393. G. Haberler to Harrod , 2 November 1934 [a]
[Replies to 388 , answered by 394 ]
League of Nations, Geneva #
2 November 1934
My dear Harrod,
Many thanks for your letter of October 29th. I agree that the question of defining saving is an important one, but only because the newfangled definition which Keynes gave of this term  has confused much of the reasoning of his adherents.
(1) Do you really think that it is possible to find out something about the real world by redefining well-established terms? If anybody has such a curiously working brain that he can only make discoveries by inventing unusual definitions, it must be possible to express this newly discovered law also in the ordinary language. The whole matter is therefore only a quibble about words or the best mode of exposition.
(2) Keynes has not made any real discovery. As Hawtrey has shown, a difference between savings and investments is ex definitione (according to Keynes' definitions) a loss, if savings are greater than investments, and a profit, if investments surpass savings. It is wrong to say that differences between S and I cause or lead to losses or profits; they are losses or profits.  I have worked this out in detail and I thought that after Hawtrey's brilliant criticism, there would be no need to publish my paper; but perhaps I shall do so.  Keynes' explanation of the business cycle in terms of differences in savings and investment comes therefore to this: the prosperity phase is characterised or caused by profits; the depression by losses. That we knew before.
As I told you, I am convinced that the Treatise contains many valuable propositions; not because of, but in spite of, and sometimes in contradiction to, the theoretical basis.
(3) It remains to be demonstrated that, under the ordinary definitions of saving and income, differences between saving and investment are possible and that such differences are equivalent to acts of hoarding or dishoarding, of inflation and deflation, to increases and decreases in demand for goods in terms of money per unit of time.
As one example--there are, of course, many others--I take the case you mention in your letter.  Let us start from equilibrium position when S = I.
Now a bank creates £100 new credit in favour of an entrepreneur who wants to spend it on new investment goods. There are two possibilities: (a) you reckon this £100 as the income of the entrepreneur or (b) you do not. (b) seems to me more in harmony with the current usage of the word "income". If you decide for this alternative you have an increase in investment, without a corresponding increase of savings. (During the following periods, after the newly created money has become wholly or in part income in somebody's hand, the situation might be different; but that is another question).
If you adopt alternative (a) (which seems to me contrary to the usual sense of the word "income") and if you assume that the new money is spent for new investment goods, then, it is true, the injection of money does not lead to a discrepancy between S and I. I has gone up, but so has S, since E has risen. But also in this case such a discrepancy can arise. If the new money is hoarded or used to pay a debt instead of being spent for investment goods, savings exceed investments. (I would not say this, because I adopt alternative (b). But if you adopt alternative (a), then you have the discrepancy).
Now take still another case: There is no new money being created. But a part of current savings disappears in the banking system, is hoarded or used to buy securities which somebody is forced to sell. Then obviously S is greater than I. (If the seller of the securities invests the money, the discrepancy between S and I is corrected during the same or, more probably, during the next period).
I must confess that I am unable to make this clearer and cannot see the slightest obscurity in this. I suggest that you show our correspondence to Opie and Marschak.  Perhaps they are able to clear up the matter. But I still hope that this letter may convince you.
I enclose the copy of a letter of mine which deals with a related problem, and I am asking Lindahl to send you the paper to which I refer in my letter . 
R.F. Harrod Esq., Christ Church, Oxford, England [b]
top of page
Return to index of this section
Go to previous page
Go to next page