theory. The key to the answer to both questions is the interpretation of
monetary developments, and particularly monetary policy in the 1930s. Consider first the situation in the U.S. By contrast with Britain, the 1920s were
a period of general prosperity, high employment, and relatively stable prices.
There was no reason to question the importance of monetary policy. Indeed, the
Federal Reserve System in the United States took for itself much of the credit
for the good performance of the economy. But then came the Great Depression.
Its initial phase, from 1929 to late 1930, had all the characteristics of a gardenvariety recession, though somewhat more severe than most, and, indeed, had it
ended in early 1930, or even early 1931, as it showed some signs of doing, it
would have gone down in history in that way, not as a major contraction, let
alone Great Depression. But the second phase, from the end of 1930 to 1933,
was very different. It was marked by a succession of banking crises, and the

Ibid., p. 1.
The most prominent of this group are the late Joan Robinson, the late Nicholas Kaldor,
in Britain, and Professor Robert Eisner, in the United States.


Federal Reserve Bank of Richmond Economic Quarterly

veritable collapse of the banking system leading to an unprecedented “bank
holiday” in March 1933, during which all the banks of the country—including
the Federal Reserve Banks themselves—were closed for business. When the
holiday ended and “sound banks” reopened, they numbered only two-thirds as
many as were in existence in 1929. This sequence of events was accompanied
by a disastrous increase in unemployment, and major declines in prices, wages,
and national income both in current and constant prices. From 1929 to 1933,
“money income fell 53 percent and real income 36 percent . . . . Per capita
real income in 1933 was almost the same as in the depression year of 1908, a
quarter of a century earlier . . . . At the trough of the depression one person
was unemployed for every three employed.”18 And what happened in the
United States was duplicated—the banking disaster partly excepted—around
the world.
To Keynes and many of his contemporaries, this sequence of events seemed
a clear contradiction of the earlier theory and of the efficacy of monetary policy.
They tended then, as many still do, to regard monetary policy as operating via
interest rates. Short-term interest rates in the United States had fallen drastically
during the contraction. In particular, the discount rate charged by the Federal
Reserve Banks on loans to banks that were members of the Federal Reserve
System was steadily reduced from 6 percent in 1929 to 1.5 percent by the fall
of 1931, though it was then abruptly increased to 3.5 percent in response to
Britain’s departure from gold in September 1931, and was still 2.5 percent in
early 1933. Judged in these terms, monetary policy was “easy,” yet it apparently had been powerless to stem the contraction, giving rise to widespread
apprehension that monetary policy was like a string: you could pull on it, but
not push on it, i.e., monetary policy could check inflation but could not offset
From another, and I would argue far more significant, point of view, monetary policy was anything but “easy.” That point of view regards monetary
policy as operating via the quantity of money. In terms of annual averages, the
quantity of money in the United States fell by one-third from 1929 to 1933—
by 2 percent from 1929 to 1930, just before the onset of the first banking
crisis, and by a further 32 percent from 1930 to 1933. Data on the quantity of
money were not published regularly at that time and were not readily available
even with some lag, whereas interest rates were readily and contemporarily
available—both effect and reinforcement of the tendency to interpret monetary
policy in terms of the interest rate rather than the quantity of money.
Keynes may well not have known what was happening to the quantity
of money, though if he had, he would also have known that “[a]t all times
18 Milton Friedman and Anna J. Schwartz, A Monetary History of the United States, 1867–
1960 (Princeton: Princeton University Press for the National Bureau of Economic Research, 1963),
p. 301.

M. Friedman: John Maynard Keynes


throughout the 1929–33 contraction, alternative policies were available to the
[Federal Reserve] System by which it could have kept the stock of money
from falling, and indeed could have increased it at almost any desired rate.”
Far from demonstrating, as Keynes concluded, that monetary policy is impotent,
“[t]he contraction is in fact a tragic testimonial to the importance of monetary
forces.”19 The contraction continued and deepened not because there were no
equilibrating forces within the economy but because the economy was subjected
to a series of shocks succeeding one another: a first banking crisis beginning in
the fall of 1930, a second beginning in the spring of 1931, Britain’s departure
from gold in September 1931, and the final banking crisis beginning in January 1933—all accompanied by a decline in the quantity of money of 7 percent
from 1930 to 1931, 17 percent from 1931 to 1932, and 12 percent from 1932
to 1933.
Even after the end of the contraction and the start of revival in 1933,
the shocks continued and impeded recovery: major legislative measures during
Franklin Delano Roosevelt’s New Deal that interfered with market adjustments
and generated uncertainty within the business community, although some of
them, particularly the enactment of federal insurance of bank deposits, reassured the community about the safety and stability of the financial institutions;
then ill-advised monetary measures in 1936 that halted the rapid rise that had
been occurring in the quantity of money and produced an absolute decline from
early 1937 to early 1938 that exacerbated if it did not produce the accompanying
severe cyclical decline.
Keynes’s readiness to interpret the U.S. experience as evidence of the
impotence of monetary policy was greatly strengthened by the British experience. By contrast with the U.S., the 1920s was a period of stagnation and
high unemployment that the severe worldwide contraction beginning in 1929
intensified. However, the contraction ended earlier in Britain than in the U.S.,
shortly after Britain left the gold standard and thereby cut its monetary link
with the U.S. Here, too, a succession of shocks played an important role: the
end of World War I and demobilization; the pressure to return to gold at the
prewar parity, which required internal deflation; the return in 1925 to gold at
a parity that overvalued the pound sterling, particularly after France returned
to gold at a parity that undervalued the franc; and, finally, the shock waves
that spread from the U.S. after 1929. The effect of steady deflationary pressure
was reinforced by “an unemployment insurance scheme that paid benefits that
were high relative to wages available subject to few restrictions . . . . Although
a few interwar observers saw clearly the effects of unemployment insurance,
Keynes and his followers did not.”20

Ibid., pp. 693 and 300.
Daniel K. Benjamin and Levis A. Kochin, “Searching for an Explanation of Unemployment in Interwar Britain,” Journal of Political Economy, vol. 87 (June 1979), p. 441.


Federal Reserve Bank of Richmond Economic Quarterly

In judging Keynes’s overall influence on public policy, it is necessary to distinguish his bequest to technical economics from his bequest to politics. Keynes’s
bequest to technical economics was strongly positive. His bequest to politics,
in my opinion, was not. Yet I conjecture that his bequest to politics has had
far more influence on the shape of today’s world than his bequest to technical
economics. In particular, it has contributed greatly to the proliferation of overgrown governments increasingly concerned with every phase of their citizens’
daily lives.21
I can best indicate what I regard to be Keynes’s bequest to politics by quoting from his famous letter to Professor Friedrich von Hayek praising Hayek’s
Road to Serfdom. The part generally quoted is from the opening paragraph of
the letter: “In my opinion it is a grand book . . . . [M]orally and philosophically I find myself in agreement with virtually the whole of it; and not only in
agreement with it, but in a deeply moved agreement.”
The part I want to direct attention to comes later:
“I should therefore conclude your theme rather differently. I should say that
what we want is not no planning, or even less planning, indeed I should say
that we almost certainly want more. But the planning should take place in a
community in which as many people as possible, both leaders and followers
wholly share your own moral position. Moderate planning will be safe if those
carrying it out are rightly orientated in their own minds and hearts to the moral
“What we need therefore, in my opinion, is not a change in our economic
programmes, which would only lead in practice to disillusion with the results
of your philosophy; but perhaps even the contrary, namely, an enlargement of
them . . . . No, what we need is the restoration of right moral thinking—a
return to proper moral values in our social philosophy . . . . Dangerous acts can
be done safely in a community which thinks and feels rightly, which would be
the way to hell if they were executed by those who think and feel wrongly.”22

Keynes was exceedingly effective in persuading a broad group—economists, policymakers, government officials, and interested citizens—of the two
concepts implicit in his letter to Hayek: first, the public interest concept of
government; second, the benevolent dictatorship concept that all will be well if
only good men are in power. Clearly, Keynes’s agreement with “virtually the

21 The rest of this preface up to the final paragraph is drawn largely from my “Comment on
Leland Yeager’s Paper on the Keynesian Heritage,” in The Keynesian Heritage, a symposium by
Leland Yeager, Milton Friedman, and Karl Brunner, Center Symposia Series CS–16 (Rochester,
N.Y.: Center for Research in Government Policy and Business, Graduate School of Management,
University of Rochester, 1985), pp. 12–18.
22 Donald Moggridge, ed., John Maynard Keynes, The Collected Writings, Vol. XXVII:
Activities, 1940–1946, pp. 385, 387, 388.

M. Friedman: John Maynard Keynes


whole” of the Road to Serfdom did not extend to the chapter titled “Why the
Worst Get on Top.”
Keynes believed that economists (and others) could best contribute to the
improvement of society by investigating how to manipulate the levers actually
or potentially under control of the political authorities so as to achieve desirable ends, and then persuading benevolent civil servants and elected officials to
follow their advice. The role of voters is to elect persons with the right moral
values to office and then let them run the country.
From an alternative point of view, economists (and others) can best contribute to the improvement of society by investigating the framework of political
institutions that will best assure that an individual government employee or
elected official who, in Adam Smith’s words, “intends only his own gain . . .
is . . . led by an invisible hand to promote an end that was no part of his
intention,” and then persuading the voters that it is in their self-interest to
adopt such a framework. The task, that is, is to do for the political market what
Adam Smith so largely did for the economic market.
Keynes’s view has been enormously influential—if only by strongly reinforcing a pre-existing attitude. Many economists have devoted their efforts to
social engineering of precisely the kind that Keynes engaged in and advised
others to engage in. And it is far from clear that they have been wrong to do
so. We must act within the system as it is. We may regret that government
has the powers it does; we may try our best as citizens to persuade our fellow
citizens to eliminate many of those powers; but so long as they exist, it is
often, though by no means always, better that they be exercised efficiently than
inefficiently. Moreover, given that the system is what it is, it is entirely proper
for individuals to conform and promote their interests within it.
An approach that takes for granted that government employees and officials
are acting as benevolent dictators to promote in a disinterested way what they
regard as the public’s conception of the “general interest” is bound to contribute
to an expansion in governmental intervention in the economy—regardless of
the economic theory employed. A monetarist no less than a Keynesian interpretation of economic fluctuations can lead to a fine-tuning approach to economic
The persuasiveness of Keynes’s view was greatly enhanced in Britain by
historical experience, as well as by the example Keynes himself set. Britain
retains an aristocratic structure—one in which noblesse oblige was more than
a meaningless catchword. What has changed are the criteria for admission to
the aristocracy—if not to a complete meritocracy, at least some way in that
direction. Moreover, Britain’s nineteenth-century laissez-faire policy produced
a largely incorruptible civil service, with limited scope for action, but with
great powers of decision within those limits. It also produced a law-obedient
citizenry that was responsive to the actions of the elected officials operating in
turn under the influence of the civil service. The welfare state of the twentieth


Federal Reserve Bank of Richmond Economic Quarterly

century has almost completely eroded both elements of this heritage. But that
was not true when Keynes was forming his views, and during most of his
public activity.
Keynes’s own experience was also influential, particularly to economists.
He set an example of a brilliant scholar who participated actively and effectively
in the formulation of public policy—both through influencing public opinion
and as a technical expert called on by the government for advice. He set an
example also of a public-spirited and largely disinterested participant in the
political process. And it is not irrelevant that he gained worldwide fame, and
a private fortune, in the process.
The situation was very different in the United States. The United States is
a democratic not an aristocratic society, as Tocqueville pointed out long ago. It
has no tradition of an incorruptible or able civil service. Quite the contrary. The
spoils system formed public attitudes far more than a supposedly non-political
civil service. And it did so even after it had become very much emasculated in
practice. As a result, Keynes’s political bequest has been less effective in the
United States than in Britain, which partly explains, I believe, why the “public
choice” revolution in the analysis of politics occurred in the United States. Yet
even in the United States, Keynes’s political bequest has been tremendously
effective. Certainly most writing by economists on public policy—as opposed
to scientific and technical economics—has been consistent with it. Economists,
myself included, have sought to discover how to manipulate the levers of power
more effectively, and to persuade—or educate—governmental officials regarded
as seeking to serve the public interest.
I conclude that Keynes’s political bequest has done far more harm than
his economic bequest and this for two reasons. First, whatever the economic
analysis, benevolent dictatorship is likely sooner or later to lead to a totalitarian
society. Second, Keynes’s economic theories appealed to a group far broader
than economists primarily because of their link to his political approach. Here
again, Keynes, in his letter to Hayek, said it better than I can: “Moderate
planning will be safe if those carrying it out are rightly orientated in their own
minds and hearts to the moral issue. This is in fact already true of some of
them. But the curse is that there is also an important section who could almost
be said to want planning not in order to enjoy its fruits but because morally they
hold ideas exactly the opposite of yours [i.e., Hayek’s], and wish to serve not
God but the devil. Reading the New Statesman and Nation one sometimes feels
that those who write there, while they cannot safely oppose moderate planning,
are really hoping in their hearts that it will not succeed; and so prejudice more
violent action. They fear that if moderate measures are sufficiently successful,
this will allow a reaction in what you think the right and they think the wrong
moral direction. Perhaps I do them an injustice; but perhaps I do not.”
Keynes did not let this analysis prevent him from serving until his death
as chairman of the New Statesman and Nation—presumably in the hope of

M. Friedman: John Maynard Keynes


influencing the moral views of its editors and writers. I regard Keynes’s analysis
as indicating that the key problem is not how to achieve a moral regeneration
but rather how either to frustrate what Keynes regards as “bad morals,” or to
construct a political framework in which those “bad morals” serve not only
the private but also the public interest, just as, in the economic market, private
greed is converted to public service.
The literature on Keynes and on the General Theory is by now immense.
Of the books specifically devoted to Keynes’s life, two stand out: the initial
authorized biography by his student and disciple, Roy F. Harrod, The Life of
John Maynard Keynes (1951); and the more recent multi-volume biography by
Robert J. A. Skidelsky, John Maynard Keynes, Vol. 1: Hopes Betrayed, 1883–
1920 (London: Macmillan, 1983), and Vol. 2: The Economist as Prince, 1920–
1937 (London: Macmillan, 1988). The Collected Writings of John Maynard
Keynes have been published under the auspices of the Royal Economic Society
in 29 volumes (Macmillan, 1971 to 1982), with a final Bibliography and Index
yet to come. This splendid collection includes not only his major work but also
his published articles on economics and politics, many previously unpublished
items, including letters, official memoranda and notes, and the like.

The Organization of
Private Payment Networks
John A. Weinberg


ne of the key roles banks have traditionally played is in the execution
of payments among participants in the economy. Liabilities issued by
banks, such as demand deposits, are a primary means of payment. The
widespread acceptability of such private liabilities requires a reliable method
for the settlement of such obligations. In a world where people and economic
activity are dispersed in space and time, settlement often requires a method
for communication between locations where purchases of goods take place and
those where payment liabilities are issued. That is, the use of bank liabilities
as means of payment requires the support of an interbank network for clearing
and settlement.
Throughout U.S. banking history, such multibank networks have played
important roles. In New England during the Free Banking period (1836 –1863),
the system that was centered around the Suffolk Bank in Boston widened the
area over which many banks’ notes could circulate at par.1 In the latter part
of the nineteenth century, banks participated in clearinghouses for the clearing
and settlement of local checks and had correspondent relationships to handle
checks over greater distances.2 While the Federal Reserve (the Fed) ultimately
took over a large part of the clearing and settlement of checks, the banking industry has developed other private, multilateral networks for handling interbank
payments. Most notable, perhaps, are the nationwide credit card associations.
Recently, interbank networks have received considerable attention. The ongoing growth and consolidation of Automated Teller Machine (ATM) networks
has stimulated discussions in the academic and public policy communities of
possible antitrust issues raised by such large joint ventures of banking organizations. Much of the discussion of possible public policy concerns regarding
This article has benefited greatly from the comments of Tom Humphrey, Jeff Lacker, PierreDaniel Sarte, and John Walter. The views expressed herein are the author’s and do not
represent the views of the Federal Reserve Bank of Richmond or the Federal Reserve System.

See, for instance, Calomiris and Kahn (1995).
A recent description of check clearing in the nineteenth century is found in Gilbert and
Summers (1996).

Federal Reserve Bank of Richmond Economic Quarterly Volume 83/2 Spring 1997



Federal Reserve Bank of Richmond Economic Quarterly

coming forms of electronic money centers on the network characteristics of
these instruments.
This article takes the position that networks are fundamental to the role
played by intermediary institutions in the payment system. Clearing and settlement, as the means of managing the financial relationship among individuals or
institutions across time and distance, are inherently network services. The characteristics of network services have important consequences for the industrial
organization of the payment system.
Arrangements for clearing and settlement of payments, whether private or
public, involve an effect that is sometimes referred to as a network externality.
Broadly, the private value to an individual of belonging to a network increases
with the number of endpoints to which that network connects. Put differently,
the private value an individual derives from participating in a network is that
the individual can communicate with the other network members. At the same
time, the individual’s participation creates value for other members by adding
to the number of endpoints.
There is an important difference between network externalities and other
forms of externality. Perhaps the most common textbook externality is pollution; the economic activity of some individuals may produce pollutants that
affect a much broader set of people. While individuals make choices about participation in the activity that generates pollution, they may have little choice as
to whether to be affected by pollution. In the extreme case of the greenhouse effect, for instance, it may be impossible to avoid incurring the costs of pollution.
Pollution is an external cost that has effects beyond the set of people engaged
in the polluting activity. Network externalities, on the other hand, are more selfcontained. An individual’s decision to subscribe to a network creates external
benefits for other subscribers by increasing the size of the network. Notice,
however, that in order to enjoy the effects of the externality, one must join
the network. The benefits to network participation, although partially external
to the individual participant, are entirely internal to the network as a group.
Since the group is composed of individuals engaged in mutually voluntary
exchange with one another, one would expect the group’s organization and
pricing arrangements to take account of the “external” benefits associated with
an individual’s participation.
In a payment network, the value of communication between two endpoints
is determined by the pattern of commerce. People at location A will place a
high value on being in a payment network with people from location B if there
is a high volume of commerce between the two locations. Since locations can
vary widely in the sets of places their people go to shop, there can be variety
among endpoints in both the private value of network participation and the
external value that an endpoint creates for others through its participation. For
a network to be sustainable, then, its services must be priced so that all of its
intended members have an incentive to join. For a network to be efficient, it

J. A. Weinberg: Private Payment Networks


must include all endpoints for which the total value of participation (private
plus external value) exceeds the resource cost of participation, and only those
This article proposes that the two standards of sustainability (which is
defined more precisely below) and efficiency can form the basis of a positive
theory of private, multilateral clearing and settlement arrangements.3 Such a
theory is quite distinct from the conventional view that network effects, as a
form of externality, are a common source of market failure. This conventional
view arises from the analysis of the behavior of network participants under the
assumption that key organizational features of networks are exogenously fixed.
By contrast, the theory presented below treats organizational arrangements as
the endogenous outcomes of interactions among participants. Understanding
the differences between these two theoretical perspectives is essential for understanding the role of central banks or other public entities in such activities.
The next section presents an abstract model of a network, gives some
possible payment system interpretations, and shows how the essential network
characteristics provide implications for network organizations. The following
sections discuss some of the elements of a more general (and complete) model
and apply some of the insights of the analysis to some historical and contemporary payment network issues. In particular, Section 3 discusses check clearing
prior to the founding of the Federal Reserve and current issues involving ATM
networks. In the former case, many observers have argued that check clearing was inefficient, as evidenced by the fact that checks sometimes followed
very indirect routes in proceeding from initial deposit to final clearing. In the
latter case, the use of surcharges (charged by an ATM-owning institution to
depositors from other institutions) has been cited as an inefficient exploitation
of market power in private networks. Section 3 will present the argument that
the empirical facts of both these cases are consistent with a theory that predicts
efficient private network organizations.

While network effects have often been said to be present in a variety of industries that are not explicitly networks, the focus here is on explicit networks.4
3 Sharkey (1985) and Henriet and Moulin (1996), for instance, follow this approach to the
theory of network structure and pricing.
4 For instance, some authors have suggested the presence of such an externality in the market
for personal computer operating systems; from a given set of alternatives, buyers prefer to have
the system that is more widely chosen. In this case, the externality comes from the indirect effect
that a system’s popularity has on the likely availability of application software. Indeed, one might
argue that a network effect is present in the retail grocery industry; the value to consumers of
shopping at a larger store (or chain) might be enhanced by the store’s ability to attract a wider set
of suppliers, thereby offering the shopper greater variety. For a general survey of network effects,
see Economides (1996).


Federal Reserve Bank of Richmond Economic Quarterly

In particular, the focus is on two-way networks where the underlying service
for which people have a demand involves transmissions between two particular
individuals. One can think of such a transmission as communication and the
underlying product as a message sent from one specific individual to another.
The model created below specifies the communication and consumption opportunities available to people in the economy.
An Economy with Consumption and Communication
To create an economic model of a network, imagine a set of individuals, each
of whom lives at a distinct location, separated from the others. One can imagine
any number of individuals, and in general one might denote the set of individual
agents as {1, 2, . . . , N}. The key insights from the analysis can be gained
from a simple example with three agents. Each individual derives utility from
sending messages to some subset of the other individuals. Again, there is a wide
array of possible patterns of desired communication. In particular, there may be
heterogeneity in the overall value agents place on communication. Some agents
may desire to send messages to a large number of recipients, while others may
wish to communicate with only a few. The pattern of desired messages that is
represented in Figure 1 has this feature. Agent 1 desires only to send a message
to agent 3. Agent 2 derives utility from sending messages to agents 1 and 3,
while agent 3 would like to send to agents 1 and 2. It is useful to assume
that, in addition to communication, agents receive utility from consumption of
a generic good and that each agent begins with an endowment of this good.
An agent’s preferences for consumption and communication can be stated
more formally as follows. Let Ji be the set of other agents with whom agent i
wishes to communicate. Hence, in Figure 1, J1 = {3}, J2 = {1, 3}, and J3 =
{1, 2}. For any set J, use the notation n(J) to denote the number of elements
in that set; for instance, n(J3 ) = 2. The agent receives utility v for each unit
of communication, and utility is linear in consumption. If xi denotes agent i’s
consumption of the generic good and Ui denotes the agent’s utility, then Ui =
n(Ji )v + xi .
A model must also specify the technology available for communication.
In particular, suppose there are two ways to communicate. A message can
be sent by direct, bilateral communication at a cost (to the sender) of c0 , in
units of the generic good. Alternatively, agents can buy access to a network. A
network is a set of “connected” agents. If an agent is connected to a network,
then messages to any other agent connected to the same network are costless.
The cost, again in units of the generic good, of connecting an individual to
a network is cs > c0 . A more general specification of network costs might
include a fixed (infrastructure) cost, a cost that is variable in the number of
agents connected (like cs per connected agent in the present specification) and
a cost per communication (assumed zero here). The essential feature that is

J. A. Weinberg: Private Payment Networks


Figure 1 A Simple Example of Demand for Network Services

Agent 1

Agent 3

Agent 2
An arrow from one agent to another indicates that the
first agent wishes to "send a message" to the second.

captured by the simple specification given here is that, by expending resources
to connect to one another, agents can reduce their costs per unit of communication. If we assume that the (utility) value of sending a message (v) is at
least c0 , and that the agents’ endowment of the consumption good is at least
2c0 , then each agent will send all desired messages even in the absence of a
network connection (the most any agent would spend on communication, if all
communication was bilateral, is 2c0 ).
An efficient network is one which includes all agents whose private and
external benefits of membership exceed the cost of connection and includes
no other agents. To state this definition formally, one needs some additional
notation. Let S be a possible network. That is, S is some subset of the agents.
This set might also be termed a coalition. The private value to individual i
of being a member of the connected set S depends on the number of other
agents in S with whom i wishes to communicate. Letting JiS denote this set,
the relevant number of connected agents for i is given by n(JiS ). Formally, JiS
is the intersection of the sets S and Ji . Network membership does not change
the actual amount of communication in which i engages; it is still worthwhile
to communicate bilaterally with all who are in Ji but not in S. Therefore, the
gross private benefit of membership (gross of connection costs) is the savings


Federal Reserve Bank of Richmond Economic Quarterly

in bilateral communication costs. For agent i, this value is n(JiS )c0, since agent
i wishes to send a single message to each of the other agents with whom he
In addition to the private value, agent i’s membership in S creates value for
other members of S. Define Hi as the set of agents who wish to communicate
with i. Hence, in Figure 1, H1 = {2, 3}, H2 = {3}, and H3 = {1, 2}. Agent
i’s membership in S creates value for all agents in Hi who are also in S. These
agents are denoted HiS , the intersection of the sets Hi and S. The external value
created by i’s membership in S is n(HiS )c0 . With the private and external values
of membership specified, one can state the following definition.
Definition 1: An efficient network is a set of agents S∗ such that5
(1) [n(JiS ) + n(HiS )]c0 ≥ cs for all i in S∗ , and
(2) [n(JiS ) + n(HiS )]c0 < cs for all i not in S∗ .

Condition (1) states that if agent i is in the network S∗ , then the private-plusexternal benefits from i’s membership equal or exceed the connection cost.
Condition (2) states that if private-plus-external benefits are less than the connection cost, then agent i is not in the network. Under the additional assumption
that 3cs < 5c0 (costs of universal connection are less than total communication
costs in the absence of a network), an efficient network for the example of
Figure 1 is one that includes all three agents.
Behavior of Agents in the Economy with an
Exogenous Organizational Structure
Given an economic environment such as the one just described, how does one
predict an outcome? In particular, what sort of network will emerge? Will it be
efficient? How will agents share the costs and benefits of network connection?
One approach to these questions is to assume a certain form of competition
among potential networks. That is, one might assume a particular game played
by the participants in the economy. To this end, it is useful to assume that
there is an additional set of agents who have access to the network technology
and derive utility only from the consumption good. These agents compete by
offering network services to the agents who have a demand for communication.
For simplicity, suppose that the services of these potential network providers
are incompatible; each provider’s network is unique and cannot be connected
to other providers’ networks. The “rules of the game” dictate the types of offers
5 To be precise, the notation in Definition 1 should specify the particular network S ∗ . For
instance, JiS should be JiS . When there is only a single network, as in Definition 1, suppressing
the extra notation introduces no ambiguity. Further, for this economic environment the assumption
of a single network is without loss of generality.

J. A. Weinberg: Private Payment Networks


the sellers are allowed to make as well as the allowed responses by network
One possible game through which network providers could compete proceeds as follows. First, each potential provider announces an access price and
stands ready to provide access to all comers at that price. Next, agents choose
whether and from whom to buy access. Finally, communication and consumption take place. This game will be referred to as the uniform price-setting game.
This name reflects the fact that sellers are not permitted to price discriminate by
offering different prices to different buyers. The predicted outcome of a game
is its Nash equilibrium, which is a set of strategies (price offers by providers
and network choices by buyers) such that each player’s (agent’s) strategy is
(privately) optimal, given the strategies of all the other players. Sellers in this
game essentially bid for the right to provide network services to all who sign
up. Since each potential provider is just as capable as all the others, price
competition will tend to drive profits to zero. With uniform price setting, this
competition can lead to extreme results, as in the following.
Result 1: In the equilibrium of the uniform price-setting game for the
environment described by Figure 1, no agent purchases network
access, so all communication is bilateral.
To see that this result is true, note first that the connection fee must be at least
cs , since sellers have no incentive to sell at a loss. Agent 1, however, will not
pay cs to join the network, since joining saves him at most bilateral costs of
c0 (if agent 3 is in the network). Without agent 1 in the network, connection
is not worth cs to the other agents, so the equilibrium network is empty.
This result is a stark example of the general finding that under certain
forms of competition equilibrium involves networks of inefficient size.6 For
the uniform price-setting game, such a result arises whenever there is at least
one agent for whom the greatest possible private benefits of connection are
less than the connection cost but for whom the private-plus-external benefits
exceed costs. Such an agent is one who has a relatively low personal demand
for communication but with whom many others wish to communicate (an agent
for whom n(Ji ) is small and n(Hi ) is large).
An Alternative Model of Behavior: Sustainability
Results like Result 1 are useful for understanding the nature of competition in
network markets. Their usefulness is limited, however, because they typically
apply to specific games. The variety of games that might be played is virtually
endless, even for the very simple economy of Figure 1. Would other games
yield different results? In particular, are there games for which the equilibrium

See Economides (1996) for a survey of such results.


Federal Reserve Bank of Richmond Economic Quarterly

network is efficient? The process of searching over all possible games is an
impractical approach to the economic analysis of such environments. An alternative is to look directly at possible outcomes (allocations of network services
and the consumption good) and ask which outcomes are sustainable, in a welldefined sense. In particular, suppose a certain allocation has been tentatively
agreed upon by the agents in the economy. That allocation is sustainable if
no subset of the agents can make themselves better off by allocating their
own endowed resources among themselves. In the present environment, such
a deviation from a proposed allocation would involve the subset of agents, or
coalition, forming its own network and communicating with all other agents
bilaterally; this coalition might also choose to reallocate its members’ endowments of consumption goods, net of connection costs.7
To state this sustainability property more formally, an allocation in this
economy can be defined as a network S and a payment for each agent, pi , toward
covering connection costs.8 In considering possible deviations from proposed
allocations by coalitions, it is sufficient to consider a coalition of any size that
deviates to form a single network; if there were an incentive for a coalition to
form two distinct networks, then each network would have its own incentive
to form, regardless of whether the other forms. Letting p = ( p1, p2 , . . . , pN ),
we can now define sustainability as follows.
Definition 2: An allocation (S, p) is sustainable if there does not exist
a network S and a payment allocation p such that

pi + n(Ji − S )c0 ≤ pi + n(Ji − JiS )c0 for each i in S , with
strict inequality for some i in S , and
i in S

pi ≥ n(S )cs .

In the above definition, the notation A − B for two sets A and B means all
elements of A that are not in B (the complement of B on A). The first condition
states that each agent in the deviating network S is made better off by joining
that network (and at least one is made strictly better off). An agent is made
better off if his total outlays for bilateral communication and connection payments are reduced. The second condition is that the network collect enough in
payments to pay for connecting all of its members.
One important fact to note about sustainability is that a sustainable allocation must be efficient. If it were not, a coalition consisting of an efficient
network could form and arrange an allocation satisfying the two conditions
above. In the case represented by Figure 1, a sustainable allocation will involve


A sustainable allocation is defined here as an allocation in the core of the economy.
In principle, one can allow for the existence of more than one network, although in this environment any allocation with more than one network is either equivalent to or (Pareto) dominated
by an allocation with a single network.

J. A. Weinberg: Private Payment Networks


a single network consisting of all three agents. The remaining task is to determine how the costs of connecting the agents (3cs ) should be shared among
the three. As we have already seen, charging each agent cs does not work;
agent 1 can be charged no more than c0 . The remaining cost (3cs − c0 ) must
be covered by charges to agents 2 and 3, with the restriction that neither can
be charged more than 2c0, the cost to each of sending all messages bilaterally.
Hence, one possible cost allocation is for agent 1 to pay c0 , agent 2 to pay 2c0 ,
and agent 3 to pay 3(cs − c0 ). These prices give no agent an incentive to leave
the network and send all communications bilaterally. The prices also leave no
incentive for any pair of agents to form a separate, two-agent network and
communicate with the third bilaterally. Hence, with these prices the efficient
network is sustainable. This result is summarized below.
Result 2: In the case of Figure 1, a sustainable allocation has an efficient
network S = {1, 2, 3} and any payments ( p1 , p2 , p3 ) satisfying
p1 ≤ c0 , p2 ≤ 2c0 , p3 ≤ 2c0 , and p1 + p2 + p3 = 3cs .
A Sustainable Allocation as the Outcome of a Game
Returning now to the question of competitive games played by potential providers of network services, does there exist a game under which the equilibrium
network is the efficient network? Suppose, as before, that a large number of
(incompatible) network service providers compete for subscribers. Instead of
requiring that competition be only in the form of nondiscriminating access
prices, suppose that the sellers can make any type of price offer they wish.
That is, price offers can be in the form of a distinct price for each buyer. Refer
to this game as the perfectly discriminating price-setting game. Equilibrium
prices for this form of competitive bidding correspond to the sustainable cost
allocations specified above.9
Result 3: Equilibria of the perfectly discriminating price-setting game are
sustainable allocations.
To see this, suppose a seller has offered a set of prices satisfying the conditions stated in Result 2. Can any other seller offer prices that win customers
from the first and yield a profit? In order to attract any individual buyer, a
competing offer must give that buyer a lower price. In order to cover costs,
however, the payment from at least one other buyer must be raised, so it is
impossible to attract more than one buyer. If all buyers do not join, network
connection is not attractive. Hence, a payment allocation satisfying the sustainability conditions cannot be undercut. On the other hand, any set of prices that
9 There are many equilibria corresponding to many core allocations. In all of them, agent 1’s
access price is no greater than c0 , agents 2 and 3 face prices no greater than 2c0 , and all agents
connect to the network.


Federal Reserve Bank of Richmond Economic Quarterly

leaves a seller with strictly positive profits can be undercut. Accordingly, the
sustainable allocations correspond with the set of equilibria.
The most notable feature of sustainable pricing arrangements is that, in
order to support an efficient network, they require the subsidization of agent
1’s connection by the other two agents; agent 1’s connection fee must be less
than the resource cost of connecting him. The other agents are willing to cover
the remainder of the cost, because agent 1’s (social) value to the network
exceeds the (private) value he places on network access. This example illustrates a general point about arrangements that support efficient networks in
environments in which agents are heterogeneous in the way they value network participation. Benefits (and possibly the costs) of network participation
have a collective component. The key to sustaining an efficient network is in
the distribution of these collective benefits and costs. This distribution must
respect the capability of agents to leave the network, either individually or in
groups. In a setting with heterogeneous agents, it can quite easily arise that the
appropriate distribution of costs and benefits require that different agents pay
different prices for essentially the same service.
The pricing arrangements that satisfy the conditions in Result 2 involve
perfect price discrimination; they require that prices be tailored for each individual buyer of network services. Perfect discrimination is not always feasible.
If, for instance, there is uncertainty about demands for network services and
an individual’s true demand characteristics are private information, then prices
cannot be as finely targeted as in the above example. In this case, private
information imposes further constraints on attainable allocations. It is possible
to incorporate such constraints into the notion of sustainable arrangements. In
such settings, pricing arrangements are likely to involve a less perfect form of
price discrimination. For instance, prices for network services may be tied to
observable characteristics or actions that are correlated with true demand. In an
environment similar to the one discussed in this section, access prices that vary
with the amount of communication might be able to achieve the desired price
discrimination in a way that allows privately informed buyers to self-select
among alternative pricing options.

The model and example of the previous section were specified in terms of a
generic communication service. The same sort of network structure, however,
can arise in a model that is specified in such a way as to capture important
aspects of payment system markets. Any noncash payment mechanism is a
communication network in a fundamental sense. An instrument presented in
payment for goods or services is an instruction to transfer monetary value from
the buyer’s to the seller’s ownership. Execution of such an instruction requires

J. A. Weinberg: Private Payment Networks


communication between the point of sale and the location or institution at which
the buyer’s value is held. This section sketches some of the ingredients of a
general payment network model. The key point is that the private and external
values to individuals of being connected to a payment network depend on the
underlying pattern of commerce.
An Economy with Payment Services
As in the above section, suppose that there is a set of N distinct locations
at which agents live and economic activity takes place. Unlike the previous
section, suppose that there is a large number of agents living at each location. Agents consume two types of goods: a generic good and location-specific
goods. Different people have different preferences for location-specific goods.
In particular, each agent desires the specific good from exactly one location.
One might, for instance, denote by φij the fraction of agents from location i who
wish to consume the specialized good at location j. These fractions determine
the economy’s pattern of commerce. Agents travel from their home location
to the locations at which they wish to consume and purchase location-specific
goods with claims on amounts of generic goods (or with the generic good
itself). In some environments, one might also imagine that these transactions
are made using government-issued fiat currency.
Making transactions across locations is costly. This cost might arise from
a number of frictions. If debt claims are created in the purchase of locationspecific goods, then there may be costs associated with communicating information about these claims across locations or in making final payment. If
buyers carry the generic good with them to make purchases, there may be
transportation costs or other losses incurred on the way. Finally, if traveling
for consumption takes time and buyers carry non-interest-bearing currency for
transaction purposes, then there may be a seigniorage cost associated with
location-specific consumption. The specific nature of the costs depends on the
details of the economic environment.
As above, suppose that there is a technology for connecting locations in
a network for the purpose of clearing and settling payment obligations. While
connection may involve a fixed cost, the variable cost of making transactions
among connected locations is lower than between locations that are not connected. For instance, if payment for location-specific goods requires shipment
of generic goods, a network that allows multilateral communication and calculation of net obligations may economize on shipping costs. Alternatively, in a
monetary economy, a network that allows people to substitute debt claims for
currency may allow agents to save on seigniorage costs.
The value to agents at a particular location of being connected to the
network depends on which other locations are connected. In particular, agents
at location i place a high value on being in a network that includes locations
j for which the fractions φij are large. By the same token, a location at which


Federal Reserve Bank of Richmond Economic Quarterly

many people want to shop ( j such that φij is large for many i) is one that brings
a high external value to any network it joins. Hence the notion of sustainable
pricing of network services, as presented in the previous section, will imply
that these popular locations receive preferential treatment in the pricing of
network services. As is the case for location 1 in the example of Figure 1,
there could easily be locations for which the private value of network services
is small while the external value is large. These would be locations that attract
many consumers from elsewhere but whose residents consume mostly their own
location-specific goods. Pricing to support an efficient network could require
that such locations pay less than the cost of connecting them, as is the case in
the example.
Payments-related models that have the features outlined in this section will
have the same implications as the example from the previous section. Network
industries will tend to be organized in ways that achieve sustainable allocations.
This means, first, that there is a tendency toward efficient network structures.
Second, the sharing of the costs and benefits of network organizations must
respect the ability of participants to form or join alternative organizations. Accordingly, network members who bring large external benefits to other members
may need to receive a share of the net benefits that appears to be out of line
with those members’ own use of the network services. Such an impression mistakenly focuses only on an individual’s private benefits of network participation
and not the benefits that an individual brings to other participants.
Barriers to Competition
An important maintained assumption in the foregoing discussion is that any
agent or group of agents that is dissatisfied with an arrangement is free to
pursue an alternative. That is, there are no barriers to entry. Various types of
barriers might arise in economic environments. For instance, if all agents do
not have access to the same technological capabilities, then it might be difficult
for agents that are dissatisfied with their network services to set up or seek out
an alternative network. Also, there may be investments in network provision
or participation that, once made, represent sunk costs. A sunk cost is a cost
that cannot be fully recovered. In this case, an incumbent network would have
a cost advantage over a competitor; while the competitor must incur the sunk
costs, those costs are no longer part of the incumbent’s decision calculus.
Other barriers to competition might arise from legal restrictions. For instance, if sellers were to face a legal prohibition of price discrimination, then
the types of network services arrangements they could offer would be sharply
limited; as seen above, price discrimination can be essential for the efficient
provision of network services with heterogeneous buyers. Other legal barriers
might take the form of restrictions on which particular sellers can offer which
particular services.

J. A. Weinberg: Private Payment Networks


A final form of potential barrier that is worth noting arises from the behavior of sellers themselves. An incumbent seller of network services might
attempt to impose rules on its buyers that make it difficult for them to switch to a
competing service. The possibility of restrictive rules set by network providers
has been a subject of interest in recent policy discussions concerning ATM
network mergers.
If there were barriers to competition, then an inefficient network structure
could persist. In such a case, can public policy intervention improve on private
market performance? The answer depends on the source of the barriers. In the
United States and other developed economies, enforcement of antitrust laws is
in part intended to guard against the anticompetitive use of restrictive rules by
sellers in their contracts with buyers. In cases where a barrier to competition
is the result of a legal restriction on the behavior of market participants, such
restrictions might have other public purposes. Here, as in the case where barriers
may have technological sources, it may be difficult or impossible for public
intervention to remove the barriers. In these cases, an incumbent provider might
extract monopoly rents, for instance, by inefficiently limiting network size.10
In such cases regulation of the pricing and product offerings of an incumbent seller might be useful in promoting network efficiency. Governments in
many economies have traditionally taken this approach to telecommunications

One can apply the logic presented above to a number of actual payment network
examples. This section will provide a brief discussion of two such examples,
one historical and one current. The historical example involves the process
of clearing checks in the United States in the period prior to the founding
of the Federal Reserve System. The current example involves the growing
geographical reach of multibank ATM networks.
Check Clearing before the Federal Reserve
By the late nineteenth century, checks had already become a dominant form of
payment. As the banking industry was highly fragmented, a significant portion
of all checks were interbank checks. Clearing of checks between the depositor’s
and the check writer’s bank occurred in one of a number of ways.11 A bank
10 Of course, even a monopoly immune to competition will not necessarily produce inefficient results. If the monopolist has sufficient ability to price discriminate, monopoly behavior can
approach full efficiency. In this case, the monopolist’s rents come at the expense of consumer
welfare but not at the expense of total welfare.
11 For a recent discussion of pre-Fed check clearing see Gilbert and Summers (1996). A
classic detailed account is found in Spahr (1926).


Federal Reserve Bank of Richmond Economic Quarterly

holding checks drawn on accounts at another bank could present the items
directly, in person, at the paying bank. By law, the paying bank was required
to make payment on such checks without imposing any presentment fee. On
the other hand, checks that were presented through the mail could be subject to
a fee imposed by the paying bank. Banks could also clear checks through the
services of an intermediary, or correspondent bank. Banks with a correspondent
relationship might have entered into a mutual agreement to accept checks from
one another without imposing fees.
At a time when bank branching was limited by law, correspondent relationships were particularly important for clearing checks in cases where the
paying bank was relatively distant from the bank in which the check was initially deposited. If both banks had correspondent relationships with the same
intermediary bank, then collection of the item could proceed free of fees. In
this sense, any two banks that were connected by a chain of correspondent
relationships belonged to a network. How was the composition of such networks likely to have been determined? The value to a bank of belonging to a
network depended on the frequency with which the bank received checks drawn
on other members of the network. Further, if there were particular institutions
that had relatively frequent and large volume interactions with many other
banks, then such institutions would naturally serve as central intermediaries in
a correspondent network. For instance, a city bank might deal with a number of
country banks in the surrounding region. The city bank might, in turn, maintain
relationships with banks in other cities that serve as correspondents for their
regions. This organization of a check-clearing network could economize on the
costs of shipping checks. A small bank in a remote town could, for instance,
send a single shipment of all its out-of-town checks to its correspondent, with
the links between larger correspondents serving as “trunk lines.”
In a correspondent network like that described above, consider the problem
faced by a bank that receives a check drawn on an institution with which it
rarely deals. The receiving bank could send the item directly to the paying
bank. In this case, however, the paying bank might charge a fee for presentment. Alternatively, the receiving bank could send the item, along with its usual
shipment, to its correspondent bank. Then, through a chain of correspondent
relationships, the check might ultimately be paid at its par value (with no
presentment fee). This indirect alternative has two potential sources of savings.
First, presentment fees might be avoided. Second, there may be savings on the
costs of transporting checks. The marginal cost of adding an item to a routine
shipment is virtually zero, certainly smaller than the postage cost of sending
a single item directly. Indeed, similar economies may have been available at
the receiving end of check shipments. A paying bank may have found it more
convenient and cost effective to receive and process bundles of checks sent by
intermediaries with whom it had a standing relationship.

J. A. Weinberg: Private Payment Networks


The Circuitous Routing of Checks
The history of check clearing during the period that preceded the founding of
the Fed contains examples of checks traveling over circuitous routes to get from
the banks in which they were initially deposited to the paying banks. A bank of
first deposit, for example, might have sent a check to its correspondent located
to the east even though the paying bank was located to the west. There are
two possible interpretations of such examples. On the one hand, such instances
might provide evidence of an inefficiency created by paying banks’ ability to
assess presentment fees. On the other hand, such cases could be consistent with
the operation of an efficient correspondent network. The pattern of links in the
network (correspondent relationships) was determined by the usual pattern of
commerce. The occasional circuitous route for check clearing resulted simply
because there were occasional exceptions to the usual pattern of commerce.
Given the existing links, it was efficient to send these occasional items together
with routine shipments. Indeed, in this view, it is possible that presentment fees
reinforced network efficiency by reducing the incentive for individual banks to
bypass the network.
The second interpretation is consistent with the analytical framework suggested above. Under this interpretation, presentment fees may have actually
reinforced efficient check-clearing relationships by discouraging the direct presentment of occasional, solitary items. Those few items that were sent directly
and on which presentment fees were paid are likely to have been items for
which the bank of first deposit could not foresee a sufficiently reliable chain of
correspondent relationships. That is, these were items for which the bank that
was due payment had little alternative to direct presentment (through the mail).
Accordingly, such items would constitute a market segment (in the market
for clearing by direct presentment through the mail) with relatively inelastic
demand. The efficiency cost of charging a high price (above marginal cost) to
market segments with inelastic demand is relatively small; with inelastic demand, quantity purchased does not decline much as the price is raised. In other
words, presentment fees allowed paying banks to price discriminate between
institutions with good alternatives to direct presentment and those without such
alternatives. It is entirely possible that the effects of such discrimination were
primarily distributional. While some buyers gain at the expense of others, price
discrimination typically increases the exploitation of gains from trade relative
to uniform pricing in the presence of market power.12
ATM Networks
An ATM transaction, like a check transaction, can require interbank clearing
and settlement. When the holder of an ATM card issued by one bank makes a
12 Each paying bank had some market power in the sense that it was the only bank that
could provide final settlement of a check.


Federal Reserve Bank of Richmond Economic Quarterly

withdrawal at an ATM owned by another bank (or perhaps by a nonbank business), the transaction must be cleared by communicating information about the
withdrawal to the card-issuing bank and settled with an appropriate transfer
of funds from the issuer to the ATM owner. In recent years, multibank ATM
networks have become increasingly important in allowing cardholders to access
their funds at ever widening sets of locations.13 A regional ATM network is
usually organized as a joint venture owned by some or all of its participating
banks. The network typically has a brand name that is placed on its members’
machines and cards.
Membership in an ATM network is valuable to a bank mainly because
access to the network’s set of ATM locations enhances the value of the ATM
services the bank offers its depositors. Clearly, membership is more valuable
the more extensive the network’s set of locations. A bank’s membership also
brings “external” benefits to other members by adding to the set of locations.
Some locations, however, are more valuable than others. For instance, banks
(and their customers) may place a high value on having access to an ATM at
a vacation resort. Such a location may be one for which the external benefits
of network participation are greater than the private benefits to the owner of
the particular ATM. The theory presented above suggests that a sustainable
network will need to price its services or share its profits so as to allow the
ATM owner to realize some of the external benefits of its participation.
There are a number of ways in which network arrangements could induce
participation of institutions that bring large external benefits. For instance, if
the network imposes membership fees, lower fees could be set for members
with more desirable locations. Similarly, if the network is organized as a joint
venture, then arrangements for profit sharing could conceivably reflect differences in the values of members’ locations. Since the values of locations are
likely to be related to the intensity of ATM use, prices based on transactions
might also be used for allocating network costs and benefits. In ATM networks,
when a cardholder from one member uses the ATM of another member, the
cardholder’s bank pays an “interchange” fee to the ATM owner. Since this fee
is typically set by the network and is usually a uniform, per-transaction charge,
its flexibility for cost and benefit sharing is limited. Owners might also be able
to realize some of the network benefits of desirable locations by imposing a
transaction fee directly on the cardholder. Cardholders’ willingness to pay such
fees, known as surcharges, would depend on the value of having access to
cash at the particular location and on the degree of competition for cash access
services at that location.

A description and history of ATM networks are found in Baker (1996).

J. A. Weinberg: Private Payment Networks


Many regional ATM networks, as well as the national networks owned by
the Visa and Mastercard associations, once had implicit or explicit agreements
among their members banning surcharges. These restrictions were ultimately
challenged, often by owners of ATMs at high-value locations.14 The debate over
surcharges centers on two opposing interpretations of the role such fees play
in the market. One interpretation holds that a ban on surcharges prevents ATM
owners from abusing the monopoly power they gain from having desirable
locations. The other interpretation is suggested by the arguments presented in
this article; surcharges support the formation of efficient networks by allowing
participants who bring large external benefits to the network to capture some
of those benefits. Under this second view, a ban on surcharges is an attempt
by network owners to impose a certain distribution of network benefits, a
distribution that may not be sustainable.
Does the second view imply that monopoly rents earned by ATM owners
with particularly advantageous locations do not create the inefficiency usually
associated with monopoly power? Not necessarily. If a network includes some
truly unique locations and there is no possibility of entry by a competitor at
those locations, then the unique locations are essentially natural monopolies.
Banning surcharges does not eliminate the rents from those monopoly positions. Rather, a ban is an attempt to spread those rents among all the banks
in the network. If a network includes valuable locations offering equal access
for customers of all network members but not for nonmembers, then members
can extract rents in the fees they charge their customers for account services
that include ATM access. In this case, the rents would be extracted from all
customers, even those with no demand for access to the highly valued locations.
Allowing surcharges, on the other hand, allows monopoly rents to be collected
in a more discriminatory fashion. Such price discrimination can have the effect
of reducing the inefficiency from monopoly power.
A second important point about monopoly rents is that an ATM owner’s
ability to extract rents is limited by competitors’ ability to enter monopolized
market segments. Hence, the primary public policy concern with market power
should be “How is it maintained?” rather than “How is it exploited?” Competitors’ incentives to enter markets by placing ATMs at particular locations
is greatest when owners can earn location-specific rents. Hence, a ban on surcharges creates a situation in which incentives to engage in location-specific
competition are muted.


See for instance Bank Network News, September 13, 1995.


Federal Reserve Bank of Richmond Economic Quarterly

The theoretical framework presented above suggests that the organization of
networks is driven by the desire of market participants to devise sustainable
multilateral arrangements. In both of the cases discussed in the previous section,
the framework leads one to interpret observed network structures and pricing
arrangements as components of an efficient arrangement. Such an arrangement
must take into consideration both the private value that participants derive from
the network and the external benefits that participants bring to the network.
Hence, in a world with heterogeneous demands for network services, price
discrimination and other means of “unevenly” distributing the net benefits of
network services can be essential for supporting efficient network structures.
Does this article’s framework necessarily imply that all actual networks
are efficient and that public intervention can never improve upon the economic performance of a private network? As discussed in Section 3 above,
this best-of-all-possible-worlds result holds strictly only when there are no
barriers that prevent groups of economic agents from pursuing the alternatives of their choice. The role of public policy, therefore, is to understand the
sources of barriers to such choice. There may be some cases in which barriers
are technological and cannot be overcome. In these cases, some regulation
of pricing practices might be called for. This argument, however, is nothing
more than the traditional justification of regulation of a natural monopoly. In
other cases, barriers may be created through the rules imposed by incumbent
providers of network services in an attempt to preserve market share. Public
intervention to eliminate such rules could be beneficial. This argument closely
mirrors traditional justifications for antitrust scrutiny of the conduct of firms
with large market shares. In all cases, the framework begins with a presumption
of efficiency. This presumption is expressed by the question, “If all economic
decisionmakers are always free to make alternative arrangements, why wouldn’t
the arrangement on which they actually agree be efficient?” This presumption
seems also to be a good place for public policy to begin.

Baker, Donald I. “Shared ATM Networks: the Antitrust Dimension,” The
Antitrust Bulletin, vol. 41 (Summer 1996), pp. 399–425.
Bank Network News. “New Math Renews Old Surcharge Debate,” September
13, 1995, p. 1.
Calomiris, Charles W., and Charles M. Kahn. “The Efficiency of Self-Regulated
Payment Systems: Learning from the Suffolk System,” Journal of Money,
Credit, and Banking, vol. 28 (November 1996), pp. 766–97.

J. A. Weinberg: Private Payment Networks


Economides, Nicholas. “The Economics of Networks,” International Journal
of Industrial Organization, vol. 14 (October 1996), pp. 673–99.
Gilbert, R. Alton, and Bruce J. Summers. “Clearing and Settlement of U.S.
Dollar Payments: Back to the Future?” Federal Reserve Bank of St. Louis
Review, vol. 78 (September/October 1996), pp. 3–27.
Henriet, Dominique, and Herve Moulin. “Traffic-Based Cost Allocation in
a Network,” RAND Journal of Economics, vol. 27 (Summer 1996), pp.
Sharkey, William W. “Economic and Game Theoretic Issues Associated with
Cost Allocation in a Telecommunications Network,” in H. Peyton Young,
ed., Cost Allocation: Methods, Principles and Applications. New York:
North-Holland, 1985.
Spahr, Walter. The Clearing and Collection of Checks. New York: The Bankers
Publishing Co., 1926.

The Case for a
Monetary Rule in a
Constitutional Democracy
Robert L. Hetzel


onstitutional democracy protects individual liberty. It does so by
placing restraints on the arbitrary exercise of power by government.
A primary restraint is the constitutional protection of property rights.
The monetary arrangements of a country either promote or undermine that
Money is unique in that its value in exchange far exceeds the cost of
producing an additional unit. On the one hand, governments have an incentive
to print additional money to gain “free” resources, or seigniorage revenues.1 On
the other hand, the central bank must limit the quantity of money in circulation
to control prices.
Through its influence on seigniorage, money creation affects how government raises revenue. It can also affect who within government decides how that
revenue is spent. Through its influence on fluctuations in the price level, money
creation influences the extent of arbitrary redistributions of wealth among
individuals. The institutional arrangements that govern the creation of money
then bear on two aspects of the protection of property rights: the taking and
disposition of wealth from the public and the distribution of wealth by government between individuals.

The opinions expressed herein are the author’s and do not necessarily represent those of the
Federal Reserve Bank of Richmond or the Federal Reserve System.

Under a commodity standard, the Mint had a monopoly over coinage. It charged a fee
called seigniorage for turning bullion into coins. Today, seigniorage is the general term for the
revenues a government gains from its monopoly over the creation of fiat money. The resources
commanded by a paper dollar (or its electronic equivalent bank reserves) far exceed the resources
needed to create that dollar. The central bank varies the amount of revenue it raises for the
government through seigniorage by varying the rate of money creation and, consequently, the
rate of inflation. Up to a point, higher inflation yields more revenue. When a central bank allows
high rates of money growth and inflation, seigniorage is commonly referred to as an “inflation
tax.” The holders of cash, who must add to their cash balances to restore the purchasing power
eroded by inflation, pay this tax.

Federal Reserve Bank of Richmond Economic Quarterly Volume 83/2 Spring 1997



Federal Reserve Bank of Richmond Economic Quarterly

A legislative mandate from Congress requiring the Federal Reserve (the
Fed) to stabilize the price level and to hold only government securities in its
portfolio would complement the rules in a constitutional democracy that protect
property rights.
A Financial Times (1990) editorial made the case for rules in the conduct
of monetary policy:2
The notion that money must fall within the domain of day-to-day politics is a
20th century heresy. . . . Painful experience with the modern manipulation of
monetary policy suggests that money is more appropriately an element of the
constitutional framework of democracy than an object of the political struggle.
Monetary stability is a necessary condition for a working market economy,
which is itself a basis for a stable democracy.

Sections 2 and 3 provide historical background showing that monetary arrangements in Britain and the United States have in the past raised issues basic to
the design of a constitutional democracy. Early British experience influenced
American thinking during the Revolutionary War, in which protection of property rights was a central issue. Section 3 reviews how during the period of
the Articles of Confederation the states’ discretionary control of money undermined property rights. The monetary abuses of this period contributed to the
convening of the Constitutional Convention.
Section 4 reviews the more recent experience with discretionary control
of inflation in the 1970s and argues that this experience produced the same
arbitrary redistributions of wealth that characterized the earlier period. The
earlier experience led the Founding Fathers to remove government discretion
over money creation by putting the United States on a specie (gold or silver)
standard. The later experience caused the Fed in practice to assign priority to
maintaining a low rate of inflation. However, in this more recent period, the
United States has not put in place institutional arrangements to ensure monetary stability in the future. Section 5 reviews fiscal transfers by the Fed made
possible through seigniorage and discusses how they can circumvent the constitutionally mandated budget process. Section 6 offers concluding comments.

The following historical review provides examples of how monetary arrangements either reinforce or relax constitutional constraints on the exercise of
power by government. Discretionary control over inflation can undermine the
2 For an economic argument in favor of rules, see Friedman (1960) and Lucas (1983). For
a constitutional argument, see Friedman (1962).

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


accountability provided for in a constitutional democracy. The power of a government to increase the revenue from seigniorage through inflation without
explicit legislation lessens democratic accountability. Moreover, the ability to
allocate seigniorage revenues discretionarily, that is, outside the constitutionally
mandated budget process, lessens accountability.
British Constitutional Democracy
The germ of constitutional democracy appeared in 1215 when British nobles
forced King John to sign the Magna Carta. The Magna Carta established the
principle that no one is above the law. Moreover, it articulated two principles
that bear on monetary arrangements. First, government should be divided into
separate parts capable of counterbalancing each other. Second, taxation “shall
be levied in our kingdom only by the common counsel of our kingdom.” The
practical working out of these principles required decisions about which part
of government would control and allocate seigniorage.
The Glorious Revolution in 1688 gave Britain its present parliamentary
form of constitutional democracy. The competition between Crown and Parliament for political power, which Parliament won, was driven in part by religion
and in part by Parliament’s desire to gain secure property rights. The desire
by Parliament to stop the arbitrary seizure of property by the Crown led to
the particular characteristics of British constitutional democracy that inspired
Americans in their Revolution. The exclusive right of Parliament to levy taxes
was especially important.
In the past, some English kings had debased the coinage as a way of
obtaining revenue. Despite protests by Parliament, Henry VIII had regularly
lowered the precious metal content of coins. Debasement of the currency gave
the Crown a source of income independent of Parliament. After the Glorious
Revolution, Parliament took over the Mint to foreclose debasement as a source
of revenue. When Parliament incorporated the Bank of England in 1694, it
prohibited the Bank from lending to the Crown without Parliament’s consent.
American Revolution
John Locke was the philosopher of the Glorious Revolution. He understood the
two major incentives to the arbitrary exercise of power: the seizure of political
power and the seizure of property. Locke ([1690] 1986, p. 180) wrote that
men “join in society with others . . . for the mutual preservation of their lives,
liberties, and . . . property.”
Locke included popular consent to taxes as a natural right. That principle,
expressed in the phrase “No Taxation without Representation,” became the rallying cry of the American Revolution. The Declaration of Rights in the Virginia
state constitution, written by George Mason and adopted in 1776, states that
“men . . . cannot be taxed or deprived of their property for publick uses, without
their own consent, or that of their representatives so elected” (Commager 1962,


Federal Reserve Bank of Richmond Economic Quarterly

p. 104). In the Declaration of Independence, Thomas Jefferson listed as one of
the “repeated injuries and usurpations” of King George his assent to “imposing
Taxes on us without our Consent.”
The arbitrary seizure of property by the British Crown became a major
factor precipitating the American Revolution. James Otis denounced writs of
assistance, which gave the Crown’s agents broad authority to search houses
and confiscate smuggled goods not restricted by “probable ground” or particular “houses specially named” (Commager 1963, pp. 46–47). The Constitution
provided for the protection of property rights in the Fourth Amendment, which
guaranteed the right of “people to be secure in their persons, houses, papers, and
effects against unreasonable searches and seizures,” and in the Fifth Amendment, which provided that no person shall “be deprived of life, liberty, or
property, without due process of law; nor shall private property be taken for
public use, without just compensation.” As explained below, the writers of
the Constitution viewed monetary arrangements as part of the constitutional
protection of property rights.

Abuse of Money Creation under the Articles of Confederation
In the period of the Confederation following the Revolutionary War, actual or
threatened civil disorder in several states, especially Massachusetts, prompted
calls for a convention to create an effective central government. The political
difficulty of imposing the excise taxes necessary to pay debts inherited from
the Revolutionary War caused many states to yield to pressure to repay debts
through the issue of paper money. Rhode Island effected a general transfer of
wealth from creditors to debtors through the inflationary issue of paper money
combined with legal tender laws forcing creditors to accept paper money rather
than specie. In a number of states, legislators used inflation in combination with
price controls to transfer income from creditors, merchants, and large planters
to farmers, debtors, and artisans. On August 7, 1786, James Madison (Rutland
1975, p. 89) wrote:
. . . the States are running mad after paper money, which among other evils
disables them from all contributions of specie for paying public debts, particularly the foreign one. In Rhode Island a large sum has been struck and
made a tender, and a severe penalty imposed on any attempt to discriminate
between it and coin. The consequence is that provisions are withheld from the
Market, the Shops shut up—a general distress and tumultuous meetings.

In a letter to Jefferson, Madison complained about the “warfare & retaliation” that arose among states due to laws allowing citizens of one state to pay
out-of-state debts in depreciated paper money (Rutland 1975, pp. 94–95). In

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


Federalist essay 10 (Beloff 1987, p. 47), Madison expressed his famous idea
that in a large republic “a greater variety of parties [would militate] against
the event of any one party being able to outnumber and oppress the rest.” One
reason was that in a large republic “a rage for paper money, for an abolition of
debts, for an equal division of property, or for any other improper or wicked
project, will be less apt to pervade the whole body of the union, than a particular
member of it” (Beloff 1987, p. 47).
The 1786 disturbance in Massachusetts called Shay’s Rebellion acted as
a catalyst for the convening of a Constitutional Convention. Armed farmers
closed courts to prevent foreclosures on properties for tax delinquency and
threatened to march on Boston to force passage of easy money legislation.
Such threats prompted the Constitutional Convention to protect property rights
by prohibiting states from issuing paper money, from making legal tender laws,
and from impairing the obligation of contracts. In Federalist essay 44 (Beloff
1987, p. 227), Madison defended the Constitution’s prohibition of the issuance
of paper money (then called bills of credit) by the states:
The extension of the prohibition to bills of credit must give pleasure to every
citizen in proportion to his love of justice and his knowledge of the true
springs of public prosperity. The loss which America has sustained since the
peace from the pestilent effects of paper money on the necessary confidence
between man and man; on the necessary confidence in the public councils;
on the industry and morals of the people, and on the character of republican
government, constitutes an enormous debt against the states.

Jefferson believed that the Constitution denied to Congress the “power of
making paper money or anything else a legal tender” (Lipscomb 1903, p. 65).
He wrote that paper money “is liable to be abused, has been, is, and forever
will be abused, in every country in which it is permitted” (Ford 1898, p. 416).
The historian Jack Rakove (1996, p. 44) argues that Madison formulated his
views on constitutional government in response to the arbitrary redistributions
of wealth caused by the paper money inflations and legal tender laws of the
states in the Articles of Confederation period.
Limiting the Power of Government to Seize Property Arbitrarily
The Founding Fathers carefully balanced the need for government to raise
revenue with the need to protect private property from arbitrary seizure. They
did so by entrusting fiscal policy to Congress, which by design encourages
open debate.3 That debate allows for the monitoring that gives substance to

Congress is composed of a large number of members. There are two houses of Congress,
each with its own subcommittees, committees, and rules committee. In order to become law, a
bill must pass through all these individual centers of power. Along the way, a bill can be killed by
a filibuster. A bill must also go through a conference committee to reconcile differences between
each house of Congress and then survive a possible presidential veto. This convoluted structure not only disperses power, but also enhances public discussion. The debate and controversy


Federal Reserve Bank of Richmond Economic Quarterly

the principle that sovereignty resides with the people. Tax legislation must
originate in the House, whose members are elected every two years. No funds
can be spent by the other branches of government without explicit authorization
by Congress. Article I, Section 9, of the Constitution states, “No money shall
be drawn from the Treasury, but in consequence of appropriations made by
law; and a regular statement and account of the receipts and expenditures of
all public money shall be published from time to time.” Article I, Section 8,
enforces congressional control of fiscal policy by allowing only Congress “to
borrow money on the credit of the United States.” To enforce the separation
of powers, however, a different branch of government, the Executive Branch,
spends the revenues raised by Congress.
In Article I, Section 8, of the Constitution, the Founding Fathers assigned
to Congress the responsibility “to coin money, regulate the value thereof.” By
assigning to Congress the responsibility to determine the metallic content of
coins, they prevented the President from following the example of European
kings and debasing the currency. Article I, Section 10, prohibited state governments from printing paper money. Taken together with Article I, Section
8, the Founding Fathers put the country on a specie standard. In this way, the
authors of the Constitution sought to limit the ability of government to abrogate
contracts by taking actions affecting the price level.
Monetary arrangements were part of the initial constitutional framework.
By removing money creation and the determination of the price level from the
political process, the Founders limited the ability of government to purposefully
redistribute wealth through inflation and inflation combined with price controls.
Those same restrictions also protected property rights by preventing recourse
by the government to the unlegislated tax of inflation.
However, the specie standard ultimately did not survive as part of the
constitutional framework. At times, the specie standard allowed for inflation
or deflation and was itself a source of instability. A mandate from Congress to
the Fed to stabilize the price level would possess the advantage of the specie
standard of eliminating discretionary control over the price level. It would
complement other protections afforded property rights. At the same time, such
a mandate would avoid the instabilities of the specie standard.
The argument for a rule requiring the Fed to stabilize the price level is not
one of original constitutional intent. And it is not a legal argument about the
constitutionality of discretionary as opposed to rule-based monetary arrangements. The Constitution does not require government to stabilize the price level.
Money is not a capital “C” constitutional issue involving a legal interpretation
of the Constitution.
generated by the multistep procedure of requiring repeated majorities among groups with widely
differing self-interests generates news and thus supplies the information citizens need to monitor
government. It also reduces the possibility of precipitate action.

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


However, the Constitution does provide broadly for the protection of property rights. A mandate from Congress to the Fed requiring price-level stabilization would buttress that broad constitutional protection. Such a rule is desirable
for the same reason that rules in general are desirable in a constitutional democracy: to constrain the arbitrary exercise of power by government. In this sense,
the sense used here, money is a small “c” constitutional issue referring to a
form of government where rules limit government power. (See the Appendix
for a brief history of money and the Constitution.)

Although monetary arrangements have disappeared as part of the constitutional
order, Congress remains constitutionally responsible for the value of the dollar.
While it has delegated that responsibility to the Fed, it has done so without
clear instructions on the desirable behavior of the price level.
The Inflation of the 1970s
In the 1960s and 1970s, the combination of pressures to make the economy
grow rapidly and a belief that inflation arose from factors unrelated to money
growth caused the Fed to pursue an inflationary monetary policy. Inflation
generated an unlegislated transfer of resources to the government through an
increase in seigniorage. More important than seigniorage, inflation generated
government revenue through its interaction with a tax code not indexed for
inflation. The arbitrary redistributions of wealth due to the inflation of this
period recall the earlier experience of the Articles of Confederation period.
Prior to indexing of the personal income tax brackets and exemptions in
1985, inflation combined with progressive tax rates to push individuals into
higher tax brackets. With inflation, capital gains taxes still fall not only on the
real gains from selling an asset, but also on the paper gains that compensate for
inflation. In addition, inflation raises corporate tax rates by lowering the real
value of historical depreciation costs and by raising the measured, but not real,
profits on holdings of inventories. Inflation also erodes the value of the estate
tax exemption. (See Feldstein [1996] and the Appendix in Hetzel [1990].) The
interaction of inflation and a tax code specified in current dollar terms created
enormous uncertainty and distortions in the 1970s. Corporations favored shortterm over long-term investment. Investors shifted capital out of the corporate
sector and into real estate in response to the effective rise in corporate tax rates.
Inflation allowed a shadow fiscal system where the combination of inflation and price controls transferred income to politically influential groups. Such
transfers were not subject to the safeguards provided by the public discussion
that accompanies explicit legislation to impose a tax. The housing industry
successfully lobbied for the use of Regulation Q to limit the interest rates


Federal Reserve Bank of Richmond Economic Quarterly

that depository institutions could pay on savings and small-denomination time
deposits. As a result, when inflation rose above Regulation Q ceilings, savers of
modest means received negative returns. Wealthy investors were unaffected as
they could invest directly in money market assets whose yields incorporated an
inflation premium. By not allowing the usury ceilings on the interest rates paid
on loans to rise with inflation, state governments discriminated against savers.
Rent controls, especially in New York and California, appropriated the wealth
of apartment owners. Price controls on domestically produced oil, instituted as
part of the Nixon price controls in early 1974 but kept until the early 1980s,
rewarded small, politically influential refiners.
Inflation and Scapegoating
When discussing the costs of inflation, economists concentrate on economic
costs, such as the increased number of trips individuals make to the bank to
economize on the holding of cash balances. The actual experience with the
inflation of the 1960s and 1970s, however, demonstrated that the problems
created by inflation can extend beyond the purely economic.
Inflation threatened the dispersion of power characteristic of the U.S. political system. It did so by creating a demand for immediate action to deal with
a problem that could only be solved through a long period of perseverance
and patience. Moreover, inflation created an incentive for the political system
to identify scapegoats and to deal peremptorily with them in ways that eroded
institutional safeguards to individual liberty. The resulting divisiveness came
on top of other divisive forces in American society arising from the Vietnam
War and the Civil Rights Movement.
Maintenance of the effective dispersion of power called for in the Constitution requires avoidance of the politics of scapegoating. The reason is that
the politics of scapegoating creates a demand for a strong leader who will
loosen institutional constraints so as to be able to deal decisively with the
offending minority. People tend to rationalize impersonal, threatening forces
seemingly beyond their control by blaming them on the actions of identifiable
minority groups. The rise of inflation in the late 1960s appeared as such a force.
Inflation added to the forces in the early 1970s that caused the middle class to
feel besieged by hostile forces beyond its control. The resulting appeal of the
politics of wage and price controls encouraged a majority to blame a minority
for inflation.
The origin of inflation lay in the high rate of growth money created by the
Fed. The political system, however, desirous of using monetary policy to promote rapid growth, found it advantageous to exploit the public’s presumption
that private parties create inflation through the exploitation of monopoly power.
That presumption led to a demand for wage and price controls. The controls,
imposed on August 15, 1971, carried the message that to control inflation

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


government had to prevent organized groups from selfishly demanding unreasonable shares of national income. The controls allowed government to make
scapegoats of organized labor, large corporations, and sectors of the economy
with politically sensitive prices such as food processors.
The combination of inflation with a fixed exchange rate created an overvalued dollar, trade deficits, and a demand for protectionism. The political
system then took advantage of worker resentment over imports and fears of
job losses to scapegoat foreigners. In the early 1970s, criticism of Japan became
a staple of the American political scene. To provide political balance for wage
controls, which were unpopular with leaders of organized labor, the Nixon
Administration imposed a surcharge on imports. Although popular with labor,
the unilateral adoption by America of an openly protectionist trade measure
threatened to precipitate a trade war and reversed the long-standing American
support for a multilateral system of world trade.4
The price controls of the 1970s were an extreme manifestation of the social costs of inflation. However, any time there is inflation, there is a political

4 The complexity of economic activity made inevitable the exercise of discretion in applying
the wage and price controls. That discretion allowed the arbitrary exercise of power by government
bureaucracy over individuals. During the initial freeze following the announcement of controls,
even individuals selling personal items in yard sales had to document the sale price of similar
items to avoid transgressing the law.
Because of the difficulty of policing a large complex economy and because of the need to
maintain political support, controls required widespread popular support. Popular support required
the appearance of fairness. For this reason, the controls entailed all kinds of destructive intervention in the economy unrelated to the behavior of the price level. For example, unions demanded
controls on interest and dividends. The general public demanded limitation on profit margins. The
government used the threat of IRS audits of profit margins to influence the price setting behavior
of corporations. The ideal in a constitutional democracy of limited government power enforced
through due process disappeared in the populist clamor to deal with powerful corporations and
Political pressures existed to make the controls permanent. The controls did not become
permanent in part because of the hostility of the Ford Administration and in part because they
did not work to prevent inflation. Without of course intending to, the Fed discredited them by
pursuing an inflationary monetary policy. If it had not, however, the public could have seen the
controls as working and they might have become a permanent feature of government control of
large corporations. Price controls on oil, which lasted through 1981, almost became permanent.
Those controls created an energy “shortage” and prevented America from producing the oil that
would have broken the OPEC price cartel. The resulting artificially high price of oil exercised
an important influence on international politics. American price controls meant that the Soviet
Union, one of the largest world producers of oil, and the oil-producing countries in the Middle
East gained the resources necessary to exercise significant power on the world scene.
Price controls criminalize socially useful activity. Inevitably, more and more individuals find
themselves breaking the law and staying out of the criminal system only through the forbearance
of a price control bureaucracy. The government can prosecute a large fraction of the population
at will or in response to political pressures to deal with an unpopular group. Controls erode the
general acceptance of law that allows a free society to function with a minimum of state coercion.
The average citizen is law abiding because he assumes most other citizens are also law abiding.
However, controls create the opposite presumption, that is, everyone else is breaking the law.


Federal Reserve Bank of Richmond Economic Quarterly

incentive to impose attenuated forms of controls through government price
fixing. That incentive arises from pressures on the political system to find
ways to redistribute wealth that do not require explicitly voted taxes. Although
the effects of inflation interacting with price controls in individual markets are
of a different magnitude from general controls, the corrosive effects are the
same. Individual vice (circumventing the laws that fix prices) becomes public
virtue. And respect for the rule of law erodes. (See Friedman in U.S. Congress,
6/21/73, p. 136.)
Bringing Monetary Arrangements into a Constitutional Framework
The contribution of inflation to the polarization of society and politics in the
1970s argues for bringing monetary policy into the constitutional framework
in a way that limits the ability of the central bank to inflate. Congress has not
established guidance for what constitutes desirable behavior of the price level.
Although the Federal Reserve Act instructs the Fed to pursue “stable prices,” it
also instructs the Fed to pursue “maximum employment” (Federal Reserve Act,
Section 2A). In practice, the absence of any instruction on how to combine the
pursuit of both these objectives has meant the lack of a meaningful mandate
for the behavior of the price level. The rise in the price level by a factor greater
than six between 1950 and 1995 amply demonstrates the ineffectiveness of the
formal mandate.
A congressional mandate to the Fed for price stability would provide an
institutional safeguard against a recurrence of the divisive experience with inflation in the 1960s and the 1970s. The more clearly stated the mandate, the
more understandable it would be to the public. Also, public opinion rallies
more easily behind a simple mandate. The clarity of a mandate to stabilize the
price level as opposed to the vagueness of a mandate to target a “low” rate of
inflation means that the former would be more likely than the latter to become
a permanent part of U.S. institutional arrangements.

Seigniorage as a Nonlegislated Tax
Money creation possesses implications not only for inflation, but also for the
level of taxation by transferring resources to the government.5 Moreover, money

For example, in 1974 (quarterly average CPI) inflation was 12.1 percent. Hetzel (1990, p.
53) estimates that federal government revenue was 17 percent higher than it would have been
with price stability. Much of the increased government revenue derived from the lack of indexing
of the tax code for inflation. The personal income part of the tax code was indexed for inflation in
1985. By reducing the incentive of government to inflate, that indexing constituted an important
institutional arrangement protecting against future inflation.

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


creation allows government to obtain resources without imposing an explicit
tax. Milton Friedman (1978, p. 27) wrote:
Since time immemorial, the major source of inflation has been the sovereign’s
attempt to acquire resources to wage war, to construct monuments, or for other
purposes. Inflation has been irresistibly attractive to sovereigns because it is
a hidden tax that at first appears painless or even pleasant, and, above all,
because it is a tax that can be imposed without specific legislation. It is truly
taxation without representation.

The seigniorage from money creation has implications for the U.S. constitutional system of limited government because of its potential for removing
fiscal policy from the recorded budget voted on by Congress. The way the
central bank handles seigniorage raises fundamental constitutional issues about
openness and the separation of powers in government.
As part of the separation of powers, the Constitution assigns to Congress
the power to tax and appropriate funds. As an institutional safeguard to keep
fiscal policy in the hands of Congress, the Founding Fathers assigned control
over the specie content of currency, and thus seigniorage, to Congress. Institutional arrangements legislated in the Federal Reserve Act also attempt to assure
that the government will not use money creation to evade the constitutional
requirement that Congress vote explicitly on taxes and appropriations.
These latter arrangements possess two aspects. One is to prevent the political system from making use of money creation as an unlegislated tax. The
Federal Reserve Act (Section 14: Open Market Operations, (b)(1)) authorizes
Fed banks “to buy and sell in the open market . . . any obligation . . . of the
United States.” The phrase “in the open market” implies that the monetization
that occurs when the Fed purchases debt should be undertaken solely to advance
monetary policy objectives rather than to alleviate the fiscal problems of the
Treasury. Specifically, the Fed should not buy Treasury debt directly from the
Treasury. It should not use the power of the printing press on request to turn
government debt into money. This aspect of independence concerns the size of
the Fed’s asset portfolio, the quantity of money, and ultimately the price level.
The second aspect of institutional arrangements designed to assure the conduct of fiscal policy through the public acts of Congress concerns the Fed policy
of transferring to the Treasury the income from the securities it holds (after
meeting its operating expenses and paying the statutory dividend to member
banks). In this way, the revenue from money creation appears as government
revenue. Consequently, that revenue can only be spent as part of the regular
appropriations process. Seigniorage revenue must be spent in ways that are
voted on by Congress and that appear on budget. As explained below, this
aspect of independence concerns the composition of the Fed’s asset portfolio.
(See also Goodfriend [1994].)


Federal Reserve Bank of Richmond Economic Quarterly

The Federal Reserve’s Fiscal Powers
The Fed creates money by purchasing securities from the public. The seigniorage value to the government is measured by the reduction in the stock of
securities held by the (nonFed) public or, equivalently, the increase in the stock
of securities in the Fed’s portfolio. For bookkeeping purposes, the value of this
seigniorage is measured by the flow of interest payments paid on the securities in the Fed’s portfolio. These bookkeeping arrangements are the heart of
the institutional arrangements that support Fed independence. The Fed achieves
budgetary autonomy from the political system by allocating to itself the amount
of interest it needs to meet its expenses. The remainder counts as tax receipts
of the government.
As part of the bookkeeping arrangements that buttress its independence,
government accounts treat the Fed as a member of the public. In order to
measure accurately the fiscal policy actions of the government, however, the
balance sheets of both the Treasury and the Fed should be consolidated. The
reason is that the Fed turns over to the Treasury the interest it receives on the
government securities it holds (above its costs). As far as the government is
concerned, interest paid on securities held by the Fed is essentially a wash. For
the purposes of fiscal policy, the key implication is that it makes no difference
whether the Treasury or the Fed sells a security. Either way, there is an increase
in the debt on which the federal government must pay interest financed by some
future increase in taxes or reduction in expenditure. In short, the Fed, like the
Treasury, can take fiscal policy actions. Consider the following examples.
Examples of Federal Reserve Fiscal Policy
The Fed could make discount window loans to an insolvent bank. For example,
in 1984 Fed loans to Continental Illinois Bank amounted to somewhat more
than $7 billion, 85 percent of the bank’s uninsured deposits. In conjunction
with such lending, the Fed sells government securities from its portfolio to
keep money and the price level unchanged. That is, it engages in a pure fiscal
policy action with no consequences for monetary policy. Government debt in
the hands of the (nonFed) public rises. Control over the composition of its asset
portfolio gives the Fed the ability to engage in fiscal policy; in this case, it is in
the form of credit allocation.6 (See Goodfriend and King [1988] and Schwartz
Consider next the direct monetization of Treasury assets that occurs when
the Fed acquires assets from the Treasury’s Exchange Stabilization Fund (ESF).
6 The credit allocation arises because of the nonmarket allocation of funds. In this example,
the Fed also transfers the liability for the insolvency from the uninsured depositors to the FDIC.
Because the premiums that banks pay to the FDIC go into general federal government revenues
and because the disbursements of the FDIC are government expenditures, the Fed transfers the
liability to the taxpayer.

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


These assets take the form either of SDRs or foreign exchange. 7 When the Fed
acquires assets from the ESF, it credits the Treasury’s deposit account at the
New York Fed. When the Treasury draws down its newly acquired deposits, the
revenues of the banking system increase. The Fed then sells U.S. government
securities out of its portfolio to offset that increase. The net result is to substitute either an SDR or an asset denominated in foreign exchange for a U.S.
government security in the Fed’s portfolio.8 Government securities held by the
(nonFed) public rise, and the Fed finances the activities of the ESF—foreign
exchange intervention and lending to foreign countries.
As a final example, consider a swap between the Fed and a foreign central
bank. For example, in a swap with the central bank of Mexico, the Fed accepts
peso deposits in exchange for dollar deposits. The Fed invests the pesos in a
peso-denominated security. When Mexico spends the dollars it receives, bank
reserves in the United States increase. The Fed then sells a U.S. government
security out of its portfolio to offset that increase. The net result is to substitute a peso-denominated security for a U.S. government security in the Fed’s
portfolio. Government securities held by the (nonFed) public rise. And the Fed
lends funds to Mexico.9
7 The International Monetary Fund periodically allocates to member countries special drawing rights (SDRs), which the ESF carries as assets.
8 In 1988 and 1989, the U.S. Treasury and the Fed engaged in coordinated sterilized foreign
exchange intervention with other central banks to counter strength in the dollar. In December
1987, the mark/dollar exchange rate was 1.6. In May 1988, the yen/dollar exchange rate was
125. By September 1989, the value of the dollar had risen so that 1.95 marks exchanged for
one dollar and the 145 yen for one dollar. In 1989, the administration became concerned about
the appreciation of the dollar given a large U.S. current account deficit. As a consequence, the
administration and the Fed began sterilized purchases of marks and yen.
The Fed and the ESF divided their purchases of yen and marks. When the ESF ran out of
dollars to sell, it obtained additional dollars from the Fed both through warehousing and through
monetizing the SDRs it held. In 1989, the Fed’s foreign-exchange-related transactions added about
$23 billion to reserves. That was more than the additions to currency that year, and the Fed sold
on net about $10 billion in government securities.
The SDRs on the books of the Fed increased from $5.0 billion at the end of 1988 to $8.5
billion at the end of 1989. In 1989, Fed warehousing of foreign currencies for the Treasury rose
from zero to $7 billion. Fed monetization of the SDRs held by the ESF increases the ESF’s
assets permanently as the ESF uses the dollars it acquires to acquire interest-bearing assets. (See
Schwartz [1997], especially Table 1.)
Figures on SDRs are from the Fed’s balance sheet reported in the Federal Reserve Bulletin.
Figures on Fed warehousing are from quarterly reports, “Treasury and Federal Reserve Foreign
Exchange Operations,” in the Federal Reserve Bank of New York Quarterly Review.
Broaddus and Goodfriend (1995) criticize such interventions for sending contradictory signals about the stance of monetary policy. In this case, the dollar was strengthening because the
FOMC had raised its funds rate peg to almost 10 percent in May 1989 to contain a rise in
inflation. By selling dollars to weaken the foreign exchange value of the dollar, the FOMC was
sending an opposite message about the desired stance of monetary policy from what it was sending
domestically by raising the funds rate. Kaminsky and Lewis (1996) make the same point.
9 For example, in September 1989, Mexico drew on its swap line with the Fed for $784.1
million dollars. At the same time, it drew on an ESF swap line for $384.1 million. Figures on


Federal Reserve Bank of Richmond Economic Quarterly

The Fed established swap lines with foreign central banks in 1962 to defend
the fixed exchange rate system without raising interest rates (Hetzel 1996). The
collapse of the fixed rate system in spring 1973 eliminated the original rationale
for swaps. The Fed, however, put them to another use. For instance, in 1973,
the administration asked the Fed to help Italy deal with the increase in its
balance of payments deficit in the aftermath of the large rise in oil prices. “The
Federal Reserve . . . came to the aid of Italy, whose chronic political instability
prevented rapid response to the energy crisis. The central bank expanded its
swap line with the Bank of Italy from $2 billion to $3 billion to help that
country finance imports in the short run” (Wells 1994, p. 125).
The use of swaps to provide short-term assistance to foreign countries
prompted a debate within the Federal Open Market Committee. In response
to a question from a governor about whether the Fed might provide longterm assistance to Italy, Chairman Burns (Board of Governors 1974, p. 783)
If the Federal Reserve were to abandon the principle that the swap lines were
available only to meet short-term needs, there would be a natural tendency for
other agencies of Government to look to the System, rather than to Congress,
for the resources to deal with a broad variety of international financial and
political problems. If the System were to provide those resources it would, in
effect, be substituting its own authority for that of the Congress. A decisive
case could then be made in support of the charge that the System was using
Federal moneys without regard to the intent of the Congress.

Limiting the Federal Reserve’s Ability to take Fiscal Actions
Congress has delegated to the Fed the right to exercise the public monopoly on
the creation of money. Through creation of money, the Fed acquires a portfolio
of government securities. Although this portfolio arises out of the conduct of
monetary policy, it allows the Fed to undertake fiscal policy actions independent
of Congress, as illustrated above.
Fiscal policy actions taken by the Fed are not subject to the open public
debate generated by congressional actions. Therefore, they limit the government accountability that is encouraged by the free flow of information. The
Fed should avoid fiscal policy actions not integrally related to its monetary
responsibilities. A restriction that permitted the Fed to hold only government
securities and to acquire them only in the open market would achieve this result.
swap line drawings are from quarterly reports, “Treasury and Federal Reserve Foreign Exchange
Operations,” in the Federal Reserve Bank of New York Quarterly Review.
The Fed does not lose interest on the assets in its portfolio as it receives interest on the
peso-denominated assets. The basic point is that the Fed can engage in the loan transaction with
Mexico because of its control over seigniorage. That is, the ability to create and extinguish fiat
money allows it to purchase peso-denominated assets.

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


Fed independence would then complement the constitutional provision: “No
money shall be drawn from the Treasury, but in consequence of appropriations
made by law.”
On the most general level, the issue is preservation of the constitutional
safeguards that give content to popular sovereignty. Those safeguards facilitate the monitoring of government by the public. Because inflation imposes
an unlegislated tax, discretionary control of inflation reduces the monitoring
ability of the public. Also, the use of seigniorage revenues by the central bank
for purposes other than financing its own operation reduces the public’s ability
to monitor government activities by limiting public discussion. If the use of
the central bank’s seigniorage revenues is directed by the Executive Branch, it
erodes the separation of powers provided for in the Constitution.

Balancing Independence and Accountability
Central bank independence can help to prevent monetary policy from becoming
subservient to fiscal policy. Preventing that subservience is an important part
of facilitating the monitoring of government by the public. A central bank ultimately controls only money creation. A legislative mandate for price stability
would limit political pressure to use money creation to achieve goals that may
be socially desirable but beyond the reach of a central bank. Although money
creation does not create real resources, it can impose a tax. A price rule would
help keep government finance honest.
An independent central bank also needs to be accountable. The Fed chairman does testify regularly before congressional committees. However, there is a
tension between the accountability provided by congressional oversight and Fed
independence. The budgetary benefits of a strong economy have, in the past,
encouraged some congressmen to pressure the Fed for low interest rates and, in
effect, inflationary money growth. A mandate requiring the Fed to stabilize the
price level would minimize this pressure, enhance independence, and provide
a clear standard with which Congress could assess the Fed’s performance.10
In a democracy, the legitimacy of central bank independence must rest on
a public belief that the central bank is accountable. That belief derives from
open debate encouraged by a transparency of the objectives of monetary policy.
Independence combined with discretion impairs accountability and encourages
political pressures that threaten the Fed’s independence. Independence combined within a rule mandating price stability would balance independence and
10 A mandate to stabilize the price level is not a complete rule unless accompanied by an
explicit strategy. If Congress were to impose such a mandate, it would probably allow the Fed to
select the strategy but require it to make that strategy explicit.


Federal Reserve Bank of Richmond Economic Quarterly

Rule of Law, Not Men
The Constitution originally provided for a commodity monetary standard. In
doing so, it restricted the power of government significantly: the government
could not manipulate the price level. In the twentieth century, the disappearance of the international gold standard that began with World War I led to a
fiat money standard. The long, painful process of learning how to manage a fiat
currency, where government sets the price level, has influenced significantly
the history of the twentieth century (Friedman and Schwartz 1963; Goodfriend
At present, monetary arrangements are working well. However, monetary
policy depends too much on the good luck of having wise policymakers. Widespread public support for a clear rule to guide the conduct of monetary policy
would provide for a continuation of the current period of monetary stability. The
major disasters of monetary policy in the twentieth century—the depression of
the 1930s and the inflation of the 1970s—derived largely from a lack of public
understanding that the central bank is responsible for the price level. A mandate
clearly assigning responsibility for the price level to the Fed and requiring the
Fed to stabilize the price level would prevent future major mistakes in monetary
A rule that required the Fed to stabilize the price level and that eliminated
its discretion over the use of seigniorage by requiring it to hold only government securities in its portfolio would complement the constitutional framework
that constrains the exercise of government power. Such a rule would protect the
public from the arbitrary redistributions of wealth that accompany unanticipated
inflation and the interaction of sustained inflation with price controls. It would
also prevent the political system from imposing an unlegislated tax in the form
of inflation and assure that seigniorage is spent only as part of the congressional
appropriations process.
The balance between rules and discretion in the exercise of power by
government is the central issue in a constitutional democracy. That issue is
also central to the design of a country’s monetary institutions. Monetary institutions should be based on rules that are thought of as part of the broad
constitutional framework. To protect property rights and the ability of citizens
to monitor government, those rules should constrain the use of seigniorage and
the recourse to an inflation tax.

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


At the Constitutional Convention, members frequently cited the issuance of
paper money by the states as a major abuse of power by government and a
source of civil discord. For example, Governor Morris (Farrand 1911, Vol. 2,
p. 299) talked about “the history of paper emissions . . . with all the distressing
effects (of such measures) before their eyes.” Madison (Farrand 1911, Vol. 1,
p. 317) stated:
He considered the emissions of paper money . . . as also aggressions. The
States relatively to one another being each of them either Debtor or Creditor;
The Creditor States must suffer unjustly from every emission by the debtor
States. We have seen retaliating acts on this subject which threatened danger
not to the harmony only, but the tranquillity of the Union.

In making the case for the checks and balances of a federal form of government, James Madison and others pointed to the overissue of paper money
as an example of abuse of unrestrained government power. Madison (Farrand
1911, Vol. 1, pp. 134–36; Vol. 2, pp. 76–77) argued that the principal reasons
for a national government were to provide
. . . for the security of private rights, and the steady dispensation of justice.
Interferences with these were evils which had more perhaps than anything
else produced this convention. . . . All civilized Societies would be divided
into different Sects, Factions, & interests, as they happen to consist of rich &
poor, debtors and creditors. . . . In all cases where a majority are united by a
common interest or passion, the rights of the minority are in danger. . . . We
have seen the mere distinction of colour made in the most enlightened period
of time, a ground of the most oppressive dominion ever exercised by man
over man. . . . The only remedy is to enlarge the sphere, & thereby divide the
community into so great a number of interests & parties that in the 1st place
a majority will not be likely at the same moment to have a common interest
separate from that of the whole or of the minority; and in the 2d. place, that
in case they shd. have such an interest, they may not be apt to unite in the
pursuit of it.
Emissions of paper money, largesses to the people—a remission of debts
and similar measures, will at sometimes be popular, and will be pushed for
that reason. . . . it is necessary to introduce such a balance of powers and
interests, as will guarantee the provisions on paper. Instead therefore of contenting ourselves with laying down the Theory in the Constitution that each
department ought to be separate & distinct, it was proposed to add a defensive
power to each which should maintain the Theory in practice.

Alexander Hamilton (Farrand 1911, Vol. 1, p. 288) argued:
In every community where industry is encouraged, there will be a division of
it into the few & the many. Hence separate interests will arise. There will be


Federal Reserve Bank of Richmond Economic Quarterly
debtors & Creditors &c. Give all power to the many, they will oppress the
few. Give all power to the few they will oppress the many. Both therefore
ought to have power, that each may defend itself agst. the other. To the want
of this check we owe our paper money.

Under the Articles of Confederation, Congress had the power to “emit
bills of credit,” that is, issue paper money. Governor Morris moved to omit
that power from the powers assigned to Congress by the Constitution. Several
delegates objected:
Col Mason had doubts on the subject. . . . Though he had a mortal hatred to
paper money, yet as he could not foresee all emergencies, he was unwilling
to tie the hands of the Legislature. He observed that the late war could not
have been carried on had such a prohibition existed.
Mr. Randolph, notwithstanding his antipathy to paper money, could not agree
to strike out the words, as he could not foresee all the occasions that might

In opposition,
Mr. Elseworth thought this a favorable moment to shut and bar the door
against paper money. The mischiefs of the various experiments which had
been made were now fresh in the public mind and had excited the disgust of
all the respectable part of America. By withholding the power from the new
Governt. more friends of influence would be gained to it than by almost any
thing else—Paper money can in no case be necessary—Give the Government
credit, and other resources will offer—The power may do harm, never good.
Mr. Wilson. It will have a most salutary influence on the credit of
the U. States to remove the possibility of paper money. This expedient can
never succeed whilst its mischiefs are remembered. And as long as it can be
resorted to, it will be a bar to other resources.
Mr. Butler . . . was urgent for disarming the Government of such
a power. . . . Mr. Read thought the words, if not struck out, would
be as alarming as the mark of the Beast in Revelations.
(Farrand 1911, Vol. 2, pp. 309–11.)

On the vote to omit “and emit bills of credit” from the Constitution, that
is, to give Congress the power to issue paper money, nine of the delegates
voted in favor and two voted against. In a later letter to Timothy Pickering,
Robert Morris, who had been a delegate, wrote, “Propositions to countenance
the issue of paper money, and the consequent violation of contracts, must
have met with all the opposition I could make. . . . to the best of my recollection, this was the only part which passed without cavil” (Farrand 1911, Vol. 3,
p. 419).
In 1862, Congress first authorized the issuance of paper money as legal tender (Greenbacks) as an expedient means of financing the Civil War. Timberlake
(1993, p. 143) wrote:

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


Up until the time of the Civil War, almost no one had seriously considered
interpreting the money clauses in the Constitution in any light except that of
prohibiting state and federal issues of currency on the basis of discretionary
authority. “To coin money” meant to provide the technical facilities for minting
coins. “Regulate the value thereof” meant only to specify a weight of fine gold
or silver as equal to a number of the units of account, which were dollars.

Eventually, however, the Supreme Court decided in favor of the constitutionality of Greenbacks and the issuance by the federal government of paper
money. At the time, these legal tender decisions were highly politicized (Timberlake 1993, pp. 133–45). President Grant appointed to the Supreme Court
individuals who favored the constitutionality of Greenbacks as legal tender.
Ultimately, the government issuance of paper money as legal tender was made
inevitable by the change in the prevailing interpretation of the Constitution. The
government’s power to issue paper money became inevitable with the demise
of the view that Congress possesses no power not expressly granted to it by the
Constitution and the emergence of the view that Congress possesses all powers
not explicitly denied to it.
In the final legal tender case, Julliard v. Greenman of 1884, the Supreme
Court decided that Congress had the power to issue paper money and make it
legal tender in peacetime as well as wartime (Timberlake 1993, p. 137). The
issuance of paper money then ceased to be a Constitutional issue (capital C)
in the sense of an issue decided by the Supreme Court. It remains, however,
a constitutional issue (small c) because of the role of seigniorage as a tax and
the insecurity of property rights engendered by inflation.

Beloff, Max, ed. The Federalist. New York: Basil Blackwell, 1987.
Board of Governors of the Federal Reserve System. Minutes of the Federal
Open Market Committee. 1974.
Broaddus, J. Alfred, Jr., and Marvin Goodfriend. “Foreign Exchange Operations
and the Federal Reserve,” Federal Reserve Bank of Richmond 1995 Annual
Commager, Henry Steele. Documents of American History, 7th ed. New York:
Appleton, Century, Crofts, 1963.
Farrand, Max. The Records of the Federal Convention of 1787, Vols. 1– 4.
New Haven: Yale University Press, 1911.


Federal Reserve Bank of Richmond Economic Quarterly

Federal Reserve Bank of New York. “Treasury and Federal Reserve Foreign
Exchange Operations,” Federal Reserve Bank of New York Quarterly
Review, vol. 14 (Winter 1989–90), pp. 69–74.
Feldstein, Martin. “The Costs and Benefits of Going from Low Inflation
to Price Stability,” in Christina D. Romer and David H. Romer, eds.,
Reducing Inflation: Motivation and Strategy. Chicago: University of
Chicago Press, 1997.
The Financial Times. “P¨ hl Throws a Gauntlet,” January 23, 1990, p. 16.
Ford, Paul L., ed. The Writings of Thomas Jefferson, Vol. 9. New York: G. P.
Putnam’s, 1898.
Friedman, Milton. Tax Limitation, Inflation and the Role of Government.
Dallas: The Fisher Institute, 1978.
. “Should There Be an Independent Monetary Authority?” in
Leland Yeager, ed., In Search of a Monetary Constitution. Cambridge,
Mass.: Harvard University Press, 1962.
. A Program For Monetary Stability. New York: Fordham University Press, 1960.
, and Anna J. Schwartz. A Monetary History of the United States,
1867–1960. Princeton: Princeton University Press, 1963.
Goodfriend, Marvin. “Monetary Policy Comes of Age: A 20th Century
Odyssey,” Federal Reserve Bank of Richmond Economic Quarterly, vol.
83 (Winter 1997), pp. 1–22.
. “Why We Need An ‘Accord’ for Federal Reserve Credit Policy,”
Journal of Money, Credit, and Banking, vol. 26 (August 1994), pp. 572–
, and Robert G. King. “Financial Deregulation, Monetary Policy,
and Central Banking,” Federal Reserve Bank of Richmond Economic
Review, vol. 74 (May/June 1988), pp. 3–22.
Hetzel, Robert L. “Sterilized Foreign Exchange Intervention: The Fed Debate
in the 1960s,” Federal Reserve Bank of Richmond Economic Quarterly,
vol. 82 (Spring 1996), pp. 21– 46.
. “A Mandate for Price Stability,” Federal Reserve Bank of
Richmond Economic Review, vol. 76 (March/April 1990), pp. 45–55.
Kaminsky, Graciela L., and Karen K. Lewis. “Does Foreign Exchange Intervention Signal Future Monetary Policy?” Journal of Monetary Economics,
vol. 37 (April 1996), pp. 285–312.
Lipscomb, Andrew A., ed. Writings of Thomas Jefferson, Vol. 10. Washington:
The Thomas Jefferson Memorial Association, 1903.

R. L. Hetzel: Monetary Rule in a Constitutional Democracy


Lucas, Robert E., Jr. “Rules, Discretion, and the Role of the Economic
Advisor,” in Robert E. Lucas, Jr., Studies in Business-Cycle Theory.
Cambridge, Mass.: The MIT Press, 1983.
Rakove, Jack N. Original Meanings, Politics and Ideas in the Making of the
Constitution. New York: Alfred A. Knopf, 1996.
Rutland, Robert A., ed. The Papers of James Madison, Vol. 9. Chicago:
University of Chicago Press, 1975.
Schwartz, Anna J. “From Obscurity to Notoriety: A Biography of the Exchange
Stabilization Fund,” Journal of Money, Credit, and Banking, vol. 29 (May
1997) pp. 135–53.
. “The Misuse of the Fed’s Discount Window,” Federal Reserve
Bank of St. Louis Economic Review, vol. 74 (September/October 1992),
pp. 58–69.
Timberlake, Richard H. Monetary Policy in the United States. Chicago:
University of Chicago Press, 1993.
U.S. Congress. “How Well Are Fluctuating Exchange Rates Working?”
Hearings before the Subcommittee on International Economics of the Joint
Economic Committee, 93 Cong. 1 Sess., June 20, 21, 26, and 27, 1973.
Wells, Wyatt C. Economist in an Uncertain World: Arthur F. Burns and the
Federal Reserve, 1970–78. New York: Columbia University Press, 1994.

Tax Disincentives to
Commercial Bank Lending
Anatoli Kuprianov


he Tax Reform Act of 1986 made sweeping changes to the U.S. tax
code. It lowered statutory tax rates on both corporate and personal income while eliminating the investment tax credit and a host of other
specialized tax deductions in an effort to ensure that all firms paid similar tax
rates. The act devoted special attention to commercial banks. Studies commissioned by Congress had found that the commercial banking industry paid
much lower average tax rates than most other firms, reinforcing a perception
that banks enjoyed many unfair tax advantages. With passage of the Tax Reform Act, the industry lost many tax preferences it had previously enjoyed.
Available evidence suggests that tax reform achieved its goal, at least insofar
as the commercial banking industry is concerned: average tax rates paid by the
U.S. banking industry rose from 24 percent in 1986 to 41 percent in 1995.
Some of the tax preferences banks lost under tax reform originally had been
intended to offset the costs of implicit taxes such as the non-interest-bearing
reserve requirements banks are obligated to hold with the Federal Reserve (the
Fed) as well as the cost of other regulations (Neubig 1984). Under the current
tax code, banks face the same treatment as all other financial intermediaries
but are still subject to the aforementioned costs. Moreover, Henderson (1987)
found that the cost of reserve requirements had not been offset by implicit
subsidies associated with the banking charter, such as access to the discount
Banks have long argued that the costs of reserve requirements and other
burdensome regulations put them at a competitive disadvantage relative to other
financial intermediaries. This assertion has received some support from McCauley and Seth (1992), who found that foreign banks had gained a 45 percent
share of the U.S. commercial and industrial loan market by 1991 and attributed
this trend to the burden of reserve requirements imposed on U.S. banks.
Mike Dotsey, Jeff Lacker, and Ned Prescott provided helpful comments on earlier drafts of
this article. Special thanks go to Leigh Ribble of the Board of Governors staff, who made
available the data needed to complete the study. Any remaining errors or omissions are the
author’s. The views expressed are those of the author and do not necessarily represent those
of the Federal Reserve Bank of Richmond or the Federal Reserve System.

Federal Reserve Bank of Richmond Economic Quarterly Volume 83/2 Spring 1997



Federal Reserve Bank of Richmond Economic Quarterly

Deposit insurance premiums levied on U.S. banks also may be viewed as
an implicit tax in certain circumstances. Congress recently enacted legislation
requiring commercial banks to help pay for the recapitalization of the thrift industry’s deposit insurance fund through a special deposit insurance surcharge.
If they are fairly priced, deposit insurance premiums represent a cost of doing
business and not a tax. But when a surcharge is imposed to fund other purposes,
deposit insurance premiums may constitute a tax on banks. One interesting
question, then, is how significant this tax is in relation to the overall effective
tax rate on banks.
These observations raise fundamental issues about the effects of explicit
and implicit taxes on the financial system. According to the U.S. Flow of Funds
Accounts, the commercial banking industry’s share of total credit extended in
the United States has fallen steadily in recent years, from almost 45 percent of
all credit market assets in 1952 to 22 percent in 1995. While financial innovation is most often blamed for this process of disintermediation, it is worth
examining whether the tax and regulatory policies may have contributed to this
This study takes a first step toward analyzing the burden of U.S. bank tax
and regulatory policies by developing a comprehensive measure of the overall
marginal effective tax rate on commercial bank intermediation. Economists
have long recognized that average tax rates do not provide a good measure of
the tax disincentives to investment. Most contemporary studies focus instead on
the marginal effective tax rate, which measures the marginal tax on investment
returns. Studies of bank taxation have been a notable exception to this rule—
existing studies have sought to measure the impact of tax reform by estimating
average effective tax rates. None of these studies has examined the marginal
tax rate on bank lending. Such an exercise turns out to be worthwhile, as it
produces some surprising results. In particular, it finds that the behavior of the
average effective tax rate has not been a good indicator of the tax disincentives
to commercial bank lending.
The discussion that follows begins in Section 1 with an examination of
the conceptual issues associated with measuring effective tax rates. Section 2
develops a financial model of banks that can be used to estimate an effective
tax rate on commercial bank intermediation. Empirical results are presented in
Section 3. The final section reviews the conclusions of the analysis.

An effective tax rate is a summary statistic that measures the tax burden associated with an activity. Tax codes stipulate not only a statutory tax rate, but also
a set of rules for calculating taxable income. These rules often do not yield a

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


measure of true economic income, however. For this reason, the effective tax
rate on investment can differ substantially from the statutory tax rate.
Effective tax rates are sometimes used to measure the impact of taxes on
incentives. Many studies compute an average effective tax rate, defined as actual
taxes paid divided by capital income, to provide a summary statistic of the tax
burden on a particular firm or industry.1 Fullerton (1984) gives several reasons
why average effective tax rates may not accurately measure the disincentives to
investment created by the tax code, however. First, relying solely on U.S. taxes
paid by a corporation ignores foreign taxes paid by the firm. Second, profits
measured for tax purposes differ from profits measured for financial reporting.
Third, taxes paid in a given year might not be related to actual profits earned
that year due to carryforwards of previous losses and tax credits. Finally, profit
measures typically used for calculating average effective tax rates are broken
down by firm or by industry rather than by asset class. Thus, while average
effective tax rates may be appropriate for measuring cash flows and distributional burden, they do not necessarily measure the disincentives to investment
inherent in the tax code.
More recent research on the incentive effects of taxation has focused on
the “marginal effective tax rate,” which measures the extra tax resulting from
a hypothetical marginal investment by a firm in a given industry. A marginal
effective tax rate measures the wedge between the marginal social return to
a capital asset and the rate of return earned by the investors who finance its
purchase. This wedge can be viewed as a measure of the disincentives to
investment created by the tax code.
Thus, marginal effective tax rates are better suited to capture disincentives
to investment. Moreover, average tax rates do not reflect the burden of implicit
taxes such as reserve requirements or deposit insurance surcharges.2 When
Fullerton and Henderson (1985) compare average and marginal effective tax
rates for 18 industries, they find almost no correspondence between the two
The effective tax rate methodology can be extended to include personal
taxes paid on dividends and interest as well as corporate taxes. Measures of
the effective tax rate that include personal income taxes can be used to analyze the effect of taxation on the intertemporal allocation of resources. If one
is interested in the allocation of capital among competing uses, however, or
among competing firms engaged in similar activities but subject to different
tax treatment—such as comparison of the tax disincentives to lending between
1 Harberger’s (1966) classic article on the efficiency effects of capital income taxes uses this
2 A notable exception here is Henderson (1987), who incorporates the cost of various implicit
taxes into a comprehensive measure of average effective tax rates.


Federal Reserve Bank of Richmond Economic Quarterly

commercial banks and other financial intermediaries—then consideration of
personal taxes is unnecessary.3
As noted in the introduction, previous studies on the taxation of commercial
banking have focused on the impact of changes in tax laws on total taxes paid
by commercial banks. 4 While such studies can be useful in evaluating the tax
burden borne by the industry, the foregoing discussion suggests that they may
not be useful in measuring the disincentives to traditional forms of bank credit
intermediation created by corporate taxes and reserve requirements. This study
differs from other studies on the taxation of commercial banking in that it
estimates the marginal effective tax rate on commercial bank intermediation.
Using this model, one can also estimate the disincentives to commercial bank
intermediation inherent in regulations such as reserve requirements or deposit
insurance surcharges. Since the analysis focuses only on the distributional impact of explicit and implicit taxes, it ignores personal taxes paid by households.
Measurements of marginal effective tax rates are typically derived using the
“user cost of capital” methodology developed by Hall and Jorgenson (1967).
Hall and Jorgenson’s measure of user cost reflects not only the financial cost
of capital—that is, the cost of financing an investment—but also the cost of
depreciation expenses and income taxes. The user cost is sometimes called an
implicit rental rate because it reflects the rental cost the owner of a capital asset
would have to charge to cover the costs of financing the purchase of the asset
along with the cost of depreciation and income taxes. In a perfectly competitive
market, this user cost would exactly equal the rental rate on capital—hence the
term implicit rental rate. Once the user cost of capital is derived, the marginal
effective tax rate can be calculated from the difference between the before-tax
return on investment and the after-tax rate of return earned by the investors
who financed the investment.5 The following section reviews this methodology
in more detail.

Based on the foregoing discussion, the measurement of effective tax rates requires a precise measure of how explicit and implicit taxes affect the user cost

If capital markets are efficient, the opportunity cost of funding will not depend on which
agent in the economy buys the bonds or equity issued by the firms. In this case, the cost of capital
to firms does not depend on the distribution of personal tax rates. See Fullerton (1984) for a more
comprehensive discussion.
4 See, for example, Henderson (1987), O’Brien and Gelfand (1987a, b), and Neubig and
Sullivan (1987). The latter three studies estimate the impact of the Tax Reform Act of 1986 on
after-tax bank profits.
5 See Bradford and Fullerton (1981) for a detailed discussion of the conceptual issues involved in measuring marginal effective tax rates.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


of capital. Understanding the impact of taxes on the cost of capital requires an
understanding of the basic theory of capital budgeting. Accordingly, the analysis
that follows first reviews capital budgeting theory and then applies the capital
budgeting model to commercial banks. This model is used to derive an expression for the cost of capital that incorporates the effects of reserve requirements
as well as deposit insurance premiums and corporate income taxes.
Review of the Basic Capital Budgeting Model
To begin, consider the simple case of a firm that finances a capital investment k
by issuing interest-bearing debt, D, and equity, S. Capital invested by the firm
earns a constant and known rate of return of ψ per period, so gross revenues
accruing to a capital stock k are ψ k. Assume capital depreciates at a constant
geometric rate δ. Then, the firm can only maintain its capital stock at a constant
level k by investing an additional δk units of capital in each period. By doing
so, the firm maintains a constant net cash flow of
X = (ψ − δ)k


in perpetuity.
The value of the firm’s future cash flows is determined by the cost of capital, which, in turn, is determined by the rate of return demanded by investors
in capital markets. For simplicity, assume the firm’s debt takes the form of a
bond issued in perpetuity that pays a fixed coupon R in each period. Let ρ1
denote the interest rate demanded by bondholders. Then, the value of the firm’s
outstanding debt is just the present value of all future interest payments:


e−ρ1 t Rdt


= R/ρ1 .


Now suppose that all returns net of investment and interest expense are
paid to shareholders as dividends, denoted E. Then
X = R + E.


Since both X and R are constant over time, so is E.
Let ρ2 denote the rate of return demanded by shareholders. Then, the value
of the firm’s equity shares will be determined by the present value of all future
dividends discounted at the rate ρ2 :


e−ρ2 t Edt


= E/ρ2 .



Federal Reserve Bank of Richmond Economic Quarterly

The value of all outstanding claims against the firm is just
V = D+S
= R /ρ1 + E /ρ2


= ρ−1 (R + E),
ρ = λ1 ρ1 + λ2 ρ2, and
λ1 = D/V


λ2 = S/V.
The variable ρ represents the financial cost of capital. It is the rate of return the
firm must earn on its investment to be able to pay the rates of return demanded
by its bondholders and shareholders.6
Substituting from (1) and (3) into (5) yields an expression for the value of
the firm in terms of its capital stock, k, and the other variables of the model:
V = ρ−1 (ψ − δ)k.


Assuming that capital markets are perfectly competitive, the equilibrium present
value of cash flows from the investment will just equal the purchase price of
the capital acquired by the firm. In equilibrium, then,
V = k.


From equation (7), this requirement translates into the condition
ψ = ρ + δ.


Equation (9) simply shows that in equilibrium the marginal rate of return on
investment, ψ , will equal the sum of the financial cost of capital, ρ, which
represents the rate of return required by investors, and the marginal cost of
depreciation, δ. The term of the right-hand side of equation (9) is the user cost,
or implicit rental rate on capital. Note that the stationary nature of this model
environment ensures that λ1 and λ2 are both constant over time with
D = λ1 k, and
S = λ2k.


6 In a more rigorously articulated model, ρ and ρ would differ because of varying degrees
of risk associated with each type of asset. For purposes of this analysis, however, I adopt the
approach common in most intermediate finance textbooks and simply assume that rates of return
on various assets can differ without explicitly modeling uncertainty.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


This last result, while not important to the foregoing analysis, will be useful
later on.
As an aside, the well-known Modigliani-Miller Theorem states that the
financial cost of capital, ρ, is independent of the firm’s capital structure when
capital markets are perfect—that is, when capital markets are perfectly competitive, transactions costs are negligible, and investors have as much information
about the firm’s investment opportunities as its managers. Under these assumptions, a firm’s investment decisions are unaffected by the mix of debt and
equity it issues. This result no longer holds when corporate income taxes are
introduced into the model, however.
Corporate Income Taxes and the Cost of Capital
The U.S. tax code defines taxable income as operating revenues less interest,
allowable depreciation, and other operating expenses. Since this analysis focuses on the effective tax rate on capital, it will abstract from any expenses not
directly affecting the cost of capital or the treatment of capital-related expenses
such as depreciation allowances. As before, let ψ k denote gross revenues and
assume that the firm maintains a constant, fixed capital stock. Let Z denote the
nominal depreciation allowance permitted under the tax code. Then, taxable
profits can be expressed as
π = ψ k − R − Z,


where R denotes nominal interest payments.
Let θ denote the corporate income tax rate. Then, net after-tax cash flow
can be calculated by subtracting corporate income taxes from net pre-tax cash
flow, as defined in equation (1):
Xa = (ψ − δ)k − θπ.


Combining (11) and (12) yields
Xa = (1 − θ)ψ k + θR − δk + Z.


Examine the term on the right-hand side of (13). Because interest expense
affects taxable income, the variable R now appears in the expression for net
cash flow. As a result, the firm’s capital structure will now influence its cost of
capital, and therefore its investment decisions. To see how, consider the relationship between interest expense and the firm’s capital stock. From equation
(2), R = ρ1 D. Together with equation (10), this implies
R = λ1 ρ1 k.


Now consider the tax deduction for depreciation. The taxable depreciation
allowance will not necessarily equal true economic depreciation. In fact, the
two will differ in most cases. The taxable depreciation allowance depends on


Federal Reserve Bank of Richmond Economic Quarterly

the rules for computing the depreciable lifetime of assets and the time path of
the capital stock. For purposes of the present analysis, assume that Z can be
factored as
Z = ζk,


where ζ is some constant. As will be seen later on, all depreciation allowances
examined in this study can be factored into such a form.
Substituting (14) and (15) into (13) yields an expression for after-tax cash
flows as a function of the steady-state capital stock, k, and the other underlying
variables of the model:
Xa = [(1 − θ)ψ + θλ1 ρ1 − (δ − θζ)]k.


The after-tax value of the investment, Va , is just the present value of its aftertax net cash flow discounted using the after-tax cost of capital:
Va = ρ−1Xa .


In equilibrium, the present value of the firm’s cash flows will equal the cost of
the initial capital stock purchased by the firm. Thus,
Va = k.
This last relation implies
ψ = γp (θ) +

δ − θζ


is the pre-tax financial cost of capital. The pre-tax cost of capital differs from
the after-tax cost of capital, ρ, in that the after-tax return to equity, ρ2 , is divided
by (1 − θ) in (19). The best way to understand this result is to note that the
presence of a corporate income tax requires the firm to earn a pre-tax rate of
return on equity of ρ2 /(1 − θ) so it can pay out an after-tax rate of ρ2 to its
Now examine the second term on the right-hand side of (18). This term
reflects the cost of depreciation, net of any taxable depreciation expenses. To
appreciate the economic interpretation of this term, note that
γp (θ) = λ1 ρ1 + λ2 (

δ − θζ
θ(δ − ζ)
The cost of depreciation in the presence of corporate income taxes is the true depreciation rate plus the cost of the tax distortion stemming from any differences
between the true economic depreciation rate and the depreciation allowance
permitted for tax purposes. In the special case where the taxable depreciation
allowance exactly equals true economic depreciation (that is, when ζ = δ), the

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


right-hand side term in (20) reduces to the true economic cost of depreciation,
δ. But when ζ < δ, the effective cost of depreciation under taxation is greater
than it would be otherwise. In this case, the second term on the right-hand
side of (20) shows how much the capital investment must earn at the margin
to pay the added tax caused by the distortion in the tax code. Conversely, an
excessively liberal depreciation allowance would effectively reduce the cost of
Taken together, the sum appearing on the right-hand side of (18) reflects
the firm’s pre-tax user cost of capital. It shows how much the firm’s capital
investment must earn at the margin so as to pay investors in bond and equity
markets the returns they expect after corporate income taxes and depreciation.

The User Cost of Capital for Commercial Bank Lending
Commercial banks are generally subject to the same tax rules as all other U.S.
companies. Thus, the foregoing model of investment and capital budgeting can
be applied to bank lending if the variables are interpreted differently. Instead
of representing physical capital, let the variable k represent the dollar value
of a portfolio of loans. Then, the marginal return on investment, ψ , can be
viewed as the commercial loan rate. Under this interpretation, ψ k denotes gross
revenues from lending.
While bank loans do not depreciate the way physical capital does, banks do
incur loan losses. Loan losses affect earnings in much the same way depreciation affects the productivity of physical capital in the model presented above:
when a borrower defaults on a loan, the lender no longer receives income from
that loan. Accordingly, let the variable δ now represent the fraction of a bank’s
loan portfolio that must be written off in each period. As before, assume δ
is constant over time. Under this interpretation the variable Z can be viewed
as the maximum loan loss provision permitted by the tax code. As with other
types of depreciation allowances, the loan loss provision permitted by the tax
code has not always equaled the true cost of loan losses.
To complete the analogy, let the variable D now denote the value of outstanding deposits. Then, the results derived above can be viewed as a first
approximation of the user cost of capital for a bank. Applied to banks, however,
the model omits at least two important features. The first is the implicit tax
imposed by reserve requirements. The second is deposit insurance premiums.
Reserve requirements obligate banks to hold non-interest-bearing reserves
in the form of vault cash or reserve accounts held with the Fed. Not all
bank deposits are subject to reserve requirements. Currently, the Fed imposes
a 10 percent reserve requirement only on transactions deposits—demand
deposits and certain interest-bearing transactions accounts such as NOW


Federal Reserve Bank of Richmond Economic Quarterly

accounts.7 In the past, however, the Fed imposed reserve requirements on
certain classes of time deposits as well.
Thus, consider a bank that issues three types of deposits as well as nondeposit debt, such as subordinated debt and bank notes. Let
D1 = transaction deposits,
D2 = reservable time deposits,
D3 = nonreservable deposits, and
D4 = nondeposit debt.


Debt of type i pays an interest rate ρi , i = 1, 2, 3, 4. The cost of equity capital—
that is, the rate of return required by the bank’s shareholders—is ρ5 . Under
these assumptions, the bank’s nominal after-tax cost of capital is


λi ρi ,



λi = Di /V,

i = 1, 2, 3, 4, and
λ5 = S/V.

As before, assume that the λi , i =1,2, . . . , 5, are fixed and constant over time,
so that each type of debt outstanding is proportional to the initial capital stock.
D1 = λ1k,
D2 = λ2k,
D3 = λ3 k, and


D4 = λ4k.
Reserve requirements reduce the bank’s interest-earning assets by the fraction of the deposits it is forced to hold as non-interest-bearing reserves. Let
α1 and α2 denote the required reserve ratio on deposits of type D1 and D2 ,
respectively. Then, total required reserves are (α1 D1 +α2 D2 ). Total funds raised
7 Lower reserve requirements apply to the first $52 million of transactions accounts outstanding at each bank, and this tranche changes each year depending on changes in the average
amount of all transactions accounts outstanding. The present analysis ignores this low-reserve

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


by the bank, k, are allocated to loans, denoted by the variable b, plus required
reserves. Formally,
k = b + α1 D1 + α2 D2 .
Substituting in for D1 and D2 from equation (23) yields
k = b + (α1 λ1 + α2 λ2)k,
which can be rewritten as
(1 − α1 λ1 − α2 λ2 )k = b.


Equation (24) expresses the relation between total funds raised and the amount
available to be invested in loans. The term (1 − α1 λ1 − α2 λ2 ) is the fraction of
each dollar the bank raises that is available for investment in loans. The remainder goes to satisfy reserve requirements. Thus, nominal interest revenues are
ψ b, the true cost of depreciation is δb, and the taxable depreciation allowance
is ζb. Using (24), interest and depreciation expenses can be expressed as a
function of the capital stock, k. The result is
ψ b = (1 − α1 λ1 − α2 λ2 )−1 ψ k,
δb = (1 − α1 λ1 − α2λ2 )−1 δk, and


Z = (1 − α1 λ1 − α2 λ2 )−1 ζk.
Banks are also required to pay deposit insurance premiums on all domestic
deposits. Let the symbol β denote the deposit insurance premium. Then, total
deposit insurance premiums paid by the bank are


Di .

Total taxable profits are gross revenues from lending less deposit insurance
premiums, interest expense, and the provision for loan losses. Letting π denote
taxable profits once again and R denote total interest payments made by the
bank to depositors and bondholders,
π = ψ b−β


Di − R − Z.


The bank’s after-tax cash flow is just its revenues less loan losses, deposit
insurance premiums, and taxes:

Xa = (ψ − δ)b − β


Di − θπ.


Substituting the expression for taxable profits (equation [26]) into (27) yields
Xa = (1 − θ)ψ b − (1 − θ)β


Di + θR − (δb − θZ).



Federal Reserve Bank of Richmond Economic Quarterly

Consider the relation between interest expenses and the capital stock. Let
Ri , i =1, . . . , 4, denote total interest payments on debt of type i. Then,
Ri = ρi Di
= λi ρi k, i = 1, . . . , 4,


ρi Di

= ρD k,


ρD =


λi ρi


is the weighted-average nominal interest cost.
Substituting from (25) and (29) into (28) yields an expression for net cash
flows as a function of the value of the initial investment, k:
Xa = (1 − θ) (1 − α1 λ1 − α2 λ2 )ψ − β(


+θρD − (1 − α1 λ1 − α2 λ2 )(δ − θζ) k.

λi )

The after-tax discounted value of this investment is Va = ρ−1 Xa . As before,
the bank’s user cost of capital can be derived by imposing the equilibrium
condition Va = k. The result is
γp (θ, β)
δ − θζ
ψ =
1 − α1 λ1 − α2 λ2
γp (θ, β) =
λi (ρi + β) + λ4ρ4 + λ5 (
denotes the pre-tax financial cost of capital.
These last two expressions are very similar to those derived in the previous
case (equations 18 and 19) except that the pre-tax financial cost of capital in
(33) now includes the cost of deposit insurance premiums. Thus, ρi + β is the
effective cost of issuing deposits of type i, i = 1, 2, 3, not including the cost of
reserve requirements.
Notice also that the expression for the user cost of capital in (32) differs
from the earlier user cost of capital presented in (18) in that the pre-tax financial
cost of capital, γp (θ, β), is now divided by the term (1 − α1 λ1 − α2 λ2 ) to reflect
the cost of reserve requirements. The firm must now earn a marginal rate of
λp (θ, β)
> λp (θ, β)
1 − α1 λ1 − α2 λ2

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


after accounting for the after-tax loan loss expense, (δ − θz)/(1 − θ) , to pay
its depositors and shareholders the return on investment they expect.

Loan Loss Allowances
Banks report loan loss reserves, also known as provisions for loan losses,
on their balance sheets as an estimate of probable future loan losses. Under
generally accepted accounting principles (GAAP), any additions to loan loss
reserves, termed loan loss allowances, are deducted from reported income in
the period the provisions are made and not in the period when the loss actually
occurs. When a bank subsequently determines that a loan is uncollectible, it
reduces its loan loss reserves by the amount of the loss. Because the impact of
the loan loss on earnings is taken into account when the loan loss reserve is
created, the act of writing off the loan has no direct impact on income reported
in that period.
For a variety of reasons, banks typically maintain loan loss reserves in
excess of their expected losses for the coming year.8 Before 1987, the tax
code permitted all commercial banks to deduct loan loss allowances, up to
a stipulated maximum, from taxable income. The Tax Reform Act of 1986
changed the rules for computing deductions for loan losses, however, reducing
the loan loss deductions available to many banks. The discussion that follows
describes the tax treatment of loan loss allowances, both before and after the
Tax Reform Act.

The Tax Treatment of Loan Loss Allowances for Large Banks
Since 1987, “large” commercial banks (banks with assets over $500 million)
have been permitted to deduct loan losses from taxable income only as they are
recognized. Many analysts feel that this rule, known as the “specific chargeoff” method, produces the most accurate measure of true economic income,
as it requires banks to recognize both interest income and loan losses in the
year they accrue.9 If one accepts this argument, the current tax treatment of
loan loss allowances accorded to large banks specifies a deductible loan loss
allowance that equals the true “depreciation” of the loan portfolio. To model
the post-1987 loan loss provision for large banks, then, set
Z = δb.


8 See Walter (1991) for a more detailed discussion of the factors determining loan loss
9 See, for example, Buynak (1987), Neubig (1984), and Neubig and Sullivan (1987). For a
dissenting view, see Henderson (1987).


Federal Reserve Bank of Richmond Economic Quarterly

In this case, the user cost of capital given in equation (32) reduces to
γp (θ, β)
+ δ,
1 − α1 λ1 − α2 λ2

(32 )

where γp(θ, β) is as given in equation (33).

Tax Treatment of Loan Loss Allowances for Small Banks
A “small” bank (one with assets less than $500 million) can choose between
the specific charge-off method and the “experience reserve” method. Under
the experience reserve method, a bank may deduct additions to its bad debt
reserves up to a maximum amount determined by the product of its eligible
loans outstanding and a six-year moving average of its historical loan loss
ratio. To see how this method works, let δ(t) denote the actual loan loss ratio
experienced in year t, and δ(t) = (1/6) 5 δ(t − i) the moving-average of the
current and past five years of loan loss ratios. Then, the maximum loan loss
reserve (LLR) permitted in year t is
LLR(t) = δ(t)b(t),
where b(t) denotes eligible loans outstanding in period t. The corresponding
maximum loan loss allowance deduction is
Z(t) = δ(t)b(t) + LLR(t) − LLR(t − 1) .


If the size of a bank’s loan portfolio does not change over time, then
the experience reserve method is roughly equivalent to the specific charge-off
method. In the case of a bank with a growing loan portfolio, however, use of
the experience reserve method has the effect of accelerating the recognition
of future loan loss deductions and causes taxable income to understate true
economic income (Neubig and Sullivan 1987). To simplify notation and better
understand the properties of these provisions, assume that the loan loss ratio is
constant over time; that is, assume that δ(t) = δ for all t. Then, δ(t) = δ and
Z(t) = δ b(t) + ∆b(t) ,


where ∆b(t) = b(t) − b(t − 1) is the change in eligible loans outstanding from
year (t − 1) to year t. Clearly, this reduces to the specific charge-off method
currently permitted to all banks when ∆b(t) = 0.
To examine the more general case where the bank’s loan portfolio may
grow over time, let
µ = ∆b(t)/b(t).

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


Substituting this last result into equation (36) yields the following expression
Z(t) = δ(1 + µ)b(t).10


Tax Treatment of Loan Loss Allowances before Tax Reform
Before 1987, commercial banks were permitted to choose among several different methods for calculating the taxable loan loss allowance: the experience
method, the specific charge-off method (both as described above), and the
“percentage method.” The percentage method was similar to the experience
method, except that the deductible loan loss allowance was 0.6 percent of total
eligible loans outstanding. As before, let LLR(t) denote the allowable loan loss
reserve in year t. Then, the allowable loan loss allowance would be calculated
as in equation (35) above, except that in this case
LLR(t) = 0.006b(t).
Substituting this last specification into equation (35) yields the result
Z(t) = δb(t) + 0.006∆b(t).


In the special case where the size of a bank’s loan portfolio stays constant
over time, ∆b(t) = 0 and the above expression reduces to Z(t) = δb(t), which
is the same as the deduction permitted under the specific charge-off method.
In the more general case where ∆b(t) = µb(t), one obtains
Z(t) = (δ + 0.006µ)b(t).


Substituting this last result into equation (32) yields the following expression
for a bank’s user cost of capital under the percentage method
γp (θ, β)
+ δ−
1 − α1 λ1 − α2 λ2

(32 )

where, as before, γp (θ, β) is as given in equation (33).
Loan Loss Reserve Recapture Provisions

In addition to eliminating the percentage method, the Tax Reform Act of 1986
also required large banks to recapture any existing loan loss reserves in excess
of actual losses. Under this provision, large banks were required to report
as income a fraction of 10 percent of excess bad debt reserves in 1987, 20
10 The astute reader will note what seems to be a logical inconsistency here, as the foregoing
analysis has assumed a constant loan portfolio size while the depreciation rules allow for a growing
loan portfolio. Interested readers are invited to verify that the results presented in the text would
remain unchanged in all substantive respects if loan portfolio growth were taken into explicit
account in the capital budgeting problem.


Federal Reserve Bank of Richmond Economic Quarterly

percent in 1988, 30 percent in 1989, and 40 percent in 1990.11 Assuming that
banks knew that their excess loan loss reserves would be subject to recapture
after 1986, these recapture provisions would effectively reduce the value of the
1986 loan loss deduction by the expected present value of future excess tax
To calculate the present value of the recapture provisions, one must take
into account any expected future changes in the statutory tax rate. In addition
to mandating the recapture of the loan loss reserve, the Tax Reform Act also
lowered the statutory corporate tax rate from 46 percent in 1986 to 40 percent
in 1987 and to 34 percent thereafter. As a result, a dollar in loan loss reserves
deducted before 1987 produced a 46-cent reduction in taxes, while the subsequent recapture of a dollar in loan loss reserves increased future taxes by a
smaller amount. Thus, the present value of 1987 taxes attributable to the loan
loss recapture would have been (0.40)(0.1)e−ρ. Similarly, the present value of
taxes due to the loan loss recapture for subsequent years would have been
(0.34) 0.2e−2ρ + 0.3e−3ρ + 0.4e−3ρ .
On net, then, taking account of the recapture provisions, the expected present
value of the loan loss allowance to a large bank in 1986 would have been
Z(1986) = δ + 0.006[µ − (40/46) 0.1e−ρ
−(34/46) 0.2e−2ρ + 0.3e−3ρ + 0.4e−4ρ ] b(1986).


The foregoing analysis has been almost entirely theoretical, focusing on the
qualitative effects of explicit and implicit taxes on the user cost of capital. A
purely theoretical analysis does not permit one to gauge the quantitative importance of specific tax rules, however. Nor can it answer questions regarding
11 The act provided exceptions for financially troubled institutions, which were permitted to
defer payment of taxes on the amount of the recapture. It also permitted banks to accelerate the
recapture. This last provision permitted banks reporting losses between 1987 and 1990 to avoid
paying at least part of the tax on the recapture (see U.S. Congress, 1987, pp. 549–57). The present
analysis ignores such considerations.
12 Even in the absence of the Tax Reform Act of 1986, banks’ authorization to use the
percentage method would have expired after 1987 (Henderson 1987). The tax reform of 1969
had instituted a gradual reduction of the maximum limit on loan loss reserve deductions, and the
expiration of the authority to use this method was expected to trigger some type of recapture.
Nor did the banking industry have reason to expect that forthcoming legislation would reinstate
this deduction. The U.S. Treasury had given the treatment of bad debt reserves special attention
during the debate over tax reform (see Neubig [1984]). Therefore, although the Tax Reform Act
was not passed until the summer of 1986, it seems reasonable to assume that commercial banks
expected they would be required to recapture their excess loan loss reserves after 1987.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


the overall impact of legislation such as the Tax Reform Act of 1986, which
lowered statutory tax rates while imposing offsetting reductions in the allowable
deduction for loan losses. To answer such questions, one needs an empirical
measure of the user cost of capital for banks.
Figure 1 depicts the behavior of the pre-tax and after-tax user cost of capital
for commercial bank lending from 1986 to 1995. This period is an interesting
one, as it includes a major change in tax laws and two separate instances where
reserve requirements were reduced. The data used to compute these series
were obtained from various reports that all insured banks must file routinely
with the federal regulatory agencies. Both series were obtained by aggregating
year-end data on all domestic commercial banks.13 As such, these series represent industrywide weighted averages. The values of the parameters characterizing tax rules and reserve requirements during this period are summarized in
Table 1.
The pre-tax cost of capital in Figure 1 is calculated using the formulas
specified in equations (32) and (33). In accordance with the earlier discussion of the tax treatment of loan loss allowances, the loan loss deduction for
1986 includes the present value of the future loan loss reserve recapture mandated by the Tax Reform Act, as characterized in equation (40).14 The loan
loss allowance for subsequent years is based on the formula in equation (34).
 δ + 0.006 µ − (40/46)0.1e − (34/40) 0.2e
−3ρ + 0.4e−4ρ
+ 0.3e
for t = 1986, and
δ for t ≥ 1987.
The after-tax cost of capital is just the pre-tax cost of capital with all
tax parameters, αi , θ, and ζ, set to zero. How best to treat deposit insurance
premiums presents certain conceptual problems. To the extent that deposit insurance reduces funding costs for banks, the deposit insurance premium, β,
just reflects the offsetting cost of the financial guarantee. If deposit insurance
were privately provided and supplied at a market-determined price, the deposit
insurance premium would not be viewed as a tax on commercial bank intermediation. As noted earlier, however, FDIC deposit insurance premiums may
not always reflect the fair market value of the underlying guarantee. If they are
set too low, they represent a subsidy. If they are increased to raise funds for
other purposes, such as rescuing a competing deposit insurance fund, deposit
insurance premiums can constitute a tax. For the present, assume that deposit

A more detailed description of data sources and calculations can be found in the Appendix.
For now, I assume that all banks are “large” banks that are subject to the specific chargeoff method and the recapture of loan loss reserves. In later sections, I will examine the marginal
impact of the loan loss recapture provisions of the Tax Reform Act of 1986 and the marginal tax
benefit of the experience reserve method, which small banks continued to enjoy throughout the
period under consideration.


Federal Reserve Bank of Richmond Economic Quarterly

Figure 1 Pre-Tax and After-Tax User Cost of Capital





1986 1987 1988

1989 1990 1991 1992 1993

1994 1995


insurance is fairly priced. Accordingly, the after-tax cost of capital in Figure 1
is calculated according to the formula
γa (β) = λ1 (ρ1 + β) + λ2 (ρ2 + β) + λ3 (ρ3 + β) = λ4 ρ4 + λ5 ρ5 + δ. (41)
The marginal impact of deposit insurance premiums on the cost of capital will
be examined in a later section.
The Taxation of Commercial Banking: 1986 –1995
There are several ways in which one can measure the marginal effective tax
rate. The simplest measure is the difference between the pre-tax and after-tax
cost of capital, which reflects the marginal cost of taxes on investment returns.
Alternatively, the marginal effective tax rate can be expressed as a percentage
either of the pre-tax or after-tax cost of capital.15 Figure 2 depicts the behavior
of the marginal effective tax rate, measured as the difference between the pretax and after-tax user cost of capital. Notice that the marginal effective tax rate
has fallen on average over the period in question, from a high of 126 basis
points in 1986 to under 70 basis points in recent years. Most of this decline took
place in the two years immediately following enactment of the Tax Reform Act
of 1986, corresponding to the period in which the reductions in the statutory tax
rate mandated by the act were phased in. By 1988, the marginal effective tax
15 See Bradford and Fullerton (1981) for an analysis of the properties of these different
summary statistics.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


Table 1 Summary of Tax Parameter Values
Statutory Tax
Rate (θ)

Present Value of the Deduction for
Loan Loss Allowances (Z(t)),
(Large Banks)
(0.2e−2ρ + 0.3e−3ρ

δ + [0.006(µ −





















(0.1e−ρ )
+ 0.4e−4ρ )b(t)]

Transaction: α1 = 12%
Time: α2 = 3%c
Transaction: α1 = 12%
Time: α2 = 3%c
Transaction: α1 = 12%
Time: α2 = 3%c
Transaction: α1 = 12%
Time: α2 = 3%c
Transaction: α1 = 12%
Time: α2 = 3%
Transaction: α1 = 12%
Time: α2 = 0%
Transaction: α1 = 10%e
Time: α2 = 0%

Before the Tax Reform Act of 1986, corporations with taxable income below $100,000 were subject to a
lower rate. In addition to lowering the statutory tax rate to 34 percent, the act changed the graduated tax
structure. Starting in 1987, the threshold for lower tax rates was lowered to $75,000. Both before and after
the Tax Reform Act, corporations with incomes exceeding the threshold were subject to a surcharge meant to
recover the benefit of lower tax rates on income below the threshold. Currently, corporations must pay a 5
percent surcharge on income over $100,000 up to a maximum of $11,750. As a result, corporations with taxable
incomes over $335,000 pay both an average and a marginal statutory rate of 34 percent. (For more details, see
U.S. Congress [1987], pp. 271–72.) In constructing the weighted-average cost of capital, it was assumed that
all banks were subject to the maximum statutory tax rate.
b The Garn-St. Germain Depository Institutions Act of 1982 required that $2 million of reservable liabilities
of each depository institution be subject to a zero percent reserve requirement. The act instructs the Board
of Governors to adjust the amount of reservable liabilities subject to this zero percent reserve requirement
each year by 80 percent of the annual percentage increase in the total reservable liabilities of all depository
institutions. In 1996, this zero-reserve tranche was raised to $4.3 million.
The Monetary Control Act of 1980 established a low-reserve tranche against which a 3 percent reserve
requirement is applied. In 1995, a 3 percent reserve requirement was applied to the first $52 million in reservable deposits. As with the zero-reserve tranche, this amount is adjusted each year by 80 percent of the total
percentage increase in the total reservable liabilities of all depository institutions. The user cost of capital
calculations ignores the zero- and low-reserve tranches, since, at the margin, virtually all banks are subject to
the higher reserve requirement listed in the table.
c During this period, reserve requirements on time deposits applied only to nonpersonal time deposits with an
original maturity less than 11/2 years. The reserve requirement on nonpersonal time deposits was reduced to
zero at the end of 1990.
d The Tax Reform Act of 1986 reduced the maximum statutory tax from 46 percent to 34 percent, effective for
taxable years beginning on or after July 1, 1987. Income in taxable years including July 1, 1987, was subject
to a blended rate. According to the methodology specified in the act, the effective statutory tax rate for the
1987 calendar year would have been calculated as (181/365)x(40%) + (184/365)x(34%) = 40%, as there were
181 days between January 1, 1987, and June 30, 1987, and 184 days between July 1, 1987, and December 31,
1987. For more details, see U.S. Congress (1987), pp. 272–73.
e The reserve requirement on transaction deposits was reduced from 12 to 10 percent in April 1992.


Federal Reserve Bank of Richmond Economic Quarterly

Figure 2 The Marginal Effective Tax Rate on Commercial Bank Lending


Including Loan Loss Recapture

Excluding Loan Loss Recapture

1986 1987

1988 1989

1990 1991 1992

1993 1994



rate had fallen by almost half, to 66 basis points, suggesting that the reduction
in the statutory corporate tax rate more than offset the loss of the loan loss
reserve deduction.
Figure 3 compares the behavior of the marginal effective tax rate with that
of the average tax rate, computed as taxes paid as a percent of pre-tax earnings.
To facilitate comparison, the marginal effective tax rate in Figure 3 is expressed
as a percent of the pre-tax cost of capital. In light of the earlier discussion, it
should not be surprising to find that the behavior of the marginal effective and
average tax rate measures differ so much. Whereas the marginal effective rate
falls dramatically after 1986, the average tax rate rises. The difference in the
behavior of these two series is in part due to the timing of the recognition
of cash flows. Recall that the marginal effective tax rate incorporates the full
present value of the future loan loss recapture in 1986—consequently, the cost
of recapture does not influence the marginal tax rate in later years. In contrast,
the measured average tax rate recognizes these taxes only as they accrue.
The behavior of the average tax rate also reflects other changes in tax
rules that did not affect the incentive of commercial banks to make loans. One
of the major provisions of the Tax Reform Act of 1986 was the repeal of
the tax deductibility of interest payments on municipal bonds. Before 1987,
commercial banks paid no taxes on interest earned on municipal bonds, while
interest payments on bank debt issued to fund such investments was tax deductible. Partly as a result of this favorable tax treatment, the average tax
rate for commercial banks tended to be low, especially when compared to the
average tax rate on most other industries. The perception that banks enjoyed

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


Figure 3 Comparison of Average and Marginal Effective Tax Rates







1986 1987

1988 1989

1990 1991 1992

1993 1994




too many tax advantages led Congress to repeal the tax deduction on interest
from municipal bonds, along with the tax deduction for bad debt reserves. 16
As Henderson (1987), Neubig and Sullivan (1987), and O’Brien and Gelfand
(1987a, b) note, however, the increase in the average tax rate due to the repeal
of the tax deduction for interest on municipal bonds is largely illusory. Because
interest paid on municipal bonds is not subject to federal taxes, interest rates
on such bonds tend to be lower than interest rates on taxable bonds—in fact,
the interest rates paid on municipal bonds tend to be comparable to after-tax
interest rates on taxable bonds. Thus, the interest rate differential between taxable and nontaxable bonds can be viewed as an implicit tax. After losing this
tax deduction in 1987, banks tended to substitute taxable bonds for municipal
bonds, with the result that taxable income increased along with taxes paid.
Although banks paid higher federal taxes on average, the impact on their aftertax return was minimal. The principal result of this change in tax laws, then,
was to substitute explicit taxes paid to the federal government for implicit taxes
that were previously paid to municipalities.
Despite these considerations, Figure 3 does hold a seeming puzzle. Note
that the marginal effective tax rate tends to be much lower than the measured
average tax rate; this despite the inclusion of the cost of reserve requirements
and deposit insurance premiums in the marginal rate but not in the average rate.
Three different factors can help explain this apparent anomaly. First, Henderson

See U.S. Congress (1987).


Federal Reserve Bank of Richmond Economic Quarterly

(1987) notes that many large banks have substantial foreign operations, which
are subject to higher tax rates. Second, to the extent that nonfinancial assets are
effectively taxed at a higher rate than financial assets such as commercial loans,
the result would be to raise the average tax rate above the marginal effective
tax rate on lending. Finally, the marginal effective tax rate calculations derived
earlier and illustrated in Figures 2 and 3 assume banks earn no pure economic
profits. To the extent that banks do earn economic profits, the marginal effective
tax rate on such profits just equals the statutory tax rate, which is currently 34
The Long-Run Impact of Tax Reform
As noted earlier, the data depicted in Figure 2 suggest that the reductions in
the statutory corporate income tax rate that took place in 1987 and 1988 more
than offset the loss of the tax deduction for the bad-debt reserve. Care must
be taken in interpreting this result, however, because the marginal effective tax
rate is influenced by many factors, not just changes in tax rules. Moreover,
the influence of the future recapture of the deduction for loan loss reserves
mandated by the Tax Reform Act exerted a significant transitory influence on
the marginal effective tax rate in 1986.
A measure of the long-run marginal impact of the Tax Reform Act on
the marginal effective tax rate can be calculated by computing the user cost
of capital for a single year under the two sets of tax rules, ignoring recapture
provisions. The results of such an exercise, performed using 1986 data, are
presented in Table 2. When the effect of the recapture provisions is excluded,
the marginal effective tax rate for 1986 falls to 92 basis points—thus, the
marginal impact of the recapture provisions was 34 basis points. Recomputing
the 1986 user cost of capital assuming a 34 percent tax rate and adopting the
specific charge-off method reduces the marginal effective tax rate another 30
basis points. From these two exercises, one can conclude that about half of the
observed decline in the marginal effective tax rate between 1986 and 1988 was
attributable to the long-run impact of tax reform. Although other factors also
contributed to the observed decline in the marginal effective tax rate during
this period, their influence was minimal.
Tax Reform and Small Banks
The last exercise ignored the differential treatment accorded to small banks
by the Tax Reform Act and assumed that all banks lost the loan loss reserve
deduction. Recall, however, that small banks were permitted to continue using
the experience reserve method in determining their loan loss deduction. What
impact did tax reform have on these institutions?

See Bradford and Fullerton (1981) for a more comprehensive discussion of this last issue.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


Table 2 The Long-Run Marginal Impact of Tax Reform on
the Marginal Effective Tax Ratea
Effective Tax
Rate, Including
(basis points)

Effective Tax
Rate, Excluding
(basis points)

Large Banksb
Before Tax Reform
After Tax Reform



Small Banksc
Before Tax Reform
After Tax Reform




Calculated using 1986 year-end industrywide financial data.
The marginal effective tax rate for large banks before tax reform was calculated using a statutory
tax rate of 46 percent and assumes that banks used the percentage method to calculate the loan
loss allowance. The tax rate after tax reform was computed assuming a 34 percent statutory tax
rate and assumes that banks use the specific charge-off method.
c The user cost of capital for small banks is based on the same data used in the large bank
example, except that the calculations assume use of the experience reserve method, under which
ζ = δ(t) + δ(t)µ, where δ(t) = 1
δ(t − i) and µ represents the growth rate of eligible
loans. As with the first exercise, the before-tax-reform user cost of capital is calculated assuming
a 46 percent statutory tax rate, and the after-tax-reform user cost is computed assuming a 34
percent tax rate.

An approximate measure of the marginal effective tax rate for small banks
can be obtained using industrywide weighted averages. Specifically, consider
a hypothetical representative “small” bank that experienced the same realized
loan loss ratios and growth rates in outstanding loans as did the industry in the
aggregate. Next, compute the user cost of capital assuming that this bank takes
advantage of its option to use the experience reserve method, as characterized in
equation (37). This last result can then be used to compute an marginal effective
tax rate measure for small banks using the experience reserve method.
Table 3 compares the marginal effective tax rates under the specific chargeoff (small bank) method with that obtained using the experience reserve (large
bank) method.18 Notice that the two tax rates differ by no more than 6 basis
points after 1986. The difference between the two in 1986 can be accounted
for almost entirely by the present value of future loan loss recoveries imposed
on large banks.
Evidently, the favorable treatment of loan loss reserves accorded to small
banks by the Tax Reform Act of 1986 has had only a small impact on the
18 For 1986, the “large” bank effective tax rate calculation assumes that banks use the
percentage method to calculate the taxable deduction for loan loss reserves.


Federal Reserve Bank of Richmond Economic Quarterly

Table 3 Comparison of the Marginal Effective Tax Rate
for Large and Small Banks
Large Banks

Small Banks




Note: All figures are expressed as basis points.

marginal effective tax rate on lending. The largest benefit to small banks conferred by the act was in exempting them from the recapture of past excess
contributions to loan loss reserves.
The Cost of Reserve Requirements
The derivation of the user cost of capital given in equation (32) showed how
non-interest-bearing reserve requirements increase the cost of funding a loan.
Figure 4 shows the net overall impact of reserve requirements on the user cost
of capital from 1986 to 1995, obtained by calculating the difference between
the pre-tax cost of capital, including the cost of reserve requirements, and the
pre-tax cost of capital net of reserve requirements:
γp (θ, β)
− γp (θ, β).
1 − α1 λ1 − α2 λ2

As Figure 4 shows, the cost of reserve requirements fell significantly after 1990,
from approximately 20 basis points in that year to just over 10 basis points in
1994. There are at least three factors that might account for this decline. The
first was the elimination of reserve requirements against time deposits in 1991
and a subsequent reduction from 12 to 10 percent in the required reserve ratio
for transaction deposits in 1992. The second is a decline in interest rates. The
third is a decline in banks’ reliance on reservable deposits—to the extent that
banks substitute nonreservable liabilities for those bearing reserve requirements,
they can effectively avoid paying the implicit reserve requirement tax.
The elimination of reserve requirements against time deposits accounted for
almost 2 basis points of the observed decline in Figure 4, while the reduction
in the reserve ratio for transaction deposits accounted for just under 3 basis

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


Figure 4 The Cost of Reserve Requirements


Effect on the
Pre-Tax Cost of Capital

1986 1987

1988 1989

1990 1991 1992

1993 1994



points.19 The remainder of the reduction can be attributed to falling interest
rates. Changes in the ratio of reservable deposits to other liabilities does not
appear to have contributed to the observed reduction in the overall cost of
reserve requirements.20
Deposit Insurance and the Cost of Capital
From 1935 to 1989 all insured U.S. commercial banks paid the FDIC an annual
statutory deposit insurance premium of 0.0833 percent of domestic deposits (or
8.33 basis points). The effective deposit insurance premium was often much
lower, however, because the FDIC frequently rebated some portion of these premiums. Such rebates ended in the late 1980s after a large increase in the number
of bank failures threatened to deplete the FDIC’s Bank Insurance Fund (BIF).
Using its newly acquired authority to increase deposit insurance assessments,
the FDIC raised its assessments to 0.2125 percent in 1991 and again in 1992
19 Estimated cost savings stemming from the 1991 elimination of reserve requirements
against nonpersonal time deposits were obtained by measuring the marginal cost of such reserve requirements at the end of 1990. Similarly, estimated cost savings associated with the 1992
reduction in required reserve ratios on transaction deposits reflect the marginal cost of holding an
extra 2 percent reserve requirement at the end of 1991.
20 Although the importance of demand and other transaction deposits has fallen substantially
in the past 25 years, transaction deposits accounted for approximately 20 percent of the value of
debt plus equity throughout the period 1986–1990, falling slightly from 1985 to 1990 and rising
modestly thereafter.


Federal Reserve Bank of Richmond Economic Quarterly

to 0.23 percent. Following a congressional mandate, the agency adopted riskbased assessments in 1993. Under this latter system, banks paid assessments
in the range of 0.23 to 0.31 percent. The minimum assessment was lowered to
0.044 percent in mid-1995 after BIF reached its mandated capitalization level,
and well-capitalized and well-managed banks received a small rebate that year.
In 1996, the FDIC reduced its risk-based premiums further to a range of zero
to 0.31 percent.21
Figure 5 shows the marginal contribution of the cost of deposit insurance premiums to the pre-tax user cost of capital, calculated by subtracting
a measure of the pre-tax user cost of capital that excludes deposit insurance
premiums from the pre-tax user cost including deposit insurance premiums.
The dramatic increase in the cost of deposit insurance after 1989 reflects the
increases in effective deposit insurance assessments imposed during this period.
These increases had a substantial impact on banks cost of capital. From 1992 to
1994, deposit insurance premiums contributed between 16 and 18 basis points
to the pre-tax user cost of capital, up from 6 basis points in 1988.
Whether deposit insurance assessments should be treated as a tax on the
banking industry depends on how fairly the FDIC’s assessments reflect the cost
of its financial guarantee. A recent study by Epps, Pulley, and Humphrey (1996)
computed “fair” deposit insurance premiums for a sample of 77 banks using
1989 data. That study found that the median fair deposit insurance premium
was 0.0107 percent of deposits (assuming one bank examination per year),
compared to the 0.0833 deposit insurance premium charged that year. At first
glance, these findings seem to suggest that deposit insurance is overpriced. A
closer look at the authors’ results reveals certain important mitigating factors,
however. The fair deposit insurance premiums for individual banks in the study
ranged from a low under 0.0001 percent to a high of 0.7749 percent. The authors
note, however, that the FDIC can reduce the effective cost of its liability to
depositors of a troubled bank through more frequent monitoring, which is the
current policy of the bank regulatory agencies.22 Nonetheless, these findings
indicate that deposit insurance requires well-managed and conservatively run
banks to subsidize banks that pose greater risks to the deposit insurance fund. To
be sure, the adoption of risk-based assessments has ameliorated this problem
somewhat. But although the adoption of risk-based assessments has reduced
the subsidy to risky banks, the findings of Epps, Pulley, and Humphrey (1996)
indicate that the current risk-based assessment scheme would not have been
sufficient to eliminate the subsidy to the riskiest banks in 1989. What Figure 5
shows, then, is the deposit insurance tax on the safest banks. Based on available


For more details, see FDIC (1995).
The fair deposit insurance premiums reported here do not include the cost of bank examinations. Epps, Pulley, and Humphrey (1996) also discuss examination costs.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


Figure 5 The Cost of Deposit Insurance


Effect on the
Pre-Tax Cost of Capital

1986 1987 1988

1989 1990

1991 1992 1993

1994 1995


information, it is not clear whether the current risk-based assessment scheme
constitutes a tax on the industry as a whole.
Legislation enacted during 1996 does use the deposit insurance system to
impose a tax on commercial banks, however. Beginning in 1997, all banks
will be required to pay a special charge of 1.29 basis points to help pay the
interest on bonds issued in 1987 to recapitalize the thrift industry’s deposit
insurance fund. That surcharge is scheduled to increase to 2.43 basis points
in 1999. Understandably, the commercial banking industry resisted legislation
requiring it to help pay for the losses incurred by the thrift industry. The banking
industry had paid very high deposit insurance premiums during the early 1990s
to recapitalize its own deposit insurance fund, and bankers did not wish to
see their premiums raised once again to help rescue a competing industry’s
fund. My model shows that these surcharges will not have a dramatic impact
on the banking industry’s cost of capital, however. Using 1995 year-end data,
the effect of a 1.29 basis point surcharge would be to increase the pre-tax user
cost of capital by less than 1 basis point. A 2.43 basis point surcharge would
produce an increase of less than 2 basis points.

In 1986, the banking industry paid an average effective tax rate of just under
24 percent. Since the enactment of the Tax Reform Act of 1986, however, the
average effective tax rate has been over 30 percent. A cursory inspection of
these data would seem to suggest that the tax burden on the banking industry


Federal Reserve Bank of Richmond Economic Quarterly

has risen in recent years. A closer look at the factors accounting for this rise
suggest otherwise, however. The increase in the average tax rate paid by banks
over the past decade is due largely to the elimination of the interest deduction
on municipal debt for banks. Banks, in return, have responded by substituting
into taxable corporate debt, which pays higher interest rates. Although banks
now pay more federal taxes, they also earn more pre-tax income. Moreover, the
elimination of the tax deduction for municipal debt had no impact on banks’
incentive to extend other forms of credit, which is influenced by the marginal
effective tax rate on bank lending.
An examination of the recent behavior of the marginal effective tax rate on
bank lending paints a much different picture, suggesting that the tax disincentives to commercial bank intermediation have fallen modestly over the past ten
years. The decline in the marginal effective tax rate is due principally to two
factors. The first is the Tax Reform Act of 1986. Although tax reform resulted
in higher average tax rates, it reduced the marginal effective tax rate on commercial bank lending. The second factor is the reduction in the implicit reserve
requirement tax, which is due partly to reductions in reserve requirements and
partly to declining interest rates.
This article began by questioning the extent to which the tax burden borne
by the commercial banking industry may have contributed to the declining
share of bank lending in credit markets. For many years, the commercial banking industry enjoyed special tax treatment, meant in part to compensate for the
burden of regulation, including the cost of reserve requirements. Commercial
banks now face the same federal tax rules as other lenders, however. At the
same time, they also continue to bear the cost of reserve requirements. Although
reserve requirements have been reduced in recent years, they still impose an
approximately 10 basis point cost penalty on banks out of a total marginal
effective tax rate of roughly 70 basis points. Even though the tax burden on
commercial banking has fallen by some measures, implicit taxes continue to
handicap the ability of banks to compete against other lenders. More importantly, recent statutory reductions in reserve requirements accounted for less
than half of the reduction in the cost of reserve requirements in recent years—
the rest was due to falling interest rates. In the absence of further policy actions,
then, an increase in interest rates could increase the marginal effective tax rate
on commercial bank lending substantially.
Many observers feel that any regulatory burden borne by banks, including
the burden of reserve requirements, is mitigated by unique benefits such as deposit insurance and access to the Fed’s discount window. The foregoing analysis
showed that changes in deposit insurance assessments contributed substantially
to the banking industry’s cost of capital from 1992 to 1995. Deposit insurance
assessments have fallen dramatically over the past year, however, and now
account for a negligible fraction of the cost of financing a loan (except for the
few banks that must pay the highest deposit insurance assessment rate of 31

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


basis points). Moreover, the estimated impact of the recent deposit insurance
surcharge imposed on commercial banks to help pay for the recapitalization
of the thrift industry’s deposit insurance fund is exceedingly small and would
appear to pose no undue burden on the industry. Whether deposit insurance
represents a subsidy to the banking industry continues to be the topic of an
active debate. Fortunately, the burden of explicit corporate taxes and the implicit
cost of reserve requirements can be quantified.

For the financial cost of capital, ρ = 5 λi ρi , estimates of interest expense
were obtained using data available in the FDIC Historical Statistics on Banking. The cost of equity capital was estimated using the basic CAPM model,
following the procedure suggested by Ibbotson and Sinquefield (1989). The
estimate for the stock market beta needed for this calculation was obtained
from Berkovec and Liang (1991). The results are summarized below.



Interest Rate





Cost of
(ρ5 )




Estimates of financial structure, as reflected by the parameters λ1 , λ2 , . . . ,
λ5 , were obtained from the Quarterly Reports of Condition and Income, or
Call Reports, and from the Federal Reserve’s Weekly Report of Transaction Accounts (FR2900). Data on loan charge-off rates came from the FDIC Historical
Statistics on Banking, while data on effective deposit insurance assessments are
from the FDIC Annual Report for 1995.


Federal Reserve Bank of Richmond Economic Quarterly

Berkovec, James A., and J. Nellie Liang. “Changes in the Cost of Equity
Capital for Bank Holding Companies and the Effects on Raising Capital.”
Finance and Economics Discussion Series 160. Washington: Board of
Governors of the Federal Reserve System, 1991.
Bradford, David F., and Don Fullerton. “Pitfalls in the Construction and Use
of Effective Tax Rates,” in Charles R. Hulten, ed., Depreciation, Inflation,
and the Taxation of Income from Capital. Washington: The Urban Institute
Press, 1981.
Buynak, Thomas M. “How Will Tax Reform Affect Commercial Banks?”
Federal Reserve Bank of Cleveland Economic Review (Quarter 2, 1987),
pp. 24–34.
Epps, T. W., Lawrence B. Pulley, and David B. Humphrey. “Assessing the
FDIC’s Premium and Examination Policies Using ‘Soviet’ Put Options,”
Journal of Banking and Finance, vol. 20 (May 1996), pp. 699–721.
Federal Deposit Insurance Corporation. Annual Report 1995. Washington:
Fullerton, Don. “Which Effective Tax Rate?” National Tax Journal, vol. 37
(March 1984), pp. 23–41.
, and Yolanda K. Henderson. “Long-Run Effects of the Accelerated
Cost Recovery System,” Review of Economics and Statistics, vol. 67
(August 1985), pp. 363–72.
Hall, Robert E., and Dale W. Jorgenson. “Tax Policy and Investment Behavior,”
American Economic Review, vol. 57 (June 1967), pp. 391– 414.
Harberger, Arnold C. “Efficiency Effects of Taxes on Income from Capital,”
in M. Krzyzaniak, ed., Effects of Corporation Income Tax. Detroit: Wayne
State University Press, 1966.
Henderson, Yolanda. “The Taxation of Banks: Particular Privileges or Objectionable Burdens?” New England Economic Review (May/June 1987), pp.
Ibbotson, Roger G., and Rex A. Sinquefield. Stocks, Bonds, Bills, and Inflation: Historical Returns (1926–1987). Charlottesville, Va.: The Research
Foundation of The Institute of Chartered Financial Analysts, 1989.
McCauley, Robert N., and Rama Seth. “Foreign Bank Credit to U.S. Corporations: The Implications of Offshore Loans,” Federal Reserve Bank of
New York Quarterly Review, vol. 17 (Spring 1992), pp. 52–65.
Neubig, Thomas S. “The Taxation of Financial Institutions After Deregulation,”
National Tax Journal, vol. 37 (September 1984), pp. 351–59.

A. Kuprianov: Tax Disincentives to Commercial Bank Lending


, and Martin A. Sullivan. “The Effect of the Tax Reform Act of
1986 on Commercial Banks,” in Compendium of Tax Research 1987.
Washington: Office of Tax Analysis, Department of the Treasury, 1987.
O’Brien, James M., and Matthew D. Gelfand. “Corrigendum: The Impact of
the Tax Reform Act of 1986 on Commercial Banks,” Tax Notes, vol. 34
(March 30, 1987a), pp. 1323–25.
. “Effects of the Tax Reform Act of 1986 on Commercial Banks,”
Tax Notes, vol. 34 (February 9, 1987b), pp. 597–604.
U.S. Congress, Joint Committee on Taxation. General Explanation of the Tax
Reform Act of 1986. Public Law 99–514, 100 Cong. 1 Sess. Washington:
Government Printing Office, 1987.
Walter, John R. “Loan Loss Reserves,” Federal Reserve Bank of Richmond
Economic Review, vol. 77 (July/August 1991), pp. 20–30.

Federal Reserve Bank of St. Louis, One Federal Reserve Bank Plaza, St. Louis, MO 63102