Economic Inquiry
vol. 22, no. 4 (October), 1984, pp. 593-96.
Deficits and Inflation: A Comment on Dwyer
Roger W. Garrison
I. Introduction
In a recent article appearing in this journal (July 1982) Gerald P.
Dwyer, Jr. offered some empirical tests of several hypotheses concerning
the relationship between federal government deficits and the rate of inflation.
Dwyer's summary contains the "tentative inference" that "deficits play
no role in determining inflation..."—an inference that runs directly counter
to the views of Buchanan and Wagner (1977). The purpose of this comment
is to suggest that Dwyer's inference cannot be justified by his analysis.
The problems with Dwyer's
analysis are conceptual, theoretical, and methodological. Each of these
aspects will be taken up in sequence. Section II shows that Dwyer's measure
of the deficit conflates the cause and the consequence of the debt-monetization
process. Section III argues that his counter-intuitive views of the relationship
between deficits and inflation are a result of undue reliance on the comparative-statics
mode of analysis and neglect of the market-dynamics mode. Section IV stresses
that the meaningfulness of Dwyer's conclusions depends on the unusual (Granger-Sims)
usage of the word "cause."
II. The Real Deficit: A Cause or a Consequence?
In Section I of his article, Dwyer obscures the relationship between
deficits and inflation by suggesting an alternative—and unconventional—measure
of the deficit. "If the real rate of interest is unaffected by inflation,
the change in the real value of the government debt is a more pertinent
measure of the deficit than the real deficit" (p. 318). This terminological
twist results in a conflation of the process that constitutes the causal
link between deficits and inflation and the ultimate consequences of that
same process. High deficits may result in debt monetization, which increases
the rate of inflation, which in turn erodes the real value of the debt.
The question of the pertinent measure of the deficit depends upon whether
one is interested in the causal relationship (in which case the conventional
measure is the more pertinent) or the ultimate consequence (in which case
the alternative measure can be defended).
But Dwyer is clearly interested
in the issue of causality. He does not hesitate to suggest that the fact
that the real value of the total federal debt has fallen as often as it
has risen since the early 1950s "...raises serious questions about the
empirical basis of the arguments by Buchanan and Wagner (1977), for example,
that inflation in this period is largely a result of deficits" (p. 316).
These "serious questions" can be answered by carefully distinguishing between
cause and consequence. Buchanan and Wagner argue in terms of the conventionally
defined deficits which lead the central bank to engage in debt monetization;
Dwyer points to statistics that reflect the ultimate consequence of that
debt-monetization process.
It is interesting to note
that the same "serious questions" might be raised in connection with the
Great German Inflation of the early 1920s. Shortly after the "London Ultimatum"
in May of 1921, which set the reparations payment in the neighborhood of
132 billion Gold Marks, the German money supply began its dramatic ascent
along a geometric growth path. Historical accounts leave no doubt that
Germany's obligation to procure fixed sums of foreign exchange at regular
intervals to pay her war debt was the proximate cause of the dramatic inflation
that followed (Graham, p. 8). But an analysis based on the Dwyer-pertinent
measure of the deficit would focus on the time profile of the real debt.
The observation that the real debt rose more slowly during the twelve months
following the London ultimatum than during the twelve months preceding
it may be seen as a basis for serious questions about the relationship
between debt and inflation.(1) The fact
that the cause of the inflation is seen as the war debt while the consequences
are seen largely in terms of domestic debt reduces the likelihood of a
misinterpretation. Nonetheless, an understanding of the distinction between
cause and consequence warns against adopting the Dwyer-pertinent measure
when causality is at issue.
III. Comparative Statics and Market Dynamics
That Dwyer chose to redefine the deficit as the change in the real
value of the debt is symptomatic of a common analytical error: the failure
to distinguish between comparative statics and market dynamics. This critical
distinction closely parallels the distinction between cause (market dynamics)
and consequence (comparative statics). Attention to these two separate
modes of analysis can put Dwyer's contribution in a new light.
To focus on the real value
of outstanding bonds for each of a sequence of years, as Dwyer does, limits
the analysis to the comparative-statics mode. It overlooks the debt-monetization
process, and the subsequent market process by which the economy becomes
adjusted to each increment of monetized debt. These processes constitute
the market dynamics that cause the real value of outstanding bonds to fall
largely offsetting the newest increment of debt. That is, the monetizing
of debt is an important part of the market dynamics that cause the amount
of debt in real terms to be no higher than before.
The distinction between
comparative statics and market dynamics offers a reconciliation of the
belief that the central bank tends to accommodate the Treasury with statements
that, on a superficial level, seem to imply such a belief has no historical
foundation. Consider Dwyer's statement, for instance, that "Even though
there were substantial deficits [during the years 1952 through 1981], federal
government indebtedness does not increase substantially because the nominal
growth rates of bonds associated with the deficits are frequently less
than the inflation rate" (Dwyer, 1982, p. 317). The statement is true,
but an accommodation central bank may be responsible for both inflation
and a constant level of real indebtedness. If so, the observation of (roughly)
constant indebtedness cannot be used as evidence that the central bank
was not accommodating. That is, the comparative-statics results cannot
be used as evidence that a dynamic process capable of yielding those results
did not take place. Dwyer's empirical observation does not, by itself,
confirm the Buchanan-Wagner view (that deficits lead to monetary growth),
but if their view is correct, the monetization process they describe would
give rise to the very pattern of inflation and real indebtedness that Dwyer
has observed.
It might be noted that the
failure to distinguish adequately between comparative statics and market
dynamics is not at all unique to Dwyer's article. The blurring of these
two modes of analysis was largely responsible for the extended controversy
about the importance of the real-cash-balance effect in monetary theory
(Archibald and Lipsey, 1958). By clearly distinguishing between a "market
experiment" (comparative statics) and an "individual experiment" (market
dynamics), Don Patinkin demonstrated that the absence of the real-cash-balance
effect in the former mode of analysis does not imply its unimportance in
the latter (Patinkin, 1965, p. 11f and passim). The arguments in
the present comment can be seen as paralleling Patinkin's.
IV. The Issue of Causality
Dwyer offers an empirical test which appears to cast doubts on the
Buchanan-Wagner view and to support the view that money creation cannot
be "explained" by deficits. He skirts the issues of market dynamics by
making use of the technique suggested by Granger and Sims. This technique
imposes no restrictions from any specific theory about the market process
which generates the text data. Each of six different variables (the level
of prices, the level of income, the nominal quantity of money, the interest
rate, the nominal quantity of government debt held by the Federal Reserve,
and the nominal quantity of debt held by the public) is written with the
use of matrix notation as a function of lagged values of these same six
variables. Of all the different tests that this technique makes possible,
the one of greatest relevance here involves the equation in which the quantity
of money is the dependent variable. The question of whether or not the
central bank tends to monetize deficits is answered by looking at the coefficients
of the lagged and contemporaneous values of the debt variables. But if
Dwyer's own caveats are taken seriously, his negative empirical results
do not have the significance that he ascribes to them. The key issue is
the concept of "causality" in economic relationships.
While the Granger-Sims technique
does not impose any particular theoretical restrictions on the relationship
among the six variables singled out by Dwyer, it does impose and important
general
restriction. Each variable is related only to contemporaneous or past values
of the other variables. No allowance is made for the effect of anticipated
future values of those variables.(2) The
Granger-Sims technique is confined to the identification of post hoc
relationships because of this general restriction. Unfortunately, this
inherent shortcoming of the technique has been disguised as a virtue (both
in Dwyer's article and in the literature in general) by imputing
causality
to the relationships identified. But to take the earlier event as the cause
of the later event is precisely to commit the logical fallacy of post
hoc, ergo propter hoc.
A dim recognition of the
post
hoc fallacy has led users of the Granger-Sims technique to retain the
word "cause" while disclaiming its conventional meaning. Dwyer clearly
recognizes this practice when he writes that "Interest rates lead inflation,
i.e., last period's interest rate ... 'Granger-causes' this period's inflation
rate, but this does not mean that interest rates 'cause' inflation in the
usual sense" (p. 323). Lamentably, the unusual sense of the word
is not specified-unless "lead" and "Granger-causes" are taken to be perfectly
synonymous, in which case the latter term is redundant at best and misleading
at worst. The truth is that "Granger-causes" does not me "causes" at all.
What does it mean to say that, for example, toting an umbrella to work
on a cloudy morning "Granger-causes" rain to fall sometime during the day?
If debt is monetized in a given period because of the anticipation of even
greater government borrowing in subsequent periods, then the monetary creation
and the subsequent inflation should be attributed to the deficits. The
anticipation of large deficits, in this hypothetical case, causes money
creation.(3) The Granger-Sims technique,
however, would identify the toting of the umbrella and the creation of
money as the causes and the mid-day rain and the large deficits as the
effects.
In the final paragraph of
the summary section, Dwyer uses the word "causal" but without the quotation
marks he had been careful to use earlier. The reader is likely to believe
that he means "causal" in the usual sense. Once it is recognized he does
not, it is easy to reconcile Dwyer's conclusions with those of Buchanan
and Wagner. He states "there is no reason to predict that a reduction of
deficits has a causal role in any policy to reduce inflation" (p. 327).
It need only be said here that there is no contradiction between the econometrician's
findings that deficits do not Granger-cause inflation and the political
economist's understanding that deficits do Webster-cause inflation.
In order to produce empirical
results, Dwyer deliberately ignored the issue of market dynamics and hence
the underlying economic processes. But only insofar as it is understood
how the different variables are related one to the other is some theoretical
and historical context can it be discovered what causes what. Identifying
the relevant incentives and constraints out of which such theory and history
are made is precisely what Buchanan and Wagner did, and what Dwyer did
not do.
References
Archibald, G. C., and Lipsey, R. G., "Monetary and Value
Theory: A Critique of Lange and Patinkin," The Review of Economic Studies,
October 1958, 1-22.
Bresciani-Turroni, Constantino, The Economics of Inflation,
Augustus M. Kelley, Publishers, Northhampton, Great Britain, 1937.
Buchanan, James M., and Wagner, Richard E., Democracy
in Deficit: The Political Legacy of Lord Keynes, Academic Press, New
York, 1977.
Dwyer, Gerald P., Jr., "Inflation and Government Deficits,"
Economic Inquiry, July 1982, 20, 315-29.
Graham, Frank D., Exchange , Prices, and Production
in Hyper-Inflation: Germany, 1920-1923, Princeton University Press,
Princeton, New Jersey, 1930.
Patinkin, Don, Money, Interest, and Prices: An Integration
of Monetary and Value Theory, 1st ed., Row, Patterson, and
Co., Illinois, 1956.
Notes:
1. This characteristic of the real
debt's time profile is recorded in Bresciani-Turroni, pp. 30-33. And empirical
study that adopted a longer time horizon and considered the debt profile
beyond 1922 would yield different results. In this later period the debt
itself rose dramatically. The exponential rise can be attributed largely
to the asymmetrical effect of inflation on the Reich's receipts and on
its expenditures (Bresciani-Turroni, pp. 64-70). But the increased deficits
during the extended period does not support the view that deficits cause
inflation any more than the initial decrease in deficits calls the causal
relationship into question.
2. Dwyer recognizes the existence (p.
323) but not the significance of this problem.
3. Similarly, if the central bank maintains
a tight-money policy in the face of high deficits and then, succumbing
to political pressures, begins to monetize the cumulative deficits, the
money creation can be attributed to the deficits. But because of the political
element, this relationship between the deficits and the money creation
may not be systematic enough to be captured by the Granger-Sims technique.
|