827. Harrod to J. M. Keynes , 6 September 1938 [a]

[Replies to 821 , continues at 829 , answered by 832 ]

6 September 1938

Dear Maynard

In any warranted position t > s, since . t here stands for the required increment of capital.

Now we have to consider the stability question. Your point is that a departure, say an upward departure from the warranted position, might cause such an increase in s as to cause a relapse. [1]

Warranted rate involves . Upward experiment involves , where t p is actual increase of capital. If t p  < t (my assumption) warranted output is unstable; if t p  > t warranted output is stable. \ we have to consider whether there is any chance of being greater than , for if it is t p will be greater than t. Take the case most favourable to you of the whole of the increment of income being saved.

Suppose we regard the experimental increase of output above warranted level = consequential increase of saving as 100 units. Then instead of considering whether

is less than, equal to, or greater than

we may consider whether

is less than, equal to, or greater than

where is amount of previous saving in absolute units and is warranted increment of output in absolute units. Now can only be greater than if is less than , that is if total saving is less than increment of output. To get such a condition you would have to take a very long period of time in computing even on the assumption so favourable to you that 100% of income is saved--a period so long that I hardly think it relevant to make this qualification.

On your p. 2 para. 1 [2] ( I return it for reference), you have used t instead of t p in the equation and I feel that this is not a slip of the pen in view of what follows. You say that t and s are forced from their normal values. Why? t p is forced away from t; that is all.

I dont of course need or want to assume constancy of s and t in the various phases of the cycle. All I need is to assume that they are determinate at any given point in order to demonstrate instability. If t p  > t there is a depressing force in operation. In due course as the level of output changes, the values of s and t may change and indeed they must change unless output is to recede to zero.

I dont quite follow why you say very violent towards the bottom of the page. [3] If warranted rate is 2%, actual rate 4% and saving , what happens is that the increase of stocks etc. is only 2.5 times value of increment of output instead of 5 times as it normally should be. Entrepreneurs will manage to rub along but they will be under an incentive to expand orders more quickly.

Unused capacity. I have said something about this, tho' no doubt there is much more to be said.

I agree that there may well be no moment in the trade cycle at which t has its normal value. I dont see why it should not approximate to normal in the later stages of revival when unused capacity may be reduced to what we may regard as a "frictional" level.

I will let you have my article again soon. I have been working through your notes.

Price History

With regard to price history [4] --I suggest that I give a summary of the financial position from their side at the next Council meeting and Hawtrey & I say something about the worth-whileness of the undertaking. It will then be up to you to present the matter from the side of the Society. Or would you prefer the 3 of us to reach agreement on the figure to be recommended to the Council beforehand?



  1. 1. Harrod discussed Keynes's criticism also in correspondence with Marschak: see footnote [i] to letter 831 , [jump to page] .

    2. Letter 821 , [jump to page] .

    3. Letter 821 , [jump to page] .

    4. Refers to the Royal Economic Society's financial support towards the publication of W. H. Beveridge, Prices and Wages in England, London: Longmans, 1939: see note 1 to letter 777 R.

    1. a. ALS, two pages on two leaves, in JMK EJ/1/5/318-19. Partly printed in Keynes, Collected Writings, vol. XIV, pp. 335-36.

Welcome page

top of page

Return to index of this section

Go to previous page

Go to next page