I have been writing recently about Keynes and his theory of the rate of interest (here, here, here, and here). Perhaps unjustly – but perhaps not — I attribute to him a theory in which the rate of interest is determined exclusively by monetary forces: the interaction of the liquidity preference of the public with the policy of the monetary authorities. In other words, the rate of interest, at least as an approximation, can be modeled in terms of a single market for holding money, the demand to hold money reflecting the liquidity preference of the public and the stock of money being directly controlled by the monetary authority. Because liquidity preference is a function of the rate of interest, the rate of interest adjusts until the stock of money made available by the monetary authority is held willingly by the public.

I have been struggling with Keynes’s liquidity preference theory of interest, which evidently led him to deny the Fisher effect, thus denying that there is a margin of substitution between holding money and holding real assets, because he explicitly recognizes in Chapter 17 of the *General Theory* that there is a margin of substitution between money and real assets, the expected net returns from holding all assets (including expected appreciation and the net service flows generated by the assets) being equal in equilibrium. And it was that logic which led Keynes to one of his most important pre-*General Theory* contributions — the covered-interest-arbitrage theorem in chapter 3 of his *Tract on Monetary Reform*. The equality of expected returns on all assets was the key to Irving Fisher’s 1896 derivation of the Fisher Effect in *Appreciation and Interest*, restated in 1907 in *The Rate of Interest*, and in 1930 in *The Theory of Interest*.

Fisher never asserted that there is complete adjustment of nominal interest rates to expected inflation, actually providing empirical evidence that the adjustment of nominal rates to inflation was only partial, but he did show that in equilibrium a difference in the expected rate of appreciation between alternative assets must correspond to differences in the rates of interest on loans contracted in terms of the two assets. Now there is a difference between the static relationship between the interest rates for two loans contracted in terms of two different assets and a dynamic adjustment in time to a change in the expected rate of appreciation or depreciation of a given asset. The dynamic adjustment does not necessarily coincide with the static relationship.

It is also interesting, as I pointed out in a recent post, that when criticizing the orthodox theory of the rate of interest in the *General Theory*, Keynes focused not on Fisher, but on his teacher Alfred Marshall as the authoritative representative of the orthodox theory of interest, criticizing Fisher only for the Fisher effect. Keynes reserved is comprehensive criticism for Marshall, attributing to Marshall the notion that rate of interest adjusts to equalize savings and investment. Keynes acknowledged that he could not find textual support in Marshall’s writings for this idea, merely citing his own prior belief that the rate of interest performs that function, consequently attributing a similar belief to Marshall. But even if Marshall did mistakenly believe that the rate of interest adjusts to equalize savings and investment, it does not follow that the orthodox theory of interest is wrong; it just means that Marshall had a defective understanding of the theory. Just because most physicists in the 18^{th} century believed in the phlogiston theory of fire does not prove that classical physics was wrong; it only means that classical physicists had an imperfect understanding of the theory. And if Keynes wanted to establish the content of the most authoritative version of the orthodox theory of interest, he should have been citing Fisher not Marshall.

That is why I wanted to have a look at a not very well known paper by Keynes called “The Theory of the Rate of Interest,” written for a 1937 festschrift in honor of Irving Fisher, *The Lessons of Monetary Experience*. Keynes began the paper with the following footnote attached to the title acknowledging Fisher as the outstanding authority on the orthodox theory of interest.

I have thought it suitable to offer a short note on this subject in honor of Irving Fisher, since his earliest [presumably

Appreciation and Interest, Fisher’s doctoral dissertation] and latest [presumablyThe Theory of Interest] have been concerned with it, and since during the whole of the thirty years that I have been studying economics he has been the outstanding authority on this problem. (p. 145)

The paper is mostly devoted to spelling out and discussing six propositions that Keynes believes distill the essentials of the orthodox theory of interest. The first four of these propositions Keynes regards as unassailable, but the last two, he maintains, reflect very special, empirically false, assumptions. He therefore replaces them with two substitute propositions, whose implications differ radically from those of orthodox theory. Here are the first four propositions.

1 Interest on money

meansprecisely what the books on arithmetic say it means. . . . [I]t is simply the premium obtainable on current cash over deferred cash, so that it measures the marginal preference . . . for holding cash in hand over cash for deferred delivery. No one would pay this premium unless the possession of cash served some purpose, i.e., has some efficiency. Thus, we can conveniently say that interest on money measures the marginal efficiency of money in terms of itself as a unit.2 Money is not peculiar in having a marginal efficiency measured in terms of itself. . . . [N]ormally capital assets of all kinds have a positive marginal efficiency measured in terms of themselves. If we know the relation between the present and expected prices of an asset in terms of money we can convert the measure of its marginal efficiency into a measure of its marginal efficiency in terms of money by means of a formula which I have given in my

General Theory, p. 227.3 The effort to obtain the best advantage from the possession of wealth will set up a tendency for capital assets to exchange in equilibrium, at values proportional to their marginal efficiencies in terms of a common unit. . . . [I]f

ris the money rate of interest . . . andyis the marginal efficiency of a capital assetAin terms of money, thenAwill exchange in terms of money at a price such as to makey=r.4 If the demand price of our capital asset

A. . . is not less than its replacement cost, new investment inAwill take place, the scale of such investment depending on the capacity available for the production ofA, i.e., on its elasticity of supply, and on the rate at whichy, its marginal efficiency, declines as the amount of new investment inAincreases. At a scale of new investment at which the marginal cost of producingAis equal to its demand price as above, we have a position of equilibrium. Thus the price system resulting from the relationships between the marginal efficiencies of different capital assets including money, measured in terms of a common unit, determines the aggregate rate of investment. (p. 145-46)

Keynes sums up the import of his first four propositions as follows:

These proposition are not . . . inconsistent with the orthodox theory . . . or open to doubt. They establish that relative prices . . . and the scale of output move until the marginal efficiencies of all kinds of assets are equal when measured in a common unit and . . . that the marginal efficiency of capital is equal to the rate of interest. But they tell us nothing as to the forces which determine what this common level of marginal efficiency will tend to be. It is when we proceed to this further discussion that my argument diverges from the orthodox argument.

Here is how Keynes describes the divergence between the orthodox theory and his theory:

[T]he orthodox theory maintains that the forces which determine the common value of the marginal efficiency of various assets are independent of money, which has . . . no autonomous influence, and that prices move until the marginal efficiency of money, i.e., the rate of interest, falls into line with the common value of the marginal efficiency of other assets as determined by other forces. My theory . . . maintains that this is a special case and that over a wide range of possible cases almost the opposite is true, namely, that the marginal efficiency of money is determined by forces partly appropriate to itself, and that prices move until the marginal efficiency of other assets fall into line with the rate of interest. (p. 147)

I find Keynes’s description of the difference between the orthodox theory and his own both insightful and problematic. Keynes notes correctly that the orthodox theory, abstracting from all monetary influences, treats the rate of interest as a rate of intertemporal exchange, applicable to exchange between any asset today and any asset in the future, adjusted for differences in rates of appreciation, and in net service flows, across assets. So Keynes was right: the orthodox theory is a special case, corresponding to the special assumptions required for full intertemporal equilibrium. And Keynes was right to emphasize the limitations of the orthodox theory.

But while drawing a sharp contrast between his theory and the orthodox theory (“over a *wide range* of possible cases *almost the opposite* is true”), Keynes, to qualify his disagreement, deploys the italicized (by me) weasel words, but without explaining how his seemingly flat rejection of the orthodox theory requires qualification. It is certainly reasonable to say “that the marginal efficiency of capital is determined by forces partly appropriate to itself.” But I don’t see how it follows from that premise “that prices move until the marginal efficiency of other assets fall into line with the rate of interest.” Equilibrium is reached when marginal efficiencies (adjusted for differences in expected rates of appreciation and in net services flows) of all assets are equal, but rejecting the orthodox notion that the marginal efficiency of money adjusts to the common marginal efficiency of all other assets does not establish that the causality is reversed: that the marginal efficiencies of all non-money assets must adjust to whatever the marginal efficiency of money happens to be. The reverse causality also seems like a special case; the general case, it would seem, would be one in which causality could operate, depending on circumstances, in either direction or both directions. An argument about the direction of causality would have been appropriate, but none is made. Keynes just moves on to propositions 5 and 6.

5 The marginal efficiency of money in terms of itself has the peculiarity that it is independent of its quantity. . . . This is a consequence of the Quantity Theory of Money . . . Thus, unless we import considerations from outside, the money rate of interest is indeterminate, for the demand schedule for money is a function solely of its supply [sic, presumably Keynes meant to say “quantity”]. Nevertheless, a determinate value for

rcan be derived from the condition that the value of an assetA, of which the marginal efficiency in terms of money isy, must be such thaty=r. For provided that we know the scale of investment, we knowyand the value ofA, and hence we can deducer. In other words, the rate of interest depends on the marginal efficiency of capital assets other than money. This must, however, be supplemented by another proposition; for it requires that we should already know the scale of investment. (p. 147-48)

I pause here, because I am confused. Keynes alludes to the proposition that the neutrality of money implies that any nominal interest rate is compatible with any real interest rate provided that the rate of inflation is correctly anticipated, though without articulating the proposition correctly. Despite getting off to a shaky start with a sloppy allusion to the Fisher effect, Keynes is right in observing that the neutrality of money and the independence of the real rate of interest from monetary factors are extreme assumptions. Given that monetary neutrality is consistent with any nominal interest rate, Keynes then tries to show how the orthodox theory pins down the nominal interest rate. And his attempt does not seem successful; he asserts that the money rate of interest can be deduced from the marginal efficiency of some capital asset *A* in terms of money. But that marginal efficiency cannot be deduced without knowledge, or an expectation, of the future value of the asset. Instead of couching his analysis in terms of the current and (expected) future values of the asset, i.e., instead of following Fisher’s 1896 own-rate analysis, Keynes brings up the scale of investment in *A*: “This must . . . be supplemented by another proposition; for it requires that we should already know the scale of investment.” Aside from not knowing what “this” and “it” are referring to, I don’t understand how the scale of investment is relevant to a determination of the marginal efficiency of the capital asset in question.

Now for Keynes’s final proposition:

6 The scale of investment will not reach its equilibrium level until the point is reached at which the elasticity of supply of output as a whole has fallen to zero. (p. 148)

The puzzle only deepens here because proposition 5 is referring to the scale of investment in a particular asset *A* while proposition 6 seems to be referring to the scale of investment in the aggregate. It is neither a necessary nor a sufficient condition for an equilibrium scale of investment in a particular capital asset to obtain that the elasticity of supply of output as a whole be zero. So the connection between propositions 5 and 6 seems tenuous and superficial. Does Keynes mean to say that, according to orthodox theory, the equality of advantage to asset holders between different kinds of assets cannot be achieved unless the elasticity of supply for output as a whole is zero? Keynes then offers a synthetic restatement of orthodox theory.

The equilibrium rate of aggregate investment, corresponding to the level of output for a further increase in which the elasticity of supply is zero, depends on the readiness of the public to save. But this in turn depends on the rate of interest. Thus for each level of the rate of interest we have a given quantity of saving. This quantity of saving determines the scale of investment. The scale of investment settles the marginal efficiency of capital, to which the rate of interest must be equal. Our system is therefore determinate. To each possible value of the rate of interest there corresponds a given volume of saving; and to each possible value of the marginal efficiency of capital there corresponds a given volume of investment. Now the rate of interest and the marginal efficiency of capital must be equal. Thus the position of equilibrium is given by that common value of the rate of interest and of the marginal efficiency of capital at which saving determined by the former is equal to the investment determined by the latter. (Id.)

This restatement of orthodox theory is remarkably disconnected from the six propositions that Keynes has just identified as the bedrock of the orthodox theory of interest. The word “saving” or “save” is not even mentioned in any of Keynes’s six propositions, so the notion that the orthodox theory asserts that the rate of interest adjusts to equalize saving and investment is inconsistent with his own rendering of the orthodox theory. The rhetorical point that Keynes seems to be making in the form of a strictly analytical discussion is that the orthodox theory held that the equilibrium of an economic system occurs at the rate of interest that equalizes savings and investment at a level of output and income consistent with full employment. Where Keynes was misguided was in characterizing the mechanism by which this equilibrium is reached as an adjustment in the nominal rate of interest. A full equilibrium is achieved by way of a *vector* of prices (and expected prices) consistent with equilibrium, the rate of interest being implicit in the intertemporal structure of a price vector. Keynes was working with a simplistic misconception of what the rate of interest actually represents and how it affects economic activity.

In place of propositions 5 and 6, which Keynes dismisses as special factual assumptions, he proposes two alternative propositions:

5* The marginal efficiency of money in terms of itself is . . . a function of its quantity (though not of its quantity alone), just as in the case of capital assets.

6* Aggregate investment may reach its equilibrium rate under proposition (4) above, before the elasticity of supply of output as a whole has fallen to zero. (Id.)

So in substituting 5* for 5, all Keynes did was discard a proposition that few if any economists — certainly not Fisher — upholding the orthodox theory ever would have accepted as a factual assertion. The two paragraphs that Keynes devotes to refuting proposition 5 can be safely ignored at almost zero cost. Turning to proposition 6, Keynes restates it as follows:

A zero elasticity of supply for output as a whole means that an increase of demand in terms of money will lead to no change in output; that is to say, prices will rise in the same proportion as the money demand [i.e., nominal aggregate demand, not the demand to hold money] rises. Inflation will have no effect on output or employment, but only on prices. (pp. 149-50)

So, propositions 5 and 6 turn out to be equivalent assertions that money is neutral. Having devoted two separate propositions to identify the orthodox theory of interest with the idea that money is neutral, Keynes spells out the lessons he draws from his reconstruction of the orthodox theory of the rate of interest.

If I am right, the orthodox theory is wholly inapplicable to such problems as those of unemployment and the trade cycle, or, indeed, to any of the day-to-day problems of ordinary life. Nevertheless it is often in fact applied to such problems. . . .

It leads to considerable difficulties to regard the marginal efficiency of money as wholly different in character from the marginal efficiency of other assets. Equilibrium requires . . . that the prices of different kinds of assets measured in the same unit move until their marginal efficiencies measured in that unit are equal. But if the marginal efficiency of money in terms of itself is always equal to the marginal efficiency of other assets, irrespective of the price of the latter, the whole price system in terms of money becomes indeterminate. (150-52)

Keynes is attacking a strawman here, because, even given the extreme assumptions about the neutrality of money that hardly anyone – and certainly not Fisher – accepted as factual, the equality between the marginal efficiency of money and the marginal efficiency of other assets is an equilibrium condition, not an identity, so the charge of indeterminacy is mistaken, as Keynes himself unwittingly acknowledges thereafter.

It is the elements of elasticity (a) in the desire to hold inactive balances and (b) in the supply of output as a whole, which permits a reasonable measure of stability in prices. If these elasticities are zero there is a necessity for the whole body of prices and wages to respond immediately to every change in the quantity of money. (p. 152)

So Keynes is acknowledging that the whole price system in terms of money in not indeterminate, just excessively volatile. But let’s hear him out.

This assumes a state of affairs very different from that in which we live. For the two elasticities named above are highly characteristic of the real world; and the assumption that both of them are zero assumes away three-quarters of the problems in which we are interested. (Id.)

Undoubtedly true, but neither Fisher nor most other economists who accepted the orthodox theory of the rate of interest believed either that money is always neutral or that we live in a world of perpetually full employment. Nor did Keynes show that the theoretical resources of orthodox theory were insufficient to analyze situations of less than full employment. The most obvious example of such an analysis, of course, is one in which a restrictive monetary policy, by creating an excess demand for money, raises the liquidity premium, causing the marginal efficiency of money to exceed the marginal efficiency of other assets, in which case asset prices must fall to restore the equality between the marginal efficiencies of assets and of money.

In principle, the adjustment might be relatively smooth, but if the fall of asset prices triggers bankruptcies or other forms of financial distress, and if the increase in interest rates affects spending flows, the fall in asset prices and in spending flows may become cumulative causing a general downward spiral in income and output. Such an analysis is entirely compatible with orthodox theory even if the orthodox theory, in its emphasis on equilibrium, seems very far removed from the messy dynamic adjustment associated with a sudden increase in liquidity preference.