View original document

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

Federal Spending
and Scientist
and Engineer
Employment
BULLETIN 1663
U.S. DEPARTMENT
OF LABOR
Bureau of
Labor Statistics
1970

L ft., 3 :




Dayton & Montgomery Co,
Public Library

OCT i 9 I97Q
DOCUMENT COLLECTION




Federal Spending
and Scientist
and Engineer
Employment
A Study
in Measurement
BULLETIN 1663
U.S. DEPARTMENT
OF LABOR
J. D. Hodgson,
Secretary
BUREAU OF
LABOR STATISTICS
Geoffrey H. Moore,
Commissioner
1970

V

 For solo by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 •Price 50 cents





Preface
Recognizing the need for an assessment of Federal Government expenditures on
the demand for scientific and technical personnel, the National Science Founda­
tion provided financial support to the Bureau of Labor Statistics (BLS) to study
methods of measuring this effect. BLS hired Dr. Robert L. Aronson to carry out
the study during his leave of absence from the School of Industrial and Labor
Relations, Cornell University. Dr. Aronson was encouraged to express his profes­
sional judgment in carrying out the study. Therefore, points of view or opinions
stated in this report do not necessarily reflect official opinions of either the Depart­
ment of Labor or the National Science Foundation.
The number of persons in various public and private organizations who provided
assistance or counsel to Dr. Aronson during this study is too large to identify in all
cases. For their help, special mention is acknowledged to Roger W. Jones, Michael
March, and Carey P. Modlin, all of the U.S. Bureau of the Budget; Murray Weidenbaum, formerly of the Department of Economics, Washington University; Allen O.
Gamble, National Institutes of Health; Leonard Lecht, Center for Priority Analysis,
National Planning Association; and Paul Downing, U.S. Department of Defense.
Various members of the staff of BLS provided technical assistance, including Neal H.
Rosenthal, Michael F. Crowley, and Daniel E. Hecker, of the Division of Manpower
and Occupational Outlook; Thomas Fleming and Richard P. Oliver, of the Division
of Economic Growth; and Douglas Schmude and Edith Andrews, of the Division of
Occupational Employment Statistics.
Robert Cain and Norman Seltzer of the Sponsored Surveys and Studies Section,
National Science Foundation, were instrumental in shaping the goals and direction
of the study and in providing general counsel and assistance.




m




Contents
Chapter:
Page
I. Introduction and Highlights ..................................................................................... 1
II. Government Programs and the Measurement of Manpower Requirements
3
The concept of Federal manpower requirements ........................................ 3
The elements of a model ....................................................... ........................... 4
Data sources: Expenditures and manpower .................................................. 7
Approaches to the measurement of Federal employment support ...............10
III. Extramural Employment of Scientists and Engineers in Federal
Programs: Budget Approach ...............................................................................11
Estimating extramural employment in Federal programs .............................11
Estimates of extramural employment in Federal programs............... ...........12
Evaluation of the budget approach
..................................................... 16
IV. Extramural Employment of Scientists and Engineers in Federal
Programs: Interindustry Transactions Approach ............................................... 17
Elements of the interindustry transactions approach ..................................... 17
Extramural employment of scientists and technicians ...................................20
Evaluation of the interindustry transactions approach...................................22
V. Scientists and Engineers in the Federal Government ....... ..................................... 25
Recent trends in Federal scientist and engineer employment .......................25
Intramural occupational employment patterns ............................................. 26
Government scientists and engineers in Federal programs.............................26
Data gaps in Federal Government employment statistics .............................27
Summary and evaluation .................................................................................28
V I Total Federal Effect and Implications for the Scientific and
Technical Manpower Information System .......................................................29
The total effect of Federal programs on employment ................................... 29
Applying the estimates....................................................................................... 31
Evaluation and Recommendations ...................................................................36
Table:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

Sources of information on the employment of scientists and engineers
in the United States, by selected characteristics ........................... .................... 9
Estimated extramural employment of scientists and engineers employed
in Federal programs, selected agencies, 1966 ........................................................13
Estimated extramural employment of R&D engineer-scientists on
Federal programs by selected agency and performer, 1966 .................................14
Percent distribution, extramural R&D employment of scientists and engineers
on Federal programs, by selected agencies and performer, 1966 .......................14
Implicit performer cost-ratios, extramural R&D scientists and engineers
in Federal programs, selected agencies, 1966 ...... ................................................ 15
Total and federally supported extramural employment of scientists
and engineers, selected performers, 1966 ............................................................. 15
Estimated employment of scientists and engineers supported by
Federal funds, by type of employer, 1965 .......................................................... 20
Federally supported employment in selected scientific and technical occupations
as a percent of total occupational employment, private industry, 1965 ............. 21
Percent distribution of scientists and engineers in private industry,
selected occupations, by source of support, 1965
..........................................22
Industrial distribution of estimated total employment,
by expenditure source, 1965 ............................................. .................................. 22
Industrial distribution of estimated engineering employment,
by source of support, 1965 .................................................................................... 23
Industrial distribution of estimated scientist employment,
by source of support, 1965 .................................................................................... 23




v

Contents— Continued
Table— Continued:

Page

Total civilian employment and scientists and engineers, Federal
executive branch 1951-65 ...............
...... ......................... .............................. 25
14. Total and Federal intramural scientists and engineers,
selected years, 1951-65 ................................................ ......................................... 26
15. Total and intramural employment of scientists and engineers,
selected occupations, 1965 ..... ........ ......... ......... ..................................... ............ 27
16. Percent distribution of agency white-collar employment in the Federal
....... .......... .....28
Government, by program, selected agencies October, 1967
17. Civilian scientists and engineers in the Federal Government,
by program, October 1966 ................................... ............................... ................ 28
18. Estimated Federal utilization of scientists and engineers,
selected programs, 1966 ....... .............. .......... ....................................................... 30
19. Estimated employment of scientists and engineers, by source
of support and sector, 1965
.......................................................................... 30
20. Federally supported employment as a percent of total employment
of scientists and engineers, 1965 .... ........ ............ .............................................. 31
21. Estimated employment of R&D scientists and engineers, by sector
and source of support, 1965 ..............................................................................31
22. Estimated utilization of federally supported scientists and engineers,
by field of science and sector,1 1965 ..................
...... .......................... 32
23. Estimated total and federally supported employment of scientists
and engineers, by sector and character of work, selected years
.......................32
24. Supply and demand for scientists and engineers, 1958-66 _______ ____ ______ 33
25. Estimated employment of scientists and engineers per billion dollars
of expenditures in defense programs, private industry,
and Federal Government, 1965 or 1966 ....................... ......................... ............ 35
26. Estimated percent of scientists and engineers on Federal work,
by selected sources of estimates .............................................................................36
Appendix:
A. A note on program planning and manpower requirements ______________ _____41
B. Estimated total employment and employment attributable to Federal Purchases
of goods and services, by source of procurement and industry, 1965
.............42
C. Estimated federally supported extramural employment of scientists and engineers
by State and local governments, selected occupations, 1965 _______ __________44
D. Discrepancies between time series employment estimates
and occupational matrix estimates ............... .......... ..... ..... ...... ...... .............45
E. Estimated scientists and engineers employed in the Federal Government,
selected occupations, by program, October, 1966 ...............................................46




13.

vi

F E D E R A L SP E N D IN G AND S C IE N T IS T AND E N G IN E E R EM P LO Y M E N T
Chapter I.

Introduction and Highlights

Relatively large amounts of information on scientific
and technical manpower in the United States have
been collected for nearly two decades. In terms of the
numbers employed, the industries and institutions in­
volved in their employment, the sources of supply, the
nature and character of their work, their education,
training and mobility, perhaps no other group of occu­
pations has been subjected to as much continuous and
detailed observation. Curiously, however, the apparent
richness of the sources of data has not been accom­
panied by a substantial body of research directed to
increased understanding of the employment and utili­
zation of scientific and technical manpower.
The influence of the Federal Government on the
employment of scientific and technical manpower de­
serves more than the casual speculation it has thus
far received, because of the degree in which the Gov­
ernment uses scientific and technical knowledge in its
programs; and has generated substantial social and
economic interest because of its policy implications.
The kinds of scientific and engineering talents en­
gaged in Government-supported programs may have
a major, if not preponderant, influence on technolog­
ical development and innovation. The technolog­
ical choices made by scientists and engineers may have
intended costs as well as benefits that reach well
beyond their immediate frames of reference. From the
economic viewpoint, awareness is also growing that
scientific and technical manpower devoted to one set
of uses, such as Federal programs of research and de­
velopment, implies the sacrifice of other socially useful
opportunities for their employment. In the absence of
a market mechanism for rationing this scarce man­
power resource, as is typically the case in public pro­
grams, other means of evaluating its deployment wall
need to be developed.
The study objectives focus on a single question:
How useful are the available sources of data on scien­
tific and technical manpower for measuring the effects
of Federal programs on employment of scientists and
engineers?
Some data sources already include source of support
as an explicit dimension, but the occupational, char­
acter of work, and institutional coverage of these




sources is incomplete or inadequate from an analytic
viewpoint. Consequently, the principal effort of the
study became directed principally towards deriving
approximate measures of Federal influence on the
employment of scientists and engineers, especially in
respect to those dimensions not presently covered by
survey design.
A major conclusion of this study is that existing
sources of data on Federal expenditures and on scien­
tific and technical manpower can provide useful results
in measuring the program—-employment relationship.
The results obtained, however, represent only broad
and approximate relationships. For manpower plan­
ning in connection with Federal programs data that
are more detailed, more precise, and more compatible
with a variety of analytic measures of activity, will
be required.
For the most part neither the sources of data nor the
methods used in this study to test their measurement
capability are novel. Expenditures per man-year and
input-output analysis have been used as complemen­
tary means of estimating employment attributable to
Federal spending. Since the latter approach converts
industrial and occupational coefficients of employment
into estimates of numbers employed, only the expen­
diture data used in this approach were unavailable
through official publications. All other estimates were
derived from published sources.
The resulting estimates of federally supported em­
ployment relate either to 1965 or 1966. At the time at
which the study was made it was not possible to obtain
employment and expenditure data which would per­
mit measurement of the government’s influence to a
comparable extent for later years. In addition to the
lapse of time between survey dates and the availability
of the data, certain essential employment and financial
data are collected independently on different time
schedules. Comparison of two different approaches to
measurement, based on independent sources of data,
further restricted the timeliness of the employment
estimates. Recommendations which could lead to more
timely estimates are made in the concluding chapter
of this report.
Conceptual problems in measuring the Govern-

1

merit’s effect on the employment of scientists and
engineers and the sources and nature of available man­
power and expenditure data are discussed in chapter
II. Estimates of federally supported employment in
the extramural sectors are developed and evaluated
in chapters III and IV. Direct employment in the
Federal Government itself is the subject of chapter V.
Chapter VI brings together the results of the three
preceding chapters, illustrates some applications of the
resulting estimates of employment, and offers recom­
mendations for changes in the concepts and proce­
dures for collecting data on the employment and
utilization of scientific and technical manpower.
Highlights

Exploration of the capability of data to estimate the
effects of Federal programs on the employment of
scientists and engineers yielded the following results
and general conclusions:
About 393,000 scientists and engineers, or nearly 30
percent of the estimated total employed in 1965,1
derived their employment from Federal funds. Only a
third of this number were employed directly by the
Federal Government.
The proportion of federally supported employment
of scientists and engineers varied markedly. The rela­
tive numbers were estimated to be less than 20 percent
in private industry and in State and local government,
but were as much as 40 percent of scientist-engineer
employment respectively, in universities and colleges
and independent nonprofit organizations.
Federally supported employment of scientists and
engineers is heavily concentrated in all sectors of
research and development (R&D). Overall the Fed­
eral Government was estimated to employ about 55
percent of R&D scientists and engineers in 1965; only
about 5 percent of non-R&D scientists and engineers
were employed in Federal programs.
Defense and space programs in 1966 were estimated

2



to employ nearly 80 percent of scientists and engineers
in the extramural sectors. More than half of Govern­
ment civilian scientists and engineers were also em­
ployed in those programs.
Federal employment of engineers in 1965—66 ex­
ceeded the employment of scientists. However, in
terms of total scientist and engineer employment, a
greater proportion of scientists than engineers were
employed in Federal programs. Federal utilization was
greatest in the life sciences.
Measures of federally supported employment devel­
oped in this study indicated that in the late 1950’s
and early 1960’s annual increases in Federal program
requirements equaled about 25 percent of the average
annual number of bachelor’s and first professional
degrees conferred in the sciences and engineering.
The foregoing estimates of federally supported em­
ployment of scientists and engineers and some illustra­
tive applications indicate that currently available data
sources have a useful but limited capability for mea­
suring the employment effects of Federal programs.
However, the estimates of employment were possible
only on the basis of rather broad assumptions about
the relationship of employment and Federal expen­
ditures.
Data on output and expenditure in Federal pro­
grams were found to be weaker than employemnt data
in most of the dimensions investigated.
The general strategy for developing data adequate
for scientific and technical manpower planning in
Federal programs should be improvement of existing
data sources rather than development of a single
integrated data system. Greater coordination and
compatibility among the sources of employment and
expenditure data, buttressed by periodic benchmark
surveys of Federal involvement, should satisfy most
needs for m anpow er p la n n in g in G overn m en t
programs.
1 Latest data available at time of this study.

Chapter II.

Government Programs and the Measurement of Manpower Requirements

The conceptual basis for measuring Federal re­
quirements for scientific and technical manpower is
essentially similar to the general basis of manpower
measurement. The underlying concept of all such
efforts is that a functional relationship exists between
the output of goods and services and the numbers of
workers employed. Where public goods and services are
involved, technical and institutional factors may require
some modification in the design of the information
system used to capture and analyze the relationship
between output and manpower. Elaboration of the
foregoing statement as background to the efforts at
measurement attempted in this study is the purpose
of this chapter.
T h e concept of Federal manpower requirements

Governments use manpower for the same general
purpose as other organizations and economic sectors
— to produce goods and services. Just as in the general
case, the Government sector has a production function
relating the quantity of goods and services produced
and the quantities of resources used, including the
numbers of scientists and engineers. The production
function for each uniquely identifiable output of goods
or services, that is, a particular Government program,
will differ from the production functions for other
Government programs. Such functional relationships
will also differ from those for other economic outputs,
such as those purchased in the markets for private
goods and services. Further, it is reasonable to assume
that manpower requirements to serve Government
programs are subject to dynamic influences more or
less common to those occurring in other economic
sectors. Technological change, changes in the supply
and prices of cooperating factors of production, and
changes in the composition of output resulting from
changes in purchasers’ preferences affect the supply
of public as well as private goods.
The application of the concept of a production
function to public programs, however, raises some
special difficulties which affect the measurement of
manpower requirements. First and foremost is the
concept of output as applied to public goods and ser­
vices. In a number of areas of Government activity,
especially in the domain of the Federal Government,




the output of public programs is objectively unmeasur­
able. The most obvious instance is the national defense
and security. No apolitical means has been found for
determining whether security from attack, or offensive
capability, is increased or decreased by a particular
level or mix of armaments and military manpower.2
Similar difficulties of concept arise in other areas,
although perhaps not to be same degree, such as in
space exploration, the development of nuclear energy,
and in the more traditional functions of Government
such as education or the maintenance of internal law
and order. In some other areas, such as transportation,
the delivery of mail, and certain health services con­
sumed directly, the concept of Government supply is
both more easliy recognizable and quantifiable.
The difficulty posed in the foregoing is especially
decisive in projecting the levels and changes in the
levels of output and employment associated with
public goods and services. Although far from com­
pletely adequate, the private sector experience tends
to support the theory of a functional relationship
between changes in expenditure, output, and employ­
ment. This relationship can be roughly determined
from economic models relating the growth and com­
position of the labor force, productivity, rates of con­
sumption and investment, and so on, to changes in
gross national product and, especially, to changes in
the demand for goods and services in particular indus­
tries. Such models often provide the means for deter­
mining manpower requirements, although they are
usually implicit in manpower forecasts.3 A functional
model of the demand for Government output, on the
other hand, has yet to be devised, largely because the
factors that determine such a demand are mediated
through a political mechanism. Except in a few cases
involving Government enterprises, no market can be
shown for Government output, and consequently no
determinate relationship exists between the “ price”
2 The evaluation of the “ quantity” of security depends on
factors such as the military posture of other countries, the
political situation, the availability of technical means, etc., as
perceived by military and diplomatic intelligence and as
interpreted by the executive and legislative branches of
Government.
3 For example see T h e L o n g - R a n g e D e m a n d f o r S c ie n tific
and
T e c h n i c a l P e r s o n n e l , National Science Foundation,
(NSF, 6 1 -6 5, 1961).

3

of such output, expenditure, and the quantities de­
manded. The absence of output measures also makes
difficult forecasting whether or not changes in expendi­
ture on Federal programs will result in corresponding
changes in employment. An increase in expenditure in
a program area may represent an increase in program
activity or output. Unlike goods and services produced
in the private sector, however, practical means of dis­
tinguishing increases in real output from rises in unit
resource costs are rare in Government programs.
Two methods of dealing with this issue can in
principle be devised. One is to develop quantifiable
measures of program achievement, such as those con­
templated as part of the Planning-ProgrammingBudgeting System (PPBS)4 or in the recent National
Planning Association study of the costs of social goals.5
The other method is to deflate program expenditures
of real resources used, such as manpower. The methods
used in this study out of practical necessity are more
closely related to the second possibility. As will be seen
later, this study is based on the assumption that
program expenditure equals program output; this
measure has been related to employment by means of
a common dollar denominator.
In approaching the measurement of manpower
requirements for Federal programs as a practical prob­
lem, account must also be taken of the institutional
and organizational complexities. The principal institu­
tional fact is that the Federal Government operates
in a so-called “ pluralistic economy.” 6 The Government
engages in direct production of its output only to a
limited extent, partly as a matter of tradition and
partly as a matter of economic and political strategy.
A substantial, if not a major part of its total output is
achieved through employing other sectors, from which
the Federal Government purchases goods and services,
or to which it makes grants-in-aid, loans, or subsidies.
Through these various financial arrangements, other
economic sectors may be co-opted into Federal pro­
grams. Some or all of these other sectors also employ
each other, through purchase or subcontract, in fur­
thering their own participation in Federal programs.
Thus, an important objective in measuring scientific
4 For a clear account and analysis of the development and
implications of PPBS, see Charles L. Schultze, The Politics
and Economics of Public Spending, (Washington, D.C.,
Brookings Institution, 1968).
5 Goals analysis is a relatively new field in which measures
of output are largely determined from normative standards.
See Leonard A. Techt, Goals, Priorities and Dollars: The next
Decade (Free Press, New York, 1966) ; and Manpower Re­
quirements for National Objectives in the 1970’s (Center for
Priority Analysis, National Planning Association, Washington,
D .C ., 1968), especially pp. 5 -7 .
6 For an interesting definition of this term and discussion
of its significance see Eli Ginsberg, Dale L. Hiestand, and
Beatrice G. Reubens, The Pluralistic Economy (New York,
McGraw-Hill, 1965).

4



and technical manpower requirements in Federal pro­
grams is to determine the sectoral distribution of these
requirements.
The organizational structure and processes of Fed­
eral decisionmaking further complicate the treatment
of the Government’s manpower requirements as func­
tional entities. The basic organizational structure for
both appropriation and administration is departmen­
tal, but this arrangement does not necessarily coincide
with distinctive functions and program goals. Compe­
tition among agencies for funds for similar, even
identical purposes makes it difficult to determine
the functional distribution of activity. The PlanningProgramming-Budgeting System, gives promise of
integrated program planning in the Federal Govern­
ment, but for the near term future at least the assump­
tion that Government programs represent a set of
possibly overlapping production functions is often
closer to reality.
T he elements of a model

An information system for measuring manpower
requirements in Federal programs should have two
broad and related attributes in its design. First, it
should supply the information necessary to its intended
uses. Second, it should result in data that more or
less accurately mirror the structure, processes and
behavior of the organizational units covered. Although
these two attributes jointly influence the design of the
system, their respective influences are probably some­
what different. Purposes will have greater influence on
the coverage of the system and the kinds of informa­
tion collected, while the structural-functional charac­
ter of the organizational units involved will tend to
govern the concepts, classification schemes, and level
of detail. The implications of the first of these attri­
butes can be dealt with briefly, since they rest on the
assumption of a production function and past experi­
ence with manpower information systems generally.
The second attribute, on the other hand, deserves more
extended consideration, since it involves novel and
possibly more troublesome elements. Most of the dis­
cussion in this section consequently will be devoted
to the structural and functional elements of Federal
programs.
A manpower information system for Federal pro­
grams, focused either on scientists and engineers or on
a broader range of occupations, could serve a variety
of purposes. Such a system could: (1) Permit compar­
ison among Federal programs in terms of the human
resource and related physical requirements as a basis
for funding; (2) permit evaluation in like terms of
either new program activities or changes in existing
program structures and support levels; (3) provide a
basis for measurement and evaluation of the adequacy
of current or future manpower resources to support

current or projected Federal programs; (4) provide
a basis for evaluating the influence of Federal man­
power utilization on the deployment and compensa­
tion of all workers; or (5) permit tracing the flows of
manpower resources used in Federal programs over
time and among industries or large regions.
Not all of these purposes need be simultaneously
satisfied, but they share certain common informational
requirements. Information would be needed on pro­
gram outputs, that is, the specific nature and quantity
of the work in terms of program objectives, to the
degree to which these can be quantified, and the cur­
rent or expected total expenditures associated with the
program activity. Insofar as scientists and engineers
are concerned, uses (3 ), (4), and (5) would also
require that the system be consistent— though not
necessarily integrated— with a system of manpower
information covering the non-Federal econom ic
sectors.
The structural and functional attributes of a Fed­
eral program oriented system of manpower informa­
tion may be more troublesome for several reasons.
Some of the difficulties perceived here seem to derive
from the “ information complexities,” 7 inherent in the
labor markets for scientists and engineers, but others
may be unique to the nature and functions of the
Federal Government and to the variety of ways in
which it gets its work done. Federal programs vary
greatly not only with regard to their respective objec­
tives, but also— and more importantly with respect to
manpower requirements— in terms of the means by
which those objectives are served. A variety of orga­
nizational patterns in carrying on Federal programs
must be accommodated in an information system.
Finally, programs may vary with respect to their dura­
tion, from those with specific timebound objectives to
those that are continuous and indefinite. Brief discus­
sion of each of these aspects as a conceptual problem
will help to appreciate the associated informational
requirements. For the most part they provide the
framework for a system of Federal manpower planning.
Federal spending covers a very broad and diverse
spectrum of activities intended to serve purposes that
have been jointly defined, approved and funded by
the President and the Congress. The budget docu­
ments, for example the Budget for Fiscal 1969, classify
outlays and expenditures on these activities in two
principal ways: By administrative agency and by func­
tion. For either fiscal planning or analysis of past
utilization of manpower resources, however, neither
classification is ideal. Manpower requirements should
be related to the end products to which they contribute,
rather than to the channels of employment and finan­
cial flows. Since with few notable exceptions, Govern­
ment programs involve the services of more than a
single agency or bureau and a number of agencies are
involved in more than one program, the administrative




classification of either fiscal or manpower inputs would
not be revealing of the input-output relationship. The
functional classification scheme employed now for sev­
eral years, exclusive of expenditures for general govern­
ment and interest payments, is made up to 10 principal
categories. Although it is a step in the desired
direction, the classification still appears to be ex post
the budgetary decisions rather than an operational
basis for allocating financial or other resources.8 Com­
parison of the expenditure totals for education given in
the main budget document and in the special analyses
reveals the ambiguity of the present scheme. For fiscal
1967, the total Federal outlay on education, including
net lending, given in the functional classification of
the budget was about $4.0 billion, but Federal expen­
diture on education in all programs in that period was
$9.2 billion. The difference between these two con­
cepts presumably is that the first relates to expenditure
on education as a final output, while the second also
includes educational expenditures as inputs in support
of other objectives— defense, for example.9 Conceiv­
ably, measurement of manpower requirements could
be desired for either type of functional classification,
and the information system should accordingly be
flexible enough to accommodate both types.
The manpower information system also must com­
prehend structural differences among programs in the
ways in which their objectives are met. In addition to
civilian and military employees of the Federal Govern­
ment, other program performers include industrial
and nonprofit contractors, State and local governments
both as contractors and recipients of loans and grants,
and various other organizations and individuals who
receive payments for goods or services. Each of these
performer groups will generate varying employment
7 The term is that of Allen O. Gamble whose discussion of
the scientist-engineering manpower information system re­
mains the leading work in this too sparse field. See his
“ Proposal for Development of An Improved ManpowerRelated Information Program,” Toward. Better Utilization of
Scientific and Engineering Talent (Washington, D.C. Na­
tional Academy of Sciences, Committee of Utilization of
Scientific and Engineering Manpower, 1964), pp. 1 06 -1 0 7.
8 See Joint Economic Committee, Subcommittee on Eco­
nomic Statistics, The Federal Budget and/or Economic Docu­
ment (Washington, D.C., 88th Cong., 1st Sess., Report No.
396, 1963), p. 9. Although some progress has been made
toward a genuine program approach, the successes have been
sporadic and mainly at the intra-agency level. See Joint
Economic Committee, Subcommittee on Economy in Govern­
ment, The Planning Programming-Budgeting System: Prog­
ress and Potentials (Washington, D .C., 90th Cong., 1st Sess.,
1967).
9 Whether this distinction is itself valid is questionable. In
The Budget for Fiscal Year 1969, aid for undergraduate and
graduate students in 1967 is shown as $421 million, but in
Special Analysis H , the sum of undergraduate and graduate
and professional student support is $925 million for that year.
No clue in either of these parts of the Budget will explain
this difference.

5

requirements for given levels of Federal expenditure,
depending on the nature of the activity involved. This
point may be simply illustrated by comparing the per­
centages of employment in the private sector directly
attributable to Federal purchases of good and services,
including the compensation of Government employees,
in different programs, during the period 1962—66:

1962
1963
1964
1965
1966

All
programs

(percent)
Defense—
atomic energy

44.6
46.4
45.1
43.2
42.7

42.6
41.3
41.3
40.3
42.3

........................
........
....................
______

Nondefense
and space
50.0
61.9
57.9
52.6
47.1

S o u r c e : Manpower Report of the President— April 1968,
appendix table G-4, p. 322.

Finally, the manpower information system must be
able to discriminate between programs, or program
elements, that are terminal and those that are contin­
uous in duration. On the basis of recent experience,
apparently the requirements for scientists and engi­
neers vary with the phases of the program as well as
with the levels of expenditure involved. Some frag­
mentary evidence of considerable variation may be
found in studies of the development of the Titan II
intercontinental ballistic missile and of a computerized
Navy command and control system. In the latter case,
for example, the engineering work force engaged on
the project varied over a 6-year period from 8 to 16
percent of the total engineering force at the contract
facility. Unfortunately, no data are given to indicate
the changes in the total engineering complement, but
the engineers employed on the computer contract in­
creased more than sevenfold over the entire contract
period.10
On the other hand, some activities of the Federal
Government are open ended fiscal commitments, but
with manpower coefficients that tend to be compara­
tively stable over time. Few such programs, however,
currently are large users of scientific and technical
manpower.
From this background it is now possible to indicate
the components and characteristics of a manpower
information system for Federal programs, focusing on
scientific and technical occupations. This “ model” is
appropriately discussed under the following headings:
10 Paul W . Cherington, “Case Studies of Titan II and
N TD S,” Toward Better Utilization of Scientific and Engi­
neering Talent: A Program for Action (Washington, D.C.,
National Academy of Sciences, Committee on Utilization of
Scientific and Engineering Manpower, 1964), pp. 127-128.
11 Planning-Programming-Budgeting: Bureau Guidelines of
1968 (U.S. Senate, Committee on Government Operations,
1968).
12 Ibid.

6



(1) Program concepts, (2) program performers, (3)
manpower and financial data, and (4) responsibility
for data collection and frequency of reporting.
P r o g r a m c o n c e p t s . In general, the manpower infor­
mation system should discriminate between groups of
activities each of which represents a common objective
or end result. The activities to be included in any
given program category, therefore, should be deter­
mined by the contribution of the activity to the pur­
pose of the program, and not by the similarity of
specific functional characteristics. Thus, for example,
employment on school construction designed to im­
prove the quality or adequacy of public education
should be classified as an element in an educational
program rather than as a public works activity, even
though it also might be undertaken to achieve the
purpose of a public works program.
The classification of any given activity (or closely
related set of activities), accordingly, cannot be estab­
lished by treating it as an isolated phenomenon.
Although each specific case of classification may pre­
sent special problems of definition, a general test of
where an activity belongs might be the response to the
question of complementarity: Will the objectives or
goals of the program be achieved in the absence of
this particular activity? An affirmative answer, of
course, implies either that the activity belongs in some
other program category or is per se questionable as a
Government program. Such a test does not preclude in
the course of time changes in program activities, either
because objectives are reformed in the light of chang­
ing conditions (including the effects of the programs)
or because new techniques of achieving program objec­
tives come to light. Indeed, one of the major objec­
tives of a program classification is to facilitate rational
analysis of such factors.
For analytical reasons, especially for assessing man­
power requirements and evaluating performance,
activities should be grouped into so-called “ program
elements.” Bureau of the Budget guidelines for the
Planning-Programming-Budgeting System11 define a
program element in terms of its characteristics— a
“ clearly definable output,” an end product for the
operating agency, rather than an intermediate prod­
uct; and output varying, though not proportionally,
with changes in inputs.12
P r o g r a m p e r f o r m e r s . Public programs in the Federal
Government are served in a variety of ways. A man­
power information system should reflect and cover the
different organizational forms involved, not only be­
cause of differences in the nature of their relationship
to the Federal Government, but also because differ­
ences in cost, manpower utilization, and program effec­
tiveness may be involved. Thus, the system should
comprehend not only the leading Federal agencies

directly involved in program management and oper­
ations, but also all other major units supplying inputs
to the program: Businesses and industries, other Fed­
eral agencies, other governments, and nonprofit and
educational institutions. In addition, because each of
these organizations purchases inputs used to supply
Federal program output, measurement of the man­
power indirectly hired is also desirable.
D a t a r e q u i r e m e n t s . The assumption is that data col­
lected by a manpower information system will be used
for analytical purposes; projection of occupational
requirements for scientific and technical personnel may
be the most important purpose. For this and most
other conceivable uses, the minimum data require­
ments would include: 13

1. Total program employment broken down by pro­
gram element and by type of performer.
2. Scientists, engineers, and supporting technical
personnel employed by each program element and cor­
responding performer, broken down by occupational
specialty.
3. Information on (2) with respect to the functional or
character-of-work assignment of the individuals employed.
4. Total expenditure or costs for each program
element, distributed among personnel, equipment and
facilities, materials, and overhead expenses.
5. A measure of the real output of the program
element.

None of the above data requirements is free of diffi­
culties of definition. Among the more troublesome is
the question of occupational classification. Any such
scheme, no matter how fine grained, is inherently
arbitrary when measured against the adaptability of
trained men and women to a variety of tasks. Even
training in those scientific and technical fields which
require long periods of formal education in particular
disciplines and technologies has not been particularly
restrictive when new tasks and requirements have
emerged. The design of an occupational classification
scheme that will fit the great variety of tasks per­
formed by scientific and technical manpower clearly
should have a high priority.
A d m in is tr a tiv e

r e s p o n s ib ility

and

freq u en cy

of

re­

Although congruence between program and
organizational structures in the Federal Government is
not necessary, for most of the existing program cate­
gories one agency is dominant, and administrative and
operating responsibility probably is more concentrated
at the program element level. Administration of a
scientific and technical manpower information system
similarly could be decentralized by following the domi­
nant agency principle; the appropriate bureaus or
divisions would be responsible for data collection and
reporting within a common informational framework.
Tabulation and analysis beyond the level of agency or

p o r tin g .




particular program need, however, could be left to an
organization with overall responsibility and interest in
scientific and technical manpower resources.
Frequency of reporting depends on the use of the
data, and may be affected by the nature of the pro­
gram. Programs with substantial expenditure and
manpower weights in the scientific and technical areas
may require more frequent monitoring than those pro­
grams that use relatively little of such input. Likewise,
programs with major short term fluctuations in re­
quirements suggest the need for a greater frequency
of reporting than programs that change more slowly
and predictably. A mixed system of data reporting
seems indicated. Comprehensive bench mark data
could be collected on all programs at common report­
ing dates, and more frequently programs of selected
interest and, probably, of major importance.
Data sources: expenditures and manpower

As an empirical matter, whether or not the effect of
Federal programs on the employment of scientific and
technical manpower can be measured depends on the
nature of the data available. In the course of this study
all of the major data sources were reviewed and their
characteristics examined in terms of the needs outlined
in the preceding section. None of these sources either
jointly or as independent data systems, of course, com­
pletely satisfies the needs of an ideal system with the
capability of measuring the manpower requirements of
Federal programs. Some of the sources do address the
issue of Federal influence directly, but others, as will
be shown subsequently in chapters III, IV, and V may
provide more meaningful results.
Discussion of data sources may conveniently be
organized in the following categories: (1) Measures
of program activity and output, (2) expenditure and
costs, and (3) employment. Particular sources of data
in some cases overlap these categories, so that repe­
tition is unavoidable. Most of the sources are available
in published form, although not all data collected in
the established information programs are published.
M e a s u r e s o f p r o g r a m a c t i v i t y a n d o u t p u t . M ea'ures
of program activity and output in the Feder**1 Govern­
ment are still in the early stage
development.
H ow ever, the stimulus to this developm ent given
by President Johnson’s August 1965 directive to install
the Planning-Progra^iming-Budgeting System on a
Government-wide basis has resulted in some progress.
S p e c i a l A n a l y s i s R , in t h e B u d g e t , F is c a l Y e a r 1 9 7 0 14

f o r the first time since the President’s directive provides

a review and analysis of the programs of 17 Federal
13
These proposed data requirements lean heavily on the
discussion by Gamble, op. cit.
1 4 S p e c ia l
Year 1 9 7 0

A n a lyses,

B u d get

of

th e

U n ited

S ta tes,

F isca l

(Washington, D .C., 1969) pp. 252-273.

7

agencies, covering about 95 percent of the proposed
budget, in detailed terms.15
The data presented are in financial terms, although
in some cases there is a statistical as well as a narrative
account of the activities to be undertaken. In other
parts of the 1970 budget, and also in the budgets for
earlier years, program activities are presented for a
smaller number of selected programs and agencies in
terms of both outlays and services provided.16
Comprehensive measures of program output never­
theless are still poorly developed, and indeed for a
number of federally supported programs the prospect
of such measures seems remote. For Government as a
whole, output in the national income accounts is
expressed in value terms as roughly equal to the sum
of the cost of employee services and the value of
Government purchases of goods and services.17 Because
of the aggregation problem, progress toward r e a l mea­
sures of total Government output is unlikely in the
near future. In some particular agencies, however,
efforts to measure physical output as a management
tool for evaluating agency performance have begun.
In addition to a limited experiment in selected agen­
cies,18 all of which confined the measure of input in
the productivity equation to in-house employment, a
few agencies regularly collect and tabulate data on
their programs, which includes extramural employ­
ment as well as corresponding measures of output. The
Atomic Energy Commission, for example, uses scien­
tific publications as a partial measure of output in its
Physical Research Program; the scientific man-years
involved as well as the associated financial outlays are
reported.19 The Atomic Energy Commission, on a
broader basis, also projects manpower requirements in
the atomic energy field, mainly on the basis of expendi­
ture plans and projections. Few other agencies as yet
attempt such measures of program output, although
progress is being made in the development of program
memoranda for PPBS presentations. In research and
development, where Federal support and utilization
of scientific and technical manpower is heavily con­
centrated, general measures of output may be practi15 Ibid, p. 254.
• P- 157. For example, hospital and medical services,
for which tn^ number of patients treated and the number of
clinic and physics n visits are indicated along with dollar
outlays.
17 See U.S. Department o£ Commerce, Office of Business
Economics, National Income, 19a-t edition (1954).
18 Measuring Productivity of Federal Government Organi­
zations (Office of the President, Bureau of the Budget, 1964) ;
also, William A. Vogeley, A Case Study in the Measurement
of Government Output (Rand Corp., 1958).
19 A Statistical Summary of the Physical Research Program
(U.S. Atomic Energy Commission, June 30, 1968).
20 Nestor E. Terleckyj, Research and Development: Its
Growth and Composition (New York, National Industrial
Conference Board, 1963), pp. 11-12.

8



cally unattainable because of the problematic outcome
of R&D effort.20
E x p e n d i t u r e a n d c o s t d a ta . In the absence of ade­
quate real measures of program output, data on
expenditure must serve as a proxy in measuring man­
power requirements. Indeed, this may be the only
general basis for estimating employment attributable
to Federal support, and extensive use has been made of
financial data in this study. Such a basis, however, is
beset by numerous risks in interpreting results, because
of cost differences in the character of work, fields of
science, and the practices of performing organizations.
The greater the detail available, the more likely the
manpower estimates will approximate reality.
Two general sources and several more specialized
sources of financial data can be used in equations for
estimating federally supported employment of scien­
tists and engineers. The most general, of course, is the
B u d g e t o f t h e U n i t e d S t a t e s . This document, which
records both outlays and proposed expenditures, covers
the total of Federal spending on a fiscal year basis.
Expenditures are tabulated both by agency and by
broad functions and program. A derivative and com ­
panion source to the Federal budget is the annual
survey, F e d e r a l F u n d s f o r R e s e a r c h , D e v e l o p m e n t a n d
O t h e r S c i e n t i f i c A c t i v i t i e s , sponsored and conducted
by the National Science Foundation. F e d e r a l F u n d s
provides expenditure data on Federal support of
research and development, R&D plant, the collection
of scientific data, and the communication or storage
of scientific information. R&D support data are broken
down by character-of-work, fields of science, per­
former, and by sponsoring or administrative agency.
In addition, for major features of Federal support each
volume of F e d e r a l F u n d s contains historical series
from 1952 onwards.
Although presently of limited use in estimating
federally supported manpower requirements, data on
Federal support to higher education is available in the
C A S E 21 series, which provides information on an
institutional basis in support of academic science and
nonscience activities. An expanded CASE program is
expected to cover additional sponsoring agencies and
to provide information on a project basis, including
the utilization of manpower.
Expenditure and some cost data are available from
National Science Foundation surveys of scientific and
technological activity in four sectors— industry, univer­
sities and colleges, State Governments, and nonprofit
institutions. The industry survey has been conducted
annually since 1957, but surveys in the other three
sectors have been irregular or less frequent. Tw o of
the surveys— the surveys of the colleges and universi­
ties and of the nonprofit institutions— also provide
considerable information on federally Funded Re-

search and Development Centers administered by the
institutions.
At to expenditure data useful for estimating feder­
ally supported employment of scientists and engineers,
all four surveys publish data on the sources of support
for research and development. Only two— the industry
and the nonprofit institution surveys— provide specific
estimates of man-year costs of scientists and engineers,
however. All four surveys provide distributions of
expenditure by field of science. Only one of the sur­
veys— the university and college survey— includes some
expenditure data on non-R&D activities in academic
science fields.
Employment data. Development of employment
data on scientific and technical manpower has been
extremely prolific, possibly exceeding that of any
other specialized occupational group. In the course
of this study more than 14 sources were investigated.22
Eight of the surveys include specific information on
federally supported employment. Some, as in the case
of the NSF sectoral surveys, report employment and
expenditure data together, while others cover only
employment of scientists and engineers either as a
major objective or as part of a broader employment
survey. A leading example of this type of survey is the
biennial survey of employment of scientists and engi­
neers in the Federal Government.23
The various surveys differ with respect to definitions
of scientists and engineers, and in the level of occupa­
tional, industrial and functional detail. Because of
differences in method of collection, certain critical
concepts, and survey dates, there are substantial diffi­
culties in linking these sources together in a compre­
hensive picture of employment of scientists and
engineers.
Since in most instances the scope and m ethod' of
data collection are explained in the original reports
on the results of the various surveys, elaborate descrip­
tion and explanation of their method are unnecessary.
Table 1 summarizes those substantive features of most
interest to this study, including whether or not Fed­
eral support of employment is explicit, the presence
or absence of occupational detail or, alternatively, the
field of science, whether or not the character of work
is indicated, and whether or not supporting detail on
expenditure and costs is included. In all cases the most
recent survey available was the basis for categorization.
The various sources of data shown in the table, both
those that provide information on employment and
those which provide expenditure or cost data, hardly
constitute an entity. Most of them have been designed
to meet various special needs without regard for uni­
formity in the scope and nature of the data. Moreover,
within any given survey, cross tabulation of one
dimension on another is not always possible, for exam­
ple, in the BLS Survey of Scientific and Technical




Table 1. Sources of information on the employment of
scientists and engineers in the United States, by selected
characteristics
Source and sector

Occupational
detail

Federal
support

Character
of work

X
X
X

X
X
X

X
X
X

X

X

X

X

X

Expenditure
or cost data

TOTAL ECONOMY
Census of the United States
National register of scientific
and technical personnel..
National engineers register.
Doctorate records f ile ...........

X

PRIVATE INDUSTRY
Survey of scientific and tech­
nical personnel in industry
Survey of industrial research
and development................
Economic information
system 1.................................
Survey of employment in
atomic energy......................
Shipments of defenseoriented industries
(M A -1 7 5 )...............................

X

X
X

X

X

X

HIGHER EDUCATION AND
NONPROFIT
ORGANIZATIONS
Survey of scientific activi­
ties of institutions of
higher education................
Survey of scientific activi­
ties of non-profit institu­
tions........................................

X

X

X

X

X

X

X

X

GOVERNMENT
Survey of State research
and development activities
Survey of scientific and tech­
nical personnel in the Fed­
eral Government................
Occupations of Federal
white-collar w orkers.........

X

<2)

X

( J)

1 Conducted by the U .S . Department of Defense and NASA; unpublished.
2 Civilians only.

Personnel, data on character of work are collected
only for all scientists and engineers rather than for
each of the occupational specialties. Different methods
of data collection, even in the coverage of virtually
similar populations, also add to the incompatibility of
the systems. Nevertheless, BLS has put together a time
series for the period, 1950—66, of science and engi­
neering employment covering engineering and the
principal natural and physical science occupations,
21 Committee on Academic Science and Engineering, Fed­
eral Council for Science and Technology. The reports are
sponsored by the Office of Science and Technology, and pre­
pared by the National Science Foundation.
22 O f the 14 referred to and listed in the table below, only
the U.S. Department of Defense and N ASA Economic Infor­
mation System is not available in published form. A number
of other scientific and technical manpower surveys are pub­
lished; for example, the surveys on medical manpower con­
ducted by the National Institutes of Health, but do not
receive as wide a distribution and attention as they deserve.
In 1964, it was reported that at least 29 Federal agencies
collect information on scientific and technical manpower, of
which 10 are “ major collectors.” Gamble, op. ct., p. 103.
23 The most recent published report in this series is “ Scien­
tific and Technical Personnel in the Federal Government,
1966,” Reviews of Data on Science Resources (NSF 68—16,
No. 14, April 1968).

9

and representing all economic sectors.24 In evaluating
the capability of the science-engineering manpower
information system for estimates of federally supported
employment, extensive use has been made of this
source both as control and, in some instances, as a
basis for estimate.
Approaches to measurement of Federal
employment support

In the next three chapters, the information sources
described above will be tested for their capability in
measuring the extent and nature of federally sup­
ported employment of scientists and engineers. Two

10



empirical approaches will be used in this test. One has
been designated as the budget approach because it
utilizes budget data on Federal expenditures as a
basis of estimate, in conjunction with published or
derived measures of expenditure per scientist or
engineer man-year. The other approach has been desig­
nated the interindustry transactions approach. This
approach uses input-output analysis and occupational
patterns by industry as the principal tools of estima­
tion. Both approaches, however, are similar in that
they relate employment and expenditure.
24

E m p lo ym en t

S ta te s , 1 9 5 0 - 6 6

dation, 1968).

o f S c ie n tis ts a n d E n g i n e e r s

in t h e

U n ited

(Washington, D .C., National Science Foun­

Chapter III.

Extram ural Employment of Scientists and Engineers in
Federal Programs: Budget Approach

In this and the next chapter efforts to develop
measures of extramural employment of scientists and
engineers in Federal programs are described. This
chapter describes the budget or unit-cost approach.
Chapter IV uses an interindustry transaction or inputoutput table approach. Both approaches are expedients
for estimating the Federal effect from sources of data
not specifically intended for that purpose. They sub­
stitute assumption and analysis for direct measurement
for the most part without a means of validating results.
A model manpower information system would obtain
data on the numbers employed, their occupations and
functions, their distribution among program categories,
and the associated costs, expenditures and other com ­
mon measures of program output, directly from extra­
mural suppliers of goods and services to Government
programs.
The budget approach simulates a model informa­
tion system by estimating employment values from
assumptions about expenditures per scientist and
engineer employed in Federal programs. Expenditure
in such programs is assumed to be proportionate, if
not equivalent, to program output. Assumed expendi­
ture or, in some instances, reported costs per scientist
or engineer are divided into program expenditure to
obtain estimates of federally supported employment.
All of the estimates discussed in this chapter were
produced in this way. The hope was early in the study,
however, to locate on-going information systems in
various Federal agencies, which would have provided
direct estimates of federally supported extramural em­
ployment. Recent developments toward such systems
are discussed in appendix A.
Estimating extramural employment
in Federal programs

Elements of the budget approach. Estimation of
federally supported employment of scientists and engi­
neers by the budget approach requires data on (1)
total expenditure and (2) expenditure or unit costs
per scientist or engineer. Each of these quantities
further should be divisible into categories representing
the activity or purposes of expenditure, the character
of the work performed, the occupational specialties
involved, and the various extramural performers or




sectors. Expenditure data, of course, must also identify
the sources of support and the appropriate Federal
budget classification. The general argument and the
assumptions underlying the employment estimates are
discussed in the following section. Subsequent sections
present and discuss the detailed estimates. All of the
estimates are in terms of 1966 data.
Estimates of expenditure. The first step in estimat­
ing federally supported employment of scientists and
engineers was to divide total expenditure into two
broad classes representing a Federal and a non-Federal
sector. In the Federal sector, however, only expendi­
tures in controllable program categories were included.
These expenditures— Federal purchases of goods and
services, grants in aid, and subsidies— correspond to
Federal Government ouput in the national income
and product accounts, and are designated here as
Federal program expenditure. Because they mainly
involve transfer payments to individuals, such as pay­
ments from the retirement and disability trust fund
accounts or interest paid on Federal debt instruments,
the remaining expenditures were included as part of
expenditure in the non-Federal sector. Utilization of
scientists and engineers as the result of such expendi­
tures, in other words, was assumed in general to be
the same as employment generated directly by other
non-Federal expenditures. The aggregate data used
are on a calendar year basis25 to make subsequent
estimates of employment congruent with control totals
of scientist and engineer employment.
Both the totals of Federal program expenditure and
of other expenditure were next divided into two cate­
gories representing the character of work. Federal
25 Calendar year data on Federal expenditures are from
Economic Report of the President (February 1968, table
B -6 2 ), p. 283.
26 The best source of information on Federal R&D spend­
ing is the National Science Foundation, F e d e r a l F u n d s fo r
R e s e a r c h , D e v e l o p m e n t a n d O t h e r S c ie n tific A c t i v i t i e s , an
annual series now in its 17th year of publication.
27 R&D spending funded by other sources initially was
obtained from the National Science Foundation, N a tio n a l
P a tte r n s o f R & D R e s o u r c e s (NSF 6 7 -7 , Washington, D .C.,
1967), table B -16, p. 23. Later, other sources for particular
sectors proved more useful in estimating employment, because
of the greater detail available.

11

spending on research and development26 subtracted
from Federal program expenditure provided estimates
of Federal non-R&D spending. Similarly, R&D spend­
ing funded by other sources— industry and business,
universities and colleges, and other nonprofit institu­
tions— was subtracted from the total of other expendi­
ture, as defined above, to estimate other non-R&D
spending.27
In summary, as the result of the foregoing opera­
tions, estimates of four classes of expenditure were
obtained: (1) Federal non-R&D spending; (2) other
non-R&D spending; (3) Federal R&D spending; and
(4) other R&D spending. The two Federal com ­
ponents include expenditure on scientists and engi­
neers directly employed by the Government as well
as those employed in the so-called extramural sectors
in Federal programs. As the following section indicates,
the difficulty this fact created for estimating federally
supported non-R&D employment in the extramural
sectors was solved by assuming a uniform utilization
ratio and estimating employment as a residual.
Expenditures per man-year and estimates of em­
ployment. Estimation of extramural federally sup­
ported employment depends on the availability of
scientists and engineer man-year expenditures appro­
priate to the particular dimension of the desired
employment estimate. No such man-year expenditures
were available for scientists and engineers working
in non-R&D functions. Consequently, the unit expen­
diture used to estimate federally supported non-R&D
employment was derived by dividing total non-R&D
expenditure— federally supported and supported from
other funds— by an estimate of the total non-R&D
scientists and engineers.28 Estimated employment in
non-R&D work was then based on a crude assumption
that the underlying expenditure ratios would be iden­
tical in both Federal programs and in activities sup­
ported by other funds.
28 Total non-R&D scientists and engineers was estimated
as a residual, subtracting R&D scientists and engineers from
total scientists and engineers, as given in National Science
Foundation, E m p l o y m e n t o f S c ie n tis ts a n d E n g i n e e r s in th e
U n i t e d S t a t e s , 1 9 5 0 - 6 6 , (NSF 6 8 -3 0, Washington, D.C.,
1968), tables A - l , A -2 , pp. 20-23. This source also provided
the basic control totals for all sectors as well as the division
of employment by character of work.
29 However, estimates based on other sources were made of
to ta l federally supported employment by particular field of
science. (See chapter 1, table 6.) The inclusion of expendi­
ture in fields of science and engineering other than those
considered in this study should not be troublesome. Such
expenditure can be regarded as the cost of a complementary
input, just as compensation of employees other than scientists
and engineers, supplies, etc. are implicitly treated in our unit
cost estimates. Adjustment of expenditures by field of science
becomes important only when employment in that dimension
is under attention.
30 Such a distribution is attempted in chapter V .

12



Estimates of federally supported non-R&D employ­
ment were obtained by applying the resulting unit
expenditure in non-R&D employment to the estimates
of both total and agency Federal non-R&D program
expenditure. Division of the latter by the former
produced estimates for each of the Federal agencies
of federally supported non-R&D employment. Since,
however, these estimates included employment in the
Federal Government as well as in other sectors, a
reduction was necessary to estimate extramural nonR&D employment. Because data on intramural scien­
tist and engineer employment were not available by
character of work on an agency basis for 1966, the
agency totals of intramural non-R&D employment
were also estimated. After deducting the BLS estimate
of non-R&D scientist and engineer employment in
the Federal Government for all agencies from the esti­
mate of total non-R&D employment, the remainder—
extramural non-R&D employment— was distributed
among the agencies proportionately to the estimated
distribution of total federally supported non-R&D
employment.
Estimates of federally supported extramural R&D
employment by sector or type of performer were made
by more direct, and somewhat more reassuring,
methods. For some sectors, expenditures per scientist
and engineer in some Federal programs specific to the
sector were available or could be derived with rela­
tively small effort. Extramural Federal expenditure
on R&D by each agency should have been adjusted at
this point to eliminate those fields— psychology and
the social and behavioral sciences— for which there
are no matching control totals in the NFS-BLS time
series on employment of scientists and engineers. Un­
fortunately, the expenditure data shown in Federal
Funds is cross classified by agency and extramural
performer and by agency and field-of-science only, but
no cross classification by performer and field exists.
Consequently, estimation of federally supported R&D
employment is based on spending in all fields of sci­
ence and engineering in the various extramural sec­
tors.29 The sources of data used and the adjustments
made are otherwise best discussed with the estimates
in the next section.
Estimates of extramural employment
in Federal programs

Estimates of federally supported extramural employ­
ment of both R&D and other scientists and engineers,
by sponsoring agency are shown in table 2. The esti­
mates for both groups were derived by the basic
method described in the preceding section, but the
methods used to estimate R&D employment differ
among the sectors and are further discussed in this
section. A distribution by program area, by bridging
agency data in some cases, could also have been

Table 2. Estimated extramural employment of scientists
and engineers employed in Federal programs, selected
agencies, 1966
[In thousands]

Department or
agency

All extramural
scientists and
engineers

R&D
scientists
and
engineers

Other
scientists
and
engineers

All agencies...............................

261.7

221.5

40.2

A griculture..............................................
Commerce................................................
Defense.....................................................
Health, Education and W e lfare.........
In te rio r......................................................
A E C ............................................................
N ASA.........................................................
N S F ............................................................
O ther..........................................................

5.2
0.9
128.3
22.7
1.6
19.3
69.4
5.2
9.1

1.5
0.3
102.5
18.5
1.0
18.8
69.4
5.0
4.3

3.7
0.6
25.8
4.2
0.6
0.5
—

0.2
4.8

NOTE: Sums of components may not equal total because of rounding error.

attempted at this point.30 Because most Federal pro­
grams are dominated, in financial terms at least, by
single agencies, an agency distribution was felt to give
results almost as meaningful on a broad level as an
approximated program distribution of employment.31
Estimates of the employment of R&D scientists and
engineers in most Federal programs rest on man-year
expenditure estimates specific to particular sectors.
The results are shown in tables 3 and 4. For the indus­
trial and business sector, costs per scientist and engi­
neer broken down by source of funds were obtained
for D O D and NASA and for all other Federal agencies
combined for 200 leading companies in selected indus­
tries.32 Estimates of federally supported R&D employ­
ment of scientists and engineers in industry were
accordingly built up by applying the appropriate man31 In fiscal year 1969, although as many as 14 different
departments and agencies contributed to activities in the fields
of commerce and transportation, and 12 were involved in
health, labor and welfare programs, less than a third of the
agencies were found in three of the functional areas or more.
Eight of the agencies were identified with only one broad
functional area, and three others included only two such areas
in their programs.
32 National Science Foundation, Research and Develop­
ment in Industry, 1966 (NSF 68—20) 1968, table 46, p. 60.
33 A two-step process was used to derive estimates of Fed­
eral R&D expenditures by industry. Percent distributions of
such expenditure for the Department of Defense. NASA and
all other agencies were first calculated from NSF, Research
and Development in Industry, table 45, p. 59. These distribu­
tions were then applied to industrial-performer R&D ex­
penditures for each agency, as shown in National Science
Foundation, Federal Funds for Research, Development and
Other Scientific Activities (Vol. X V I , 1967) table C-7, pp.
112-113.
34 The following National Science Foundation reports were
used: Scientific Activities in Colleges and Universities, 1964
(NSF 67—17, May 1968) and R&D Activities in State Gov­
ernment Agencies, (NSF 67-1 6, September 1967).
35 Op. cit.
36 This procedure was possible because the sources of R&D
funds in each performer category were available in the vol­
umes cited in footnote 34.
37 National Science Foundation, Scientific Activities of
Nonprofit Institutions (NSF 6 7 -1 7 ), p. 5, 1964.




year expenditure estimates to agency R&D expendi­
tures on industrial performers. Only the estimates for
all industries are shown in tables 3 and 4, however.33
The employment estimates for other extramural
performers, because of lack of detailed man-year
expenditure data, in each case rest on a cruder basis.
Considerable adjustment of data were required to
derive expenditure per man-year and estimate employ­
ment. In each nonindustrial performer sector, reported
expenditures from all sources on research and develop­
ment and full-time equivalent employment of R&D
scientists and engineers for the most recent year 34 were
first adjusted to conform to the BLS-NSF concepts, by
excluding expenditures and employment in the fields
of pyschology, the social sciences, and unclassified
areas. Further adjustments, in the case of the col­
leges and universities, independent nonprofit organiza­
tions, and their affiliated Federally Funded Research
and Development Centers, were made to convert total
employment into full-time equivalent units. These
adjusted totals of money and manpower were then
extrapolated, usually by assumption, to take account
of changes between the date of observation and 1966.
An adjusted ratio of employment to expenditure was
then calculated for each of these sectors.
Expenditures per scientist and engineer obtained
after these adjustments were subsequently applied to
the respective extramural R&D performer expenditures
for each agency, as reported in Federal Funds.35 The
estimated employment obtained by applying a com ­
mon sectoral cost-ratio to each agency’s expenditure,
however, gave unreasonable results for some per­
formers. Except for independent nonprofit organiza­
tions, the alternative procedure adopted for the other
nonindustrial R&D performers was, first, to estimate
the overall proportion of R&D employment in the per­
former category attributable to Federal funds.36 This
proportion was applied to the BLS-NSF estimate of
total R&D scientists and engineers in the sector in
1966. The estimated total R&D employment for the
sector was then distributed among the respective agen­
cies shown in table 3, in accordance with each agency’ s
relative share of R&D expenditure in the performer
category.
A common expenditure per man-year for all agen­
cies was used to estimate Federal employment of R&D
scientists in independent nonprofit organizations.
More than 90 percent of expenditure and manpower
in these organizations is devoted to research and devel­
opment and its administration.37 About 75 percent of
the work of these agencies is federally financed, 80
percent of the funds came from two agencies— the De­
partment of Defense and the Department of Health,
Education, and Welfare. It was consequently believed
appropriate to apply a uniform average expenditure
per scientist and engineer for each agency, after adjust­
ment for increases in both expenditure and employ13

ment since 1964, to Federal R&D funds in 1966.
Because data on expenditure and employment in
federally funded Research and Development Centers
were available for colleges and universities and in­
dependent nonprofit organizations, employment in
these wholly supported components needed to be
adjusted mainly for fields of science and time. The
results have been combined in each case with those
for their administratively affiliated category, however.
Because Federal Funds does not identify support of
R&D in State and local governments, an arbitrary
allocation of 50 percent of expenditure in the “ Other
performer” category was made to that sector.
Table 5 is derived from the results obtained with
regard to federally attributable employment of R&D
scientists and engineers. The implicit expenditure
ratios, of course, are the quotient of expenditure
divided by employment in each agency performer
category. As between performer groups, these implicit
ratios appear reasonable; for example, since much of
industrial employment of scientists and engineers is in
development or applied research,38 plausibly expendi­
tures per scientist or engineer in that sector may be
above average. In colleges and universities, on the
other hand, proportionately more effort is devoted to
basic research, which typically requires less plant and
equipment, supporting personnel, and so forth. In gen­
eral, the numbers of R&D scientists and engineers
employed in an agency performer category depend on
the character of the work and the fields of science
represented. However, except in a fragmentary way
data are not readily available on the latter basis.
Although the relative magnitudes shown in the pre­
ceding tables are probably correct for the various
agency and performer categories, the estimated num­
bers, in general, may understate the Federal involve­
ment. The estimates for the NASA are reasonably
close to and consistent with the agency’s own data.39
On the other hand, our estimate of industrial scientists
and engineers employed in Department of Defense
Table 3. E stim a te d e x tra m u ra l e m p lo y m e n t o f R&D
engineer-scientists on Federal programs by selected agen­
cy and performer, 1966

Department or
agency

All R&D
extra­
mural

Industry1

Colleges
and
universi­
tie s 1

Inde­
pendent
nonprofit1

State
govern­
ment

All agencies

221.450

161,630

47,400

10,720

1,700

Agriculture...........................
Commerce............................
Defense.................................
HEW........................................
In te rio r..................................
A EC........................................
NASA.....................................
N S F.........................................
O ther......................................

1,540
285
102,470
18,470
955
18,840
69,460
5,030
4,400

50
170
87,400
730
250
6,500
63,600
<2)
2,900

1,475
100
10,190
13,000
490
11,475
5,350
4,750
570

<*>
<*>
4,880
3,540
( 2)
865
420
270
700

<’ )
< 2)
<2)
1,200
180

1 Includes federally funded research and development centers.
2 Less than 5 0; included in total.

14



—

90
—

230

Table 4. Percent distribution, extramural R&D employ­
ment of scientists and engineers on Federal programs, by
selected agencies and performer, 1966
Performer

Department or
agency

All
extra­
mural

Industry1

Colleges1
and
universi­
ties

Indepen.
non­
profits 1

State and
local
govern­
m e n t1

All agencies

100.0

100.0

100.0

100.0

100.0

A griculture...........................
Commerce............................
Defense.................................
Health, Education and
W elfare.............................
In te rio r..................................
A EC........................................
N A SA .....................................
N S F ........................................
O th e r......................................

0.7
( 2)
46.3

<2)
( 2)
54.1

3.1
<2)
21.5

<2)
<2)
45.5

( 2)
<2)
( 2>

8.3
( 2)
8.5
31.4
2.3
2.0

0.5
1.5
4.0
39.3
( 2)
1.8

27.4

33.0
( 2)
8.1
3.9
2.5
6.5

70.6
10.6

1.0
24.2
12.9
10.0
1.2

—

5.3
—

13.5

1 Includes federally funded research and development centers.
0 .5 percent.

1Less than

NOTE: Detail may not add to 100 because of rounding.

programs, approximately 128,000, falls a bit short of
the 130,000 reported for December 1965 by a highly
specialized group of defense contractors, which in­
cludes only a portion of subcontract employment and
no direct employment.40 Since the definitions of scien­
tists and engineers are the same for the BLS-NSF time
series data used as control and the DO D -N A SA Eco­
nomic Information System,41 if the former provides
an accurate estimate of total scientists and engineers
either of two possibilities may explain the discrepancy.
Our man-year expenditure estimate in the defense
space field may be too high, or there has been over
reporting of federally supported employment in the
EIS surveys. Neither possibility could be investigated
in the course of this study, however.
The obviously missing dimension in these cost-based
estimates of federally supported extramural employ­
ment is occupational specialty. For evaluation of Fed­
eral influence in the context o f demand and supply of
38 In 1966, 96 percent of industrial R&D expenditure was
allocated to applied research and development. National
Science Foundation, Research and Development in Industry,
1966 (NSF 6 8 -2 0 ), table 68, p. 82.
39 If allowance is made for the absence of employment data
for the Jet Propulsion Laboratoiies, NASA reported a total
of 66,800 extramural scientists and engineers for June 30,
1967. Our own estimate for a date approximately 18 months
earlier is 69,400; N ASA’s budget at that time was higher
than in June 1967.
40 Calculated from unpublished data supplied by the U.S.
Department of Defense, P l a n t w id e E c o n o m i c R e p o r t (Eco­
nomic Information System). The overall proportion of scien­
tists and engineers employed on Federal work in the reporting
establishments has been close to 90 percent in recent years.
41 The reporting instructions for EIS specify that the defi­
nition of scientists and engineers is consistent with the
National Science Foundation definition, also used in the timeseries prepared by BLS. EIS, however, does not exclude
psychologists, but this alone would hardly account for the
difference.

scientists and engineers, detail on either the occupa­
tions or on the related fields of science and engineering
involved is essential. Except for the State and local
government sector, however, the difficulty is not an
absence of sufficient occupational employment detail.
What is lacking is a division of expenditure, both total
and Federal, along comparable lines. Until expendi­
ture or man-year outlays can be matched to occupa­
tional or field of science groups, the budget approach
permits no more than estimates of aggregate scientist
and engineer employment.
This overview of extramural employment of scien­
tists and engineers should end with a few substantive
comments on the pattern of Federal utilization. Keep­
ing in mind the crudity of the techniques of estimation
and resulting errors of estimation, it is nevertheless
clear that in 1966 Federal spending contributed to a
very sizable share of employment in the non-Federal
economic sectors. Table 6 shows that an estimated 20
percent of all scientists and engineers in the nonFederal sectors owed their employment to Federal
program expenditure. Although private industry as a
whole used only about a third of its scientists and engi­
neers on research and development, the estimates in
this study indicate that almost 85 percent of private
employment supported by Federal programs was de­
voted to R&D. In particular sectors, especially the
colleges and universities and the independent non­
profit oi'ganizations. Federal spending may affect more
than 70 percent of the scientists and engineers engaged
in research and development. Only about 5 percent of
non-R&D employment of scientific and engineering
manpower, on the other hand, is directly affected by
Federal dollars.
Agency (or program) patterns of extramural util­
ization of scientists and engineers also show significant
variation. Although less than 15 percent of such em­
ployment is in functions other than R&D, in the older
established departments and agencies— the DepartTable 5. Im plicit performer cost-ratios, extramural R&D
scientists and engineers in Federal programs, selected
agencies, 1966
Performer
All
extra
mural
performers

Industry1

Colleges1
and
Universi­
ties

All agencies

$53,770

$57,175

$41,500

$53,395

$22,700

A griculture...........................
Commerce............................
DO D........................................
HEW .....................................
In te rio r..................................
AEC........................................
N A SA.....................................
N S F ........................................
O ther......................................

45,975
46,300
50,050
45,100
51,600
63,300
60,150
42,550
51,900

52,000
50,000
50,650
50,000
58,400
102,630
61,760
( 2)
54,170

41,800
38,000
41,365
41,350
40,200
41,290
41,270
41,100
57,720

( 2)
( 2)
52,830
52,850
( 2)
52,830
52,860
53,300
55,700

<*>
( 2)
( 2)
20,800
46,670

Department or
agency

Inde­
pendent
non­
profit 1

State and
localgovern
m ent3

—

34,450
—

9,130

1 Includes federally funded research and development centers.
2 Estimated number of scientists and engineers too small to compute cost
ratio.
3 Federal R&D funds for State governments in FY 1966 estimated at 50 per­
cent of the "o th e r” performer category.




Table 6. Total and federally supported extramural em­
ployment of scientists and engineers, selected performers,
1966
Character of work
and performer

Total
Federally
extram ural supported

Percent
Federal

All activities (in thousands)............................

1279.2

261.7

20

Research and developm ent..........................................
Industry1........................................................................
Colleges and universities1........................................
Independent nonprofit organizations1..................
State and local governments...................................
Non-R & D .............................................................................

454.5
370.9
65.6
213.6
4.4
824.7

221.5
161.6
47.4
10.7
1.8
40.2

49
44
72
79
41
5

’ Includes federally funded research and development centers.
2 Estimated from Scientific Activities of Nonprofit institutions (NSF 6 7 -1 7 ,
March 1969).
SOURCES: See text and appendix C.

ments of Agriculture and Commerce, for example—
the reverse is the case. (See table 20.) As will be shown
in chapter V, such agencies also tend to have a rela­
tively high proportion of their total R&D activity in
intramural employment. Agency influence on the R&D
performance sectors also varies. The Department of
Defense affects more than half of R&D employment
on Federal programs in industry, but influences only
one-fifth of R&D scientists-engincers employment in
colleges and universities, where the Department of
Health, Education, and Welfare and the Atomic
Energy Commission are more prominent sources of
activity. (See table 4.)
An effort to distribute Federal agency employment
in R&D by industry produced rather indifferent
results,42 sufficient only to indicate some broad rela­
tionships. Industrial employment of R&D scientists
and engineers is highly concentrated in the programs
of the Defense Department, which result in half or
more of federally supported R&D employment in all
but one industry covered in the Census-NSF survey.43
Other agencies have correspondingly smaller effect on
federally supported R&D employment. Federal effect
on total R&D employment, on the other hand, is much
more varied among the industries, partly because of
the dependence of the industry on Federal expenditure.

Evaluation of the budget approach
As demonstrated in this chapter, the budget or manyear expenditure approach does permit estimation of
federally supported employment of scientists and
engineers. Sufficient information exists to construct
42 The principal factor inhibiting this exercise is the con­
ceptual difference in reporting employment by industry.
Bureau of Census industry groups, in their reports on R&D
manpower and expenditure, are based on the c o m p a n y as the
reporting unit, while the Bureau of Labor Statistics surveys
of science and engineering manpower are on an e s ta b lis h m e n t
basis.
43 Chemicals, machinery, electrical equipment, motor
vehicles, aircraft missiles, instruments, and all other. See
R e s e a r c h a n d D e v e l o p m e n t in I n d u s t r y (NSF 68 -2 0, 1966),
table 46, p. 60.

15

plausible, if not wholly accurate, estimates of such em­
ployment. The various dimensions for which estimation
is possible, moreover, reveal a substantial amount of
information about the structure of the Federal employ­
ment effect in all major extramural sectors.44 In only
two, but very important, dimensions— occupation a n d /
or field of science and function— are the sources
incomplete or seriously deficient in the ingredients
needed to estimate employment.
The principal weakness in the budget approach was
found to be on the side of financial and expenditure
data. Rather strong assumptions had to be made in
allocating aggregate expenditures among Federal pro­
grams and agencies, particularly in areas other than
research and development. Only two extramural
sectors provided any detailed estimates of expenditure
per scientist and engineer and thereby necessitated
crude assumptions to derive employment estimates for
other sectors. Employment data, on the other hand,
appear to be adequate for most sectors. Indeed, much
more detailed estimates encompassing at least R&D
employment by occupation or field of science would
have been possible if such data could have been
bridged to financial and cost data, appropriately dif­
ferentiated by funding sources. Efforts to integrate
employment and financial data in the fields of science
and engineering should therefore be given careful
consideration.
No approach to a problem of estimation such as is
attempted in this study can exceed the limitations
of the sources of data. Apart from the question of
modification and improvement in the data sources,
however, the merits and limitations of the budget
approach as such should be recognized. Its principal
merits are, first, that currently it provides the only
comprehensive means of estimating Federal employ­

16



ment influence in science and engineering, and, second,
that the kinds of data required are already being
collected, although not in ztll sectors. The principal
limitation is that the approach is relatively inflexible
as a tool of manpower planning in Federal programs.
It can provide information on the manpower impli­
cations of current changes in support levels within a
given program structure. It could also be used to
project changes in program requirements, but only if
there is a suitable means of projecting both expen­
diture and man-year outlays.45 On the other hand, to
visualize the utility of the approach for estimating the
requirements for scientists and engineers in so-called
“ new initiative” programs is difficult. The levels of
detail on total and man-year expenditures required
for such programs, in particular, may make it worth­
while to consider a more direct attack.46
44 Previous efforts to utilize man-year expenditures as esti­
mators of employment have covered fewer sectors and provide
no functional and agency detail. For an example of such an
earlier effort, see Melville S. Green, Studies in Scientific and
Engineering Manpower (U.S. Department of Commerce,
Office of Assistant Secretary for Science and Technology,
Staff Report 6 3 -1 , October 1963).
45 A pioneering effort to project Federal program expendi­
ture is the work of Gerhard Colm and Peter Wagner, Federal
Budget Projections (Washington, D.C., Brookings Institution,
1966). The conceptual and measurement problems involved
in developing £fh index of R&D costs are discussed in Helen
S. Milton, Cost of Research Index (Research Analysis Corp.,
McLean, Va., March 1966) ; also see Kathryn S. Arnow,
“ Indicators of Price and Cost Change in R&D Inputs,” and
Allan D. Searle, “ Measuring Price Change in R&D Pur­
chases,” in Proceedings (American Statistical Association,
Business Economics Section, 1966).
46 This is not to say that a very rough first cut could not be
taken by assuming certain levels of expenditure and unit costs
of scientist and engineer employment. See the illustration pro­
vided in chapter V I, pp. 29-39.

Chapter IV.

Extram ural Employment of Scientists and Engineers in Federal Programs:
Interindustry Transactions Approach

The results described in the previous chapter provide
only a broad picture of the effects of Federal spending
on the deployment of scientific and engineering man­
power. They confirm more casually constructed or
indirect evidence of the concentration of the Govern­
ment’s demand for scientists and engineers in research
and development and among certain extramural R&D
performers. On the other hand, they provide only a
limited view of the industrial structure of program
manpower requirements, even in the R&D area, and
virtually no occupational detail. Both of these latter
dimensions are essential to purposes such as planning
for major changes in program support levels or for the
introduction and development of substantial new
programs.
The relatively new tool of input-output analysis
offers a method of overcoming some of the limitations
of the budget approach. Input-output analysis consists
of manipulating a tabular model of the economy,
which displays the transactions among all producing
industries and between them and the final demand
sectors, including government. The technique has been
used for studies of a variety of economic relationships,
including studies of employment and manpower re­
quirements.47 This chapter briefly describes the essen­
tial features of the interindustry transactions approach,
provides an estimate and some analysis of extramural
employment in scientific, engineering and allied occu­
pations resulting from Federal purchases in 1965, and
evaluates the approach as a manpower planning tool.

Elements of the interindustry transaction approach
Estimates of the employment of scientists and engi­
neers attributable to Federal expenditure may be
derived from so-called interindustry employment tables.
These tables show the total employment generated by
the delivery of a billion dollars of an industry’s output to
final demand or to other industries. Total employment
is the sum of employment in the industry providing the
goods or services and the employment in all supporting
industries. The total of a particular industry’s employ­
ment, therefore, is the sum of employment generated
by its gross output, that is, the industry’ s delivery to
final demand and to all other industries resulting from
a billion dollars of sales. Application of appropriate




occupational patterns or coefficients to this employ­
ment total provides estimates of occupational employ­
ment in the industry per billion dollars. If the total
purchases of the Federal Government, or any of its
component programs, and their industrial distribution
are known, then the corresponding occupational em­
ployment in each industry and in all industries may
be estimated.
Acceptance of the results achieved through appli­
cation of the interindustry transactions approach
depends on the acceptability of certain assumptions
and limitations about the process and the data. In this
section, these assumptions and limitations are set forth.
Succeeding sections (1) present the resulting occupa­
tional distributions and derived total requirements for
scientific and engineering manpower in certain Federal
programs, and (2) evaluate the usefulness of the
approach in terms of the objectives of a scientific and
technical manpower information system.
The interindustry employment tables. Extramural
employment of scientific and technical manpower on
47 A number of employment studies based on input-output
analysis have appeared in the Monthly Labor Review. See
Jack Alterman, “ Inter-industry Employment Requirements,”
(July 1965), pp. 8 4 1-85 0; Richard P. Oliver, “ The Employ­
ment Effect of Defense Expenditures,” (September 1967),
pp. 9 -1 6 ; Max A. Rutzick, “ Worker Skills in Current Defense
Employment,” (September 1967), pp. 17-21. Projections of
employment either in terms of the numbers expected as the
result of economic growth or in terms of hypothetical social
and economic goals, analogous to federally supported pro­
grams, are other uses of input-output analysis. See Projections
1970: Interindustry Relationships, Potential Demand, Em­
ployment (BLS Bulletin 1536, 1966) ; and Leonard A. Lecht,
Manpower Requirement for National Objectives in the
1970’s (Center for Priority Analysis, National Planning
Association, Washington, D .C., February 1968).
48 Input-output tables are essentially a matrix or grid relat­
ing all transactions in the country, including industry sales to
other industries and sales, to a final demand sector such as
business, households, and so on. Each transaction between
industries, which make up the processing sector of the table,
is counted twice to reflect the fact that it represents an input
as well as an output of an industry. A good introduction to
the subject is William H. Miernyk, The Elements of InputOutput Analysis (New York, Random House, 1965). A more
extensive and more mathematical treatment may be found in
Hollis B. Chenery and Paul G. Clark, Interindustry Economics
(New York, Wiley, 1959)

17

Federal work may be estimated from interindustry
employment tables for the Federal sector of final
demand. Such tables are generated from the basic
input-output table 48 and show for each billion dollars
of Federal purchases the equivalent employment in the
supplier industries and their supporting industries.49
The product of the industry employment coefficients
and expenditure on the output of an industry equals
the estimated total employment in the industry associ­
ated with the expenditure.
The employment estimates used reflect the industrial
distribution of total private employment in 1965 and
were prepared by the Interagency Growth Project.
The input-output tables used to generate these esti­
mates of employment are based on the 1958 table;
the technical or product coefficients were adjusted to
1965 and reflect subsequent changes in the technol­
ogies of production, changes in the product mi:-:, and
changes in factor prices. Such adjustments are neces­
sarily rough, because they have been based on frag­
mentary information on industrial and technological
developments rather than intensive industry studies
required for greater precision of estimate.50
T h e F e d e r a l B i l l - o f - G o o d s . Given industry employ­
ment coefficients, the first step in the estimating process
is to obtain industrial distributions of Federal purchases
of goods and services. Such distributions conventionally
referred to as the Federal “ bill-of-goods,” consist of

49 l.e., the industries supplying raw materials, power, serv­
ices, etc., for current production. Input-output tables usually
do not include purchases of capital goods as an input coeffi­
cient. See Projections 1970, etc., p. 20; however, much
research has been undertaken to solve the problem of empir­
ical estimation of capital coefficients. See Chenery and Clark,
op. cit., pp. 149-153.
50 For further discussions of the nature of such adjustments,
see Chenery and Clark, pp. 85-100. A completely revised
table of interindustry relations for 1963 was published after
this study had been completed.
51Budget of the United States, Special Analysis, Fiscal
Year 1970, table 0 -3 , p. 209. Probably only about one-third
of State-local aid consisted of transfer payments under public
assistance and medicaid programs.
52 The exclusion of these Federal payments is necessary to
maintain consistency between the input-output tables and the
national income and product accounts.
53 Projections 1970, table IV -4, pp. 63-64.
54 Data on State and local government expenditures are
from Joint Economic Committee, Subcommittee on Economic
Progress, State and Local Public Facility Needs and Financ­
ing, December 1966, p. 37.
55 Conversations with officers of the General Accounting
Office, which warrants the payment of the Government’s bills,
revealed that there is no coding system which identifies expen­
ditures with programs and the supplier industries as classified
by the Standard Industrial Classification scheme. It had been
hoped that this would be a primary source relating purchases
to program activity.
56 Budget of the United States, Special Analysis, Fiscal
Year 1970, table A-6, p. 18.

18



payments for Federal purchases of goods and services
in the private industrial sector. Except for purchases
from private educational and nonprofit organizations,
this sector corresponds to the industrial and business
sector of the private economy. Expenditures on per­
sonal services hired directly from the household sector
and expenditures in the government sector itself, in­
cluding government enterprises, are not included in
the Federal bill-of-goods.
In addition to purchases of goods and services, Fed­
eral aid to State and local governments constitutes a
significant and growing proportion of government out­
lays. In 1965, Federal outlays to such governments
were nearly 10 percent of total Federal outlays and
about 19 percent of Federal outlays on domestic
programs.51 As presently constituted, Federal grants,
loans, and transfer payments to State and local gov­
ernments are not included in the Federal bill-of-goods,
however.52 On the other hand, there is a bill-of-goods
for State and local governments,53 which includes the
expenditure of Federal funds in the industrial aggre­
gates. T o estimate the corresponding federally sup­
ported employment, a crude bill-of-goods representing
the expenditure of Federal funds was constructed from
unpublished data adjusted to a calendar 1965 basis.
Transfer payments to individuals, under federally sup­
ported public assistance and some manpower and
poverty programs, and shared revenues were excluded.
The amount constituting the bill-of-goods— about $6.8
billion— represents about 9 percent of 1965 State and
local government spending,54 more than two-thirds of
which was allocated to the construction of highways
and other facilities.
In principle, the total purchases of the Federal Gov­
ernment can be subdivided into any convenient set of
industrial distributions representing various functions
or programs. In practice, however, such distributions
are not easily accessible, but must be painstakingly con­
structed from various sources of information, including
special studies.55 Consequently, the only ready-at-hand
bills-of-goods available divide Federal purchases be­
tween defense and nondefense purchases.
Defense purchases, o f course, have dominated Fed­
eral purchases of goods and services during the past
decade; in fiscal year 1965, the year of estimate of
scientists and engineers on Federal work, the defense
effort absorbed 75 percent of such purchases.56 The
defense bill-of-goods used here includes only those
purchases made by the Department of Defense. Some
additional defense purchases made by agencies such
as NASA and AEC are included in the category
“ Other government,” because the total purchases of
those agencies could not be segregated functionally.
Estimating “ Federal” Employment. Estimation of
the employment attributable to Federal purchases of
goods and services in the private sector, thanks to the

computer, is a comparatively simple mechanical pro­
cess. Each bill-of-goods is run against an employment
inverse, that is, a matrix of employment coefficients
arranged by industry, to produce estimates of total
employment by industry. These estimates become the
base for further estimation of occupational employ­
ment in each of the sectors considered; in this case:
Final demand, total Federal expenditure, Department
of Defense expenditure, and other Government expen­
ditures. Estimates of federally supported employment
of scientists and engineers in the State and local gov­
ernment sector, however, could be broken down by
Federal department or agency.
Once such employment estimates have been made,
it is then comparatively easy to calculate the percent­
ages or ratios of federally supported employment to
total employment in each industry and occupation.
The percent distribution of the Federal share between
defense and nondefense employment in each industry
as well as the industrial distribution in each program
sector, as noted above, are then calculable as are sim­
ilar distributions of occupational employment.
Each employment estimate is composed of two parts
— direct and indirect employment. For any given in­
dustry direct57 employment is the employment created
by direct delivery to final demand or a sector of final
use such as the Government. The indirect component
of total employment represents employment in the
industry created by Government purchases in all other
industries. Sales of steel billets to the Governmentowned ordnance plants, for example, create direct
employment, but Government purchases of shovels
create indirect employment in the steel industry. A
breakdown of the two types of employment could be
useful for manpower planning in Federal programs,
although no attempt was made for this study.
A p p l i c a t i o n o f O c c u p a t i o n a l P a t t e r n s . The final
step short of analysis is to apply occupational patterns
or coefficients to the estimates of total industry employ­
ment. Federally supported and total industry employ­
ment used for this purpose are shown in appendix B.

57 Sometimes referred to as primary employment.
58 Occupational Employment Patterns for I960 and 1975
(BLS Bulletin 1599, December 1968).
59 A conversion table developed by the BLS Division of
Occupational Employment Statistics was used in this study.
The SIC groupings used in the industry-occupation matrix
are shown in BLS Bulletin 1599, appendix B, pp. 83 -8 5. The
interindustry classification scheme is in Projections 1970,
table IV -1, p. 58.
60 The patterns in the BLS occupation-industry matrix are,
therefore, the patterns of the average firms in the respective
industries.
61 For example, in fiscal year 1965, 25 percent of procure­
ment by the Department of Defense was through a sole
source. The remaining purchases were made by more competi­
tive procurement methods. U.S. Department of Defense,
Annual Report (Washington, D .C., 1967), p. 415.




The occupational patterns used included 21 distinctive
occupations: Eight scientific occupations, including
mathematicians; 10 engineering occupations, and 3
technician groups. The occupational patterns used,
and the method of their construction, were adapted
from the BLS occupation-industry matrix to correspond
to the industry groups of the input-output tables.58
The resulting estimates of occupational employment
attributable to Federal purchases of goods and services
deviate from “ true” estimates, since the correspon­
dence of the two schemes of industrial classification is
only approximately identical. The selection and classi­
fication of the 116 industries in the industry-occupation
matrix were governed by interest in particular occupa­
tions (mainly those requiring extensive education and
training) and by the availability of related historical
data sufficient to permit projections of employment.
Consequently, the occupational patterns for some in­
dustries, for example, construction, are based on broad
industrial categories, while others relate to finer in­
dustrial groups. The interindustry tables consolidate
detailed SIC industries into 86 groups, 77 of which
were employed in this study. This scheme of grouping
is governed largely by the composition of final demand
and its distribution among producing industries. Be­
cause of these differences in purpose, the conversion
of the industry-occupation matrix to the interindustry
base cannot be complete with respect to the relevant
occupational patterns, although both schemes exhaust
total employment.59 Until occupational patterns for a
greater number of detailed industries are developed
on a common base, estimates of occupational employ­
ment must be regarded as only a first approximation.
An additional source of error in estimating occupa­
tional employment may result from the selective char­
acter of Federal purchases among firms in supplier
industries. Because of lack of information about the
intraindustry distribution of such purchases and corre­
sponding differences in occupational patterns,60 the
assumption has been made that the patterns for the
industry as a whole are identical with those of Federal
suppliers. In fact, of course, Federal Government pro­
curement is not randomly distributed within indus­
tries. In defense procurement especially, a particular
company or establishment commonly is designated as
“ sole source.” 61 Such suppliers may have occupational
patterns that are unique because of their unique
product-mix and the applicable technology involved.
Extramural employment of scientists
and technicians

The resulting estimates of extramural employment
of scientists and engineers supported by Federal funds
are necessarily incomplete without estimates from other
sources. As explained earlier, the Federal bill-of-goods
covers only a portion of employment outside of the

19

Federal Government. Consequently, the overall results,
shown below in table 7, include adjusted estimates
based in part on the man-year expenditure estimates of
employment developed in chapter III.
Several crude adjustments had to be made to esti­
mate federally supported employment in the uncov­
ered sectors. Ratio estimates were used to estimate
total employment in publicly controlled universities
and colleges to reflect the fact that employment in
private higher education is included in the estimates
based on the Federal bill-of-goods.62 Federally sup­
ported employment in both publicly controlled higher
education and in State and local government, public
employment was estimated by applying ratios calcu­
lated from the scientists and engineers’ registers to
control totals in the BLS-NSF time series.63 This
method of estimate was necessary because the registers,
despite certain shortcomings, are the only primary
source of data on total federally supported employ­
ment of scientists and engineers in those sectors. The
estimate for the nonprofit institutions sector, on the
other hand, is the same as developed in chapter III.
Estimates of federally supported extramural em­
ployment in the State and local government sector
were made on the basis of a crude bill-of-goods sup­
plied by the Bureau of the Budget. The resulting
estimates of employment represent those portions of
employment generated in private industry by the pur­
chases of State and local governments and which are
assumed to be attributable to grants-in-aid from the
Federal Government. Highway and educational facili­
ties construction absorbed more than 90 percent of
such funds in calendar year 1965. Despite the substan­
tial amounts involved, the effect on the shares of
federally supported employment in private industry,
as shown in table 7, is fractionally small.
The addition of estimated employment of federally
supported engineers in that sector increases the all­
sector total of extramural employment by less than
1 percentage point.
Appendix C shows for the detailed occupations the
estimated total employment and federally supported
employment resulting from State and local govern­
ment purchases from private industry. The estimates
of occupational employment are questionable, however.
In view of the substantial role of the Federal Govern­
ment in State highway construction, for example, the
number and percent (6 percent) of employment of
civil engineers are remarkably small. Federally sup­
ported mining engineers, on the other hand, are esti62 The ratios used were taken from National Science Foun­
dation, S c ie n tific A c t i v i t i e s at U n iv e r s itie s r a n d C o l le g e s , 1 9 6 4
(NSF 68-22, Washington, D .C., 1968), table II, p. 16.
63 E m p l o y m e n t o f S c ie n tis ts a n d E n g i n e e r s in th e U n i t e d
S t a t e s , 1 9 5 0 - 1 9 6 6 (NSF 68 -3 0 ) table A - l .
64 O c c u p a t io n a l E m p l o y m e n t P a tte r n s (BLS Bulletin 1599)
appendix C., p. 86.

20



mated at 27 percent of extramural employment in
State and local government programs. Although these
estimates could be close to the actual relationship,
their implausibility strongly points to the general need
for more detailed program-based bills-of-goods and
possibly finer industry-occupation patterns.
Several observations of substantive interest may be
made on the basis of table 7. First, the overall Federal
share of extramural scientist-engineer employment is
significantly large. Some comparisons may help to
evaluate this share. In 1965, the Government’s budget,
exclusive of transfer payments and net interest, repre­
sented about 12 percent of Gross National Product.
Estimates of the direct and indirect effect of Federal
spending on total employment in 1965 indicate that
the Government’s share, again only in private industry
economy is about 5 percent. In the broad category,
professional, technical, and kindred workers, the Fed­
eral portion still only approaches 6 percent of the
estimated total. Further, scientists and engineers in all
sectors of employment, public and private, were only
about 1.5 percent of total employment in 1960, and
are projected to increase to a little over 2 percent by
1975.64 Thus, in 1965 the federally supported share of
scientist and engineer employment was 11—14 times
greater, relatively, than the total proportion of such
employment in the economy.
The Federal effect in particular sectors is even larger
than its total effect on the employment of scientists and
engineers. The estimates shown in table 7 indicate very
large shares of federally-supported scientist-engineer
employment in the publicly controlled universities and
colleges, and in the privately controlled nonprofit
organizations. In these sectors the shares of federally
supported employment are 2 to 4 times larger than
the sectoral proportions of total science and engineer­
ing employment. On the other hand, State and local
governments experienced the least effect; even with
the addition of federally supported employment in
private industry the Federal share is only 13 percent.
Evaluation of the significance of Federal support,
however, must look not only at aggregates, but also at
its influence on the utilization of particular occupa­
tional groups. The degree of utilization and, by impliTable 7. Estimated employment of scientists and engi­
neers supported by Federal funds, by type of employer,
1965
Type of employer

Total

All employers.......................................................
Private industry1..............................................................
Scientists2......................................................................
Engineers........................................................................
State and local governm ent..........................................
Publicly controlled universities and colleges . . . .
Nonprofit organizations.................................................

1,290,300
1,088,400
224,743
863,657
83,200
104,700
14,000

Federally
supported
249,907
169,827
21,965
147,862
12,480
59,300
8,300

Percent
Federal
19
16
10
17
15
57
9

1 Includes medical, educational services, and nonprofit organizations other
than State and local government units.
2 Includes mathematicians; excludes social scientists.

Table 8. Federally supported employment in selected
scientific and technical occupations as a percent of total
occupational employment, private industry, 1965
[Employment in thousands]

Occupation

Total
private
industry

Total
Federal
Govern­
ment

Department Other
of
Federal
Defense Govern­
ment

In percent
A ll occupations............................
Professional, technical and kindred
w orkers...................................................
Scientists and engineers........................
Engineers, technical .............................
Engineers, aeronautical..........................
Engineers, chemical.................................
Engineers, c iv il..........................................
Engineers, electrical................................
Engineers, m echanical...........................
Engineers, m etallurgical........................
Natural scientists.....................................
Chemists.....................................................
Agricultural scientists............................
Biological scientists.................................
Mathem aticians........................................
Physicists...................................................
Technicians, except medical and
dental.......................................................
D raftsm en...................................................

62,326.0

5.2

3.3

1.8

6,026.8
1,088.4
863.6
48.3
45.3
127.6
186.0
166.4
25.1
224.7
92.9
22.0
23.4
25.7
22.4

7.7
15.6
17.1
62.3
10.6
7.8
20.0
16.6
15.2
9.8
8.7
6.5
5.5
18.0
16.6

5.1
11.0
12.4
51.1
6.4
4.1
14.7
11.8
10.3
5.9
4.9
1.8
2.5
13.1
11.7

2.6
4.6
4.8
11.2
4.2
3.4
5.3
4.8
5.0
3.8
3.8
4.7
2.9
4.8
4.9

746.0
246.4

11.8
12.2

8.1
8.6

3.7
3.7

cation, the goods and services to which their services
contribute necessarily underlie any judgment about the
size of Federal utilization of scientific and technical
manpower. Table 8, which shows the estimated shares
of federally supported employment in particular scien­
tific and technical occupations as a percent of total
occupational employment provides some insights. The
table shows considerable occupational variation in the
federally supported shares of occupational employment,
ranging from under 6 percent (biological scientists) to
over 60 percent (aeronautical engineers) of employ­
ment. Federally supported employment was generally
higher among engineers than among scientists.
Table 8 also shows how the estimated Federal shares
of extramural employment were distributed between
employment in programs of the Department of Defense
and all other Federal programs. Not too surprising is
the dominance of the Defense Department in almost
every occupation. In only a few occupations, notably
civil engineering and agricultural scientists, a major
share of the Federal Government’s impact is generated
by other programs.
For planning purposes, differences in occupational
patterns among types of programs may be more rele­
vant than the levels of expenditure involved. Unfor­
tunately, the data available allow an analysis only of
broad differences between Federal programs and other
patterns of utilization of engineers and scientists. Table
9 indicates that the two Federal occupational patterns

are generally similar to each other and to the private
economy.
Differences in occupational patterns between the
Federal Government as a whole and the total economy
clearly appear to be influenced mainly by defense
spending. The higher utilization ratios in certain
engineering specialties— aeronautical and electrical
engineers— is largely because of the relatively large
numbers employed in defense programs. In other Fed­
eral programs, the engineering pattern more nearly
approximates the pattern of the total economy. Federal
utilization of scientists and mathematicians is also sim­
ilar to the total pattern, both in defense and non­
defense programs. Such differences in influence among
the occupations in part are attributable to the heavy
Federal support of applied research and development,
on the one hand, and the relatively smaller allocation
of Federal R&D funds to basic scientific research.65
The differences in occupational patterns between
the various sources of expenditure are probably the
joint result of two factors. One is the occupational
patterns of employment in the various industries,
which may differ substantially. The other factor is the
industrial distribution of procurement in particular
Federal programs. Since, in the inter-industry trans­
actions approach, the individual industry occupation
patterns are necessarily assumed to be unchanged from
program to program, the industrial mix of the pro­
gram becomes the principal determinant of the overall
occupational pattern of any given program.
The force of the industrial influence is illustrated in
tables 10, 11, and 12. Table 10 shows the estimated
distribution of total employment at large and in the
defense and nondefense areas of Federal program
activity. In 1965, the Federal Government procured
Table 9. Percent distribution of scientists and engineers
in private industry, selected occupations, by source of
support, 1965

Occupation

A ll
sources
of
support

Scientists and engineers (num ber) . .

1,088,400

169,827

120,134

49,704

Total (in percent).......................

100.0

100.0

100.0

100.0

79.4
4.4
4.2
11.7
17.1
8.8
15.3
2.3
1.8
5.3
8.0
20.6
8.5
2.0
2.1
1.8
2.4
2.1
1.6

87.1
17.7
2.8
5.7
21.9
10.0
16.3
2.2
0.6
4.0
5.8
12.9
4.8
0.8
0.7
0.8
2.7
2.2
0.9

88.9
20.5
2.4
4.4
22.8
10.4
16.4
2.1
0.5
4.0
5.4
11.1
3.8
0.3
0.5
0.7
2.8
2.2
0.7

82.6
10.8
3.8
8.7
19.7
9.3
16.0
2.5
0.7
4.0
6.8
17.4
7.1
2.1
1.4
0.8
2.5
2.2
1.3

Engineers, technical................................
Engineers, aeronautical..........................
Engineers, chemical.................................
Engineers, c iv il..........................................
Engineers, electrical................................
Engineers, industrial...............................
Engineers, m echanical...........................
Engineers, m etallurgical........................
Engineers, m ining....................................
Engineers, sales........................................
Other engineers, technical....................
Natural scientists.....................................
65
In fiscal year 1965, of the total of $15,731 millions of Chemists.....................................................
Agricultural
scientists............................
Federal obligations for research, development, and R&D
Biological scientists.................................
plant, 31 percent was allocated to research and 11 percent
Geologists and geophysicists................
Mathem aticians........................................
to basic research. The Department of Defense was responsible
Physicists...................................................
for more than half of total Federal development spending.
Other natural scientists.........................

These percentages were calculated from F e d e r a l
X V I , tables C-78, C-81 to C-84, pp. 212-214.




F u n d s,

Total
Federal
Govern­
ment

Depart­
ment
of
Defense

O ther
Federal
Govern­
ment

Vol.
NOTE: Figures may not add to total because of rounding.

21

much more of its goods and services from the manu­
facturing industries than from other industrial sectors.
Federal programs outside of the defense area, however,
were relatively large purchasers of goods and services
from the nonmanufacturing industries, notably in con­
struction and the service industries. The industrial
pattern for engineers and scientists, shown in tables
11 and 12, are generally similar to the pattern of
Federal effect on total employment. On the other
hand, there are substantial differences between the two
occupational groups. Engineers in federally supported
programs are more heavily concentrated than scien­
tists in the manufacturing industries, and scientists
have a much more substantial representation in the
services and other nonmanufacturing industries. Pre­
sumably, further breakdown of Federal programs
would show even more varied industrial and occupa­
tional patterns.
Table 10. Industrial distribution of estimated total em­
ployment, by expenditure source, 1965
Industry*

All
expenditure

Department Other Federal
of Defense Government

A ll industries (n u m b er).......................

62,326,607

1,918,300

1,154,773

Total (in percent)..................................

100.0

100.0

100.0

Agriculture, forestry and fisheries...............
M ining.....................................................................
Construction..........................................................
M anufacturing......................................................
Transportation.....................................................
Communications and public u tilitie s ............
Trade.......................................................................
Finance, insurance and real estate...............
Services and miscellaneous.............................

7.5
2.9
5.6
29.4
4.3
2.4
24.4
5.4
17.7

1.3
1.5
4.6
77.6
( 2)
3.9
3.7
0.3
7.0

4.8
1.8
15.1
51.3
7.1
5.9
4.5
1.8
10.1

1 For S IC-related industries, see 6LS Bulletin 1536, table IV -1, p. 58.
2 Less than 0 .0 5 percent.
NOTE: Figures may not add to total because of rounding.
SOURCE: Appendix table A.

E v alu ation of the interindustry transactions
approach
As a method of measuring Federal influence on the
employment of scientific and technical manpower and
as a tool for manpower planning, the interindustry
transactions approach on balance is a promising tech­
nique. Conceptually, it consists of a set of simultaneous
equations representing a set of interrelated production
functions, in which output is treated uniformly between
industries and among various bills-of-goods. For pro­
gram planning in which the focus is on manpower
requirements, this feature is an analytical advantage.
Although some limitations exist, the major difficulties
arc found not in the technique itself, but in the under­
lying information system. Therefore, separating these
two aspects is essential in evaluating the approach.
Given the availability of the basic input-output table,
the interindustry transactions approach potentially has
several very strong technical advantages: Comprehen­
siveness, internal consistency, analytic compatibility,
and adaptability. The first and second advantages are

22



inherent in the input-output tables. Such tables, by
definition, must account for all transactions in the
economy.66 Hence, the resulting employment relation­
ships are comprehensive, irrespective of the varying
quality of the data in the cells of the employment
matrix. Both direct and indirect employment are also
estimated. By the same token the employment coeffi­
cients and estimates generated by an input-output
table are internally consistent, that is, the same method
of estimating employment is not only applied to each
of the industries but also resulting employment in the
rows of the table must equal the column totals. This
will be the case whether total employment or employ­
ment in a particular occupation is under consideration.
An especially strong feature of the approach is its com­
patibility with various analytic sources of data, partly
by design as in the case of the national income
accounts with which the input-output tables share a
common source of data as well as conceptual similar­
ity, and partly by linkages to employment and other
data via the Standard Industrial Classification system.
Because of this feature, the interindustry approach
readily lends itself to the analysis of dynamic changes
in the growth and structure of the economy, including
projections of employment requirements. Finally, in
principle the interindustry transactions approach is
highly adaptable, since both the sources of demand
and the resources required to satisfy varying sources
of demand can be disaggregated to a relatively fine
level of detail. In employment analysis, the resulting
detail of structure can be especially useful.
As presently designed, the interindustry transactions
approach seems to have at least two disadvantages,
viewed from the perspective of the occupational em­
ployment effect of Federal programs. One of these is
that the occupational patterns related to any partic­
ular source of employment are purely a function of the
industry mix. In this respect, the same occupational
pattern applies whether the delivery of the industry’s
output is to total final demand or to any segment of
that total such as the Federal defense sector. Yet,
occupational patterns can vary markedly within indus­
tries as the result of differences in product mix, as in
electronics, for example.67 This difficulty might be
66 For the purpose of measuring Federal effect, the national
economy usually is the relevant base for the input-output
model. However, it is also possible to construct input-output
models representing transactions within a region and between
regions in a national economy. For a brief discussion of
regional input-output analysis, see William Miernyk, op. cit.,
chapter 4, pp. 58-77.
67 The occupational distributions of the military-space and
the consumer products sectors differ greatly in electronics
manufacturing. See Employment Outlook and Changing Oc­
cupational Structure in Electronics Manufacturing (BLS Bul­
letin 1363, 1963), p. 37. See also the discussion in Joseph F.
Fulton, “Employment Impact of Changing Defense Pro­
grams,” Monthly Labor Review (May 1964), pp. 511-513.

reduced, but not altogether eliminated, by a larger,
more detailed input-output table and correspondingly
more detailed employment coefficients and industry
employment patterns. However, this difficulty is not
unique to input-output analysis, but is a limitation on
most other practical approaches to estimating occupa­
tional requirements.
The other disadvantage, which, however, may be
particularly apt to Federal program activity, is that the
approach is relatively insensitive to changes in the job
content of occupations, especially the character of
work. Occupational requirements for a planned Fed­
eral program derived by the interindustry approach in
the manner described in this chapter will not be mean­
ingful if the related job content is radically different.
The NASA experience is opposite to this point, namely
the need to develop a unique scheme of occupational
classification reflecting the job content of its own work
force. Similarly, the approach has no practical way of
descriminating among the various uses of scientific and
technical manpower, especially between R&D and
other work performed by scientists and engineers.
For practical purposes, however, the limitations just
discussed appear to be tolerable. Other approaches to
estimating manpower requirements for new Federal
programs and developments would encounter similar
problems of undifferentiated occupational patterns and
significant differences in job content. Supplementary
information and special care in interpreting results
should be sufficient in most cases to deal with these
problems.
Deficiencies and gaps in the basic sources of data
may impose a more serious limitation on the potential
use of the interindustry transactions approach in the
near future. The results illustrated in this chapter
cover only private industry and a segment of State
and local government. Other sectors in which there is
substantial Federal support, such as higher education
and other nonprofit organizations, remain practically
uncovered. Estimates of employment in these sectors
not only had to be derived by other means, but also
Table 11. Industrial distribution of estimated engineer­
ing employment, by source of support, 1965
Industry

All
sources

Department Other Federal
of Defense Government

All industries (n u m b e r)......................

863,657

106,803

41,058

Total (in percent)..................................

100.0

100.0

100.0

Agriculture, forestry and fisheries...............
M ining....................................................................
Construction.........................................................
M anufacturing.....................................................
Durable goods.................................................
Nondurable goods..........................................
Transportation.....................................................
Communications and public utilitie s ............
T rad e......................................................................

0.3
4.4
8.7
64.4
55.4
9.0
(■)
4.3
1.8
04
14.7

(•)
0.9
1.9
91.2
88.8
2.4
O)
1.0
O)
(i)
4.8

0.3
1.4
7.4
80.1
74.2
5.9
O)
2.0
0.2
(i)
8.0

Services and miscellaneous.............................

1 Less than 1 /1 0 of 1 percent.
NOTE: Figures may not add to total because of rounding.




Table 12. Industrial distribution of estimated scientist
employment, by source of support, 1965
Industry

All
sources

Department Other Federal
of Defense Government

All industries (n u m b e r)......................

224,743

13,311

8,646

Total (in percent)..................................

100.0

100.0

100.0

Agriculture, forestry and fisheries...............
M ining..............................................................
Construction.........................................................
M anufacturing......................................................
Durable goods.............................................
Nondurable goods......................................
Transportation.....................................................
Communications and public utilitie s ............
Trade..................................................
Finance, insurance and real estate...............
Services and miscellaneous.............................

7.6
77
0.8
53.3
20.0
33.3

1.5
4.8

77.2
61.3
15.9

12.1
3.0
07
67.2
34.1
33.1

0.7
2.0
0.4
27.5

0.3
0.3
O)
15.4

0.3
0.2
(> )
16.1

(l)

' Less than 1 /1 0 of 1 percent.
NOTE: Figures may not add to total because of rounding.

lacked corresponding occupational detail. Presumably
both aspects for those and similar sectors could be
developed and the sectors integrated into a basic inputoutput table. So far as is known, data sources adequate
for this purpose do not yet exist.
A subsidiary issue of possible future importance is
the reliability of the occupational coefficients used in
estimating occupational employment. How important
this issue may be depends on the degree to which
occupational detail is wanted. Total industry employ­
ment generated by the interindustry technique is de­
rived from coefficients based on establishment payroll
reports. The general basis for the occupational matrix,
on the other hand, is the decennial census adjusted for
broad occupational groups and a few detailed occupa­
tions with data from the Monthly Report on the Labor
Force, to reflect changes in intercensal periods. The
occupational coefficients for all of the scientific occu­
pations and for total engineering employment applied
in this study, however, were based on establishment
data, which are believed to be more reliable. Only the
breakdown of the engineering specialties was based
on census sources.68 Should the need arise for more
extended occupational coverage of federally supported
employment, increases in the error of estimated occu­
pational employment are likely because of the present
patchwork character of the industry-occupation matrix.
68
For a complete explanation of the construction of the
occupational employment patterns, see BLS Bulletin 1599,
pp. 4—8. Consideration was given to the BLS S u r v e y o f S c i e n ­
tific a n d T e c h n i c a l P e r s o n n e l in I n d u s t r y as an alternative
source of occupational patterns. The necessary ingredients—
total employment in an establishment and occupational detail
— are collected in the scientific, professional, and technical
personnel programs within the framework of the Standard
Industrial Classification. The degree of occupational detail is
much less than in the patterns used, however, consisting of
four distinctive physical science occupations, two life science
occupations, and mathematicians, but no detailed engineering
specialties. Considering this limitation, the very large amount
of preparation required to construct alternative patterns did
not seem worthwhile.

23

In contrast to the foregoing, the data base and in­
dustrial classification scheme of the interindustry tables
may conceal useful information. As noted, the Federal
bills-of-goods currently available include only pur­
chases of goods and services ; other Federal programrelated expenditures, especially grants-in-aid to State
and local governments, are excluded. It was possible
to construct a partial Federal bill-of-goods represent­
ing Federal expenditure in this sector, but the results
aggregate scientists and engineers employed in publicly
controlled universities and colleges with those em­
ployed in other State and local governmental functions.
Scientists and engineers employed in privately con­
trolled institutions of higher education are also lumped
with those serving in other institutions, including non­
profit organizations. Employment in public and quasi­
public institutions in 1965 probably constituted over
10 percent of total scientist and engineer employment,
and played a proportionately even larger role in Fed­
eral programs. It is an anomaly that there is substantial

24



detail on the disbursement and use of Federal funds
in these performer sectors, but little or no information
on the resulting deployment of manpower. An ex­
panded bill-of-goods including colleges and universi­
ties and nonprofit organizations as distinctive sectors
and corresponding occupational patterns might be
considered as one solution to this problem.
The foregoing considerations tend towards one over­
all conclusion. The interindustry transactions approach
is a useful and adaptable tool for estimating or measur­
ing the manpower implications of Federal programs,
but its effectiveness is highly dependent on the quan­
tity and quality of the data employed. This chapter
has indicated some of the deficiencies in the data avail­
able insofar as they relate to Federal programs. Over­
coming these deficiencies would not in itself make use
of the interindustry transactions approach superfluous-,
but reserved for later discussion and judgment is the
practical question of how the data deficiencies might
be remedied.

Chapter V.

Scientists and Engineers in the Federal Government

In terms of accuracy of estimate and occupational
detail, information on the employment of scientists
and engineers in the Federal Government is unmatched
in any other sector. Since at least 1954, the U.S. Civil
Service Commission in cooperation with the National
Science Foundation has developed a continuing survey
of scientific and technical manpower covering all but
a few of such employees in Federal departments and
independent agencies. The occupational classifications
are convertible to those used in other sectors, but pro­
vided in greater and more specialized detail. With the
introduction of a scheme of functional or character of
work classification in 1967, the results of future surveys
should be analytically even more useful than those
already available.69
The distribution of Federal scientists, engineers and
allied occupations by program categories is a signifi­
cant dimension of the total effect of Federal program
activity and expenditure. Employment of scientific and
technical manpower in the Federal Government has
been increasing much more rapidly than total Federal
employment. Federal scientists and engineers consti­
tute a substantial fraction of total employment in the
scientific and technical occupations as well as in some
particular occupations, although in recent years this
fraction has not been growing. In some particular
occupations intramural employment may represent a
very sizable, even a major share of federally supported
science and engineering employment.
Several sources of data are combined in this chapter
to achieve two objectives. One of these, is to estimate
by scientific and engineering specialty the share of
in-house Federal employment in total occupational
employment. The other objective is to distribute
employment in selected science and engineering occu­
pations into categories representing the structure of
Federal programs. As background, a brief account of
trends in the employment of scientists and engineers
in the Federal Government is provided.

social and behavioral sciences, the employment of
131,000 remaining scientists and engineers is impres­
sive. Federal employees were nearly 10 percent of the
total employed in scientific and engineering occupa­
tions in 1966; Federal employment in all occupations
was only about 4 percent of total nonagricultural
employment.
Growth of employment of scientists and engineers in
the Federal Government during the post-World War
II years has been markedly greater than the growth of
the Federal civilian work force. Table 13 shows that
the total Federal work force was nearly stable through­
out the period, increasing by less than 2 percent. At
least since 1957, on the other hand, Civil Service
scientists and engineers have been increasing by an
annual rate of more than 4 percent. The number of
physical scientists, mathematicians and engineers in
the Federal service in 1965 was nearly twice the
number in 1951. Scientists more than doubled their
numbers, and engineers increased their numbers by
about 75 percent. By 1965, the proportion of scientists
and engineers in the Federal civilian work force was
more than twice the proportion of scientists and engi­
neers in the country.
Science and engineering employment in all sectors,
however, has been growing more rapidly than Federal
employment in these occupations. The estimated share
Table 13. Total civilian employment and scientists and
engineers, Federal executive branch 1951-65
[In thousands]

Total
1951......................................
1953.....................................
1955......................................
1957
1959
1961......................................
1963.....................................
1965.....................................

Total
employ­
ment

Scientists
and
engineers

2,456.0
2,532.0
2,371.0
2,391.0
2,355.0
2,407.0
2,490.0
2,496.0

70.7
84.9
81.5
90.1
97.7
102.2
120.3
132.0

Percent
of
total
2.9
3.4
3.4
3.8
4.1
4.2
4.8
5.3

Scientists1 Engineers
25.6
33.1
32.7
36.5
39.2
40.4
47.3
52.4

45.1
51.8
48.8
53.6
58.5
61.8
73.0
79.7

Recent trends in Federal scientist
and engineer employment

' Excludes social scientists, psychologists, health personnel.
SOURCE: Total Federal employment, Budget of the United States, Special
Analysis I, 197 0 , table 1-4, p. 109; engineers and scientists, from NSF, Em­
ployment of Scientists and Engineers in the United States, 1 9 5 0 -6 6 , (NSF
6 8 -3 0 ), tables A1 to A4.

In 1966, 193,000 scientists, engineers and certain
health personnel were reported to be employed by the
Federal Government.70 Even on a more restricted defi­
nition, mainly excluding health personnel and the

69 See discussion in section on “Data Gaps in Federal
Employment Statistics.”
70 S c ie n tific a n d T e c h n i c a l P e r s o n n e l in th e F e d e r a l G o v ­
e r n m e n t , 1 9 6 6 , National Science Foundation (NSF 68-16,
1968), p. 1.




25

Table 14.

Total and Federal intram ural scientists and engineers selected years, 1951-65
[In thousands]

Year
1951......................................................................................
1953......................................................................................
1955......................................................................................
1957......................................................................................
1959......................................................................................
1961....................................................................................
1963......................................................................................
1965......................................................................................

Total
scientists
and
engineers
606.1
741.9
806.3
952.7
1,051.3
1,144.6
1,273.5
1,361.3

Federal
scientists
and
engineers

Percent
Federal

70.7
84.9
81.5
90.1
97.7
102.2
120.3
132.1

11.7
11.4
10.1
9.4
9.3
8.9
9.4
9.7

All
scientists
159.0
189.6
208.5
248.9
287.5
316.0
355.1
395.5

Federal
scientists
25.6
33.1
32.7
36.5
39.2
40.4
47.3
52.4

Percent
Federal
16.1
17.4
15.7
14.7
13.6
12.8
13.3
13.2

All
engineers
447.0
552.2
597.8
703.8
763.8
828.5
918.3
965.8

Federal
engineers
45.1
51.8
48.8
53.6
58.5
61.8
73.0
79.7

Percent
Federal
10.1
9.4
8.2
7.6
7.6
7.5
7.9
8.2

SOURCE: Employment of Scientists and Engineers in the United States, 1 9 5 0 -1 9 6 6 (NSF 6 8 -3 0 ).

of intramural employment fell from 12 to under 10
percent of total science and engineering employment
during the 1951 to 1965 period, despite very large
increases in the absolute numbers during the period.
(See table 14.) The decline appears to have slowed
in 1957, however, and the shares of Federal scientists
and engineers have since remained more or less stable.
A contributing factor may have been the relatively
slow rate of growth of intramural Federal spending on
research and development, which has been increasing
at roughly half the rate of total Federal support of
science and technology.71
Intramural occupational employment patterns

Employment of scientists and engineers in the Fed­
eral Government varies among occupations as a share
of total occupational employment. Table 15, which
shows the extent of such variation, was constructed
from NSF-BLS estimates of occupational employment
by sector.72 To maintain consistency with occupational
estimates developed in chapter IV, the estimates are
for 1965.
Engineers employed in the Federal Government
outnumber scientists in the ratio of 3.2. Flowever, only
about 8 percent of all engineers are Federal employees,
but approximately 13 percent of scientists in 1965 were
in the Federal service. In some particular occupations,
71 In the period 1955 to 1965, total Federal support of
R&D increased 477 percent, but intramural R&D spending
increased by only 245 percent, resulting in a reduction of the
latter from 37 percent in 1955 to 22 percent of Federal R&D
spending in 1965. National Science Foundation, Federal
Funds for Research, Development, and Other Scientific Activ­
ities (Vol. X V I , NSF 6 7 -1 9 ), table C-86, p. 216. A variety of
reasons could account for the shift in the composition of
federally supported R&D employment. A sensitive manpower
information system could detect some of these factors, such
as program developments affecting the functional require­
ments for scientists and engineers, or shifts in the structure of
Federal programs.
72 As reported in National Science Foundation, Employ­
ment of Scientists and Engineers in the United States,
1950-66 (NSF 6 8 -3 0, September 1968).
73 For example, see Budget of the United States, Special
Analyses, Fiscal Year 1969 (table F -l, p. 71).

26



the Federal share is more than twice the overall share
of all scientific occupations, as in the case of agricul­
tural scientists. Such differences undoubtedly reflect
differences in the scientific and technological require­
ments of different Federal programs. Accordingly, the
next section is an effort to distribute occupational
employment in the Federal Government by broad pro­
gram categories.
Government scientists and engineers
in Federal programs

Data on employment of scientists and engineers in
the Federal Government currently are available only
in terms of the agencies in which they work. To
convert agency employment of an occupation to a
program basis, each occupation was assumed to be
distributed among programs in the same proportions
as total white-collar employment in Federal agencies.
The distribution of white-collar employment among
programs, however, also had to be estimated. To make
a program distribution of employment it was assumed
that if an agency contributes directly to more than one
program, its white-collar employment is proportionate
to its current expenditure, exclusive of trust and sim­
ilar funds, in each program.
Table 16 shows how agency white-collar employ­
ment was distributed in program categories. The
employment data on which the percentages are based
is for October 31, 1967, because the U.S. Civil Service
Commission did not conduct a survey of Federal whitecollar employment in 1965. A finer departmental or
agency breakdown, such as is available for total civil­
ian employment in the Budget,73 undoubtedly would
produce somewhat different distributions among the
program categories, including fewer employees in the
“ Other Programs” category. Such a breakdown is not
available for white-collar workers or for any of the
scientific and engineering occupations, however.
Table 17 summarizes the program distribution of
scientists and engineers employed in the Federal Gov­
ernment in October 1966. The occupational data are

from a recent National Science Foundation report,74
based on information supplied by the U.S. Civil Serv­
ice Commission. The occupational classes used have
been adjusted, principally by grouping detailed occu­
pations, to conform to those used in the historical
employment series,75 and thus arc comparable with the
definitions of scientists and engineers in other sectors.
Before grouping, however, each detailed occupation
was distributed in accordance with the white-collar
employment distribution shown in table 16 and dis­
cussed previously. This method undoubtedly results in
some distortion from a true program distribution of
Civil Service scientists and engineers. Although no in­
dependent means of checking the validity of the result­
ing distributions is available, apparently the numbers
actually employed in the areas of national defense and
of commerce and transportation may be larger than
indicated, and the agricultural resources and health
and welfare areas may be smaller than these estimates.
In terms of relative orders of magnitude, total inhouse employment of scientists and engineers generally
is similar to the occupational pattern of federally
supported employment in other sectors. Employment
in national defense programs makes up nearly half of
the total for all such occupations. Defense employment,
furthermore, is more heavily concentrated among the
engineering occupations than among the scientific
specialties except for mathematicians and physicists. A
detailed breakdown of occupational employment by
program is shown in appendix E.
Data gaps in Federal employment statistics

From a methodological viewpoint, the degree of
detail on occupational employment in agencies or pro­
grams considerably exceeds the level available in other
sectors. There are no open-ended categories. Thus the
“other occupation” category for scientists and engi­
neers in the Federal service does not have the same
meaning as in the other sectors.76 Again, this is an
instance of data “ loss” resulting from the incompati­
bility of definitions and classifications of different Fed­
eral statistical systems.
Two data gaps limit the coverage and the analytic
utility of data on scientists and engineers employed
in the Federal Government. One is the problem of
estimating the uniformed scientists and engineers em­
ployed in the Armed Forces. The other is the issue of
the functional distribution of scientists and engineers
in all branches of the Government.
Military personnel working as scientists or engineers
may constitute a significantly large group of federally
supported scientific and technical manpower. Although
accurate and meaningful data on employment and
occupation are not available, the number of uniformed
personnel currently working as scientists and engineers
may approach 25,000. Recent estimates from the




Table 15. Total and intramural employment of scien­
tists and engineers, engineers, selected occupations, 1965
[In thousands]

Occupation

All
Federal
employees employees

Percent

Federal as
percent
of
total

All scientists and engineers...

1,361.3

132.1

100.0

9.7

Engineers, technical................................
Scientists....................................................
Chemists............................................
Agricultural scientists....................
Biological scientists........................
Geologists, geophysicists ...........
M athem aticians...............................
Physicists..........................................
Other scientists...............................

965.8
395.5
116.0
44.7
55.6
23.6
48.9
39.0
67.7

79.7
52.4
8.2
15.5
6.4
2.8
3.8
5.6
9.3

60.3
39.7
6.2
11.7
4.8
2.1
2.9
4.2
7.0

8.3
13.2
7.1
34.7
11.5
11.9
7.8
14.4
13.7

SOURCE: National Science Foundation, Employment of Scientists end Engi­
neers in the United States, 1 9 5 0 -1 9 6 6 (NSF 6 8 -5 0 , September 1 9 6 8 ). Tables
A -l to A l l .

national registers show nearly 6,000 scientists employed
in the Armed Forces in 1966,77 and 1,300 engineers in
the military and public health services in 1964.78
However, a special study, covering so-called engineer­
ing occupations, indicates that the four services in 1967
had manning requirements for nearly 18,000 advanced
degree holders, but an additional 7,000 with bachelor’s
and first professional degrees—mainly in the Air Force
— also may have been working in scientific and tech­
nical engineering assignments.79 On the other hand,
nearly 34,000 military personnel held scientific and
engineering degrees in 1967. Discussions with Depart­
ment of Defense manpower specialists indicate that
neither the Department as a whole nor the individual
services have the information capability for estimat­
ing scientist and engineer employment. There is no
uniform occupational classification scheme among the
services, and no means of converting the categories
used to a common scheme, principally because no
74 “ Scientific and Technical Personnel in the Federal Gov­
ernment, 1966,” Reviews of Data on Science Resources (NSF
68 -1 6, No. 14, April 1968), table A, pp. 14—16.
75 The detailed occupations included, excepting medical
scientists and the groupings follow those in National Science
Foundation, Employment of Scientists and Engineers in the
United States, 1950-1966, pp. 49-51.
76 “Other” categories in most employment surveys usually
reflect either specialized occupations too small in number to
justify separate classification or, more often, occupations
whose unique characteristics present unusual problems of
classification. Ambiguity of response may also contribute to
classification in this open-ended category. In contrast, the
“ other” categories used to distribute Federal Government
scientists and engineers in this study resulted only from the
exclusion of certain detailed occupations which, however,
have distinctive classification in the Civil Service scheme.
77 American Science Manpower, 1966, National Science
Foundation (NSF 6 8 -7 , Washington, D .C., 1967), tableA-8,
p. 67.
78 Engineering Manpower in Profile, Engineers Joint Coun­
cil (New York, 1964), table 5, p. 16.
79 “ Engineers in Uniform,” Engineering Manpower Bulletin
(Manpower Bulletin No. 9, December 1967).

27

Table 16.

Percent distribution of agency white-collar employment in the Federal Government, by program,
selected agencies October, 1967
[In thousands]

Department or agency

All
programs

National
defense

In te r­
national
affairs

1,926.1
24.5
629.9
53.5
85.2
24.8
9.5
96.9
7.0
14.5
29.2
6.7
116.2
828.2

32.8
—
98.0
—
—
—
—
—
100.0
—
—
—
—
—

1.8
99.0
—
—
13.0
—
—
—
—
—
—
—
—
—

All agencies.................................................
S ta te ............................................................................
DO D .............................................................................
In te rio r.......................................................................
Agriculture................................................................
Commerce..................................................................
Labor...........................................................................
HEW .............................................................................
A E C ..............................................................................
H U D ..............................................................................
N A S A ...........................................................................
T V A ..............................................................................
V A ................................................................................
Other agencies or departm ents..........................

Space
1.5
—
—
—
—
—
—
—
—
—
100.0
__
—
—

A gri­
culture
3.4
0.6
—
—
78.0
—
—
—
—
—
—
—
—
—

Natural
resources

Commerce
trans­
portation

Housing
develop­
ment

Health
labor and
welfare

Veterans
affairs

Other
programs

3.5
—
2.0
91.0
9.0
—
—
—
—
—
—
100.0
—
—

3.9
—
—
—
—
100.0
—
—
—
—
—
—
—
—

0.7
—
—
—
—
—
—
—
—
100.0
—
—
—
—

5.8
—
—
9.0
—
—
100.0
100.0
—
—
—
—
—
—

6.0
—
—
—
—
—
—
—
—
—
—
—
100.0
—

40.4
0.4
—
—
—
---—
—
—
—
—
—
—
100.0

SOURCE: U.S. Civil Service Commission, Occupations of Federal W hite-Collar Workers, October 31, 1967 (Washington, D .C ., 1 9 6 8 ), table E.

operational purpose would be served by such a system.
Since the primary duty of uniformed personnel is gen­
erally defined as a military function, less attention is
paid to the technical requirements of particular mili­
tary jobs with exception of those which may be unique
to the services.
The other gap in the data on employment of scientists
and engineers in the Federal Government is the lack
of a functional breakdown. Until recently, only very
crude estimates of the functional breakdown of Fed­
eral scientists and engineers have been made. In 1967,
the U.S. Civil Service Commission in cooperation with
the National Science Foundation instituted a func­
tional classification of scientists and engineers, using a
20-category code to identify various functions.80 Pre­
liminary results of this new reporting scheme indicate
that in May 1967 about 30 percent of scientists and
engineers in the Federal Government were engaged in
research and development. A final report of the results
is expected to be published in the near future.
Table 17. Civilian scientists and engineers in the Federal
Government, by program, October, 1966

Program

Scientists
and
engineers

Natural
scientists1 Engineers

All programs.........................................................

131,087

54,255

76,832

National defense..............................................................
International a ffa irs ........................................................
Space research and technology...................................
Agriculture and agricultural resources.....................
Natural resources............................................................
Commerce and transportation.....................................
Housing and developm ent............................................
Health, labor and w elfare..............................................
Veterans affairs................................................................
Other programs................................................................

61,492
3,097
13,646
15,605
16,097
9,098
554
5,803
1,527
4,168

14,468
2,338
4,769
13,202
8,352
4,208
5
4,346
1,021
1,546

47,024
759
8,877
2,403
7.745
1,890
549
1,457
506
2,622

1 Includes mathematicians.

28



Summary and evaluation

Since by definition and institutional relationship all
scientists and engineers employed in the Federal
Government are also federally supported, there is no
ambiguity in determining the relationship of their
employment to Federal spending. The effect, so to
speak, is 100 percent. The data source available for this
group, moreover, is detailed and amenable to con­
solidation into categories comparable with the occupa­
tional definitions applied to other sectors. At the time
of this study, data on the functional character of Fede­
ral employment were being developed. The absence
of data on the utilization of scientists and engineers in
the Armed Forces, on the other hand, is also of some
consequence.
The principal deficiency in the intramural sector,
however, is the absence of a program classification. A
crude rearrangement of agency employment in 1966
has been attempted, to show the relationship of science
and engineering employment to the broad goals of the
Federal Government for planning purposes. Indeed, it
is only in this sector that plausible distributions of
scientist and engineer employment both among pro­
grams a n d in sufficient occupational detail were pos­
sible. If similar and comparable constructions were pos­
sible for all other sectors, even at the broad level
achieved in the Federal Government sector, it would
be an important step toward a manpower information
system capable of measuring Federal influence.
80
A description of the functional classes and the reporting
requirements are in U.S. Civil Service Commission, F e d e r a l
P e r s o n n e l M a n u a l S y s t e m , F P M L e t t e r , No. 293-9, March
24, 1967.

Chapter VI.

T otal Federal Effects and Implications for the Scientific
and Technical Manpower Information System

The attempt to estimate employment of scientists
and engineers supported by Federal funds in each
of the various economic sectors has been completed.
Three tasks remain: (1) To consolidate the results,
bringing together intramural and extramural employ­
ment estimates to measure the total effect of Federal
spending; (2) to provide illustrations of some applica­
tions of the methods of estimate with data presently
available; and (3) most important of all, to evaluate
the scientific and technical manpower information
system and to recommend lines of improvement and
further study of the Federal effect. These are the ob­
jectives of this chapter.
T otal Federal effect on employment

Federal programs vary with respect to both their
shares of the total utilization of Government-supported
employment and the proportion of scientists and engi­
neers employed in the Federal service. Table 18 shows
the estimated employment of scientists and engineers
according to the source of support and the program
categories of those supported by Federal funds. Both
employees of the Federal Government (except military
personnel) and those employed in other sectors are
included in the estimates of federally supported em­
ployment. Although a little less than an estimated 30
percent of all scientists and engineers employed in
1966 were supported by Federal funds, the table readily
indicates that more than half of this number were em­
ployed in national defense programs. This proportion
is probably on the conservative side, since some scien­
tists and engineers classified in other areas, for ex­
ample, space research, are known to be working in
activities directly or indirectly related to defense. Finer
program detail and identification of expenditure and,
at least, intramural employment would have permitted
a more accurate distribution of employment.
Estimates of the proportions of scientists and engi81 For example, agricultural or the industrial and business
community. In such programs, a substantial part of the pro­
gram output is information rather than physical services.
82 See pp. 30-33.
83 A number of such nonprofit organizations are almost
completely dependent on government contracts and grants.
See chapter II, this bulletin, pp. 3-16.




neers employed in the Federal Government in each
of the program areas are also shown in table 18. Dif­
ferences in the extent of in-house employment among
the programs are quite marked. There is no easy gener­
alization about these differences, which range from
16 percent of program employment in space research
and technology to more than 90 percent in commerce
and transportation. The nature of the complementary
inputs used in the program, for example, experimen­
tal space vehicles, the nature of the program output,
especially the clientele directly served,81 the existance
of established government-owned facilities, such as
laboratories and agricultural experiment stations, and
the pace of legislative and environmental changes
in the area comprehended by the program, are un­
doubtedly among the factors contributing to such
differences. Direct and intensive study, is needed to
establish the particular factors involved and to evalu­
ate their respective influences. For the time being,
however, even estimates as crude as these have some
analytic value, as will be demonstrated in the next
section.82
Tables 19 and 20 are distributions of federally sup­
ported employment of scientists and engineers by sec­
tor as well as source of support. The employment
estimates in this instance are for 1965, instead of 1966,
because of the need to key the estimates of all other
sectors to the largest sector, private industry. Estimates
of Federal effect for that sector are based on the inter­
industry transactions approach, for which data were
available only for 1965. Table 20 again shows that the
overall Federal effect is nearly 30 percent of the total
of employed scientists and engineers. Although precise
data are not available, this estimated share probably
would not be significantly affected by the addition of
scientists and engineers in the armed services. Such
additional data, of course, would alter the proportions
of in-house employment and the occupational profile of
Federal manpower utilization; very likely the engineer­
ing occupations, mainly located in the defense pro­
grams, would show heavier concentrations of Federally
supported employment.
The percent of federally supported extramural em­
ployment, as shown in the tables varies between the
sectors, the largest extramural sectoral shares occur in
the universities and colleges and the independent non­
profit organizations.83
29

Table 18. Estimated Federal utilization of scientists and
engineers, selected programs, 1966
[In thousands]
All
scientists
and
engineers

Percent

All sources of support........................................

1,412.5

100.0

9.3

Federally supported........................................................
National defense..............................................................
Space research and technology...................................
Agriculture and agricultural resources.....................
Natural resources.............................................................
Commerce and transportation......................................
Health, labor and w elfare2............................................
Other Federal programs.................................................

392.8
209.1
83.0
20.8
17.9
10.0
33.7
18.3

27.8
14.8
5.9
1.5
1.3
0.7
2.4
1.3

33.4
29.4
16.4
75.0
89.9
191.0
17.2
51.4

Source of support and program

Percent
employed
in Federal
Govern­
ment

1 Excludes employees of the Post Office.
1 Includes National Science Foundation.
SOURCES: Tables 3, 5.

Two of every three scientists and engineers involved
in Federal programs are located in one or another of
the four extramural sectors, but, as earlier analysis of
federally supported R&D employment indicates, they
are apparently concentrated in a few program areas.
Data gaps affecting the functional and occupational
dimensions partially block further analysis of the total
Federal effect on the employment of scientists and en­
gineers. The blockage is partial because data deficien­
cies in these two dimensions vary among the sectors.
Occupational or field-of-science estimates of Federal
effect are more or less satisfactory for private industry,
universities and colleges, and the civilian branches of
the Federal Government. Such a breakdown, on the
other hand, is not possible for either the independent
nonprofit organizations or State and local govern­
ments.
In both of these sectors, there are no Federal expen­
diture data by field of science, even for R&D scientists
and engineers.84 Sectoral breakdowns of employment
of scientists and engineers in terms of the character of
the work are possible on a crude basis for all sectors,
but the estimated employment of R&D scientists and
engineers in the Federal Government is at present
dependent on an extrapolation of trends in R&D costs
per scientist-engineer man-year and Federal spending
Table 19. Estimated employment of scientists and engi­
neers, by source of support and sector, 1965
[In thousands]

Sector

All
sources

Federal
support

Other
support

All sectors..............................................................

1,361.3

393.4

967.9

Federal Government1......................................................
Other sectors.....................................................................
Private industry................................................................
Universities and colleges...............................................
Independent nonprofit organizations........................
State and local governm ent..........................................

132.1
1,229.2
960.9
171.1
14.0
83.2

132.1
261.3
169.8
68.3
10.7
12.5

967.9
791.1
102.8
3.3
70.7

_

1 Civilian only.
SOURCE: Scientists and engineers, all sources of support, and scientists and
engineers employed in the Federal Government are from National Science Founda­
tion, Employment of Scientists and Engineers in the United States, 1 9 5 0 -6 6 ,
NSF 6 8 -3 0 , table A - l. All other sectors based on derived estimates.

30



on research and development.85 The estimates of fed­
erally-supported employment by character-of-work
and by field-of-science, and in each case by sector
involved, are shown in tables 21 and 22. Regrettably
these two dimensions cannot be concurrently analyzed
in the framework of Federal spending and programs,
since there is ample evidence of field-of-science varia­
tion in cost per man-year.86
Little comment is required on the substantive aspects
of table 21. The principal difference between this table
and a similar table in chapter III is the addition of the
R&D scientists and engineers estimated to be employed
in the Federal Government. The estimates of federally
supported employment are derived from the ratios
developed in that chapter, but applied insofar as pos­
sible to 1965 data in order to maintain a rough tem­
poral consistency. Clearly, the influence of Federal
programs and funds on the utilization of scientists and
engineers engaged in research and development is
much greater than their influence on total scientistengineer employment.
The field-of-science distribution shown in table 22
requires more extensive comment. The consolidation
of the estimates of Federal use of scientists and engi­
neers into broad fields of science was necessitated by
the differences between employment and expenditure
classifications. Detailed estimates of occupational em­
ployment are available, or can be derived for the Fed­
eral Government, private industry, universities and
colleges, and for the independent nonprofit organiza­
tion sectors. However, only data for universities and
colleges on R&D expenditures are segregated by source
of funds and reasonably well matched to the employ­
ment data in each of the detailed fields of science.87
In all other sectors R&D expenditures are available, in
published form at least, for three broad fields of
science only. Expenditure data in all sectors for nonR&D functions are also available only on a broad field84 Tabulations of Federal expenditures on science and tech­
nology in the National Science Foundation series, Federal
Funds, show expenditures by field of science and by per­
former, but no cross-tabulation of these two dimensions. In
the case of private industry and the universities and colleges,
however, the surveys conducted in those sectors include suffi­
cient information on total as well as R&D spending by source
of funds to permit the estimates of employment developed in
this study.
85 National Science Foundation, Employment of Scientists
and Engineers in the United States, 1950-66, p. 51. Only
intramural spending provided the base and trend data.
86 For example, in the independent nonprofit organization
sector, annual costs per scientist or engineer man-year in
January 1965 varied from $19,300 for the field of mathe­
matics to $48,900 in the engineering sciences, as calculated
from data in National Science Foundation, Scientific Activi­
ties of Nonprofit Institutions, 1964 (NSF 67 -1 7, 1967).
87 It is assumed that the employment and expenditure data
in a given sector match, though it is possible that an indi­
vidual working as a physicist actually might be employed in
another field of science.

Table 20. Federally supported employment as a percent
of total employment of scientists and engineers, 1965
[In thousands]

Sector

Federally
supported

Percent
fed eral1

Percent of
total
federally
supported

All sectors..............................................................

393.4

29

100.0

Federal Government........................................................
Extramural sectors...........................................................
Private industry................................................................
Universities and colleges...............................................
Independent nonprofit organizations........................
State and local governments........................................

132.1
261.3
169.8
68.3
10.7
12.5

100
21
18
40
76
15

33.6
66.4
43.3
17.3
2.6
3.2

1 Civilian only.
SOURCE: Table 6.

of-science basis. Thus, the problem of data lies more
in identifying funding sources, expenditures and costs
than in the area of occupational employment per se.
As a consequence of the foregoing data problems,
the sectoral contributions to each of the fields-ofscience are not strictly additive. In addition, data for
the Federal Government sector relate to October 1966,
while the estimated contributions of the other sectors
are centered near January 1965. At most, it is possible
to say that relative Federal effect is least in the engi­
neering fields and greatest in the life sciences resulting
in about 40 percent of 1965 employment of scientists
and engineers in the latter.
Ideally, occupational detail to the level used in the
interindustry transactions approach should be avail­
able for all sectors. At present, however, that approach
provides only for estimation of federally supported
occupational employment in the private industry sec­
tor. Comparable occupational and functional detail—
and then only for the sciences, not engineering— is
available oniy for one other sector, the universities and
colleges. Should these difficulties be overcome, either
by developing consistent technical coefficients and cor­
responding occupational patterns for other sectors or
by developing and broadening the scope of surveys of
science and engineering manpower in other sectors,
the functional dimension of employment would be lost
in using that approach. At present, research and de­
velopment is treated in the input-output tables as a di­
rect sale from the various producing industries to the
88 For a more complete explanation of the method of treat­
ing research and development, see P r o j e c t i o n s 1 9 7 0 (BLS
Bulletin 1536, 1966), p. 38.
89 For example, in 1966, 41 percent of employed scientists
engaged in research and development held the Ph.D. degree;
only 29 percent of those whose highest degree was the bacca­
laureate were so engaged. Calculated from National Science
Foundation, A m e r i c a n S c ie n c e M a n p o w e r , 1 9 6 6 , (NSF 6 8 -7 ,
December 1967), table A -9 , p. 70.
90 This is not alone a matter of selective publication of
survey results, but a question of design. Lack of coverage of
costs and employment in the non-R&D scientific and engi­
neering work force, in most sectors covered by the NSF sur­
veys, is an example.




purchasers in the final demand sectors, such as the
Federal Government.88 Consequently, no distinction
can be made between personnel engaged in R&D and
those engaged in other types of work. This would not
be serious if it could be safely assumed that scientists or
engineers in any given occupational category are func­
tionally interchangeable. Limited information on this
point, however, suggests otherwise.89
A pplying the estim ates
The critical test of an information system is the
degree to which it can provide answers to significant
questions within the system’s domain. A few of the
manpower information systems considered in this study
apparently have been designed to meet specific needs,
such as the Economic Intelligence System sponsored
and managed by NASA and the Department of De­
fense. However, the majority of the systems are general
purpose systems, even though concerned in each case
with particular sectors. Consequently, they suffer from
the practical limitations of such systems, including a
lack of sufficient analytic detail,90 while providing gen­
eral answers to a variety of issues.
The information system synthesized in this study,
despite sharing the limitations referred to above can
answer some specific questions in the area of Federal
program planning. Three have been chosen as a test
of the data sources and the applicability of the ap­
proaches used. The answers provided are necessarily
broad, and moreover, there is no measure of the
degree of precision. Nevertheless, they may also serve
to illustrate further needs for change or improvement
in the information system. The questions to be an­
swered, in order of increasing specificity, are:
1.
Has utilization of scientists and engineers in
Federal programs increased more rapidly than total
employment in those occupations? Subsidiary to this
question are related questions such as the nature of
the change in the structure of the federally-supported
scientist and engineer work force, such as the division
between intramural and extramural employment, or
Table 21. Estimated employment of R&D scientists and
engineers, by sector and source of support, 1965
[In thousands]

R&D
scientists
and
engineers

Federally
supported

All sectors..............................................................

504.3

280.6

56

Federal Government........................................................
Private industry................................................................
Universities and colleges...............................................
Independent nonprofit organizations........................
State and local governments........................................

65.7
357.0
64.5
13.3
J3.2

65.7
157.1
46.4
10.1
*1.3

100
44
72
76
41

Sector

Percent
Federal

1 Civilians only.
2 Full-tim e equivalent employments
SOURCE: NSF, Employment of Scientists and Engineers in the United States,
1 9 5 0 -6 6 , table A -2, Page 2 2 . Table 11-4 ratios were applied to 1965 estimated
employment.

31

Table 22. Estimated utilization of federally supported
scientists and engineers, by field of science and sector,1
1965
Sector

All fields

Physical
Engi neering sciences2

Life
sciences

and engineers expect to be absorbed, or will they be
vulnerable to sustained unemployment?
F e d e r a l p r o g ra m s a n d g r o w th o f d e m a n d fo r sc ien ­

Answers to the first question raised
above were obtained by constructing a time-series built
on the budget model described in chapter II. The
employment estimates cover the period 1954—66, but
are cruder than those developed in the earlier chapters.
Data necessary to estimate employment in Federal pro­
grams by man-year expenditure estimates in particular
sectors are not available on an annual basis, although
for some sectors there are enough observations to
permit more refined annual estimates by interpolation.
Table 23 shows the trends in the growth of employ­
ment of scientists and engineers and in the level and
structure of federally supported employment in these
occupations. Four different years were selected to
exhibit these trends— the 2 terminal years, 1954 and
1966; 1957, which is the beginning for most sectors of
systematic and probably more reliable employment
data;91 and 1964, which represents the period just
prior to the Viet Nam buildup of defense program
demand. The period 1957—64 also spans the buildup
of the NASA manned lunar landing program.
Such a table permits analysis of changes in the pat­
tern of utilization of scientists and engineers and its
relationship to Federal programs. From the com­
pounded annual growth rates, it is apparent that fed­
erally supported employment over the period 1954—66
has been growing nearly a fifth more rapidly than total
employment. Growth has been especially rapid among
R&D scientists and engineers, whose total employment
growth rate substantially exceeds that of total science
and engineering employment. Between 1957 and 1966,
the proportion of R&D scientists and engineers rose
from 32 to nearly 37 percent of total scientist-engineer
employment. Of the increase in R&D scientists and
engineers— about 218,000— federally supported em­
ployment is estimated at 184,000, or about 84 percent.
Federally supported R&D employment was 98 percent
of the increase in total federally supported employ­
ment. The table also indicates a relative shift in
federally supported employment toward increased
utilization of extramural scientists and engineers.
Among R&D scientists and engineers in Federal pro­
grams, the rate of increase of those employed outside
of the Federal Government was nearly a third greater
than employment of so-called in-house personnel. By
1966, two of every three federally supported scientists
or engineers were employed in the extramural sectors.
The findings summarized above could be extended
to show the distribution of changes in Federal demand
tists a n d e n g i n e e r s .

Em ploym ent (in thousands)

1360.7

965.8

253.7

141.2

Percent of total fed erally supported em ploym ent
Federal G o v e rn m e n t3...............................
Private in d u s try ...........................................
Universities and co lle g es........................
In d ep en d en t nonprofit organ izations.
S tate-local g o v e rn m e n t4..........................

9.3
12.4
5.0
0.7

0.1

7.7
15.3

12.3
7.0
8.7
( 5)
( 5)

1.0
( 5)
( 5)

14.4
2.9
25.6
( 5)
( 5)

1 In c lu d in g em ployees o f fe d e ra lly funded research and developm ent centers
a d m in istered by extra m u ral sectors.
2 Inclu des m ath e m atics.
3 October 1 9 6 6 em p lo ym en t.
4 R & D personnel only.
5 Inclu ded in sector to ta l.
SOUR CE: Federal G overnm ent fro m N SF 6 8 - 3 0 , tab le s A - l to A l l .

between research and development and other activi­
ties. A satisfactory measure of the annual change in
Federal demand can also be compared with the
changes in the supply of scientists and engineers.
2. Given a major change in the structure of Fed­
eral programs but little change in the volume of ex­
penditure, what are the implications for science and
engineering manpower requirements? Specifically, how
many scientists and engineers, would be released by a
reduction in spending on the Viet Nam conflict, and
how many could be absorbed by compensatory spend­
ing in other areas? What economic sectors are likely
to be affected by postulated changes in program struc­
ture, and to what degree?
3. Given a change in the level of spending in a
particular Federal program, what changes would be
expected to occur in the detailed occupational require­
ments for scientists and engineers? At current rates of
change in employment and in the aggregate supply of
these occupations, can most of the displaced scientists
Table 23. Estimated total and federally supported em­
ployment of scientists and engineers, by sector and
character of work, selected years
[In thousands]

Sector

1954

1957

776.6

952.7 1,273.5 1,412.5

5.1

Federally supported................................ 209.7 238.7 431.3 426.8
79.7
90.1 126.4 134.1
Federal Government1....................
130.0 167.9 304.9 292.7
Extram ural........................................
R&D scientists and e n gineers.. . 237.1 302.3 490.5 520.5
Federally supported................................ 127.8 162.1 342.1 346.1
Federal Government1....................
30.1
31.4
62.0
66.8
97.7 130.7 280.1 279.3
E xtram ural........................................
Civilian labor force2............................... 63,643 66,929 73,091 75,770
Nonagricultural em ploym ent2............. 53,898 58,123 64,782 68,915
Professional, technical and kindred
w orkers3................................................. 5,588 6,468 8,550 9,322

6.1
4.4
7.0
6.8
8.6
6.9
9.1
1.5
2.0

All scientists and engineers...

1964

1966

Compound
annual rate of
growth, 1954-66
(Percent)

4.3

1 Civilian only.
2 16 years and over.

3 14 years and over.
SOURCE: Total employment ana total R and D employment, NSF 6 8 -3 0 , table
A -l; labor force, nonagricultural employment and professional, technical and
kindred, U.S. Department Labor, Manpower Adm inistration, Manpower Report of
the President, April 1967, tables A - l, A -9.

32



91
Data collected before 1957 on science and engineering
employment are probably less reliable than in subsequent
years. Hence, discussion is confined to the latter period.

Table 24. Supply and demand for scientists and engi­
neers, 1958-66
[Numbers in thousands]

Year
1958..........................
1959..........................
1960..........................
1961..........................
1962..........................
1563..........................
1964........................
1965..........................
1966..........................

Science and
engineering
degrees
conferred1
96.9
105.7
110.1
111.6
116.2
121.6
135.6
145.9
150.9

Change in employment
from previous year
All sources
of support
42.4
56.2
46.0
47.3
59.7
69.2
46.6
41.2
51.2

Federally
supported2
20.9
45.8
8.1
30.6
32.6
39.5
15.1
— 1.7
— 2.8

Federally
supported
as percent of
degrees
conferred
21.6
43.3
7.4
27.4
28.0
32.5
11.1
—

1 Bachelor’ s and first professional, masters’ and doctoral degrees in engineer­
ing, agriculture, forestry, mathematics, biological science, physical sciences,
and general sciences programs. Year indicates end of academic year.
1 Civilian only.
SOURCES: 1 9 5 8 -6 3 , degrees conferred from National Science Foundation,
Scientific and Technical Manpower Resources NSF 6 4 -2 8 , tables V -1 2 , V—15,
and V—18, November 1964; 1 9 6 4 -6 6 , from Office of Education, Projections of
Educational Statistics to 1 9 7 6 -7 7 (19 6 7 ed ), tables 20, 21, 22.

for scientists and engineers by broad program cate­
gories. Since the interest here is primarily in method
rather than substance, the necessary exercises were
not performed. On the other hand, the method does
not provide for estimates of employment changes in
the occupational specialties, or even by broad field-ofscience.
The pressures generated by changes in Federal pro­
grams and spending can also be evaluated by compar­
ing the year-to-year changes in federally supported
employment with the annual output of graduate scien­
tists and engineers. Table 24 shows the results of such
a comparison. The extfnt to which the annual re­
ported output of the universities and colleges repre­
sents the effective supply cannot be ascertained, since
some portion of the graduates at the bachelor’s, first
professional, and master’s degree levels continue their
formal education. Other science and engineering grad­
uates in recent years have entered the Armed Services
where, as indicated in chapter V, only about half (or
less) may perform scientific or engineering tasks. The
additional numbers who obtain scientific and engi­
neering jobs by informal means, such as on-the-job
training and upgrading, is practically unascertainable.
On the other hand, the year-to-year changes in em­
ployment are exclusive of the changes in requirements
resulting from retirement, deaths, or movement out of
scientific and engineering fields.
Relative to the annual output of formally educated
scientists and engineers federally supported demand
appears from the estimates shown to vary greatly from
year to year. On the average for he 9-year period
shown in table 24, including 2 years in which federally
supported demand showed an absolute decline, the
annual Federal demand equaled almost a quarter of
the earned degrees in the field. This proportion may
understate the Federal effect on the effective supply of
scientific and technical manpower, since substantial




and increasing proportions of bachelor degree recip­
ients in those fields continue their education instead
of entering industrial, academic or public employ­
ment.92 The decline in the last 2 years of the period
may reflect the reduction in the space research pro­
gram. Most of this reduction, in any case, occurred in
the extramural sectors, which experienced a reduction
of nearly 8,000 federally supported scientists and engi­
neers between 1965 and 1966, mostly in the non-R&D
extramural work force.
C h a n g e s in p r o g r a m s t r u c t u r e a n d s c i e n c e a n d e n g i ­

Major changes in the
structure of Federal programs may effect comparably
large short-term dislocations of manpower, even if the
total volume of Federal program spending (in con­
stant dollars) remains about the same. Scientists and
engineers released from Federal programs that have
been cut back will need to find employment in their
occupations either in the strictly civilian economy or
in expanding Federal programs. If, in fiscal terms,
such programs offset reduced spending elsewhere, will
their manpower requirements exceed or fall short of
the numbers of released workers? This section is an
effort to answer such a question.
Estimates of the savings resulting from a reduction
in Viet Nam military spending and proposals for com­
pensatory spending can be tested for their manpower
implications in science and engineering. These esti­
mates are based on those in the report of the Cabinet
Coordinating Committee on Economic Planning for
the End of Viet Nam Hostilities.93 The Cabinet Com­
mittee estimates a fiscal year 1972 peace-and-growth
dividend— a combination of reduced spending on Viet
Nam and normal revenue growth— at $22 billion.94
Since, however, there is no suitable method for pro­
jecting man-year costs of science and engineering em­
ployment, a substitute assumption is made that the
spending on Viet Nam in fiscal year 1967 becomes
immediately available for other uses. In addition, if it
is assumed that there has been no rise in expenditures
per scientist-engineer man-year between fiscal 1966
and fiscal 1967, the manpower implications of reduced
defense spending and spending on new initiatives can
be examined.
n e erin g m a n p o w e r r e q u ir e m e n ts.

92 This is evidenced in part by the general decline in the
proportion of bachelors’ and first-professional to all degrees
granted by institutions of higher education. According to one
special study of June 1958 degree recipients and candidates,
47 percent of baccalaureate degree recipients in the natural
sciences and mathematics and 22 percent of those awarded
first degree in engineering enrolled in graduate schools for
one term or more after receiving their degrees. National
Science Foundation, T w o Y e a r s A f t e r th e C o l l e g e D e g r e e
(NSF 63-26, Washington, D .C., 1963), table 16, p. 29.
93 In E c o n o m i c R e p o r t o f th e P r e s i d e n t , (January 1969),
pp. 181-212. The report is dated December 31, 1968.
94 Ibid., pp. 199-201.

33

A decline in defense spending of more than $22
billion in fiscal 1967 would have reduced employment
in the national defense programs by approximately
79,000 scientists and engineers. This estimate was
obtained by dividing the reduction in expenditures by
the implicit expenditure per man-year of defense
scientist and engineer employment— approximately
$276,000 in 1966 dollars. This employment reduction,
accepting the ratios of intramural to total federally
supported employment in defense, would consist of
about 23,000 scientists and engineers employed in the
Federal Government and 56,000 employed in other
sectors.
Let it now be assumed that $22 billion will be spent
to expand existing programs or to undertake “ new
initiatives” along the lines and in the magnitudes
indicated in the Cabinet Committee’ s report.95 Selec­
tion of the particular programs was governed by the
availability of appropriate man-year expenditure
ratios, and do not represent any preference regarding
social and political priorities.96 At the prices and costs
assumed to prevail in fiscal 1967, new programs and
program expansion in various nondefense areas would
require nearly 39,000 more scientists and engineers
than would be released by the reduction in Viet Nam
spending. Nearly 4,000 of this increase in require­
ments, or 10 percent, would be added to the Federal
Government’s payroll if current ratios of intramural
to total employment in each of the program areas
remain constant. Requirements for the economy as
a whole for scientists and engineers would exceed the
numbers released because of the so-called “ peace
dividend” by almost 55,000, if the assumed reduction
in taxes is converted into consumer spending. These
estimates of requirements, of course, are conservative,
since no allowance has been made to replace losses
95 Ibid., table 3, pp. 204—205.
90 Originally it was hoped to locate back-up data on the
industrial distribution of purchases in each of several new
program areas. Such data would have made possible a su­
perior estimate of the manpower implications, including occu­
pational detail, through use of the interindustry transactions
approach. Planning in most of these program areas, so far as
could be ascertained, has not yet developed such data.
97 In 1966, R&D funds from all sources— public and pri­
vate— were $5,460 per employee in the aircraft and missiles
industry as compared with $760 of company funds per em­
ployee in that industry. In nonmanufacturing industries, on
the other hand, the respective R&D costs per employee were
$410 and $120. National Science Foundation, Research and
Development in Industry, 1966 (NSF 68—20 ), table 56, p.
70.
98 Case studies of the labor market experiences of defense
workers have been conducted under the sponsorship of the
U.S. Arms Control and Disarmament Agency in the aircraft
and missile industries. The findings of these studies have been
summarized and subjected to intensive analysis in U SACD A,
Reemployment Experiences of Defense Workers, A C D A /F 113 (December 1968).

34



resulting from occupational mobility, retirement, and
death during the post-Viet Nam transition.
The arithmetic of this exercise is as follows:

Program

Total
Em ploym ent Employed
in
change
expenditure Cost per
Federal
(In
(In billion scientist
dollars) or engineer thousands) Government

Health, labor
and welfare .___ 15.1
Science and
space research
1.0
Natural resource
1.4
Development
Transportation .. ...... 1.0
Surplus or tax
3.5
reduction
22.0
All programs
Deduct:
Defense
Viet Nam
22.0

141.8

106.5

18.3

72.0

13.9

2.3

175.0
208.3

8.0
4.8

7.2
4.4

227.3
-

-1 5 .4
117.8

-5.1

276.0

78.9

23.2

38.9

3.9

Net increase in Federal program
Requirements ..
—-—-

27,1

Various objections can be raised against the results
of such an exercise. Most of the objections, however,
relate to data gaps rather than to the logic involved,
at least in the context of this study. For example, the
wide difference in expenditure per scientist and engi­
neer between the defense program and other programs
implies that the latter generally are more laborintensive. Although independent sources of data sup­
port this implication,97 no source, of course, actually
provides observed expenditure and related employ­
ment data on a Federal program basis. Objection also
can be made against the implicit assumption that the
manpower resources used in defense programs are
transferable to other types of programs. Differences
in the occupational and functional composition of
various programs, however, cannot be accepted a
priori as barriers to the mobility of scientists and
engineers. An information system which also provides
for measurement and appraisal of the sources of
supply of such manpower, by occupation, industry
and type of Federal program is clearly indicated.
E f f e c t o f c h a n g e s in p r o g r a m e x p e n d i t u r e . Changes
in levels of support in particular programs may have
significant economic and manpower implications,
especially if reductions take place. Although several
studies indicate that professional and technical
workers generally fare better than other occupational
groups in finding new jobs and maintaining pre-layoff
wage and earning levels, in addition to the social
costs, mass layoffs as the result of program changes
can still result in considerable personal and family
hardship.98 In general, although work force charac­
teristics play some part in the adjustment process, the
more important factors affecting the effect of layoffs
are controlable and permit advance planning to mini-

Table 25. Estimated employment of scientists and engi­
neers per billion dollars of expenditures in defense pro­
grams, private industry, and Federal Government, 1965
or 1966
Total
defense

P rivate1
Industry

All scientists and engineers...............

3,370

2,230

1,140

Engineers, technical...........................................

2850

1980

870

aeronautical...................................
chemical...........................................
c iv il....................................................
electrical..........................................
industrial..........................................
mechanical......................................
other..................................................

520
70
260
750
260
510
480

460
50
100
510
230
370
260

60
20
160
240
30
140
220

Natural scientists...................................

520

250

270

Chemists................................................................
Agricultural scientists.......................................
Biological scientists............................................
Geologists and geophysicists..........................
Mathem aticians...................................................
Physicists...............................................................
Other natural scientists....................................

130
10
30
30
110
130
80

80

Occupation

Engineers,
Engineers,
Engineers,
Engineers,
Engineers,
Engineers,
Engineers,

Federal
Government2

50
3

3

10
20
60
50
30

20
10
50
80
50

1 Department of Defense only; 1965.
2 Department of Defense and Federal Government scientists and engineers in
Atomic Energy Commission. Civilians only. October 1966.
3 Less than 10.

mize adverse employment and earnings effects."
Knowledge of the levels and occupational distribution
of the work force expected to be displaced by a pro­
gram cutback would be useful in such planning.
Occupational coefficients developed along the lines
of this study, can be used to illustrate the expected
effect of a given reduction in defense spending on the
employment of scientists and engineers. In that pro­
gram area, however, such coefficients are available
only for the private industry and Federal Government
sectors, although these two sectors make up the bulk
of defense employment in those occupations. In
practice, reductions in spending would probably be
governed by technical as well as political priorities.
Consequently, the application of a reduction in expen­
diture would be concentrated in particular sectors
rather than across-the-board, as assumed here.
Table 25 estimates occupational employment per
billion dollars for scientists and engineers employed in
the defense program in 1965. These estimates, as indi­
cated, exclude employment in universities and colleges,
independent nonprofit organizations, and State and
local governments. Military personnel working as sci­
entists or engineers also are excluded. However, given
the estimated expenditures per man-year and the
defense program structure prevailing in 1965 or 1966,
the derived coefficients for the two sectors imply dis­
placement (or growth) of about 3,400 scientists and
engineers for each $1 billion reduction (or increase)
99 Ibid., pp. 220-221.
100 In part, this is because the estimates of employment in
the private sector include indirect as well as direct employment.
No account has been taken of indirect employment generated
by the government sector per se, although it probably would
be quite small.




in defense spending. The two sectors would be affected
differently by changes in expenditure. The incidence
of a general reduction would fall much more heavily
on scientists and engineers employed in private industry
than on employees of the Federal Government,100
although not in each occupation.
Whether or not reductions in defense spending
result in serious unemployment and other hardships,
of course, depends on a number of factors. The magni­
tude of the reduction in spending, the rate of growth
of employment in other programs or economic sectors,
the rate of increase in the supply of the affected occu­
pations, and the general state of the scientist-engineer
labor market are among the more important factors
to be considered in efforts to plan for such contingen­
cies. This study, however, is limited to crude estimates
of a few variables, mainly the trends in employment
in the occupations and the corresponding occupational
employment coefficients in table 25. By dividing the
estimated average annual growth for the period 1950
to 1966 by the per-billion-dollar employment in each
of several occupations, the size of a reduction in
defense spending that could adversely affect employ­
ment and earnings opportunities can be inferred.
The results of such computations show that a reduc­
tion in the defense budget of approximately one
quarter of its size in fiscal 1966 would slow the growth
of scientist and engineering employment to nearly zero.
For engineers alone, a $20 billion reduction would
tend to freeze employment. In contrast, among chem­
ists, the reduction would have to be nearly $50 billion
to produce a similar result: for mathematicians the
equivalent figure would be a little more than $40
billion.
E v alu ation recom m endations
G e n e r a l c o n c l u s i o n s . The results of the exercises in
estimation of the effect of Federal spending on the
employment of scientists and engineers may be sum­
marized in five broad conclusions and some recom­
mendations.
The conclusions are:
1.
A limited capability exists for measuring the
employment effects of Federal programs in the field
of scientific and engineering manpower. Given certain
assumptions about the relationships between expendi­
tures and employment, it was found possible to achieve
plausible estimates of the employment magnitudes for
the larger economic sectors, along functional lines and,
to a probably lesser degree of accuracy, in terms of
broad program categories. Estimation within program
categories in terms of occupational specialization was
least successful. In general, as the level of detail
desired increases in any of the dimensions, the capabil­
ity of the data sources for measurement of manpower
tends to diminish.

35

Table 26.

Estimated percent of scientists and engineers on Federal work, by selected sources of estimates

Sector

Unit-cost
estim ate(a)

Input-output(d)

All functions and performers................................
N on-R&D......................................................................
R & D .............................................................................
Industry........................................................................
Federal Government.................................................
Colleges and universities........................................

28
12
49
44
100
72

30
( ‘)
( ')
( ')
( ‘)
( ')

BLS-STP(c)
( ')
(> )
( ‘)
45.6

C ensus-N S F(b)
(>>
( ')
( ‘)
44.0

—

—

( ‘)

C)

National
Engineers
Register(e)

National
Scientists
Register(f)

242.0
( ‘)
< l)

46.3
35.4
57.9
C)

—

N S F-IH E
Survey(g)
< ‘)
( ‘)

—

—
—

( ‘)

*76.6

1 Adjusted for not employed and no report.
2 1964.
3 1 9 6 4 -6 5 .
SOURCES: (a) Tables 11-4: (b) tables 11-4, 111—1; (c) U.S. Department of Labor, Bureau of Labor Statistics. Survey of Scientific and Technical Personnel,
1966; (d) National Science Foundation, Research and Development in Industry, 1966; (e) Engineers Joint Council, Engineering Manpower in P rofile, 1964; (f)
National Science Foundation, American Science Manpower, 1966; (g) National Science Foundation, Reviews of Data on Science Resources, (August 196 6 ).

2. The foregoing conclusion is closely related to
the general problem of measuring manpower require­
ments. Although measurement of Federal effect adds
another—and extremely important— dimension, the
basic problem is the measurement of the demand for
scientists and engineers.
Until concepts and techniques of measurement of
demand are further developed, the overall capability
of the science manpower information system will
remain weak.101 Measures of demand or requirements
for manpower must be directly related to measures of
output such as total employment or the output of
goods and services. In this study such a relationship
could be established with reasonable certainty only for
two broad aggregates of Federal program activity and
expenditure, and then only for one economic sector.
Estimates for other sectors at a gross level were made
by assumptions that are plausible but not verifiable.
3. Data problems in estimating the effect of Fed­
eral spending on employment in science and engineer­
ing were encountered mainly in the measurement of
output or expenditure. These problems, however, were
encountered not because of a paucity of financial and
expenditure data in most of the important sectors, but
because of the absence of information on man-year
expenditures specific to occupational groups, fields of
science or engineering, and character of work. Indeed,
rather than paucity, under present conditions there is
an effective “ loss” of manpower information in the
process of estimating the effect of Federal spending on
employment. Expenditures per man-year related both
to funding sources and to some dimensions of employ­
ment and utilization of scientists and engineers were
most adequate in the private industry and indepen101 A Study of Scientific and Technical Manpower, Com­
mittee on Science and Astronautics, U.S. House of Repre­
sentative, 86th Cong. 2d Sess., January 4, 1960, p. 33.
102 Joint Economic Committee, Subcommittee on Economic
Statistics, The Coordination and Integration of Government
Statistical Programs (90th Cong. 1st Sess., 1967), pp. 1-2.
A similar conclusion and recommendation was reached by the
Killian Committee. National Academy of Sciences. Committee
on Utilization of Scientific and Engineering Manpower, op.
cit., p. 14.

36



dent nonprofit organization sectors, less so or not at all
in the university and college and government sectors,
including the Federal Government.
Employment data, on the other hand, were reason­
ably adequate for most sectors, although the degree of
detail varies among the sectors. Occupational employ­
ment measures are satisfactory in the Federal Govern­
ment and private industry sectors, but only the latter
sector relates the functional and occupational dimen­
sions. A functional breakdown of employment of scien­
tists and engineers in the Federal Government,
expected to be available in the near future, should
augment capability to measure that dimension. Occu­
pational and functional measures of employment in
the State and local government sector are virtually
lacking. Scientists and engineers in the military ser­
vices constitute an informational void.
4.
For most practical purposes, measurement of
the Federal employment effect is also inhibited by sev­
eral conceptual problems. The most significant of these
is the lack of a systematic, rationalized and opera­
tionally feasible conceptual framework for relating the
physical services of Federal programs to measures of
employment. The inability to penetrate this problem
is a major failing of this study, although the general
direction of a solution has been indicated.
To only a slightly lesser degree, the lack of coordi­
nation and integration of statistics on manpower and
related analytical data is a conceptual problem in need
of solution. The difficulties encountered in interrelat­
ing various sources of data, including manpower data
purported to measure the same population of scientists
and engineers, underscores a major conclusion of the
Joint Economic Committee’s Subcommittee on Eco­
nomic Statistics.
. . . further significant improvements in our statistical
services depend upon a higher degree of integration and
coordination of our statistical programs. Indeed, there
are strong indications that this is the aspect of the sta­
tistical system where progress is needed most. . . .
Nothing less than the quality of our public and private
economic policies is at stake.102

The virtual absence for most sectors of information
about man-year expenditures and output of scientists

and engineers engaged in non-R&D functions is still
another conceptual weakness of the science manpower
information system. Although our estimates indicate
that the overall Federal impact among such scientists
and engineers is relatively minor, in the aggregate they
constitute more than 60 percent of total science and
engineering employment. A significant shift in the
structure of Federal programs would mean the need
for more information about this group of scientists and
engineers.
5.
Given the present state of information on the
employment of scientists and engineers, the preference
for one or the other of the estimating approaches used
in this study cannot be easily made. Either the budget
or method of estimation or the interindustry transac­
tions approach gives results that appear reasonable,
although to cover all major dimensions of scientist and
engineer employment in this study both approaches
were necessary. In addition, as table 26 shows, the esti­
mates of federally supported employment derived by
these approaches match comparable estimates based
on direct surveys in two of the larger sectors— industry
and universities and colleges.103 In part, this consis­
tency is attributable to the common control totals,104
although the effect of this factor on the estimates
cannot be measured. The budget approach at present
provides more insight into the sectoral and functional
dimensions of the Federal employment impact, but is
weak on the occupational and program dimensions.
The interindustry transactions approach, in contrast,
is somewhat stronger in these two respects, although
the estimates of occupational employment diverge—
substantially in some cases— from those provided in
the time-series data for 1965. An explanation of the
probable sources of these discrepancies is given in
appendix D.
Under more ideal data conditions, the choice be­
tween the approaches would depend on the purposes of
an analysis of Federal employment impact. Assuming a
continuing flow of current manpower and related
financial data, the budget-type approach would appear
to be somewhat better suited to the analysis of short­
term problems generated by changes in the levels of
Federal program support. For the projection of man­
power requirements in Federal programs, or for plan­
ning new initiatives and undertakings, the interindustry
transactions approach seems more useful. With this
approach better account can be taken of general eco­
nomic trends that may affect future programs, such
as the rate of technological change; indirect as well as
direct manpower requirements can be estimated; and
occupational employment patterns are readily avail­
able and adaptable to any given program “ mix.”
Recommendations for changes

1.

The general strategy for improving the capabil­




ity of the science manpower information system for
measuring Federal employment impact should be based
on the existing data sources and subsystems. No prob­
lems encountered in this study seems to compel con­
sideration of an entirely new system of data on the
employment of scientific and technical manpower. In
addition, the various surveys and subsystems of em­
ployment and financial data already serve a variety of
important interests and specialized groups of clientele.
It is doubtful that a single, integrated system could
meet all needs concurrently, including measurement
and analysis of manpower requirements in Federal
programs.
2.
Substantial improvements can be made in the
existing system, which will make measurement and
analysis of Federal program requirements practically
feasible. In general, much can be achieved by further
integration and coordination of the relevant financial
and employment systems. Careful attention should be
given to the following:
a. A scheme of classification of fields of science
and engineering common to all sectors and including
both financial and employment data. It should be pos­
sible, for example, to bridge expenditure data reported
in F e d e r a l F u n d s with employment and expenditure
data in the various sectoral surveys of employment of
scientific and technical personnel. Detailed occupa­
tional employment data should be collected in all
sectors, but within a common conceptual framework of
broader field-of-science groupings.
b. A common concept of reporting units for surveys
of employment and expenditure in the private industry
and business sectors. Although statistical arguments
can be marshalled to support the company or enter­
prise concept used in studies conducted by the Bureau
of the Census, technological and industrial develop­
ments strongly indicate that the establishment unit
used by BLS may be analytically more rewarding.
c. Data collection centered on a common reference
date. Employment and related data gathered by sec­
toral surveys should uniformly be related to the mid­
point of the Federal fiscal year, that is, December or
103 Note should be taken of the fact that the engineers’
register and the scientists’ register cover only their respective
populations, while our estimates and the scientific, profes­
sional, and technical survey figures are composites of scientists
and engineers. Using the input-output estimates for private
industry and Federal Government data, estimates for scientists
and for engineers are separately available. For those two
subjects combined, federally supported employment of engi­
neers is estimated at 24 percent : for scientists, the estimate is
27 percent.
104 Both the Survey of Scientific and Technical Personnel
in Industry and employment inverse generated from the
input-output tables used to ta l employment by industry as
controls and/or data sources. Estimates of total employment
are comparable to those collected in the establishment survey
reported in E m p l o y m e n t a n d E a r n in g s (BLS Bulletin 1312—

6 ).

37

January. Although a few industries are subject to
strong seasonal patterns of activity, and others such as
colleges and universities function on other than a
calendar-year basis, so far as scientific and technical
manpower is concerned the results would be little
affected by a common date of survey or reference.
3. To further the recommendations in (2) above,
serious consideration should be given to consolidation
of some statistical programs in the field of science and
technology. Specifically, the following possibilities
should be considered and carefully evaluated:
a. Merger in one program of the Survey of Scien­
tific and Technical Personnel, now conducted by the
Bureau of Labor Statistics and the Survey of Industrial
Research and Development, now conducted by the
Bureau of the Census. These two programs cover vir­
tually the same population in the largest sector of
federally supported employment of scientific and tech­
nical manpower. The principal barrier to consolida­
tion appears to be that occupational employment data
are better collected at the establishment level, and
financial and cost data are usually a central headquar­
ters responsibility. No substantial reason, however,
would bar collection of estimates of expenditures per
man-year at the establishment level. Differentiation of
these expenditures by function and field of science, as
well as by source of support as indicated below, would
help to develop useful employment estimators for vari­
ous types of Federal programs.
b. The Economic Information System, which now
covers a selected group of defense contractors, should
be terminated. This program, at least on the employ­
ment side, duplicates in part the coverage of the STP
program. Its main function appears to be to provide
information on the regional distribution of defense
contract employment, but in other important respects
—occupation, character of work, employment costs,
and especially, Federal support of employment— EIS
either provides no information or employment esti­
mates which may contain a substantial bias. The
regional distribution of employment in defense and
other Federal programs can easily be ascertained from
the establishment-based STP data.
4. Further improvements in the data base can be
made by efforts to fill gaps and lacunae in the present
system. Among those to which early consideration
should be given are:
a. Occupational and related employment charac­
teristics of engineers. At present, detailed occupational
specialization in the engineering profession is available
only from the decennial census and from the records
of the Federal Civil Service. It is recognized that defi­
nitions of engineering occupations may be more diffi­
cult than definition of the various scientific specialties,
although the latter are also not free of ambiguity. The
problem, however, is more a matter of the level of
education or training than a question of job content
38



and the nature of the services provided. Special studies
in selected industries may help to resolve the difficul­
ties in surveying engineering employment on an estab­
lishment basis.
b. Increased attention should be given to establish­
ing appropriate definitions and concepts for measuring
employment and requirements in the social and behav­
ioral sciences. These occupations were excluded from
consideration in this study, and financial and expendi­
ture data were adjusted (accordingly) to maintain
consistency with the current BLS-NSF definition of
scientists and engineers. Continued omission of the
social and behavioral sciences, however, may not be
warranted. Political and social developments indicate
that Federal and private activity will increasingly
develop toward greater utilization of those professions.
c. Classification and measurement of personnel
serving in the Armed Forces as scientists and engineers
should be explored. Scientists and engineers in uniform
may constitute an additional 4 to 5 percent of federally
supported employment of scientists and engineers, but
they are not accounted for in either the estimates of
total or Federal science and engineering employment.
An effort to measure such personnel in this study failed
mainly because the Department of Defense lacks an
information system adequate for measuring employ­
ment and utilization of uniformed scientists and engi­
neers. Differences in the classification schemes among
the four service branches appear to be a major obstacle.
5.
This study’s greatest obstacle and most conspic­
uous failure proved to be the development of employ­
ment estimates in a detailed program framework. A
solution to this problem will depend, first, on the devel­
opment of a conceptual framework and corresponding
measures of program activity or output. This part of
the problem will necessarily have to be solved coopera­
tively between the operating agencies, the Bureau of
the Budget, and the agencies responsible for the collec­
tion and analysis of data on scientific and technical
manpower. Once such a framework becomes available,
the second part of the problem will be to establish a
means of collecting and analyzing data on federally
supported employment of scientists and engineers.
It is proposed that benchmark employment and
related financial and output data in major Federal
programs be collected by periodic surveys. The collec­
tion of such data should be the direct responsibility of
the dominant agencies in each of the designated pro­
gram fields, for examples, the Department of Defense,
the Department of Health, Education, and Welfare
and so on. However, the surveys should be under the
general guidance and direction of the National Science
Foundation to ensure uniformity of manpower, fieldof-science, occupational and functional concepts as
well as sectoral coverage and survey dates. Should such
benchmark surveys be developed and established on a
coordinated basis with the present sample surveys,

further elaboration of the details of Federal support
would not appear to be necessary. Indeed, under such
ciicumstances, a case might be made for eliminating
this issue from employment surveys now on an annual
or biennial cycle, such as the STP. Experience in the
general behavior of occupational employment relation­
ships indicates that a survey cycle of 3, or even 5,
years may be sufficient to meet most of the needs of the
Government for such data.
6. Further experimentation should be made with
the interindustry transactions approach to investigate
the manpower implications of various Federal pro­
grams, especially in areas of proposed Federal activity.
The major effort needed here will be to estimate billsof-goods for each such program. Additional study also
may be needed to evaluate the validity of the occupa­
tional patterns.
7. For manpower planning in the fields of science
and engineering, there is a need for reliable estimates
of total requirements. In addition to changes in the




requirements generated by Federal programs, occupa­
tional coefficients reflecting changes resulting from re­
placement needs— retirement, out-mobility, and death
— should also be developed.
8.
Although not an objective of this study, for the
purposes of measuring Federal employment effect,
work on various aspects of the behavior of the
labor market for scientists and engineers should be
encouraged. Among the topics in need of further inves­
tigation and analysis are the industrial, occupational
and functional mobility of such workers; the sources
of supply of labor to the scientific and engineering
occupations; the effects of earnings on changes in the
supply and demand for scientific and technical man­
power ; and the relationship between training and edu­
cation and the utilization of scientists and engineers.
Such studies would not only be valuable in under­
standing the labor market, but essential for analysis
and interpretation of data on federally supported
employment in this field.

39




Appendix A.

A Note on Program Planning and Manpower Requirements

This study investigates the degree to which manpower requirements were being
estimated as part of the Planning-Programming-Budgeting System in various Federal
Government agencies. The investigation was neither exhaustive nor systematic;
approached were only a few agencies to which the researcher was directed because
these agencies were thought to have made some progress toward projecting man­
power requirements as an integral part of the PPBS exercise.1
PPBS is an effort to achieve efficiency in the allocation of resources in government
programs.2 Its proponents visualize its concepts and techniques as substitutes in the
public sector for the price and profit indicators used in investment decisions (or
long-term planning) in the private, market-oriented sector. Charles Schultze has
described PPBS as an effort to replace the practice of incremental budgeting or plan­
ning with a scheme that reviews and evaluates the costs and benefits of government
programs as a whole.3 PPBS was established on a Government-wide basis by a Presi­
dential directive in August 1965. Manpower requirements and costs are among the
more important elements to be evaluated under the system.
Our investigation of the manpower aspect of PPBS may be briefly summarized in
the following conclusions:
1. Manpower planning as an integral component of PPBS has made relatively
little progress in the agencies surveyed, although there is general interest in the
development of this component.
2. Assessment and costing of future manpower requirements in most Federal
agencies is limited to intramural employment. A number of agencies, for example,
the Bureau of Mines, concern themselves principally with manpower planning as
an internal management or personnel administration tool. Agencies which have
made more inclusive projections of professional and technical manpower require­
ments in their areas of responsibility vary with respect to the integration of projected
requirements into their PPBS programs. Mostly, these agencies have education and
training of scientific and technical manpower as a substantial responsibility— for
example, AEC, NASA, the National Institutes of Health, and the Department of
Agriculture. Only a few, notably AEC, have attempted to include projected man­
power requirements in their PPBS memorandums.
3. Lack of an adequate information system on manpower inputs, especially in
the extramural sectors, may be a significant barrier to progress in Federal manpower
planning under PPBS.
4. Another imporatnt limiting factor is a shortage of personnel in the Federal
Government with skill and experience to develop the operating concepts and
manpower information systems required for PPBS.
1 Among the agencies visited or contacted by telephone were the Department of Agricul­
ture, the Department of Defense, the Department of the Interior, the Department of Labor,
the Department of Health, Education, and Welfare, the Federal Aviation Administration, the
National Aeronautics and Space Administration, the Atomic Energy Commission, and the
Bureau of the Budget. Discussion in each case was quite general and open-ended, although
several agencies generously provided documents relating to their manpower information systems.
2 There is a growing literature on PPBS. In addition to the Government’s own studies (see,
for example, Budget of the United States, Special Analysis, Fiscal Year 1970, Pt. 4, pp. 2 5 3 273), there are several good introductory explanations of objectives and techniques. See David
Novick, ed., Program Budgeting, (Rand Corporation, Santa Monica, California, 19 65 );
Charles L. Schultze, The Politics and Economics of Public Spending (Washington, D .C ., The
Brookings Institution, 1968), especially chapter 2, pp. 1 9 -3 4 ; and Edward Sussna, “ Plan­
ning, Programming, and Budgeting Systems— A New Approach to Government Spending,”
Pittsburgh Business Review, May 1968.
3 Schultzer, op. cit., p. 23.




41

Appendix B.

Estimated total employment and employment attributable to Federal purchases of goods and services,
by source of procurement and in d u s try ,1 1965
[ In thousands]

Federally supporter
Industry titles

Total
employment

Department
of
Defense

Total

All other
Federal
Government

All industries................................................................................................................................................................

62,325.4

3,220.4

2,079.2

1,141.3

Livestock and livestock products........................................................................................................................................
Other agricultural products...................................................................................................................................................
Forestry and fishery products..............................................................................................................................................
Agricultural, forestry and fishery services.......................................................................................................................
Iron and ferroallory ores m ining.........................................................................................................................................

1,936.1
2,395.2
109.1
220.3
293.5

17.6
41.1
12.7
9.9
2.4

8.3
14.1
1.9
1.6
1.5

9.3
27.0
10.8
8.4
0.9

Crude petroleum and natural gas.......................................................................................................................................
Stone and clay mining and quarrying.............................................................................................................................
Chemical and fertilizer mineral m ining.............................................................................................................................

56.5
147.6
307.2
988.2
17.8

13.2
7.8
20.0
5.4
1.2

4.1
43
15.8
2.6
0.6

9.2
35
4.2
2.8
0.6

New construction......................................................................................................................................................................
Maintenance and repair construction................................................................................................................................
Ordnance and accessories....................................................................................................................................................
Food and kindred products...................................................................................................................................................
Tobacco manufactures............................................................................................................................................................

2,556.4
1,114.0
225.9
1,795.1
87.0

118.7
103.0
185.6
18.1
0.6

35.1
53.7
101.7
6.7
0.4

83.6
49.3
83.8
11.5
0.2

Broad and narrow fabrics, yarn and thread m ills .........................................................................................................

603.5
115 4
1 541.7
162.5
622 2

14.9
34
10 9
6.7
32 3

11.8
23
56
4.1
17 2

3.2
11
53
2.6
15 2

Wooden containers...................................................................................................................................................................
Household fu rn itu re .................................................................................................................................................................
Other furniture and fixtures..................................................................................................................................................
Paper and allied products, except containers.................................................................................................................
Paperboard containers and boxes.......................................................................................................................................

37.5
332.5
133.6
435.0
198.8

2.5
11.8
5.6
15.7
15.8

1.4
8.5
1.7
11.8
6.0

1.0
3.3
3.9
3.9
9.8

Printing and publishing..........................................................................................................................................................
Chemicals and selected chemical products.....................................................................................................................
Plastics and synthetic m aterials..........................................................................................................................................
Drugs, cleaning, and toilet preparations..........................................................................................................................
Paints and allied products.....................................................................................................................................................

1,046.1
407.3
194.8
225.5
64.9

33.7
36.9
23.1
8.8
8.6

29.5
15.7
6.6
3.2
3.0

4.1
21.2
16.5
5.6
5.6

Petroleum refining and related products..........................................................................................................................
Rubber and miscellaneous plastics products..................................................................................................................
Leather tanning and industrial leather products...........................................................................................................
Footwear and other leather products................................................................................................................................
Glass and glass products.......................................................................................................................................................

182.6
469.6
35.8
327.7
170.3

12.2
21.0
13.1
4.0
7.5

10.2
18.7
5.4
3.8
6.6

2.0
2.3
7.7
0.1
1.0

Stone and clay products.........................................................................................................................................................
Primary iron and steel m anufacturing..............................................................................................................................
Primary nonferrous metals manufacturing......................................................................................................................
M etal containers.......................................................................................................................................................................
Heating, plumbing and structural metal products........................................................................................................

440.2
917.8
357.6
70.6
432.7

17.5
65.5
60.9
16.4
11.7

14.1
51.6
33.7
1.0
10.9

3.4
13.9
27.2
15.4
0.7

Stampings screw machine products and bolts...............................................................................................................
Other fabricated metal products.........................................................................................................................................
Engines and turbines............................................................................................................................................................
Farm machinery and equipm ent.........................................................................................................................................
Construction, mining and oil field machinery.................................................................................................................
Materials handling machinery and equipm ent.............................................................................................................
Metalworking machinery and equipm ent........................................................................................................................
Special industry machinery and equipm ent....................................................................................................................
General industrial machinery and equipm ent.................................................................................................................
Machine shop products...........................................................................................................................................................

318.8
418.7
90.3
138.6
174.8
76.2
316.1
195.6
252.1
209.7

45.9
33.6
20.8
4.8
4.2
5.0
25.5
12.6
17.4
37.7

24.1
20.1
7.7
1.8
4.0
3.0
24.2
3.1
15.9
33.0

21.8
13.5
13.1
3.0
0.3
2.0
1.3
9.5
1.5
4.8

Nonferrous metal ores m ining............................................................................................................................................

Miscellaneous fabricated textile products........................................................................................................................

42



Appendix B.

Estimated total employment and employment attributable to Federal purchases of goods and services,
by source of procurement and in d u s try ,1 1965— Continued
[ In thousands]

;ederally supporter
Industry titles

Total
employment

Department
of
Defense

Total

All other
Federal
Government

Office, computing and accounting machines...............................................................................................................
Service industry machines..................................................................................................................................................
Electric industrial equipment and apparatus...............................................................................................................
Household appliances
Electric lighting and wiring equipm ent..........................................................................................................................

190.1
111.7
355.4
159.1
169.5

20.0
5.1
42.9
4.0
15.5

14.2
3.2
30.8
2.6
9.7

5.8
1.9
12.1
1.4
5.8

Radio, television and communication equipm ent..........................................................................................................
Electronics components and accessories.......................................................................................................................
Miscellaneous electrical machinery and supplies........................................................................................................
Motor vehicles and equipm ent..........................................................................................................................................
Aircraft and parts....................................................................................................................................................................

549.3
306.1
100.4
842.9
624.2

226.5
98.7
8.8
18.2
446.9

181.0
71.2
6.1
13.5
373.0

45.5
27.5
2.7
4.7
73.9

Other transportation equipm ent.......................................................................................................................................
Scientific and controlling instrum ents............................................................................................................................
Optical, ophthalmic and photographic equipm ent........................................................................................................
Miscellaneous manufacturing..............................................................................................................................................
Transportation and warehousing......................................................................................................................................

268.5
258.9
128.0
451.8
2,695.1

48.2
43.9
14.3
12.3
183.5

45.4
27.1
12.8
6.7
145.8

2.7
16.8
1.5
5.6
37.6

Communications: except broadcasting.........................................................................................................................
Radio and television broadcasting.....................................................................................................................................
Electric, gas, water and sanitary services.......................................................................................................................
Wholesale and retail tra d e ..................................................................................................................................................
Finance and insurance.........................................................................................................................................................

771.5
108.1
633.0
15,224.0
2,582.6

41.5
6.1
32.8
202.6
50.0

24.5
3.4
18.9
120.2
31.3

17.1
2.7
13.9
82.4
18.7

Real estate and rental.............................................................................................................................................................
Hotels; personal and repair services, except auto........................................................................................................
Business services.....................................................................................................................................................................
Research and developm ent...................................................................................................................................................
Automobile repair and services..........................................................................................................................................

767.0
2,747.6
2,161.6
101.7
493.3

12.8
82.2
122.7
52.5
13.5

8.1
36.5
70.5
38.7
6.3

4.7
45.7
52.2
13.7
7.2

Amusements..............................................................................................................................................................................
Medical, educational and nonprofit organizations........................................................................................................

706.0
4,851.0

13.5
176.1

8.7
95.3

4.8
80.8

1 SIC equivalents of industry groups defined in Bureau of Labor Statistics,




Projections 1970, B ulletin 1536 (Washington, D .C ., 1966) table IV—1 p. 5 8.

43

Appendix C.

Estimated federally supported extramural employment of scientists and engineers by State and local
governments, selected occupations, 1965

Occupation

Total employment

Federally supported

Percent
federally supported

All scientists and engineers.....................................................................................................................

70,795

7,653

Engineers, technical.................................................................................................................................................

59,754

6,147

10.3

Aeronautical.......................................................................................................................................................
Chem ical.............................................................................................................................................................
C iv il......................................................................................................................................................................
Electrical...........................................................................................................................................................
Ind ustrial............................................................................................................................................................
M echanical.........................................................................................................................................................
Metallurgical......................................................................................................................................................

627
2,633
22,618
8,766
4,955
9,155
1,612
862
8,849

94
357
1,476
999
594
1,048
228
232
1,167

15.0
13.6
6.5
11.4
12.0
11.4
14.1
26.9
13.2

Natural scientists......................................................................................................................................................

11,041

1,506

13.6

Chemists.............................................................................................................................................................
Agricultural scientists.....................................................................................................................................
Biological scientists.........................................................................................................................................
Geologists and geophysicists........................................................................................................................
M athem aticians.................................................................................................................................................
Physicists............................................................................................................................................................
Other natural scientists..................................................................................................................................

5,011
1,127
827
1,163
1,221
1,001
588

561
98
76
208
155
129
52

11.2
8.7
9.2
17.9
12.7
12.9
8.8

44




10.8

A ppendix D . Discrepancies between time series em ploym ent estimates
and occupational m atrix estimates
Discrepancies are noted between estimates of occupational employment in private
industry published in E m p lo y m e n t o f Scientists and Engineers in the U n ited States,
National Science Foundation, NSF 68—30 and those reported in the BLS occupa­
tional matrix used in chapter IV of this study. These discrepancies are shown in the
following table:
Interindustry
Matrix

Time
Series

Difference

.............. ............................................. ...............

863,657

785,800

77,857

Natural scientists__ ___ _________ __________ ...............
Chemists ..................................................... ...............
Agricultural scientists ........................... ...............
Biological scientists ........ ......................... ...............
Geologists and geophysicists ................. ...............
Mathematicians ............................. ........... ...............
Physicists ..................................................... ...............
Other natural scientists........................... ......... .

224,743
92,927
22,038
23,394
19,405
25,708
22,387
17,264

175,300
83,900
6,100
9,400
13,400
26,500
15,300
20,200

49,443
9,027
15,938
13,994
6,065
-7 9 2
7,087
-2 ,9 3 0

Occupation
Engineers

The discrepancies may occur for a variety of technical reasons, primarily dif­
ferences in industry and worker classification schemes used in the two studies.
Differences may also be due to the different survey methods. Time Series data are
based on employer enumeration; occupational matrix data was developed from
the U.S. Census of Population which is based on individual employer enumeration.
In the Time Series, all government employees are enumerated under the govern­
ment sector. In the matrix concept, government employees, except those in public
administration functions, are classified in the appropriate industry category for
activities commonly carried on also by the private sector. These differences can be
better illustrated perhaps by using agricultural scientists as an example.
About 6,100 agricultural scientists were employed in private industry in 1965
according to the Time Series (S T P ). The occupational matrix shows 22,038 agri­
cultural scientists, a difference of almost 16,000 (15,938). Over one-half of all
agricultural scientists shown in the matrix are in the agriculture, forestry, and fish­
eries sector of the economy. Most are in the forest products industry. Under the
matrix concept, the agriculture, forestry and fisheries sector included over 36,000
government workers in 1960. Some of these workers are agricultural scientists. In
the matrix, foresters and conservationists are included in the totals for agricultural
scientists. Therefore, some Federal “ agricultural” scientists show up as being em­
ployed in the agricultural, forestry, and fisheries sector under the matrix concept.
In the Time Series, on the other hand, all Federal Government workers are included
in the Federal Government sector. In 1965, about 15,500 agricultural scientists were
employed by the Federal Government. State and local governments employed an
additional 5,600 agricultural scientists.




45

Appendix E.

Estimated scientists and engineers employed in the Federal Government selected occupations, by Program
October, 1966

National
Defense1

In te r­
national
affairs

Space

Agriculture

Natural
resources

131,087

61,492

3,097

13,646

15,605

16,097

9,098

554

76,832

47,024

759

8,877

2,403

7,745

4,890

549

3,076
853
8,942
13,101
1,860
7,424
510
52
11,206

14
395
75
9
15

5,478
9
62
1,110
22
186
226

Engineers, other .” .............................

8,898
l'358
181606
18,323
2,089
8,796
811
434
17,517

4
247

Natural Scientists......................................

54,255

14,468

2,338

Mathem aticians . ' .............................
Physicists.............................................
Other natural scientists...................

8 135
14739
6933
2638
4J09
5 763
li;9 3 8

2 685
'214
932
738
2,708
4 228
2^963

156
1 767
336
28
8
6
37

Occupation
Total Scientists and Engineers
Engineers, Technical.................................

Engineers,
Engineers,
Engineers,
Engineers,

c iv il...................................
e lectrical.........................
industrial........................
mechanical.....................

' Excludes Selective Service.
2 Commerce and FAA.
1 M aterials engineering.




All
programs

Commerce2
and
transport

324
15
2,247
1,572
25
132
11

1,784

67
1,522
181
19
62
1
16
535

349
3,686
1,472
42
357
53
303
1,483

564

4,769

13,202

8,352

4,208

119
2
76

939
10 005
l|851
132
48
38
189

1 027
2 562
2 203
1311
146
201
902

351
1
19
236
340
745
2,516

643
381
3,548

Housing
Health
Education
and
development and Welfare

Veterans
affairs

Other
programs

5,803

1,527

4,168

1,457

506

2,622

104
24
13
34
1

230

44
946
111
2
78
2
1
273

330

20
7
410
671
96
488
7
58
865

5

4,346

1,021

1,546

1,956
153
1,176
128
164
83
686

624

278
34
14
65
38
33
1,084

292
6
1
20

1
4

326
10
48
13

Source: N ational Science Foundation, S cientific and Technical Personnel in
the Federal Government, 1966, NSF 6 8 - 1 6 , A pril 1968.

f t U . s . G O V E R N M E N T P R I N T I N G O F F I C E : 1 9 7 0 O ------3 8 4 .9 6 9




U.S. DEPARTMENT OF LABOR
BUREAU OF LABOR STATISTICS
WASHINGTON, D.C.

20212

OFFICIAL BUSINESS




POSTAGE AND FEES PAID
U.S. DEPARTMENT OF LABOR

THIRD CLASS MAIL