Printer Friendly

Some real evidence on the Real Bills doctrine versus the Quantity Theory.

Researchers interested in testing between the Real Bills doctrine and the Quantity Theory approach to inflation in the face of rapid, deficit-financing money growth are confronted by an observation equivalence problem. This paper identifies a data set which resolves the dilemma and tests the two inflation models. The results provide clear evidence supporting the Real Bills doctrine, that the value of assets backing money determines its value, over the Quantity Theory.


One of the more interesting developments in monetary theory in the early 1980s was the re-emergence of the Real Bills doctrine. The doctrine identifies the value of assets backing the money supply as the prime determinant of the money's value. Its adherents claim that the monetization of "worthless" debt is responsible for inflation, and not the increase in the money supply itself.[1] That is, an increase in the money supply will not lead to inflation so long as the money issue is backed by "real" assets. Sargent [1982], attributes the ends of four major inflations to exactly this phenomenon.[2] In contrast, the Quantity Theory claims that changes in the money supply should have inflationary consequences regardless of its backing.

The new interest in the old doctrine parallels a quite different line of work centering on what looks like a similar theme. This second track focuses, too, on the rate of inflation and the fiscal wellbeing of the government, though here the transmission mechanism from deficit to inflation is rather conventional. In this essentially monetarist vision, if the deficit is "out of control" and the monetary authority is to some degree subservient to the fiscal authority, there must be some eventual monetization of the resultant debt, resulting in inflation. To the extent that higher rates of inflation are expected in the future, higher rates of inflation may occur now.[3]

Both lines of thought argue that a deficit that is "out of control" is responsible for inflation. It is the transmission mechanism between this excessive deficit and the price level that distinguishes the two models.

The Real Bills doctrine proposes a radical departure from traditional monetary theory by suggesting that money is, in effect, efficiently priced. The central bank is seen as being similar to any other financial intermediary, purchasing debt and issuing liabilities, and those liabilities (money) are then priced according to the value of the assets they were used to purchase (the value of the assets backing the money). Thus money gets priced efficiently, and its value is not determined by its relative or expected relative quantity, but rather by the "real" characteristics of its backing. That is, ceterus paribus, an increase in the money supply will not, per se, be inflationary. The increase in the money supply will only be inflationary if the assets purchased with the money issue are of such low value as to reduce the average value of the assets backing the currency.

This paper provides some evidence in favor of the Real Bills vision of price level determination over the Quantity Theory. The next section describes the observational equivalence problem confronting researchers interested in distinguishing between the Real Bills and Quantity Theory approach to inflation in the face of rapid deficit-financing money growth and the institutional setting which resolves the observational equivalence dilemma. Section III describes some tests designed to discriminate between these two hypotheses of the inflation process and their resuits, and section IV contains my conclusions.



A central theme of both of the Real Bills and Quantity Theory hypotheses is that fiscal deficits that are out of control will be inflationary, though for very different reasons. Apart from any insight gained with regard to the inflation process, testing the basic proposition that deficits are inflationary is problematic for two reasons:

First, there is the standard problem of government accounting standards and practices in general and, consequently, of the appropriate definition of the fiscal deficit.

Second, there is the problem of defining what constitutes a deficit that is "out of control." Feasible deficits may be large and (expected to be and actually) short lived, or relatively small and continuous. In neither of these cases is the deficit out of control, yet the contemporaneous identification of either condition is far from easy.[4]

Even if it can be shown that an unambiguously out-of-control deficit causes inflation, there is an observational equivalence between the Quantity Theory and Real Bills hypotheses. It is to this problem that we now turn.

The Problem

The current discussion of the Real Bills doctrine was motivated by Sargent's look, among others, at inter-war European hyperinflations, which he claimed were caused by central bank monetization of a vast amount of bad debt. The argument is that the inflations were caused not by the large increase in the money supply, per se, but rather by the fact that this additional money was backed by worthless government bonds. As the value of the central bank's asset portfolio decreased, so too the value of its liabilities decreased.

Consider this simple example focusing on the interaction between the monetary and fiscal authorities. The actions of the monetary authority are described by the balance sheet equality (this term is used since the version of the Real Bills doctrine discussed here focuses on the asset side of the central bank's money liabilities):

(1) Delta (Central Bank Money)[.sub.t] =

Delta (Bonds in Central Bank Portfolio)[sub.t].

Equation (1) simply describes the effect of open-market operations on the central bank's T-accounts: a change in central bank liabilities (high-powered money) must equal its change in assets.

The fiscal authority's actions are summarized by their budget constraint:

(2) [Delta] (Bonds Outstanding)[sub.t] =

(Government Spending)[sub.t] - (Tax Revenue)[sub.t].

In the hyperinflation cases, the lefthand side of equation (1) is frequently equated to the right-hand side of equation (2), forming a single fiscal budget constraint where tax revenue is augmented with seigniorage. Seigniorage is separated here, since the Real Bills argument revolves around the balance sheet of the monetary authority, and this separation is an important part of the Real Bills argument.

In the case of the inter-war hyperinflations, the fiscal deficit of the countries involved, as represented by the right-hand side of (2), were both quite large and "outof-control" in the specific sense that there seemed to be no mechanism in the existent political regime that would bring them to an end. As a consequence, the bonds issued to finance the deficit were worthless, i.e., no one would willingly buy them at prices and in quantifies sufficient to finance government expenditure. The monetary authority was either factually or functionally under the control of the fiscal agent, and thus the monetary authority was forced to monetize the deficit. In effect, the central bank acquired the worthless debt as a portion of its portfolio, making the change in government debt outstanding equal to the change in the stock of bonds held by the central bank, and thus making base-money growth large.s

The hyperinflations ended, Sargent argues, some sort of regime change functionally separated the monetary authority from the fiscal authority, with the result that further increases in central bank liabilities were thereafter backed exclusively by "good" debt. Further, Sargent claims that the central bank did not essentially otherwise change its remarkable rate of base-money creation. That is, it was frequently the case that the central bank's rapid acquisition of new debt (high rate of money growth) would continue for some time well after the inflation came to an end. Thus he observes a large increase in money now backed by "real" assets, that did not result in inflation. This, he reasonably claims, is distinctly at odds with a Quantity Theory of money approach to price level determination.

In the case of the inter-war hyperinflations, the regime change that resulted in a balanced budget clearly stopped the deterioration of the value of the central bank's (and everyone else's) portfolio of government bonds. If the price level then immediately stabilized without a change in the growth rate of money, however, this does not immediately lend support to the Real Bills doctrine over the Quantity Theory. There may be an observational equivalence.

Consider the simple, quite standard money demand formulation


with the left-hand side representing current real money balances, and the demand for money a function of real output-(Y), the real interest rate (r), and the time t expectation of future money, Et(M[ ] i), which, in the Quantity Theory world, represents a proxy for expected inflation. Suppose the economy works in a Quantity Theoretic way, such that the price level is determined by something like (3). Suppose further that the government is faced with a large revenue shortfall and is forced to monetize its deficit, effectively combining equations (1) and (2) above. In this case, the "excessive" money issue results in inflation. More importantly, however, the inflation is not driven just by current money growth, but also by expected future money growth.

As a consequence, a political regime change that brings government spending in line with tax revenue (balancing the budget in equation (2)) will shift money demand due to an expectation of lower rates of monetary growth in the future. Therefore, at the time of the fiscal regime change, the resultant increase in the demand for real money balances makes a substantial increase in the nominal supply on the part of the monetary authorities desirable, or a substantial decrease in the price level inevitable. That is, we may see a large increase in the money supply that has no inflationary consequences because the change in fiscal conditions has increased money demand. This is a standard result from the rational expectations, obstacles-to-curtailing-inflation literature (see Taylor [1975] or Phelps [1979]).

While the fiscal reform stops the devaluation of government debt, that same fiscal reform removes the necessity for "excessive" seigniorage and thus serves to lower expected future money growth, even if such a change in money growth is not immediately evident. The same sort of fiscal reform that would lead to a reduction in expected monetary growth rates, satisfying the monetarist vision of the inflation process, would also eliminate the problem of the central bank having to monetize more worthless debt, thereby stabilizing prices in the Real Bills vision of the inflation process.

That is, in the case of the hyperinflation stabilizations, the Real Bills and Quantity Theoretic explanations of the inflation process are observationally equivalent. After the inflations ended it was frequently the case that monetary base growth continued at extraordinary rates for quite some time. This rapid expansion of the monetary base after the fiscal reform (and after the inflation ended) is not necessarily inconsistent with the monetarist paradigm, since the dynamic driving money growth had changed with the regime, and so too, then, might money demand. Thus the distinction between the hyperinflation ending due to a radical money demand shift brought about by a perceived future change in the money supply process and an end to the inflation due to the change in the backing of the money is one that may be difficult to make--in the case of the hyperinflations they are observationally equivalent.

A resolution to this problem can be found by examining an economy that behaves in the manner described by Sargent as associated with post-hyperinflationary economies, i.e., one where the central bank is rapidly expanding the monetary base for some reason that does not involve the acquisition of large amounts of debt at a price other than that established by the market. That is, the Real Bills doctrine suggests that an economy that for some reason has a rapid expansion in the money supply that is not accompanied by a deterioration of the average asset value of the central bank's portfolio (where the central bank is rapidly monetizing only "real" assets), should not have an inflation problem, whereas clearly the Quantity Theory suggests that they should. The recent experience of the Republic of China on Taiwan fits this description and can provide a test of the Real Bills versus Monetarist propositions.

The Solution: The Curious Institutional Case of Taiwan

The fiscal history of the Republic of China on Taiwan can loosely be divided into two parts. Immediately after establishing the government in Taipei, Taiwan issued a substantial amount of government debt that was used largely for infrastructural development. This period of government cash-accounting deficits lasted about a decade. Since then the government has tended to run a slight annual budget surplus. Lumpiness in tax revenue collections and a lack of sophisticated cash management techniques on the part of the government lead to a periodic issuance of short-term treasury bills, but rarely for a duration of more than nine months. Consequently, the government debt to income ratio has been falling until the slight and intentional deficit of 1988.

Up until July of 1987, currency markets were non-existent, and the economy was functionally closed financially. The central bank was the sole legal currency trader and would buy and sell foreign exchange at a specified rate.6 The private holding of foreign currency was restricted. The effect of this institutional arrangement was to force the central bank to monetize Taiwan's trade surplus. Export industries' (trading with the United States) commercial banking representatives obtained payment in dollar-denominated accounts in U.S. bank demand deposits, and the commercial banks sold these accounts to the central bank in exchange for high-powered money. Thus the immediate result of a trade surplus was an increase in the base-money supply, essentially backed by U.S. dollars.

Of course, any country with a fixed exchange rate loses control over its domestic money supply. The case of Taiwan, however, is interesting in that institutional arrangements made the loss of control of money manifest itself purely in a loss of control of the quantity of high-powered money. Further, the size and persistence of the balance-of-trade surplus made for remarkable rates of base-money growth. The inflation that is supposed to result from this process and which would bring to an end. the chronic trade surplus never occurred.[7]

Given these institutional arrangements, it is useful to think of the monetary base growth function as

(4) A (Base Money)t - (Trade Surplus)t + (Other OMOs)t.

The "Other OMO" (open-market operation) source of base growth was limited by the lack of domestic financial instruments with which to perform the operations. The government did not issue longterm debt, and corporate debt markets were functionally non-existent. Thus the trade surplus was responsible for the bulk of the change in the monetary base.[8] The forced monetization of the trade surplus resulted in some remarkable money growth rates, as illustrated in Figure 1 (DRESSA).[9]

Clearly the case of the Republic of China seems to fit the requirements of a situation capable of resolving the observational equivalence dilemma. The institutional arrangements in Taiwan mean that the central bank having to monetize the trade deficit is essentially equivalent to other central banks having to finance the budget deficit. The continued double-digit rates of money growth unaccompanied by a government deficit, along with a reasonably well-defined backing of the money, make Taiwan an ideal case with which to test the Real Bills versus Quantity Theory propositions.


Three Test Forms

Granger Tests. With rapid money growth backed by foreign exchange reserves consisting largely of U.S. dollars, one way to distinguish the Quantity Theory from the Real Bills doctrine is through a standard Granger-causality test. In the case considered here, two sets of Granger tests are conducted. The first set is the case of


where P is inflation and M is either money growth or the change in the value of money's backing. This examines whether money or the money's backing, independently, can be said to Granger-cause domestic inflation. The Quantity Theory would suggest that the [Gamma.sub.i.s] are significant when M is money, while the Real Bills doctrine would suggest that they are significant when M is the backing.

Estimations of these equations as specified may bias the tests against the Quantity Theory due to an omission of any adjustment for the income elasticity of money demand. Certainly the money series used is quite volatile, but with a small open economy, income may also be quite volatile, and so would the demand for money. Thus an adjustment for changes in income may be needed. An additional set of tests is included, subtracting the rate of change of an index of industrial production (Figure 3) (a proxy for income) from the rate of change of the money supply.[10]

The second set of Granger tests is generalized further by including both money growth and its backing together:


The idea is to test whether money's backing Granger-causes inflation when its quantity is included in the regression, or whether money Granger-causes inflation when its backing is included. Again, the Quantity Theory would suggest that the

[Gamma.sub.1,i] set would be significant and the 2d set not, while the Real Bills doctrine would suggest the converse.

A VAR. The second test considered is a VAR; essentially a multivariate version of the Granger test above. In order to discriminate between the Real Bills and Quantity Theory hypotheses, a likelihood ratio test is performed on the exclusion of domestic money and industrial production from a system containing domestic and U.S. inflation. The question tested is whether domestic money and income add any information to a forecast of domestic inflation (and U.S. inflation) otherwise based on a system of just domestic and U.S. inflation. That is, is the price system exogenous with respect to domestic money and output?

The VAR provides a more robust test form than the univariate Granger tests with respect to, for example, the problem of misspecification of money demand, in terms of possibly rejecting the quantity theory. On the other hand, the VAR cannot make inferences about the causal ordering of domestic versus U.S. inflation, as is the case with the Granger test, and thus may be somewhat less powerful in terms of its implications about the Real Bills doctrine. Variance Decompositions. As a final test, a variance decomposition of forecast errors can provide some additional evidence in determining the relative explanatory importance of the variables included in the VAR above. The variance decomposition gives the n-period ahead percentage of forecast error in a VAR that is attributable to a shock in one of its variables.

As is the case with the two tests above, the Real Bills approach would suggest that innovations in the value of the backing of the money would explain a greater percentage of the innovations in the inflation rate than would money and/or output innovations (which, presumably, would have little explanatory power). Conversely, the Quantity Theory would suggest that innovations in the backing of the money would explain none of the innovation in the inflation rate, while innovations in money and output would explain a great deal.

Despite their individual potential weaknesses, the results of these tests taken together are capable of providing substantial insight into the distinction between the Real Bills and Quantity hypotheses.

The Data

The tests require four data series to be used. The rate of change of the price level in Taiwan is measured as a percentage change in the domestic consumer price index. The money measure used is the percentage change in the stock of bank reserves, i.e., the high-powered money stock issued by the central bank. The industrial production index series is essentially the IMF's International Financial Statistics (IFS) series. Since U.S. dollars form the bulk of the reserves backing the high-powered money, I take as a proxy measure of the change in the value of the assets backing the unit of account, the percentage change in the U.S. Consumer Price Index.[11]

Though the Republic of China is no longer a member of the IMF, the Central Bank of China continues to publish the same data in a publication replicating the IFS format. All series are monthly and begin in January 1970. The sample period ends in May of 1987, when substantial structural changes occurred in Taiwan's financial markets.[12] Time-series plots of the data series appear in Figures 1 through 3. Figure 1 plots money growth (DRESSA) against domestic inflation (DCPISA). Figure 2 presents U.S. inflation, while Figure 3 shows growth in domestic industrial production.

Seasonal adjustment of the data series turns out to be a non-trivial problem. Taiwan is not an especially large economy, and overall activity is thus potentially volatile. In addition to inherent volatility, the government never undertook any serious macro-stabilization programs (for example, there is a substantial spike in interest rates at the time of the lunar new year), and thus whatever seasonal volatility exists in the private sector is largely undamped by actions of the fiscal or monetary authority. As a result, the large volatility in the raw series produces volatility in the seasonally adjusted series that may not be directly temporally associated with the underlying data.

Two standard seasonal adjustment procedures were undertaken here: the standard regression of the series on monthly dummies, and exponential smoothing of the series. Both yield similar results, and only the monthly dummy results are reported. This seasonal adjustment is accomplished by using eleven monthly leads over the data interval, and thus shortens the sample period to 86:04.


Results of the Granger test estimations are contained in Tables I and II, while the VAR results are in Table III and the variance decomposition results are in Table IV.

In Table I, results of the test restriction that the coefficients on money are jointly equal to zero (in equation (5)) are reported in the first line of each particular lag length, while results for the restriction that the coefficients on the backing of the money are jointly zero are reported in the second line of the same table. The bottom half of the table contains results for the post-oil shock period.

Table II replicates Table L but for equation (6) and with the change in the money supply adjusted by the change in output.

Standard tests of lag length specification indicate that little information is added at lengths n > 13 (thus the lag length -- 12 in VARs unless otherwise stated). Nonetheless, the reported results run from n- 6 to n--18 to allow a monetarist money-to-price transmission mechanism at least six quarters in which to work.[13]

The tables look strikingly alike in terms of their indication of significance of blocks of explanatory variables. Thus, the interpretation of the results is quite straight forward.

In none of the cases is there any evidence that changes in the rate of domestic reserve money growth have any power to explain changes in inflation rates regardless of whether the change in the value of the money's backing is included in the explanatory equation. That is, there is no evidence that money Granger-causes inflation. Further, with the exception of long lag lengths (n [equal to or greater than] 14) in the post-oil shock period with income-adjusted money included in the regression, innovations in the value of the backing of the money are clearly Granger-prior to innovations in the value of the money itself. That is, innovations in the relative quantity of money do not Granger-cause innovations in its value, while innovations in the value of the backing of the money do.

Further support for the Real Bills doctrine over the Quantity Theory is found in the results of the VAR. The chi-square statistics in Table III result from a likelihood ratio test (with Sims's small-sample correction) that domestic and U.S. inflation are exogenous with respect to domestic money and output. The results for both the entire sample period and the post-oil shock period strongly fail to reject this hypothesis. While the VAR results add little with respect to the Real Bills doctrine other than a confirmation of the Grangerordering tests, the test does add confidence to the rejection of the pure quantity theory, since the atheoretic nature of the test provides a result more robust to the possible specification errors discussed earlier.

The variance decomposition results in Table IV show that the backing of money dominates its relative quantity in explaining innovations in the inflation process. Since forecast error decompositions are sensitive to variable ordering in the decomposition, Table IV presents the best and the worst results of all ordering permutations in terms of supporting the backing approach over the quantity theory, labeled strongest case and weakest case, respectively. They do not look very different. For the entire sample period, after a passage of six quarters, about 19 percent of innovations in the depreciation of the domestic currency is explained by innovations in the backing of the currency, while domestic money and output, combined, explain about half of that. In the post-oil shock period the difference between the backing of the money and its relative quantity is less dramatic, with about 27 percent of the innovation in the inflation process explained by innovations in the value of the numeraire's backing, and about 25 percent explained by the combination of both money growth and industrial production.

Overall, the results of various tests tend to persistently and frequently strongly favor the backing version of the Real Bills doctrine over the Quantity Theory.


The remarkable conditions found in 1970-87 Taiwan offer some reasonable empirical experience to support the backing version of the Real Bills doctrine over the Quantity Theory of money. Taiwan experienced a prolonged period of substantial money growth with little inflationary consequence. A priori, this condition would seem to speak against the Quantity Theory.

More importantly, as the rapid rate of monetary expansion was not motivated by a fiscal budget constraint, Taiwan provides a resolution to the observational equivalence problem associated with distinguishing between the Real Bills doctrine and the Quantity Theory. The tests presented here clearly indicate that a very rapid expansion of the domestic money supply may not have inflationary consequences so long as teh money is adequately backed. Indeed, the results indicate that innovations in the value of the money's backing are responsible for innovations in the value of the money itself. This implies that money may be efficiently priced, as suggested by the Real Bills doctrine. Indeed, when the change in the value of the central bank's portfolio is included in an explanatory equation for inflation, adding money growth provides no additional information. I interpret this as some clear support for the Real Bills doctrine over the Quantity Theory.
COPYRIGHT 1992 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Cunningham, Thomas J.
Publication:Economic Inquiry
Date:Apr 1, 1992
Previous Article:Takeovers: managerial incompetence or managerial shirking?
Next Article:Does industrial diversity always reduce unemployment? Evidence from the great depression and after.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters