Full text of Summary of Proceedings of Conference on Productivity, October 28-29, 1946 : Bulletin of the United States Bureau of Labor Statistics, No. 913

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

UNITED STATES DEPARTMENT OE LABOR
L. B. SCHWELLENBACH, Secretary
BUREAU OF LABOR STATISTICS
E w an C lague , Commissioner

Summary of Proceedings
of
Conference on Productivity
October 28-29, 1946

Bulletin No. 913

For sale by the Superintendent of Documents, U. S. Government Printing Office
Washington 25, D. C. - Price 15 cents






L etter o f T ran sm ittal

U nited S tates D epartment of L abor,
B ureau of L abor S tatistics ,

Washington, D. C,f May 1, 1947.

T he S ecretary of L abor :

I have the honor to transmit herewith a report which summarizes the
Proceedings of the Conference on Productivity, held in Washington on
October 28 and 29, 1946. This summary of the voluminous proceedings
was prepared by Mrs. Celia Star Gody, of the Bureau's staff, and was
submitted to the participants in the conference, for their editorial ap­
proval. The report also includes an evaluation of the conference by
Solomon Fabricant, of the National Bureau of Economic Research, who
acted as chairman of the executive committee which planned the details
and agenda of the conference.
The conference brought together an outstanding group of research
workers from labor, industry, government, and the universities for a
mutual exchange of views on a topic of great current importance. The
conference may confidently be expected to have a beneficial effect on the
content and direction of research activities in the field of productivity for
many years to come. This report is designed to make generally available
the discussions and results of the conference.
E w an C lague , Commissioner.
H on . L. B. S chw ellenbach ,




Secretary of Labor.

11!




Bulletin No. 913 of the
United States Bureau of Labor Statistics

Preface
After the close of the war, in 1945, increasing attention was given in
public discussion to the subject of productivity, and in some instances
sharp controversies arose. Some differences of opinion seemed rooted
in divergent interpretations of scanty data; others could apparently be
traced to such basic difficulties as lack of agreement on the meaning and
significance which should be attached to the term productivity. Many
persons felt that under these circumstances a mutual exchange of views
on the conceptual and measurement problems in the field of productivity
might help materially to substitute agreement for argument.
The Bureau of Labor Statistics of the U. S. Department of Labor
and the Division of Statistical Standards of the U. S. Bureau of the
Budget joined in sponsoring a Conference on Productivity to be attended
by representatives of labor, industry, private research groups, and gov­
ernment agencies. The agenda and details of the conference were planned
by an executive committee which included members representing labor
and industry, selected with the advice and assistance of the major labor
and business organizations.
The conference was held on October 28 and 29, 1946, at Washington,
D. C. It was attended by about 100 active participants and approximately
an equal number of observers.
This bulletin is intended to make available to others the ideas which
were developed during the two days of the conference. The following
pages present an evaluation of the conference by Mr. Solomon Fabricant
of the National Bureau of Economic Research, who acted as chairman
of the executive committee; the official program of the conference; and
a digest of the proceedings of the conference. The digest was prepared
by Mrs. Celia Star Gody of the Bureau of Labor Statistics, a member
of the executive committee. This digest was submitted to the participants
in the conference for their editorial approval.
The membership of the executive committee which made arrangements
for the conference was as follows:
Henry B. Arthur, Swift & Co.
Solomon Barkin, Textile Workers Union of America, CIO.



George Brown, United Association of Plumbers and Steamfitters, AFL.
Katherine P. Ellickson, Congress of Industrial Organizations.
W. Duane Evans (Secretary), U. S. Bureau of Labor Statistics.
Solomon Fabricant (Chairman), National Bureau of Economic
Research.
Celia Star Gody, U. S. Bureau of Labor Statistics.
Everett Hagen, National Planning Association.
Douglas H. Holmes, Westinghouse Electric & Manufacturing Co.
Thomas Mills, U. S. Bureau of the Budget.
Margaret Scattergood, American Federation of Labor.
Ernst Swanson, U. S. Chamber of Commerce.
Samuel Thompson, U. S. Department of Commerce.
Charles Young, Westinghouse Electric & Manufacturing Co.




vi

Contents
Page

P refa ce....................................................................................................................................... iii
An evaluation of the Productivity Conference, by Solomon Fabricant................... 1

OCTOBER 28—MORNING SESSION

Chairman, Leon Henderson, Research Institute of America

Welcome to conferees, by Stuart A. Rice........................................................................
Scope of conference, by Ewan Clague.............................................................................
Productivity and its general economic setting, by Robert Nathan...........................
Problem of concepts and measurements in the field of productivity, by
Hiram D a v is..................................
Concepts and measures of productivity at the job level:
Martin Gainsbrugh .........................................................
Nathan Spero ...................................................................................................................
Discussion:
Solomon B arkin...... ........................................................................................................
H. B. Maynard ...............................................................................................................
Lazare Teper ...................................................................................................................
Rufus Tucker .................................................................................................................
Ewan Clague ...................................................................................................................
Gardner Means ...............................................................................................................
Hiram D a v is.....................................................................................................................
Duane Evans ...................................................................................................................
Robert W. B urgess......................

OCTOBER 28—AFTERNOON SESSION

9
9
10
11
12
12
13
13
14
14
14
15
15
15
15

Chairman, Robert Nathan, Robert Nathan Associates

Concepts and measures of productivity at the plant and company levels:
Benjamin Haskell ........................................................................................................... 16
H. B. M aynard............................................
17
Charles E. Young ......................................................................................................... 18
Concepts and measures of productivity at the industry level:
Solomon Barkin ....................................................................................................
19
Charles E. Young ........................................................................................................... 19
John D. Gill ..................................................................................................................... 20
Discussion:
Nathan Fine ..................................................................................................................... 21
Charles E. Y o u n g ........................................................................................................... 21
Harry M agdoff................................................................................................................. 21
John D. Gill ........................................................................................................................22
Solomon Barkin ..................................................................................................
22
Everett Hagen ................................................................................................................. 22
Duane Evans ................................................................................................................... 22

OCTOBER 28—EVENING SESSION

Chairman, George W. Taylor, University of Pennsylvania

Concepts and measures of productivity at the national level:
Marion Hedges ............................................................................................................... 23
Everett Hagen .......................................................
24
Robert W. B urgess......................................................................................................... 25




vii

Page

Discussion:
Duane Evans ......................................................................................................................
Solomon Barkin ...............................................................................................................
Everett Hagen ...................................................................................................................
Concepts and measures of productivity at the international level:
Julius Hirsch .....................................................................................................................
Charles M erw in...........................................................................................
Hans Staehle .....................................................................................................................
Discussion:
Prof. J. R. Hicks ...........................................................................................................
Solomon Fabricant ...........................................................................................................
Charles Roos .....................................................................................................................
E. J. Riches .......................................................................................................................

26
27
27
27
29
31
31
31
31

OCTOBER 29—MORNING SESSION

Chairman, Isador Lubin, President of the American Statistical Association
Scope and limitations of existing measures of productivity:
Duane Evans .....................................................................................................................
Katherine Pollack Ellickson...........................................................................................
Rosalind Schulman ...........................................................................................................
Andrew Court ...................................................................................................................
Discussion:
Boris Stern .....................................................................................................................
Frank R. Garfield ...........................................................................................................
Solomon Fabricant .........................................................................................................
John D. Gill .....................................................................................................................
Martin Gainsbrugh .........................................................................................................
Celia Star Gody ............ ................................................................................................
Duane E van s.....................................................................................................................

39
39
40
40
41
41
42

OCTOBER 29—AFTERNOON SESSION

Chairman, Thomas Blaisdell. U. S. Department of State
Need for additional productivity measures, by Lazare T eper..................................
Presentation of productivity measurements, by Solomon Fabricant.........................
Discussion:
Irving H. Siegel .............................................................................................................
George Brown .................................................................................................................
Lyle Cooper ................................................................................................ ....................
Henry B. A rth ur........................................................................................ ....................
Solomon Barkin .............................................................................................................
Margaret Scattergood ..................................................................................................
Everett Hagen .................................................................................................................
Hiram Davis ...................................................................................................................
Duane E van s.....................................................................................................................
R. H. Bookbinder .........................................................................................................
Ewan Clague .................................................................................................................
Summary of the conference, by Samuel H. Thompson................................................




32
34
37
38

viii

43
46
47
48
48
48
49
49
49
50
50
50
51

51

29

CONFERENCE ON PRODUCTIVITY,
OCTOBER 28-29, 1946
An Evaluation of the Productivity Conference

The aims of the conference, as set forth by the executive committee
for the conference, in inviting participation, were as follows:

The subject of productivity is generally recognized to be of unparalleled current
importance. It is directly related to the employment and unemployment prob­
lem. It is at the center of any discussions relating to wages, prices, and profits. It
affects the ability of the United States to maintain its position in international trade.
It is the means whereby a constant improvement in the living standards of American
families is achieved. Despite the importance of productivity, there is lack of agree­
ment on the concepts and measures which are appropriate to its evaluation and on
the implications of current measures.
In this situation, a technical conference is planned as a forum for the discussion
of productivity concepts and measurements, in the hope that this will clarify and
resolve genuine differences in viewpoint which now exist in the field. The conference
will also provide a vehicle for discussing the current state of knowledge regarding
productivity and labor cost, the adequacy of present programs, and related questions.
The conference is sponsored jointly by the Bureau of Labor Statistics of the
U. S. Department of Labor and the Division of Statistical Standards of the U. S.
Bureau of the Budget. The conference is not intended to act as an official advisory
body, but rather to be a discussion meeting. Papers will be presented by invitation
on selected subjects, but there will be ample provision for general discussion. Further
conferences may be scheduled, depending on the wishes of the participants.
A limited number of participants will be invited from labor, industry, government,
private research agencies, and academic institutions, and will be present in their
individual capacities, not as official representatives of organizations. The participants
will be selected from among those who may be expected to contribute to the discus­
sion of productivity.

Parents are always a little startled to find their infants grown up and
with a bulk and personality of mature proportions. The sponsoring
agencies and the executive committee were similarly surprised to dis­
cover what had come out of this invitation. Instead of a small group of
technicians mulling over some rather abstruse concepts and measurements,
there assembled over 200 people, all told, from labor, business, govern­
ment, research, and the academic world, engaging enthusiastically in a
broad discussion. Interest was fully as great at the end as at the beginning
of the two solid days of discussion.
Yet it should not have surprised us too much. The American people
have come tQ see the value of a wide and firm basis of accepted fact
and mutual understanding as a common meeting ground. They know that
74 5867— 17---- 2




0)

2
it is the only sure foundation for threshing out common problems. They
realize that concepts and measurements are essential in the building of
this foundation, and therefore that concepts and measurements are well
worth discussion. The conference provided a forum for a free discussion
and demonstrated the value of joining persons of diverse points of view
in it. This was a real accomplishment, perhaps the outstanding accom­
plishment of the conference. It showed how wrong are those who stress
only differences and clashes of interest, who deny the possibility or value
of meeting on a common ground of accepted fact and standard usage,
who see all things simply as matters of opinion.
The conference of course also yielded fruits of immediate bearing on
the question at issue, namely, concepts and measurements of productivity.
The abstract of the proceedings which follows summarizes fairly exten­
sively the successive statements and remarks of the various participants,
and clearly indicates the value of the discussion. The reader will want to
study this record with the care it deserves. It is desirable, however, to
preface the abstract by a review of and commentary on what came out of
the conference, along other than chronological lines; and to indicate, at
the same time, plans for the future work declared so necessary by all who
attended.

Definitions and Measurements of Productivity
First of all, the discussion clearly brought out how general is the term
“productivity.” The term relates to a whole family of concepts rather
than to any specific member of that family. Almost any comparison of
output with input is covered by it. Output may be defined in various ways,
however, and input may be measured by one factor or another or by
several factors in combination. Further, productivity may be the relation
between output and the input of one factor, all other input factors being
kept constant, as in an experiment under controlled conditions; or it may
be the relation between output and the input of one factor, with changes
occurring in all other factors. Still further, productivity may be the rela­
tion between the total output and input of a period, or it may be the rela­
tion between the increment in output associated with the addition of one
unit of a given factor of production; that is, it may be “average” or
“marginal.” Thus, economists writing on the “theory of marginal pro­
ductivity” use the term quite differently from statisticians who compute
indexes of productivity. Even statistical measures of productivity may
differ among themselves, fundamentally, as well as in detail. To avoid
confusion, therefore, it was agreed that the term “productivity” be used
only in a general way to describe the area with which the conference was
concerned, and to use more specific terms in referring to particular con­



3
cepts or measurements. “Physical output per man-hour,” for example, is
better than “labor productivity’" as a term descriptive of most currently
published indexes. Since no reasonably short term can be completely
descriptive, however, it was felt that terminology should always be defined
as explicitly as possible whenever used.
Stemming from the acceptance that there is a whole group of produc­
tivity concepts was the realization that a variety of measures is possible.
Measures may be made of output per unit of capital equipment or per
unit of labor; of output per man or per man-hour; of physical output
per man-hour or value output per man-hour; of “net” physical output
per man-hour or “gross” physical output per man-hour; and so on.
There is no way of deciding which measure is appropriate, except with
relation to a particular purpose. For different purposes different measures
may be suitable. Even for a particular purpose, several different measures
may be better than one; they may complement, rather than compete with,
one another. With a variety of measures possible and desirable, it is of
course all the more necessary to label each carefully and define each
accurately. Just what these purposes are and what indexes are most
suitable for particular uses remain subjects for further investigation.
Here the conference pointed up a job to be done.
Considerable attention was given to currently prepared measures of
output per man-hour, and also to some historical measures; and various
questions were raised concerning their scope and accuracy. It goes with­
out saying, of course, that no current or even historical statistical index
can hope to attain complete coverage and perfect accuracy, but there are
degrees of accuracy and coverage and it is well to strive for the highest
in both. The discussion and critical comments brought out at the con­
ference will, I am sure, stimulate even greater efforts on the part of
compiling agencies to better their indexes. Almost as important, it will
lead them to indicate as fully as possible the margins of error and the
limitations of representation inevitably still resident in their measures.
A number of the criticisms raised at the conference had already been con­
sidered by the compiling agencies and had been determined to be of small
significance. However, the discussion taught one clear lesson: results of
such considerations and evaluations of limitations should be made public,
preferably in advance of the appearance of criticisms. The range of
doubts that beset consumers of statistics concerning their validity must
be narrowed.
Many of the points raised in criticism were, in essence, that alternative
methods and measures are possible and perhaps desirable. Some of these
criticisms can be satisfied by a variety of carefully defined measures of
productivity. Others require further examination and discussion before



4
their comparative merits and demerits can be understood, and the discus­
sion thus provides a list of topics for further investigation, such as the
following:
Choice between value weights and man-hour weights, in combining diverse
physical units of output.
Adjustments for changes in coverage of output or employment by samples of
products or samples of establishments.
Hours paid versus hours worked.
Machine-hours versus other measures of capital input, such as deflated capital
assets.
Quality changes in products and materials, and the development of entirely new
products and materials.
Ways of combining different types of input and the related problems of
choosing between weighted and unweighted aggregates of labor input.
Treatment of overhead, maintenance, and other indirect labor.
Analysis of the representativeness of the sample group of industries for which
current indexes of productivity can be prepared.
Attainment of consistency of measures at the national level with measures at
subsidiary levels, such as the industry, the plant, or the job level.
Measurement of productivity in long-cycle or custom industries.
Measurement of productivity in service and trade.

We need, especially, to learn the quantitative importance of some of
these questions.

Limitations of Basic Data
Discussion of the limitations of current indexes indicated their obvious
dependence on the availability of basic statistics. No index can be much
better than the data which it summarizes. A deeper appreciation thus
came out of the conference of the great need for more and better basicdata collections. Many of the questions raised concerning the accuracy
of the current indexes, for example, can be answered finally only when
we can check our current sample statistics against complete surveys such
as the Censuses of Manufactures and Mines. Indeed, if such “bench­
mark” data were made available at reasonably frequent intervals, many of
these questions would never arise. We need not only more Census sur­
veys, but also to collect more information in them; for example, informa­
tion on capital assets, equipment, and kinds and quantities of materials
consumed. This information should be presented in detail—for example,
by size of plant. In addition to complete surveys, there is of course
need for broader and more intensive current surveys by government and
private organizations, if the information on contemporary changes in
productivity is to possess the accuracy and scope desired. One of the
interesting facts brought out was the rather widespread lack even of
private plant records on productivity changes. It is probably fair to report



5
the conference as favoring wholeheartedly the most comprehensive statis­
tics possible, as a basis for learning from the past, reporting on the
present, and guidance to the future.

Factors in Productivity
With the discussion ranging over the full list of different kinds of
input—materials, fuel, and use of capital equipment, for example, as well
as labor—attention was drawn to the possibilities of reducing the require­
ments, per unit of product, of factors of production other than labor.
Besides labor savings there could be—and the historical records show
have been—savings of materials and other factors. The more efficient
extraction of energy from coal is an example.
This may be put differently, to fit the view of many of the discussants.
Labor can be and has been saved directly, by cutting man-hour require­
ments for a given job or volume of production. Labor can be and has
been saved also indirectly, by cutting fuel and capital and other require­
ments for a given job or volume of production. Fuel, for example, em­
bodies in large part labor put into production at earlier stages of produc­
tion. To save a ton of coal means to save the labor of coal miners, of
railroad workers, of distributors. These indirect labor savings contribute
to our advancing standard of living as definitely as do direct labor
savings themselves. The gain from increased productivity is equal to the
sum of all these direct and indirect savings, and not merely to the savings
of one factor of production.
With this recognized, conference discussion advanced to another, re­
lated, point. When material or capital savings occur, these are to be
added to direct labor savings to get a measure of productivity increase;
but when more materials or more capital are required to turn out a unit
of product, then direct labor savings must be offset correspondingly. To
what extent increases rather than decreases in unit material or capital
requirements have occurred is determinable only by appeal to the facts. It
was suggested by some speakers that, on the whole, material require­
ments per unit of product have been reduced rather than increased; and
that the period of increasing capital investment relatively to output, so
characteristic of early industrial development, had more recently been
followed by a period in which many industries’ capital requirements per
unit were reduced. It is important to distinguish carefully between long­
term trends and shorter movements, however, and between one industry
or plant and another. Broad generalizations always imply exceptions.
Here, too, there is open a field for serious research and historical study.
The recognition that in an industry or plant various factors cooperate
in the job of production, that these influence one another, and that they



6

are also influenced by factors outside the industry or plant, advanced
understanding in still another way. The discussion at the conference high­
lighted the dangers of imputing responsibility for advance—or decline—~
in productivity to any one factor. To isolate the influence of a single
element in a complex process has always been the most difficult problem
of science. It has been solved, satisfactorily, only in the physical sciences
where experimentation under controlled conditions has been developed.
In the social sciences, statistical methods to accomplish the same aims are
still in the process of development, and in any case, the data to which to
apply these methods and the skill with which to interpret the results are
largely lacking. This does not mean that now and then the immediate
cause of a particular change cannot be determined; but this is probably
rare, and determination of the ultimate cause is quite another thing. As
one speaker so well put it, shall we try to figure Lord Kelvin’s contribu­
tion in assessing the factors aiding in recent developments in electric
motors? To try to determine credit or responsibility is to enter an endless
maze. Only in Heaven will judgments of this kind be made and grace­
fully accepted.
As planned, emphasis was placed in the conference on questions of
concept and measurement. Effort was made to avoid such controversial
and speculative matters as how to divide the gains from productivity. In­
evitably, however, the talk at the conference touched—even if lightly—
on such matters as wage rates, profits, and prices, as well as numbers of
workers, volume of equipment, and volume of goods produced, for ob­
viously there are important questions of developing and maintaining in­
centives to raise productivity. There are questions of the bearing on pro­
ductivity of “balance” among industries, plants, and factors of produc­
tion, of “equilibrium” of costs and prices, incomes and expenditures.
The important fact could not be overlooked, that raising the Nation’s pro­
ductivity is an economic problem, not merely a problem of technology or
engineering. More must be learned not only about concepts and measure­
ments but also about the economics of productivity, if the factors deter­
mining our economic progress are to be ascertained. The conference
members showed great awareness of our ignorance of the economics of
productivity. It is for this reason that they stressed the need for more
work in analyzing this intricate set of relationships. Only by coming to
grips with this problem can a decision be made as to the concepts and
measurements needed.

Work Still to be Done
It is hoped that this quick review will give the reader a notion of what
went on and stimulate him to read the more detailed storv that follows.



7
Although, naturally enough, the main points of agreement have been
stressed, in a sense the conference may be summarized also by saying that
it accomplished not so much explicit agreement on particular points as
better understanding and appreciation of diverse points of view. This
too, however, implies agreement that more work has to be done.
This means more work of the type that has been done. Understanding
of the facts concerning productivity and the factors stimulating (or re­
tarding) advances in it is furthered by the preparation of current indexes
reporting changes in such quantities as output per man-hour and labor
cost per unit of product. These indexes can be increased in number and
improved in construction. They are valuable. Understanding can also be
broadened and deepened by other types of work. Studies of marginal costs
and of the functional relation between production, labor and capital (the
“production function”), are interesting approaches which should be con­
tinued. Studies of such relationships as the following also need to be con­
tinued and expanded:
Changes in productivity in relation to changes in selling prices, employment,
wages, and other economic quantities.
Variation in productivity among plants of different size, age, and organization.
The extent of and reasons for international differences in productivity.
The origin and diffusion of technological developments of all sorts.
The economic histories of particular industries and of the relations between
industries.
Obstacles to the introduction of technological advances.
Cyclical changes in productivity.
Productivity differences among individuals.

In attacking a set of problems with such wide ramifications, no ap­
proach or aspect should be neglected.
It became apparent during the conference that the participants not only
realize that much work remains to be done in this field but expect the con­
ference, acting through the executive committee, to carry on a good deal
of this work and stimulate its undertaking in other quarters. It is appro­
priate therefore to report the plans now being made by this committee to
carry out the mandate laid upon it. It is hoped to start a series of round­
table discussions at which experts will be asked to present papers or
otherwise lead discussions bearing on particular and well-defined ques­
tions related to productivity. These questions will be drawn from those
given above and from additions to them. Participating in the discussions
will be not only members of the committee but also other persons qualified
to contribute. It is hoped to keep the groups small and active—essentials
for intensive discussions of technical subjects. The discussions will be
made available to the public in the form of papers or abstracts published
in appropriate journals or bulletins. It is expected, also, that as experience



8

with the discussions accumulates, it will again be desirable to arrange a
large-scale conference, open to a wider audience, to discuss the past work,
bring to public attention the work currently being done in this area, and
plan for the future.
Obviously, no one group can hope to do all the work required. Con­
sideration should and must be given to these various questions by men and
women throughout industry, labor, academic circles, and government. Ex­
perience with particular industries and access to special bodies of data
are real advantages possessed by many of these persons. It is hoped that
they will bring these advantages to bear effectively on the common prob­
lem. It is hoped, too, that they will advise the sponsoring agencies, through
the chairman or the secretary of the executive committee, of the work
they are doing. With all thus working intensively, in constant touch with
one another, there is hope of enlarging the broad conceptual and factual
basis so valuable as a common meeting ground for persons and groups
with diverse points of view.




S olomon F abricant ,

Chairman, Executive Committee,
Conference on Productivity.

Summary of Proceedings
OCTOBER 28—MORNING SESSION

Chairman, Leon Henderson, Research Institute of America

Welcome to Conferees
S tuart A. R ice (Assistant Director of the U. S. Bureau of the
Budget) : A warm welcome is extended to the members of the conference.
The conference was arranged through an executive committee composed
of representatives of management, organized labor, the Department of
Labor, the Department of Commerce, and the Bureau of the Budget. The
work of the National Income Conference in clarifying and developing
concepts in that field should be noted. It is hoped that the Productivity
Conference will contribute to greater agreement on productivity concepts
and to the improvement of measures of productivity.

Scope of Conference
E w an C lague (Commissioner of Labor Statistics) : The conference
was not called to discuss means of improving productivity or the methods
whereby the social benefits of production economies should be divided or
distributed. Its purpose is to reach agreements, if possible, on what
physical facts of the economy the word “productivity” should cover. The
conference may wish to differentiate the various concepts now encom­
passed by the term “productivity,” both in definitions and measurements.
The participants will discuss the indexes and measures of productivity
now available, what they cover in the way of physical facts, and what
physical facts they ignore or obscure. The meeting will also consider
means of improving measures of productivity, extending them, and mak­
ing them more useful to greater numbers of persons. It is hoped that the
conference will give direction to all research in this field, both private and
public.
The conference is to be a discussion meeting in which the members
participate. A transcript of the proceedings will be taken but will not
be made available for general distribution. A summary of the proceedings
will be distributed to all participants. After approval of the participants,
this summary may be made available to the public. A limited number of
publications representatives have been admitted to the conference, with
the understanding that no quotations will be attributed to individuals
except with their express permission. Moreover, all individuals represent
themselves only and not their organizations,
745867—47---- 3




9

10

Productivity and Its General Economic Setting
R obert N ath a n (Robert Nathan Associates): Although the con­
ference is concerned only with the questions of concepts and measure­
ments, clarification of these questions would help in policy considerations.
Over the past half century, there has been a tremendous increase in
productivity in terms of the relationship between output and every factor
of production. The increase in productivity has not been limited to a rise
in production in relation to man-hours worked. It is probable that pro­
ductivity has also increased as regards the relationship between produc­
tion and capital equipment, provided capital equipment is measured not in
dollars but in terms of the efforts and resources put into capital
equipment.
This increase in productivity has many implications, one of the fore­
most being its relationship to the question of employment. In the past,
when the problem of technological unemployment was discussed, it was
thought that there was only a given volume of production to be created
and that when increased output per unit of effort was achieved, the total
effort previously used would not be required. However, as a consequence
of analysis of what people can consume, the concept of technological
unemployment has begun to be replaced with positive and constructive
thinking of full employment. The benefits of increased productivity should
be considered in terms of standards of living rather than of technological
unemployment. The more rapid technological development and the more
speedy increase in productivity that may take place in the future is not
a threat and must not be regarded as a threat to full employment. Society
must make certain that, as output increases, employment is maintained, so
that participation in consumption and production shall not be narrowed
to a limited and decreasing percentage of people.
On the basis of the speaker’s experiences overseas, it appears that the
rest of the world is just beginning to recognize the importance of pro­
ductivity in the development of the tremendous economic power of the
United States. In France, for example, there is a keen recognition of the
need for industrial development, modernization, and higher productivity
if the nation is to be able to compete in world activities with the United
States, Great Britain, and other countries. Moreover, other countries
look to the United States for leadership, for guidance, for “know-how,”
for help.
Productivity also has significant implications with respect to domestic
policy on wages, prices, and profits. Increased consumption can emerge
only through production. Monetary benefits to any group mean nothing
unless expressed in terms of real output and real consumption.



11

Problems of Concepts and Measurement in the
Field of Productivity
H iram D avis (Director of Industrial Research, Wharton School,
University of Pennsylvania) : Most persons, in using the term “produc­
tivity” have meant the physical output obtained for a given physical
input. The usual measure of productivity—output per input of labor—
is inadequate, however, since one input factor is used to represent all
input. The change in output per man-hour may be greater than the change
in output per unit of total input, for example, when power or machines
have been substituted for labor. On the other hand, even though man­
hour output shows no change, the input of other factors may be changing
materially. Thus, capital consumption may be speeded up by obsolescence
or reduced by improved machines with a longer service life. Input factors
cannot be measured by money costs, since money costs are both a measure
of input and a measure of what the input agents receive.
Several methods of measuring input have been suggested. The em­
bodied labor method involves the expression of machinery and other
factors in terms of the labor content represented. One difficulty is that
it is not possible to express risk-bearing, for example, in terms of labor
content. Moreover, it is extremely difficult to express capital consumption
in terms of labor content. Another proposal is the use of costs at constant
prices. However, there is the practical objection that the compilation of
appropriate indexes for correcting the value series is extremely difficult.
Because of the many difficulties in the measurement of total input, it
might be more useful, for analytical purposes, to develop a series of ratios
relating output to each of the important input items.
Change in the composition of output, the substitution of new products
for old, and changes in quality all make it extremely difficult to measure
output. The equivalent-unit method—the method of reducing output of
different grades to an equivalent grade—involves the problem of the
method of conversion used. If prices are used as weights, the output
measure may be influenced by differences in marketing methods or differ­
ences in demand for the various classes of products. If labor time is used
as weight, the output series would correspond to the man-hour series.
Another method proposed is the potential service of the commodity, for
example, tire-miles. Although this measure is useful for certain purposes,
it has its limitations. In the case of tires, it would credit the tire industry
for improvements in automobiles, in highways, and in the quality of
crude rubber, that were brought about by other industries. Many indexes
of physical production are weighted by value or value added. Since value
added changes from year to year, depending upon profits, it may be



12

desirable, where possible, to refine the value-added weights to the point
where they represent manufacturing costs.
Since the problem of measuring output falls particularly at the indus­
try level, it is hoped that out of the conference may come a decision to
develop working committees in each of the important industries.
In connection with input measurement, it might be desirable to consider
not only the input used up, but also the input supply assembled for an
undertaking. Another important concept is the measurement of output not
only in relation to physical factors, but in relation to real costs.
In relating output to man-hours, it might be preferable to express the
results in terms of labor requirements per unit of output rather than
output per man-hour. A similar form might be used in showing the ratios
between output and capital and other production factors. The term “pro­
ductivity” might well be restricted to those cases in which it is possible
to present more than one measure of input or in which the broad sweep
of developments over a substantial period is considered.

Concepts and Measures of Productivity at the Job Level
M artin G ainsbrugh (National Industrial Conference Board) r1 A
questionnaire was sent out to the NICB panel of business executives,
concerning changes between the prewar and current period in labor
productivity on comparable jobs performed under comparable conditions.
The reports indicated a general dissatisfaction in industry with the term
“labor productivity.” Some replies indicated a belief that the number
of labor hours per unit was determined primarily by the type of equip­
ment and not by human factors. Many companies had no measures either
of productivity on comparable jobs or of output per man. Several
measures were used in the replies—output per man, dollar value of
products per productive man-hour, dollar sales per employee, etc.
Although the survey indicated that there is an interest in labor produc­
tivity, few plants have quantitative information on this subject.
N athan S pero (United Electrical, Radio and Machine Workers,
CIO) : Management has been carrying on a Nation-wide publicity drive,
singling out worker effort as the important factor in productivity and
implying that worker effort has decreased. This campaign obscures the
fact that the effort of the individual worker is inseparable from many
of the other important factors that affect productivity. Worker effort is
affected by such factors as patriotism, the cost of living, profits, and
other influences. A constant flow of materials of the right quality is
1E
’ N
—The remarks of Mr. Gainsbrugh, scheduled for the morning session, were
actually delivered in the afternoon. They are given here for the sake of clarity and in order to
round out the presentation of the subject. The same was done in a few other cases.
ditor s

ote




13
extremely Important to efficient production at the worker’s level. Short­
ages of materials have resulted from speculative stock piling to anticipate
price increases or hoarding to force price increases. In addition, the
nature of the plant’s physical equipment determines, to a large extent,
the worker’s efficiency. The efficiency of the workers also depends on
the efficiency of management. Frequently management has raised the cry
of decreased workers’ effort to hide its own shortcomings. Similarly,
systems of wage payment influence worker effort. Incentive systems
provide some estimate of the extent of productivity increase and of the
extent to which management passes along the gains resulting from
increased productivity. However, even incentive systems can sometimes
be turned into their opposites, resulting in a lack of incentive for produc­
tion. Finally, relations between the company and union have an extremely
important effect on worker efficiency—for example, the way in which
negotiations are conducted, the way the grievance procedure operates,
and the nature of supervisory relationships.

Discussion
S olomon B arkin (Textile Workers Union of America, C IO ): Man­
agement has employed numerous measures which allegedly measure out­
put. The one which is very widely used is the time study. It is not a true
measure of productivity. It provides a method of judging, through inade­
quate means, the increase in output expected for varying additions in the
application of human effort and skill. Many incentive systems are based
on the premise that labor should be compensated only for the additional
output which arises from more than normal effort and skill. They there­
fore recapture all increases in output arising from causes other than
human effort and skill. They do not measure output. Measures of physical
output which do not seek to distinguish the source of the increased output
more nearly conform to the measure which is being sought. Thus, for
example, incentive systems which provide payment for output and which
do not allow for the change of this rate for any cause more nearly
approximate compensation for increased productivity from all sources.
During the war certain plant-wide incentive systems were based on
this principle.
H. B. M aynard (Methods Engineering Council) : Difficulties are in­
volved in establishing performance ratings. With fixed methods, the
contribution of the worker to production is dependent on his skill and
on the effort with which he works. While performance ratings are based
on a process of judgment, it is possible to achieve a fair degree of
consistency by using carefully worked-out definitions, motion pictures
representing different performance levels, etc. Another approach to this




14
problem is effort rating, which endeavors to arrive at an evaluation of
performance level by judging the speed of the motions made by the
operator. A different approach being evolved establishes, as the result of
considerable research, predetermined motion time standards which, if
once accepted, will take the element of judgment out of the problem.
L azare T eper (International Ladies' Garment Workers' Union,
AFL) : Mr. Maynard has discussed the question of measuring efficiency
at the job level under fixed conditions, but the main problem arises when
conditions change. If a particular operation is changed, for example,
from a 5-motion job to a 4-motion job, efficiency is not always measured
in relation to the former 5-motion standard but in relation to a new
4-motion standard, and the worker loses the benefits of increased physical
productivity. If labor is not to pass judgment on management's efficiency,
performance on the job must be judged in terms of the physical units
produced.
H. B. M aynard : When a worker, through his own ingenuity, reduces
the number of motions involved in a job, he should have a substantial
award. However, the standard should be reset and rewards for ingenuity
should not be included in the incentive plan. If they are, the earnings
structure is distorted, with some workers earning considerably more than
others for the same expenditure of effort; the workers themselves con­
sider such a wage structure unfair.
R ufus T ucker (General Motors Corporation) : Although it is per­
fectly true that bad morale and bad labor relations result in decreased
productivity, output is reduced only because such conditions cause the
laborer either deliberately or unconsciously to hold back. Consequently,
such conditions should not be charged to management, since the effect in
reducing output depends wholly on the laborers' attitude.
M r. C lague : The problem of measurement of productivity at the job
level leads quickly to the problem of shares in the output. For example,
assume that man-hours per unit were reduced because of the workers'
ingenuity, wages were raised proportionately, and the same number of
people were at work. Production would be increased, but it would not be
possible to sell the whole output at the same prices. One is thus forced
to a consideration of wages, interest, rent, and payments for other
factors of production. Nevertheless, it is desirable to work out methods
of measurement that will not be confused by the issue of sharing the
profits. Another difficulty is that the job is constantly changing, and any
measure of productivity at the job level cannot be put into a long series.
In view of these difficulties, it appears that measurements of productivity
at the job level are the concern of labor and management at the particular




15
plant, and both labor and management will have to get into the study
of job output.
G ardner M eans (Committee for Economic Development) : In a boom
period there is a small proportion of overhead workers relative to pro­
duction workers. In a depression, the number of overhead workers may
be the same, but the number of operating workers is reduced, with the
result that total man-hours per unit of output are increased, even though
each worker is just as productive in his particular job. There is a question
as to whether such a change should be called a decline in productivity
or whether the effect of changes in the proportion of overhead workers
should not be separated from other changes.
H iram D avis : It would be possible to develop a series of ratios, relat­
ing output to the various categories of labor and not merely to labor
input as a whole. Changes in the proportion of overhead workers may not
be only a cyclical relationship; it has been suggested that there is a decided
trend in this country toward, an increase in the proportion of overhead
workers.
(Seth Levine of the CIO Maritime Committee, Charles E. Young of
Westinghouse Electric & Manufacturing Co., and Rosalind Schulman
of the Industrial Union of Marine and Shipbuilding Workers of
America also discussed briefly the relationship between productivity and
the level of output.)
D uane E vans (U. S. Bureau of Labor Statistics): Cyclical changes
are not so likely to affect measures of productivity at the job level as
at the industry level. Physical output per unit of labor input is of primary
importance, rather than per capital input or input of other factors. On
the one hand, the physical output produced determines the standard of
living. On the other hand, the labor force is the one resource whose
utilization cannot be deferred. Tremendous numbers of separate thing?
determine the output of the individual worker—management factors,
mechanical factors, machine factors, materials factors, and worker
factors. A very big element in the high productivity of the United States
is the accumulated skill and experience of the labor force, which is not
wholly industrial but is due, in part, to familiarity with mechanical
devices in daily life.
There is no typical situation with respect to reports concerning worker
effort. Some plant managers have stated that their labor force is better
today than when we entered the war, while others hold strong opinions
to the contrary.
R obert W. B urgess (Western Electric Co.) : Referring to remarks



16
by Mr. Barkin and others which questioned the accuracy of available
measures of output at the job and plant level, as a matter of concept,
measures of productivity (according to the orthodox definition of output
divided by man-hours worked) could readily be constructed in plants
which use the piece-work system with predetermined “standard” costs
and standard times for each job. The point is that the concept of pro­
ductivity in such cases is clear-cut and its measurement available Without
question. How accurately and fairly standard costs and standard times
have been established in any plant is another question, separable from
that of the concept of productivity. There is still another question of the
operating type: whether day-by-day adjustments should be made in the
application of piece-work rates to recognize unusual current irregularities
in production conditions.
S olomon B arkin : A piece-rate system does provide for payment in
proportion to output. In contrast, engineered incentive systems are based
on the concept that there should no!t be compensation for increased out­
put if the increase is attributable to some factor other than the worker’s
effort. Thus confusion arises because of the attempr to identify worker’s
effort and output.
OCTOBER 28—AFTERNOON SESSION

Chairman, Robert Nathan, Robert Nathan Associates

Concepts and Measures of Productivity at the Plant
and Company Levels
B e n ja m in H askell (United Textile Workers, AFL) *. The statistical
approach to productivity—the measurement of the relationship between
output and labor—has very serious and dangerous inadequacies. The
desire of the statistician for some single over-all measure is a stumbling
block to obtaining genuine knowledge. For example, a variety of very
different products may be called by the same name. The averages pre­
sented by the statistician often obscure rather than disclose what should
be measured.
The statistical approach ignores the meanings and explanations. How­
ever, the explanations and qualitative factors are just as significant as
the figures. For instance, when a strike broke out at a cotton mill, it was
discovered that the employer had bought a poor batch of cotton. On the
basis of production during that period, one might conclude that labor
was falling down on the job, but the factor of poor cotton had as much
importance as the time input. In the same way, the organization of the
plant, labor relations, the kind of machinery, and the flow of materials



17
affect the volume of output. Those who handle statistics have a moral
obligation to give explanations, because the tendency has been to imply
responsibility whenever the statistics show an apparent decline in physical
output per unit of labor input.
The farther productivity statistics depart from the plant level, the
more obscure and meaningless they become. The statistical approach
should therefore be supplemented by the case-history approach at the
plant level. The case-history approach will provide the qualitative factor,
the explanation of why things are as the statistics claim them to be. The
information will be based on actual experience of management and labor
and not on reported statistics which have many flaws. A valuable by­
product of this approach will be to increase the cooperation of local labor
and management in matters of common concern, such as productivity.
H. B. M aynard (Methods Engineering Council) : In considering pro­
ductivity at the company level, it is necessary to include not only the
productivity attributable to the application of the worker but also that
attributable to methods. First, productivity is affected by methods deter­
mined by job conditions, including design. For example, in the dress
industry, productivity will depend on the presence or absence of sleeves,
collars, pockets, frills, and the like; in coal mining, on the presence or
absence of gas, the height of the seam, the hardness of the floor, and
similar factors. Second, methods determined by physical equipment affect
productivity—whether the operations are performed by hand or machine,
and the types of machines used. The third method variable, under given
job conditions and physical equipment, is the sequence of motions em­
ployed by the operator.
Because of these factors, the measurement of productivity is difficult.
The various yardsticks used (volume per man-hour, volume per man-day,
volume per crew-day, volume per machine-hour, dollar value of sales
per man-year) measure productivity incompletely and must be used
with understanding. General units like tons, molds, and pieces do not
correlate well with the time and effort required to produce them, if the
range of sizes, weights, etc., is varied. In many instances difficulties have
been experienced with incentive plans because differences in the design
of the products handled resulted in large variations in earnings.
The solution is generally to establish additional units of measure, with
a standard for each. In coal mining, for example, instead of a single
tonnage per man-day figure, it would be necessary to establish standards
for gaseous and nongaseous mines, for various ranges of seam height,
for various grades of coal, and for various types of physical equipment.
Management officials, through long experience, are able to establish
empirical relationships so that even incomplete measures take on value

745867—47-— 4



18
and meaning. However, in order to obtain productivity figures that are
comparable and useful at the plant level, it would appear necessary to
establish well-defined classifications of product, job conditions, and
physical equipment. Then any differences may be attributed to labor
effectiveness, including in the latter term such factors as skill, effort,
heat, light, ventilation, training procedures, morale, managerial skill, and
so on. Such information would appear to be most useful in raising pro­
ductivity throughout a given industry.
C harles E. Y oung (Westinghouse Electric C o.): Output per man­
hour of factory labor is only one of several interrelated productivity
concepts that are important in managing a manufacturing enterprise. For
purposes of reaching valid business judgments, comparisons of results
with costs must ultimately be expressed in monetary terms. Analysis of
physical output per man-hour is useful primarily as a refinement of tech­
nique aimed at eliminating the vagaries of money as a means of measure­
ment. However, the concepts and measurements must eventually be rein­
terpreted into monetary terms.
Physical measurement of productivity at plant and company levels is
necessarily inexact. Even if a plant or company produces a single product
all the units of which are identical, problems may arise. If the plant has
been generating its own electricity but later buys its power, for example,
the man-hours per unit of finished product will be reduced, but it is
questionable whether this change should be called an increase in produc­
tivity. Measurement becomes more complex when a second product is
introduced, for this involves a weighting problem. This lack of homo­
geneity also applies to man-hours, since there are differences in skill and
effort between different individuals and in different occupations. Another
problem concerns what part of the total hours worked in a plant or
company shall be used. Because of these difficulties, comparisons pur­
porting to show differences in productivity must be subjected to the most
careful scrutiny. Possibly the proper function of physical measurement
of productivity at the plant or company level is to delineate broad trends
rather than to support detailed conclusions as to management action
or wage policy.
Many of the improvements in manufacturing efficiency come about
product by product, or even part by part. There is no reason to suppose
that each and every operation included in an average has become more
efficient in the same degree. An example of productivity change operating
virtually on a plant-wide level is the redesign of the electric-motor line
of the Westinghouse Electric Co., which resulted in a 30-percent
reduction in the man-hours required to produce a given motor. This pro­
gram involved investigation of the characteristics that customers desire



19
in motors, extensive development work, and the investment of 20 million
dollars in a new pl^nt. The number of parts required for the new motors
was reduced by over 90 percent. This example indicates that substantial
improvements in productivity do not accrue merely by the passage of time,
but must be conceived, studied, implemented, and backed by substantial
investment.
Many of the forces which operate to increase productivity have little
connection with the earnestness and skill of the individual worker. Among
the foremost requisites of growing efficiently is alert aggressive manage­
ment. Friction between management and workers, however, can hinder
and even undo many promising technological gains. If friction is replaced
by mutual confidence and respect, management and labor can go forward
together, to their mutual advantage and to the benefit of the general public.

Concepts and Measures of Productivity at the Industry Level
S olomon B ark in (Textile Workers’ Union of America, CIO) : This
is a summary of two papers already distributed to the participants.
One, “Measuring Man-Hour Output in the Synthetic-Yarn Industry,”
discusses the deficiencies of the index prepared by the Bureau of Labor
Statistics as well as of an alternative measure prepared by the Textile
Economics Bureau. The Textile Economics Bureau production index,
which is based on the total yardage of rayon yarn, is invalid, for there
is no direct relationship in the synthetic-yarn industry between yardage
and man-hours. The Textile Economics Bureau and Bureau of Labor
Statistics indexes have a number of deficiencies in common. The com­
panies in the industry produce many products in addition to rayon yarn
and the proportion of these products has continuously increased; for
example, two plants are devoted exclusively to the production of nylon.
It is incorrect to derive output per man-hour by dividing an index of
rayon-yarn production by a man-hours index which includes nylon and
other products, as well as rayon yarn. Moreover; the man-hours figures
are primarily the number of man-hours paid for and not man-hours
worked. As unions obtain provisions for more liberal vacations and
sickness and lunch-period pay, the discrepancy between the index of
man-hours paid for and the actual hours worked is constantly widened.
M r. Y o u n g : The experience of Westinghouse Electric Co. is that “net
allowed hours” are about the same as in 1941. Elapsed hours for the
incentive workers are somewhat lower, and indicate a gain in efficiency.
However, for hourly paid workers, there has been a considerable reduc­
tion in the number of “net allowed hours” as against the elapsed hours.
In addition, there has been a large increase in the number of salaried




20
and staff people, brought in on the expectation of a high volume of out­
put. The company is currently making inefficient ,use of some labor;
this is no reflection on the willingness or skill of the workers, but may
be bad management.
Per capita income has been growing at an accelerated rate and it is
important that the remuneration of labor be reviewed quite frequently.
This should be done by considering total productivity and sharing the
increases, making only such distinctions between workers as are possible
through very detailed studies of performance at the job level. The
general movement of wages must be based on the general movement in
productivity. It is possible, by studying the changes in real per capita
income of the country as a whole, to determine what changes should be
made in the income of labor as a whole to provide equilibrium in the
economy. About a year ago the speaker made an attempt of this kind.
Assuming that wages, prices, and productivity were in balance as of
January 1941, making allowances for the increased cost of living, and
projecting the annual productivity increase of the years before the war,
it was found that as of August of last year [1945] there was a needed
adjustment of 18 or 19 percent in the general wage level. Although this
study yielded only a first rough approximation, it was a starting point.
J ohn D. G ill (Atlantic Refining Co.) : Repeated attempts to make
measurements of labor productivity in the petroleum industry have been
unsuccessful because of changes in equipment, in processes, in the
quality of output, and in the quality of raw materials. The chief reason
for attempting to measure labor productivity is with respect to the re­
muneration of labor. However, it makes no difference whether labor
productivity can be measured at the plant and industry level. It is possible
to depend on competition from department to department, from plant to
plant, and from industry to industry, for high labor productivity, in the
sense of the workers’ attitude and qualities of mind and muscle. It is
absurd to think that there can be vast differences in the remuneration of,
for example, Number 1 machinists in different plants because of differ­
ences in productivity at the plants. The apparent differences in produc­
tivity between the plants may be almost entirely the result of management
efforts, which are quite irrelevant to the value of the machinists.
The original interest in man-hour output centered in the question of
displacement and reemployment, but current interest centers in the
creation of new wealth for distribution. Therefore, it is necessary to
consider how the results of our industrial activity may be more rationally
distributed. Output per man-hour gives one phase of the approach to
productivity, but it is also necessary to consider other phases of produc­
tivity in the sense of creation of wealth at a lower cost. Enormous sayings



occur in industry through more economic use of materials and savings of
overhead items.
The second paper, “Measuring Productivity Rather Than Man-Hour
Output In an Industry,” discusses changes in total productivity in the
cotton-textile industry between 1937 and 1939. The value of products,
less wages and profits, corrected for price changes, is used to show the
input of factors other than labor. Output per unit of input of these factors
must be considered along with output per man-hour. This paper is pre­
sented only as an approach to the problem of measuring total productivity
and may need refinement.

Discussion
N atha n F in e (Research Director, Metal Trades Department, Ameri­
can Federation of Labor): Mr. Young's implication was that shares
should be based on the contribution of the various productive factors. The
common assumption is a very static one. It is assumed that the worker
stays pretty much on the same level, but that the research division has no
limitation. The status of the industrial arts, which is so largely determined
by the universities and Government agencies, is taken for granted. In
the last 4 or 5 years, the Government has invested hundreds of millions
of dollars in machinery and has given it mostly as free goods to industry.
That contribution will automatically be assumed to be the contribution
of research of the companies. No effort is made to determine what part
of the workers' products goes into the company treasury for depreciation
or research, which in turn becomes a company contribution to further
development. When analysis is made from the viewpoint of relative
contributions and it is assumed that management takes over everything
except the workers' contribution, the whole question of productivity
implies something which is a little pernicious from the labor standpoint.
If studies were made plant by plant, or job by job, and all the factors
were analyzed, the common assumptions would not necessarily be shown
true.

M r. Y oung : It is agreed that the integration of research activities make
it difficult to determine the specific contribution of any one group.
H arry M agdoff (U. S. Department of Commerce) : There is need
for relating what is being measured to the purpose of measurement.
Different measures are needed for different purposes. There is the
national aspect of distribution of income, from the standpoint of adequate
balance to maintain full production and full employment. Secondly, a
measure is needed for negotiations with respect to wages. Third, measures
are needed for management problems with respect to controls and effi­



22
ciency. In the opinion of the speaker, there is no scientific method of
measuring performance on the job. The only constructive way is for
labor and management to get together and to settle on techniques and
standards.
M r. G ill : Industrial engineers can do much to affect productivity, in
the sense of power to produce, but the individual who might be operating
at 80 or 85 percent of his personal effectiveness cannot have that effective­
ness greatly increased. The main concern is productivity, in the larger
sense of capacity to produce. The problem is to increase total production;
this can be done only by attaining and maintaining equilibrium in the
economy.
M r. B arkin : Industrial profit figures are not real indicators of pro­
ductivity. There is, first, no set rule as to how much industry should
receive; nor is there uniformity as to the methods of calculation. The
figures usually discount increases in productivity which have been passed
on through lower prices and reflect the bargaining powers of the various
economic groups. It is therefore necessary to study the relations of
various factors that enter into productivity at the industry level.
E verett H agen (National Planning Association) : There is a method
by which it is possible to get double entry bookkeeping into our thinking.
The changes in productivity—total physical output divided by man-hours
—together with the change in prices and the shift in the nature of the
output should add up to the change in the value of output per man-hour.
A study made by Sam Cohn, under the direction of the speaker, on
the change in the value of output per man-hour between 1939 and 1945,
both for all nonagricultural industries and for manufacturing showed
that for all nonagricultural industries the increase was 55 percent. The
increase in prices may be estimated at 28 percent. Thus, the increase in
physical output per man-hour, plus shifts in the nature of the products,
must have accounted for a 20-percent rise in the value of nonagricultural
output per man-hour. It is possible to estimate the change in physical
productivity, the effect of the shift in the composition of output, and
the change in prices. If all three will not add up to the change in the
value per man-hour, then something is wrong with one of the four
measures.
D u a n e E vans (U. S. Bureau of Labor Statistics): One way of ap­
proaching economies in the use of input factors other than labor is by
the use of the interindustry relationship technique. The Bureau of Labor
Statistics, for the year 1939, divided the entire economy into major
industry groups and investigated quantitatively the kinds and amounts
of products which proceed from each industry to the others for further



23
processing or for use. For example, over the past 20 years there has been
a marked reduction in the amount of coal required to produce a kilowatthour of electric power. This change does not show up in the productivity
figures either for the coal industry or for electric power, but it dotes show
up on the interindustry balance sheet.
Many different types of measures can be useful. In the tire industry,
for instance, one can measure the number of tires produced or the total
tire-mileage. These may be related to total man-hours, skilled and un­
skilled, together and 'without differentiation, or it is possible to separate
the man-hours worked by skilled workers and unskilled workers and to
weight them to derive an adjusted figure. None of these measures is
intrinsically wrong, and each may convey useful information, but the
user must understand what is included, what is excluded, and what
is ignored.
OCTOBER 28—EVENING SESSION

Chairman, George W. Taylor, University of Pennsylvania

Concepts and Measures of ^Productivity at the National Level
M arion H edges (International Brotherhood of Electrical Workers,
AFL) : Whether the United States operates under a free economy, a
planned economy, or a mixed economy, its achievements depend upon
production. Productivity is a secondary concept, of importance because
it bears directly on the question of equitable distribution of the results
of production.
The salient fact about American industry is that it is mechanized. Our
weakness has been that we have never clearly visualized our destiny with
relation to mechanized production. Mechanized production enables in­
dustry to pass from piecemeal production with high unit costs to mass
production with low unit costs. Mass production with low unit cost, how­
ever, must depend upon high income for each citizen. The United States
has elected to follow the free-enterprise concept in the operation of our
economy. Our economy can be operated under this concept, but not
without controls.
An article by Benjamin Graham, in the Commercial and Financial
Chronicle, presents a point of view in business that may be of great
moment. Mr. Graham traces mass unemployment in time of prosperity to
a persistent tendency for productivity to expand faster than per capita
income and to a recent tendency for the working force to expand relative
to population. Mr. Graham states that wages should advance with’ pro­
ductivity, while the workweek should decline to the extent needed to
maintain full employment.



24
Some way must be found to take into consideration, in any productivity
index, intangible values. Moreover, in some industries, as in building
construction, productivity cannot be measured quantitatively.
Labor can hope to profit by a more scientific operation of industry,
with national goals in mind. Labor believes generally that labor is not
now receiving and never has received a just share of the fruits of pro­
duction. Labor wishes to see the gap closed between the growth of
productivity and the income distributed. Labor does not believe produc­
tivity can be quantitatively measured, since intangibles are all-important.
Labor believes that any gauge of productivity will be advantageous—
even physical output per man-hour, but labor wants such a gauge, when
promulgated, to be seen for exactly what it is, and advocates that a com­
plete description be published with each set of periodic figures.
E verett H agen (National Planning Association) : The first concep­
tual problem in the measurement of productivity at the national level
relates to the weights to be used to combine into one index the trends
for the different types of goods, when the proportions of the different
goods change. As an example, assume that the two types of goods are
automobiles and tables, with tables representing forest products in
general, and that the wage payment per man-hour and the value of output
per man-hour are higher in automobile production. The increase in pro­
ductivity in the automobile industry has been much more rapid than in
the forest-products industry. If the two industries are weighted together
in accordance with the number of man-hours in each, there will be one
result for the combined productivity index. If the weights are the wages
in each industry, on the basis that a man-hour at $1.50 is not the same
as a man-hour at $1.00, a different index will be obtained. If the price
that the final purchasers of the products put upon them and weight by
value of product are considered more relevant, still a different combined
index is obtained. In general, weighting by value of product is most
appropriate.
Another problem is, which year is to be selected in weighting the two
series together. If automobiles and forest products are weighted in
accordance with their relative importance in 1940, automobiles will be
weighted much more heavily than if relative importance in 1900 deter­
mined the weights, and the index will therefore go up faster.
The problem of shifts in the proportions of different goods arises
between the goods-producing and service-producing industries. In the
services, the growth in productivity has been much more slow than in the
goods industries. In the thirties, the service industries grew greatly in
relative importance, because of disguised unemployment of persons who
shifted into small businesses and attempted to sell services. A marked



25
reverse shift occurred during the war. If the trend in productivity for
the economy as a whole is estimated by comparing the deflated value of
output with man-hours, the effect of shifts between industries is included
in the estimate. In the opinion of the speaker it is correct to include the
effect of such shifts. Because of such shifts, it was estimated in the
Hagen-Kirkpatrick study that the increase in over-all productivity
between 1939 and 1950 would be 2.3 percent per year, compared with
1.7 percent between 1923 and 1939.
Another conceptual problem which enters at the national level is the
difficulty of measuring output in some industries. There is no definable
unit of output in such fields as the law, services of government workers,
banking, and the like. A final problem is the lack of data for some fields,
even where there are no conceptual difficulties.
The conference might consider why there are no data in certain areas,
whether there are conceptual problems and, if so, what problems. In this
way, it could make some technical advance in the problem of measuring
productivity.
R obert W. B urgess (Western Electric C o.): Studies of productivity
at the top level may lead to the formulation of decisions on particular
operating questions; studies at the plant level may facilitate the task of
management; studies at the industry level may furnish a standard of
comparison of industries. At the national level two types of measures of
productivity would be desirable. To measure the typical gain in produc­
tivity that the economy as a whole is making, a composite should be
constructed by the use of constant weights applied to the productivity
series for the different industries. On the other hand, to meet the prob­
lems raised by Mr. Gill and others concerning changes in total product to
be distributed, a measure based on current varying weights for different
industries is needed. This measure is probably one of the most funda­
mental in the application of productivity studies to wage questions, since
it is the over-all gains in national productivity which have justified, during
the past century, the continued improvement in our standard of living
and similar over-all gains which are expected to justify corresponding
improvements in the future.
The concept of gross national product is more helpful in this problem
than the concept of national income. Gross national product is the
primary concept, since national income is derived by subtracting from
gross national product depreciation accruals; i. e., the capital assets
supposed to have been consumed. Annual depreciation accruals never
were very satisfactory as measures of the capital value used up per year
since they were never designed for that purpose, but are simply the
results
of convenient and conventional accounting procedures for spread­



26
ing capital costs by years. During the next few years, annual depreciation
accruals, as recorded in the books of manufacturing companies, are going
to be still less satisfactory measures of capital value used up since many
companies are going to use, without any depreciation charges, equipment
that was paid for and amortized during the war, and in addition, de­
preciation accruals based on plant cost will be much lower than the value
of plant used up as measured by replacement cost.
A defect in both national income and gross national product is the
inclusion of interest on the public debt. In the speaker’s opinion, it would
be preferable to consider only nongovernment output, even though it is
possible to justify the inclusion of some of the government expenditures
because they represent consumer services. However, most government
services are not valued in the market place and the only value that can be
assigned to them is the conventional one of what is paid for them.
It is important that measures of national production reflect gains
through technological improvements and replacements. Natural gas will
perhaps be brought to the eastern part of the country, replacing some
coal. If it serves the same purpose, gross national output is increased
or is maintained at less input.
In some industries, it is very difficult to consider productivity as due
to the efforts of the workers immediately concerned. In the exhibition
of motion pictures, for instance, there is, according to the formula,
greater production per usher where the theater is filled than when there
is a small audience. In fishing, again, production depends mostly on
natural factors which the individual fisherman merely accepts. It is diffi­
cult to think of productivity as having anything to do with efficiency of
labor and management factors in such cases.

Discussion
D uane E vans (U. S. Bureau of Labor Statistics): An illustration
will show the difference in results attained by weighting with man-hours
and with value added. The number of man-hours required to produce
a thousand cigars decreased about 30 percent from 1939 to 1945, and
the number required per thousand cigarettes decreased about 20 percent.
If we determine how many man-hours it would have taken in 1945 to
reproduce what we actually did make in 1939, we will get a figure between
20 and 30 percent as our average. If we determine the value per man-hour,
in constant prices, created in these two industries together, we get a
60-percent increase.
Value-added weighting is useful for some purposes but not for others.
One cannot conclude, on the basis of such a measure, that it is possible
to increase profits, to lower prices, or to raise wages to the extent of




27
a 60 percent increase in efficiency. The change is due partly to a shift
of production from industry to industry, and no single plant in either
industry is necessarily 60. percent more efficient.
For the national economy, it becomes virtually impossible to measure
physical productivity and it is necessary to use the value concept. Each
has its own field of use. The physical concept has a much more immediate
relation to economies in the usual sense.
S olomon B arkin (Textile Workers Union of America, CIO) : As
suggested earlier there are sources of increased efficiency which are not
reflected in the measures of output per man-hour hut which are relevant
to the process of trying to distribute increased productivity among the
various components of our productive system. For example, in a number
of industries chaotic competition has been ruinous. As a result of stabiliz­
ing controls there will be fair-trade practices. In some industries, as
for example in the apparel group, the unions have contributed to greater
stability, and the result of such stabilization is a diminution of the amount
of capital that gets destroyed.
Another illustration is the cotton-textile industry, in which the function
of selling is being integrated with the function of manufacture. The
integration of these two operations will not show up in any of the figures
that we have. However, it will mean a reorganization of the systems of
production and designing, of fabrication, and of marketing, which will
stabilize the industry and possibly result in less loss and less capital
involved in the industry for the same output. In wage negotiations, the
union will be tapping this economy because it will tend to show up in a
higher level of profits. Another example is the tendency in the men's
clothing industry for the manufacturer to have his own retail store.
M r. H a g en : Where there is an economy in a given industry which
is associated with a saving of material, without any saving of labor, a
study of man-hour output in the separate industries will not reflect it.
However, because of the reduction of employment in a previous industry
which was producing the material, without reduction in the Nation's
total output of finished products, the economy as a whole will show
such a saving.

Concepts and Measures of Productivity at the
International Level
J ulius H irsch (Consulting Economist) : Professor Rostas has tried
to show that for manufacturing industries as a whole, output per person
employed was roughly twice as great in the United States as in the
United Kingdom, and a similar result was obtained for the textile indus­



28
try by the Textile Mission. He states that it is possible that American
supremacy in manufacturing is obtained only at the cost of relative
inefficiency in the other four-fifths of the economy.
Comparisons made by the speaker for the year 19442 showed that
British labor received, in dollars, approximately two-thirds of what
American labor received, but purchasing power probably would not have
been so far from that of American labor.
It is greatly one-sided to measure productivity only by considering
output in relation to human labor hours. For national and international
comparisons various other standards have been developed, especially
those of net output per kilowatt-hour and net output per installed
horsepower. Data from the British Census of Production for 1935 show
that where there is much human labor and little machinery, net output
value per kilowatt-hour is high. Where there is little labor and much
machinery, the reverse is true.
A specific German approach to the question of efficiency was a sys­
tematic nation-wide comparison of costs and performances in industry.
Professor Schmalenbach developed schemes for uniform cost accounting
for more than 60 industries, and industry and government assisted in this
work. In this way, it was possible to compare every individual cost ele­
ment in a business with those in other businesses in the same industry.
These methods brought about an enormous increase in efficiency. German
statistics show an increase in value added by manufacture per man-hour
of about 50 percent from 1926 to 1936.
In the Soviet Union, there is a decided attempt to bring about an
increase in productivity with all means available to the concentrated
power. Production quotas are allocated to every factory, every depart­
ment, and ultimately, every laborer. The goals of the 5-year plan called
for an average annual rate of increase in productivity of 12 percent in all
industries, and 15 percent in the heavy industries. The Russian publica­
tions say that in most cases the performance of the Russian industry is
somewhat better than the plan; the statistical evidence, as far as we have
it, does not yet show how far this has been the case.
In the field of international comparisons, the first work was done on
distribution—wholesaling and retailing. The great area of retailing has
been neglected in the usual discussions of productivity. It is, however,
probably the area where there is the greatest need for a change. The
Harvard Business School has stated that the costs of distribution, includ­
ing transportation, absorbed 54 to 59 percent of the total cost of produc­
ing and distributing goods. -International collaboration in eliminating
sources of waste has been adapted to great advantage by the members
* Mr. Hirsch's comments were based on tables on “Productivity and Productivity Measurement
and International Cost Comparisons/* which he had distributed to the members of the conference.



29
of the International Association of Department Stores. There are new
attempts in the United States, as in Sweden and Switzerland, to lower
the increasingly heavy burden of distribution costs, though none has as
yet had decisive results.
Comparison of materials, wages, overhead, and profit in the main
manufacturing and mining industries for Sweden, England, and the
United States in three different years at different stages of the business
cycle shows that the share of materials in the total is almost identical.
The share of wages is highest in England and lowest in the United States.
The best international exports we can make are those of knowledge—
how the suffering nations may soon reach the level of highest produc­
tivity. We have ourselves reaped enormous profits through such ex­
changes. In showing others how to increase productivity, we teach them
how to help themselves to better and better standards of living.
C harles M erw in (International Monetary Fund): All of the diffi­
culties encountered in making comparisons of productivity at the national
level, the industry level, the plant level, and the job level arise in an
accentuated form in international comparisons. The problem of index
numbers is even more difficult to resolve on a spatial basis than on a
temporal basis. In using broad measures, there is the difficulty of deter­
mining what exchange rates to use. Finally, differences in definition and
classification usually present more practical problems in an international
comparison than in a temporal comparison.
Nevertheless, a comprehensive comparison of productivity at the inter­
national level should be attempted, and the appropriate time is the
present, when we can profit from wartime experience. It would be neces­
sary to have an international working party. This working party would
try to define the problem, uncover usable data, and determine differences
in productivity for the economy as a whole, for particular industries, for
particular products, and even for particular jobs. Finally, such a working
party would appraise the strategic factors accounting for the differences
in productivity.
Such a study could be sponsored on a nongovernmental basis by, say,
philanthropic foundations in two or more countries, or it could be done
on the governmental level. Since the Board of Trade in Great Britain
has engaged Professor Rostas to study the productivity of competitor
countries, an obvious international working party would be, say, the
Board of Trade and the U. S. Department of Labor. Private business
enterprise should participate in such a study. Coriporations which operate
on an international scale should be an ideal source of information as to
details concerning productivity on the job and in the plant.
H ans S taehle (International Monetary Fund) : The usual procedure




30
in dealing with the question of comparisons of the change in productivity
over time within one country with the change in another country, is to
compute a production index, which is weighted by value added at one
particular, moment of time. On the other hand, the employment index
usually is the unweighted, unadjusted sum of all man-hours worked.
In any kind of index-number computation it is always conducive to
clarity to consider an index number as a weighted average of some
variable. Since the variable usually measured is the ratio of output to
man-hours, such an index gains in clarity if it can be built up as a
weighted average of such ratids, and then combined into some aggregate
expression.
Interpretation of the usual type of index (based on a production index
weighted by base-year value added) in terms of the original industry
indexes is extremely difficult and leads to a weighting system which looks
at least awkward in symbols. If a weighted average of indexes pertaining
to particular industries is desired, then of necessity weighting by man­
hours should be used. In this case, the index is clear-cut, and its compu­
tation would not exclude such other extra measures as might be found
useful or necessary for other purposes. Such an index would measure
the average change in productivity per man-hour actually worked. The
weights should be the man-hours worked currently, year for year, in
each industry.
For the purpose of checking on this idea, the speaker had taken
IS out of 17 of the industry groups that Dr. Fabricant presents in his
work, and had computed output per man in each one, weighting them
together with shifting weights, year for year, by the number of men
actually at work. Finally, an over-all and very rough correction was made
for the disproportional change in total man-hours as compared to total
employment in terms of the number of men.
For the period 1899 to 1939, the speaker computed from Dr. Fabricant's productivity index an average annual rate of increase of 2.98
percent. Taking the index computed with changing employment weights,
the result of 1.95 percent a year—a tremendous difference—was ob­
tained. For the period 1899 to 1929, Fabricant’s index leads to an average
increase of 2.94, whereas the speaker got 1.79 percent.
The speaker by no means subscribes to the idea of a normal continuing
increase in productivity. Conceptually, there is no doubt whatsoever that
there can be no automatic, virtually self-perpetuating increase in produc­
tivity of that sort.
Any rates of increase, over time, of productivity in a country are only
loosely related to the increases in real income as it can be measured, for
example, by deflating money incomes. Thus, if a country imports a large
part of its food supply and food becomes cheaper, real incomes will



31
increase. However, if that decrease in food prices is due to increased
efficiency in other countries, it is one more instance of the world solidarity
in greater efficiency.

Discussion
P rof. J. R. H icks (University of Manchester, England) : The more
price and value are used as weights, the closer you bring the discussion
to the general field of national income. Possibly the next thing which
needs doing is to get the national accounts divided up properly by indus­
tries, so that the contribution of different industries becomes clearly
appreciated. A good deal of the discussions at the conference have tended
more in that direction than in the direction of productivity. Productivity
work proper does have a further function, in the direction of trying to
explain why the real national income, or contributions of particular
industries to the real national income, move the way they do.
S olomon J abricant (National Bureau of Economic Research) : The
differences that Mr. Staehle mentioned in the indexes of manufacturing
output compiled with different weights were very much larger than those
found by the speaker's own experimental computations along the same
line, and it would be well for Mr. Staehle to check his computations.
The consistency criterion mentioned by Mr. Staehle can be met in several
ways. One way is to define the output figures for the individual industries
so as to attain that consistency. An index of “net physical output,"
something analogous to net value added, will yield a measure for an
individual industry such that when averaged for all industries the result
is consistent with deflated national product. Moreover, such an index of
net physical output takes into account some of the points Mr. Barkin
mentioned. For example, savings in materials would result in a greater
rise in net output than in the usual indexes of “gross physical output."
C harles R oos (Econometric Institute, Inc.) : It would be desirable
first to define the purpose for which a productivity index is to be used
and then to build one to meet that purpose. Since the greatest productivity
changes are in the newer and growing industries, national productivity
might be the summation of growth functions and it is thus wrong to use
a logarithmic function to describe productivity. With respect to costs of
distribution, if the- customer is willing to pay for services, it is up to
industry to provide the service and it is not our province to say that the
cost of distribution should be lowered. Finally, it is important to state
very clearly how any index is constructed. If that is done, there will be a
use for almost any weighted index.
E. J. R iches (International Labor Office) : The ILO would welcome



32
suggestions on means to insure that the basic data collected in different
countries are as nearly uniform as possible and that methods followed in
making measurements of productivity are standardized internationally.
In drafting proposed international standards for employment statistics
for submission to the Sixth International Conference of Labor Statisti­
cians, which will convene in Montreal in August 1947, careful con­
sideration will be given to the discussions of this Productivity Conference
and to suggestions which members in the conference may send to
the ILO.
OCTOBER 29—MORNING SESSION

Chairman, Isador Lubin, President of the American Statistical Association

Scope and Limitations of Existing Measures of Productivity
D uane E vans (U . S. Bureau of Labor Statistics): All the participants
have received a bibliography which attempts to list the main references
to quantitative work in the field of productivity. The Bureau of Labor
Statistics should be informed of important omissions, since the bibliog­
raphy may be reissued later.
The Bureau of Labor Statistics has been working in the field of pro­
ductivity since at least 1898. Its present work owes a substantial debt
to the activities of the National Research Project on Reemployment
Opportunities and Recent Changes in Industrial Techniques. Continuing
work in the Bureau of Labor Statistics was started in 1941 with a staff
of 25 people, but within the past year the number has grown to about 65.
At one time or another, the Bureau of Labor Statistics has done some
work on almost every aspect of productivity. That part of its work
which has probably received the most attention is the preparation of
productivity indexes, or indexes of physical output per man-hour. Before
the war, there was the invaluable help of a Census of Manufactures
every 2 years, but such a census has not been taken since 1939. There
may be one covering the year 1947, but the results will probably not be
available until 1949. The census is especially valuable because it presents
necessarily comparable production and employment statistics for the
same establishments.
Since 1939, employment reports have been obtained from one source
and production reports from others. The question of comparability be­
tween the production and labor reports presents continual difficulties.
The Bureau has systematically examined the statistics collected by all
agencies, trade as well as government, to see whether there is any
material which can be used to prepare additional indexes of output
per man-hour, and this work will continue.



33
None of the indexes is perfect. It is a commonplace to the statistician
that all figures of broad coverage include imperfections. The real ques­
tion is the permissible range of uncertainty. The guide on this question
is the following: Would anyone be likely to reach a different policy deeision because of errors or imperfections in the figures ?
The Bureau naturally receives criticisms of its indexes and sugges­
tions for improvement. These are all carefully examined and appraised.
Moreover, there is a substantial amount of self-examination and selfcriticism, and a continuous effort to improve the validity and field of use
of productivity figures.
One of the criticisms made is that the man-hour figures include, to
some extent, hours which are paid for but not worked. If such hours
could be excluded, the indexes of output per man-hour would be some­
what higher. The Bureau’s figures on employment and pay rolls are
not collected primarily for the purpose of computing productivity
measures. The series are in part designed so that there will not be erratic
variations in average hourly earnings. The inclusion of paid vacations in
the man-hour figures tends to balance out the average hourly earnings
figures during those periods of the year when vacations are common.
It may be possible that some correction can be made to exclude hours
paid for but not worked, but productivity work is not the prime factor
in determining what type of information is collected.
Another problem is that the employment series collected by the Bureau
are based on month-to-month changes in identical groups of plants. Such
series require periodic adjustments to a benchmark. At present, many
of the series are not adjusted, and the general direction of the bias is
downward. The speaker is more concerned about this problem than any
other single source of bias which might affect the indexes.
Several statements which had been made, to the effect that the field
of use of these indexes is limited, are heartily endorsed. The indexes have
been incorrectly used, and as a result there has been criticism of the
BLS. It may be that the BLS has failed to qualify the indexes sufficiently,
but even when extensive and lengthy qualifications are attached to an
index, the index is sometimes stripped of the qualifications and presented
as immutable fact. For example, the Bureau has not been able to prepare
any measure of output per man-hour which covers all manufacturing
during the war period. Measures have been prepared for a limited
number of nonwar industries, mostly those producing nondurable goods
or materials. The average for that limited number of industries has been
used in some instances to represent manufacturing as a whole. This is
misuse or misrepresentation.
The Bureau of Labor Statistics is compiling a brief report for each
industry for which it has published indexes of output per man-hour.



34
These reports will include all the basic figures, all the weights, and such
analysis as can be made of the effects of various possible revisions. The
BLS will send these reports to interested people in the unions and com­
panies in the industries concerned; their criticisms and comments will
be invited. No promise can be made that this will be done for all the
industries, but an effort will be made to complete the task before final
indexes covering the year 1946 are published.
A new program, now under way, is designed to yield! for a number
of industries more satisfactory measures of annual change in output per
man-hour than have been available in the past, and more particularly it
will give productivity measures for durable-goods industries about which
nothing previously was known. In each industry selected for study, a
group of products is chosen to represent the industry. Careful specifica­
tions are drawn up for each product, and a sample of plants manufac­
turing each product is selected. From each plant, the Bureau is attempting
to secure an annual report on the average number of man-hours required
to produce the specified product. The direct reporting arrangement elimi­
nates many questions of comparability, makes productivity measurement
independent of secondary sources of information on production and
employment, and permits the collection of collateral information explain­
ing the reasons for changes in output per man-hour. The survey in each
industry can be designed to accommodate the special problems and
operating characteristics of the industry.
Statistics of productivity, or of anything else, are tools; they are not
themselves things. Recognizing that figures are tools, it is up to us to
understand them, to polish them as well as we can, and to apply them to
the particular jobs which they are best fitted to do.
LIMITATIONS OF PRESENT BLS MEASURES OF PRODUCTIVITY
AND LABOR COSTS

K atherine P ollack E llickson (Assistant Director of Research,
Congress of Industrial Organizations) : The BLS indexes of produc­
tivity and unit labor costs have serious shortcomings and have expanded
unduly to fill the void in this important field. It is the responsibility of
the BLS to try to avoid this result, first, by hemming in the indexes with
a sufficient explanation of their limitations and, second, to seek to fill
in the gap in other ways so as to present a properly balanced picture.
In addition, the indexes should be carefully reviewed in order to improve
them where possible, or to terminate them if they are unreliable.
The following are specific limitations of the indexes which the Bureau
should both seek to remove and explain fully in all releases and state­
ments :




35
(1) Production patterns within many industries were greatly changed
by the war, so that the degree to which the output series represented
each industry undoubtedly fluctuated considerably. Government-procured
items generally were made to higher specifications and required more
work than civilian items, as for example paratroopers' boots as com­
pared with ordinary shoes. It would seem to be incumbent on the BLS
to make an analysis of this matter generally available.
(2) Various systems of weighting may be used to combine the compo­
nent parts of the production series for each industry. It would be helpful
to explore the effect of different bases for weighting, to indicate the
possible range in which the indexes might vary if different methods
were employed.
(3) If the employees covered in the man-hour series are making
products not represented in the production series, then obviously errors
result. Many companies changed their products during the war but con­
tinue to be classified according to the 1939 basis in the compilation of the
BLS employment and man-hours data. Although the BLS dropped
indexes for industries in which it knew that a considerable shift in the
nature of products had occurred, no thorough canvass was undertaken in
the remaining industries to make sure that the employment and man-hours
figures actually matched the products with which they were compared.
It should be possible to make some check for at least some of the
industries, in the same manner as attempted by Mr. Barkin for the
synthetic-yarn industry. At the very least, this limitation should be men­
tioned, not just here but in all publications.
(4) The BLS employment and man-hours data are artificially inflated
by the inclusion of time paid for but not worked; that is, vacations,
holidays, call-in time, lunch time, sickness for which benefits are paid,
and so on. The BLS specifically asks the reporting firms to include in
their employment and man-hours reports all hours which were paid for
but not worked. Since unions have won very marked gains along these
lines in recent years, a definite downward bias results. Thus, each week
of vacation with pay during which no work was performed adds 2
percent to the man-hours reported for the employees affected. The BLS
Industrial Relations Branch reports 2 million workers as covered by
union contracts with provisions for paid vacations in 1940, but 11%
million in November 1944. Vacation time is only one of the types of time
that are paid for but not worked. This question requires careful investi­
gation and possibly some adjustments; it could mean an error of 3 to
5 percent.
(5) Also, the BLS employment and man-hour series are affected by a
downward bias that results in part from delay in securing information
on new firms. Unemployment-compensation 'data are used to check BLS




36
employment reports for major industry groups, and some subgroups
are also being adjusted by these benchmarks, but some downward bias
remains. This operates to offset the upward bias in the man-hours data
etc., but there is no reason to assume that the errors cancel each other oht.
(6) The BLS indexes of unit labor cost are derived by relating pay
rolls to output per man-hour. All the errors already mentioned therefore
are carried over into the labor-costs series, and some may well be com­
pounded. The pay-roll figures include items, such as premium pay for
night shifts, which are granted labor not merely because of the incon­
venience to the workers but also because of the savings resulting to
employers. Obviously, running two or three shifts produces savings in
overhead which are likely to more than offset the premium pay, so that
including it in unit labor costs produces a misleading result.
(7) Output per man-hour as measured by the BLS is too narrow a
concept of productivity. Economies in production brought about by better
integration of industries, substitution of improved materials, and more
economical use of fuel and power are not taken into account. The BLS
should place its series in proper perspective by discussing these factors.
(8) The 29 nonmunitions industries for which the BLS has issued
indexes in recent years are unrepresentative of all manufacturing be­
cause they are industries in which the effects of the war tended to
impede rather than enhance productivity advances. Autos, aircraft, ship­
building, steel, and all the metal-fabricating industries are otiiitted. A
good many nonessential industries are covered, which suffered especially
from shortages of manpower, materials, repairs, and equipment. As a
result, the BLS release for 1945 and earlier releases have erroneously
conveyed the impression that productivity did not rise during the war,
and in 1945 was only slightly above 1939 levels. Since the data have
already been released, the BLS should immediately issue an official
statement on their proper interpretation.
Although the preprint of the December 1946 Monithly Labor Review
article partly covered the speaker’s proposals about interpretation of these
indexes and putting them in proper perspective, the article still relied too
heavily on the indexes and implied that they had a degree of accuracy
which cannot be substantiated. All releases should carry an explanation
of these limitations.
Special industry studies are being undertaken by the BLS, which will
follow through changes in man-hours required for specific operations or
production processes. This approach is worth exploring, but there is
grave danger that the products of operations selected will be considered
representative of the industry when they are not. The most stable opera­
tions are those which can be most easily studied. In areas where rapid
change is taking place, continuous measurement may well prove impos­



37
sible. The reporting companies know very well that labor may use the
results to seek wage increases. There will be a temptation to err on the
side of understating increases in productivity, not through misrepresen­
tation of the figures so much as through selecting processes which, for
some reasons, the employer may think are more representative of the
industry, but which in many ways are not.
The planning of these special studies, including the choice of operations
and the review of results from time to time, should be done in consulta­
tion with union representatives. Although the BLS may be placed in the
dilemma of making the studies on the employers’ terms or not making
them at all, the Bureau should take a firm stand against employer pres­
sure and should explore the possibilities of compulsory reporting of
essential information.
The BLS should explore new methods of approach along lines sug­
gested by this conference. It should offset the erroneous impression
created by its indexes of output per man-hour and labor costs by explain­
ing the wartime factors which interfered with full development of pro­
duction potentialities. It should also stress the tremendous increases in
productivity which occurred in war industries, the vast effects that can
be expected from plant expansion, the significant increase in machine
tools with harder cutting edges, the pooling of research information, the
stimulation of research by the Government, the acquisition of German
patents, etc.
Instead of simply assuming that the long-time trend of a 3-percent
increase in productivity will continue after the war, the BLS should
be alert to explore whether this long-time trend may not have been
accelerated by the wartime developments. Only through a broader ap­
proach can the BLS place its indexes in proper perspective and avoid the
harmful, even though unintentional, underestimation of the increase in
productivity in our economy.
PROBLEM OF PRODUCTIVITY MEASUREMENT IN THE
SHIPBUILDING AND SHIP-REPAIR INDUSTRY

R osalind S chulman (Research Director of Industrial Union of
Marine & Shipbuilding Workers, CIO) : In peacetime shipbuilding, like
large locomotive production and large turbine building, is custom produc­
tion. There has not been, except during wartime, any measure or attempt
at measure of production in this industry. The Bureau of Labor Statistics
has published a study of productivity changes during the war, when
large numbers of vessels were built, but this study has absolutely no use
for peacetime production.
In any form of custom production, certain operations remain compara


38

lively stable over a decade or so. It might be possible to develop a series
of ratios measuring operational cost as adjusted to quantity of opera­
tional production. Such a series of measures would reflect, over a short
period, major technological change and improvement and major changes
in labor utilization.
Rapid technological changes in phases of construction, which perhaps
cannot be measured on an operational basis, could make the index for
one particular operation meaningless, but if a series of commondenominator operations were developed, the limitations of any single
operation might be overcome. Another limitation is that the earnings
figures may be inadequate because of modifications of rate structures
and modifications of the operations themselves. Continual changes in
design will also invalidate this approach.
However, if a series of operational ratios is developed and used over
not too long a period, valuable information may be obtained on the
custom-production industries which cannot be derived in any other way.
SCOPE AND LIMITATIONS OF EXISTING PRODUCTIVITY MEASURES

A ndrew Court (General Motors Corporation) : The speaker cannot
agree that Government wartime research has made a large contribution
to all types of manufacturing industries. It has in some areas, but it has
not in others.
In the auto industry, the poundage of product per man-hour, computed
according to a crude formula developed by the U. S. Department of
Commerce, showed 18y2 pounds per man-hour in 1941. For 1946, the
output per man-hour has been running less than 14 pounds.
It is possible to compare accurately the productivity of different indi­
viduals on the same job. Where the union permits, such comparison
enables the employer to reward individual effort and skill. This tends
to maximize the income of the workers. Comparison of different jobs at
a given period, or of jobs which change from period to period, is ex­
tremely difficult or impossible. For any industry that is evolving, pro­
ductivity comparisons over a long time span are likely to be crude, at
least. In the speaker’s opinion, attempts to measure or compare plant
productivity are practically useless as factual tools in negotiating changes
in wage levels or other union-management economic issues. The whole
idea of attempting to base wage rates on productivity in a specific
company or industry is inconsistent with equal pay for equal work.
Much of the change which occurs in productivity is due to forces
developed outside the industry in which the results appear. Major
factors in the high productivity of the American motor-vehicle industry
include a large market frefe of tariff walls, good roads, and progressive



39
supplying industries such as rubber, steel, glass, and machine tools.
Another important factor is the American ideal of free competition.
Industry-wide production planning and pricing have probably restricted
competition and the growth of the industry in England and Europe.
Standardization is also of great importance in the motor-vehicle industry.
It is possible because Americans are willing to buy cars which look alike.
Thus, it is very difficult to interpret industry productivity data.
On the national level, there is a great delusion that productive capacity
is unlimited and that the only problem is to provide mass purchasing
power. Actually, the farms and factories of this country have never pro­
duced nearly enough to provide a fully satisfactory standard of living
for everybody. For example, the amount of meat and fat and cheese
required by the BLS maintenance budget was about twice the amount
that could be purchased with ration points in 1944 and 1945.

Discussion
B oris S tern (U. S. Bureau of Labor Statistics): The productivity
data published in the past by the Bureau of Labor Statistics have been
extremely useful in describing, in general terms, developments with
respect to machinery, increased productivity, higher wages, and shorter
hours. The work the Bureau may do in the future may become more
effective because it is now a more-integrated agency, it has more skills,
and the need for productivity data is more generally recognized.
There is a serious danger, however, in that this new tool is being used
for purposes for which it is not intended and cannot be effectively used—
collective bargaining. The data hitherto available cannot be used for col­
lective bargaining except in a general way for describing the history of
the industrial arts or the effect of the use of machinery on our standards
of living.
If productivity information is desirable for collective bargaining, labor
and management should agree, industry by industry, on what methods
should be used to develop such data. With such agreement, the Bureau
of Labor Statistics would be ready and able to work with labor and
management in the development of the data, provided labor and manage­
ment also agree that the results obtained will be acceptable irrespective
of whether they show a decline or an increase in productivity. Unless
there is such agreement, it must be recognized that although productivity
data are useful to illustrate certain developments nationally or in various
types of industries, they should not be used for collective-bargaining
purposes.
F rank R. G arfield (Federal Reserve Board) : The concept of pro­
duction is basic to productivity measures. If the concept of production



40
is considered in relation to the final product and the scale of living, it
is clear that many factors enter into the production process. In any par­
ticular industry, materials and fuels are consumed and transposed into
finished products by workers, machinery, and management. Electricpower consumption obviously is not a satisfactory measure of the whole
production activity in an industry, and neither is man-hours. In the tex­
tile industry, it takes a great many man-hours to produce $10 worth of
net product; in the chemicals industry, very few man-hours. For meas­
ures such as gross national product and national income, the whole com­
munity of statisticians is agreed on including all of the product and in­
cluding it in value terms.
It is true that when values added are used as weights, elements of
the market enter into the picture. Value is influenced by decisions of
the customers as to what they are willing to pay and by the decisions
of the producers as to what they are willing to receive for their services
and their goods. Granted that there are elements of monopoly and of
Government price fixing, the value-added figures reflect, as nearly as
anything, the combined judgments of the community. Therefore value
added for the most representative period for which data are available
should be used as weights.
S olomon F abricant (National Bureau of Economic Research): The
National Bureau of Economic Research has published studies on pro­
ductivity, production, and employment in manufacturing, agriculture, and
mining, and is now working on other sectors of the economy. Its indexes
are simply byproducts of these studies. The National Bureau is inter­
ested in the factors affecting the growth of our industries, or, in some
cases, the decline. Studies are made not only of the factors affecting
productivity, but also of the relations between trends in productivity, in
employment, in production, in capital, in prices, in wages, and so on.
J o h n D. G ill (Atlantic Refining Co.) : Regarding Dr. Stern’s re­
marks, we can use the present productivity measures in bargaining, but
we need to understand that they are not accurate and to use them as
approximations. Unless there is a balance between our power to pro­
duce and the ability to clear the market of that production by appropriate
measures of compensation, there will be severe dislocations; thus we
need to know something about the increase in production which becomes
the basis for collective bargaining.
Economic measures which lack the dollar mark should be used only
when it is impossible to make a report based on dollar magnitude. Be­
cause of competition, it is possible to lean heavily on accounting profitand-loss statements which make possible a comparison of the results of
any operation with the operations of comparable firms elsewhere.



41
If the general change in production—cither gross national product or
national income—is used to measure the general change in compensa­
tion, there remains only the problem of administration. A 10-percent
increase in wages in a mass-production industry may be possible with­
out an increase in prices, hut a 10-percent increase in wages in a firm
hand-tooling belts, for example, may require an increase in prices and
may result in* the loss of the market. We should anticipate such changes
and prepare either to substitute machine-worked belts for hand-worked
belts or go out of business.
M artin G ainsbrugh (National Industrial Conference Board) : One
desirable result of the conference might be that the BLS prepare
a reply to the CIO criticisms of the BLS indexes of productivity.
C elia S tar G ody (U. S. Bureau of Labor Standards): The ques­
tions raised by Mrs. Ellickson concerning the BLS productivity indexes
might be separated into three categories. The first question is whether
the measures are correct for what they are intended to show. It is agreed
that the indexes are not correct to the last decimal place, but many of
the factors mentioned are much less significant than they are generally
believed to be. For example, on the question of the time paid for and
not worked, if every worker had a 1-week vacation, there would be
an error of something less than 2 percent in the man-hour series. The
actual average is doubtless considerably less, since many workers are
not covered by vacation provisions, many who are covered have not
worked long enough to be eligible, etc. The error in the man-hours
index is the difference between the extent to which there are paid vaca­
tions in the base year and the extent .to which there are paid vacations
in the year under consideration, and may be in the neighborhood of 1
percent. When there have been more significant changes in the reporting
of man-hours, the BLS has made explicit adjustments; for example, in
coal mining, for which, after the change to portal-to-portal pay, the re­
ported hours figures included travel time.
It is true that there have been, in many industries, changes in the
specifications of products, new types of goods, and changes in the pro­
portions of different products. The BLS has not continued to show
indexes for those industries in which such changes were so great that
it was impossible to make any comparison. For the remaining indus­
tries, the Bureau attempted to learn all it could about the changes in
the nature of the products and to make tests of the accuracy of the pub­
lished indexes. For example, it has been stated that the index for the
shoe industry is too low, since military shoes are bigger and heavier
and presumably better than civilian-type shoes. The information ob­
tained indicates that the military work-type shoe does require more



42
labor than a civilian work-type shoe, but less labor than a civilian dresstype shoe. If the index is reweighted to take this factor into account,
the result differs by only 2 percent from the published figures.
In cases in which the indexes incorporate weights for some old period,
tests have been made, changing the weights to the year 1939. In prac­
tically all cases, this reweighting has resulted in only small changes in
the indexes.
Two problems raised by Mrs. Ellickson—the possible bias in the em­
ployment series, and the comparability between the production and
employment series—cannot be solved completely until there is another
census. Many of the industries for which indexes are shown are fairly
well defined and it is possible to state with confidence that the indexes
include their total production and that the production figures are compar­
able with the employment series. In other cases, varying amounts of
products are not included in the production series and there is no infor­
mation on whether there has been any shift in the importance of such
products. A complete evaluation cannot be made until there is another
census; until then, all that is possible is to determine from available
sources whether there is any indication of such shifts.
A second question concerns the factors which make the indexes go up
or down. These are not limitations, but rather explanations, of the in­
dexes. For example, where there are special wartime factors which make
for declines in some of the indexes, it is important to know them, not
because the indexes are not good measures but because it should not be
assumed that the wartime levels will be the levels in the future when
those factors have disappeared. In the building-materials industries, for
example, there were substantial drops in output per man-hour during
the war simply because of the decline in production. It is important to
know the reason, because it can definitely be expected that as soon as
production increases, output per man-hour will also rise. This has actu­
ally happened during 1946.
A third question is what additional measures would be desirable.
Output per man-hour is only a partial measure; the ratio of output to
man-hours in itself does not give a complete story of what is happening
in the economy. For example, data on unit labor cost must be used to­
gether with other information. It is necessary to consider changes in costs
of materials, in overhead, and in all the factors which enter into the cost
structure. The Bureau would like to do as much as possible in the way
of providing additional measures of productivity in this broader sense.
Whether it can do so depends, first, on the availability of information on
these other factors — materials, overhead, etc. — and, secondly, on the
availability of staff for such work.
D uane E v a n s : Mr. Garfield's comments on the desirability of weight


43
mg with value added were probably based on his experience with the
Federal Reserve index of industrial production. Conceptually such a
measure tells us what the value of our production would be, relative to
the base period, had there been no changes in the prices paid for all
factors used in production, including labor. The speaker knows no better
method of measuring the general level of industrial activity than the one
used by the Federal Reserve Board.
In the use of man-hour weights, there is an attempt to answer a dif­
ferent, clearly stated problem: If we were to reproduce the physical
amounts produced in some base period, how many man-hours would
it take relative to the man-hours used in the base period? This physical
approach is useful at the plant and industry level. The bigger the sphere
encompassed, the less the utility of this approach because the necessary
figures cannot be obtained.
It is believed that the other points which have been raised will be met
by the analyses of the indexes which will be sent to the interested parties,
and that these statements will also answer Mr. Gainsbrugh’s point.
OCTOBER 29—AFTERNOON SESSION
Chairman, Thomas Blaisdell, U. S. Department of State

Need for Additional Productivity Measures
L azare T eper (International Ladies’ Garment Workers’ Union,
A F L ): The significance of the physical output man-hour 'ratio will be
examined, first under a set of theoretically controlled conditions and
then under conditions complicated to bring them into relation with
reality.
First, consider a single plant which turns out a single product and
assume that the product manufactured by this plant did not undergo
any changes in its specifications; that the raw materials used remained
similarly unchanged; and that there were no changes in any of the
manufacturing methods or equipment. Under such conditions, varia­
tions in the physical output/man-hour input ratios can be interpreted
as indicators of changes in the relative average efficiency of the labor
force. Of course, the ratio does not, in any way, throw light on the
causes responsible for the changes.
Next, assume that the theoretical plant changed its methods and tech­
niques of internal operations (including installation of different capital
equipment). Since the factors responsible for changes are no longer
exclusively on the labor side, this concept of efficiency changes must be
deemed a function of the many independent and interdependent forces
which take labor, management, and equipment into account.



44
If the specifications of the product turned out by the plant have also
been evolving, the ratio of physical output to man-hour input will reflect
not only the changes in efficiency, but the effect of changes in the specifica­
tions of the product as well. When additional factors are permitted to
influence the quanta of the physical output/man-hour input ratios, the
meaning of the resultant statistics is complicated to an even greater extent.
Thus, the nature of integration between the productive processes of a
given plant and those of its suppliers and customers changes over a
period of time, and such changes either increase or decrease the need
for workers. Similarly, changes in the specifications, character and
nature of raw materials may also affect the plant's worker requirements.
As long as it is impossible to segregate the various factors which may
be responsible for the fluctuations in the physical output/man-hour
ratios, the designation of such ratios as measures of productive efficiency
seems hardly tenable.
On the other hand, a statistical series showing the ratios of man-hour
input to physical output does serve as an indicator of manpower utiliza­
tion—measured in units of time input—within the confines of a partic­
ular plant, needed to turn out a unit of product by changing specifica­
tions under the conditions of changing efficiency of the various factors
of production and of changing position of the plant in the economy.
These conclusions apply equally well to a group of plants manufacturing
a single article.
If a plant is engaged in the production of two or more commodities,
the meaning of the data is further affected by the inability to combine
noncommensurate entities without converting them into some other
quantities, commensurate with each other as a matter of statistical con­
venience. Thus, the technique recommended by the National Research
Project and the Bureau of Labor Statistics permits the computation of
a combined index for a group of products by weighting “manpower
utilization" figures for the individual commodities by the units of physi­
cal output for some preselected “weight" base year. An index so com­
puted for more than one product, in accordance with the NRP-BLS
techniques, should be described in terms which denote its significance as
an index of manpower utilization (measured in units of time input)
needed to produce “weight" base-year units of physical output, as in­
fluenced by changes in the product specifications, efficiency of the dif­
ferent factofs of production, and the nature of the plant's (or industry's)
relationship to the economy.
The theoretical advantage of an index computed on a fixed-weight
base lies in the fact that it measures but a single element of change, to
the exclusion of others. Yet, in this case, there is the infusion of many
factors, each of which changes in time. As long as such influences are



45
not eliminated, one can properly raise a question as to whether an index
number, which would incorporate in its make-up the effect of all of
the dynamic changes which affect the economy, including changes in
the relative importance of the different products, would be of greater
significance for economic analysis. Such an index would reflect changes
in the relative importance of the different products in the course of time,
and would permit the inclusion of new products as they make their
appearance. It could be described, in terms of its significance, as an
index of manpower utilization (measured in units of time input) needed
at different times to produce a unit of physical output in accordance
with specifications, efficiency, and nature of economic organizations at
each of the periods under consideration.
Since “productivity” indexes do not portray changes in productive
efficiency, their designation as such should be abandoned and full qualifi­
cations made as to their meaning at the time of publication. At the same
time, it would seem desirable for the investigators of productive effi­
ciency to direct their attention, to a greater degree than heretofore, to
the collection of descriptive information which is a necessary prerequisite
before refinements in the productivity measures are attempted on a large
scale. Research should be undertaken on the plant level to find out, in
greater detail, how changes in productive efficiency are brought about
and the influence thereon of machinery, management, individual effort,
flow of work, demand for goods, specifications and variety of product,
nature of raw material, and so on. Such studies should investigate how
changes within the plant affect -the skills and composition of the labor
force, and what displacements, if any, changed methods of production
bring. Statistical investigations, dealing as they do in averages, may actu­
ally conceal the full importance of such changes.
More information is needed about the effects on workers of manage­
ment pressure for higher output, and the necessity for earning a living.
Are there socially desirable optimum levels of human effort beyond which
it is undesirable to go? What is the effect of the existing methods of
incentive payments on the workers' well-being? What is the effect of
management's behavior on the efficiency of workers and the enterprise?
More information is needed on the influences, from outside the plant,
on its productive efficiency. To what extent do changes in the demand
for goods or competitors' activities induce variations in the production
methods? How rapid is the process of change as between the different
plants? What is the relation between capacity output and productive
efficiency? How is productive efficiency affected by changes in the na­
ture of integration between the plant and other businesses? What is
the effect of developments in the field of communications and trans­
portation on the plant?



46
The present interest in the question of productivity stems from a
universal desire to find the clue to a better standard of living. If it is
held that the answer lies in greater productive efficiency, there is no
question but that econometric studies must be supplemented by investiga­
tions of cause, effects, and byproducts of such change. It is Ito this
area that the attention of investigators of productive efficiency must
turn.
In making studies at the plant level, it is particularly important that
such plant studies be undertaken by an impartial agency such as the
Government, with joint cooperation of advisory groups representing
both labor and management.
(Mr. Teper presented an additional paper criticizing the formula used
by the Bureau of Labor Statistics in computing its indexes of unit labor
cost. The paper indicated that the appropriate weights for a production
index to be used in deriving unit labor cost were pay rolls per unit of
output and not man-hours per unit.)

Presentation of Productivity Measurements
S olomon F abricant (National Bureau of Economic Research):
The speaker, in behalf of those who produce* productivity measure­
ments, believes they must do the best possible job of presenting such
measurements. They should continue to review terminology and should
define terms. They should make clear what the figures mean and note
some of the things they do not mean. They should be explicit about the
sources of the data, about accuracy, about limitations. They should note
the methods of combination, the weight systems, the weight-base pe­
riods. They should compute alternative measurements and publish them
now and then, if only to dispose of questions concerning the importance
of the difference between them and the figures issued regularly. They
should have a base book, revised and expanded every* so often, to which
they can refer in the shorter public statements.
They should call attention to the representativeness and homogeneity
of their samples. When the samples are not homogeneous, they should
avoid combining the individual items into an average. When an aver­
age for a homogeneous group is prepared, but the group is not rep­
resentative of the whole of which it is a part, they should avoid imply­
ing that it is.
When they don’t know, or are not sure, they should say so clearly
and emphatically. Whenever they revise, they should spell out the rea­
sons for the revisions, and the uncertainties still 'remaining even in
the new figures. They should dissociate questions of fact from matters



47
of policy. They should dissociate guesses about the future from state­
ments about the past.
It is hoped that the consumers of productivity data really read the
notes and textual qualifications and refer back to the more-extended dis­
cussions provided before raising questions; that when they think of a
reason why the figures might be off, they won’t jump to the conclu­
sion that they are in fact off, or if off, that the error is always and
necessarily important.
Producers of productivity data welcome questions and suggestions,
but hope that such questions and suggestions would not be made, at least
in the first instance, via the public press. They have a weakness in par­
ticularly liking constructive suggestions.
All of us must have some common meeting ground, if our kind of
economic and political system is to survive. Established fact is an essen­
tial ingredient of this common meeting ground. There are enough dif­
ferences dividing us on which it is not easy to come to common agree­
ment. Facts need not be among them. Our fact-finding agencies should
be kept nonpartisan, for our mutual advantage.
From the discussion, during the conference, of the limitations of the
BLS indexes of output per man-hour, it is feared that some people may
get the impression that these indexes are of little value. That impression
would be quite wrong. The indexes for the period since 1939 cover only
a few industries. It is true, too, that they are subject to adjustment when
the censuses come out. Indeed, like all statistics, they are surrounded
by a margin of error. This does not mean, however, that BLS indexes
are subject to so much error that they are worthless. On the contrary,
for -the industries to which they relate and which they are intended to
cover, they convey useful and valuable information,

Discussion
Irving H. S iegel (U. S. Veterans Administration, formerly with the
National Research Project and the Bureau of Labor Statistics): Con­
sumers of productivity statistics have apparently expected too much,
or have ignored the cautions accompanying such measures as long as the
indicated changes conformed to their intuitions or interests. Though the
available indexes are probably the best that can be made with available
data, they have outstanding limitations which must be taken into ac­
count especially if they are used beyond their original purpose. In any
case, the differences concerning the definition and measurement of pro­
ductivity do not seriously impede solution of the important practical
problem which motivates much of the theoretical discussion, namely, the
maximization of output, however defined, of a given work force through



48
the improvement of production processes and of labor-management
relations.
G eorge B rown (United Association of Plumbers and Steamfitters,
A F L ): Labor people want to see accurate productivity figures developed
and want those figures to reflect efficiency. The study of productivity
can occur at two levels—first, in the area of academic interests, in the
development of industry and national figures; and, second, in the area
of practical application, at the plant level. Enough deficiencies have been
pointed out to indicate there is a large area for work at the plant level.
L yle C ooper (United Packinghouse Workers of America, CIO) :
The speaker's union has not found the BLS index for slaughtering and
meat packing good for bargaining purposes. The index is computed by
weighting with fixed prices. The Federal Reserve index, which is on a
1935-39 base, shows that man-hour output went up substantially more
after 1939 than the BLS index indicates. There is a good reason for
using several years as the base period, particularly in meat packing,
where prices of the different products fluctuate considerably. Between
1944 and 1945, the BLS index went up about 9 points, but a company
negotiator pointed out that this change was due to the fact that hog
slaughtering had decreased and beef (which is higher priced) therefore
received a greater weight. The BLS itself has indicated in correspondence
many limitations of the type described.
H enry B. A rthur (Swift & Co.) : The explanation of the change in
the BLS index for meat packing is the fact that BLS uses the total
value of the product as weight, rather than something approximating
value added. This gives importance to the two species—beef and hogs—
which represent the total input of the farmer as well as the packing­
house worker. Between 1944 and 1945, there was no change in the
total weight produced, but there was a decline in the amount of work to
be done because pork production, which requires a great deal of proc­
essing, declined.
In the presentation of productivity ratios, we have conventionally
shown a ratio in which the whole (production) is divided by the part
(man-hours), and this convention has occasionally led us into thinking
in terms of labor productivity. If a series of ratios is prepared, with
the various inputs in the numerators and the output in the denominator,
one might say that the relationship which changed least from one year
to the next was probably the factor which most nearly accounted for
the change in production. As an example, assume that man-hours per
unit of product decreased 20 percent and the amount of all the other
input factors per unit decreased 20 percent except power, with the kilo­
watt-hours used showing no change. As a first step it might be pre­



49
sumed that the increase in production was a result of the application
of kilowatts. These ratios show that it was something other than the
factor in the numerator that caused the change in the denominator.
In working with measurements, we are operating within a part of the
total field. Even the broadest measures omit about a third of our total
economy—the output in households. From a social point of view, there
should perhaps be an appraisal of the total values of our economy.
S olomon B arkin (Textile Workers Union of America, C IO ): The
emphasis which originates from the collective-bargaining process and
from the study of national incomes is on the desirability of obtaining
more all-embracing mathematical techniques for gaining insight into
increased productivity. It is also necessary to evaluate the importance
of our neglect to use some of our resources—both capital and labor—
and the importance of that part of our output, such as that of house­
holds, which is not converted into monetary business terms. Perhaps
what is needed is an assimilation of the concepts of measurement of
physical output with some of the concepts in the national income field,
plus some approach on the problems of untapped resources and un­
tapped manpower.
M argaret S cattergood (American Federation of Labor) : The AFL
members met during the noon period and made unanimous recommenda­
tions on two points: (1) They wish to emphasize the importance of the
case-study method and to urge its use, with representation of unions and
management on committees cooperating with the government agents.
(2) They believe the work of the conference should be continued through
a continuing committee on which labor, management, and government
are represented. They would like to refer to such a committee the ques­
tion of whether a set of figures for the economy as a whole can be de­
veloped and, secondly, the problem of defining or perhaps changing
some of the specifications of the present Bureau of Labor Statistics
indexes.
E verett H agen (National Planning Association): One of the rea­
sons for dissatisfaction with the use that has been made of productivity
data is failure to agree on the meaning of the term “productivity.” In
economic theory, the term means the contribution of a particular factor
to production; statisticians in the field use the term simply to refer to
output per man-hour, whatever the reason for the change. Another dif­
ficulty arises in the use of data covering more than one product. Such
an index can change simply because of shifts in the proportion of the
commodities, depending on the weights' used. We should not criticize
the figures because we are not sufficiently mathematically trained to un­
derstand them, but should realize that a figure cannot be all things for



50
all purposes and a technician is required for its proper use. A third dif­
ficulty is the inadequacy of the figures because of failure to take account
of all relevant circumstances.
The conference should create a continuing smaller body to continue
to examine the existing productivity series, to clear up misunderstand­
ings, and to investigate changes which might be desirable for new pur­
poses.
H iram D avis (University of Pennsylvania) : It might be more pro­
ductive to develop a series of ratios relating various input factors to
output than to attempt to measure total input. The conference might
consider whether it would be worth while to form working parties in
some of the industries in which there is disagreement over how output
should be measured.
D u a n e E vans (U. S. Bureau of Labor Statistics): Even the most
precise productivity figures are not necessarily a definite guide to the
way in which the production economies of the system can be divided.
In the long :run, output per man-hours has been going up, but there are
segments of the economy in which productivity has not increased, such
as some of the service industries. It cannot be maintained that wages
should not increase in such an industry if, over a period of time, gen­
eral earnings have gone up. If society wants the products, it has to pay
a high enough wage to keep people in that industry. On the other hand,
there are some industries, such as rayon, in which there have been ex­
tremely rapid increases in productivity. If wage levels in such an in­
dustry were increased accordingly, it would be necessary to have a guild
system to keep people from entering the industry, because wages would
be so much above the general level.
M r. B arkin : Those who suggest the need of data for collective bar­
gaining certainly do not have the ideas imputed by Mr. Evans. To pose
these theoretical alternatives does no service to the discussion of the
problem, but bars constructive thinking in this field. The problem is to
get a measure of productivity, so that the parties can then rationally
discuss and possibly bargain about the methods of distributing the gains.
H. H. B ookbinder (Amalgamated Clothing Workers of America,
CIO) : The speaker represents the workers in the shirt industry, where
the union expects wages to irise in the future, but hopes that productivity
as affected by worker effort will decline. Mechanization seems to have
reached its maximum limit, and the rate of production of a shirt de­
pends mostly on the fingers of the operator and not on the machine
itself. Because of the piece-rate structure of the industry, the worker
has driven herself beyond reasonable human levels in an attempt to



51
increase her earnings. On one important operation, for example, the
top stitch on the collar, a girl may work on as many as two to three
thousand shirts a day. This type of work has led to countless cases of
physical and psychological breakdown. In the future, it is expected, of
course, that general productivity in American industry will continue
to increase. Shirt workers should benefit from this general increase.
However, it is hoped that some decent production standards can be
agreed upon for industries like the shirt industry. As for sharing the
benefits of increased productivity, there will be times when labor will
ask for the extra share at the plant level, at times the industry level.
In the long run, however, it must be the economy level, either in the
form of generally higher wages or generally lower prices.
E w an C lague (U. S. Commissioner of Labor Statistics) : A govern­
ment agency can do effective work in studies of individual plants or
jobs, but only if there is general management-labor agreement on such
studies. Such data, while valuable locally, have very little significance
nationally. On the other hand, the general indexes were not necessarily
drawn up for collective-bargaining purposes. When any of them are
used for specific purposes, they should be reexamined to see if they
will fit. It is hoped that some of these measures can be made useful for
collective bargaining, at least for a first approach.

Summary of Conference
S am uel H. T hompson (U. S. Department of Commerce) : The con­
ference has been made up of people who believe in progress, in im­
provements, in the continued movement upward of the human race and
of our own industrial civilization and of the United States. The con­
ferees have agreed on several major areas.
The first is that, since it is desirable in the interests of all of us to
increase productivity, measurements are needed that will help find out
how to increase it. We need measures adapted for use as tools in the
analysis of the productive process that will help us improve the total
quantity and quality of production. We need many different kinds and
styles of tools. We certainly need physical measures of productivity,
based on physical production and the number of man-hours or manyea'rs. We also need monetary measures. There are many other meas­
ures we need—for example, kilowatt-hours per unit of output, pounds
of coal per kilowatt-hour, etc. The conference has agreed that we must
be clear as to what the measures we use mean and what they are useful
for. In index numbers, we necessarily put together things which are not
commensurate; we have to use index numbers, but must not forget that
fact. Speaker after speaker has emphasized the need for a clear under


52
standing of the purpose and then a careful design of the instrument. At
the same time, it is interesting to note how amazingly few people are
keeping systematic records from which the tools could be forged.
It has also been emphasized that we need to do a better job with what
we already have. The conferees have agreed that we use these measures
to analyze the factors that enter into production. It has been emphasized
that there are many factors which affect production and productivity.
The conferees have also agreed that the ultimate responsibility for
improvement in productivity or production cannot be assigned to any
single factor. The technological progress of our industrial civilization
arises from the whole of our society, perhaps the whole of mankind.
Mathematically, every individual of the race contributes in some meas­
ure, to either his own generation or the next. We can improve our per­
formance, and ultimately we all gain from each man’s individual effort.
There are other areas which were omitted or in which there was some
disagreement. One is the matter of the use of indexes in collective bar­
gaining. There has been general agreement on the need for further anal­
ysis leading to the design of better tools for specific use in reaching
collective bargains. The conference has not discussed very thoroughly
the need for both physical and monetary indexes and therefore physical
and monetary weights. It has not discussed at any length the actual
techniques that are available or should be devised for getting the basic
data.
The executive committee wishes to announce that there will be a
transcript of the meeting available for inspection, but there will be no
attempt to make it generally available. A summary will be prepared by
the executive committee which will be sent to all the participants, for
their approval or correction, which may later be published.
The executive committee will, if the conference approves, meet and
discuss ways and means of continuing the work of this conference—
whether through subcommittees, through recurrent meetings of this
size, or through a combination of these and other means.




☆ u. S. G O V E R N M E N T PR IN TIN G O FFIC E: 194 7-74 58 6 7