The post-Keynesian view

One of the most significant post-Keynesian developments has been the emphasis on net private wealth as well as income as a factor influencing real flows of expenditures. The connection between net wealth of the private sector and consumption was first pointed out by Pigou ² and Haberler,³ and in a more rigorous manner by Patinkin,⁴ in the form of the real cash balance effect: changes in the real quantity of money could affect real aggregate demand even if they did not alter the rate of interest. The central feature of the real cash balance effect is the assumption that the stock of money is a component of the net wealth of the economy. However, the stock of money that these economists considered as part of net wealth was not the usually defined concept of narrow money (currency outside banks plus demand deposits) but rather the monetary base, or, in Gurley and Shaw’s terminology, outside money alone.⁵ The justification for excluding demand deposits (inside money) as part of wealth is that these deposits are claims of the public on the banking system that are counterbalanced by the debts of the public to the banking sector. Therefore, when the balance sheets of all economic units are consolidated, inside money disappears and, hence, should not be considered as part of wealth.

This justification, however, brings out immediately the question of why the same logic should not be applied to outside money, which is, after all, the noninterest-bearing debt of the government. If, indeed, outside money is government debt, as it has been treated in the post-Keynesian literature,⁶ a consolidation of the balance sheet of the private and government sectors must result in the cancellation of outside money as an item of net wealth. It then follows that a change in outside money cannot exert a wealth effect, since the change cannot cause a simultaneous change in net wealth. To establish that the change in outside money does indeed have economic consequences, one has to provide an explanation other than that outside money is part of wealth. The explanation that has had widest support has been that the government, unlike other debtors, is unconcerned about the size of its debt and makes its economic decisions accordingly.⁷ Thus, the only effect of the change in outside money is the effect of the increase in assets of the private sector. This explanation reduces the wealth effect of changes in outside money to a distribution effect of a wealth transfer between the private and government sectors. This explanation also suggests that the traditional distinction between inside and outside money is based not on a measure of wealth applicable to all types of asset but rather on asymmetric responses of various economic decision-making units to changes in assets and debts.⁸

Later developments have extended this wealth effect beyond money to other forms of wealth, such as the real market values of equities and interest-bearing government debt. It is fairly easy to appreciate the wealth effect or distribution effect associated with outside money. However, it is not evident whether the same argument could be applied to interest-bearing government debt. There is an important difference between the two assets. Unlike outside money, the interest burden on interest-bearing government debt must be financed by future taxes. Hence, if the private sector discounts its future tax liabilities in the same way in which it discounts future interest receipts, the existence of government bonds represents an asset as well as a liability to the public and will, therefore, not generate any net wealth effect.⁹ However, if the public considers only a constant fraction of total interest-bearing government debt as a liability, then an open market purchase of government bonds will increase net private wealth and thereby directly affect aggregate demand.

At the theoretical level, the link between the net wealth of consumers and real consumption has been refined as in the life cycle hypothesis of Ando, Brumberg, and Modigliani, which holds that consumers allocate consumption over their lifetime, given initial net worth, a rate of time preference, and expectations regarding labor income.¹⁰

The Keynesian approach uses the rate of interest on long-term bonds as the representative rate on all types of earning asset. This implicitly assumes that equities, government bonds, and private debt are all perfect substitutes for one another.¹¹ As a result, the Keynesian analysis considers only substitution between money and bonds important but ignores entirely the substitutability between money and real assets or real expenditures. The intellectual importance of the real cash balance effect lies in the fact that it allowed the possibility of substitution between money, on the one hand, and real expenditures, on the other hand, in macroeconomic analysis.¹² This fact has, to a great extent, contributed to the explicit treatment of real capital goods in portfolio analysis and to a re-emphasis on the role of money.

The Pesek and Saving thesis

In a recent book by Pesek and Saving,¹³ the authors argue that commercial bank demand deposits (inside money) should also be treated as part of the net wealth of the economy, because demand deposits are an asset produced by banks and sold by them to the public in exchange for the latter’s statements of indebtedness.¹⁴ Demand deposits, unlike outside money, carry an “instant repurchase clause”—an obligation by the banks to repurchase them with outside money. But this characteristic, Pesek and Saving argue, does not affect the basic fact that demand deposits are part of net wealth.

If Pesek and Saving’s thesis is valid, then their analysis will have important implications for the effects of monetary policy. Their analysis literally means that monetary authorities can directly create or destroy nominal private wealth at will through monetary policy. This, in turn, implies that monetary policy can affect aggregate demand directly and in theory can have strong effects on the level of economic activity. For instance, a central bank’s open market purchase of government bills, given the money supply multiplier, will generate a multiple expansion of the initial Increase in outside money. The increase in money supply should be considered an increase in nominal private wealth, which will, in turn, increase consumption expenditure. This means that the open market operation has a direct wealth effect in addition to the conventional liquidity effect¹⁵ and the interest-rate-induced wealth effects (capital gains or losses on government bonds and equities) on total spending. For this reason, Pesek and Saving’s argument deserves careful examination.

Pesek and Saving’s view is based on the principle that the economically relevant measure of wealth is the capitalized value of a stream of net income. When this measure of wealth is applied to outside money, it is shown that outside money is a component of net wealth because it yields a net flow of services to the user—not because the government is unconcerned with its outstanding debt. The flow of services of outside money is the saving of time in barter transactions, which stems from the role of money as a medium of exchange. The saving of time may be used either for leisure or for the production of capital goods. Hence, outside money generates a positive stream of income; the capitalized value of this income is then an addition to the net wealth of the economy. The same reasoning applies to demand deposits, the basic difference between government fiat money and demand deposits being one of institutional arrangements. The arrangement is that government fiat money is produced by the government, whereas demand deposits are produced by the government agents, namely, commercial banks that have received the monopoly right of producing money. From this point of view, there is no difference between outside money and demand deposits. To put it differently, inside money also represents a stream of net income, since demand deposits, as a medium of exchange, yield a flow of services that banks have agreed to provide. For inside money, however, the net income takes the form of the earnings of the banks through the process of money creation or intermediation. The capitalized value of the earnings is then an increase in the net wealth of the economy.¹⁶

To elaborate further on this point, let us assume that a commercial bank receives $5,000 of demand deposits and that the bank is subject to a reserve requirement of 20 per cent. The bank then retains $1,000 in cash reserves and creates demand deposits amounting to $4,000 by lending this amount to the private nonfinancial sector of the economy. Suppose that there is a single rate of interest, say, 5 per cent, in this economy. This means that, ceteris paribus, the bank is earning an additional $200. The present value of this interest earning, given the 5 per cent interest rate, is $4,000, which is exactly equal to the increase in demand deposits held by the public. This present value represents an increase in the net worth of the bank. Since the bank is a private one, the increase should be reflected also in the balance sheets of its stockholders; ceteris paribus, the increase in the net worth of the bank will be reflected in an increase in the market value of the bank’s stocks, which is equivalent to an increase in the public’s wealth.

The argument so far is based on the assumption that the costs of producing (or servicing) demand deposits are negligible. However, in reality the commercial banking industry has operating costs just like any other industry and is fairly competitive. To the extent that there are operating costs connected with demand deposits (or outside money, for that matter), the whole value of demand deposits is not necessarily part of the net wealth of the economy. In the extreme case where the banking industry is fully competitive and is not subject to government controls and regulations, Patinkin shows that the capitalized value of the banks’ operating costs (the sum of the present values of the current operating costs and of the imputed annual interest charge on the fixed assets) is equal to the value of demand deposits and that, consequently, demand deposits should be excluded from net national wealth.¹⁷ The implication of the perfect competition case is, of course, that the true origin of the net worth of the banking sector is in its monopoly right of creating demand deposits.

However, if the assumption of zero costs of managing demand deposits is unrealistic, so is the assumption of perfect competition in the banking sector. A reasonable assumption would then be that, since entry into the banking industry is restricted by the necessity of obtaining a license, the present value of the income from creating demand deposits should be regarded, in part, as a component of net private wealth. This conclusion in turn raises an important question as to whether the conventional specification of the economically relevant wealth equation for theoretical as well as empirical studies is a proper one, and if it is not, how it should be modified.¹⁸

Net private (nonhuman) wealth (W) is generally defined as the sum of outside money and the market values of interest-bearing government debt and of the existing stock of physical capital:

W = PK + M_o + B,

where P = price level; K = real stock of capital; M_o = outside money; B = supply of government bonds.¹⁹

The market value of real assets or equities (V) has been defined, for instance, as

where Y_cd = corporate dividend payments; R_d = dividend/price ratio on common stock (percentage).²⁰

On the other hand, Pesek and Saving define real nonhuman wealth for the purpose of general analysis as

where M = nominal quantity of money (narrow definition of currency plus demand deposits); P = price level; y_n = real nonhuman income; g = real nonhuman income yielded by government securities; r_n = capitalization rate applicable to real nonhuman income; r_g = capitalization rate applicable to real government interest payments.²¹

Brunner and Meltzer, following Pesek and Saving’s reasoning, also explicitly introduce the net worth of the commercial banking sector into their definition of nonhuman wealth:

W = PK + B + (1 + σ)M_o,

where σ = the banking system’s net worth multiplier; σM_o = the net worth contribution of the banking sector, or the capitalized value of net earnings of the banking system.²²

The basic difference between the definition of Pesek and Saving (or of Brunner and Meltzer) and the conventional one is, of course, that the former includes demand deposits (in whole or in part) as a component of net private wealth whereas the latter does not. At first sight, it may appear that the appropriate definition of wealth is the specification of either Pesek and Saving or Brunner and Meltzer. But that is not so. It was pointed out that any increase in the net worth of the banking sector would, ceteris paribus, lead to a corresponding increase in the net worth of the nonbanking private sector in the form of an increase in the market value of bank stocks. This means that the value of demand deposits in Pesek and Saving’s definition, or the net worth contribution of the banking sector (σM_o) in Brunner and Meltzer’s is already included in the market value of equities (PK, V, or Y_n/r_n) to the extent that demand deposits (in whole or in part) are part of net wealth; adding the net worth of the banking sector to the market value of equities amounts to a double counting of the market value of commercial bank stocks. For this reason, the wealth definitions of Pesek and Saving and of Brunner and Meltzer should probably be rejected. One might then conclude that the conventional definition of net wealth is economically appropriate regardless of the validity of Pesek and Saving’s thesis and that the direct wealth effect of monetary policy implied by their analysis has been accounted for in traditional analysis without being explicitly recognized.

The monetary sector

Given the assumption of unitary real income elasticity of the demand for money, equation (4) can be rewritten as

In order to simplify the exposition of the model, let us assume that the supply of money is determined exogenously:

Then equations (4)′, (5)′, and equilibrium conditions (6) yield

Equation (7) can be written as

The Fisherian distinction between the nominal and real rate of interest is given by the following identity:

where q = the real rate of interest; $\left(\frac{1}{p}\frac{dp}{dt}\right)=$ the rate of change of the price level.

From equation (9), it also follows that

where the variables with an asterisk refer to anticipated (or expected) values.

Following Keynes’ argument that market rates of interest are determined largely by speculators with firmly held expectations, Friedman assumes that

From the identity, Y = yp, it follows that

Combining equations (10), (11), and (12) we obtain

where $g*=\left(\frac{1}{y}\frac{dy}{dt}\right)*=$ the anticipated rate of growth of real income.

Substitution of equation (13) into (8) yields

Equation (14) states that the level of income Y is determined by q*, g*, $\left(\frac{1}{Y}\frac{dY}{dt}\right)$*, and M. Friedman assumes that the difference between q* and g* is a constant. In a static framework, the expected rate of growth of nominal income $\left(\frac{1}{Y}\frac{dY}{dt}\right)$* may be treated as a predetermined variable. Then equation (14) determines the level of nominal income for a given supply of money without any reference to the real sector of the model.⁵⁷ In a dynamic framework, however, it would be natural to regard ${\left(\frac{1}{Y}\frac{dY}{dt}\right)}^{*}$ as determined by the past history of nominal income. Since the past history of nominal income is in turn a function of the past history of money as implied by equation (8) for earlier dates, equation (14) becomes a relation between the level of nominal income at each point in time and the past history of the quantity of money.

This dynamic character of the model may be better understood by analyzing an example given by Friedman. Take the logarithm of equation (14) and differentiate with respect to time.

This gives

where $s=\frac{1}{H}\frac{dH}{dr}=$ the slope of the regression of log H on the rate of interest.

Suppose that the expected rate of growth of nominal income is determined by an adaptive expectation process:

Substituting equation (16) into equation (15) and solving for $\left(\frac{1}{Y}\frac{dY}{dt}\right)$, we have

When

$\left(\frac{1}{Y}\frac{dY}{dt}\right)*=\frac{1}{M}\frac{dM}{dt}$, equation (17) gives the quantity theory result that nominal income changes at the same rate as money supply.

Friedman states that his monetary model of nominal income corresponds to the broader framework implicit in much of the theoretical and empirical work that he and others have done in analyzing monetary experience in the short run and is consistent with many of their empirical findings.⁵⁹

However, Friedman’s model sheds little light on his view concerning the precise mechanism through which changes in the quantity of money affect income. This is so because the model is designed primarily for empirical analysis of the relation between money and income. What this model reflects is rather his view on the appropriate empirical approach to evaluating the role of money—a view that is in sharp contrast to the more common one held by nonmonetarists. Contrary to the impression that one might gather from his model—possibly a rigid and mechanical connection between money and income—his view on the transmission mechanism in a conceptual framework is a complicated portfolio adjustment process that involves many uncertain channels and impinges on a wide array of assets and expenditures. The process ⁶⁰ following an exogenous change in the supply of money begins with changes in the prices and yields of financial assets and spreads to nonfinancial assets. These changes in the prices of financial and nonfinancial assets influence spending to produce new assets and spending on current services. At the same time, these changes alter real wealth of the public relative to income and thereby affect consumption. This is, in a simple fashion, the way in which the initial impulse is diffused from the financial markets to the markets for goods and services. The exposition stresses portfolio adjustment and is strikingly similar to that described by many economists.

The transmission process is also essentially consistent with the Keynesian liquidity preference doctrine as to how money affects income. As in the income/expenditure theory, interest rates play a key role. This being so, it has been pointed out that Friedman cannot be saying anything different from Keynes ⁶¹ and that there is no clear reason why one should look at money supply as a target as Friedman insists rather than at interest rates directly.⁶² However, these criticisms miss the fundamental points of Friedman’s view. The difference between Friedman and the nonmonetarists concerning the transmission process is not whether changes in the supply of money operate through interest rates but rather (i) the range of interest rates considered and (ii) the empirical approach to estimating the actual influence of monetary policy. Friedman argues that the impact of monetary policy is likely to be understated in magnitude and narrowed in scope in the Keynesian income/expenditure theory. One reason is that, since monetary policy impinges on a broad range of capital assets and a correspondingly broad range of associated expenditures, the Keynesian practice of looking only at recorded market interest rates, which are only part of a much broader spectrum of rates, makes one underestimate the actual impact of monetary policy. The rates of interest that influence investment decisions are for the most part implicit yields and hence not observable, so that one cannot hope to obtain useful results by looking at relations between market interest rates and the categories of spending associated with these rates. Also, recorded market interest rates may not provide an appropriate measure of the cost of capital, since these interest rates are not real rates of interest that reflect the basic forces of productivity but nominal rates that are influenced by the expected rate of inflation. Moreover, monetary influences may work through channels that we have not been able to identify. In fact, it may not be possible to trace through any particular channel, as monetary policy operates through an extremely complicated process of portfolio adjustments.

For all these reasons, Friedman considers that even the most complex structure of a general equilibrium model cannot be expected to capture actual monetary influences adequately. A more reliable empirical approach would be to pursue the methodology of positive economies, the essence of which is to select the crucial and simple theoretical relationships that allow one to predict something large (such as GNP) from something small (for instance, the supply of money), regardless of the intervening chain of causation. One such relationship is claimed to be the velocity function relating income to money, which is the essence of the quantity theory; another is the multiplier relationship relating income to autonomous expenditure, which is the essence of the income/expenditure theory.⁶³

Friedman argues that the velocity function (that is, the relationship between money and nominal income) has been shown, on average, to be more stable and less affected by institutional and historical change than the multiplier relationship and that, consequently, the velocity function may be the key relationship in understanding macroeconomic developments.⁶⁴ It then follows that a much more promising approach to the question of evaluating the effects of monetary policy on the economy is to try to relate changes in income directly to changes in the quantity of money. Friedman’s monetary model of nominal income seems to reflect this point of view.

Friedman’s view, however, raises two important issues. One issue is concerned with the conditions under which one could derive a simple relationship between income and money. It has been argued frequently that Friedman’s view is valid only if the demand for money is not significantly influenced by the rate of interest. As indicated earlier, Friedman now states clearly and repeatedly that the rate of interest is an important determinant of the demand for money. If indeed the demand for money depends on the rate of interest, then one cannot—critics of Friedman argue—hope to find such a simple relationship between money and income, unless one specifies an independent theory of the determination of the rate of interest.

The Keynes and Fisher synthesis that Friedman incorporates into his monetary model appears to be one such independent theory. It is an independent theory in the sense that the determination of the nominal rate of interest depends on relationships outside the system of six equations. According to Friedman’s monetary model, the rate of interest is determined solely by the anticipated rate of growth of money income $\left(\frac{1}{Y}\frac{dY}{dt}\right)$ *, given the assumptions that r = r* and that q* – g* = k_o. While these assumptions are crucial to a model that allows one to relate current money income directly to current and prior quantities of money, it should be realized that they are also responsible for several defects of the model. One serious defect is that the model rules out liquidity or substitution effects of a change in the quantity of money on the interest rate.

In Friedman’s monetary model, a change in the rate of change of money supply $\left(\frac{1}{M}\frac{dM}{dt}\right)$ directly affects the rate of growth of nominal income $\left(\frac{1}{Y}\frac{dY}{dt}\right)$ (see equation 17) with little direct effect on change in the interest rates, that is, the change does not initially produce the liquidity effect but immediately produces an income effect. The change in $\frac{1}{Y}\frac{dY}{dt}$ then causes a change in the expected rate of growth of nominal income $\left(\frac{1}{Y}\frac{dY}{dt}\right)$ * (equation 16) which in turn influences the nominal rate of interest (equation 13). In other words, the change in the rate of change in money supply affects market rates of interest only as it influences the courses of current and expected nominal income and, in consequence, the expected rate of inflation. This process does not appear to be supported by the existing empirical evidence on the effects of changes in money supply on the interest rates and income.⁶⁵ Nor does it seem to be consistent with Friedman’s own earlier exposition on the transmission mechanism of monetary policy, in which he argues that a change in the supply of money has its first impact on the financial markets and much later on the market for goods and services.⁶⁶

However, the failure of the model to explain the liquidity effect is clearly not a point of disagreement with Friedman, for he admits that the model neglects the effects of changes in the nominal quantity of money on interest rates.⁶⁷ Yet, it is not clear to what extent he appreciates the analytical significance of relaxing either of his two critical assumptions to permit a strong liquidity effect in the model. Relaxation of these assumptions would mean that Friedman’s model would be modified in such a way that the rate of interest is determined within the model through the interaction of demand for and supply of money together with other behavioral relationships in the real sector of the economy. Such a modification would then suggest that one may not utilize velocity to derive a simple relation between money income and the quantity of money.

The other issue that Friedman’s monetary model raises is whether the methodology of positive economics embodied in the model is a scientifically acceptable method. The general consensus seems to be that the methodology is seriously inadequate. We shall return to this topic in some detail in the final section of this paper.

Direct substitutability between money and real assets: the Fisherian view

As was pointed out in Section I, in Keynesian income/expenditure models, real capital goods do not appear as an asset in the portfolio because they are assumed to be a perfect substitute for long-term bonds. Owing to this assumption, the direct substitution relationship between money and real capital as opposed to the same relationship between money and bonds is not explicitly recognized. Several economists seem to misinterpret this Keynesian position as implying that money is a direct substitute for only financial assets, such as bonds, but not for real assets. A reaction to this misinterpretation has in turn led to the recent argument that money is a good (close) substitute not only for bonds but also for real assets and that, consequently, monetary policy operates “directly” in a framework in which real assets are explicitly introduced along with other assets. This argument is said to represent the monetarist view ⁷³ or to be associated with the ideas of Irving Fisher and other classical economists.⁷⁴ For the lack of better terms, we will call this argument the Fisherian view.

Although it seems doubtful, money may indeed be a good substitute for real capital goods. However, the basic question is whether this assumption leads to the conclusion that a change in the quantity of money affects real spending directly or not. In fact, it will be shown that the substitution relationship between money and real assets has no relevance whatsoever to the question of whether monetary policy works directly or indirectly. In order to show this and to elaborate further on the Fisherian view, we will begin our discussion by quoting three authors who support this view.

The monetary view assumes that money (or liquidity) is a direct (and good) substitute for real assets.⁷⁵ Since money is held primarily as a temporary abode of purchasing power it is quite reasonable to expect an OMO [open market operation] to affect spending on real goods, directly. In other words, since money is a good substitute for real assets, the increased volume of money in the portfolio will be spent on goods and services.⁷⁶

Monetarists favor a transmission mechanism in which an increase in money may directly affect expenditures, prices, and implicit yields on physical asets. They suggest that money may be substituted not only for bonds but also for other assets, and that individuals may re-establish portfolio equilibrium by purchasing either a financial or a physical asset.⁷⁷

In contrast to Keynes, Fisher believed that close substitution occurred between money and real capital, so that people in a Fisherian model respond to an increase in the supply of money by purchasing real commodities.⁷⁸ In the Fisherian model, with close substitution between money and real capital, people respond to the increase in the supply of money by purchasing goods.⁷⁹

All these assertions assume a high degree of substitutability between money and real assets. Given this assumption, it is then argued that changes in the supply of money,⁸⁰ through the process of portfolio substitutions, directly affect spending on real assets (or real goods or real commodities). However, it is not clear whether these economists mean that changes in the quantity of money directly affect (i) “flows” of real expenditures on goods and services (the flow of consumption and investment expenditures, namely, the flow of aggregate demand) or (ii) spending on “stocks” of secondhand real capital goods, such as plants, equipment, machinery, and houses. If they mean the former, (i), the argument is simply incorrect. If they mean the latter, (ii), the argument is no more than an explanation of the initial effect of changes in the quantity of money on balance sheets, namely, the substitution effect. The argument has nothing to say about how the initial substitution effect is translated into an effect on the income account—the essential aspect of the transmission process of monetary policy. It will be also shown that, regardless of whether the Fisherian view refers to (i) or (ii), the assumption most relevant to the argument is not whether money is a good substitute for real assets but whether any asset that the monetary authority deals with in the open market is a close substitute for real assets.

To substantiate these points, let us consider an open market purchase of government bills by the monetary authority. The immediate consequence of this operation is that owners of wealth are willing to hold more money relative to other assets in their portfolios because of their adjustment to the fact that the bill rate has been reduced. However, at this stage the interest rates on other assets have not been reduced. They will seek to correct this disequilibrium situation by converting money into other financial and real assets and so will tend to raise the prices of real assets and to reduce interest rates on financial assets. Unless the open market operation simultaneously changes the public’s net worth or its income—which it does not—the initial effect of the operation is not on income at all, but on the prices of existing assets. This effect, the substitution effect, is then subsequently transmitted to the real sector, and only at this stage are the levels of income and employment affected. On this transmission process of the substitution effect, the Fisherian view has little to say. Neither the explicit treatment of real assets in the public’s portfolio nor the degree of substitutability between money and real assets alters the nature of this process, although both factors may have different quantitative implications for the impact of monetary policy. Specifically, the introduction of real assets into the public’s portfolio suggests that the substitution effect of the open market operations may affect the implicit rates of return on physical assets, including consumer durables, and need not be restricted to a set of conventional interest rates on financial assets as in Keynesian income/expenditure models. A more adequate treatment of these yields in the specification of the consumption and investment functions may show that the influence of monetary impulses on the level of economic activity could be stronger and more rapid than has been found in the past empirical studies.

The Fisherian view could, of course, be interpreted as meaning that changes in money supply directly affect spending on “stocks” of real assets, not the “flows” of goods and services. Even in this case it is ambiguous whether the assumption of a high degree of substitutability between money and real assets has any particular bearing on the conclusion. When the monetary authority purchases treasury bills in the open market, it is making more funds available for the public to purchase a wide variety of financial and real assets. The pertinent question to ask in this regard would then be which asset (or assets) the public will “first” turn to. The presumption is that they would attempt to acquire first the assets comparable to those that they have sold to the monetary authority, namely, the assets that are close substitutes for treasury bills, not for money. This means that owners of wealth will purchase real assets first, if they regard bills as better substitutes for real assets than for other marketable financial assets, such as government and private bonds.

In the opposite situation, they would react to the operation by acquiring various financial assets rather than real assets and only when, or if, the prices of these financial assets have gone up sufficiently would they turn to equities, real property, and other capital goods. This means that if bills and real assets were poor substitutes for each other, the substitution effect of the open market operation on the demand for real assets, ceteris paribus, could be negligible, and hence a statement such as “people trade money directly for real capital goods” would lose much of its significance. The discussion so far suggests that the crucial factor in the Fisherian view is not, ceteris paribus, the substitution relationship between money and real assets but that between bills (or any assets that the monetary authority deals with in the open market) and other assets that are not money. Indeed, if money and real assets were very good substitutes for each other, the public would have less inducement to acquire real assets as a result of an open market operation. This is so because the public considers real assets to be similar to money, which is abundant relative to other assets in their portfolios, and, hence, less attractive candidates for portfolio diversifications; that is, for reducing the overall risk of the portfolio.

The Fisherian view—the second interpretation—also seems to suggest that the initial increase in money supply following the open market operation leads immediately to an increased demand for real assets and that, then, the interest elasticity of the demand for real assets is irrelevant.⁸¹ This argument would appear to be incorrect. The open market purchase of bills initially reduces the rate of interest on bills, otherwise the public would not be satisfied with the new portfolio composition. The reduction in the bill rate, ceteris paribus, will increase the demand for alternative financial and real assets. In these conditions, if the public responds to the lower bill rate by purchasing real assets instead of other financial assets, it must mean that the demand for real assets is not only elastic to changes in the bill rate but also much more sensitive to the bill rate than is the demand for financial assets other than treasury bills.⁸² This is, of course, another way of saying that treasury bills are better substitutes for real assets than for financial assets.

On the question of substitutability between all types of asset, there seem to be two opposing views. Tobin argues that all categories of government debt (money, treasury bills, long-term bonds, etc.) are good substitutes for one another, since their risk characteristics are quite similar, but they are poor substitutes for equities and other real capital goods because the risks associated with these two types of asset are substantially independent.⁸³ This view suggests that open market operations may not be powerful stabilization weapons, since relatively larger changes in interest rates are required to produce any given change in equity yields.

Friedman and Meiselman take the view that money is a temporary abode of purchasing power, which makes it a poor substitute for both real and financial assets. However, they argue that there is a high degree of substitutability among various financial and real assets.⁸⁴ If the Friedman and Meiselman view were correct, we would expect a stronger impact of changes in money supply than Tobin would claim.

The financial constraint: the crowding-out effect

A group of economists at the Federal Reserve Bank of St. Louis has for some time been engaged in the estimation of reduced form equations to evaluate the impact of money on economic activity. They estimate the response of GNP to monetary and fiscal actions by statistically relating GNP to a set of exogenous variables, such as the supply of narrow money (or the monetary base) and high-employment federal government expenditures and receipts.⁸⁵ The response of GNP to monetary actions was stronger, quicker, and more reliable than its response to fiscal actions. Their results would be more convincing if they could provide some explanation of how such actions affect total spending. They suggest that the reduced form equations they tested are potentially consistent with both Keynesian and quantity theory models.⁸⁶ In some of their recent writings, they seem to suggest that the potency of monetary policy comes from its ability to alter the financial constraint of the economy. According to Andersen, Federal Reserve actions influence total spending in two ways. First, its purchases and sales of government securities change the monetary base (high-powered money), which in turn varies the financial constraint of the economy, resulting in an immediate impact on total spending. A further impact results from the induced changes in income, interest rates, and prices, as the economy moves to a new equilibrium level of total spending. But the first impact accounts for a relatively fast response of economic activity to monetary actions.⁸⁷

The ineffectiveness of fiscal actions is also explained by this constraint. Given the financial constraint, “government spending unaccompanied by accommodative monetary expansion, that is, financed by taxes or borrowing from the public, results in a crowding-out of private expenditures with little, if any, net increase in total spending.”⁸⁸ According to Andersen, “a financially constrained economy … is one in which a decision to spend more than current income (i.e., borrow) by one economic entity can only be fulfilled by a reduction in the level of spending out of income by another entity.” ⁸⁹ Without explaining what the financial constraint represents, Andersen goes on to claim that the Federal Reserve System is an economic unit that is capable of varying this constraint through its monetary policy. Further research on this concept will undoubtedly result in a precise definition of the financial constraint. However, it is also clear that even when the term is defined the idea of the constraint has the distinct flavor of the credit rationing channel of post-Keynesians. At this stage, it is then fair to say that the economists at the Federal Reserve Bank of St. Louis do not appear to have developed any new channel of monetary policy that other economists did not recognize.