## 472. J. M. Keynes to Harrod

, 14 September 1935[a]

[Replies to 471 , answered by 473 ]

14 September 1935*
** [b]**
*

I absolve you completely of misunderstanding my theory. It could not be stated better than on the first page of your letter. Also I think the construction in the note, of which I attach a copy to remind you of it, is both correct and very useful as a help to exposition, and I should like to appropriate it. But what you say seems to me to fail either as a statement of what the classical theory says or as a justification of its subject to special assumptions. I make a great distinction between the classical theory of employment, which does make perfect sense and works [c] all right on certain special assumptions, and the classical theory of the rate of interest which makes no sense on any assumptions whatever. I have now thought about the matter enough (having reached these chapters in my revision) to express myself more clearly.

Let me start out from your note which I understand as follows:--[fig. 1]

r the rate of
interest is measured horizontally, the quantity of investment (or
saving) vertically.
is the schedule of the marginal efficiency of capital showing how
much investment will take place at each rate of interest. The curves
Y* 1* Y* 2* Y* 3 * show the relation between the
propensity to save and the rate of interest, assuming successively
that income is Y* 1* , Y* 2* , Y* 3* . Now you
argue that, if income is Y* 1* , the rate of interest is given
by the intersection of the Y* 1 * saving curve with
, the schedule of the marginal efficiency of capital; and that this
is the classical theory of the rate of interest.

I will return to this in a moment. But, first of all, I should like to point out that what the diagram does show (and most elegantly) is the relation between the rate of interest and the level of income, given the schedule of the marginal efficiency of capital and the propensity to consume.

The reason that
this is not a theory of the rate of interest in any circumstances is
that the assumption of a fixed
income Y* 1
* is too all-embracing and does not allow for those things to
vary which the Classical theory assumes to vary. For, if either the
propensity to save or the marginal efficiency curve changes, the
level of income changes; so that the assumption of a given level of
income involves us in assuming that neither of the classical theory's
own chosen variables is capable of change,--unless there are changes
in certain other factors which happen by a miracle to be such as to
leave income unchanged (and in this case it is the nature of the
miracle which determines the rate of interest).

Take the case
most commonly contemplated by the classical theory in which there is
a change in the schedule of the marginal efficiency of capital so
that the
curve is replaced by the curve
. The classical theory assumes that the rate of interest will then be
given by the intersection of
with the Y* 1 * curve. But, of course, this is not so; for if
the marginal efficiency curve shifts, income cet. par. will change,
so that, if we assume income fixed, we must also assume that the
marginal efficiency curve is fixed. The only circumstance in which a
change in the marginal efficiency curve can leave income unchanged
(assuming that the propensity to save does not change), is in the
event of liquidity-preference happening to change by just the amount
necessary to balance the change in the marginal efficiency curve. But
in this event it would still be the change in liquidity preference
which was determining the rate of interest.

The question whether or not the real wage is equal to the marginal disutility of labour does depend on whether or not there is full employment. But no similar reconciliation is possible between the theory that the rate of interest depends on the intersection of the propensity to consume with the schedule of the marginal efficiency of capital, and the theory that it depends on liquidity preference. The fault of the classical theory lies, not in its limiting its terrain by assuming constant income, but in its failing to see that, if either of its own variables (namely propensity to save and schedule of marginal efficiency of capital) change, income must also cet. par. change; so that its tool breaks in its hand and it doesn't know and can't tell us what will happen to the rate of interest, when either of its own variables changes

I say, therefore, that it is nonsense to assume at the same time that income is constant and that the propensity to save and the schedule of the marginal efficiency of capital are variable.

Look through again my various quotations from classical economists and tell me which of them is not nonsense.

I shall be here from Sept. 22nd at least up to the end of the month. Will you come down for a night or two after you are back from Ireland? 1 It is probable that by then I shall have finished re-writing the chapters dealing with rate of interest.

You will see that I have, in a sense, shifted my ground in the above. I have gained a great deal from your hard knocks, and would like some more. Since I wrote the above I have finished my redraft of the relevant chapter and it should be back from the printer if you can come as above.

R. F. Harrod Esq., Clandeboye, near Belfast, Ireland.

- a. TLS with autograph correction, diagram, and addition, five pages on five sheets. CcI without autograph emendations in JMK GTE/1/386-90, original in HP II-56 (the latter folder also encloses a typed transcription of Harrod's "Note to page 4" of his previous letter--here at [jump to page] ). Printed in Keynes, Collected Writings, vol. XIII, pp. 557-59. Reproduced by kind permission of the Provost and Scholars, King's College, Cambridge.
b. The letter was originally dated 10 September (in type), and later corrected in Keynes's hand into 14 September.