earnings rates in different industries · 87 drygoods 29.110 investments, various 88 groceries 59...

This PDF is a selection from an out-of-print volume from the NationalBureau of Economic Research

Volume Title: Industrial Profits in the United States

Volume Author/Editor: Ralph C. Epstein assisted by Florence M. Clark

Volume Publisher: NBER

Volume ISBN: 0-87014-025-6

Volume URL: http://www.nber.org/books/epst34-1

Publication Date: 1934

Chapter Title: Earnings Rates in Different Industries

Chapter Author: Ralph C. Epstein, Florence M. Clark

Chapter URL: http://www.nber.org/chapters/c5024

Chapter pages in book: (p. 70 - 113)

CHAPTER 3

EARNINGS RATES IN DIFFERENT INDUSTRIES

1. VARIATION IN PARTICULAR YEARS

IF ONE takes any single year of the 1919—28 andexamines the rates of net income upon capitalization earnedby the numerous specific industries into which Manufacture,Trade, Finance and Mining can be divided, he finds a sub-stantial variation. The purpose of the present chapter is toexamine the discrepancies that prevail in individual yearsand to see if they disappear over a period.

In 1919, a year of pronounced post-War prosperity, theaverage return for the 3,144 Manufacturing, Trading, Fi-nancial and Mining corporations belonging to our ten-yearseries' was 17.7 per cent.2 Dividing Manufacture, Trade,Finance and Mining, however, into specific individual in-dustries or trades such as Bakery Products, Castings andForgings, or Department Stores, we obtain 73 groups inManufacturing, 22 in Trade, 5 in Finance, and 6 in Min-ing, or 106 separate groups in all. These we may call either

The larger corporations for which average figures were discussed inthe preceding chapter.

2 This is not precisely equal to the average given in Ch. 2, Table 2. Thatfigure, 17.1 per cent, was a deliberately weighted average, applied to thesample for the four divisions (see note 5, Ch. 2). The present average of17.7 per cent is weighted only because it is an aggregate, i.e., is weightedonly in the first of the three senses described in note 6. Deliberate weightingof the present average was avoided because of the mathematical treatmentto which data that were compared with it were subjected.

[70]

RATES IN DIFFERENT INDUSTRIES [71]

'industries' or 'minor groups'.3 The number of corpora-tions contained in each depends upon the nature of thegroup; it ranges from nine or ten in some groups to 100 ormore in others.4 But in every case, identical corporationsare classified in the same groups from year to year. Cor-porations with losses as well as those with gains are thusincluded. The classification of industries is given in Table12

Examining the average rates of return in these 106 in-dustries during 1919, we find the median minor group earn-ing 21.2 per cent, as compared with the arithmetic mean

8 In contrast to the broader 'major groups', for example, Food Products,Metals or Retail Trade, to which they belong respectively and which arediscussed in the next chapter.

Three-quarters of the groups, however, contain 16 or more corporationseach. The complete list is given in Table 12.

In iota, of course, these 106 minor groups comprise the 3,144 companiesseries for which general averages were discussed in the preceding chapter.See, for qualifications, Ch. 43, 46.

It will also be noted in Table 12 that the numbering of the groups runsfrom 1 through 111. This is because of the necessity of combining severalgroups into miscellaneous categories in preparing the original Source-Booktables. The numbering scheme of the Source-Book, however, is retained inthe present volume, although only 106 separate groups are available foranalysis. (For illustration, consult Table 12 where Group S does not ap-pear separately but is shown in combination with Group 11, as 'Group11—8'.) The original Group 8 was Cane Sugar manufacturing, but thesample contained too few corporations to be significant as a separate classi-fication, hence Cane Sugar was thrown into the Miscellaneous Food Productscategory.

It may here also be observed that in the analysis of these 106 groups, theterm 'median industry' or 'median group' is often because it has aless abstract connotation than merely 'the median' or 'median value'. By itis meant the industry that stands in the middle of the list when the 106groups are arrayed according to their rates of return. To be sure, in anarray containing an even number of items, such as 106, the mid-point isstatistically not item 53 or 54 but 53T/2. But with the 53rd industry showinga return of perhaps 12.6 per cent and the 54th industry showing a returnof 12.5, the median is taken as 12.6, which is the value obtained by round-ing off 12.55 per cent.

'When the upper or lower quarter of these 106 industries is discussed,reference is to the first 27 industries from the top or bottom of the list andnot to an interpolated value..

[72] INDUSTRIAL PROFITS

TABLE 12

MINOR GROUP CLASSIFICATION, 3,144 CORPORATIONS,SHOWING SIZE OF SAMPLES

MINORGROUP

Manufacturing1 Bakery products2 Flour3 Confectionery4 Package foods5 Dairying6 Canned goods7 Meat packing9 Beverages

10 Tobacco11—8 Miscellaneous food

products12 Cotton spinning13 Cotton converting14 Cotton weaving15 Weaving woolens16 Silk weaving17 Carpets18 Men's clothing19 Knit goods20 Miscellaneous clothing21 Miscellaneous textiles22 Boots and shoes23 Miscellaneous leather

products24 Rubber products25 Lumber manufacture26 Planing mills27 Millwork28 Furniture (non-metal)29 Miscellaneous lumber

products30 Blank paper31 Cardboard boxes32 Stationery33 Miscellaneous paper

products34 Newspapers and

periodicals

35 Book and music publish-I ng

32 36 Job printing21 37 Miscellaneous printing19 and publishing26 38 Crude chemicals16 39 Paints23 40 Petroleum refining11 41 Proprietary preparations23 42 Toilet preparations

43 Cleaning preparations27 44 Miscellaneous chemicals12 45 Ceramics18 46 Glass49 47 Portland cement31 48 Miscellaneous stone and17 clay products

49 Casting and forgings

4250 Sheet metal

2351 Wire and nails52 Heating machinery

25 53 Electrical machinery54 Textile machinery

29 Printing machinery26 56 Road machinery64 57 Engines26 58 Mining machinery17 59 General factory55 machinery 23

60 Office machinery 13

28 61 Railway equipment 25

62 Motor vehicles 32

63 11

20 64 40

65 30

23 66 Bolts and nuts 15

67 Miscellaneous machinery 3220 68 Non-ferrous metals 48

NUMBEROF

CORPO- MINORRATIONS GROUP

17

NUMBEROF

CORPO-RATIONS

1746

179

425256

9

16264818

21

2799

2020425418122211

12

FirearmsHardwareTools


TABLE 12 (continued)MINOR GROUP CLASSIFICATION

NUMBER NUMBEROF OF

MINOR CORPO- MINOR CORPO-

GROUP RATIONS GROUP RATIONS

Manufacturing (con!.) 91 Building material and69 Jewelry 24 hardware 13

70 Miscellaneous metal 92 Paper 18

products 45 93—85 Miscellaneous wholesale71 Scientific instruments 23 trade 82

72 Toys 12 and Retatil73 Pianos 11 95 Commission dealers74 Miscellaneous special 96 Coal, wood and fuel 10

manufacturing 43 97 Hardware 16Trade—Retail 98 Lumber 11

75 Automobiles 15 99—95 Miscellaneous wholesale76 Men's clothing 17 and retail trade 5277 Department stores 93 Mining78 Drygoods 27 100 Bituminous coal 3379 Furniture 12 101 Gas and oil wells 1180 Groceries 14 102 Stone quarrying 9

Si Jewelry 11 103 Clay, sand and gravel 9

82 Building material and 104- Metals, various 15hardware 27 105 Miscellaneous mining,

83 Lumber and coal 15 other than metal 11

84 Miscellaneous retail Financetrade 52 106 Savings banks 18

Trade—W/iolesale 107 Commercial banks 3085 Coal, fuel and wood 108 National banks 10586 Drugs 25 109 Trust companies 5687 Drygoods 29 .110 Investments, various88 Groceries 59 sorts89 Hardware 43 111—110 All other finance (ex-90 Importers and exporters 23 cept life insurance) 137

figure of 17.7 per cent above cited. But some industriesearn more than 40 per cent, while others earn less than 1per cent. The upper quarter of these 106 industries all showrates of over 29 per cent; the lowest quarter, under 16 percent. The average departure or 'spread' of all 106 minorgroups above or below the median figure is 8.5 per cent 6

° The expression 'per cent' is here, of course, used in an absolute, not a


a spread equal to about two-fifths as much as the medianitself. In appraising these figures, it is of course to be re-called that 1919 was an exceptionally prosperous year.

Turn, however, to a year of depression such as 1921.The median industry here earns only 7.1 per cent on itscapitalization. Fifteen groups suffer actual losses. At thesame time five others earn 20 per cent or more. The actualvariation, in absolute terms, is less than in 1919; the aver-age departure from the median is 5.2 per cent as against8.5 per cent. But in relative terms the variation betweenthese 106 industries in 1921 is substantially greater thanin 1919; their average departure of about S per cent fromthe median represents an amount two-thirds as large as themedian figure itself.

Finally, as a third example, take a prosperous year suchas 1928. Here the median industry earns 11 per cent. Theupper quarter earn 14.8 per cent or over; the lower quarter,8.5 per cent or under. The average departure from themedian is 3.8 pci. cent, a smaller figure than in either 1919or 1921. Even so, it amounts to about one-third of themedian figure.

These three distributions of earnings rates, together withthose for other years, are shown in Chart 3. In no year isanything approaching a uniformity of return seen betweenthe different industries, although it is true that in the lastrelative, sense. The rate of return upon which attention is fastened in thischapter is the percentage of total net income to capitalization. In someways, the rate of total profits to total capital (as these terms are definedin Ch. 2 and in the Glossary) would be preferable as a measure of com-parative earning power. It is not employed in this chapter because our datashow funded debt figures only for 1924—28, and ten—year comparisons arethus not available in terms other than those of total income upon capitaliza-tion. The two ratios, however, are not very different for the purpose inhand. On the close correspondence between them see Table 4, Ch. 2; alsoCh. 8.


five years of the period the majority (slightly over half)of those 106 minor groups earn between 8 and 15 per cent.7The earnings rates of each individual industry, for 1921 and1928, are given in Table 13.8

TABLE 13

EARNINGS RATES, 106 MINOR GROUPS IN 1921 AND 1928, RANGEDIN ORDER OF PROFITABLENESS FROM THE LOWEST TO THE

HIGHEST FIGURES

Trade Groups are Distinguished from Manufacturing by the LettersR (Retail), W ('Wholesale) or W—R (Wholesale and Retail)

1921 1928PERCENTAGE PERCENTAGEINCOME TO . INCOME TO

CAPITALIZATION CAPITALIZATION

—12.6 Rubber products 1.3 Rubber products—12.4 Sheet metals 1.5 Weaving woolens—7.0 Meat packing 3.4 Railway equipment—6.6 Miscellaneous leather products 4.5 W—Building material and—5.4- Miscellaneous food products hardware—3.1 Tools 4.5 Bituminous coal—2.9 Miscellaneous metal products 4.6 Meat packing—2.7 W—Paper 4.7 Cotton weaving—2.6 Miscellaneous machinery 4.7 Silk weaving—2.5 Stationery 5.1 Miscellaneous leather products—2.3 Metals, various 5.1 Blank paper—1.8 Bolts and nuts 5.1 W—R Lumber—L5 W.—Hardware 5.8 Gas and oil wells—0.4 Castings and forgings 5.9 Pianos—0.4 Wire and nails 6M Castings and forgings0.8 Cardboard boxes 6.7 Carpets0.9 Beverages 6.8 Lumber manufacture1.1 Gas and oil wells 6.9 Engines1.4- Miscellaneous R—Trade 7.4 Groceries

The shape of the distributions varies somewhat from year to year, al-though in several years a rough approach to symmetry is approximated.Whether or not a symmetrical pattern would be approximated more closelyif the samples were larger is debatable; but even so, the ranges of varia-tion would scarcely be narrowed. As to the degree of reliability attachingto the present samples, see Ch. 43—45.

Data for other years may be found in the chapters which contain de-tailed discussion of particular major and minor groups (the several chaptersof Book II).


TABLE 13 (continued)

EARNINGS RATES, FROM THE LOWEST TO THE HIGHESTFIGURES

1921PERCENTAGEINCOME TO

CAPITALIZATION

1,5 Miscellaneous stone and clayproducts

1.8 Millwork1.9 Blank paper1.9 Hardware1.9 Non-ferrous metals2.0 Miscellaneous W—Trade2.0 W—R Hardware2.1 Firearms2.1 Jewelry2.3 Pianos2.5 Lumber manufacture2,5 W—Building material and

hardware2.6 Groceries3,4 Mining machinery3.8 General factory machinery4.2 R—Lumber and coal4.3 Miscellaneous clothing4,3 Railway equipment4.7 R—Jewelry4.9 Paints5.1 Heating machinery5.3 Miscellaneous paper products5.3 Miscellaneous chemica's5.96.0

6.16.46.46.46.66.66.7

6.8

7.07.27.3

Stone quarryingAll other finance (except life

insurance)W—DrygoodsCeramicsElectrical machineryOffice machineryConfectioneryCrude chemicalsMen's clothingR.—Building material and

hardwareW—R Coal, wood and fuelTrust companiesW—R Lumber

/928PERCENTAGEINCOME TO

CAPITALIZATION

7.6 W—R Hardware7.7 Miscellaneous textiles7.9 Miscellaneous mining, other

than metal8.1 Commercial banks8.3 Cotton converting8.3 Millwork8.4 R—Building material and

hardw are8.4 W—Drygoods8.5 Planing mills8.5 Ceramics8.5 Metals, various8.6 Beverages8.6 W—Hardware8.7 Miscellaneous W—Trade9.0 Miscellaneous paper products9.2 Heating machinery9.4W Miscellaneous food products9.4 R—Departrnent stores9.4 National banks9.5 Cotton spinning9.7 Dairying9.7 Miscellaneous W—R Trade9.8 Petroleum refining9.8 R—Furniture

10.1 Miscellaneous lumber products10.2 Mining machinery10.2 W—Paper10.410.5 Furniture (non-metal)10.5 Stationery10.6 R—Lumber and coal10.7 W.—Importers and exporters10.8 W—R Coal, wood and fuel10.8 Stone quarrying10.9 Trust companies11.0 Hardware11.2 Knit goods11.2 Savings banks



EARNINGS RATES, FROM THE LOWEST TO TFIE HIGHESTFIGURES

192.1 1928PERCENTAGE PERCENTAGEINCOME TO INCOME TO

CAPITALIZATION CAPITALIZATION

7.6 W—Drugs 11.4 Men's clothing8.0 Miscellaneous lumber products 11.8 Glass8.1 Portland cement 11.8 Sheet metal8.1 Clay, sand and gravel 11.9 Crude chemicals8.3 Dairying 11.9 R—Drygoods8.3 Silk weaving 11.9 All other finance (except life8.3 Cleaning preparations insurance)8.4 Furniture (non-metal) 12.0 Flour8.4 Motor vehicles 12.1 Book and music publishing8.5 Printing machinery 12.1 Jewelry8.7 Road machinery 12.3 Printing machinery8.9 Flour 12.4 Miscellaneous clothing9.0 Bituminous coal 12.4 Miscellaneous special manu-9.0 National banks facturing9.1 Canned goods 12.5 Clay, sand and gravel9.1 R—Men's clothing 12.6 R—Men's clothing9.2 W—Importers and exporters 12.7 Canned goods9.4 Cotton weaving 12.8 Cardboard boxes9.4 Engines 12.8 Portland cement9.5 Glass 13.4 General factory machinery9.9 Planing mills 13.5 Miscellaneous machinery

10.3 R—Furniture 13.8 Wire and nails10.4 Cotton spinning 13.9 Job printing10.6 Miscellaneous special manu- 14.6 Textile machinery

facturing 14.8 Boots and shoes10.7 Miscellaneous W—R Trade 14.8 Paints11.1 R—Automobiles 14.8 Non-ferrous metals11,2 Toys

. 15.1 Confectionery11.4 Weaving woolens 151 R—JeWCIr11.4 Boots and shoes 15.2 Tools11.4 Proprietary preparations 15.2 Miscellaneous R—Trade11.4 Savings banks11.6 Commercial banks 15.6 Tobacco

11.7 R—Drygoods 15.6 Miscellaneous metal products11.9 Cotton converting 16.1 Motor vehicles12.2 Miscellaneous textiles 16.1 Toys12.6 R—Department stores 16.5 Electrical machinery13.0 Miscellaneous mining, other 17.0 Road machinery

than metal 17.1 Miscellaneous printing and13.1 Tobacco publishing



1921PERCENTAGEINCOME TO

CAPITALIZATION

13.1 Job printing13.1 Scientific instruments13.5 Book music publishing14.5 Petroleum refining14.9 Package foods15.0 Bakery products15.9 Carpets16.5 Knit goods21.5 Newspapers and periodicals22.6 Textile machinery24.4 Miscellaneous printing and

publishing26.1 R—Groceries29.2 Toilet preparations

TI-IE HIGHEST

.1928PERCENTAGEINCOME TO

CAPITALIZATION

17.1 Miscellaneous clay and stoneproducts

17.5 Bakery products17.6 Package foods17.7 Office machinery17.8 Miscellaneous chemicals18.2 R—Automobiles19.0 Bolts and nuts19.4 Cleaning preparations19.6 Firearms21.6 R—Groceries21.8 Proprietary preparations25.4 Toilet preparations26.5 Newspapers and periodicals27.3 Scientific instruments

The degree of departure of the earnings rates of these106 industries from the median figure, in particular yearsof the period 1919—28, is highest in 1919, 1920 and 1921.In all years subsequent to 1921 the relative variation re-mains almost constant from year to year, at a figure aboutequal to one-third of the median itself. These figures—'coefficients of variation'—are printed with the diagrams ofChart 3•9 They indicate that, over the period 1922—28, nosignificant change in the extent of the variation betweendifferent industries took place. Chart 4 affords anothermeasure of the relative variation of earnings in differentyears by presenting the 'interquartile ranges' to supplementthe broader figures.

° The term 'coefficient of variation' is employed in the sense of the ratioof the arithmetic mean deviation to the median, and not to represent, as ismore usual in statistics, the ratio of the standard deviation to the arithmeticmean.

EARNINGS RATES, FROM THE LOWEST TOFIGURES

RATES IN DIFFERENT INDUSTRIES

2. VARIATION IN TEN-YEAR RATES

[79]

evidencenot—at least, over the ten-yearcontinuous data.

their 'ten-year

But neither the fact that the variation between industriesis less in 1928 than in years before 1922, nor the fact thatit remains much the same from 1922 to 1928 affords evi-dence as to the degree in which the alleged tendency of thereturn in different industries to approach an equality overa period is realized in practice. For it is conceivable that theseveral industries might constantly be changing their posi-tions, yet the average amount or extent to which their ratesvaried from the median remains much the same. This mightbe true even were a pronounced tendency to equality indeedpresent; for it might be exactly through such a shifting ofthe high industries of one year into the low positions of an-other year, and vice versa, that the aggregate average earn-ings of a particular industry for a period would correspondwith those of other industries for the same period. Oneindustry might enjoy prosperity during one set of years,other industries earn above average rates in different years.To reiterate, a marked variation during any given yearcould conceivably be the mechanism through which approxi-mate equality was realized in the long run.'°

But is this what takes place? The availablewould indicate that it is

period for which we haveThe 106 industries, the net incomes of each totalled for

ten successive years and then divided by the aggregate capi-talizations shown by each industry over the ten successiveyears, display a wide diversity indeed in

10 Were the frequency distributions entirely symmetrical, such shifting aswould equalize earnings rates over a period could conceivably result notonly in the absolute maintenance of a marked annual variation during theperiod but also in yearly variation coefficients of exactly the same magnitude.

[80]

CHART 3

INDUSTRIAL PROFITS

FREQUENCY

0

0

Dz

DISTRIBUTIONS OF

20

lb

(B) YEAR 1920

'0.0

'2 -14.1

Q3 19.5

0

COEFF.V

PERCENTAGE INCOME TO CAPITALIZATION

.482

m


03 $0.7

COEFF.V. .732

2)TO CAPITALIZATION,

I

PERCENTAGE OF INCOME

106 MINOR GROUPS

(A) YEAR 1919

6

$6.6

21.2$2 —29.2

COEFF.V. .404

nnn..m.—tO -5 0 5 10 IS 20 25 .30 35 40


a.

0

I..0

U,lfl

z

20(C) YEAR 1921

$6

0

Ui

Dz

-2.1

7.1

8

4

-10 -5 0 5 tO IS 20 25 35 40

CHART 3 (CONT.)

FREQUENCY DISTRIBUTIONS OF PERCENTAGE OF INCOME

U)0.20

(U

2z

U)a.700I,-0

(U

27z

(0) YEAR 1922

Q1 10.0

13.6

17,6

COEFF.V. .394

0-5 0 5 10 15 20 25 30 35 40


YEAR 1924



TO CAPITALdZATION, 106 MINOR GROUPS20

16

12

8

4

0

20

16

CE) YEAR 1923

11.3

14.3

a3 17.5COEFF.V. .239

111.1 ..11i 11i

U)a.20

12

0

2z

4

0

20-

16

(F)

-tO -5 0 5 10 IS 20 25 30 35 40

INCOME TO CAPITALIZATION

12

8

4

0

8.611.4

.I.

15.5

COEFF.V. .354

r

. i—i•—i—•i•••l

-tO -5 0 5 ID 15 20 25 30 35 4011 fl

[82] INDUSTRIAL

CHART 3 (CONT.)

PROFITS

FREQUENCY DISTRIBUTIONS OF PERCENTAGE OF INCOME

0

0

z

TO CAPITALIZATION, 106 MINOR GROUPS

20

16a0U 12

U-0

z4

0-10 -5 0 5 10 15 20 25 30 35 40


-0-50

Q3 13.4COEFF.V. .329

10 15 20 25 30 35 40

-tO -5 0 5 tO 15 20 25 30 35 40

PERCENTAGE INCOME TO CAPIIALIZATION

YEAR 1926

a' e.gQ2 12.5

L18

(I) - YEAR 1927

8.9

... -

U,a.D0UU.0

Iz

IC

8

PERCENTAGC INCOME TO CAPITALIZATION


CHART 3 (CONT.)FREQUENCY

'I,D0

L.0

Li

z

DISTRIBUTIONS OF PERCENTAGE OF INCOME

TO CAPITALIZATION

Q1 10.6

12.6

Q3 15.4

COEFF.V. .264

D0

12U-0

Li

Dz


20

TO CAPITALIZATION,

16

106 MINOR GROUPS

(J) YEAR 1928

8.511.0

- . 14.8

COEFF.V. .342

- i0"PERCENTAGE INCOME TO CAPITALIZATION

20(Ft) TEN-YEAR AGGREGATE 1919-28

U)0.

0

0

Li

Dz

16

12

8

4

0

r

20

-10 -5 0 5 10 IS 20 25 30 35 40PERCENTAGE INCOME

-tO -5 0 5 tO 15 20 25 30 35 40


rates' of earnings. Some industries, for example, MeatPacking, Castings and Forgings, or Rubber Products, earnless than 6 per cent upon capitalization for the period.Others, such as Scientific Instruments, ToiletPreparations, earn over 25 per cent. The lowest quarter ofthe 106 industries earns under 10.6 per cent; the highestquarter, over 15.4 per cent. The median return is 12.6 percent, as compared with an arithmetic mean of 10.7 per cent.Chart 3, section K shows the distribution of all of these'ten-year aggregate' figures.

A comparison of this distribution with those for the sev-eral individual years of the period (in other parts of Chart3) does, however, establish the fact that although the ten-year earnings of the different industries in no sense ap-proach uniformity, they do vary somewhat less widely thanthe earnings rates of any one year. The average departurefrom the median, in the case of the ten-year figures, is 3.3per cent, or about one-fourth of the median itself, whereasthe average departure in individual years is equal to one-third of the median figure. Likewise the interquartile rangefor the period is somewhat less than in all individual years,save one. Yet, in spite of this, we are not able to concludethat equality of return in different industries over a periodis at all realized. It is not attained, at least, over the ten-year period covered by this investigation. All that can besaid is that the rather wide discrepancies found in all indi-vidual years are narrowed somewhat over a run of years.But so substantial a variation still remains that no 'levelingmechanism' can be said to operate very effectively, over ourten-year period.

While we do not have continuous data for any longerperiod it will be of interest briefly to examine those for asomewhat shorter span. The first few years of the period1919—28 are characterized by rather violent fluctuations in

RATES IN DIFFERENT INDUSTRIES

I-zUU

U0.

CHART 4INTERQUARTILE RANGE, PERCENTAGE INCOME TO

CAPITALIZATION, 106 MINOR GROUPS

[8S]

business conditions. On the other hand, the last six years,constitute a period of general expansion, broken

by recessions far less severe than during the earlier period."'We may, therefore, survey the 'six-year rates' for this rela-tively stable period and observe to what extent they differfrom those for the ten-year figures.

It is to beK. The six-year distribu-

tion is roughly symmetrical: the 106 industries are groupedon either side of the median (which is 12.1 per cent) insomewhat more regular pattern than in the case of the

This statement refers, of course, to industrial activity and not tosecurity speculation.

30

25

2

i92O 1921 1923 1924 1925 1926 1927 1928

1923—28,

Chart 3, section L shows this distribution.compared with Chart 3, section

a


other distributions. The quartiles, however, are not verydifferent, and the coefficient of variation from the median isonly slightly higher than for the ten-year figures.

We leave these distributions without extended commentat this point. But it seems clear that the 'tendency towardsequality over a period' is scarcely more effective over a ten-year period than over a six-year one. It would also appear,so far as the evidence goes, that the distorting influenceswhich prevent this tendency from actually working itselfout are quite as much in evidence during a stable and pros-perous period as over a longer period which includes ex-treme expansion and depression phases (1919—21) in addi-tion to years of relatively stable prosperity.'2

3. INVALIDITY OF THE RISK EXPLANATIONS OF DIFFERENTEARNINGS RATES

Without endeavoring in this chapter to account for thenon-equality of earnings rates which has been established,we must, however, comment upon one matter. It is fre-

12 This conclusion contradicts the tentative hypothesis set forth by R. C.Epstein in Statistical Light Upon Profits, As Analyzed in Recent Literature,Quarterly Journal of Economics, February 1930, in which it was said:"Crum (in his Corporate Earning Power) alludes to 1926 as 'a fairlytypical year' (p. 13S). Probably he is right. In the sense of being relativelyfree from price fluctuations, and from customarily accepted boom or depres-sion characteristics, 1926 was 'normal'. But profits and losses in businessappear not merely during such normal years, but in years less typical aswell. And it may indeed be that the principal distorting elements whichperhaps prevent profits from reaching either relative equality or relativestability, as between different industries and trades, occur during the non-typical years which every business repeatedly goes through."

Distorting elements which prevent profits from reaching equality doindeed exist, as our present data show; but they are no more operativethe entire period under survey than in that portion which contains 'normal'years chiefly. In fact, upon the basis of the above ranges and coefficients,they seem even somewhat less operative during the shorter period; but themargins of possible error in the data prevent attaching much importance tothis slight difference.

RATES IN DIFFERENT INDUSTRiES [87]

quently said that a tendency to equality between the severalindustries is operative, 'differences in risks considered'.Such statements are loose and betray a misconception of theproblem.

Differences in the degree of risk to which industries aresubject, to be sure, do exist. If the hazards are susceptibleof close estimation in advance, they may be allowed for.The charge then becomes one for insurance and is deductedfrom gross income before net earnings appear. If, as ismore often the case, no actuarial estimate of such risks ispossible, then in the profits of any one (good) year theremight appear a component that would properly be charac-terized as compensation for risk, or more accurately, forthe uncertainty attaching to the economic circumstances ofan industry in any particular year.'3 But here what holdstrue of the short run is emphatically not true of the longrun. An excess of return in any one industry over the aver-age rate in a good year, in so far as it constitutes a riskdifferential, would be received in order to offset a deficiencyof return in a bad year; and if the principle which holdsthat differences in profits serve to equalize differences in riskswere true, the 'differential surpluses' of the good yearswould, over a period, exactly equal the 'differential deficits'of the poor years.

If this did not occur, then it would always be advan-tageous to invest capital in the most risky, or rather themost 'uncertain', industries, for they would consistently re-turn earnings (over a time period) at higher than averagerates! In other words, the classical doctrine of compensa-tion for risk can properly be applied in the explanation ofsuch differences in earnings rates as exist in any one year,but not in the explanation of discrepancies that prevail over

Frank H. Knight, Risk, Uncertainty and Profit (1921).


a long period. Interpreted correctly, the principle of "thetendency towards an equality of profit rates in different in-dustries differences in risk considered", in so far as it oper-ated perfectly, would result in absolutely no differences inreturns, extraordinary gains and losses combined, over aperiod sufficiently long to include years in which the hazardswere realized.'4 Accordingly, 'differences in risk' can scarcelybe said to account for the variation in earnings shown bythe ten-year aggregate rates of the 106 industries hereunder examination.'5

4. PERMANENCE OF THE POSITIONS OF 'HIGH' AND 'LOW'INDUSTRIES

The evidence afforded by the discrepancies found to existamong the 'ten-year earnings rates' of different industriesmay be supplemented by a somewhat less summary set ofanalyses. Earnings rates in general, we have found, fail toattain equality over a decade. But exactly how great andhow continuous are the differences which specific minor

N Smith and Ricardo were perhaps not so explicit upon this score asmight be desired, but a careful reading of Mill or a fair interpretation ofMarshall leaves no doubt as to the logic of the position here taken: differ-ences in risk, if compensated accurately, would equalize, not vary rates ofreturn.

This is not to say that capital is not withdrawn from industries inwhich, over a ten-year or longer period, it receives remuneration at sub-stantially less than 'average' rates; in this sense, continued and recurrentuncertainties may affect the decisions of entrepreneurs just because theyare not compensated for. The whole question of the calculation of risk is,however, involved, and in most discussions of it, gratuitous assumption andguesswork have played all too prominent parts. The question of the re-newal or non-renewal of capital is again mentioned in Ch. 47.

Professor F. C. Mills, in commenting upon the general argument of thetext, agrees that differences in risk do not account for the wide variationsin earnings shown by our ten-year figures but suggests that the averagereturn in a risky business might in the long run exceed that in a less riskyenterprise, "the differential representing a reward for the high (year-to-year) variability of the return".


groups are able to maintain year by year? Do the influenceswhich in some instances serve to reduce the extreme dif-ferential advantages enjoyed by specific industries operatesteadily throughout a period or only at certain times? Anddo any industries manage to escape entirely the 'erosive' in-fluences of competition upon persistently high profit rates?

Numerous statistical measures were applied to our datain an effort to answer these questions as completely as pos-sible. From the several analyses made, two types of ap-proach were finally selected as most suitable for the pur-pose at hand, in terms of fruitfulness and comparativesimplicity alike. Neither method, however, avoids makingsome demand upon the reader's patience. For the benefitof the non-statistically minded reader, each is explainedrather fully.

The first method of approach is to list the 106 industries,for each of the ten years separately, in the order of theirrespective rates of return upon capitalization in that year.Each list or array is then divided into four equal parts, thepoints of division being termed 'quartiles'. The 27 indus-tries that are highest in point of net return may be said tostand 'above the first (or upper) quartile'. Similarly, in thelist for any year the 27 lowest ranking industries occupypositions below the third (or lower) quartile. The industryat the second (or middle) quartile is the median industryand its rate of return is the median figure.'6

Thus any particular industry, or set of industries, may10 In a series of 106 items the twenty-seventh item (counting from either

end of the array) is theoretically merely a point of division between groupsand does not, strictly speaking, belong to the group of items above or belowit. In order, however, to include all of the 106 industries in the analysis,both the twenty-seventh item and the eightieth item have been included inthe groups above the first quartile and below the third quartile. There arethus actually 27 items in each of these two groups, and 26 items in each ofthe two groups between the median and the upper and lower quartiles. Seenote 5, Ch. 3.


be compared with the median industry in any one year, orits position relative to the quartiles in any year may betraced throughout the period. The same industry might, ofcourse, stand above the first quartile in one year but belowit in another. Both with individual industries and with dif-ferent sets of industries it is of interest to know not merelywhether they consistently remain above the highest (orbelow the lowest) quartiles in position, but also the extentto which either their individual or their averaged rates ofearnings regularly stand above or below the median figure.Our first method of measuring the degree and the per-manence of profit differentials, therefore, consists of count-ing the changes in position, and examining the earningsrates, of the various minor groups that consistently re-mained above or below certain division points in the list.

Of the 106 industries four are to be found above thefirst quartile in all ten years: Scientific Instruments, ToiletPreparations, Miscellaneous Printing and Publishing, andRetail Automobiles. The first three are manufacturinggroups, the last a trading group. The Scientific Instrumentgroup includes 23 corporations, with an aggregate capital-ization of 149 million dollars (annual average). Its netincome upon capitalization ranges from 13.1 to 33.1 percent per year. In six years of the ten its rate is equal totwice or more that of the median industry in the sameyear.'7 In no year does the return fail to exceed the medianby about one-half.

In the case of Toilet Preparations the group includesonly nine corporations. Their aggregate capitalization an-nually averages about nine million dollars. But on this capi-talization three million dollars or over is earned in six ofthe ten years. The result is a rate of return that is ordinarily

IT That is, the coefficient of variation for this one item is always 1.00 ormore.


two or three times the median figure in each year. Beyondquestion, the great differentials enjoyed by the companiesin this group rest upon the combination of low manufactur-ing expense and the high market prices that can be obtainedfor strongly advertised trade marked products.

The Retail Automobiles group comprises 15 corpora-tions with a total investment of 13 million dollars (annualaverage). This group's variation from the median rate isin some years much like that of the Scientific Instrumentsgroup, but shows greater year-to-year fluctuation. Since1925, however, its differential advantage has declined.Doubtless this is an instance in which the advantage wouldnot be of permanent character if a period longer than tenyears were taken, for we know that the secular trend ofmotor vehicle consumption has changed during the last halfof the period. It is also to be noted that the corporations inquestion ai-e relatively large for retail trading establish-ments and may not be very typical of motor car and dis-tribution experience as a whole, even for this period.

Of the Miscellaneous Printing and Publishing groupnothing need be said here other than to state that in mostyears the earnings of the 17 companies involved stood at afigure half as large again as the median.

Table 14 shows the ratios by which the earnings ratesof the four groups in question exceeded the median rates.In each case the 'median difference coefficient' represents thedifference between the group's earnings rate for the yearand the median rate, divided by the median rate for thesame year.

Although only four of the 106 minor groups are foundabove the first quartile in every year, two others, News-papers and Retail Groceries, remained there in all years ofthe period save one. Details for these and other groupsdiscussed in this section appear in Tables 14 and 15.

TA

BL

E 1

4M

ED

IAN

DIF

FER

EN

CE

CO

EFF

ICIE

NT

S, R

EL

AT

IVE

LY

PRO

FIT

AB

LE

.47

.85

.56

2.86

IND

UST

RIE

S'

2.23 .3

9.5

71.

42

2.44

3.11 .8

5.56

.92

2.91

1,26

1.15

.31

1.48

1.00 .91

.74

1.63

1.35

1.71

.53

1.30

1.36 .85

.66

1.58

1.31 .61

.55

1.31

1.48 .6

5

.56

.25

.47

.31

.85

.27

.83

.56

2.86 .40

.29

1.13

2.23 .2

1.39

.56

.30

.57

1.42 .14

GROUP NAME

1919

1920

1921

1922

1923

1924

1925

1926

Abo

ve f

irst

qua

rtile

dur

ing

ten

year

sM

isce

llane

ous

prin

ting

and

publ

ishi

ng.5

5

Toi

let p

repa

ratio

ns2.

30Sc

ient

ific

inst

rum

ents

.4-7

Ret

ail a

utom

obile

s1.

22A

bove

fir

st q

uarl

ile d

urin

q ni

ne y

ears

New

spap

ers

..

1.13

2.03

1.64

.87

1.14

.48

1.38

1.17

1.41

Ret

ail g

roce

ries

.40

..

2.68

1.01

.55

1.06

.71

.74

1.02

.96

Abo

ve m

edia

n du

ring

ten

year

sB

oots

and

sho

es.4

4.3

5

New

spap

ers

1.14

1.41

Mis

cella

neou

s pr

intin

g an

d pu

blis

hing

.55

.55

Prop

riet

ary

prep

arat

ions

.72

.98

Toi

let p

repa

ratio

ns2.

301.

31

Roa

d m

achi

nery

.66

.55

Mot

or v

ehic

les

.52

.46

Scie

ntif

ic in

stru

men

ts.4

71.

48

Ret

ail a

utom

obile

s1.

22.6

5

Ret

ail g

roce

ries

1.06

.96

Abo

ve m

edia

n du

ring

nin

e ye

ars

Furn

iture

.32

.34

.22

.07

..

Job

prin

ting

.36

.21

.16

.17

.26

Pain

tsG

lass

.30

.52

.29

.26

.24

.09

.19

..

.35

.07

Tex

tile

mac

hine

ry.

..2

0.1

5.4

6.3

3

Toy

s.3

8.2

6.1

2.3

7.4

6

Mis

cella

neou

s sp

ecia

l man

ufac

turi

ngSe

e te

xt f

or e

xpla

natio

n..0

4.

..2

6.2

1.1

3

.61

2.03

2.44 .61

3.11 .2

3.1

8.8

5.5

62.

68

.50

1.64

.92

.92

2,91 .0

3.8

61.

261.

151.

01

.13

.87

.31

.35

1.48 .34

.45

1.00 .91

.55

.31

.48

.74

.71

1.63 .48

.93

1.35

1.71 .7

1

.22

1.38 .5

3.7

81.

30 .40

.66

1.36 .85

.74.

.61

1.17 .6

61.07

1.58 .4

4.4

81.31

.61

1.02

.46

.73

.86

.28

.17

.38

1.28

.29

.22

.24-

.74

.24

.28

.18

.85

.34

2.18 .5

8.4

9

.49

.23

.47

.29

.64

.49

.28

.09

.20

.66

.28

.10

.24


The six groups just discussed all maintained positionsabove the first quartile for either nine or ten years. Butwhat is the situation with respect to the groups that re-mained continuously, or almost continuously, below thethird quartile?

Only one industry, Meat Packing, persistently retainssufficiently large (negative) differentials to place it belowthe lowest quartile in all ten years.18 Regularly over theten-year period the rate of return in this group falls farshort of the median figure. For the ten-year period the ag-gregate rate of return in Meat Packing is 1.9 per cent ascompared with the median group's ten-year aggregate rateof 12.6 per cent.

Four groups remained in the lowest division in everyyear except one. Two are manufacturing groups, Beveragesand Castings and Forgings. The latter contains 99 corpora-tions with an aggregate capital of about $3.4 billion (an-nual average). On this capital it earns in most years lessthan 8 per cent. From 1923 to 1928, the degree to which itsearnings stand below the median changed hardly at all.

The other two industries that remained below the thirdquartile in all years except one are mining groups, Gas andOil Wells, and Metals (ferrous and other). In the case ofGas and Oil Wells, however, the regularly low earningsmay be more apparent than real because of the accountingpractices in the matter of depletion that are peculiar to thisfield.1° It is here somewhat difficult to distinguish the causesof low return due to the aleatory character of the industryfrom those due to the peculiar circumstances of accountingand valuation methods.

In the case of these negative differentials the 'median difference coef-ficient' is, of course, the amount by which the group rate falls short of themedian, divided by the median figure.

See Ch. 25, where so-called 'discovery depletion' and related mattersare discussed,


5. PERMANENT POSITIONS OF GROUPS ABOVE AND BELOWTHE MEDIAN

A similar analysis may be made of the comparative profit-ableness of those industries that remained above the medianin all years. This set of industries comprises ten groups. Ifwe average their several median difference coefficients, wefind that these ten industries regularly earned from aboutone and one-half to two times the median rate, and that,from 1924 to 1928, the differential scarcely narrowed at all.

If to these ten industries we add the seven that remainedabove the median in all years but one, we have 17 groupsin the 'almost permanently-above-the-median' category.

Take now those groups that remained below the medianin all years, as well as those which did so in all years butone. In the first category are seven groups, in the second,six. In some instances, for example, in Railway Equipmentand in Rubber, declining secular rates of growth in the in-dustry may partly account for the persistently poor show-ing, but in most instances such an obvious explanation isnot apparent. The seven groups that remain below themedian in all years average a 'median difference' which inmost years amounts to at least 50 per cent of the medianitself. Data for all groups discussed in this section appear inTables 14 and 15.

6. SPECIFIC PROGRESS OF THE LEADING GROUPS OF EACI-IYEAR

C,

• Another method of analysis, in our search for measuresof the permanence or non-permanence of differences inearnings rates, may now be employed. This is not to isolatethe industries that remain above or below any quartile or

TA

BL

E 1

5

ME

DIA

ND

IFFE

RE

NC

E C

OE

FFIC

IEN

TS

FOR

RE

LA

TIV

EL

YU

NPR

OFI

TA

BL

E I

ND

UST

RIE

S

GR

OU

P N

AM

E19

191920

1921

1922

1923

7924

7925

7926

1927

192

8— z

1.17

.98

1.99

.86

.73

.61

.74

.50

.79

.58

1.09

.99

.87

.76

.57

.74

.46

.48

.59

..

.46

..

1.06

.85

.52

.59

.54

.38

.59

.45

.76

..

.85

1.08

1.13

.89

.84

.69

.72

.47

.93

.91

1.32

.96

.78

.78

.62

.61

.64

..

1.17

.98

1.99

.86

.73

.61

.74

.50

.79

.58

Z1.

09.9

9.8

7.7

6.5

7.7

4.4

6.4

8.5

9.2

2.1

8.5

11.

76.4

3.3

5.1

5.4

5.2

7.3

5.1

5—

.46

.26

1.06

.85

.52

.59

.54

.38

.59

.45

Z.2

0.2

5.3

9.4

7.0

6.2

6.57

.39

.49

.69

.76

.23

.85

1.08

1.13

.89

.84

.69

.72

.47

.93

.91

1.32

.96

.78

.78

.62

.61

.64

.23

.10

.85

2.77

.74

.58

.36

..

.31

.37

.88

.38

..

.73

.51

.31

.08

.44

.47

.48

.54

•.

.62

.14

.15

.12

.35

.26

.45

.17

.24

.01

..

1.21

.32

.98

.40

.30

.38

.29

.22

.15

..

1.38

.27

.07

.07

.15

.33

.23

.07

•.

.67

.63

.15

.23

.05

.24

.38

.11

.33

Cl)

Below third quartile during ten years

Meat

pack

ing

Bel

owthird quartile during nine years

Beverages

Cas

tings

and

for

ging

sG

as a

nd o

il w

ells

Met

als

(min

ing)

Bel

owmedian during ten years

Meat

pack

ing

Bev

erag

esIv

lisce

llane

ous

food

pro

duct

sC

astin

gs a

nd f

orgi

ngs

Rai

lway

equ

ipm

ent

Gas

and

oil

wel

lsM

etal

s (m

inin

g)B

elow

median during nine years

Rubber

prod

ucts

Bla

nk p

aper

Who

lesa

le d

rygo

ods

Who

lesa

le h

ardw

are

Who

lesa

le p

aper

Who

lesa

le g

roce

ries

See

text

for

exp

lana

tion.


other division point throughout successive years, but to takethe highest or the lowest industries of each year and, what-ever their positions in succeeding years, ascertain the degreeto which they maintain or lose their advantage or disad-vantage.

Specifically, we take the 27 highest ranking industries in,say, 1919 (those above the first quartile) and tabulate theamount by which the return in each exceeds the median rate.We then total these differences and divide by the number ofindustries (27). This gives us the average amount by whichthe return of the 27 industries exceeds the median rate. Toillustrate, in 1919—a highly prosperous year—the medianreturn is 21.2 per cent. Some of the highest 27 industriesearn over 45 per cent, others around 30 per cent; they aver-age an earnings rate of 37.5 per cent as compared with themedian of 21.2 per cent. The average departure from themedian, as we have termed it in another connection earlierin this chapter, we may now, in accordance with the lan-guage of statistics, term the averaye deviation.20

But for purposes of comparison with the deviations pre-vailing in other years, such an average deviation can bestbe expressed as a ratio to the median itself. In this instancewe regard the difference of 16.29 not as 16.29 but as aboutthree-quarters (.77) of the median rate, 21.2. Again in thelanguage of statistics this figure of .77 we may term thecoefficient of variation. It expresses the degree to whichthe several rates of earnings in the highest industries

20Deviations may, of course, be based upon other averages than themedian as statistically conversant readers are aware. To such readersmuch of the explanation contained in this and other chapters of Book Iwill prove somewhat tedious because of its elementary character. But inview of the needs of other readers, some prolixity of explanation as wellas a relative simplicity of statistical methods seems desirable, at leastthroughout Book I. However, for the sake of accuracy of expression, it isperhaps wise to employ the terminology of statistics in most instances whererepeated use is made of the same statistical device.


vary, on an average, from the median rate. It will be notedthat this comparison is relative rather than absolute—themedian is divided into the average of the deviations fromit, and the resulting coefficient simply shows the proportionof the median which the average deviation is. For this rea-son, coefficients of average deviation afford better compari-sons, between data of different years, than do the absolutedeviations themselves, and will thus be employed through-out the remainder of this chapter.

We take, then, the industries that are the 27 highest inpoint of earnings in 1919, and analyze the earnings rates ofthose same industries in each of the succeeding nine years.If a tendency towards equality between different industrieswere effective, the coefficient of variation for these 27 indus-tries would steadily decline, and to the extent that thistendency operated perfectly, would approach zero. Chart 5shows that the coefficient does indeed decline at first; duringthe first five years the excessively large variation of these27 groups' earnings from the median rate is reduced. Theratio falls from .77 in 1919 to .33 in 1923. But from 1923through 1928 the average advantage of these 27 industriesis consistently maintained; their earnings are always from.33 to .46 in excess of the median figure (that is, are from1.33 to 1.46 times it). The course of these coefficients maybe traced in Chart 5, while Table 16 gives them in numericalform.

In similar fashion coefficients are computed for the 27industries that stood highest in 1920 (some of which werefound above the first quartile in 1919, others not), andthese same industries are then followed through the remain-ing years of the period.

Exactly the same analysis is applied to the highest indus-tries of 1921, 1922, 1923 and 1924 respectively. The re-suits of the several analyses, data for all of which are

-3

TA

BL

E 1

6

AV

ER

AG

ED

EV

IAT

ION

S A

ND

CO

EFF

ICIE

NT

S O

F V

AR

IAT

ION

FO

RPE

RC

EN

TA

GE

S O

F N

ET

IN

CO

ME

TO

CA

PIT

AL

IZA

TIO

N,

27 H

IGH

EST

MIN

OR

GR

OU

PS'

1919

1920

1921

1922

1923

1924

1925

1926

1927

1928

Ave

rage

devi

atio

n16

.289

8.86

35.

878

7.25

94.

737

5.24

15.

211

4.85

63.

930

4.23

0C

oeff

icie

nt.7

68.6

29.8

28.5

34.3

31.4

60.4

24.3

88.3

61.3

85

Ave

rage

dev

iatio

n12

.759

5.65

97.

000

5.54

14.

985

5.02

64.

833

4.31

94.

470

ZC

oeflI

cien

t.9

05.7

97.5

15.3

87.4

37.4

09.3

87.3

96.4

06

Ave

rage

dev

iatio

n7.

900

8.70

75.

081

5.60

45.

244

5.74

85.

281

5.41

9C

oeff

icie

nt1.

113

.640

.355

.492

.426

.460

.484

.493

Ave

rage

dev

iatio

n10

.741

5.77

85.

930

5.93

35.

719

5.07

04.

978

Coe

ffic

ient

.790

.404

.520

.482

.458

.465

.453

—3

Ave

rage

dev

iatio

n7.

111

5.96

76.

311

5.80

74.

770

5.25

2C

oeff

icie

nt.4

97.5

23.5

13.4

65.4

38.4

77t!

3C

,,

Ave

rage

devi

atio

n8.

137

6.94

86.

341

6.02

25.

896

Coe

ffic

ient

.714

.565

.507

.552

.536

See

text

for

exp

lana

tion.

"C "C


shown in Chart 5 and Table 16, are identical in that theearnings differentials are in all cases permanent. In thethreeseries starting before 1922 a marked decline in thecoefficient takes place after 1921, but in nearly all of theseries the coefficients exhibit no appreciable fall from 1923on.

That the coefficients in several instances do show a sub-stantial decline after the first year or two might seem toindicate that competition, while not serving to equalize profitrates, reduces excessive discrepancies. But a careful ex-amination of these figures, and other rather complicateddata related to them, reveals that two fortuitous factorsare mainly responsible for the initial, but really specious,declines in the coefficients. The first is the presence, in anygiven year, of a few very high earnings rates enjoyed byindustries that are not persistently 'big money-makers'. Thesecond is the excessively low average or median rate ofearnings in 1921. A coefficient of variation affords a bettermeasure of relative disparities than does the mere averagedeviation itself, but the coefficient may nevertheless be un-stable and misleading if considered apart from fluctuationsof the average on which it is based. In this case, that aver-age is the median return for our 106 groups, and this figurefell from 21.2 in 1919 and 17.9 in 1920 to 7.1 per cent in1921. It then jumped to 13.6 per cent in 1922.

In other words a large part of the explanation of theinitial fall in the coefficients of Table 16 is that a year ofdepression such as 1921 brings about temporarily greaterrelative differentials in the standing of profitable industriesrelative to the median group because the latter in such ayear shows a very much lower figure.21 When recovery oc-curs the following year, the median return rises more sharply

21 Mrs. John D. Sumner has given valuable criticism and aid in the inter-pretation of the misleading initial declines of these coefficients.


than the earnings rates of the most profitable groups (ingeneral) ; and the relative differential between the medianand the rates of the 'high groups' is reduced—not becausethe earnings rates of the latter (in general) decline but be-cause the median rate rises. Competition does not greatlyreduce the earning power of the high groups, but generalcyclical improvement disproportionately enhances that ofthe others. This is evident if one glances at Table 17. Themedian increased from 7.1 in 1921 to 13.6 in 1922, ornearly doubled. (Similarly, the arithmetic mean of the 106groups jumped from 4.4 to 10.6.) The average earningsrates of the 27 groups which were highest in 1921, however,increased only from 15.0 to 21.8, or by less than one-half.

As final evidence that the initial declines in the coefficientsare largely spurious, so far as signifying any very effectivetendency towards an equality of profit rates is concerned, wemay 'run back' all of our computations into earlier years,that is, in the case of the ten industries that were highestin 1920, calculate their average deviations and coefficientsfor 1919; similarly, in the case of the highest industries for1921, calculate their coefficients for 1920 and 1919, etc.\Vhen this is done, the 1919 coefficients for all series aremuch smaller than those for 1921; most of them are roughlyhalf as large. And, roughly again, all of the 1919 coefficientsare the same size as the 1928 figures. In other words, nomarked reduction in the average deviation of any of theseseries from the median, in relative terms, took place overthe ten years in question. Table 18 repeats the data ofTable 16, but contains, in addition, data for the years pre-ceding that in which the industries of each series are theten highest. The figures for that year—the year of originfor each series, as it were—are shown in italics.

We are thus forced to conclude that the industries inwhich earnings exceed the median return by the largest rela-

TA

BL

E 1

7

AV

ER

AG

EE

AR

NIN

GS

RA

TE

, 27

HIG

HE

ST M

INO

R G

RO

UPS

IN

EA

CH

YE

AR

191

9-28

, TR

AC

ED

TH

RO

UG

H S

UC

CE

SSIV

EY

EA

RS

AN

D R

UN

BA

CK

TH

RO

UG

H 1

919

(Ari

thm

etic

mea

n an

d m

edia

n fi

gure

s sh

own

firs

t are

for

all 1

06 g

roup

s)19

191920

1921

1922

1923

1924

1925

1926

1927

1928

2919 Arithmetic

mea

n17

.1M

edia

n21

.2A

vera

ge r

ate,

27

grou

ps37

.517

.99.

619

.416

.913

.615

.714

.613

.813

.1

1920 A

rith

met

ic m

ean

11.6

Med

ian

14.1

Ave

rage

rat

e, 2

7 gr

oups

27.0

26.9

8.7

17.9

17.3

14.2

15.1

14.5

12.5

13.1

1921 A

rith

met

ic m

ean

4.4

Med

ian

7.1

Ave

rage

rat

e, 2

7.gr

oups

27.8

16.9

15.0

21.8

18.1

15.5

16.2

15.7

14.6

14.2

1922 A

rith

met

ic m

ean

10.6

Med

ian

13.6

Ave

rage

rat

e, 2

7 gr

oups

29.8

17.8

13.4

24.3

19.7

17.0

17.2

16.8

15.0

14.8

1923

0A

rith

met

ic m

ean

10.9

Med

ian

14.3

I.A

vera

gera

te, 2

7 gr

oups

27.8

20.4

11.6

22.7

21.4

17.2

17.7

17.2

14.8

14.6

1924 A

rithm

etic

mea

n9.

6M

edia

n11

.4A

vera

ge r

ate,

27

grou

ps27

.417

.811

.720

.819

.719

.518

.818

.616

.316

.2

1925 A

rithm

etic

mea

n11

.3Z

Med

ian

12.3

Ave

rage

rat

e, 2

7 gr

oups

27.0

16.7

10.5

21.1

19.1

18.0

19.7

19.0

16.7

16.3

1926 A

rithm

etic

mea

n11

.5M

edia

n12

.5A

vera

ge r

ate,

27

grou

ps27

.216

.410

.020

.318

.817

.118

.619

.617

.117

.0

1927 A

rithm

etic

mea

n9.

8M

edia

n10

.9A

vera

ge r

ate,

27

grou

ps27

.017

.010

.920

.018

.517

.219

.019

.218

.317

.5

1928 A

rithm

etic

mea

n10

.7M

edia

nij.

oA

vera

ge r

ate,

27

grou

ps26

.815

.99.

312

.617

.616

.618

.419

.017

.218

.2— t1 C

',

Ave

rage

dev

iatio

nC

oeff

icie

nt

TABLE 18

AVERAGE

DEVIATIONS AND COEFFICIENTS OF VARIATION FOR

PERCENTAGES

OF

NE

T I

NC

OM

E T

O C

API

TA

LIZ

AT

ION

,27

HIG

HE

ST M

INO

R G

RO

UPS

1

1919

1920

1921

1922

1923

1924

1925

1926

1927

1928

-4 z U) 0 0)

16.2

898.

863

5.87

87.

259

4.73

75.

241

5.21

14.

856

3.93

04.

230

.768

.629

.828

.534

.331

.460

.424

.388

.361

.385

Ave

rage

dev

iatio

nC

oeff

icie

nt10

.322

.487

.12.

759

.905

5.65

9.7

977.

000

.515

5.54

1.3

874.

935

.437

5.02

6.4

094.

833

.387

4.31

9.3

964.

470

.406

Ave

rage

dev

iatio

nC

oeff

icie

nt9.

422

.444

6.83

7.4

857.

900

1.11

38.

707

.640

5.08

1.3

555.

604

.492

5.24

4.4

265.

748

.460

5.28

1.4

845.

419

.493

Ave

rage

dev

iatio

nC

oeff

icie

nt11

.041

.521

7.20

4.5

116.

693

.943

10.7

41.7

905.

778

.404

5.93

0.5

205.

933

.482

5.71

9.4

585.

070

.465

4.97

8.4

53

Ave

rage

dev

iatio

nC

oeff

icie

nt10

.096

.476

7.75

2.5

506.

163

.868

9.07

8.6

687.

11.1

.497

5.96

7.5

236.

311

.513

5.80

7.4

654.

770

.438

5.25

2.4

77

See

text

for

exp

lana

tion.

Ave

rage

dev

iatio

n9.

152

9.08

55.

681

7.85

65.

763

8.13

76.

948

6.34

16.

022

5.89

6C

oeff

icie

nt.4

32.6

44.8

00.5

78.4

03.7

.14

.565

.507

.552

.536


tive amounts in any one year lose no significant part of theirhigh relative earning power over a six-, eight.. or even ten-year period. Indeed except for such temporary 'changes inaverage levels of earnings as accompany a year of markeddepression, they retain a rather,constant measure of advan-tage. This dictum does not necessarily apply to every indi-vidual industry that enjoys a large advantage in a particularyear, but is based upon averages illustrative of the generalsituation.22

Much the same sort of process is at work in the case ofthe industries which stand, not highest, but lowest in eachyear. Taking the 27 minor groups standing under the lowestquartile in 1919, we find their coefficient of variation to be.50. By 1922 it falls to .37, but it continues regularly tostand at about that figure. Table 19 presents the data forthe 1919 'low' groups and those starting with other years.

In the second and third of these cases, those of the indus-tries standing lowest in 1920 and '1921 respectively, thereis a sharp fall in the coefficient after one or two years—that is, the average return in the 'low' industries risesgreatly. But for the remainder of the period this. returnremains definitely and steadily below the median rates. Inthe case of the last three series, those starting in 1922, 1923and 1924 respectively, the fall (denoting an increase in therelative prosperity of the 'low' industries) is not somarked. Perhaps this is because 1922—24 were not asextreme years, cyclically speaking, after which readjust-ments could be expected to take place, as were 1919—21.

22 Although these coefficients are derived from arithmetic means, theyare not as non-typical as most arithmetic averages. It is to be noted thatthe selection of the items is such that all the components of each 'series standabove the first quartile in the initial year. Thus the mean of the deviationsfrom the median is not so much influenced by the presence of extremevariants from the central figure, or by an extreme range, as it would bewere all items of the entire range employed as components—as was donein some earlier sections of this chapter.

TA

BL

E 1

9

AV

ER

AG

E D

EV

IAT

ION

S A

ND

PER

CE

NT

AG

ES

OF

NE

TC

OE

FFIC

IEN

TS

OF

VA

RIA

TIO

N F

OR

IN C

OM

E T

O C

API

TA

LIZ

AT

ION

,27

LO

WE

ST M

INO

R G

RO

UPS

'

Ave

rage

dev

iatio

nC

oeff

icie

nt9.

674

.686

5.28

5.7

444.

544

.334

4.20

4.2

943.

915

.343

3.58

1.2

914.

252

.340

3.82

6.3

514.

022

.366

Ave

rage

dev

iatio

n8.

763

5.73

74.

993

3.57

04.

185

3.81

93.

504

3.85

6C

oeff

icie

nt1.

234

.422

.31-

9.3

13.3

40.3

06.3

21.3

51

Ave

rage

dev

iatio

n6.

896

5.58

93.

985

4.03

73.

930

3.69

33.

852

Coe

ffic

ient

.507

.391

.350

.328

.314

.339

.350

1919

1920

1921

1922

1923

1924

1925

1926

1927

1928

Average

devi

atio

n10

.530

6.55

64.

019

5.00

44.

448

3.71

93.

959

4.11

94.

081

3.61

5C

oeff

icie

nt.4

97.4

65.5

66.3

68.3

11.3

26.3

22.3

30.3

74.3

29

Ave

rage

dev

iatio

nC

oeff

icie

nt

Ave

rage

dev

iatio

nC

oeff

icie

nt

See

text

for

exp

lana

tion.

z Cl,

6.48

54.

207

4.14

14.

122

3.30

43.

622

.453

.369

.337

.330

.303

.329

5.36

74.

470

4.88

13.

589

4.21

5.4

71.3

63.3

90.3

29.3

83

0 Cl,


Our conclusion here is thus more tempered than in the caseof the highest ranking industries above discussed, but itremains true that the industries showing low earnings inany one year, while tending to lose some of their disadvan-tage, still permanently retain a substantial and rather con-stant measure of disadvantage.

If, as corroborative evidence of a more extreme sort, weperform similar analyses upon, not the 27 highest or lowestof our 106 industries, but upon the ten highest and lowestgroups of each year, we again obtain approximately thesame results. These are presented in Tables 20 and 21.The coefficients are, of course, larger. The initial declinesthat occur are sharper. But even towards the ends of theseveral spans of years in question the coefficients never gobelow .24, and the advantage or disadvantage of both setsof groups is permanent for the period. Here again the sharpinitial declines suggest that competitive influences, cyclicalreadjustments, technological changes, and consumptionshifts all do indeed force some reduction of extreme differ-entials, whether positive or negative. But the regular per-sistence of the coefficients above zero also affords clearevidence that competitive influences do not serve to bringabout equality, over the period of years under survey, indifferent industries. Whether any greater tendency towardssuch an approximate equality would assert itself over longerthan a ten-year period cannot, of course, be told.

Undoubtedly the highly specialized character of muchfixed capital in modern industry, making it difficult if notimpossible to withdraw all of the capital of an enterpriseover a ten-year period, is one factor that prevents whatever'tendencies towards equality' that may exist from becom-ing effective. Many enterprises that do not earn returnsat rates that accord with the expectations of their ownersstill continue in business, for a smaller return is better than

TA

BL

E 2

0

AV

ER

AG

ED

EV

IAT

ION

S A

ND

CO

EFF

ICIE

NT

S O

F V

AR

IAT

ION

FO

RPE

RC

EN

TA

GE

S O

F N

ET

IN

CO

ME

TO

CA

PIT

AL

IZA

TIO

N,

10 H

IGH

EST

MIN

OR

GR

OU

PS1

1919

1920

1921

1922

1923

1924

1925

1926

1927

1928

00 LJ

24.2

57.

515.

089.76

5.83

6.89

7.02

5.40

4.46

4.54

1.14

4.5

33.7

15.7

18.4

08.6

04.5

71.4

32.4

09.4

13

Ave

rage

dev

iatio

nC

oeff

icie

nt19

.86

1.40

96.

73 .948

8.46 .622

6.15 .4

305.

26 .461

5.38 .4

375.

07 .406

4.71 .4

324.

41 .401

Ave

rage

dev

iatio

nC

oeff

icie

nt12

.96

1.82

514

.88

1.09

47.

47 .522

8.02 .7

046.

06 .493

7.49 .5

997.

61 .698

6.90 .6

27

Ave

rage

dev

iatio

nC

oeff

icie

nt18

.21

1.33

910

.02

.701

9.35 .8

2010

.02

.815

9.43 .7

548.

38 .769

8.79 .7

99

Ave

rage

dev

iatio

nC

oeff

icie

nt11

.27

.788

9.41 .8

258.

79 .715

8.08 .6

466.

69 .614

7.34 .6

67

Ave

rage

dev

iatio

n12

.24

8.83

8.40

8.22

8.02

1.07

4.7

18.6

72.7

54.7

291

See

text

for

exp

lana

tion.

Ave

rage

dev

iatio

nC

oeff

icie

nt

Coe

ffic

ient

-I z CI, 0

TA

BL

E 2

1

AV

ER

AG

ED

EV

IAT

ION

S A

ND

CO

EFF

ICIE

NT

S O

F V

AR

IAT

ION

FO

RPE

RC

EN

TA

GE

S O

F N

ET

IN

CO

ME

TO

CA

PIT

AL

IZA

TIO

N,

10 L

OW

EST

MIN

OR

GR

OU

PS

z :1

1923

1924

1925

1926

1927

1928

8.00

6.15

6.60

5.80

5.49

3.77

.559

.539

.537

.464

.504

.343

See

text

for

exp

lana

tion.

1919

1920

1921

1922

Ave

rage

dev

iatio

n15

.25

8.19

5.69

8.07

Ave

rage

dev

iatio

nC

oeff

icie

nt

Ave

rage

dev

iatio

n14

.87

7.94

6.30

6.07

6.62

5.68

6.67

5.96

5.87

Coe

ffic

ient

1.05

51.

118

.463

.424

.581

.462

.534

.547

.534

12.8

86.

925.

433.

24.

3.66

3.71

2.62

3.70

1.81

4.5

09.3

80.2

84.

.298

.297

.240

.336

Ave

rage

dev

iatio

nC

oeff

icie

nt10

.35

.761

8.31 .5

815.

69 .499

5.86 .4

765.

18 .414

5.02

.461

4.30 .3

91

Ave

rage

dev

iatio

n9.

866.

145.

955.

344.

843.

89C

oeff

icie

nt.6

90.5

39.4

84.4

27.4

44.3

54

Ave

rage

dev

iatio

n8.

006.

785.

064.

51C

oeff

icie

nt.7

02.5

68.5

42.4

64.4

10

-4 z C,, '-3 -'-I


none at all; were production entirely abandoned, all fixedcapital investment of a specialized character might be com-pletely lost. In extreme cases operations may be continuedfor considerable periods even though the product is soldat an absolute loss, if prices that repay something morethan prime costs (dii-ect labor and material) are realized.

Then, too, under the corporate form of enterprise, re-organization is far more frequent than liquidation. Oldenterprises not discontinued in one field and new onesreadily started in another. The old shell—the frameworkof the enterprise—continues; but policies are drasticallyaltered, methods are changed, executive personnel is re-placed. The corporation may remain in the same industry,but its activities are reorganized and reoriented. New typesof product within the same general field may be developed;and these new products or methods may prove no moresuccessful competitively than their predecessors. At allevents, it is often a long time before an unsuccessful cor-poration gives up the struggle entirely. Even then, the enter-prise frequently disappears in name only; if the establish-ment is of any size, it may be absorbsome other concern.

When reorganizations or mergers of this sort take place,

ed by, or merged with,

'capital values' are written down on the books of the un-successful corporation, or the fact of failure to earn asatisfactory return is recognized by the payment of farsmaller amounts, in cash or stock, for the securities of theunsuccessful companies, than the sums that have been in-vested in them. In other words, declines in capital valuesare often formally recognized by owners or purchasers ofunsuccessful enterprises, but the physical plant is not aban-doned or even necessarily devoted to other uses. It maycontinue to produce the same products, but since book in-vestment has been lowered, the same earnings as before will

of competition—thedesigns of products,

sorts—serves also toequality is not' realized.

value—the standard ex-planation of the prices of individual commodities—assumedthat the price of an article could not long remain muchbelow cost of production because capital would be with-drawn from the field of enterprise, so did it also assumethat new producers would soon enter the field were pricelong above cost of manufacture. It was thus tacitly assumedthat what one producer could do, another could and woulddo; that identical products, in other words, are offered bydifferent producers competing for the consumer's favor.The manufacture and sale of modern trade-marked productsand specialty goods, however, do not fit this description.Great numbers of consumer goods commonly used today—automobiles, radios, vacuum cleaners, washing machines,electric cooking devices, furniture, even many articles ofclothing and numerous food specialties—are given highlydistinctive characters by virtue of their differentiated de-signs and particular specifications. The same is true of some


show as a higher rate of return on the capital as currentlyvalued. Practices such as these do not affect our figures forindustries of low return in the sense that the discrepancybetween them and the, median industry is accentuated, for,as just stated, the 'new' rate of return after capital valuesare written down is higher than the old and tends to raisethe apparent earnings level of the industry in question. Butsuch practices do throw light upon the reasons why ten-yeardifferences in earnings rates prevail in spite of such recapi-talizations: the recapitalization device is utilized justbecause large, specialized corporate capitals usually can notbe withdrawn in the physical sense.

Finally, the imperfect characterprevalence of trade-marks, specialquasi-monopoly advantages of allexplain why the tendency towardsJust as the classical doctrine of


producers' goods—machinery of different types designedto perform the same kinds of work. So long as differentsets of purchasers either see or are led to believe they seeadvantages of the one product over the other, the marketsfor such products will not be fully competitive in the ortho-dox sense of that term. And competition, while preventingthe very large net revenues that might accrue under condi-tions of real monopoly, will not necessarily equalize returnsbetween either industries or enterprises.

The character of modern manufacturing and merchandis-ing, therefore, furnishes some explanation of the lack ofuniformity in the average profit rates of large corporations,when classed by industries, for two important conditionsessential to perfect competition—a quick mobility of capitaland a homogeneity of competing products—are absent ina large proportion of industries. To be sure, a substantialmobility of capital exists over a period of time. But, on theside of withdrawal, this mobility exists in the possibility ofnot creating new plant and organizations of particulartypes rather than in withdrawing from service those thatat any given time are under way. We do not thus undertaketo say that the classical doctrine of equality of profit ratesis invalid as a 'tendency'; but we do say that under modernconditions, as they existed in 1919—28, the evidence seemsclear that the tendency was scarcely effective over this ten-year period.23

23 Whether or not a 'long-run tendency' ought to show an effect over aten-year period depends upon one's definition of 'long-run'. The writer'sfeeling is that a fairly considerable effect ought to be noted within tenyears if over, say, fifteen or twenty years the tendency really works out asalleged. But present data relate to ten years only.

Probably the persistence in an individual industry of an earnings ratefar lower than the median figure would indeed not continue for severaldecades if the rate fell much short of the 'normal' interest rate or the costof borrowed capital. Nor might an earnings rate in an individual industry,even if equal to this cost, persist for several decades at a level far belowthe median—assuming that the median 'were knav.'n. Either or both of these


situations might cause the withdrawal of capital or its 'non-renewal' inone industry, and its investment in another. To grant deductively that sec-ular trends in one industry may serve to cause the eventual withdrawal ofcapital is one thing; but to allege that this process results in an equalityof the rate of earnings upon capital beiween industries over a period isquite another. Sometimes it results, as in the case of wagon manufacture, ina gradual collapse of the entire industry. And returns in the successor in-dustry can hardly be added to losses of the capital employed in the predeces-sor and 'averaged' so as to see if equality with other industries over aquarter-century or more perhaps results.

Statistically it would be interesting to have data for twenty or thirtyyears running, but the problems of classification and analysis would bemore difficult, and the results in some ways accordingly less significant,than with data covering only a ten-year period. See Ch. 46 on industrialclassification; see also Ch. 47, where the question of the non-renewal ofcapital is again touched upon.

earnings rates in different industries · 87 drygoods 29.110 investments, various 88 groceries 59...

Documents