## 464. Harrod to J. M. Keynes

, 12 August 1935[a]

[Replies to 463 , answered by 465 ]

as from Sennen Cove Hotel, Sennen Cove, near Land's End, Cornwall

I move to the above address on August 16 and shall be there for about a week.

I am very glad to hear of your simplification of the chapters dealing with user cost, income etc. because, tho very interesting, they did provide a serious stumbling block for the average educated reader or the journalist-economist, which coming at an early stage would limit the appeal of your book.

Prime factor cost [b]

One small point. You say you are adopting my suggestion of "marginal factor cost." [1] But I expect this is a mistake for what I did suggest, viz. prime factor cost [2] (which can of course be further divided into marginal and average). I had in mind almost a hyphen between prime & factor. Prime costs proper, viz. those which can be avoided by not undertaking output in the short period, divide into prime-factor costs, viz. those involved in employing concurrently ultimate factors of production, labour, short loans etc. and user costs, or supplementary factor costs, viz. those involved in using machinery,--generally, the products of prime factors employed in the past. They are prime costs because they can be avoided by not using the machinery, but they are not prime-factor costs because they do not involve the concurrent employment of ultimate factors of production (and arent necessarily paid out to anyone as income).

I have appended another note to my original note on galley 29 . 1

A few words about the equation of saving & investment. I agree that there is no sense in saying that interest equates S to I in sense that if rate of interest were different there would be an excess of S over I or of I over S. But I hold there is sense in saying that interest equates the demand for investment to the supply of saving. And I also hold that the fact that S must be equal to I 2 does not in itself invalidate the proposition that interest is the price which makes them equal. I further hold that if there was some other mechanism for securing constancy of income, which the old classical doctrine assumes 3 (i.e. it implicitly assumes constancy of income and does not envisage the rate of interest as the mechanism for keeping incomes constant) then the classical doctrine that it is the rate of interest which equates the propensity to invest to the propensity to save would not only make sense but also be true. 4

Where anyhow I am sure you must be on unsure ground is when you say that the fact that S must be equal to I invalidates the notion that the rate of interest equates the propensity to save to the propensity to invest. [3] The price of a German lesson does not ensure [c] merely that the amount spent on such lessons is equal to the amount received by teachers: for that would be true if the lessons were compulsorily ordained by the state and the fees were compulsory like unemployment insurance payments. No: the price of the lessons ensures that the amount of lessons which people freely choose to take is equal to the amount that they freely choose to give, and this equation of supply to demand by price is possible, in spite [d] of the fact that the amount of lessons given must necessarily be equal to the amount received.

The essence of your point I feel to be that the cet. par. clause of the supply & demand analysis, which in this case includes the level of income, is invalid. The classical theory \ is invalid but not nonsense.

- 1. Letter
463 ,
[jump to page] .
2. Letter 458 , [jump to page] .

3. Harrod is probably referring to the proofs in Collected Writings (1971-89), vol. XIV, pp. 475-76 (The General Theory, 1936, pp. 183-84). Keynes, however, was surprised by Harrod's use of such an expression: see letter 465 , [jump to page] .

1. "Quasi-rent is the (excess of price over av. cost) ¥ by number of units. I assumed in this note that you were in your text assuming that av. prime cost = marg. prime cost. For if marg. cost > av. prime, no difficulty arises about quasi rent for (apart from question of marg. revenue differing from price) we have a quasi-rent equal to (marg cost - av. cost) ¥ no. of units. Where then marg cost > av prime your absurdity does not occur. But I was taking the case, more difficult from the old fashioned point of view, in which marg prime av. prime. This difficulty is overcome by the doctrine of imperfect competition in which marg. prime = marg. revenue, and in which even if marg prime av. prime, there is still a quasi-rent due to the excess of price over marg. revenue.

So that tho what you say about user cost is true and important, I dont think it necessary to explain the existence of quasi-rent, even in the case favourable to your argument in which av. prime = marg. prime." [Harrod's additional note is pencilled on page 5 of his letter of 31 July, under Keynes's remark here reproduced on [jump to page] ].