The General Theory of Employment, Interest and Money

By John Maynard Keynes

Appendix 2

From The Economic Journal, September 1936


In my General Theory of Employment, Interest and Money, Chapter 8, I made a brief attempt to illustrate the wide range of fluctuations in net investment, basing myself on certain calculations by Mr Colin Clark for Great Britain and by Mr Kuznets for the United States.

In the case of Mr Kuznets' figures I pointed out (Chapter 8) that his allowances for depreciation, etc., included 'no deduction at all in respect of houses and other durable commodities in the hands of individuals'. But the table which immediately followed this did not make it sufficiently clear to the reader that the first line relating to 'gross capital formation' comprised much wider categories of capital goods than the second line relating to entrepreneurs' depreciation, etc.'; and I was myself misled on the next page, where I expressed doubts as to the sufficiency of the latter item in relation to the former (forgetting that the latter related only to a part of the former). The result was that the table as printed considerably under-stated the force of the phenomenon which I was concerned to describe, since a complete calculation in respect of depreciation, etc., covering all the items in the first line of the table, would lead to much larger figures than those given in the second line. Some correspondence with Mr Kuznets now enables me to explain these important figures more fully and clearly, and in the light of later information.

Mr Kuznets divides his aggregate of gross capital formation (as he calls it) for the United States into a number of categories as follows:

(1) Consumers' Durable Commodities

These comprise motor-cars, furniture and house equipment and other more or less durable articles, apart from houses, purchased and owned by those who consume them. Whether or not these items should be included in investment depends (so far as the definition is concerned) on whether the expenditure on them when it is initially made is included in current saving or in current expenditure; and it depends (so far as the practical application is concerned) on whether in subsequent years the owners feel under a motive to make provision for current depreciation out of their incomes even when they are not replacing or renewing them. Doubtless it is not possible to draw a hard-and-fast line. But it is probable that few individuals feel it necessary in such cases to make a financial provision for depreciation apart from actual repairs and renewals. This, in combination with the difficulty of obtaining proper statistics and of drawing a clear line, makes it preferable, I think, to exclude such equipment from investment and to include it in consumption-expenditure in the year in which it is incurred. This is in accordance with the definition of consumption given in my General Theory, p. 54.

I shall, therefore, exclude this category from the final calculation; though I hope to deal with the problem more thoroughly at a later time. Nevertheless it may be interesting to quote Mr Kuznets' estimates, which are of substantial magnitude:

  (Millions of dollars.) 
  1925 1926 1927  1928 1929  1930 1931  1932 1933 
Consumer's durable commodities










The above figure for 1929 includes 3,400 million dollars for motor-cars, whilst the depreciation in respect of the same item for that year is estimated at 2,500 million dollars.

(2) Residential Construction

This is an important and highly fluctuating item which should undoubtedly be included in investment, and not in consumption expenditure, since houses are usually regarded as purchased out of savings and not out of income, and are often owned by others than the occupiers. In the Bulletin from which these figures are taken Mr Kuznets gives no estimate for the annual rate of depreciation, etc. More recently, however, his colleague, Mr Solomon Fabricant, has published such estimates, which I have used in the following table:

  (Millions of dollars.) 
  1925 1926 1927  1928 1929  1930 1931  1932 1933 
Residential construction








Depreciation* 1,554 1,676  1,754 1,842 1,911  1,901 1,698 1,460  1,567
Net investment 1,496 1,289  1,102 1,253 216  -679 -798 -1,149  -1,291

* These figures are calculated in terms of current (reproduction) costs. Mr Fabricant has also provided estimates in terms of original cost, which for the years prior to 1932 are considerably lower.

(3) Business Fixed Capital

Mr Kuznets here distinguishes expenditure on new producers' durable goods and business construction from the net change in 'business inventories,' i.e. in working and liquid capital; and we shall, therefore, deal with the latter under a separate heading.

The amount of the deduction to obtain net investment in respect of parts, repairs and servicing, and repairs and maintenance of business construction as distinct from depreciation and depletion, which is not made good, depends, of course, on whether the former have been included in gross investment. Mr Kuznets gives a partial estimate for the former but the figures given below exclude these items both from gross and from net investment. But whilst the result of deducting both the repairs item and the depreciation item probably corresponds fairly closely to my net investment, the two deductions taken separately do not closely correspond to my deductions for user cost and supplementary cost; so that it is not possible to calculate from Mr Kuznets' data a figure corresponding to my (gross) investment.

The following table gives in the first line 'the formation of gross capital destined for business use, exclusive of parts, repairs and servicing, and repairs and maintenance of business construction, and excluding changes in business inventories'; and in the second line the estimated 'depreciation and depletion' on the same items:

  (Millions of dollars.) 
  1925 1926 1927  1928 1929  1930 1931  1932 1933 
Gross business capital formation
(as above) 










Depreciation and depletion*








Net investment 3,385 3,546  3,243 3,572 4,357  2,624 -221 -1,887  -2,077

* These figures are not taken from Mr Kuznets' memoranda, hut from Mr Fabricant's later and revised estimates. As before they are in terms of current (replacement) cost. In terms of original cost they are appreciably lower prior to 1931 and higher subsequently.

(4) Business Inventories

For the financial gains or losses arising out of this item there appear to be fairly adequate statistics in the United States, though not in this country. Mr Kuznets' figures are as follows:

  (Millions of dollars.) 
  1925 1926 1927  1928 1929  1930 1931  1932 1933 
Net gain or loss in business inventories










This table covers not only manufacturers' stocks but also stocks of farmers, mines, traders, government agencies, etc. From 1929 onwards the figures given in Mr Kuznets' memorandum of 1934 proved to require correction. Those given above are provisional and approximate estimates, pending the publication of revised figures by the National Bureau.

(5) Public Construction and Borrowing

The relevant figure in this context is not so much the gross (or net) expenditure on construction, as the amount of expenditure met out of a net increase in borrowing. That is to say in the case of public authorities and the like, their net investment may be best regarded as being measured by the net increase in their borrowing. In so far as their expenditures are met by compulsory transfer from the current income of the public, they have no correlative in private saving; whilst public saving, if we were to find a satisfactory definition for this concept, would be subject to quite different psychological influences from private saving. I have touched on the problem in my General Theory, footnote. I propose, therefore, to insert in place of the figures of public construction the 'loan expenditure' of public bodies.

Mr Kuznets has very kindly supplied me with figures for the net changes in the amount of public debt (Federal, State and local) outstanding in the United States, which, except for minor changes in the Government's cash balances, represent the amount of public expenditure not covered by taxes and other revenues. This is given below in parallel with his estimates of the amount of construction by public authorities. The interesting result emerges that up to 1928 there was a net reduction in the public debt in spite of a large expenditure on public construction, and that even up to 1931 some part of public construction was met out of revenue. The excess of borrowing over construction in 1932 and 1933 represents, of course, various measures of public relief.

  (Millions of dollars.) 
  1925 1926 1927  1928 1929  1930 1931  1932 1933 
Public construction








Net change in outstanding public debt**










* See Mr Kuznets' Bulletin, Table II, line 22, brought up to date on the basis of more recent data.

** See col. 9 of the table given in the appendix below.

(6) Foreign Investment

Finally, we have the net change in claims countries, estimated by Mr Kuznets as follows:

(Millions of dollars.) 
1925 1926  1927 1928  1929 1930  1931 1932  1933
428 44 606  957 312 371  326 40 293 

(7) Aggregate Net Investment

We are now in a position to combine the above items into a single aggregate. This total is not quite comprehensive, since it excludes construction by semi-public agencies, and a small amount unallocable construction. But Mr Kuznets is of the opinion that both omissions are quite minor in character and could not much affect the movements of net investment in the table which now follows.

  (Millions of dollars.) 
  1925 1926 1927  1928 1929  1930 1931  1932 1933 
Residential construction








Business fixed capital








Business inventories








Net loan expenditures by public authorities 










Foreign investment








Aggregate net investment









It is evident that this table is of first-class importance for the interpretation of business fluctuations in the United States. In matters of detail the following points stand out:

(a)  The arrears of residential construction at the end of 1933 must have been enormous. For there had been no net investment in this field since 1925. This does not mean, of course, that the actual state of housing was so bad as this. Some gross investment in housing continued throughout, and the gradual deterioration in the state of accommodation, through obsolescence and decay not made good, does not impair forthwith to an equal extent the actual accommodation available for the time being.

(b)  The part played by fluctuations in business inventories is very marked, especially in accentuating the depression at the bottom of the slump. The increase in inventories in 1929 was probably for the most part designed to meet demand which did not fully materialise; whilst the small further increase in 1930 represented accumulations of unsold stocks. In 1932 and 1933, manufacturers met current demand to an extraordinary extent out of stocks, so that effective demand fell largely behind actual consumption. But this, fortunately, is a state of affairs which could not continue indefinitely. A further depletion of stocks on this scale could not possibly take place, since the stocks were no longer there. A level of business inventories so low as that which existed in the United States at the end of 1933 was an almost certain herald of some measure of recovery. In general an aggregate of net investment which is based on an increase in business inventories beyond normal is clearly precarious; and it is easy to see in retrospect that a large growth of inventories in 1929, coupled with a decline in residential construction, was ominous. The figures for 1934, 1935, and 1936 will be most interesting when we have them. One would expect that the recovery of the two former years has been based on a return of inventories to normal and on public loan expenditure, but that by 1936 durable investment was beginning to supplant inventories in making up the total. It is on the continued steadiness of the first two items of the above table at figures not less than those of 1925 to 1928 that the maintenance of prosperity must depend; and it is for this reason that a low long-term rate of interest is so vitally important.

(c)  The manner in which the changes in public loan expenditure came in to moderate the fluctuations, which would have occurred otherwise, is very apparent. The manner in which from 1931 Federal borrowing took the place of State and local borrowing, as shown in the Appendix below, is striking. From 30 June 1924, to 30 June 1930, Federal loans outstanding fell from 21 to 15 billions, whilst in the same period State and local loans rose from 10 to 16 billions, the total remaining unchanged; whereas from 30 June 1930 to 30 June 1935, Federal loans rose from 15 to 26 billions and the others from 16 only to 17 billions. The appendix, which gives the figures of public borrowing up to 30 June 1935, shows¾contrary, perhaps, to the general impression¾that public borrowing was at its height in 1931, and that in 1934-35 it was but little more than in 1929-30.

(d)  When comparable figures of income are available, we shall be able to make some computations as to the value of the Multiplier in the conditions of the United States, though there are many statistical difficulties still to overcome. If, however, as a very crude, preliminary test we take the Dept. of Commerce estimates of income (uncorrected for price changes), we find that during the large movements of the years from 1929 to 1932 the changes in money-incomes were from three to five times the changes in net investment shown above. In 1933 incomes and investment both increased slightly, but the movements were too narrow to allow the ratio of the one to the other to be calculated within a reasonable margin of error.


A Note to Appendix 2


  (Millions of dollars.) 
  Total outstanding issues  Net outstanding issues 

(30 June)





State, county, city, etc.








State, county, city, etc.





Net change 


Average for calendar year


1924 20,982  11,633 32,615 20,627  9,921 30,548   -   -
1925 20,211  12,830 33,041 19,737  10,975 30,712 +164  -43
1926 19,384  13,664 33,048 18,790  11,672 30,462 -250  -280
1927 18,251  14,735 32,986 17,542  12,610 30,152 -310  -244
1928 17,318  15,699 33,017 16,522  13,452 29,974 -178  -10
1929 16,639  16,760 33,399 15,773  14,358 30,131 +157  +441
1930 15,922  17,985 33,907 14,969  15,887 30,856 +725  +1,712
1931 16,520  19,188 35,708 16,098  17,457 33,555 +2,699  +2,822
1932 19,161  19,635 35,796 18,673  17,828 36,501 +2,946  +2,565
1933 22,158  19,107 41,265 21,613  17,072 36,685 +2,184  +2,796
1934 26,480  18,942 45,422 25,323  16,771 42,094 +3,409  +2,173
1935 27,645  19,277 46,922 26,137  16,895 43,032 +938    -

(Source: Report of the Secretary of the Treasury for year ended 30 June 1935, p. 424.)

Total outstanding issues exclude a small volume of matured and non-interest bearing obligations (see ibid., p. 379).

Net outstanding issues are equal to total outstanding issues less those held in U.S. Government trust funds, or owned by U.S. Government or by governmental agencies and held in sinking funds.

The table above does not include the contingent debt of the Federal Government, i.e. obligations guaranteed by the United States. These, comprising largely debt issues of the Federal Farm Mortgage Corporation, Home Owners Loan Corporation and the Reconstruction Finance Corporation, were as follows:

Date Millions of dollars 
30 June 1934 691
31 December 1934 3,079
30 June 1935 4,151
31 December 1935 4,525

(See Cost of Government in the United States, by the National Industrial Conference Board, pub. no. 223, New York, 1936, Table 26, p. 68.)