===== PAGE 1 =====

1938]

Commodity Stocks in the Trade Cycle
By L. M. Lacemany and F. Snapper

I

In this paper we endeavour to make use of the statistics
of commodity stocks in order to throw some light upon the
Trade Cycle and the issues arising from it.

Our main problem is: Do commodity stocks move in
positive or inverse correlation with the Cycle ? Important
issues as to the momentum of the ‘cumulative process’
hinge upon the answer to this question. For, in the case
of positive correlation, investment in commodity stocks
would be an important accelerating force in the mechanism
of booms and depressions, tending to make any increase
in investment activity somewhere in the economic system
the impelling force of a cumulative process. By analogy,
in the case of inverse correlation changes in the size of stocks
would be a retarding force.

It would, of course, be most desirable to be able to make
use of statistics of the stocks of finished as well as of unfinished
goods. For then it might be possible to say something
about the relative size of stocks at different stages of produc-
tion in different phases of the Cycle, a very important
problem to all those who, unyielding to the attractions of
‘ macrodynamics ”’, refuse to see in crises simply fluctuations
in total investment. Unfortunately, we have at our disposal
statistics of unfinished commodities only.* There is, however,
reason to believe that the stocks of finished products move in
positive correlation with the cycle, because they are kept by
producers and merchants as a constant percentage of turnover.

We thus shall have to confine ourselves to the study of
raw material stocks and try to find out what light they
throw upon the Trade Cycle. So far, Mr. Keynes has been
the only one to formulate a precise and logical theory of
the cyclical fluctuations in commodity stocks.? According

LThereare a few statistics of stocks of finished products, but they are not very satisfactory, e.g. :
U.S.4., Survey of Current Business, Annual Supplement, 1936, p. 10.
S. Kuznets, National Income and Capital Formation, New York, 1937, p. 40, table 10,
No. 2¢ “ changes in business inventories ”’, and p. 120 (for steel sheets).
Jan Tinbergen in De Nederlandsche Conjunctuur, March, 1933, pp. 11-20.
2 J. M. Keynes, 4 Treatise on Money, Vol. II, ch. 29, ““ Liquid Capital”. Also General
Theory of Employment, Interest and Money, pp. 318-19.

435


===== PAGE 2 =====

436 ECONOMICA [NOVEMBER
to him, “ surplus stocks ”’ must be cleared before recovery
is possible, and therefore the depletion of stocks during the
depression is a subsidiary force of disinvestment. We shall
have to examine this thesis in the light of the statistical
facts. On the other hand, if it can be shown that stocks
as a rule reach their lowest level immediately before the
outbreak of the crisis, this may conceivably give some indica-
tion of the causes of the latter. It would purport to show
that the point at which “surplus stocks” are exhausted is
not the lower, but the upper turning point of the cycle.

II

We present below two tables of statistics of movements
in stocks of certain raw materials and foodstuffs, the first
referring to the pre-war period and the second to the years

1919-37.

Tare 1

Pig Iron Pig Iron Copper Tin | Wheat | Cotton

End | North of | End | North of in in 0000
of England of | England | Europe| U.S.A. 000,000| bales
Year | ooo tons | Year| ooo tons ooo tons ooo tons | bushels | 1 Aug.
(2) 2) (1) 3) @]| © (s)

1869 116 1885 517 — 16 — 134
1870 118 1886 652 47 14 — 144
1871 68 1887 637 56 19 — 147
1872 42 1888 473 35 19 — 129
1873 80 1889 262 96 17 — 132
1874 90 18go 256 95 14 126 108
1875 74 1891 263 62 16 175 174
1876 182 1892 164 53 19 194 282
1877 271 1893 163 52 20 206 226
1878 337 1894 216 43 30 200 213
1879 283 1895 306 51 36 186 320
1880 331 1896 163 43 40 151 193
1881 378 1897 97 31 35 126 192
1882 266 1898 138 27 24 113 324
1883 253 1899 71 22 21 160 400
1884 302 1900 44 28 21 163 246
1901 139 22 26 169 267

1902 121 16 17 142 267

1903 97 I3 17 133 292

1904 199 16 17 146 277

1905 707 13 16 140 504

1906 537 17 16 151 418

1907 89 20 15 127 569

1908 132 57 57 23 132 458

1909 385 109 66 23 96 568

1910 519 84 55 20 132 473

1911 526 57 41 19 185 484

1912 241 40 48 14 — 681

1913 230 21 41 16 — 636



===== PAGE 3 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 437

TasLe II
Copper Tin Zinc Sugar Wheat | Cotton | Rubber

End in in 000 0000

of Europe U.S.A. 000 000 tons ooo,000| bales 000
Year 000 tons tons tons 1 Oct. bushels | 1 Aug. | tons

(1) DW ®e 6 © | 6 [H®06

1919 18 546 20 01 — 347 675 —
1920 20 587 18 81 1338 317 634 281
1921 23 427 24 72 2039 238 968 296
1922 33 344 26 20 868 285 561 358
1923 37 409 21 34 791 339 338 306
1924 51 374 25 42 542 363 276 233
1925 65 322 18 9 2500 323 327 234
1926 48 358 16 20 1919 378 550 311
1927 14 330 16 38 1780 400 784 334
1928 10 315 25 42 1850 565 521 301
1929 12 439 28 75 4047 583 452 382
1930 12 585 42 144 5522 583 619 505
1931 34 722 51 130 7022 608 897 644
1932 48 747 45 125 7099 598 | 1326 641
1933 39 673 23 106 6264 539 1181 668
1934 69 567 13 120 5396 §16 | 1070 724
1935 104 496 12 84 4128 487 909 637
1936 63 377 20 45 2983 318 696 473
1937 42 505 20 79 3102 346 624 496

(1) Year Books of the American Bureau of Metal Statistics.

(2) Monthly Trade Supplement of the Economist.

(3) Special Memoranda of the London and Cambridge Economic Service, Nos. 32 and 43.
(4) World Tin Statistics, 1938.

(3) Cotton Year Book of the New York Cotton Exchange, 1937.

(6) Broomball’s Corn Trade Year Books.

(7) Statistische Zusammenstellungen der Metallgesellschafi.

(8) London and Cambridge Economic Service, May Bulletin, 1938.

(9) Special Memorandum No. 32, table on p. 19.

IIT

Economists before the war assumed that Sauerbeck’s
Index Numbers were a fairly good barometer of the General
Trade Cycle. Now there is reason to believe that for the
period before the war this contention holds true, although
we prefer a production index which is derived from the
English unemployment figures. We do not need to explain
why certainly after the war the American production index
is greatly preferable.

The movements of agricultural raw materials require,
however, a separate explanation. We observe that during
the period 1873-1913, whereas the general trade cycle
reaches its peak in 1881, 1891, 1900, 1907 and 1913, the
index numbers of prices of foodstuffs behave somewhat


===== PAGE 4 =====

438 | ECONOMICA [NovEMBER
Chart 1

1870 1875 1880 1885 1890 1895 1900 1905 1810

pa ees

7] STOCKS of PIG IRON oT
7001 °/, of 1899 Level (orth of England) # ha
S00} wos
400 ——

300} A —-
200} i.
0 n
+00} SS of pac we]

y/ of 1899 Level (us 4 ena cuROPE)
30014 /° \_ —
200} MNL \ i.
100}= “ne —-

SU Ig
o e®e rn ~~ ——-
#7 Te STOCKS of TIN
00 yA 1899 Level ee IPP OL RET Ol yf
lo!
5° 1875 1880 1885 1890 i835 1800 1805 210
7 NS Sm 7
Jo © Treo ~ EMPLOYMENT
SOF a Numbers =
er]
85h ed

oo The peaks of this cycle are in the years 1877,
1891, 1900 and 1912. At first sight it may seem that agri-
cultural production as a whole has a cycle of its own, which
consists of two components: the general industrial cycle
and changes in natural conditions. By natural conditions
we mean all those atmospheric and climatic factors which
influence the size of crops.! The absence of a production
function in the strict sense, i.e., the fact that in agriculture

1 Some writers have attributed these changes to sun spots. They believe that there is a
sun spot cycle of about 11 years, which causes a cyclical fluctuation in the size of the crops.
Among the outstanding writers who have taken this view are W. Stanley Jevons and Mr. S. de
Wolff. The latter in his book, Het Economisch Getij, states that there is a longer cycle of 45
years too. Mr. D. H. Robertson (4 Study of Industrial Fluctuations) mentions this theory
without. committing himself.


===== PAGE 5 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 439

output quantities are not uniquely correlated with input
quantities, makes the supply of agricultural produce a
relatively independent variable at least in the short period,
when acreage and methods of cultivation are given. Hence,
if there are cyclical fluctuations in the size of output per
acre this would be a sufficient condition for a separate
agricultural cycle. So much for the supply side.

On the demand side, of course, the agricultural cycle is
linked up with the general trade cycle. Industrial demand
for agricultural raw materials depends, of course, on the
state of trade. The demand for agricultural produce for
consumption is governed by the level of income and employ-
ment unless the income elasticity of demand is very low.
It 1s clear, e.g., that fluctuations in wheat (demand for
which is very inelastic) will be entirely governed by the
acreage and the output per acre, whereas demand for cotton
will depend on factors partly germane to the industrial
situation.

We may therefore conclude that the whole conception of
an agricultural trade cycle is somewhat doubtful. However,
there are very good reasons, as we have seen, to expect
production and prices to deviate from the general trade
cycle each in its own way. It would be useless, therefore,
to correlate stocks of such commodities to the general trade
cycle. We correlated the movements of cotton stocks
with the price of cotton in the period 1885-1913, and we
found an inverse correlation, after eliminating trend, of
— 82. Some economists will find this result very satisfactory,
but we beg to differ. We prefer the simple method of
graphical illustration to the dubious niceties of correlation
analysis.

For the period 1928-1936 a recent inquiry has shown how
the prices of different commodities like wheat, cotton, coffee,
tea, rubber, silk and tin move inversely with their respective
stocks.? Besides, a diagram is presented in which the total
stocks of the commodities mentioned move inversely with
their average price level. To this we added the American
index of industrial production (chart III). It is, of course,
well known that after the war the Trade Cycle was more
intimately connected with investment in the production of
raw materials, particularly in overseas countries.

1 « Eenige Gegevens betreffende Grondstoffenmarkt” in De Nederlandsche Conjunctuur,

February, 1937, pp. 14-20.
B


===== PAGE 6 =====

440 . ECONOMICA [NOVEMBER

YI

CT, &
— 3 7 5 8 @
— »eE LE —1a
°C =
SS ERE vO a I
23343 13
-

BT EO 5 po
— 3% 38 2
[72] J—
2S 2® .
J 1 a
| | in}

~ — | | | ] | | i I |

a 4
jeu] [id Fo ly 3 Fo.)
l= a3 ot 2
2 Qg | een E
a) [@] [ed reese cones”
£8 5 > ks
0 o& wn | . uv Q
> * a ~~. x = T . 5 —15
x 9 h 9
N 2 5 : ®
T < i
hd n
Ow © po ® $
oF g 2 : &
g g / s :
< 2 / < .
‘ th PN ‘
e a e 0
& _-— re} :) IN



===== PAGE 7 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 441

Sugar shows a peak in 1925 (Table II, Col. 6). The
explanation is given in a Special Memorandum of the London
and Cambridge Economic Service.! The cause must be
sought in the decision to increase the acreage of plantations
in Cuba in 1923. We cannot deal here with all the very
interesting questions with which that Memorandum deals.
But in our opinion the inverted movement of sugar prices
with the stocks in those years is perfectly clear.

The steady rise of stocks of cotton (Table II, Col. 8) from
1924-1927 is accompanied by a steady decline in price.? The
fall in stocks in 1928 was due to a rise in price. The same
holds true for the years 1921-1923, whereas the fall in price
from 1920-1921 is again accompanied by a rise in stocks.

Stocks of wheat, as Mr. Keynes has told us,? show maxima
in 1896, 1899, 1907, 1923 and, we can add, 1933. But when
we look at the prices we shall find them in these years at
their minimum level, with the difference that with regard
to the years 1907 and 1933 the price reaches its minimum
one year earlier. When we examine those years which
according to the Stanford University Wheat Studies (the
statistics to which Mr. Keynes refers) are minimum years
for stocks, they appear to be maximum years for prices ;
we have to add that since then Stanford University has
compiled world stocks for the period 1922-1934.¢ Moreover,
Professor Tinbergen gives also Stocks for wheat for the
period 1890-1911 which do not differ much from the Stanford
Statistics.®

Finally it should be noticed that the stocks of wheat
increase during the year 1937 (Table II, last line) ; the stocks
of rubber and sugar do the same, and when we compare
the year end stocks of cotton, we also see a rise®: from
6 million bales at the end of 1936 to 8.4 million bales at
the end of 1937.

The stocks of pig iron, copper and tin before the war
show the inverse correlation with the Cycle rather well
as 1s seen from our chart I.

Pig iron stocks after 1896 exclude makers’ stocks” (Table I,

LJ. W. I. Rowe. Special Memorandum No. 31, Sugar, London and Cambridge Economic
Service.

2 Cotton Year Book of the New York Cotton Exchange, 1937.

3 Treatise on Money, Vol. II, p. 124n.

4 IW heat Studies, Vol. IV, p. 180, and Vol. X| p. 134.

5 Jan Tinbergen in De Nederlandsche Conjunctuur, March, 1933, pp. 11-20.

6 London and Cambridge Economic Service, May Bulletin, 1938, p. 207.

“The Economist, Monthly Trade Supplement, December, 1896.


===== PAGE 8 =====

442 ECONOMICA [NOVEMBER

Col. 4). The sudden fall in 1878-9 (Col. 2)—otherwise a time
of intense depression—was due to a coal strike in Durham,
but it was slight owing to the attempts of ironmasters to raise
prices.! The enormous increase in 19o5 was due to heavy
speculation in “Middlesboro’ N. 37.2 As will become clearer
later on, it was a typical instance of the case where opinions
differ widely and dealers were more “ bullish ” than either
ironmasters or iron consumers.

Copper had a special overproduction crisis (Table I, Col. 5)
in 1889.2 Afterwards stocks have a falling trend for about 17
years owing to the steady expansion of the electrical industries.

After the war (Table II) the stocks of copper, tin and zinc
show great irregularities, but by 1929 we find them all having
resumed the inverse movement to the Cycle. Copper and
zinc stocks increased in 1937, tin (where production was
most rigidly controlled) already in 1936.

It remains for us to discuss the influence of monopolistic
restrictions on the size of stocks, taking the case of copper,
stocks of which after the war show a very close inverse
correlation with business activity, with a lead of one year.
The period 1923-29 appears to be very suitable for this
purpose. After 1929 the phenomenon is disturbed by the
Great Depression.

A combine of copper producers was formed in 1926.
Until March, 1929, although stocks remained very small
throughout this period, this does not seem to have had
much influence on the size of stocks. But then a speculative
boom broke out. The combine, more interested in high prices
than in the stabilisation of production, at first allowed stocks
to reach a minimum level. After a month the boom collapsed,
but the high prices had by then induced producers to increase
output, and because of the American anti-trust laws the com-
bine was unable to prevent this. They tried in vain to keep
the price high by accumulating stocks. After 1930, however,
the accumulation of stocks was no longer deliberate.

Between 1923 and 1929 there was a cartel which tried
to restrict the output of spelter. Its efforts, however, were
not very successful. Production rose from 960,000 tons
in 1923 to 1,440,000 tons in 1929. Moreover, as far as we
know, it took no measures to influence the size of
stocks.

1 The Economist, 1879, pp. 421 and 559.
2 Ibid., 1908, p. 1072.
3 Ibid., September 12th, 1908, p. 481.


===== PAGE 9 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 443

Iv

We shall now have to examine the distribution of raw
material stocks between different classes of holders. For
this purpose we shall divide them into :

Producers (of raw materials) ;

Dealers (in raw materials) ;

Manufacturers (of finished commodities in so far as
they are buyers of raw materials).

We would expect that because of the costs involved in the
storing of commodities, everybody wants to keep his stocks
as low as possible. Moreover, we know that entrepreneurs
as a rule keep their stocks in a certain relation to their
turn-over. We therefore have to find an explanation why
the stocks of raw materials do not obey to that rule. The
obvious reason is that in the case of sudden and unforeseen
changes in demand, agricultural and mineral production
can only be readjusted with a certain time-lag.

Sometimes another reason is mentioned : Speculation ;
about this point we shall have to say more in a later section
of this paper.

If our explanation of the inverted movement of the
stocks of raw materials is correct, we shall expect producers
to bear the burden of these surplus stocks, for the dealers
and manufacturers, who wish to maintain their proportion
between output and stocks, are not responsible for the
production of raw materials.

In some cases, however, not the producers but the dealers
and manufacturers hold these surplus stocks. Here the
explanation has to be sought in the relative ease with which
in different industries different classes of entrepreneurs can
obtain credit. In some fields of production the producers
have a relatively large reserve of capital available, or they
may have an easy access to the credit market. In other
fields of production producers do not enjoy these facilities
and have always to sell to dealers. It is also conceivable
that the dealers may be unable to provide enough capital
and that they may have to sell a part of those stocks to
the manufacturers. We thus can imagine the enormous
extra fall in price when at a certain moment during the
slump the capital reserves of a certain field of production
are becoming insufficient to finance the growing surplus
stocks,


===== PAGE 10 =====

444 ECONOMICA ~ [NoveEMBER

Thus the stocks of sugar in Java were negligible before
1929. ‘‘ For the crop was always sold forward in its entirety
before the grinding season began”, as we read in Special
Memorandum No. 45 of the London and Cambridge Economic
Service.! After 1929 the stocks in Java (and in Cuba)
and in the hands of dealers move very strongly in an inverse
relation to the trade cycle.

In the same Memorandum is an interesting table concerning
the division of rubber stocks during that period.? We
combined the stocks of U.K. public warehouses with the
stocks of U.S. dealers and manufacturers. Then we subtracted
these combined figures from the total amount.

STOCKS OF RUBBER AT THE END OF THE YEAR: (coo Tons).

On Estates
Year U.S. and U.K. Total Difference in Malaya
1928 88 301 213 35
1929 192 382 190 28
1930 320 505 185 26
1931 449 644 195 20
1932 472 641 169 22
1933 457 668 214 21
1934. 496 724 228 12
1935 476 634 158 23
1936 301 473 172 26

The conclusion 1s that in the depression the manufacturers
and dealers in these commodities bear the brunt of the
burden. It is interesting to observe how the stocks in the
hands of producers (on estates in Malaya) move in opposite
direction to the total of stocks. It should further be noticed
how during the period total stocks of rubber moved inversely
towards the Trade Cycle.

The division of the stocks of copper, on the other hand,
presents entirely different features. Although there are no
statistics of copper in the hands of manufacturers we know
the stocks in the hands of producers (American smelters
and refiners) and (European) dealers.

The stocks of copper are mostly in the hands of producers,
but there is a tendency for dealers’ stocks to move in the
same direction. The reason obviously is that producers
have to carry the bulk of these stocks in order to maintain
prices. They are able to do it because of the credit facilities

~¥ Stocks of Staple Coinmodities, by J. W. F. Rowe and others ; November, 1937, p. 24.
2 Ibid., p. 30, Table I. :


===== PAGE 11 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 445

they enjoy; they are supported by the Morgan group.!
The dealers, as a rule, have some financial reserves and
their stocks exercise a buffer function as regards changes
in demand. At any rate, we are entitled to conclude that
the stocks will be held in the strongest hands.

Vv

The main conclusion emerging from the statistics we
have presented appears to be that our stocks are inversely
correlated with the cycle. As a rule, they reach their lowest
level very shortly before the outbreak of the crisis, while
their peak level is to be found towards the end of the depres-
sion. If this reading of the facts is correct, it is difficult
to accept Mr. Keynes’ theory according to which surplus
stocks must be exhausted before recovery can start. On
the contrary, the conclusion that seems to suggest itself
is that raw material stocks must have reached a certain
size if they are to support a lasting recovery. This shows
the highly artificial character of the division of stocks into
“ working capital” and ‘liquid capital” according to
“the normal requirements ” of production and illustrates
the ambiguity of the concept of “ surplus stocks”; for
stocks that may have been surplus with regard to the
level of activity at the trough of the depression may
be insufficient to sustain a major recovery. It is here that
the buffer function of stocks?—at least of those goods the
supply of which can only be adjusted with a time-lag—
comes out most clearly: without ample reserves of raw
materials, recovery may soon be checked by all sorts of
““ bottlenecks ”.

On the other hand, our statistics seem to show that it is
at least not impossible that prosperity should come to an
end owing to the scarcity of certain factors of production.
It is no doubt difficult to generalise from the material
presented, because different commodities show different
“leads ” against the Trade Cycle, and it is, of course, by
no means necessary that all the crises brought about by
scarcity should be brought about by scarcity of the same
factor. The coefficients of production being fixed in the
short run, scarcity of ome factor may suffice to stop all

1 London and Cambridge Economic Service, Special Memorandum No. 32, Stocks of Staple
Commodities, by J. M. Keynes, J. W. F. Rowe and G. L. Schwartz, pp. 10-11.

2 Cf. L. M. Lachmann, ¢ Commodity Stocks and Equilibrium,” Review of Econ. Studies,
Vol. 111, n. 3, June, 1936.


===== PAGE 12 =====

446 ECONOMICA [NOVEMBER

investment activity. Moreover, as the Austrian Theory
has shown, scarcity at one stage of production is quite
consistent with unsaleable stocks at another stage. For
these and similar reasons, until we have more accurate
knowledge about the distribution of stocks between capital
goods—and consumption goods industries in general, and
of raw material stocks between producers, dealers, and
manufacturers of finished commodities in particular, extreme
caution in the cyclical interpretation of the low level of
stocks seems advisable.

If the statistics presented are representative of the behaviour
of stocks of unfinished goods in general, what conclusions
have we to draw with regard to the theory of the Trade
Cycle ? If these stocks diminish during the upswing they
evidently offer no scope for investment. It follows that
for the source of that investment activity which characterises
the upswing we have to look elsewhere, i.e., we probably
have to seek it in investment in fixed capital. The obverse
applies to the downswing, and we are therefore entitled to
conclude that investment and disinvestment in staple
commodities’ stocks, so far from being secondary forces in
the mechanism of the cumulative process, actually are
retarding forces, offsetting to a certain extent the effects of
investment in equipment.

This statement has to be qualified in several respects.
In the first place, we have to remember that our statistics
refer to unfinished goods only and that, as we said, there
is reason to believe that stocks of finished commodities
move in positive correlation with the cycle?!, i.e., that
producers and merchants of consumption goods tend to
keep their stocks in a certain proportion to their turnover.
In so far as changes in raw material stocks merely offset
opposite movements in the stocks and production of finished
goods—owing to the lag with which raw material production
is adjusted to changes in demand—raw material stocks
serve as a kind of excess reserve for the industries producing
finished commodities (buffer function).

Secondly, the size of stocks has a direct causal influence
on the production of raw materials. Not only will output
be restricted as long as stocks are accumulating—very
much against the wishes of the producers who have to carry

1 Ralph H. Blodgett, Cyclical Fluctuations in Commodity Stocks (University of Pennsylvania
Press), Philadelphia, 1935, Appendix C, p. 171.


===== PAGE 13 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 447

them—but even while they are falling it is unlikely to
recover before, indeed, stocks have again reached a normal
size. Thus, as long as stocks are large, an increase in demand
will not immediately lead to an increase in supply ; in this
case the buffer function of stocks will check the cumulative
process.

Another point to be considered in this connection is the
relationship between the size of stocks and investment in
the production of raw materials. To the extent to which
the investment activity characteristic of periods of prosperity
is due to investment in raw material production our inverse
correlation may not hold. Unless demand grows more
rapidly than supply is forthcoming stocks will increase and
prices tend to fall. This need not cause a general collapse
as long as investment in other parts of the economic system
remains satisfactory ; it may even stimulate expansion in
raw material consuming industries. Still; it is true that
investment in the production of raw materials undertaken
in the expectation of a rise in demand which does not
immediately occur will lead to a, perhaps temporary, increase
in stocks. There is every reason to believe that the steady
rise in raw material stocks between 1923 and 1929—in
positive correlation with the Trade Cycle!—has to be
ascribed to similar causes.

If it is true that in the past recovery has usually been
preceded by an accumulation of stocks of industrial raw
materials, it follows that all schemes aiming at a restriction
of output by means of monopolistic control have to be
regarded as potentially dangerous. One has, of course, to
beware of the post boc ergo propter hoc fallacy, and it follows
by no means that accumulation of stocks of a definite size
is a necessary condition of Recovery. It may well be that
in the past Recovery would have occurred even with
much smaller stocks. But we beg to submit that then it
may have been much shorter. At any rate, the danger of
‘“ bottlenecks ” being encountered would be greater.

For these reasons we are unable to follow Mr. Keynes
in his advocacy of restriction schemes. It seems to us that
such schemes are justifiable only where it is impossible for
a price fall to lead to readjustment, i.e., where the following
three conditions are fulfilled :

(1) demand is very inelastic;
(2) prime costs are either constant or falling ;


===== PAGE 14 =====

448 ECONOMICA [NOVEMBER

(3) all producers work under identical conditions
so that there are no high cost and low cost
producers.

Mr. Keynes argues that the present economic system
offers no mechanism for the carrying of surplus stocks,
hence the necessity of restriction schemes. The material
we have presented seems to indicate that in spite of his
contention that the holding of large stocks is too costly
to be feasible such holding did and does in fact take place,
and that stocks do exercise a buffer function.?

VI

Out of the foregoing there arises the interesting problem
of the scope and significance of commodity speculation
during the Trade Cycle. From all we know from the reports
and descriptions of contemporary observers of past booms,
commodity speculation has always been a strong, and in
some instances a decisive factor. But, if stocks of unfinished
commodities actually diminish during the upswing, how is
commodity speculation possible ? The possibility of specula-
tion in finished goods may be dismissed as it is obvious that
because of their low “ plasticity 2 large speculative trans-
actions in them are not feasible. Moreover, we know from
experience that speculation is usually most intense on the
big markets for staple commodities which offer speculators
the greatest facilities. How, then, is the riddle to be solved ?

A given change in price is not necessarily correlated with
a given volume of transactions. From the financial press
we all know instances where “ the movement of prices was
out of all proportion to the volume of dealing”. The
volume of transactions necessary to bring about a certain
price movement is an indication of the division of opinions

1 After this paper had been completed, Mr. Keynes once more took up the subject of com=
modity stocks in a paper read at the Cambridge meeting of the British Association in August,
1938 (“The Policy of Government Storage of Foodstuffs and Raw Materials,” Economic
Fournal, September, 1938). While his diagnosis has remained essentially the same,—* The
competitive system abhors the existence of stocks, with as strong a reflex as nature abhors a
vacuum, because stocks yield a negative return in terms of themselves,” op. cit., p. 449—his
therapy is new. He seems to have grown sceptical of output restriction which “is apt to be
objectionable in general, even when it is highly desirable for the particular purpose of meeting
fluctuations, because it may be part and parcel of conditions of almost uncontrolled monopoly ”.
What he proposes is, briefly, a government subsidy for the carrying of stocks. To this we have
no objection. But it still seems to us, in the light of the statistics we have presented, that his
fears about insufficient stock-carrying in a competitive world are a little exaggerated. The
whole issue has, however, now become a question of degree. .

2 H. Makower and J. Marschak, * Assets, Prices and Monetary Theory,” Economica,
August, 1938, pp. 280-1.


===== PAGE 15 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 449

in the market. If everybody expects prices to rise, they
will rise without any transactions taking place. Hence,
the more * bullish ” the market during the boom the less
transactions are necessary in order to bring about a given
price rise. In other words, commodity speculation during
the boom will not lead to an accumulation of stocks, if
sellers, buyers and consumers are all equally “ bullish ”.
It follows that stocks can increase only to the extent to
which producers and speculators are more * bullish ” than
consumers.!

Let us restate the same thing in the terminology of the
forward market most appropriate where we have to deal
with intertemporal price- and quantity-relations. Stocks
can accumulate only if the forward price exceeds the spot
price by more than the carrying costs, for only then it will
be profitable to carry them. Hence, changes in stocks are
determined by changes in the forward price relatively to
the spot price. It follows that an increase in stocks during
the boom can occur only in so far as the spot market tends
to lag behind the forward market, i.e., to the extent to which
operators in the spot market are less “ bullish > than those
in the forward market. Where there is no division of opinions
between the two markets, and the spot price immediately
follows every movement of the forward price, there can be
no change in the size of stocks. Such changes are propor-
tionate to the dispersion of opinions.

We have now seen why commodity speculation during
the boom need not lead to an increase in stocks, if optimism
is sufficiently widespread. But the inverse correlation
between commodity stocks and the price level conceals
even more interesting problems. We have found that in
the upswing stocks of raw materials actually decrease, i.e.,
forward prices tend to fall relatively to spot prices. As
we pointed out, this may be due to actual shortage of supply.
Where production can only be adjusted with a time-lag,
a situation may be reached in which present supply is short
but future supply plentiful, and where therefore nobody
will carry stocks. But the explanation of our inverse
correlation in terms of increasing physical scarcity during
the upswing is not the only possible one. It may be due
to the superior skill of operators in the forward market
who in this phase of the cycle already anticipate the next.

1 Cf, above p. 442 about the 1905 boom in * Middlesboro’ N. 3”.


===== PAGE 16 =====

450 ECONOMICA [NOVEMBER

If a decrease in stocks indicates a tendency of the forward
price to fall relatively to the spot price, this shows an increas-
ing divergence of opinions. If this occurs during a boom,
it means that forward market operators are less ‘ bullish ”
than their colleagues in the spot market. In which case
we shall reach the astonishing conclusion that the commodity
speculation characteristic of the boom is just speculative
activity of people who ordinarily do not speculate”
i.e., operators in the spot market, which drives up the spot
price relatively to the forward price.

If this were true, it would be, in fact, a tribute to the
superior foresight of the professional speculators operating
in the forward markets, and it would, moreover, show why
a free exchange economy with a well-developed system of
intertemporal markets operated by specialists can weather
many storms. We must not forget, however, that the
relationship between the size of stocks and the difference
between forward and spot price applies to covered stocks only.

This makes it extremely difficult to use our theorem for
the depression. The large stocks characteristic of the
downswing are, as we saw, mostly carried by producers
very much against their own wishes and for the purpose of
preventing a complete collapse of prices. They are probably
‘““ unhedged ”’, since if they were sold forward they would
affect prices. Therefore the accumulation of stocks in the
depression cannot very well be ascribed to the superior
foresight of speculators.

This is, of course, not to say that the existence of large
uncovered stocks outside the market will not affect the
latter. We know from experience that the existence of
huge stocks kept outside the market will by forcing down
the forward price cause the liquidation of “hedged * stocks.

In applying this theory to the tin market during the
last 13 years we find that it is, on the whole, borne out by
the facts.! Broadly speaking, boom periods are accompanied
by a backwardation and falling stocks, depressions by a
contango and stock accumulation.? In 1926, 1927 and 1928
visible supplies of tin were extremely scarce and did not
reach 20,000 tons until October, 1928. Throughout this
period there was a backwardation which in 1926 and 1927
averaged [6.16 and occasionally reached fro.

1 The following figures are taken from Tin (annually published in London by the Inter-
national Tin Producers’ Association). Total visible supplies as estimated by W. H. Gartsen.
2 In some markets contango and backwardation are called premium and discount,


===== PAGE 17 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 451

From February, 1929, till the end of 1932 total visible
supplies rose from 26,000 tons to 46,000 tons while the
contango was never less than [I1.

By February, 1934, stocks had fallen to 21,000 tons
while spot and forward prices hardly deviated. This fall
in stocks was due to the liquidation of the International
Tin Pool.

In 1934, with an average backwardation of 19s. 2d.
stocks fell to 12,600 tons. This must have been a minimum
level, for whilst throughout 1935 and 1936 the backwardation
was never less than [2, stocks remained at a little above
13,000 tons. In November-December, 1936, the backwarda-
tion disappeared and in January-February, 1937, gave
way to a contango which by March had brought up stocks
to about 20,000 tons.

Monopolistic interference with the tin market started in
1928 : “ During the year a syndicate which came to be
known as ‘The Group’ was formed to hold tin off the
market and the rise in price which began in August may
mark the beginning of their operations.” Their activity
would explain why stocks rose suddenly between August,
1928, and February, 1929, in spite of a large backwardation.

The International Tin Pool was formed in September,
1931, and by the end of January, 1932, had acquired 21,000
tons. Liquidation began in July, 1933, and lasted until
the early months of 1934. There can be little doubt that
but for its sales of spot tin the backwardation which did
not exceed [1 before March, 1934, would have appeared
earlier.

VII

At last we have to discuss what light, if any, is thrown
by our investigations upon some modern trade cycle theories.
Of course, how one expects commodity stocks in general
and raw material stocks in particular to behave during the
Trade Cycle depends on the type of theory one happens
to hold. We have found that the cyclical behaviour of
(industrial) raw material stocks conforms to a definite
pattern which, it would seem at first sight, must rule out
at least some theories.

In fact, however, practically all those theories which

1 J. K. Eastham. * Rationalisation in the Tin Industry,” Review of Econ. Studies, October,
1936, p. 20.


===== PAGE 18 =====

452 ECONOMICA [NOVEMBER

stress the importance of fluctuations in investment in fixed
capital as the outstanding feature of the Trade Cycle are
borne out by our material. The reason is that an increase
in investment activity of this kind involves an increase
in the demand for mineral products such as iron, copper,
tin, the supply of which can only be adjusted with a time-
lag. Hence, in the meantime stocks are bound to decrease
as they are likely to augment in the case of a setback in
investment activity. It follows that all over-investment
theories are consistent with the results of our investigation.
On the other hand, Recovery cannot possibly start in the
raw material producing industries, hampered as they are
by large and increasing stocks which have to be cleared
before their production can recover.

Unfortunately we are not in a position to judge the relative
merits of various over-investment theories. In particular,
our statistics neither corroborate nor disprove the so-called
“ monetary over-investment theories’ of the Austrian
School, as we do not know enough about the distribution
of stocks among the different stages of production. As
we mentioned above, only two of our series include manu-
facturers’ (rubber and cotton) stocks. Mr. Blodgett, how-
ever, has pointed out that stocks of raw materials in the
hands of manufacturers tend, on the whole, to move in
inverse correlation to the Trade Cycle! What would be
required in order to verify ‘ monetary over-investment
~ theories ” is exact knowledge about the relative movements
in the stocks of the manufacturers of capital and consumption
goods respectively, but it is here that we are almost completely
ignorant. Where the ““over-investment” of the boom is
due to investment in the production of raw materials, stocks
may, of course, increase with production, and then their
inverse correlation with the Trade Cycle will be broken.
As we pointed out above, this actually was the case
between 1923 and 1929 when all stocks were increasing
rapidly.

With regard to Mr. Keynes’ views on these and similar
subjects we have to distinguish between his general trade
cycle theory? and his thesis about the cyclical fluctuations
in commodity stocks. Whereas the former may be described

1 Blodgett, op. cit., pp. 5-8.
2 General Theory of Employment, Interest and Money, ch. 22 : “ Notes on the Trade Cycle ”.
3 Treatise on Money, Vol. 11, ch. 29 : * Liquid Capital ”.


===== PAGE 19 =====

1938] COMMODITY STOCKS IN THE TRADE CYCLE 453

as an over-investment theory the latter does not form a
necessary part of it, l.e., the investment activity of the
boom is mostly due to investment of fixed capital. The
depletion of “liquid stocks ” which in his opinion is a pre-
requisite of Recovery as well as the increase in “ working
capital ” accompanying the phase of expansion are, on the
whole, secondary forces in the mechanism of the cumulative
process which originates from and centres in the investment
in equipment. Thus, even if our statistics show that fluctua-
tions in stocks so far from being secondary factors are actually
retarding forces, this does not affect Mr. Keynes’ argument.
Moreover, we must not forget that our theory refers to
strictly industrial raw materials only.

What are our conclusions to be with regard to under-
consumption theories ? It is well known that this doctrine
cannot be disproved or proved by a mere comparison of
total quantities like, e.g., investment and consumption.
What is required in order to decide whether or not under-
consumption was the cause of a crisis is a knowledge of
events in their chronological order. In other words, what
we would have to know is what increased first, stocks of
finished consumption goods, of finished capital goods, or
raw material stocks (the problem of “leads > and lags ”’).t
Raw material stocks alone are an insufficient criterion.

There is, however, a version of under-consumption theory
which has a direct bearing on the demand for industrial raw
materials. In the * under-consumption cum acceleration ”
theory of Professor J. M. Clark and Mr. Harrod investment
is linked up with the rate of increase of consumption. Hence,
every slowing down in this rate of increase entails a setback
in investment activity which explains the increase in the
stock of mineral products. This version of under-consumption
theory is therefore entirely consistent with the statistics
we have presented.

The only trade cycle theory which, at first sight, it seems
difficult to reconcile with the results of our investigation
is that of Mr. Hawtrey, who emphasises investment in
stocks as the impelling force of the Trade Cycle. But, since
his theory refers to stocks of manufactured commodities
only and not to agricultural and mineral products? it does
not affect our argument. Moreover, it is highly probable

1 Blodgett, op. cit., p. 103.
2 R. G. Hawtrey, Capital and Employment, London, 1937, pp. 116-17.


===== PAGE 20 =====

454 ECONOMICA [NOVEMBER

-that stocks of finished commodities in the hands of whole-
salers and retailers do move positively with the cycle. Still,
if our statistical conclusions are correct, the forces released
by Mr. Hawtrey’s movements in finished commodity stocks
are largely offset by the economic forces which we have
endeavoured to describe.