Concept and Statistical Measurement of Vertical Integration · 2013-04-25 · VERTICAL INTEGRATION...
Transcript of Concept and Statistical Measurement of Vertical Integration · 2013-04-25 · VERTICAL INTEGRATION...
This PDF is a selection from an out-of-print volume from the NationalBureau of Economic Research
Volume Title: Business Concentration and Price Policy
Volume Author/Editor: Universities-National Bureau
Volume Publisher: Princeton University Press
Volume ISBN: 0-87014-196-1
Volume URL: http://www.nber.org/books/univ55-1
Publication Date: 1955
Chapter Title: Concept and Statistical Measurement of Vertical Integration
Chapter Author: M. A. Adelman
Chapter URL: http://www.nber.org/chapters/c0965
Chapter pages in book: (p. 281 - 330)
CONCEPT AND STATISTICAL MEASUREMENTOF VERTICAL INTEGRATION
M. A. ADELMANMASSACHUSETTS INSTITUTE OF TECHNOLOGY
WE OFTEN speak of a firm as being highly integrated vertically, ofone industry as being more integrated than a second, of either afirm or an industry as becoming more integrated or less integratedin the course of time. Once we speak in terms of greater and less,we have spoken in terms of quantity and have indicated the needof measuring quantity.
This need is emphasized by the fact that one commonly usedmeasure of size and of concentration, the amount of sales, has a well-known shortcoming. This measure does not indicate the degree ofvertical integration; for example, a manufacturer selling $i millionworth of goods is a much larger firm than a retailer selling the sameamount. Firms of such dissimilar structure should not be classifiedtogether upon the basis of size of sales. It is a mistaken though com-mon practice to analyze firms statistically by computing the per-centage of their sales that is allocated to research, wages, or otherpurposes. Also, it is a regrettable fact that the "concentration ratios"developed by the Temporary National Economic Committee can bestated only in terms of sales; however, as a measure of oligopolyrather than of concentration, this is not nearly so objectionable.Furthermore, ambiguities arise when attempts are made to measurethe trend of concentration through time.'
It seems clear that there is some kind of relationship, at least ofa formal kind, between integration and concentration. The objectof this paper is to develop the formal relationship as an aid to thestudy of the real one. Accordingly, two requirements have been setup for defining any measure of vertical integration. First, it mustbe an extension of, and consistent with, accepted economic doctrine.Second, it must be operational and capable of statistical treatment.Two such measures will be proposed and used here. Their use sug-gests that size is positively correlated with the degree of verticalintegration, but the relation is not a simple one.
1. Ratio of Income to SalesTHE first proposal employs the concept of the ratio of income to
1 M. A. Adelman, "The Measurement of Industrial Concentration," Review ofEconomics and Statistics, November 1951, pp. 272, 291.
281
VERTICAL INTEGRATION
sales. Every firm is confronted by a choice between purchasing orselling on the one hand and additional processing2 on the other:make or buy, sell or process further. The decision depends uponthe particular economies of each course of action. Through verticalintegration the firm by-passes or, more accurately speaking, encom-passes a market nexus. Administrative direction replaces the bar-gaining of the market. In the widest sense all firms, even small ones,are vertically integrated in that they could conceivably be dividedinto two or more firms, earlier stages selling to the later.8
Now, were all firms and industries completely integrated, therewould be no sales except to final consumers. A tableau of the Leon-tief type4 (but arranged on a firm rather than on an establishmentbasis) would be collapsed into the lowest line and the extreme right-hand column. The total sales of business firms to consumers wouldbe equal to the income originating in the business sector of theeconomy. There is, of course, a whole family of income measure-ments, each with a different degree of netness; but assuming con-sistency in use, complete vertical integration would mean that theratio Y/S (where Y denoted income, and S sales) would equal unity.The less integrated the business system, the more interfirm transac-tions there would be, the larger would S become, and the smallerwould be the ratio.
If the firm instead of the whole economy is considered, the situa-tion is largely but not wholly the same. The sales of the firm,whether to consumers or to other firms, are equal to the total in-come generated up to that point of sale. The income originatingwithin the firm is its own contribution to that total. Thus the ratioY/S defines the degree of integration only up to the point of saleand takes no account of operations past this point. For example,suppose that in a given industry there are three firms: a primaryproduction firm, a manufacturing firm, and a distribution firm; eachcontributes one-third of total value added by the industry. Theprimary producer-_on the unrealistic assumption that he buysnothing from other firms—would have a ratio of 1.0; the manu-facturer, a ratio of .5o; the distributor, one of .3g. If the manufac-turer integrated backward to absorb primary production, the new
2 The word "process" or "make" means any productive activity, not merelyextractive or manufacturing activity.
S R. H. Coase, 'The Nature of the Firm," Economica, November ig', p. 389.4 W. W. Leontief, The Structure of American Economy (Harvard University
Press, 1941), pp. 15-20, and Tables 5 and 6.
282
VERTICAL INTEGRATION
firm would have a ratio of 1.0; if he integrated forward to absorbdistribution, the new firm would have a ratio of .67.
Thus the ratio of income to sales, when calculated for the singlefirm, does not have the simplicity and accuracy it has when calcu-lated for the whole business economy. The nearer we go to primaryproduction, the less sensitive the index becomes to changes in thedegree of integration. This is a serious limitation, but it can also beturned to good use. Thus, if our completely integrated primary pro.ducer has integrated forward and his ratio of income to sales hasactually decreased, it follows that he must now be very asymmetrical,with the later stages buying a considerable amount from outsidesources. Again, if two firms of apparently identical function showa significant difference in ratios, the identity must be spurious.
The measure of integration in an industry is in concept midwaybetween those involving the economy and the firm. Thus, if everyindividual producer in the industry were completely integrated or if,as a special case of complete integration, the industry were monop-olized, there would be no sales within the industry. But if the in-dustry were finely subdivided into as many firms as there were suc-cessive processes, the amount of sales would be very much larger.Thus the ratio of sales to value added is again an index of the de-gree of vertical integration. But for the industry the ratio reflectstwo separate characteristics: one is the "stretch" of the whole indus-try from material entry to product exit; the other is the degree ofsubdivision between these two points. And for the purposes of anyparticular investigation, it may make a good deal of difference whichcharacteristic is responsible for the ratio.
2. Ratio of Inventory to SalesA SECOND proposed or alternative measure of vertical integration isthe ratio of inventory to sales. The longer the production line andthe more successive processes are operated by one firm, the higherthe ratio. This is a derived measurement rather than a direct one.There is no limiting value which would signify complete integration,and all figures are of purely relative significance. The accuracy ofthis measure would probably be improved if it included only goodsin process, since this would be closer to the length of the productionline. The ratio of inventory to sales has one particular virtue as ameasure of integration: it is not distorted by the nearness of the firmto primary production; the other ratio or measure is. On the other
283
VERTICAL INTEGRATION
hand, the ratio of inventory to sales is more susceptible to meaning-less comparisons.
3. Divergent, Convergent, and Successive FunctionsAT THIS point it is useful to consider an earlier attempt at quantita-tive measurement of integration, both horizontal and vertical. Mon-ograph 27 of the Temporary National Economic Committee classi-fied multiplant firms into five broad types.5 The most common typeconsisted of firms performing a "uniform function"; this illustratesthe most obvious form of horizontal integration. The least commontype engaged in "unrelated functions," or conglomerate integration.Of the 5,600 multiplant firms recorded by the Bureau of the Censusin 1937, 2,100 controlled plants in more than one industry, but onlyg firms controlled plants without visible functional relationshipamong themselves. Even this number the monograph considers asan overestimate resulting from the lack of time and money to delvefurther.6 Types of integration involving "uniform" and "unrelated"functions are outside the scope of this study. The other three broadclassifications may be considered in detail, since they are involvedto some extent in this analysis.
The "divergent functions" performed by one group of firms in-clude: (a) "Joint products," which are defined as goods made fromthe same raw material or subassembly. This would seem to be hori-zontal rather than vertical integration. (b) "By-products," which arereally joint products as commonly defined in economic theory; thatis, they are technically inseparable, so that the firm cannot produceone without the other. The possession of an establishment for fur-ther processing a by-product is vertical integration. (c) "Like proc-esses," or producing goods which are different in the physical andmarket sense, such as woolen and cotton woven goods. This consti-tutes horizontal integration.
The "convergent functions" of another group are also subdivided:(a) "Complementary products," where two or more specialized plantssupply the several components of the finished product. (b) "Auxil-iary products," where one plant supplies a product or products ofanother. Both of these seem to involve that progression from earlierto later stages of production which constitutes vertical integrationand would be reflected in our two ratio measurements. (c) "Like
5 Willard L. Thorp and Walter F. Crowder, The Structure of Industry, Tem-porary National Economic Committee, Monograph 27, 1941, Part ii, Chap. .
6 Ibid., pp. 206-207.
284
VERTICAL INTEGRATION
markets," where physically unlike products result. This would beclassified as horizontal integration.
The "successive functions" of the fifth group illustrate verticalintegration in the narrowest sense.
The monograph lists, by major industrial groups, the number ofcentral offices (of multiplant firms) operating in each group, accOrd-ing to the function of the central office. Although this informationis interesting, it is not a measure of the degree of integration. Eachcentral office was counted once, whether it controlled two establish-ments or two hundred; there is no indication of the number of em-ployees or of the amount of value added by the establishment con-trolled by the central office. The total number of establishments in-volved is not given. Even if these data were available, this would byno means solve the problem. A "plant" is sometimes an arbitrarygrouping and may as properly be called two (or more) as one. Asingle plant does not necessarily perform a single function or pro-duce a single product. Thus most large firms in the rubber industrywould on common-sense grounds be considered vertically integrated,but the integration or the succession of functions takes place withinvery large plants, so that the mere counting of plants and of centraloffices performing vertical functions would not indicate verticalintegration.
4. Statistical DataWITH these various concepts in mind, we turn to the statistical evi-dence. Table i summarizes census data on multiplant production,which reflect both horizontal and vertical integration for 1939 and1947. The change between these years is not significant; the numberof establishments of the multiplant firms, as would be expected inso expansionary a period, did not increase as much as the total num-ber of establishments of all firms, but the percentage of value addedof multiplant firms8 scarcely diminished. The establishments of themultiplant firms are substantially larger than the average establish-ment in terms of the number of employees and still larger in termsof value added. Thus since multiplant production is unmistakablyassociated with size and also with vertical and horizontal integration,these data constitute evidence of some positive association between
7 Ibid., pp. 196-197.8 This is not a measure of change in concentration, however, because the
census does not give us the total number of manufacturing firms or the num-ber of multiplant firms for either year.
285
VERTICAL INTEGRATION
size and vertical integration. But this association is extremely loose;nothing more precise can be obtained from the census data.
Table 2 presents a comparison of the income-sales ratios for asample of 183 large manufacturing corporations, and Table 6 showsthe same comparison for the manufacturing corporate universe. Thesample was selected from available data in annual reports; for thisreason the proportion of income thus accounted for varies widelyamong industries. For both the sample and for the universe, thedenominator of the ratio is sales. This would require for strict com-parability a similarly gross concept of income; hence the best numer-ator would be one which comprehends the whole spread betweenpurchases of goods and services from other firms and sales. Unfor-tunately, such data are available in only a few of the corporate re-ports; therefore, income has been defined as the total of (a) payrolls,including supplemental employee payments and "fringe" benefits,(b) profits before federal income taxes, (c) interest, and (d) de-preciation.
For the universe and the industry subdivisions, the denominatoris corporate sales as estimated by the Department of Commerce.9The numerator is, in effect, income originating in the corporatesector by major industrial groups.1° The largest element is the Com-merce estimate of "wages and salaries" by industry. For each indus-trial grouping, the estimate has been multiplied by a factor repre-
9 The advantage of using Commerce data arises from their conformity withthe national income concept. But they arc based on Statistics of Income, Bureauof Internal Revenue; and to the extent that corporate income tax returns arenot completely consolidated, certain amounts appear as sales which are reallyintracorporate transfers. Hence there must be an upward bias in the sales figureand a downward bias in the Y/S ratio.
The extent of this bias is suggested by comparison of Statistics of Incomewith the Quarterly Industrial Financial Report Series, 1949, Federal Trade Corn.mission and Securities and Exchange Commission, since large corporations in thelatter are completely consolidated. For all manufacturing, the Quarterly Indus-trial Financial Report Series, total is about 15 per cent below that of theBureau of Internal Revenue. Unfortunately, there seem to be some factorsother than consolidation involved here. The discrepancy is least in the big-business industries and largest in the small-business industries, although the im-portance of consolidation is just the contrary. It has so far proved impossible todevise an adjustment.
10 This is the first attempt to construct such a table, so far as I am aware,and some revision is necessary before it can be considered satisfactory. Such re-finement would be well worth while, in my opinion, because these statistics couldthen be used for other purposes—most notably to extend John Lintner's valu-able study of corporate profits and national income, Corporate Profits in Per-spective (American Enterprise Association, Inc., 1949).
286
VERTICAL INTEGRATION
senting the corporate share of total employee compensation in theindustry. The factor is derived largely from Census of Manufactures,1947 and from some miscellaneous subsidiary sources. Corporateinterest is the figure taken from the Bureau of Internal Revenuereports, since the Commerce estimates of interest are much too net—they measure "the payments less the receipts of relevant payergroups" rather than the interest outpayments of corporations, withwhich we are concerned. Corporate profits are those estimated bythe Quarterly Industrial Financial Report Series, since the adjust-ments made by the Department of Commerce to fit the national in-come concept make its estimates less comparable with the sampleand the Bureau of Internal Revenue data were not available at thetime of writing.
There is a question as to whether interest should be counted asincome originating within the firm, since it might be consideredsimply a payment for services, perhaps comparable with a paymentfor electric power service. Our view is that creditors should becounted as members of the corporate family. Profits plus interestinclude the return on the property of the corporation; to excludeinterest would be to have income determined by the fortuitous effectof the company's particular capital structure, which is a matterlargely of the discretion of management.
Corporate reports usually do not show the amount of rents androyalties paid out, but these items would be excluded in any event,since they are payments for services by outside persons or agencies.If the firm bought rather than rented, it would be more highlyintegrated and the return to capital would be higher; similarly, thefirm has a choice between paying patent royalties or setting up aresearch department, so that payment would be shifted from roy.alties to salaries and to property income. Nor is the matter differentin the case of the firm without any such choice. The best-knownexample of this is found in shoe machinery," where the paymentsto the lessee of shoe machinery are clearly to be imputed to the latter.
No two corporate reports are exactly alike, and a host of smalladjustments were necessary to keep the data comparable. An impor-tant defect of Table 2 results from the volatility of corporate profitsover time and their variability among firms. The more capital-intensive the firm and the higher the ratio of property income to
11 United States v. United Shoe Machinery Corp., Commerce Clearing House,1953, par. 67436.
287
VERTICAL INTEGRATION
labor income generated, the wider the margin of possible error. Andthe more unsettled the general price level, the less precision ofmeaning do the profit data have: for this reason, the year 1949 wasused in Tables 3A, 3B, and 4. If there were a systematic and strongassociation between size and profitability, there might be as a dimin-ished echo an association between size and vertical integration. Thisdoes not seem to be the case during periods of high employment.1'
For most branches of manufacturing, the Commerce industrygroupings are at least as broad as the fields of activities of the largecorporations in the sample. But some difficulties were encountered.Employee income, which is the major part of total income, is givenon an establishment basis in our national income statistics, so thatemployee income originating in the nonmanufacturing activities ofa predominantly manufacturing concern appears in the nonmanu-facturing sector. But corporate profits, as recorded by the Bureau ofInternal Revenue and adapted by the Department of Commerce,are shown' by firms and not by establishments. This is comparableto the reports used in our sample but not comparable to the othernational income statistics.18
The best procedure might have been to estimate, for each non-manufacturing industry, the total of all activities operated by con-cerns predominantly engaged in manufacturing, in order to obtaina "mixed" universe comparable with the "mixed" sample. Since thiswas impossible, it was necessary to include in the activities of cer-tain manufacturing industries those of the whole nonmanufacturingindustry which supplies them, as explained in the notes to Table 3A.
Table 4 presents the ratio of the value of inventories to sales com-puted from data given in Statistics of Income for the years 1940 and1949. Since inventories are a relatively unstable item of assets, af-fected not only by changes in business activity and in price but alsoby anticipations of both, it is desirable to choose years when thesewere at a minimum. During 1940 the sum of the absolute values ofthe inventory valuation adjustment, for all corporations and foreach component of the business economy, was the lowest for anyyear of the period 1929-1951 (with the exception of '935, for which
12 Joseph L. McConnell, 'Corporate Earnings by Size of Firm," Survey of Cur-rent Business, Dept. of Commerce, May 1945, and "1942 Corporate Profits bySize of Firm," Survey of Current Business, January 1946; Sidney S. Alexander,"The Effect of Size of Manufacturing Corporation on the Distribution of theRate of Return," Review of Economics and Statistics, August 1949, pp. 233-25.
11 National Income Supplement, 1951, Survey of Current Business, Dept. ofCommerce, p. 85.
288
VERTICAL INTEGRATION
statistics are not otherwise as satisfactory).14 The data for 1940 are
unsatisfactory for two reasons: (i) 1940 is more than a decade past;(2) because consolidated tax returns were not permitted between1934 and 1941, the figures result in an artificial equalization of firmsize and an artificial increase in sales. For these reasons, it is desirableto use data for a later year. There was the smallest inventory distor-tion for 1949 of all nonwar years since 1940.
5. Size AssociationsTABLE 1 suggests that large corporations are more integrated verti-cally than are small corporations; this is not satisfactorily verifiedin Table 2. The reasons for this inconclusive result are fairlyobvious: (a) the diversity within each two-digit industry group and(b) the small number of corporations within each subgrouping.Certain such subgroups are purely formal or residual; for example,the group "Stone, clay, and glass" contains four diverse and non-comparable kinds of enterprise.
Some of the industry detail is comparable, however. In meat pack-ing and in dairy products, there seems to be a mild association be-tween size and degree of integration. The rest of the food firms andthe tobacco firms are too diverse in their output to allow any mean-ingful comparison. No textile group shows any trend. No relation isobservable in paper production or in chemicals, rubber, or petroleumrefining. The homogeneity of the major oil companies is striking,but Standard Oil of California is substantially more integrated thanany other. In primary iron and steel, among the first eight concerns,there appears to be a positive relation; and the same may be said ofelectrical machinery, if we exclude a rather specialized firm likeRaytheon.
The transportation equipment group is perhaps the most interest-ing. It has long been known that General Motors (like Ford) isconsiderably more integrated than Chrysler; according to the table,this is in the proportion of to 3. Yet one would hesitate to concludethat there was any tendency in general for larger automobile com-panies to be more integrated than smaller, for there is no significantdifference between Chrysler and the other, much smaller, automobileassemblers (even if Nash-Kelvinator is disregarded as being too mucha part of the electrical machinery group). The large parts makerslike Briggs, Budd, and Borg-Warner are, as was to be expected, moreintegrated than the automobile builders other than General Motors.
14 Ibid., Table 22.
289
TAB
LEi
Cor
pora
te M
ultip
lant
Firm
s by
Man
ufac
turin
g In
dust
ry G
roup
s, 19
39 a
nd 1
947
(PER
CEN
TAG
ES R
EFER
TO
TO
TAL
OF
IND
UST
RY
GR
OU
P)A
VER
AG
E N
UM
BER
OF
PRO
DU
CTI
ON
WO
RK
ERS
Per C
ent
Per C
ent
Inof
All
of A
llC
orpo
rate
ESTA
BLI
SFIM
ENTI
OPE
RA
TED
Prod
uctio
nW
ages
and
Mul
tiuni
tIn
All
Num
ber
Per C
ent
Wor
kers
Sala
ries
Esta
blis
hmen
tsa
Oth
ers
Indu
stry
Gro
up'9
39'9
47'9
39 '9
47'9
39 '9
47'9
39 '9
' '"
'939
'947
Food
and
kin
dred
pro
duct
s9,
273
8,oi
621
220
.153
.354
.1n.
a.54
.946
7411
i6To
bacc
o rn
anuf
actu
resb
119
2ii
15.6
19.3
73.3
78.9
n.a.
8o.g
539
386
5125
Text
ile m
ill p
rodu
cts
1,92
11,
742
29.8
214
59.5
55.0
n.a.
54.1
335
364
7So
App
arel
and
rela
ted
prod
ucts
1,43
11,
669
7.1
5.4
28.1
23.9
n.a.
19.5
148
139
2925
Lum
ber a
nd lu
mbe
r pro
duct
s
©
(exc
ept f
urni
ture
)cFu
rnitu
re a
nd fi
xtur
esc
Pape
r and
alli
ed p
rodu
cts
1,95
72,
169
i6.i
8.46
345
99.
16.
o1,
275
1422
38.9
34.7
39.6
30.3
27.4
26.1
66.
68.6
n.a.
33.9
n.a.
27.5
n.a.
69.2
So8
115
i6i
i8i8
8
2317
3029
4446
Prin
ting
and
publ
ishi
ng1,
223
1,20
24.
94.
125
.229
.3n.
a.32
.667
107
1011
Che
mic
als a
nd a
llied
pro
duct
s3,
047
2,85
838
.128
477
.773
.3n.
a.72
.073
120
1017
Petro
leum
and
coa
l pro
duct
s64
361
465
.044
.392
.985
.5n.
a.85
.915
223
621
32R
ubbe
r pro
duct
si6
ii8
o27
.120
.669
.474
.7n.
a.76
.752
089
085
78Le
athe
r and
leat
her p
rodu
cts
627
647
17.9
12.2
48.9
44.5
n.a.
414
256
240
5842
Ston
e, c
lay
and
glas
s pro
duct
s1,
835
1,52
026
.113
.057
.755
.7n.
a.58
.890
149
23i8
Prim
ary
met
al in
dust
riesd
(571
)1,
261
21.7
23.5
54.7
77.2
n.a.
77.5
223
6i8
5156
Fabr
icat
ed m
etal
pro
duct
sc(1
,318
)1,
523
13.7
9.1
44.0
47.4
fl.B
.47
.014
425
629
29M
achi
nery
(exc
ept e
lect
rical
)1,
253
1,55
913
.28.
752
.253
.4n.
a.53
.021
842
630
35El
ectri
cal m
achi
nery
441
8821
.922
.367
.073
.2n.
a.74
.738
952
954
55Tr
ansp
orta
tion
equi
pmen
t56
461
526
.8i6
.680
.680
4n.
a.81
.398
6 1,
290
706
Inst
rum
ents
and
rela
ted
prod
ucts
c18
327
5i6
.oio
.650
.558
.1n.
a.59
.221
138
439
33M
isce
llane
ous m
anuf
actu
resc
505
992
6.6
7.0
27.6
33.8
n.a.
35.6
103
135
1920
Tota
l29
,500
29,
566
i6.o
12.3
54.6
54.8
564
146
221
29.4
25
z T1 71 0 z
See
page
292
for f
ootn
otes
.
TAB
LE i
(con
tinue
d)(d
olla
rs in
thou
sand
s)A
VER
AG
E V
ALU
E A
DD
EDB
Y M
AN
UFA
CTU
RE
AV
ERA
GE
VA
LUE
AD
DED
PER
PRO
DU
CTI
ON
WO
RK
ER
Per C
ent o
f Tot
alIn
Cor
pora
tein
Cor
pora
teV
alue
Add
ed b
yM
ultiu
nit
Mul
tiuni
tM
anuf
actu
rers
Esta
blis
hmen
tsa
in A
ll O
ther
sEs
tabl
ishm
ents
1939
1947
1939
1947
1939
1947
939
947
57.7
59.4
$ 21
8$
66g
$ 43
$115
$ 4.
7$
9.0
87.7
89.9
2,58
12,
720
6775
4.8
7.1
59.5
55.7
564
1,71
016
336
81.
74.
724
.820
.624
054
855
120
i.6.g
Indu
stry
Gro
upFo
od a
nd k
indr
ed p
rodu
cts
Toba
cco
man
utac
ture
sbTe
xtile
mill
pro
duct
sA
ppar
el a
nd re
late
d pr
oduc
tsLu
mbe
r and
lum
ber p
rodu
cts
(exc
ept f
urni
ture
)eFu
rnitu
re a
nd fi
xtur
esc
Pape
r and
alli
ed p
rodu
cts
Prin
ting
and
publ
ishi
ng in
dust
ries
Che
mic
als a
nd a
llied
pro
duct
sPe
trole
um a
nd c
oal p
rodu
cts
Rub
ber p
rodu
cts
Leat
her a
nd le
athe
r pro
duct
sSt
one,
cla
y, a
nd g
lass
pro
duct
sPr
imar
y m
etal
indu
strie
sdFa
bric
ated
met
al p
rodu
ctsc
Mac
hine
ry (e
xcep
t ele
ctric
al)
Elec
trica
l mac
hine
ryTr
ansp
orta
tion
equi
pmen
tIn
stru
men
ts a
nd re
late
d pr
oduc
tsc
Mis
cella
neou
s man
ufac
ture
sc
Tota
l
In A
ll O
ther
s19
3919
47
$4.0 1.
33.
01.
74.
61.
94.
8T1
42.0
35.4
146
407
3957
i.84.
930
.530
.128
790
466
533
2.5
5.6
70.5
72.7
481
1469
ia8
293
3.5
7.8
29.3
33.4
423
i,86
5310
26.
311
.1
77.4
74.8
477
1405
6g18
76.
511
.792
.683
497
32,
738
144
432
6.4
si.6
72.8
75.6
1,83
554
7425
545
73.
56.
i49
.943
.246
41,
024
102
187
i.84.
367
.163
.033
395
658
843.
76.
462
.578
.164
63,
570
107
308
2.9
5.8
45.7
474
469
1,53
088
170
3.2
6.o
50.8
514
799
2,57
811
723
23.
76.
i72
.575
.31,
643
3,31
517
531
14.
26.
383
.28i
.2,
646
7,75
819
655
53.
36.
o57
.860
496
22,
371
134
184
4.6
6.2
a8.o
37.0
aS'
780
4810
02.
5.8
59.4
58.2
$49
7$1
464
$65
$147
$3.
4$
6.6
1.7
2.2
2.9
5.1
6.6
6.7 3.0
1.7
2.5
2.1
3.0
3.9
3.2 2.8
3.4
2.5
$ 2.
2
5.9
4.6
6.4
9.2
so.8 5.9
4.5
4.8
5.5
6.o
5.6
5.7
5.0
$.8
VERTICAL INTEGRATION
n.a. not available Notes to Table ia Production workers" and "multiunit" are designations used in 1947. They
are treated as equivalent to the "wage earners" and "controlled by central office"designations used in ig.
Multiunit establishments include only those owned by corporations. Thoseowned by individuals and partnerships have been classified with "all others,"because they resemble the latter much more closely. In 1947 two industry groups."electrical machinery" and "miscellaneous manufactures," include under multi-unit all types of organization in order to avoid disclosure. In both groups, how-ever, the distortion is negligible.
b This includes in 1947 the designation "tobacco stemming and redrying,"which was not included in 1939 and for which no içjg figures are available. Ofs.o86 firms in "tobacco manufactures," 163 were in this subgroup; therefore thedata for the two years are not completely comparable.
C These industry groups were reclassified from 1939 to 1947, and as a conse-quence a retabulation of 1939 data was necessary. In a few cases figures for theindividual industry groups were not available. The error introduced on accountof this difficulty is, however, less than io per cent.
d These industry groups were reclassified from 1939 to 1947. Because of thedisclosure rule, data for certain of the industries for 5939 were not available;hence the total given here is incomplete and not comparable with the 1947 data.Where the column entries are ratios rather than absolute numbers, the 1939entries can be considered as a sample, although probably not a representativeone, of the 1939 industry group.
Source: Census of Manufactures: 1939. Bureau of the Census, Vol. s, Chap. v;Census of Manufactures: 1947. Vol. s, Chap. iv.
Table 4 is based on the complete universe of manufacturing cor-porations, and the trend which is indicated in Tables 2 and 3 ismuch more striking here. Table 4 shows that in every case there is astrong trend toward higher ratios between the value of inventoriesand sales as the size of the firm increases. But for total manufactur-ing and for fifteen out of twenty industry groups in 1949,15 firms withthe largest inventories had a lower ratio than the group immediatelypreceding. Possibly this might be explained by the correlation ofLIFO (last-in, first-out) accounting (which would understate in-ventories) with size of firm.1° But the same tendency is observable inthirteen industry groups for 1940, despite the artificial equalizationof firm size. Assuming, at least for the sake of the argument, that amore normal period would show the same phenomenon, how canthis be interpreted? One explanation might be that, where the mar-keting of the product becomes an important consideration, the ad-vantages of carrying a full line impel the larger firms not only toprocess to completion but also to buy semifinished products and to
15 This is true also for 1948, although the data for that year have beenomitted.
i6J, Keith Butters, Effects of Taxation: Inventory Accounting and Policies(Harvard University Press, 1949), Chap. II.
292
TAB
LE 2
Rat
io o
f Cor
pora
te In
com
e to
Cor
pora
te S
ales
for i
8 La
rge
Man
ufac
turin
g C
orpo
ratio
ns, 1
949
(dollars
in m
illio
ns; r
atio
s in
per c
ent)
INC
OM
E
CO
DE
NA
ME
OF
FIR
M
Wag
esPr
ofits
and
befo
reIn
tere
stSa
les
Sala
ries
Taxe
sPa
idFo
od a
nd K
indr
ed P
rodu
cts
RA
TIO
OF
INC
OM
ETO
SA
LES
Tota
lEx-
Tota
l In-
Ex-
In-
clud
ing
clud
ing
clud
ing
clud
ing
Dep
re-
Dep
re.
Dep
re.
Dep
re-
Dep
re-
ciat
ion
ciat
ion
ciat
ion
ciat
ion
ciat
ion
201
Swift
and
Com
pany
$2,213
$249
$46
$ 3
$t3
$298
$311
134
14.1
201
Arm
our a
nd C
o.1,
848
203
26
8211
219
114
11.9
201
201
Wils
on a
nd C
o.C
udah
y Pa
ckin
g70
955
975 49
7...
7i 1
2
83 4386 45
si.6 7.6
12.1
8.i
201
John
Mor
rell
293
30
a1
131
32
10.5
10.9
202
Nat
iona
l Dai
ry8g
B15
757
215
216
231
24.0
25.7
202
Borden Co.
614
110
35
110
146
156
23.7
25.4
202
Beatrice Foods
192
26
81
3435
17.8
18.2
202
Pet tilkb
138
19
7a
226
28
i8.8
20.3
203
204
California Packing Co.
General Mills
169
410
45 .
s6
i8
a a4
62
62
66
65
36.6
15.1
39.1
15.9
204
Cor
n Pr
oduc
ts R
efin
ing
Co.
145
2623
a49
5233
435.9
205
Nat
iona
l Bis
cuitb
agb
101
39a
614
014
647
.349
.3205
Suns
hine
Bis
cuit
ioi
27a
139
4039
.039
.6205
Con
tinen
tal B
akin
g15
451
9a
26o
6239
.140
.320
6A
mer
ican
Sug
ar R
efin
ing
277
2313
a2
3739
13.2
14.1
206
Nat
iona
l Sug
ar R
efin
ing
132
84
aa
1313
9.5
g.8
206
Am
eric
an C
ryst
al S
ugar
Co.
326
19
25.1
28.1
208
208
208
Sche
nley
Indu
strie
sB
row
n Fo
rman
Dis
tribu
tors
Anh
euse
r.Bus
ch
201 27 135
48 4 30
40 7 24
3 a a
4 1 4
92 11 53
g6 12 57
45.6
42.4
39.4
47.8
44.4
42.2
209
Gen
eral
Foo
ds47
569
451
5114
119
24.1
25.1
TOTA
L SA
MPL
E$s
o,os
8$1
,400
$409
$19
$91
$i,8a8
$1,9
1918
.219
.2
Toba
ccoc
2111
2111
P. Lorillard
Philip Morrisd
$ 85 13
1$
8 11$1
1 25$1 2
$1 1
$20
38
$21 39
234
29.3
24.7
29.8
Tota
l Sam
ple
$216
$ig
$36
$$2
$58
$6o
26.9
27.8
O z I
See page 301
forfootnotes.
TAB
LE 2
(con
tinue
d) INC
OM
E
RA
TIO
OF
INC
OM
ET
OS
ALE
S
242
Wey
erha
euse
r Tim
ber C
o.24
2Lo
ng B
ell L
umbe
r$156
$i
$gn.
a.$11
$88
$99
Si
20
11
a31
35
$237
$69
$50
$15
$119
$i4
a$i
$48
$i
$48
261
International
Pape
r26
1Crown
Zelle
rbac
h26
1St.
Reg
is P
aper
e26
1Champion
Pape
r and
Fib
re
$416
$94
i68
4514
433
8i27
$89
n.a.
$15
$s83
a6
8o
24
$i
458
s6
a44
$igS
44.1
47.6
86
474
51.2
6s40
.243.1
47
54.3
58.0
221
221
222
223
227
227
227
Tota
lEx-
Tota
l In-
Ex-
In-
clud
ing
clud
ing
clud
ing
clud
ing
IJep
re-
Dep
re-
Dep
re-
Dep
re-
ciat
ion
ciat
iOn
ciat
ion
ciat
ion
CO
DE
NA
ME
OF
FIR
MSa
les
Wag
esPr
o/Its
and
befo
reSa
larie
sTa
xes
Inte
rest
Paid
Dep
re-
ciat
ion
Text
iles
J. P.
Stevens Co.
Pacific
Mill
s$278
99
$ 71 27
$35 6
n.a. a
$ 4 2
$io6 33
$iio 35
Wes
t Poi
nt M
anuf
actu
ring
8423
10a
i33
34D
an R
iver
Mill
s65
26
5n.
a.2
3133
Big
elow
San
ford
Car
pet C
o.A
lexa
nder
Sm
ithA
rmst
rong
Cor
k
67 70 163
26 29 54
5 3 17n.
a.n.
a.
1 1 5
31 31 71
32 32 76
Tota
l sam
ple
$826
$256
$81
$16
$336
$aLu
mbe
r
Tota
l Sam
ple
251
Sim
mon
s Co.
Tota
l Sam
ple
38.1
33.6
39.5
47.5
45.5
44.2
43.4
40.7
56.3
38.5
50.2
39.6
35.4
40.5
50.8
47.8
45.7
46.6
42.6
63.5
43.2
Furn
iture
$iio
$38
$g
$110
$8$9
Pape
r
$49
$49
43.2
44.5
43.2
44.5
261
Wes
tV
irgin
ia P
ulp
and
Pape
r26
1R
ayon
ier
264
Mar
atho
n C
orp.
267
Con
tain
er C
orp.
of A
mer
ica
8827
14
a42
4747
.153.4
4955
ga
2528
51.5
57.1
6o
237
a30
33
50.3
55.0
115
2415
n.a.
338
4133
.435
.7
$1,121
$288
$208
Chemicals
$42
$500
$542
$1,0
25 200
364
j66
121
148 90 53 38 48 6o 33 51
$335 61 86 45 38 35 27 14
16
17
11 8 8
Tota
l Sam
ple
282
F.
I. du
Pont
282
Dow
Che
mic
al28
2Allied
Che
mic
al28
2Monsanto
282
Hercules
Pow
der
282
Am
eric
anH
ome
Prod
ucts
282
Air
Red
uctio
n28
2International
Min
eral
s28
2Liquid
Car
boni
c28
2Diamond
Alk
ali
282
Virginia-Carolina
282
Commercial
Solv
ents
282
U.S.
Indu
stria
l Che
mic
als
283
Abb
ott L
abor
ator
iesb
283
Mer
ck a
nd C
o.28
3E.
R. S
quib
b28
3RexalI
Dru
gb28
4C
olga
te P
alm
oliv
e28
9 Ea
stm
an K
odak
289
Gen
eral
Ani
line
and
Film
289
Am
eric
an V
isco
se28
9C
elan
ese
Cor
p.
Tota
l Sam
ple
$334 41 62 28 17
19
11 7 2 5 6 5
$1
n.a.
$57
$669
$220
104
n.a.
17147
19
74n.
a.6
55a
254
14
39a
21a
2
a3
22a
i8n.
a.1
13
a1
10
44.6
48.3
65.2
70.1
51.9
62.0
40.5
45.1
44.5
50.0
45.1
50.4
36.5
37.8
43.0
47.8
40.1
45.3
47.4
52.6
45.3
52.1
29.3
33.3
40.1
42.4
19.5
21.6
$726
124
164
83 6i 56 43 24 20 25 20 14
11
-I z '-I Ni
C) -I C 2
7420
iga
139
4052
.654
.174
22ii
na.
334
3746
.150.0
82
i812
a1
3031
36.3
37.8
154
64
12
4143
26.5
27.9
204
34
i6na
.2
5052
24.5
25.5
396
187
76
a17
263
280
66.3
70.1
8o33
51
38
4147
.551
.3195
71
34
n.a.
ii10
411
553
.559
.017
16i
343
11gS
rog
57.1
63.7
$3,8
27$1,183
$749
$g$1
78$1,941
$2,1
1950
.755
.4
TAB
LE 2
(con
tinue
d)R
ATI
O O
FIN
CO
ME
INC
OM
ETO
SA
LES
Tota
lEx-
Tot
al In
-Ex
-In
-W
ages
Prof
itscl
udin
gcl
udin
gcl
udin
gcl
udin
gan
dbe
fore
Inte
rest
Dep
re.
Dep
re-
Dep
re-
Dep
re-
Dep
re-
CO
DE
NA
MEOF
FIR
MSa
les
Sala
ries
Taxe
sPa
idci
atio
nci
atio
nci
atio
nci
atio
nci
atio
nPe
trole
um a
nd C
oalt
2911
Stan
dard
Oil (N.J.)
$2,892
$537
$442
$10
$184
$g88
$1,172
34.1
37.1
2911
Stan
dard
Oil
(md.
)1,
158
219
139
744
364
408
31.4
35.2
2911
Soco
nyVacuum
1,227
228
,i65
4234
9391
28.4
31.9
2911
Tex
asCompany
1,077
,68
158
559
331
390
30.7
36.2
2911
Stan
dard
Oil (Calif.)
743
131
i8i
376
315
391
42.4
52.6
2911
Gul
f970
i68
113
552
s86
29.5
34.8
2911
Shel
l8i6
159
102
348
264
312
32.3
38.2
2911
Sinc
lair
584
86
74
436
165
201
28.1
34.4
2911
Phill
ips
486
846o
330
147
177
30.2
36.4
02911
Sun
462
g831
a21
129
150
27.8
32.5
2911
Atla
ntic
Refining
446
78
33
127
112
139
25.1
31.2
2911
Tide
wat
er355
5j
34a
2286
io8
24.0
30.4
2911
Con
tinen
talOil
315
3446
ai8
8i99
25.6
314
2911
Stan
dard
Oil (Ohio)
255
4122
112
6577
25.3
30.2
2911
Uni
onOil
(Calif.)
200
3422
225
588
29.1
41.5
2911
Ske
Ily164
i632
a13
4962
29.8
37.8
2911
Ric
hfie
ld12
32O
g25
19
4655
37.0
44.7
2911
Lion
669
121
422
2633
.339
.429
9K
oppe
rsCo.
191
5012
15
6368
33.0
35.6
299
Flin
tkot
eCo.
68
20
a2
2931
42.9
45.6
Tota
l sam
ple
$12,
598
$2,2
31$1
,663
$52
$729
$3,9
49$4
,678
31.3
37.1
oii
Goodyear
Tire
3011
U.S.
Rub
ber
301
iFirestone
Tire
3011
General
Tire
and
Rub
ber
Tota
l sam
ple
314
International
Shoe
Tota
l sam
ple
t'3
322
Pittsburgh
Plat
e G
lass
322
Libb
ey O
wen
s32
2Anchor
Hoc
king
3241
Lone
Star
Cem
ent
3241
Lehigh
Portl
and
Cem
ent
3272
U.S.
Gyp
sum
3272
National
Gyp
sum
3292
Johns
Man
ville
329
Minnesota
Min
ing
329
Harbison.Walker
329
Car
boru
ridum
Co.
Tota
l sam
ple
Rub
bir
$634
$190
$42
$4$2
0$2
37$2
5751
818
423
313
210
223
8o
167
352
15205
220
93
292
a2
3133
$i,8
a$7
o$102
$9
$5o
$683
$733
$igg
$77
.519
Leat
her
$199
$77
$19
n.a.
37.3
404
35.3
34.0
374
40.1
43.1
37.9
35.5
40.2
$a$g
7$9
948
.749
.7
$2$g
7.5
9948
.749
.7St
one,
Cla
y, a
nd G
lass
$281
i4 71
$88 46 27
$g 8 8
n.a.
n.a. a
$7 6 2
$147
84 35
$154 90
37
52.2
624
49.0
54.8
67.2
52.1
tzl
64 4114
12
iS
10
n.a.
n.a.
2 2
30
23
32
25
47.0
55.2
50.0
6i.o
138
2635
n.a.
56i
6644
.147
.8Z
59i6
a2
2628
44.2
47.5
163
6i22
a8
8851
.154
.011
532
24n.
a.5
66i
49.0
53.0
417
8n.
a.2
2527
57.9
62.8
3919
1a
121
2253
.856
.4
$1,148
$358
$230
$39
$591
$630
51.5
54.9
TAB
LE 2
(con
tinue
d) INC
OM
ER
ATI
O O
F IN
CO
ME
TO S
ALE
S
Prim
ary
Iron
and
Ste
el
Wag
esan
dSa
les
Sala
ries
Prof
itsbe
fore
inte
rest
Taxe
sPa
id
Tota
lEx-
Tota
l In-
Ex-
In-
clud
ing
clud
ing
clud
ing
elud
ing
Dep
re-
Dep
re-
Dep
re-
Dep
re-
Dep
re-
ciat
ion
ciat
ion
ciat
ion
ciat
ion
ciat
ion
CO
DE
NA
MEO
F FI
RM
i U.S
.St
eel
331
Bet
hleh
em S
teel
i Rep
ublic
Ste
eli J
ones
and
Lau
ghlin
i Nat
iona
l Ste
el33
1Y
oung
stow
n Sh
eet a
nd T
ube
331
Arm
co S
teel
331
Inla
nd S
teel
331
Whe
elin
g St
eel C
orp.
331
Col
orad
o Fu
el a
nd Ir
on33
1C
ruci
ble
331
Alle
ghen
y Lu
dlum
Ste
el C
orp.
331
Pittsburgh
Shar
on S
teel
Cor
p.33
1Lu
kens
331
Gra
nite
City
331
Alan
Woo
d33
1C
oppe
rwel
d33
1In
terla
ke Ir
on33
1C
ontin
enta
l33
1R
otar
y El
ectri
c33
1D
etro
it St
eel
Tota
l sam
ple
332
Alu
min
um C
o. o
f Am
eric
a33
2C
alum
et a
nd H
ecla
b33
2Sc
ovill
Man
ufac
turin
g
Tota
l sam
ple
$2,2
931,
267
652
386
425
335
341
346
143
138 99 io6 8o 90 46 36 43 57 23 17 28
$946
$295
$2$7
$1,2
4347
717
15
3365
420
78i
215
291
147
352
17
184
105
77i
i18
385
521
1013
799
512
1015
288
412
1013
18
141
673
51i6
a3
6843
21
345
403
n.a.
243
331
a2
3529
a2
3417
41
21
105
115
11a
215
10a
i8
10
n.a.
2i8
72
n.a.
19
2a
a
35
n.a.
18
$1,3
40 687
306
201
196
147
162
141 79 71 48 45 37 22 16 17 14 20 10 5 9
71 z -I tTJ C)
71 '-I 0 z
54.2
51.6
44.5
47.6
42.9
40.9
44.4
37.9
o.6
48.9
454
41.0
43.4
38.3
37.7
32.1
41.0
29.5
19.3
41.1
30.6
28.7
48.2
52.9
46.4
49.9
52.4
58.4
54.2
46.9
52.1
46.1
45.9
47.5
40.8
55.2
51.4
48.5
42.5
46.3
40.0
39.3
34.8
47.2
32.6
35.1
43.5
29.4
32.1
51.5
57.6
51.1
54.1
56.8
$7,o
o7$2
477
$879
$19
$232
$,77
$3,6
09
Prim
ary
Non
ferr
ous M
etal
s$3
47$1
43$4
1n.
a.$i6
$184
$200
4714
a2
22
24
74
344
n.a.
37
40
$468
$191
$52
$21
$245
$266
341
iA
mer
ican
Can
3411
Con
tinen
tal
Can
343
Am
eric
an R
adia
tor
Am
eric
an S
teel
Fou
ndrie
s49
Yal
e an
d To
wne
Tota
l sam
ple
Fabr
icat
ed M
etal
s$4
68$112
$48
a$i
i$i
6o$1
7134
.236
.5
336
86
19
$27
107
114
32.0
33.9
49.5
52.0
474
206
70
29
a3
9910
247.7
7527
11na
.1
3839
50.4
5724
2a
a26
2745
.8
Mac
hine
ry (e
xcep
t Ele
ctric
al)
Elec
trica
l Mac
hine
ry
$4,191
$1,655
$451
$14
$1,142
$319
$109
$2
$23
$430
$453
37.7
39.7
352
Inte
rnat
iona
l Har
vest
er$g
o$3
25$g
4$i
$ig
$420
$41g
46.2
352
Dee
re a
nd C
o.36
291
70s
6162
s68
44.8
352
Alli
s-C
halm
ers
351
100
331
313
413
738.1
357
Nat
iona
l Cas
h R
egis
ter
167
io6
23
i4
129
133
77.6
352
Cat
erpi
llar T
ract
or25
577
301
5io
8ii
42.2
357
Rem
ingt
on R
and
148
8217
25
100
105
67.4
352
J. I.
Cas
ei6
4530
a3
7578
48.2
c35
735
635
2
Bur
roug
hs A
ddin
g M
achi
neLi
nk B
elt
Oliv
er C
orp.
82
88
sos
54 35 30
11 13
10
n.a.
n.a. a
1 1 3
65 47 41
66 48 44
78.8
53.5
40.2
355'
Food
Mac
hine
ry a
nd C
hem
ical
8226
7a
3336
39.9
356
Fairb
anks
Mor
se78
27
31
3132
40.3
Dre
sser
Indu
strie
s8i
21
6a
228
30
34.2
353
Buc
yrus
-Erie
62i6
iin.
a.1
2728
43.4
Tota
l sam
ple
$2,9
24$1
,035
$358
$7$'
j$1
400
47.9
361
Gen
eral
Ele
ctric
$1,6
14$6
50$2
04$
$47
$858
$905
53.1
361
Wes
tingh
ouse
946
376
111
414
490
504
51.8
366
Wes
tern
Ele
ctric
858
348
6g2
1742
0437
48.9
366
Rad
io C
orp.
of A
mer
ica
396
141
421
7184
191
46.4
366
Elec
tric
Aut
oLite
218
75i8
14
g4
g8
366
Sylv
ania
Ele
ctric
103
39
51
245
47
43.6
66R
ayth
eon
Man
ufac
turin
g56
262
aj
2829
50.2
z
48.3
46.4
390
79.6
44-3
70.9
50.0
8o.
54-5
43.6
43.9
41.0
37.0
45.2
49.8
56.1
53.3
50.9
48.2
45.0
45.6
51.8
Tota
l sam
ple
$92
$2,119
$2,211
5o.6
52.8
TAB
LE 2
(con
tinue
d) INC
OM
ER
ATI
O O
F IN
CO
ME
TO S
ALE
S
Tran
spor
tatio
n Eq
uipm
ent
Tota
lEx-
Tot
al In
-Lx
-In
-cl
udin
gD
epre
-ci
atio
n
Wag
esPr
ofits
clud
ing
clud
ing
clud
ing
and
befo
reIn
tere
stD
epre
-D
epre
-D
epre
-D
epre
.C
OD
EN
AM
E O
F FI
RM
Sale
sSa
larie
sTa
xes
Paid
ciat
ion
ciat
ion
ciat
ion
ciat
ion
371
Gen
eral
Mot
ors
$5,7
0137
1C
hrys
ler
2,08
537
1St
udeb
aker
h503
371
Nas
h-K
elvi
nato
r36
4371
Packard
215
371
Willys-Overland
142
7iFr
ueha
uf T
railo
r78
571
Mac
k Tr
ucks
7837
1K
aise
r-Fr
azer
104
371
Brig
gs M
anuf
actu
ring
338
371
The
Bud
d C
ompa
ny26
757
137
1B
org-
War
ner
Thom
pson
Pro
duct
s25
2io
837
1Ea
ton
Man
ufac
turin
g102
371
Hou
daill
e H
ersh
ey37
1St
ewar
t War
ner
55
$1,441
$1,125
$9$110
$2,574
$2,6
8445
.147
.1349
213
a19
563
582
27.0
27.9
110
23
13
133
i626
.327.0
8645
13
132
135
36.3
57.1
40
13
n.a.
355
625.0
26.3
345
n.a.
239
41
27.3
28.9
22
41
127
2834
-535
.929
—6
223
2529
.932
.141
—39
211
4314
41.3
13.5
123
22n.
a.5
146
151
43.0
44.7
8a26
15
108
113
40.6
42.3
686
a10
310
741
.042
.541
a2
5052
46.9
48.1
2713
n.a.
140
4139
.540
.223
a1
2627
49.4
50.1
224
n.a.
126
2746
.749
.130
713
524
a1
159
i6o
51.7
52.1
183
79i8
n.a.
397
100
55.0
54.6
124
6g12
a81
82
65.0
66.i
117
68
'1a
2g
8,67
.169.2
ii8
67
a6g
7059
.059.3
6o25
7n.
a.31
3151
451
.7
7835
8n.
a.1
4344
55.3
564
286
6i
n.a.
276
7826
427
.314
739
ga
248
5032
.554
.011
96
a1
4142
34.4
35.5
76o
i6n.
a.1
4647
6o6
6,.8
92o
6n.
a.3
5740
39.8
45.5
$12.
15o
$3,2
07$i
,6$1
5$1
91$4
,853
$5,0
4439
.941
.5
I37
2B
oein
g A
irpla
neb
372
Ben
dix
Avi
atio
n37
2N
orth
Am
eric
an A
viat
ion
372
Douglas
Airc
raft
572
Lockheed
572
Gru
mm
an57
3N
ewpo
rt N
ews S
hipb
uild
ing
and
Dry
dock
574
Pullm
an, m
ci37
4 A
mer
ican
Loc
omot
ive
374
Bal
dwin
Loc
omot
ive
574
Wes
tingh
ouse
Air
Bra
ke57
4 A
mer
ican
Bra
ke S
hoe
Tota
l sam
ple
Inst
rum
ents
3821
The
Sper
ry C
o.$i
i6$6
3$8
a$2
$72
$75
61.9
64.7
584
John
son
and
John
son
i640
i6n.
a.2
5557
40.6
41.9
Tot
alsa
mpl
e$2
52$i
o$2
4$4
$128
$132
50.8
524
Mis
cella
neou
s39
83D
iam
ond
Mat
ch$7
6$s
$a
$i$2
9$3
037
.739
.539
89R
emin
gton
Arm
s8
17n.
a.1
2021
53.3
55.3
Tota
l sam
ple
$114
$40
$8$2
$g$'
43.0
44.7
n.a.
=no
tav
aila
ble.
a Le
ss th
an $
500,
000.
b19
50 d
ata.
sale
s fig
ures
are
exc
lusi
ve o
f fed
eral
exc
ise
taxe
s.d
Fisc
al y
ear e
ndin
g M
arch
31,
195
0.el
g47d
ata.
C C
orpo
rate
sale
s fig
ures
are
exc
lusi
ve o
f fed
eral
and
stat
e ta
xes o
n ga
solin
e an
d lu
bric
ants
.z
gEst
imat
ed.
h 19
51 d
ata.
I igj
8 da
ta.
Not
e: T
he in
dust
ry c
ode
num
bers
app
roxi
mat
e th
ose
used
in th
e C
ensu
s of M
anuf
actu
res,
Bur
eau
of th
e C
ensu
s. To
tal i
ncom
e is
com
pute
d fr
om u
nrou
nded
dat
a so
that
subt
otal
s will
not
nec
essa
rily
add
to to
tals
.So
urce
: Ann
ual r
epor
ts.
VERTICAL INTEGRATION
TABLE 3ARatio of all Corporate Income to all Corporate Sales by Manufacturing Industry Groups, 1949
(dollars in millions; ratios in per cent)Profits Ratio of
Wages and before Interest Total IncomeIndustry Group Salaries Taxes Paid Income Sales to Sales
Food and kindred products $4,1o3 $i,600 $71.2 $,77 $36,167 i6.oTobacco manufactures 218 250 20.6 488 1,714 28.5Textile mill products 3,134 596 31.6 3,761 10,602 35.5Apparel and related products 2,086 142 7.4 2,235 7,896 28.3Lumber and lumber products
(except furniture) 970 228 7.6 1,204 3,061 39.4Furniture and fixtures 1,218 97 6.5 1,320 3,082 42.8Paper and allied products 1,496 547 22.6 2,066 5,301 39.0Printing and publishing 2444 249 13.3 2,707 6,067 44.6Chemicals and allied products 2,504 1475 38.7 4,017 13,355 30.0Petroleum and coal products 2,041 2446 126.8 4,614 18,450 25.0Rubber products 785 s8i bA 977 3,088 31.6Leather and leather products 900 83 5.6 988 2,750 35.9Stone, clay, and glass products 1,421 522 8.7 1,952 3,917 49.8Iron and steel and their products 6,123 2,042 831 8,253 19,921 41.4Nonferrous metals and their
products 1,604 477 a8. 2,109 5.587 37.8Machinery (except electrical) 4,635 1,305 21.9 5,962 13,139 45.4Electrical machinery 2,495 629 i6.6 3,140 8,466 37.1Transportation equipment 4426 2,199 i6.o 6,642 18,963 85.0Miscellaneous manufactures 1,366 290 18.3 1,673 3,229 51.8
Source: Census of Mineral Industries: 1939, Bureau of the Census; Statistics ofInca me for 1946, Bureau of Internal Revenue; Quarterly Industrial Financial Re-port Series for 1946 and 1949, Federal Trade Commission and Securities and Ex-change Commission; Census of Manufactures: 1947, Bureau of the Census; Na-tional Income Supplement, 195!, Survey of Current Business, Department ofCommerce.
Derivation of the data for mixed mining-manufacturing industries: The indus-try group "Iron and steel and their products" includes the iron-ore mining industryand part of the bituminous-coal mining industry, while "Nonferrous metals andtheir products" includes nonferrous metal mining. The data for the narrowerdefinitions of these industry groups, excluding the mining industries, were ob-tained in the following manner: (i) Wages and salaries for 1949 were taken fromthe National Income Supplement, 1951, Survey of Current Business. (2) To thisfigure was applied the ratio of corporate to total wages and salaries, derived froma previous amalgamation of figures for the individual industries of Census ofManufactures: 1947, into those for industry groups corresponding to "Iron andsteel and their products" and "Nonferrous metals and their products." In 1947,97 per cent of all wages and salaries paid in the iron and steel industry and 95per cent in nonferrous metals were paid by corporations. It was assumed thatthere had been no change up to 1949, and the National Income Supplement, 1951total industry group figures were so divided. () Profits before taxes were alsotaken from the National Income Supplement, 195!, and adjusted upward a percent to compensate for the discrepancy between figures in the National IncomeSupplement, xi and in the Quarterly Industrial Financial Report Series, the
302
VERTICAL INTEGRATION
latter being used for most of the other industry groups. () Interest paid in 1949was derived as follows: Interest paid in 1946 was obtained from the Statistics ofIncome for 1946 and inflated according to the percentage change in interest-bearing liabilities, obtained from a sample of published reports, from 1946 to 1949.
To the figures obtained for the narrow industry definitions for iron and steeland their products and for nonferrous metals and their products were added thosefor appropriate portions of the mining industries. The procedures used for com-puting figures for metal mining and bituminous coal mining were similar tothose outlined above, with the exception of the second step. An estimate of theapproximate division between corporate and noncorporate shares of wages andsalaries was made, using the Census of Mineral Industries: 1939. This is the mostrecent source of any figures or statistics of this kind, although at some futuredate a new census of mining will be published in conjunction with the 1950census. Corporate firms paid 94 per cent of the total payrolls in bituminous coalmining and 95 per cent in metal mining. Since there were no later figures orindications that these percentages had been substantially altered, they wereapplied to the 1949 data in the National Income Supplement, 1951.
Metal mining data were divided between iron and steel and their products andnonferrous metals and their products on the basis of the proportion of the wagesand salaries paid by iron mines to the total industry wages and salaries. Thus, 30per cent of industry wages was paid in iron and steel and 70 per cent in non-ferrous metals. Approximately 19 per cent of wages paid in bituminous coalmining was assumed to be paid in the iron and steel industry on the basis of theratio of coal production to coal used in iron and steel production in 1949.
The figures for the industry group "Petroleum and coal products" were de-rived in the same way as the figures for the majority of the industrial groups.However, the figures for crude petroleum and pipeline transportation industrieswere added to this group. The data for the crude petroleum and natural gasindustry were obtained in the same manner as those for iron and steel and non-ferrous metals, again with the exception of the second step of the procedure.Corporate firms paid 88 per cent of total wages and salaries. For pipelines thefigure was an estimate based on data in Petroleum Facts and Figures, £950, Amer-ican Petroleum Institute, and the 1939 census; corporations paid about 94 percent of the total. The classification of natural gas was removed from the crudepetroleum and natural gas industry and natural gas transmission from pipelinesby using the ratio of gas wells to oil wells and to oil and gas wells in 1939, asgiven by the Census of Mineral Industries: £939. The resulting figures werechecked by later figures given in Petroleum Facts and Figures, 1950. Thus, 94 percent of wages paid in each industry was credited to the industry group "Petro-leum and coal products."
assemble on a large scale. This is particularly true of the automobilecompanies; indeed, the most striking decline of the ratio in the topsize classes, both in 1940 and in 1949, is observable in the auto-mobile group.
However, if we are too cautious to accept the idea of an actualdecline in the degree of vertical integration as one approaches thetop size class, it does seem clear that we cannot speak of any ob-servable increase. It may not be too farfetched to say that, even ifsuch a trend toward a decline existed, we would not know it. Afterall, there are only about 140 manufacturing firms with total assets
303
VERTICAL INTEGRATION
TABLE BIllustration of the Derivation of Corporate Income and Sales
by Manufacturing Industry Groups, 1949(dollars in millions)
Food and Kindred ProductsOperation and source Datai. Total wages and salaries, 1949, Department of Commerce, Na-
tional Income Supplement, 1951 $4,632.002. Total wages and salaries, 1947, Census of Manufactures 3,789.003. Corporate wages and salaries, 1947, Census of Manufactures 3,357.004. Line divided by line 2 88.60%
5. Line 4 multiplied by line s: estimated corporate wages $4,103.006. Profits before taxes, 1949, Quarterly Industrial Financial Report
Series, ip.p 1 ,6oo.oo7. Corporate interest paid, 1946, Statistics of Income, 1946 59.308. Long- and short-term indebtedness, 1st quarter 1947, Quarterly
industrial Financial Report Series, iv 1,708.00
9. Long- and short-term indebtedness, 4th quarter 1949, Quarterly
Industrial Financial Report Series, 1949 2,0920010. Line 9 divided by line 8 1.2%ii. Line 7 multiplied by line io: estimated interest paid 1949 $ 71.2012. Income originating in corporate business: lines plus 6 plus ii 5,774.0013. Corporate sales, 1949, Department of Commerce, National In-
come Supplement, 'p5' 36,167.0014. Ratio of income to sales: line 12 divided by line 13 .1597%
Source: Statistics of Income for 1946, Bureau of Internal Revenue; QuarterlyIndustrial Financial Report Series, 1946 and '949, Federal Trade Commissionand Securities and Exchange Commission; Census of Manufactures: 1947, Bureauof the Census; National Income Supplement, 1951, Suroey of Current Business,Department of Commerce.
TABLE 4Ratio of Value of Inventories to Gross Sales by Manufacturing Industry Groups
and Asset Size Classes, 1940 and 1949
Total Food and KindredAsset Size Class Manufacturing Products Beverages
(thousands of dollars) 7940 1949 1940 1949 1940 1949
I. Under $o .007 .075 o5o .042 .o8o o94
II. $o.$ioo .104 .094 .064 .o78 .097IlL 100-250 J24 .105 078 .o6 093 091
IV. 250-500 .147 .ii6 .089 .065 .109 101
V. 500-1000 .164 .127 .099 .070 41 .113
VI. 1000-5000 .191 .148 .123 o83 .128 .122VII. 5000-10,000 .216 .167 .145 .107 .250 .146
VIII. 10,000-50,000 .224 .175 .i6o .102 .267 .222IX. 50,000-100,000 .260 .173 .221 .119 .417 .ig6
X. Over $ioo,ooo .194 .157 .i8i .094 .224
Total .189 .152 .ii6 .091 .177 .177
304
VERTICAL INTEGRATION
TABLE 4 (continued)Tobacco Textile Mill Apparel and
Asset Size Class Manufactures Products Related Products(thousands of dollars) 1940 5949 1940 1949 1940 1949
I. Under $50 .ii6 .148 .075 .064 .056 .055II. $5o-$loo .162 .197 .114 .094 .o8 .085
III. 100-250 .220 .191 .137 .113 .104 .103
IV. 250-500 .195 .199 .165 .126 .146 .121
V. 500-1000 .300 .238 .189 .i8 .151 .i38
VI. 1000.5000 .347 408 .219 .163 .206 .157VII. 5000.10,000 .238 .286 .272 .184 .295 .192
VIII. 10.000-50,000 .521 409 .327 .223 .247 .193
IX. 50,000-100,000 .433 .486 .355 .234 .249
X. Over $ioo,ooo .424 .490 .184
Total .421 .502 .233 .179 .129 .131
Lumber and FurnitureLumber Products and Finished Paper and
Asset Size Class (except furniture) Lumber Products Allied Products(thousands of dollars) 1940 1949 1940 5949 5940 1949
I. Under $o .098 .o8i .117 .099 .102 .075
II. $50-$100 .140 .io8 .159 .126 .114 .og6
III. 100-250 .170 .125 .176 .126 .125 .103
IV. 250-500 .191 .144 .195 .143 .141 .109
V. 500-1000 .229 .157 .210 .152 .155 .110
VI. 1000-5000 .243 .i86 .236 .159 .178 .ii8VII. 5000.10,000 .i8g .179 .257 .190 .176 .141
VIII. 10,000.50,000 .262 .164 .260 .i6i .182 .i8IX. 50,000-100,000 .151 .272 .136 .145 .130
X. Over $100,000 .154 .069 .202 .117
Total .208 .157 .208 .148 .167 .125
Printing and Chemicals and Petroleum andAsset Size Class Publishing Allied Products Coal Products
(thousands of dollars) 1940 1949 5940 1949 1940 5949
I. Under $50 .044 .042 .122 .120 .o6i .064
II. $o'$ioo .067 .055 .130 .112 .045 .053
III. 100-250 .o8o .o6 .128 .io8 .052 .o6o
IV. 250-500 .090 .071 .156 .105 .072 .062
V. 500-1000 .094 .o8 .i8 .122 .107 .074
VI. 1000.5000 .io8 .097 .195 .141 .103 .075
VII. 5000-10,000 .071 .094 .223 .157 .164 .101
VIII. 10,000.50,000 .071 .086 .195 .172 .i6g .157IX. 50,000.100,000 .107 .077 .216 .171 .171 .115
X. Over $ioo,ooo .043 .046 .i8 .i66 .i8 .ii
Total .082 .o8o .182 .157 .174 .127
305
VERTICAL INTEGRATION
TABLE 4 (continued)Rubber Leather and Stone, Clay, and
Asset Size Class Products Leather Products Glass Products(thousands of dollars) 1940 5949 1940 1949 1940 1949
I. Under $o .o8i .o8 .091 .090 .125 .og8IL $504100 .126 .091 .109 .101 .127 .094
III. 100-250 .o86 .125 .107 .146 .100IV. 250-500 .153 .103 .170 .117 .163 .o96
V. 500-1000 .158 .087 .202 .149 .154 .107VI. 1000-5000 .165 .115 .306 .175 .187 .132
VII 5000-10,000 .193 .102 .218 .217 .237 .i68
VIII. 10,000-50,000 .327 .207 .271 .i8i .239 .172
IX. 50,000-100,000 .290 .296 .252 .138 .i6X. Over $100,000 .250 .199 .247 .144 .101
Total .229 .177 .215 .i66 .183 .136
Nonferrous
Iron, Steel, and Metals andAsset Size Class Their Products Their Products Primary Metals
(thousands of dollars) 1940 '949 1940 1949 1940 1949
I. Under $50 .ioi na. .099 n.a. n.a. .o66
II. $o-$ioo .127 n.a. .122 n.a. n.a. .063III. 100-250 .147 na. .137 n.a. n.a. .076IV. 250-500 .159 na. .159 n.a. n.a. .095V. 500-1000 .i8i n.a. .i8o n.a. n.a. .103
VI. 1000-5000 .207 n.a. .191 n.a. n.a. .133VII. 5000-10,000 .218 n.a. .173 n.a. n.a. .143
VIII. 10,000-50,000 .219 n.a. .211 n.a. n.a. .170IX. 50,000-100,000 .293 n.a. .233 n.a. n.a. .151X. Over $ioo,ooo .277 na. .277 n.a. n.a. .i6
Total .229 n.a. .201 n.a. n.a. .151
Electrical MachineryAsset Size Class Fabricated Metal Products and Equipment
(thousands of dollars) 1940 1949 1940 1949
I. Under $o n.a. .og6 .149 .131II. $o-$ioo n.a. .114 .142 .147
III. 100-250 na. .125 .155 .151IV. 250-500 n.a. .142 .155 .159
V. 500-1000 n.a. .155 .182 .172
VI. 1000-5000 n.a. .172 .201 .184Vu. 5000.10,000 n.a. .i96 .205 .190
VIII. 10,000-50,000 na. .186 .212 .174IX. 50,000-100,000 na. .i6o .125
X. Over $100,000 n.a. .i66 .223 .199
Total n.a. .165 .ao6 .184
306
VERTICAL INTEGRATION
TABLE 4 (continued)Machinery Automobiles
(except Electrical and and EquipmentAsset Size Class Transportation Equipment) (except Electrical)
(thousands of dollars) 1940 1949 1940 1949I. Under $o .ii ,io8 .isa .117
IL $504100 .162 .137 .130 .539Iii. 100-250 .177 .156 .125 .128
IV. 250-500 .tg8 .187 .iV. 500.1000 .206 .193 .159 .145VI. 1000-5000 .250 .224 .152 .188VII. 5ooo-1o,000 .s86 .236 .187 .147
VIII. 10,000.50,000 .5s0 .247 .153 .i6iIX. 60,000.100,000 .296 .237 .i8o .151X. Over $100,000 .355 .245 .117 .103
Total .267 .228 .132
TransportationEquipment Manufacturing
(except Other NotAsset Size Class Automobiles) Manufacturing Allocable
(thousands of dollars) 1940 1949 1940 1949 1940 1949I. Under $o .112 .097 .105 .io6 .126 fl.Z.
H. $o-$ioo .118 .141 .145 .124 .542 n.a.UI. 100.250 .152 .153 .163 .133 .i8 n.a.
IV. aro.5oo .s813 .i78 .i88 .148 .164 na.
V. 500-1000 .220 .163 .208 .166 .213 n.a.VI, 1000.5000 .273 .185 .254 .193 .201 n.a.
VII. 5000.10,000 .235 .248 .328 .196 .a66 na.VIII. i0,O00.0,0Oo .s88 .179 .293 .205 .327 n.a.IX. 50,000.100,000 .495 .241 .299 n.a.X. Over $100,000 .307 .244 .351
Total .220 .230 .174 .191 n.a.
Scientific Instruments,Ordnance and Photographic Equipment.
Asset Size Class Accessories Watches, Clocks(thousands of dollars) :949 5949
I. Under $o .s6g .187II. $o-$ 100 .205 .114
IU. 100.250 .195 .i88IV. 250.500 .107 .577V. 500.1000 .423 .229
VI. 1000-5000 .220 .239VII. 5000-10000 .674 .260
VIII. 10,000'50,000 .321 .277IX. 50000-100,000 .300
X. Over $100,000 .378 .228
Total .352 .249
na. (not available)Source: Statistics of Income for 5940 and 1949, Bureau of Internal Revenue,
Complete tabulations for years following 1947 were unpublished in 1952. Accessto the unpublished data was by courtesy of Joseph R. Pechman of the Treasuryepartment.
307
VERTICAL INTEGRATION
over $zoo million; within the group, the variation from lowest tohighest is over 4,000 per cent; and if we subdivide by industries,the samples become, in almost every case, so small as to lose anyreliability. Or, what amounts to the same thing, the large firm isso much a historically individual one that the general influences aremerged and hidden in the particular situation. And, of course, thelarge firms often cross industry boundaries, which makes the sampleseven less reliable.
6. Trends over TimeTABLES 5 to 8 measure the trend of integration through time. Table5 indicates no perceptible change in the degree of vertical integra-tion among manufacturing plants (not firms) since 1849; Table 6,covering all firms since 1929, also shows no change. This is our onlyevidence for the long-term trend of integration over large areas ofthe economy. It is obvious but bears repeating that these are broadaverages that indicate nothing about individual subgroups.iT
Table 7 measures vertical integration of the U.S. Steel Corpora-tion since 1902. Fortunately, both variants of income are available:(i) the spread between total sales and total payments to other firms
and (2) the total of wages, profits before taxes, interest, and depreci-ation. There are no significant discrepancies between the two seriesshown in columns 4 and 12 in Table 7. As noted earlier, the ratioof income to sales is much more sensitive to changes in backwardintegration than to changes in forward integration. If U.S. Steel hadbecome more integrated after 1902, it would probably have beenforward integration, for it began operations nearly self-sufficient inraw materials. Hence an increase in the ratio would be evidence ofa real increase in integration greater than the apparent increase. Butno significant increase is discernible. U.S. Steel has acquired manycompanies since 1902 and built many plants for itself, but it is nomore nor less self-sufficient, that is, no more nor less dependenton the market, than it was fifty years ago.
However, this stability of the statistics is consistent with at leasttwo hypotheses. Suppose that the product mix of U.S. Steel changed,with products requiring relatively little fabrication replacing prod-ucts requiring much fabrication. This would tend to lower theratio. But over the same period, suppose also that U.S. Steel acquiredmore capacity in later stages, becoming more integrated verticallyand increasing its ratio of income to sales. Two such opposing
17 See the three principal qualifications discussed below.
308
VERTICAL INTEGRATION
tendencies could exist and precisely offset each other, leaving theratio unchanged. This hypothesis is certainly not absurd, but untilit is supported by further evidence, it is less tenable than the muchsimpler hypothesis that there has not been much change in thevertical integration of U.S. Steel.
Table 8 presents the results (omitting the computation) for tenmajor steel companies other than U.S. Steel. The data are notamenable to any formal manipulation, but for the larger producers,from Bethlehem to Inland, there seems to be no marked trend overtime. Some of the smaller companies, like Wheeling Steel Corpora-tion, may be increasing their degree of vertical integration.
Certain qualifications must be borne in mind when using dataof this kind.1S The ratio of value added to value of products or ofincome to sales can be affected not only by the integration of manu-facturing processes but also by three broad types of fortuitous de-velopments which are discussed below.
i. The first of these involves price movements. The price changesof raw materials, manufactured fuels, and imported semimanufac-tures purchased may differ from the price movements of manu-factured goods. An increase in the prices of raw materials pur-chased would result in a change of roughly the same order ofmagnitude in value of products but would have a much smallereffect, and that only of an indirect kind, upon value added. Overthe long run, this is not an important qualification, because thereis no reason to believe that since 1849 the long-term price move-ments of raw materials and other purchases have differed signifi-cantly from the long-term movement in prices of manufacturedgoods. The only way to test this hypothesis is to compare the sta-tistics on agricultural and nonagricultural prices; the comparisondoes not support such a hypothesis.
However, it is obvious that the short-term price movements ofmanufactured products differ markedly from those of nonmanu-factured raw materials. This is essentially a cyclical movement. Ifthe census of manufactures were taken annually, or oftener, com-parisons could be made from peak to peak or trough to trough.Since this is not the case, one might look for cyclical movementswhich happened to coincide roughly with the census years. A betterprocedure would be to take averages of prices for decades or foroverlapping decades, as the National Bureau of Economic Research
18 My obligation to Maxwell R. Conklin is great, but he bears no responsibil.ity for any errors in the following discussion.
309
VERTICAL INTEGRATION
usually does. It is a nice question how many such overlappingdecades one needs to consider. In general, the less variable the ratio,the fewer the number of observations needed to establish its con-stancy apart from random fluctuations.
2. The second possible distortion of these ratios would be causedby shifts in the pattern of output.
a. First, it might be the case that in each individual industryintegration was, say, decreasing but that there was a shift from theless to the more integrated industries. In such a situation the ratiofor every industry would show a decline, although the ratio formanufactures as a whole might be stable or even show an increase.Each individual industry would be becoming less integrated, but theeconomy as a whole would be becoming more integrated.
b. In the second situation, if the industries in which the pricesof materials consumed were very high were to expand more rap-idly than those industries in which the prices of materials werevery low, the ratio of value added to value of products would de-crease, and vice versa. This would be not a paradox but a sta-tistical mirage. However, there are two reasons for doubting thatthe results would be significantly distorted in this way. First, aconstant shift from lower- to higher-priced raw materials, or viceversa, would, over the years, be reflected in an upward or down-ward movement of raw material prices relative to others. Thismust be the case, if the average of raw material prices is a weightedaverage. Second, even if there were a real possibility of such dis-tortion occurring between any two given industries, it would be-come of negligible importance when scores or hundreds of groupsare involved. The larger the number of separate forces, the less theinfluence of random events.. Finally, the Bureau of the Census may classify as two plants aproduction unit that was treated as a single plant in earlier censuses.The effect of such action would be to increase the value of productswithout any increase in value added. It is hardly conceivable thatsuch distortions could affect the figures for manufactures as awhole, but they might well be significant for a given minor in-dustry group.
In summary it can be said that the ratio of income to sales as ameasure of integration may be distorted by random "errors" (inthe statistical, not the computational, sense), and it certainly isdistorted by cyclical divergences in price movements. Obviously,comparisons among small industrial groups are dangerous unless
310
VERTICAL INTEGRATION
it can be established that each group is homogeneous in all aspectswhich may affect the ratio. Even for all manufactures, comparisonsover a short period are misleading as often as not.
Our aim at this point is not the measurement of secular trendbut the much more modest objective of determining whether thetrend differs significantly from zero. For this very limited purposethe length of the time period, especially for Tables 5 and 7, makesit difficult to imagine an accumulation of random errors whichwould just offset some tendency one way or the other, leaving thehorizontal trend line we are able to observe. For more preciseanswers, the ratio must be adjusted in various ways; if these ad-justments cannot be made, the ratio cannot be used.'9
The ratio is, even for the individual firm, an aggregative measurein that it reflects the net outcome of all internal and external forcesinfluencing the degree of vertical integration. It indicates nothingabout any particular market or source of supply.
7. Paradoxes of Technical ChangeTHE statistical measures of vertical integration proposed here arenot necessarily better or worse than the more traditional concepts.But they are different from those measures and do not serve onlyas an approximation to them. Furthermore, the ratio Y/S dependson a certain concept of size in economics; if that concept is unac-ceptable, the ratio is invalid.
The size of any part of the economic world is defined as theamount of income generated there or of factor cost absorbed.°Measurement of the size of fixed assets by the cost of acquisition(adjusted for price changes if need be) is only a variant of the in-come approach: it is the cumulation of past factor cost totals. Sizeis a value magnitude, and relative size is a ratio of two or morevalues.
Thus, it appears that the large firms' percentage of total value addedis about the same as their share of total employee compensation (seeTable ). The large firms use more capital per employee but
19 Because the meaning of the total "value of products" standing alone is lim-ited and because this figure was continually misused over the years, the Bureau ofthe Census, after extensive consultation, reluctantly decided not to publish thefigure for the 1947 and later censuses of manufactures. The Bureau has, how-ever, informed the author that "the 1947 ratio, as compared to that for 1939,does not show evidence of increasing integration of plants."
20 For our purpose, we need not consider the problem of income at marketvalue or factor cost.
311
VERTICAL INTEGRATION
TABLE 5Ratio of Value Added by Manufacture to Total Value of Products,
Selected Years, 1849-1939(dollars in billions; ratio in per cent)Total Value Value Added Ratio of Value
Year of Products by Manufacture Added to Total Value1849 $ 1.0 $ 0.4 40.01859 19 o.8 42.11869 3.4 14 41.21879 5.4 2.0 37.0i88g 4.2 44.7i899a 13.0 57 43.81899b 11.0 4.6 41.81904 14.3 6.o 41.91909 19.9 8.2 41.2g4a 23.4 9.4 40.1
igi4b 23.0 9.2 40.01919 6o.o 23.7 39.51921 41.6 17.3 41.61923 58.2 24.6 42.31925 6o.8 25.7 42.3
1927 60.3 26.3 43.61929 68.o 30.6 45.01931 39.8 i8.6 46.71933 3o.6 14.0 45.81935 45.0 18.6 41.3
1937 60.7 25.2 41.61939 56.8 24.7 43.5
a Old basis.b New basis.Source: Census of Manufactures, Bureau of the Census. Data for 1947 are not
available. See discussion in text.
no more capital per unit of wage. If the measure of capital intensityis the hybrid ratio of value units to physical units, then large firmsare more "capital intensive"; if the measure is the ratio of two valueUnits, they are not. The latter ratio seems more meaningful anduseful. In this case, it has a clear implication for the study of costbehavior. The cost structures of large and small manufacturingfirms are not significantly different; large firms do not usually havehigher overhead or capital costs, and lower labor. costs, per unit ofoutput than do small firms in the same broad industrial group.
Some paradoxes arise in studying technical change. Suppose thatthe steel industry installs the continuous casting process, eliminatingthe pouring of steel ingots, reheating, and rough shaping beforefabrication. Costs and prices would fall, in the long run, by aboutthe same amount; so would Y and S, and hence the ratio Y/S, as
shown in tables such as 7 and 8, would decline; steel manufacture
812
VERTICAL INTEGRATION
TABLE 6Ratio of all Corporate Income to all Corporate Sales, 1929-1951
(dollars in billions; ratios in per cent)Ratio of
Year Sales Total Income Income to Sales1929 $138.6 $45.2 32.6igo 118.3 38.2 32.31931 92.4 28.1 30.41932 69.2 18.3 26.41933 73.0 17.2 23.51934 8g.6 23.2 25.9
1935 102.0 27.0 26.51936 119.5 32.0 26.81937 128.9 37.3 29.01938 io8.6 32.0 29.41939 120.8 36.0 29.81940 135.2 42.2 31.21941 176.2 56.5 32.01942 202.8 72.9 35.91943 233.4 88.2 37.91944 246.7 90.8 36.8
1945 239.5 82.8 34.61946 270.9 87.2 31.11947 347.8 105.8 3041948 388.7 1214 31.21949 370.1 116.5 31.5
1950 423.9 131.2 31.01951 484.9 152.3 31.4
Source: National income Supplement, 1951, Survey of Current Business, De-partment of Commerce; Survey of Current Business, July 1952, Tables 12, 29.
would be said to be less integrated. Most of the reduction wouldcome in capital costs; the labor used might well be more skilledand highly paid; the steel industry would be less capital-intensiveand more labor-intensive.
Another example: a grocery supermarket is a larger and costlierdistribution unit than any of the stores it replaces. Yet the cost ofmoving a given unit of food through the supermarket is roughlyhalf that of moving it through a scrvice store;21 hence an integrateddistributor who was also a manufacturer (as a few chains are) be-came less integrated by changing to supermarkets. And the wagesof clerks are greater per unit of "output" in a supermarket than inan old-fashioned store, so the glittering modern supermarket is morelabor-intensive than its predecessor.
Such decreases in integration, I would maintain, are genuine and21 The reason is not only that capital and labor are used much more efficiently
but also that the consumer has taken over the function of collection of goods,of delivery, and to a considerable extent, of holding inventory.
313
Inco
me
Rat
io o
fan
dPr
ofits
AIn
com
eW
ages
Prof
itsEx
cess
befo
re(M
argi
n)A
and
afte
rPr
ofits
Taxe
sPu
rcha
ses
Sale
s(2
_I)
toSa
les
Sala
ries
Taxe
sTa
xes
(6+7
)(x
)(2
)(3
)()
()(6
)()
(8)
$i6o
.8$4
23.1
$262
.362
.0$1
20.5
$90.
3$9
0.3
164.
139
8.2
234.
158
.812
0.8
55.4
554
142.
332
4.9
182.
656
.210
1.0
30.2
30.2
189.
852
3.7
333.
963
.817
7.3
75.9
i.877
.726
5.3
902.
363
7.0
70.6
263.
927
1.5
12.2
283.
7
345.
91,
284.
693
8.7
73.1
347.
922
4.2
233.
545
7.7
339.
21,
344.
61,
005.
474
.845
3.0
125.
327
4.3
399.
636
4.5
1,12
2.6
758.
167
.547
9.7
76.8
52.0
128.
8
Rat
io o
fIn
com
eIn
com
eB
BIn
tere
stD
epre
cia-
(5+8
+to
Sal
esPa
idtio
n9±
10)
(11±
2)()
(so)
(ii)
(52)
$21.
3$2
7.8
$259
.961
425
.629
.323
1.1
58.0
30.1
18.2
179.
555
2
29.8
28.0
254.
562
.229
.435
.631
0.9
64.2
29.4
35.1
329.
965
.431
.323
.822
1.3
66.7
31.5
31.8
294.
066
.7
30.6
32.5
325.
566
.231
.127
.827
5.8
63.9
32.6
33.7
310.
158
.1
33.3
34.0
357.
963
.833
.226
.624
7.1
59.9
32.8
34.3
322.
161
.532
.043
.063
0.2
69.8
31.0
83.3
919.
971
.630
.798
.898
2.1
73.0
30J
89.9
728.
564
.9
TAB
LE 7
Rat
io o
f Tw
o V
aria
nts o
f Inc
ome
to S
ales
, U.S
. Ste
el C
orp.
, 190
2.19
52(d
olla
rsin
mill
ions
; rat
ios i
n pe
r cen
t)Fe
dera
lIn
com
e
Yea
r19
0219
0319
04
1905
1906
1907
1908
'.19
09
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
151.
140
9.2
258.
163
.112
8.1
68.6
168.
748
4.0
315.
365
.114
7.8
98.1
169.
150
4.4
335.
366
.5i6
o.8
104.
610
4.9
331.
622
6.7
68.4
120.
545
.713
8.4
441.
130
2.7
68.6
151.
779
.0
157.
149
1.8
334.
768
.t17
5.0
87.4
146.
343
1.7
285.
466
.ii6
i.65.
321
4.3
533.
931
9.6
189.
654
.219
1.6
560.
836
9.2
65.8
207.
581
.2$
1.9
153.
741
2.2
258.
562
.716
2.7
23.4
1.2
68.6
98.1
104.
645
.779
.0
87.4
55.3
54.2
83.1
24.6
rTj
I1 0 z
413.
61,
290.
687
7.0
68.o
581.
810
9.7
249.
972
6.0
476.
165
.633
2.2
36.6
334.
78o
.o58
.632
3439
.6
3774
1,09
6.5
719.
16.
647
0410
8.7
266.
992
1465
4.5
71.0
443.
685
.1
37.5
147.
224
39.0
2.5
42.1
15.8
124.
5ii.
696
.7
29.3
8o.o
838.
365
.028
.540
.143
9.8
60.6
28.4
47.1
441.
054
.528
.056
.967
9.8
62.0
27.3
53.2
620.
867
.4
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
13.3
103.
917
.113
3.8
11.5
99.4
15.0
129.
1
17.2
214.
7
12.0
116.
40.
113
.1
0.1
—71
.10.
1—
364
2.6
—19
.1
3.9
5.0
11.2
61.7
29.5
124.
42.
9—
4.8
13.0
54.1
26.3
128.
511
8.7
234.
915
3.1
224.
384
.314
6.9
65.0
125.
8
30.0
88.o
32.0
120.
691
.021
8.1
109.
023
8.6
126.
029
1.9
234.
044
9.5
398.
058
2.3
117.
026
0.7
6i.6
650.
863
.770
470
0.3
64.7
644
602.
662
.773
.263
0.9
62.8
69.8
709.
664
.7
5.6
63.8
557.
567
.35.
550
.432
7.4
59.7
5.3
41.6
114.
339
.75.
245
.318
2.0
48.5
5.1
46.4
247.
258
.7
5.0
49.8
313.
758
.2
4.9
59.0
464.
658
.85.
164
.164
0.7
62.3
8.3
50.3
348.
257
.09.
364
513.
360
.7
13.6
72.6
679.
062
.96.
o98
.696
7.8
59.7
6.2
128.
21,
1414
61.3
6.3
134.
01,
200.
16o
.85.
013
9.0
1,22
7.0
58.9
3.5
123.
41,
040.
459
.54.
868
.789
8.6
60.1
2.5
114.
01,
238.
258
.324
146.
014
22.7
57.3
2.3
119.
71,
359.
859
.1
2.2
143.
91,
775.
06o
.o2.
016
2.1
2,12
1.0
60.2
1.9
176.
91,
761.
656
.1
27.1
25.8
26.1
25.7
14.9
333.
61,
022.
068
8.4
674
458.
290
.634
6.7
1,08
2.3
735.
668
.o46
9.3
116.
732
3.1
960.
563
7.4
66.4
412.
787
.933
841,
005.
366
6.9
66.3
402.
911
4.1
350.
01,
097.
474
7.4
68.i
410.
219
7.5
234.
882
8.4
593.
671
.737
1.7
104.
4
187.
254
8.7
361.
565
.925
8.4
13.0
141.
828
7.7
145.
950
.713
8.5
—71
.216
1.4
375.
021
3.6
57.0
167.
9—
36.5
140.
542
0.9
280.
466
.621
4.8
—21
.7
191.
253
9434
8.2
64.6
253.
91.
1
287.
579
0.5
503.
063
.633
9.0
50.5
342.
61,
028.
468
5.8
66.7
447.
194
.922
8.3
6ii.t
382.
862
.629
4.4
—7.
729
3.5
846.
055
2.5
65.3
386.
541
.1
358.
31,
079.
172
0.8
66.8
464.
310
2.2
604.
61,
622.
31,
017.
762
.762
8.3
116.
267
3.4
1,86
3.0
1,18
9.6
63.9
782.
771
.273
0.6
1,97
2.3
1,24
1.7
63.0
912.
962
.681
4.4
2,08
2.2
1,26
7.8
60.9
957.
260
.8
670.
11,
747.
31,
077.
26i
.682
5.5
58.0
560.
414
95.1
935.
762
.570
4.5
88.6
839.
42,
122.
81,
283.
46o
.90
3.6
127.
1i,o
o8.g
2481
.514
72.6
59.3
1,03
5.7
129.
688
5.7
2,30
1.7
1416
.061
.594
5.9
165.
9
1950
i,ii8
.82,
956.
41,
837.
662
.21,
179.
421
5.5
1951
1,32
7.9
3,52
4.1
2,19
6.2
62.3
1,37
4.5
184.
3
1952
1,30
7.6
3,13
7.4
1,82
9.8
58.3
1,32
2.1
143.
7
0 z
Sour
ce: A
nnua
l Rep
ort f
or 1
950
and
1952
,U
.S.
Stee
l Cor
p. D
ata
in c
olum
n 7
supp
lied
by U
.S. S
teel
Com
ptro
ller's
Off
ice.
(The
yva
ry sl
ight
ly fr
om S
tudy
of M
onop
oly
Pow
er, H
ouse
Sub
com
mitt
ee o
n M
onop
oly
Pow
er, S
eria
l No.
14,
Par
t4-
13,
Exh
ibit
S-23
9,p.
570
.)
TAB
LE 8
Rat
io o
f Inc
ome
to S
ales
, Ten
Maj
or S
teel
Com
pani
es, 1
905-
1952
(per
cent
)Y
oung
stow
nW
heel
ing
Shar
onA
llegh
eny
Bet
hleh
emR
epub
licN
atio
nal
Arm
coSh
eet
Jone
s and
Inla
ndSt
eel
Stee
lLu
dlum
Yea
rSt
eel
Stee
lSt
eel
Stee
lan
d Tu
beLa
ughl
inSt
eel
Cor
p.C
orp.
Stee
l Cor
p.19
0547
.619
0631
.419
0727
.519
088.
s19
0945
.7
1910
49.8
1911
47.7
1912
48.1
C,3
1913
51.0
d19
1451
.2a)
'119
1532
.219
1654
.719
1751
.3ig
i851
.043
.619
1964
.148
.12
1920
59.6
50.6
1921
57.0
24.5
1922
54.3
39.9
1923
56.0
53.2
1924
564
53.9
1925
56.4
48.8
1926
56.7
52.9
55.1
1927
58.2
52.7
53.5
1928
57.4
n.a.
49.8
3929
58.6
na.
44.0
1945
1946
1947
1948
1949
59.9
45.9
41.5
614
504
50.5
53.2
47.6
47.3
534
47.8
47.9
54.9
49.8
50.3
1950
58.1
514
50.2
1951
56.2
50.8
50.1
1952
47.3
44.5
Not
e: F
ully
con
solid
ated
stat
emen
ts w
ere
secu
rity
taxe
s and
frin
ge b
enef
its) h
ave
been
414
414
42.3
38.7
42.3
44.3
484
46.2
44.3
47.1
47.2
47.3
49.0
504
47.5
52.6
41.9
43.0
53.2
42.1
41.0
49.3
37.2
43.6
48.3
43.9
52.7
43.7
49.3
444
47.8
40.2
42.7
59.7
43.9
44.1
59.4
44.5
45.!
56.3
39.5
42.8
1930
193!
1932
1933
1934
59.2
55.0
47.1
52.6
56.8
59.5
39.3
344
41.6
45.0
35.6
32.7
39.0
39.6
40.9
1935 ig6
1937
1938
1939
59.9
584
57.6
6o.i
6.8
47.6
46.8
49.2
48.9
51.1
46.6
46.3
48.3
48.8
45.5
-.1
1940
1941
1942
1943
1944
55.6
57.6
59.6
6o.8
62.3
50.1
49.7
52.1
46.6
484
45.9
46.9
49.6
44.3
43.7
41.8
44.4
44.4
42.6
44.8
48.3
50.6
52.5
47.9
50.3
no.8
56.1
46.9
54.2
42.5
52.3
46.8
49.9
n.a.
49.5
47.2
50.3
45.2
50.4
43.9
35.8
43.1
43.0
42.0
42.1
37.7
44.8
47.3
47.0
48.5
48.3
50.2
40.8
48.3
41.7
48.2
45.3
53.2
42.6
54.0
44.3
52.5
42.8
564
45.8
49.8
41.5
56.8
45.0
514
40.8
54.2
T1 z Iii
51.8
48.3
used
for a
ll co
mpa
nies
. Whe
re re
porte
d, to
tal e
mpl
oym
ent c
osts
(incl
udin
g so
cial
incl
uded
. Sal
es fi
gure
s are
net
in a
ll ca
ses.
VERTICAL INTEGRATION
not merely statistical. If the input of resources necessary for a givenprocess should decline and the process occupy less economic spacerelative to earlier processes than it formerly did, then the firm wouldrely more on the market and less on its own contribution for thefinal product. It would be less self-sufficient and less integrated.
The same paradox relates to labor or capital intensiveness. It isdifficult to avoid the impression of capital intensiveness when theoperating units are physically large and impressive. But, again, sizein economics is not a physical but a value dimension, and in ourcapital-rich civilization we can and do treat as cheap, because theyare plentiful, capital instruments which are scarce and thereforeexpensive in other societies. In commenting on the ancient protec-tionist argument against cheap foreign labor, one wit has pointedout that foreign protectionists might with equal logic inveigh, againstbeing flooded with the products of cheap American capital. I sus-pect that research would show many American exports and perhapsexports as a whole to be really labor-intensive, as that term is de-fined here.
8. Possible Correlation with ExpansionASSUMING, then, that the ratios of income to sales and of inventoryto sales are logically sound, what interpretation can be made of theapparent rough correlation between size and the degree of verticalintegration?
George J. Stigler has come nearer than anyone else to formulatinga law of vertical integration.22 Treating integration as the oppositeof specialization, Stigler expects disintegration to be characteristicof an expanding industry, integration of a contracting one. As theindustry (the market) expands, economies of scale become possiblein the various processes and these tend to split off to be separatelyperformed. Hence, in the absence of attempts at market control,there should be decreasing integration; this would be reflected infalling ratios.
Stigler doubts that we need a distinctive theory of vertical integra-tion; this leads me to interpret the above situation as a pattern ofindustry development rather than a logical necessity. As a pattern,it is plausible and will doubtless be borne out in many instances.But in my opinion Stigler's analysis (correctly) contrasts a mature orlarge-scale industry with a small-scale industry, not an expanding
22 'The Division of Labor Is Limited by the Extent of the Market," Journalof Political Economy. June 1951, pp. 185-193.
318
VERTICAL INTEGRATION
industry with a contracting one. The distinction between processand result seems to be of crucial significance: I would guess that anexpanding industry is more highly integrated than a relatively stableone. If we start with an industry in its earliest years, when it is aninnovation, it is at first adapted to and fills a niche in the existingstructure of markets and of factor supply. It is essentially a rearrange-ment of known and available resources. Few can discern its largepossibilities for growth and for pushing the capacity of supplyingindustries and firms. The railroads were originally feeders to canalsand turnpikes, and, later, pipe lines and trucks were considered asfeeders to railroads; the automobile was a rich man's toy; wirelesstransmission of signals was intended for ship-to-shore telegraphy;and many other examples might be given.
As the firms and their industry grow, they do so under the forceddraft of demand chronically in excess of supply at prevailing prices.This economic tension is transmitted to the factor markets as thefirms bid not only for increasing amounts but for changing composi-tion of factors. As larger quantities are needed, some factors becomerelatively scarce and substitution must be resorted to, often by pain-ful trial and error. Economies of scale now appear, as Stigler rightlyiisists; my point is that they appear unforeseen and generally lag-ging behind a keenly felt need. A sluggish response will often forcethe growing firm to provide its own supplies and/or marketingoutlets.
It may be regarded as axiomatic that integration takes place onlyin response to imperfections of competition in supplying or receiv-ing markets. A firm does not normally integrate into a market whereit can buy unlimited quantities at the going price and where theproducers are receiving a normal (or subnormal) return. A firm doesintegrate into a broadening market whose service is scarce and ex-pensive. The scarcity and the high price may be the result of monop-oiy control in the invidious sense, by a single seller or by a groupwhose several minds have but a single thought. But the scarcity mayexist simply because of the time lag in supplying the new factor. Asmall, uncertain, and fluctuating supply is peculiarly subject to re-current "corners" and extortion. Also, there may be considerablemonopoly profit. Even when this is not the case, if the factor is ex-pensive or unsuitable, so that it takes additional costly processingbefore being ready for use or is uncertain in amount, the impact onthe expanding firm is much the same. Thus the very expansion ofdemand which induces economies of scale in the associated markets
319
VERTICAL INTEGRATION
also induces the firms in the growing industry to occupy the asso-ciated markets. Once established, the pattern perpetuates itself, un-less the (private) diseconomies of integration are considerable.
An industry in rapid growth throws the process into boldest re-lief, but it is only the most extreme example of a more general prob-lem. Given an expanding and changing economy, there must neces-sarily at any instant be a host of markets out of equilibrium intowhich it becomes profitable to integrate. This would explain theexistence of vertical integration even in the absence of attempts topre-empt an essential resource in order to prevent competitors fromusing it or to insure the "right" kind of price policy at later stages.
Given imperfect competition and no sharply defined loci of least-cost output, so that a firm may be well away from the optimum scalewithout ceasing to exist, then the half-forgotten history of an in-dustry plus the power of inertia may largely explain its existingpattern of vertical integration. Chance alone may be no small partof the explanation. Imagine an industry comprising n stages withno economies or diseconomies of vertical integration. The joiningof functions would then be purely random. There would be 2—'possible varieties or degrees of vertical integration; the "average"firm would encompass (n+l)/2 stages, with '.JhT/2 standard 4e-viation.2 Thus, with only a half dozen stages, one would find 32possible patterns; the average number of stages encompassed wouldbe s½ but this could be in any one of several patterns; further-more, there would be one chance in three of any given firm encom-passing less than 2½ or more than 4½ stages.
9. Four Developmental PatternsTHE foregoing sketch is not, I think, inconsistent with the evidence.The smallest firms appear to be specialists in a particular process.As the firm grows, it does not merely duplicate its activities; it takesover additional functions, performing some of the services it for-merly purchased. Hence, both ratios would increase. But for thefirm which has grown into a large part of its available market, thetrend to self-sufficiency may be reversed by the marketing necessityto carry a full line. Hence the firm purchases many finished ornearly finished goods to be marketed in conjunction with those itprocesses over a greater length of production line. This decreasesthe two ratios, but even with this reversal the large firm is moreintegrated than the small. A similar development takes place in the
28 My thanks on this point are due to Robert M. Solow.320
VERTICAL INTEGRATION
fully developed industry, where with the passage of time certaindiseconomies of integration are slowly realized and certain functionsare discontinued. For any growing industry or firm, the trend is toincreasing integration, but this is counteracted both by the constantgrowth of new, less integrated industries and firms and by the re-versal of the trend in highly mature industries.
Thus we have about four developmental patterns of vertical inte-gration. For any given firm at any given time, they may be mutuallyexclusive but not over longer periods. Nor are they exhaustive. Itwould be most unfortunate if we looked for the typical pattern andtried to make of it a general theory of vertical integration. Whatwe need is to increase our knowledge of various patterns and tomultiply hypotheses while trying to practice some orderly house-keeping among them. The ratios proposed here, when used in con-junction with other evidence, may serve as useful tools in this task.24
AppendixCERTAIN sources other than those discussed above were investigated,but they proved unworkable. They are briefly indicated in this Ap-pendix, as a warning to those interested in further research in thisarea. The Structure of the American Economy25 contains two inter-esting tabulations of census data. One of them concerns the 200largest manufacturing concerns, grouped by fives; the other givesconcentration data (the largest four and the largest eight producers)in several hundred industries. For each group the value added andalso the value of products is indicated. Unfortunately, the latter isgiven on a combined rather than consolidated basis, so that salesare overstated, in ratio loosely proportional to the number of plants.It is impossible to calculate and allow for bias from this source,since a firm with many horizontally related plants would show littleor no overstatement, while a plant with even two vertically relatedplants might show a very large overstatement. If Monograph 2726showed the number of establishments involved in the several rela-tionships discussed above, it might be possible to calculate the bias;unfortunately, as already seen, the monograph shows only the num-ber of central offices (firms) involved.
Another source which was not useful was tabulation by the Office24 Cf. the able Ph.D. thesis by Frederick E. Balderston, Scale, Vertical Integra-
tion, and Costs in Residential Construction Firms (Princeton University Press,1953).
25 Leontief, op. cit.2eThorp and Crowder, o. cit.
321
VERTICAL INTEGRATION
of Price Administration of wartime manufacturing corporate finance,which was published by the FTC in 1947.27 The basic trouble withthese data, for our particular purpose, is that the gross is too gross,and the net too net. Only purchases of raw materials are indicated,so that the spread between total purchases and sales includes morethan value added by the firm. Proceeding contrariwise, corporateprofits are available, as well as interest; but the wage total is incom-plete and a very large part of total cost is unclassified general manu-facturing expense. Hence, income generated is understated. Anotherdefect, for our particular purpose, is that certain large subsidiarieswere not consolidated with their parents; this results in an over-statement of sales.
Monograph 27 also presents concentration data for 1,807 narrowlyclassified industries, each producing what approximates a "product"in the market sense. Unfortunately, concentration is given in termsof sales in dollars and in physical units but not in terms of valueadded, so that no ratio can be calculated.
Finally, there are the tables computed for the Celler Subcommitteefrom the 1947 census.28 Here, too, concentration is shown in termsof sales, except for twelve industries for which it is given in termsof value added. Nowhere, however, do we have both.
27 Report on Wartime Profits and Costs for Manufacturing Corporations, FTC1947.
28 Letter from the Secretary of Commerce to Representative Emanuel Celler,December i, 1949, Table v appended.
COMMENTIRSTON R. BARNES, Federal Trade Commission
VERTICAL integration is normally understood to refer to an organiza-tion of production under which a single business unit carries onsuccessive stages in the processing or distribution of a product whichis sold by other firms without further processing. Vertical integra-tion results in by-passing a market or having a position on bothsides of the market: making a product instead of buying it, carryingit to a later stage instead of selling it. Thus, bringing together underone managerial direction a raw material producer and the manu-facturer using that raw material, a producer of an intermediateproduct and a manufacturer using that intermediate product, or aNora: The views expressed in this comment are the writer's and not necessarilythose of the Federal Trade Commission.
822
VERTICAL INTEGRATION
manufacturer of the finished product and a distributor of that prod-uct, are examples of vertical integration. In a practical sense, inte-gration must be examined with respect to the particular product orproducts which a firm manufactures, not with respect to productsgenerally; full integration with respect to all products which a manu-facturer fabricates probably exists nowhere in the real economy.
It is, of course, commonplace that a large proportion of industryinvolves two or more successive steps in production which mighttheoretically, though not practicably, be split among two or moreproducers. However, attention is generally focused on those situa-tions where successive stages of production are brought under asingle managerial supervision and where markets are by-passed orstraddled.
It is not fruitful for present purposes to make a prolonged in-quiry as to whether vertical integration is fundamentally different,either in objectives or in consequences, from other coherent formsof integration. We may simply note in passing that the questionsraised by all coherent forms of integration are essentially similar.
RATIO OF INCOME ORIGINATING TO SALES
STATISTICAL attempts to measure vertical integration have yieldedless than satisfactory results. After noting many of the deficiencies,Adelman proposes two measures of vertical integration. He wouldmeasure vertical integration by the ratio of income originatingwithin the corporation (or within the industry) to its sales. He alsooffers an alternative measure of integration: the ratio of inventoryto sales.
Throughout his paper Adelman appears to identify his conceptof integration with the measures which he advocates. These meas-ures are offered as general-purpose yardsticks, and the implicationis that the magnitude measured may be accepted as an accurateindex of integration for any and all purposes.
Adelman appears to be drawn to his proposed measures becausetheoretically and conceptually the larger the value added by themanufacturer, or the greater the income originating within the firm,the farther the firm carries its processing of the product. Unfor-tunately this conceptually neat identification of integration doesnot yield results which are useful in resolving the real problemswhich arise with respect to integration.
Adelman recognizes most of the deficiencies inherent in the ap-plication of his two measures; yet he nevertheless advocates the re-
323
VERTICAL INTEGRATION
finement of the statistical series which he uses and holds out thehope that the measures may be made to work. Where the author hasbeen so diligent in pointing out the deficiencies in his own analyticaltools, it is somewhat gratuitous for the critic to affirm the defectswhich have been noted. Nevertheless, a skepticism amounting almostto conviction that Adelman's ratios of income originating to salesand of inventory to sales will not prove useful in resolving the prac-tical problems associated with integration prompts a warning thatothers refrain from following this path. In short, neither of thesemeasures of integration yields reliable or consistent results. Indeed,each reflects a complex of factors, many of which—such as the profitlevel of the firm or industry—are quite unrelated to integration.
Some of the deficiencies associated with the income-sales ratiomay be examined briefly:
1. Adelman offers a theoretical example of integration involvinga primary producer, a manufacturer, and a distributor, each ofwhich by assumption contributes one-third of the total value ofthe product. The application of the income-sales ratio yields anindex of ioo for the primary producer, 50 for the manufacturer,and for the distributor; yet by definition all are equally "inte-grated." In this instance the ratio reflects the stage in the produc-tive process which is being measured rather than the degree ofintegration. This characteristic alone deprives the index of anyreal value in making comparisons between industries or even incomparing different producers in the same industry, where they arenot at exactly the same stage in the productive process.
The question commonly arises as to what effect an acquisition ormerger has upon the degree of integration before and after the mer-ger. The same example illustrates another deficiency. If the manu-facturer absorbs the primary producer, that is, if he integrates back-ward, the index of integration increases from 50 to ioo and themanufacturer is fully integrated. On the other hand, the primaryproducer, whose index of ioo indicates that he is already fully inte-grated, might absorb the manufacturer; if so, the index of integra-tion would still be 100. However, if the manufacturer absorbs thedistributor, that is, integrates forward, an equal degree of forwardvertical integration yields an index of 67. Thus in these instances theindex of integration measures, not the degree of integration butthe direction of integration, yielding higher magnitudes for back-ward integration than for forward integration.
2. The infirmities revealed by the theoretical example lead Adel-
324
VERTICAL INTEGRATION
man to conclude that comparisons are possible only when dealingwith the same economic function or with closely similar functions,and he warns against comparing the integration of a mining com-pany with that of a retailer. Yet there will seldom exist that simi-larity or identity of function which would render the use of thisindex dependable. Indeed, it is the fundamental purpose of verticalintegration to combine different functions, and different mergerswill effect different combinations of functions. However, the indexis not available for these simple and direct comparisons. If oneknows enough to use the income-sales index with safety, he knowsmore that is significant about the companies concerned than theindex can ever reveal.. A high index of integration as indicated by the income-salesindex may reflect the intensiveness of the productive process as wellas the vertical extension of the productive operation over successivestages. The employment of skilled labor and expensive machinerynormally gives rise to a greater value added by the manufacturingoperations, even though only a single productive stage is involved.Thus, without calculating the index of integration, we should ex-pect to find a higher index for a watch company than for a com-pany engaged in a relatively simple metal-fabricating operation.
4. Income originating, or value added, includes sales less expendi.tures for raw materials, fuel, and power. Specifically it includes cor-porate profits. Hence the greater the corporate profits, the higherthe degree of integration which the index will show, and, con-versely, the larger the corporate losses, the lower the degree of inte-gration which the index will record. Thus, of two companies carry-ing on identical manufacturing operations, the more successful willshow a higher index of integration. In this instance differences inthe indexes of integration measure differences in the competence ofmanagement or in the profit-making possibilities of the two com-panies. Hence the index is not a reliable measure of integrationeven with respect to two companies operating at the same stage inthe same industry.
The lack of comparability between companies performing thesame function renders Adelman's caution against comparisons be-tween companies in different industries somewhat misleading. Infact, the same lack of similarity which he accepts with respect tointerindustry comparisons is inevitably present within most two-digit industries. Furthermore, comparisons might be rendered in-valid by integration which carries different companies in the same
325
VERTICAL INTEGRATION
industry across different industry lines. In short, a statistical measureof integration which is applicable only within the narrowest con-fines has very limited utility. Many critical questions respecting inte-gration which demand answers refuse to be neatly pigeonholed, asthe proposed index would require.
5. When the income-sales measure is applied to reveal changes inintegration from year to year, new difficulties are encountered. Dif-ferential changes in the level of prices introduce aberration. With-out any change whatever in the productive processes carried onwithin the firm, a lower index of integration may result from anincrease in the costs of materials, fuel, and power. Or without anychange in the physical processes of production, a higher indexof integration may result from an increase in the prices at whichthe product is sold, from a greater increase in prices than in costs,or from a greater reduction in costs than in prices.
6. Adelman presents an array of indexes of integration for two-digit industries. In the light of the deficiencies already noted, com-parisons between industries have no discoverable significance.Moreover, the two-digit industries comprehend so many differentproductive operations that it may not be assumed that the industry'sindex is an average which is characteristic of any of the companiescomprising the industry.
7. Adelman presents a tabulation of the indexes of integrationfor the U.S. Steel Co. from 1902 to 1950. To the uncritical reader,the index yields curious results. The index is lower during the de.pression years than during the years before and after the depression.What changes in the realities of integration occurred during theseyears is not known, but if there were no changes, the same resultsmight be expected from changes in cost-price relationships.
RATIO OF INVENTORY TO SALES
AN ALTERNATIVE index of integration, the ratio of inventory to sales,is presented as a derived measure of integration. The theoreticaljustification lies in the observation that the longer the productiveline and the more successive processes performed within the samefirm, the higher the ratio will be. Adelman notes that this measureis "more susceptible to meaningless comparisons" than the income-to-sales ratio. His warning is necessary, for this ratio appears to be nobetter than the income-to-sales measure. It is subject to many ofthe objections already considered as well as to a number of ad-
328
VERTICAL INTEGRATION
ditional deficiencies. Only three new objections are commented onhere:
1. The inventory-sales index is simply the ratio of inventory tosales without respect to the stage at which the inventory is held.Thus, if the inventory is at the final-product stage, the index couldbe as high for the unintegrated producer of the final product as foran integrated producer whose operations cover several stages inproduction.
2. The index yields confusing results where the productive proc-ess is continuous. If there are no intermediate products or if pro-duction moves continuously, as in some of the chemical and papercompanies, the amount of the inventory is minimized. Hence themore highly integrated producers, whose visible inventory consistsprimarily of low-value raw materials, may show a lower index ofintegration than a partially integrated firm whose first inventoriesare in the form of intermediate products. An index which makes apartially integrated firm appear to be more highly integrated thana fully integrated firm is a measure which should be discarded, notrefined.. The index of integration will vary according to the marketingpractices of companies within the same industry. For example, ifthey are successful in moving finished cars directly from the pro-duction line to their dealers, the larger and more highly integratedautomobile manufacturers may show lower indexes of integrationthan smaller, less integrated competitors who are less successful ininducing their dealers to carry their inventories.
In summary, it does not appear that either of the two ratios pro-posed, income originating within the firm to sales or inventory tosales, can be helpful in dealing with any of the real problems ofvertical integration, whether the problems arise with respect to spe-cific firms or to industries.
A SIMPLE PROPOSAL
MANY problems in connection with vertical integration relate to theeffects of such integration upon a competitive organization of in-dustry and upon competitive markets. Some of these problems canbe illuminated by the development of measures which would showthe degree to which different companies are dependent upon mar-kets at specific stages in the processes of production and distribution.An unambiguous and simple measure of the degree of vertical inte-gration of the individual firm with respect to different market levels
327
VERTICAL INTEGRATION
can be constructed for any product or class of products, with sepa-rate measures showing the degrees of forward and backward inte-gration. To make the example specific, one might seek to measureintegration with respect to the pulp-producing and pulp-consumingoperations of a paper company. In this instance one would measurethe interplant transfers of pulp by the company as a percentage ofits total shipments and interplant transfers of pulp; this wouldsupply an index of the degree of forward integration. Or, the con-sumption of pulp by a paper company from its own pulp millsmight be expressed as a percentage of its total consumption of pulp.This would reflect the degree of backward vertical integration. Thismeasure would facilitate comparisons of the degree of integrationof different companies in the same industry and would be equallyserviceable in showing the degree of integration of the same com-pany at different periods of time.
Where suitable universe data are available, as they are in thepaper industry, the same type of measure may be used to determinethe degree of vertical integration between successive market levelsor between the same markets at different periods of time. Or, onemay compare the degree of vertical integration of specific companieswith the degree of vertical integration for the market as a whole.
The complement of the measure of vertical integration for anindividual company shows its dependence on the market for salesor purchases of the products in question. The complement of themeasure of forward integration indicates the degree to which theproducing firm at any particular stage is dependent on the marketfor the disposition of its product. The complement of the measureof backward integration shows the degree to which any consumingunit in the company depends upon the open market to supply itwith the product in question.
The complement of the measure of forward vertical integrationfor the market as a whole states the percentage of the industry's out-put of the product which is available to open-market purchasers.And the complement of the measure of backward vertical integra-tion for the market shows the percentage of the consumption ofthe material or product which is derived from open-market pur-chases.
No statistical measure of vertical integration is without its disad-vantages. Two immediately apparent disadvantages of the measurehere proposed may be briefly noted:
i. Suitable universe figures are collected for only a limited num-
328
VERTICAL INTEGRATION
ber of industries, and within those industries for only a limitednumber of products. Thus, figures for total shipments or for thetotal of sales or purchases in markets, particularly regional mar-kets, may not be available.
2. The measurement of interplant transfers does not yield anunambiguous figure, for different companies, even within the sameindustry, may value their interplant transfers on different bases.The valuation of interplant transfers is essentially a matter of mana-gerial decision. Some companies value interplant transfers at cost,others value them at market price at the time of the transfer, andothers appear to split the final sales price according to an individ-ual management formula.
The principal advantages of the proposed alternative measurearise from the fact that it is concerned with real problems of inte-gration:
i. The percentage of market measure will yield consistent resultsboth through time and as applied to different companies.
2. Full forward or backward vertical integration with respect toa product will yield an index of 1. This would mean that a com-pany sold none of its output of the product in question on theopen market or made no purchases of material on the market.. A company selling all its output of a product and using noneof it, or buying all its requirements of a product, would have avertical integration measure of zero.
4. The indexes are additive, that is, the individual companyfractions can be totaled to give a sum which indicates the degree ofmarket integration. Thus, if the degree of vertical integration forsome of the largest producers or consumers of a product is knownand if corresponding figures are available for the market, it is pos-sible to infer the degree of vertical integration for other companiesas a group. This may be helpful in characterizing particular mar-kets, and it may also illuminate some of the competitive problemsfacing other producers and consumers in that market.
5. The percentage of market index can be applied either to quan-tity figures or to dollar figures, whichever are available.
6. The index will not be distorted by changes in price levels,since equivalent values are used for any given period of time.
7. Finally, the percentage of market index expresses the com-monly accepted concept of vertical integration rather than definingvertical integration in terms of the measure employed.
There are many problems relating to vertical integration where329
VERTICAL INTEGRATION
judgments must rest on some fairly reliable measure of the positionof individual companies with respect to other companies and to themarket as a whole. The critical questions relating to vertical inte-gration must be asked before suitable measures of integration canbe devised. Further work in the construction and refinement ofmeasures of integration should await a clinical examination of theproblems of vertical integration and should be directed at yieldinguseful and unambiguous results.
330