|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 303
APPENDIX
E
Projections of Demand and
Supply in Occupations
This appendix describes the purposes to which projections of worker
demand and supply are put and the characteristics these projections must
have if they are to serve these purposes. The various methods that have
been used for making projections are summarized, and the limitations of
each are discussed. The accuracy and limitations of the methods used by
BLS are also considered. A final section points to needed research and
suggests how the projections can best be understood and used.
PURPOSES OF PROJECTIONS
Economic history amply demonstrates the rise and fall of industries and
of occupations. Fluctuations in supply are most likely in those occupations
that require long training periods; this pattern occurs because the supply
of workers in a particular field that develops in response to market signals
may take years to get through the educational pipeline. Workers investing
time and money in education, employers concerned about the availability
of skilled workers, and a public interested in stability of wages and prices
and in getting services when they need them all have an interest in our
ability to anticipate changes in employment at least a few years in the future.
Projections may be made for a variety of purposes, among which are
the following:
· evaluating the adequacy of training or education programs in light of
the potential need for workers;
.
303
OCR for page 304
304
APPENDIX E
· estimating the feasibility of major proposed programs for government
expenditure (such as defense, public works, or facilities) in terms of the
availability of skilled workers to accomplish or staff them; and
· providing information on future employment opportunities for the
guidance of individuals choosing courses of education or training.
Examples of the first of these purposes include the insistence of Congress
.
.
that federally supported programs of vocational education and training of
the unemployed or the disadvantaged be planned with future employment
opportunities in mind. Similarly, the congressional consideration of such
programs as highway construction, community mental health facilities, and
the Strategic Defense Initiative ("Star Wars") programs to name a few-
included an inquiry into the availability of the highly skilled personnel that
are required for these projects. BLS launched its occupational outlook
research program in 1940 in response to the concern of guidance profes-
sionals that young people have adequate information with which to choose
among careers. The same motivation lies behind the efforts of state gov-
ernments to provide local projections of employment growth by occupation.
The rationale and assumptions underlying the projections may differ
depending on the purposes these projections must serve. Both vocational
guidance and evaluations of the adequacy of training programs to meet
future needs for skilled workers call for a realistic estimate of future eco-
nomic demand in the occupation. Estimating the feasibility of proposed
human resources programs, on the other hand, calls for the translation of
program goals whether or not they are realistic into personnel, and
adding to these requirements a realistic estimate of the demand for the
same types of workers in the rest of the economy.
On the supply side, vocational guidance purposes require projections of
the most probable worker supply in comparison with the economic demand.
These types of projections give the best picture of future employment
opportunities and the competitive situation in each field. For evaluating
the feasibility of a proposed program a forecast of the most probable supply
is also desirable; such a forecast would show whether the program can be
accomplished without special measures to attract more workers to the field.
For appraisals of the adequacy of present training programs, however, a
major element of the estimate of future supply the number of trainees-
is the quantity for which the exercise is undertaken, the unknown in the
equation, and there is no need to estimate it independently. One way to
look at the supply is to treat the losses to the occupation resulting from
death, retirements, and net mobility to other occupations as components
of"replacement needs." These replacement needs should be added to the
estimated growth of the occupation to get the total demand that has to be
satisfied by the flow of trainees.
OCR for page 305
APPENDIX E
305
In all of the above, we have discussed demand and supply as though
they were independent of each other; in fact, they are interdependent. An
increase in demand, by raising wage rates, elicits an increase in supply, and
supply also affects demand through its effects on wages and costs. Only
when there are constraints on demand, such as those imposed by the
technology of an industry (a steel mill cannot employ pastry cooks to roll
steel), or constraints on supply, such as limited educational facilities or
licensure, is the adjustment of worker demand and supply impeded.
Yet in those occupations that require long periods of education or train-
ing, it may take several years for the signal of an increase in demand to
fill the educational pipeline and produce an increase in graduates; it is for
these occupations that projections are particularly useful in facilitating the
adjustment of demand and supply. In the absence of projections, young
people have only the current market situation to guide them. If they react
strongly to a current shortage of graduates and high salary offers, they
may find that when they graduate, 4 years later, the field has become
overcrowded and salaries are dropping, conditions that may cause the
current year's entrants to avoid the field and precipitate a shortage 4 years
later. (The operation of"cobweb" patterns in the labor markets for highly
trained workers is demonstrated in a number of papers by Richard Free-
man.)
Projection Methods
A variety of methods have been used to project demand and supply.
The simplest has been to ask employers how many workers they expect to
employ in the future. This method appeals to many people as a straight-
forward way to tap the expert knowledge of the people who will make the
decisions. Yet it has produced such poor results that, after years of use, it
was abandoned early in the 1970s. Researchers found that few employers
could make the necessary projections of their sales and of technological
changes in their industries to develop good estimates of their future oc-
cupational requirements. (Indeed, most employers do not reply to the
surveys or give casual, off-the-cuff answers.) There is some tendency for
each firm to assume it will gain a larger market share; and an offsetting
tendency for companies to report that their personnel requirements 5 years
in the future will be the same as they are now. Finally, this method makes
no allowance for employment in new firms, which, according to some
research, are and will be a major provider of additional employment.
A second method that has been used to project demand and supply is
to extrapolate past employment trends in the occupation. This method is
based on the assumption that, whatever factors have operated in the past
will continue to operate. Unfortunately, history is full of instances in which
OCR for page 306
306
APPENDIX E
the employment situation changed radically as any buggy whip manu-
facturer will attest. Another deficiency of this method is its tendency to
treat the occupation as if it were in a vacuum and unrelated to other events
in the economy and in society. This limitation is illustrated by the attempt
in the early 1950s to extrapolate the growth of the engineering profession
by assuming that the exponential growth it had shown would continue.
With such growth, engineers would have exceeded the total labor force in
a short time, leaving no draftsmen to prepare drawings, no bookkeepers
to pay salaries, and no trash collectors to haul away their refuse.
A more sophisticated approach of late has been to associate the growth
of an occupation with causative variables that can themselves be projected
independently. For example, projections of the population by age have
been used to project the demand for teachers: the pupils in elementary
grades 6 years hence have already been born, as have those who will be
high school students 14 years hence. Changes in pupil-teacher ratios or
other strategic variables can be used to modify the results of these projec-
tions. Similar methods have been used to project the demand for physicians
(Graduate Medical Education National Advisory Committee) and nurses
(Western Interstate Committee on Higher Education). In some cases,
regression analysis has been used to measure the relative effects of the
variables on the result: the projection of employment.
This method may be used to yield estimates of the need for workers in
the occupation rather than estimates of the economic demand. If the rel-
evant ratios (e.g., the pupil-teacher ratio in the projection of employment
for teachers) are set at an ideal level that is nevertheless in line with what
experts in the field consider optimum, the resultant projections can be
viewed as projections of need. To the extent that they are based on current
ratios, which in turn reflect the current market situation, or are adjusted
for the future to reflect expected changes in the market situation, the
resultant estimate will be closer to an estimate of demand. Each approach
serves a different purpose.
The advantages of this approach over the simple extrapolation of past
trends are obvious. This method attempts to take into account some of the
strategic factors affecting employment. This is not an easy task, however;
demand in an occupation may be affected by technological changes; market
changes; the way consumers spend their money and the amount of income
they have to spend; government expenditures on education, health, high-
ways, and military material; and the capital expenditures of industry. Even
more important than these factors are the context of the growth of related
occupations and industries and the entire interwoven structure of the econ-
omy and of society. When we consider the combination of factors that
affect employment in health occupations—for example, the importance of
population trends, social trends, income and expenditure patterns, the
A,
OCR for page 307
APPENDIX E
307
science and technology of medical practice, the financing of medical care,
training and licensure, and the growth and attractiveness of alternative
occupations it becomes apparent that a comprehensive approach is called
for.
BLS, which began its research in this area in 1940 and issued its first
occupational projection ~ years later, at first tried the approach of studying
individual occupations but concluded that a comprehensive analysis was
needed. With support from the Veterans Administration (which wanted
information to help in the vocational choices of the millions who studied
under the World War I G.I. Bill), BLS published outlook information on
hundreds of occupations beginning with the first Occupational Outlook Hand-
book in 1949. The handbook has been a biennial publication since the mid-
1950s.
The broad occupational coverage, frequent publication, and wide use of
the projections (150,000 copies of each edition of the handbook are bought
by high schools, colleges, libraries, and community agencies) have had
important implications for the research program. Spreading research costs
over so many occupations has allowed a more comprehensive approach
than could be supported if interest were focused only on a few occupations.
The continuing research effort has led to accumulating experience, deep-
ening knowledge of each- occupation, and ongoing contacts with industry,
professional organizations, unions, and research institutes that are familiar
with each field. It has also permitted regular appraisals of the accuracy of
the projections and analyses of the possible reasons for errors. As a result
of this experience, new research programs and data collection systems have
been instituted, an example of which is the occupational employment sta-
tistics program begun in the early 1970s. Research is also being conducted
on tables of working life and on how people move from one occupation
to another, one purpose of which is to develop insight into some of the
elements of supply. Over nearly five decades of experience, occupational
research methods have changed and improved. The wide publication of
the results of such research has ensured that industry and professional
groups in each occupation have cooperated with the bureau in giving
information and carefully reviewing drafts. The use of research results in
schools and in vocational guidance undoubtedly influences the perceptions
of students about employment opportunities and the occupational choices
they make.
The basic approach followed by BLS is to estimate the employment in
each occupation that will be generated by economic demand. This estimate
is based on the demand for the goods or services the occupation provides,
which in turn is affected by the total spendable income available to con-
sumers and governments and by the changing patterns of what they spend
it on. These patterns are influenced by a wide variety of social and economic
OCR for page 308
308
APPENDIX E
factors, including changing tastes and styles, scientific discoveries and tech-
nological change affecting both what is produced and how it is produced,
the growth and changing composition of the population, taxation and
government expenditure policies ("guns or butter"), and what other coun-
tries are buying from and selling to us.
Producing such estimates is a formidable task, and predicting what will
happen in the future on so many different fronts is hazardous. Natural
disasters, social cataclysms, and business cycles are hard to predict. Yet
some of the changing factors move relatively slowly: there are lags between
scientific discoveries and the commercial exploitation of new technology,
between the initiation of a new style and its widespread adoption, between
the first Japanese automobile sold in the United States and iapan's sub-
sequent market success. These lags mean that useful projections can be
produced, provided certain conditions are met: (1) projections are confined
to a relatively short time horizon (about 10 years is enough to guide ed-
ucational policy and the career choices of individuals); (2) sets of alternative
projections are made to show the effect, for example, of alternative as-
sumptions as to the state of the economy or the business cycle; (3) events
are constantly monitored; (4) the projections are revised at frequent in-
tervals; and (5) continuous research is carried out on the accuracy of the
projections and on the adequacy of the methods.
BLS projections begin with the population projections made by Census
Bureau demographers. The census data give the number of consumers
and are a basis for BLS's projections of the labor force, which are based
on the trends in labor force participation by each age, sex, and race group.
From the total human resources thus projected, BLS estimates the gross
national product (GNP) that will be generated by making assumptions about
the growth of output per worker, changing hours of work, and the level
of unemployment that must be taken into account. To provide for the
uncertainties of the business cycle and to suggest the range of error to
users of the projections, three sets of projections are usually made: a "high,"
"moderate," and "low" forecast. The BLS assumptions about productivity,
hours, and unemployment are adjusted to yield an estimate of GNP growth
under these three conditions.
This somewhat simplified recital of an elaborate process may give the
impression of a mechanical juggernaut that rides roughshod over the entire
economy of 110 million people with all its complexity, nuances, and infinite
variety, mashing up the professions in which we are interested with masses
of coal miners, factory workers, and fast-food slingers. What has not been
said is that, at each step of the BLS process, special knowledge is introduced
whenever it is available, and the factors that enter into the calculations are
adjusted on the basis of information on developing and newly emerging
trends in the industry. In the most recent projections, for example, forecasts
for the mining industries took into account the latest petroleum import
OCR for page 309
APPENDIX E
309
analyses for the target year from the Department of Energy. Projections
for the machinery and computer manufacturing industries incorporated
analyses of the market situation and foreign competition. Projections for
health services considered such developments as cost containment policies,
the shift of many surgical procedures to physicians' offices and outpatient
facilities, the growth of new group practices and nursing and personal care
facilities, and the aging of the population. The bureau's extensive research
program on productivity and technological development yields insights
about the growth of overall productivity and of productivity in each in-
dustry, and the technological developments that affect the numbers and
kinds of occupations that are employed. The advantage of the compre-
hensive interactive approach is that special information or analyses on any
aspect of our complex economy can be inserted and implications drawn-
not only for a particular occupation or industry but for all of the others
as well.
In contrast, there is no unifying and systematic method for projections
on the supply side. The supply of workers in an occupation is affected by
two factors: (1) the inflow of trainees and of persons who acquire the
necessary skills by experience or work in related occupations or by the
study of related subjects and (2) the outflow of persons retiring, dropping
out of the labor force temporarily, dying, or transferring to other occu-
pations. Supply, of course, is also affected by the relative wages in this and
other occupations available to workers.
Projections of the number of college graduates in each field have been
published by the Department of Education; they were based on the pro-
jected population of the appropriate age and on mathematical extrapola-
tion of trends in the proportion of the population completing college. The
total degrees awarded were distributed by field (college majors) using math-
ematical extrapolation of past trends. Because there was no attempt to take
into account the effects of social and market factors on the decisions of
young people (except insofar as these factors were embodied in projected
past trends), these projections cannot be considered realistic. They do,
however, serve a useful purpose: they can be used to illustrate what would
happen to the outflow of graduates, an important component of the supply
of workers, if nothing happened to change the choices people make about
future careers. If such estimates are compared to independent estimates
of future demand or the requirements for attaining some national goal
such as a proposed community mental health program, a disparity between
the projected demand and the projected supply could point to policy mea-
sures that might be required to attain the goals (e.g., scholarships or other
inducements to undertake training for the occupations).
To determine the outflows and inflows that affect occupational supply,
BLS has pursued a number of avenues of research. For example, the
bureau has developed tables of working life (similar to life tables), showing
OCR for page 310
310
APPENDIX E
the annual attrition to a population at each age, to estimate losses resulting
from deaths and retirements. These age-specific rates were then applied
to the age composition of each occupation to estimate annual losses. The
tables take no account of differences in work life patterns among occu-
pations, however, nor of losses resulting from transfers to other occupa-
tions. More recently, studies have been made of transfers into and out of
occupations (Eck, 1984), and more complete attrition rates for each oc-
cupation have been estimated, including shifts into unemployment and
withdrawal from the labor force (either retirement or temporary with-
drawal).
BLS does not make projections of worker supply in occupations. It does
publish estimates of annual attrition or replacement rates. This information
is offered, together with the projected rate of growth in each occupation
and information on the unemployment rate, as clues to the employment
opportunities in the occupation. The inclusion of information on replace-
ment rates makes clear the point that projected growth alone does not tell
the whole story about employment opportunities.
Projections of employment demand for more than 300 occupations are
published in technical articles and bulletins. (The most recent projections
of general economic growth, industrial growth, and occupations were pub-
lished in the September 1987 issue of Monthly Labor Review). Brief articles
on each of about 200 occupations involving relatively long periods of train-
ing are published in the Occupational Outlook Handbook; profiles of the basic
numbers- employment, projected employment growth, unemployment rates,
replacement rates, and number of people completing training in a recent
year for about the same number of occupations are published in a series
of bulletins called Occupational Projections and Training Data, of which the
most recent (BLS Bulletin 2206) was issued in 1984.
State and Local Projections
In most states, employment projections for the state and major geo-
graphic areas within the state are made by state agencies, most commonly
employment security agencies, but sometimes universities or other eco-
nomic analysis organizations. Until a few years ago, there was a cooperative
federal-state relationship in this work, with BLS providing technical con-
sulting and sometimes tabulation work, but this cooperation has been dis-
continued as a result of budget cuts. The states are continuing their work,
however. The National Occupational Information Coordinating Commit-
tee, which is composed of representatives of the Departments of Labor
and Education, and its affiliated state occupational information coordinat-
ing committees give leadership to these efforts.
The states use varying methods, but they all have a few elements in
common. The national projections of the growth of industries are generally
OCR for page 311
APPENDIX E
311
taken as a framework, and past changes in each state's share of national
employment in the industry, together with projections of the state's pop-
ulation and any available input from the economic development agency of
the state, are used to project the industry's growth locally. Industry occu-
pational composition data from the Occupational Employment Survey (which
is conducted by the state agencies in cooperation with BLS) are used to
project employment by occupation. Replacement rates provided by BLS
are also published.
Evaluation of BLS Projections and Methods
Any evaluation of BLS's projection methods should begin with a look at
the scorecard that is, at how accurate the projections have been. The
bureau has published a number of evaluations of the accuracy of its pro-
jections, comparing them to actual employment in each industry and oc-
cupation when the target year's statistics become available. The two most
recent evaluations will be analyzed here: those for the 1960-1975 projec-
tions (Carey, 1980) and the 1970-1980 projections (Carey and Kasunic,
1982~. (No more recent evaluations have been published, in part because
changes in the classification system for occupations have made it difficult
to compare earlier projections with employment data gathered since 1983.)
Comparing a projection that purports to reflect demand, without regard
to supply, with the actual employment in the target year is not entirely
logical. It is justified only if one can assume that the supply will adjust itself
to match the demand, which does not always happen.
There are a number of ways to consider the accuracy of projections. One
is to compare the number of workers employed in the target year with the
number projected. The purpose of the projections, however, is to anticipate
change, to distinguish occupations that are growing rapidly or slowly, and,
especially, to perform the more difficult task of identifying occupations
that shrink while the economy as a whole is growing. Our evaluation will
therefore concentrate on how well the rate and direction Or change in em-
ployment was projected.
To begin with, we must look at the degree of variabilty in growth rates
among occupations. If growth rates vary in a narrow range around the
average, we would expect projections to be fairly accurate; if growth rates
are widely dispersed, the projections may be judged by more lenient stan-
~ . ~ . .
uarus. 1 able ~-1 arrays the actual changes In employment In occupations
included in the two BLS evaluation studies referred to above according to
broad groupings of their rates and directions of change as compared to
the average change for all occupations.
This little table could well have been made the preface of this paper: it
powerfully demonstrates the variability of occupational change, the risk
undertaken by anyone who invests in long and expensive training for an
OCR for page 312
312
.
APPENDIX E
occupation, and the difficulties of the forecaster. In a 10- or 15-year period
when the average occupation grew by about 30 percent, between one-fifth
and one-third of the occupations under study actually experienced declines
in employment. The number of occupations that grew at a rate triple the
average was about the same as the number that grew at a rate less than
the average. There was virtually no clustering around the average. Ob-
viously, occupations are highly volatile in their rates of employment and
subject to diverse economic forces.
An evaluation (Goldstein, 1983) of how well the BLS projections for
these occupations succeeded in predicting the actual changes shown above
concluded that, first, users of the projections had some warning of the
declines: 5 of the 16 occupations that declined from 1960 to 1975 had been
predicted to decline, and small increases of less than the average had been
predicted for the other 11. In the 1970-1980 period, 6 of the 20 occu-
patians that declined had been projected to decline, and small increases of
less than the average had been projected for 7 more.
Second, did the projections identify the occupations that were growing
rapidly and that needed special attention in planning training programs?
In the first period, 21 occupations grew at more than twice the average
rate; 15 of them had been projected to grow that fast. In the second period,
14 occupations grew at more than twice the average rate, but in only 2 of
them had such growth been projected.
Taking all of the projections together, how close did they come to the
actual employment changes that occurred? Going back to the class intervals
shown in Table E-1, we might say that if the predicted change was in the
TABLE E-1 Growth Rates in Employment in Occupations with
Increases and Decreases in Emolovment. 1960- 1975 and 1970- 1985
Item
1960-1975 1970-1985
Average (weighted) change (in percentage) for
all occupations
Total number of occupations compared
Occupations with declines in employment
Occupations with increases in employment
Below average (more than 10 percent below
the average)
About average (within 10 percent above or
below the average)
Somewhat above the average (between 10
percent above the average and twice the
average)
Twice to triple the average
More than triple the average
32.6
76
16
60
11
17
11
11
10
28.9
64
20
44
0
9
11
5
9
SOURCE: Carey (1980); Carey and Kasunic (1982).
OCR for page 313
APPENDIX E
313
TABLE E-2 Projected and Actual Employment Changes (percentage)
for Six Health Occupations, 1960-1975
Occupation
Projected Actual
Nurses, professional 73.5 68.5
D. . .
~et~t~ans and nutritionists 35.1 44.6
Optometrists 17.6 10.0
Attendants, hospital and other institutions 140.7 122.4
Dentists 43.S 23.1
Physicians, medical and osteopathic 66.7 40.2
SOURCE: Carey (1980); Carey and Kasunic (1982).
same interval as the actual change, it was on target. For the first period,
40 percent of the predictions were on target; for the second, 33 percent
were on target. If we consider as reasonably close those predictions that
were in the class intervals adjacent to the actual change, we find that 40
percent of the predictions in the first period and 27 percent of those in
the second period were reasonably close. By these standards, which are
perhaps somewhat lenient but whose leniency is justified by the variability
of economic employment changes, we would consider 80 percent of the
projections in the first period and 60 percent in the second to be either
on target or reasonably close.
Another question that must be answered is: Were the errors biased so
that projections were consistently too high or too low? Of all the projections
in the first period that were not on target, one-third were too low; in the
second period, roughly half were too low. Thus, there is some evidence of
a pessimistic bias in the second period.
Our concern in this report is somewhat more narrow, however. We must
thus consider how well the method predicts the growth of the the allied
health occupations. It is a reasonable hypothesis that the economic, tech-
nological, social, and institutional factors that are peculiar to the health
industry and its occupations may make the general projection method used
by BLS inappropriate for use in these fields.
The evaluation studies we have cited do not include many of the allied
health professions, largely because they included only occupations for which
the statistics were comparable over the 10- or 15-year spans between the
original projections and the target years. The comparison data needed for
the allied health professions, with their dynamic changes over recent de-
cades, are not available. Yet we can still test the hypothesis of peculiarity
with evaluations of the accuracy of the projections for other health occu-
pations (Tables E-2 and E-3.
It appears that the projections captured the general magnitude of the
employment changes in these fields rather better than they did for all of
the occupations evaluated earlier in this section, although one could wish
OCR for page 314
314
APPENDIX E
TABLE E-3 Projected and Actual Employment Changes (percentage)
for Five Health Occupations, 1970 - 1980
Occupation Projected Actual
Optometrists 20.0 19.4
Osteopaths 43.7 39.3
Physicians and surgeons 48.5 43.3
Registered nurses 42.7 59.9
Dentists 32.0 22.3
SOURCE: Carey (1980); Carey and Kasunic (1982).
for more accurate projections for dentists and physicians in the first period
and for nurses in the second. From these data, the hypothesis of peculiarity
of the health fields is not supported. Let us turn, however, to some of the
aspects of the projection method that raise questions or present problems.
Demand or Requirements?
In traditional economic analysis, demand and supply are equated at a
price or wage. Yet there is no explicit evidence of this process in the BLS
projection methods. Instead, the employment estimates for future years
may be seen as requirements generated by the levels of production or
services that the projected economic changes will engender. Indeed, chang-
ing relative prices throughout the system could change the projected eco-
nomic relationship—for example, in tracing the demand for raw materials
generated by the production of finished goods. However, the adjustments
made at various steps in the projection process to introduce technological
change and changes in markets and foreign trade have the effect of in-
serting price and market changes into the system.
At the end of the process, it is true that there is no systematic attempt
to modify the employment estimates for each occupation by a consideration
of supply. Indeed, as the forecasters lack projections of supply, this cannot
be done. The projections of occupational employment will be consistent
with actual employment in the target year only if the supply of trained
workers (perhaps forewarned by publication of the estimates or, in the
1960s, responding to policy measures that were designed to raise supply
to meet increased demand resulting from new entitlement programs) ad-
justs to the employer's requirements. Although the projections are not true
estimates of demand in the sense of traditional economic concepts, they
do come close to the goal of a realistic estimate of the number of jobs that
will be offered, as distinct, for example, from an estimate of ideal needs.
Occupational Composition of Industries
Evaluations of the accuracy of the projections made by BLS staff have
concluded that the subject industries' total employment was more accurately
OCR for page 315
APPENDIX E
315
projected than was employment by occupation. From the foregoing dis-
cussion, we might suspect that the lower accuracy of occupational projec-
tions may have resulted from the fact that demand had not yet been
confronted with supply. If it had. a different level of emolovment would
have emerged.
1 J
Less accuracy could also have been the result of the quality of the oc-
cupational data; until recently the only reasonably complete source of data
on the occupational composition of each industry was the decennial pop-
ulation census. In household surveys such as censuses, people report their
occupation by whatever name they have to describe it, telling the census
enumerator briefly what activities they perform. These reports are classified
by census clerks into the approximately 400 occupations the census tabu-
lates. There is potential error first in the respondent's report: some people
overstate their occupational status, as is evident from independent data.
Second, the census clerks do not always have enough information to classify
the occupations correctly; terminology varies across the country. (The same
comments apply to another source of occupational employment data, the
Current Population Survey, which is conducted by the Census Bureau. The
occupational estimates of this survey are based on a smaller sample than
those in the population census and thus have larger sampling errors and
somewhat less occupational detail; however, they are available annually.)
To improve the accuracy of occupational composition data, BLS initiated
an Occupational Employment Statistics (OES) survey, early in the 1970s in
cooperation with state agencies. Employment by occupation is collected
from employers by means of a separate questionnaire for each industry
that lists the occupations found in that industry. The questionnaire contains
brief definitions of the occupations that have been worked out in consul-
tation with employers to ensure understanding and accurate reporting.
The sample plants in the survey are chosen to represent all size classes in
the industry and to yield accurate estimates.
The survey is limited to wage and salary workers in each industry; BLS
adds the self-employed in each occupation using data from the Current
Population Survey.
Because it is based on reports from employers, the OES counts each
worker more than once if he or she has more than one job at a time. This
practice introduces a small inaccuracy in the occupation employment es-
timates; in the series of surveys of dual job-holding that was made from
1958 to 1980, the number of persons with more than one job averaged
about 5 percent of the total employed the exact figures were 6 percent
for men and 3 percent for women.
.
The BLS estimates count workers whether they work full-time or part-
t~me and do not distinguish between these two categories. Thus, in any
occupation, there could be many part-time workers in the figures. In 1986,
18.7 percent of all persons at work were working part-time 5.3 percent
OCR for page 316
316
APPENDIX E
for economic reasons (no full-time work was available or they had been
temporarily assigned to part-time work) and 13.4 percent because they
preferred part-time work. There was more part-time work among women-
27.5 percent of women worked part-time, 6.5 percent for economic reasons
and 21 percent voluntarily. The incidence of part-time work varies among
occupations, in the occupation group "technicians and related support
personnel," in which many allied health professions are included, 12.9
percent of workers were part-time (2.2 percent for economic reasons):
among women technicians, 20 percent were on part-time (3.4 percent for
economic reasons). (All data in.this paragraph are from the Current Pop-
ulation Survey.) There are therefore fewer FTE (full-time equivalent) jobs
than the number of people employed in an occupation implies.
The definitions of each occupation worked out for the OES were, as
noted above, designed in cooperation with employers to facilitate reporting.
The definitions must be understood within the culture of each industry
and must be consistent across industries so that the employment estimates
for each occupation are additive. This qualification, however, may not
always provide the nuances in definition that professional societies con-
cerned about qualifications, licensure, and similar matters would like to
have. Appendix C lists allied health occupations and related occupation
definitions.
We have suggested two reasons for the lower degree of accuracy of the
occupational employment projections compared with those for industry
employment: (1) the demand projections are not tested against occupational
supply and (2) the basic data on the occupational composition of industries
used in past projections were inaccurate. We should consider a third reason:
the way in which occupational composition is changing is not well under-
stood, and the adjustments inserted into the system to allow for the effects
of technological and other changes thus are not adequate.
The theory underlying the use of occupational composition data in fore-
casts is that the technology of each industry and the way it does its business
calls for a unique mix of occupations. In a gross sense, this is certainly true:
pastry cooks are not employed in steel-rolling mills. But there could be
differences among plants in the same industry that result from differences
in processes, in equipment, in the way the work is organized, and in the
local supply of trained workers and the extent to which less-trained workers
are substituted for them. For those familiar with hospitals and other health
service institutions, there is no need to belabor the point that occupational
composition may differ from one to another for many reasons.
When the acting commissioner of the Bureau of Labor Statistics first
testified before Congress on the request for funds to conduct occupational
outlook research, he stated that the research would consider the occupa-
tional composition of the most technologically advanced plants in each
OCR for page 317
APPENDIX E
317
industry for clues as to the way composition would be changing. Now,
nearly half a century later, this kind of analysis is made possible for the
first time by the OES. Not only are the occupational statistics better but
the collection of reports from individual plants offers a potential that has
never before been available except from a few industry wage surveys: The
chance to analyze why the occupational composition differs among plants
in the same industry and how it is affected by the size of the plant and by
new technology- analyses that may lead to better projections of occupa-
tional employment.
Staffing of the Projections Research
The number of BLS occupational outlook research staff has been re-
duced over the past few years as a result of budget cuts, and the burden
on individual staff members has therefore increased. With some 200 oc-
cupations to cover with articles in the Occupational Outlook Handbook, staff
are spread thin. Nevertheless, when the National Academy of Sciences staff
visited them to discuss their projections, it was found that no fewer than
four economists were working on health occupations. They were in touch
with developments in their fields and in the health care industry generally
and were familiar with the issues and findings of recent studies.
Use of the Projections and Further Research Needs
It should be apparent that forecasting for years in advance is always
hazardous and that this truism applies particularly to employment by oc-
cupation. While there is always hope that the data and methods will improve
in the future, the best we can realistically expect is that the degree of error
will be somewhat reduced. The user of projections must bear this in mind
mar] take them tic oniv rush indications of the direction and general
~,,~ am. _ =,,_,,. rev ~ , , ~ ~ C,
magnitude of changes.
Of the projection methods we have reviewed, that of the Bureau of
Labor Statistics appears to be the best in its ability to take into account
multiple factors. BLS staff have been doing such projections continuously
for a long time; they have accumulated experience, knowledge, and contacts
in each field; and they check their errors and are innovative in improving
data collection and analysis methods.
For any projections of employment in the allied health professions the
Institute of Medicine committee would be well advised to build on the work
BLS has done not necessarily to accept the projections without question
but to take advantage of the analysis of the framework of the U.S. economy
within which the health industry operates, and to examine the assumptions
and judgments made by BLS staff in the health fields, modifying them if
necessary. Our discussions with BLS staff made it clear that they are ear-
OCR for page 318
318
APPENDIX E
nestly searching for understanding and would welcome any insights that
would improve their projections.
Before we can have any assurance that occupational supply can be un-
derstood or projected, more research needs to be done on occupational
mobility and the factors that determine how people shift employment among
occupations. The same may be said about the factors affecting occupational
choice.
On the demand side, the weakest link has been in converting projections
of employment by industry, which have a fair degree of accuracy, into
projections by occupation. Now, however, analysis of the factors that affect
the occupational composition patterns of industries can be performed be-
cause for the first time we have occupational data for individual plants.
BLS practice in publishing its projections has been to issue only 10-year
or longer projections (without the intermediate years). Yet intermediate-
year projections are likely to be more accurate because they are closer to
what we now know; in addition, they are useful for many purposes. They
also lend themselves to more frequent evaluations of accuracy, a practice
that, if adopted, would enable BLS to correct its more distant projections.
REFERENCES
Carey, M. L. 1980. Evaluating the 1975 projections of occupational employment. Monthly
Labor Review 1 03June): 1 0-20.
Carey, M. L., and K. Kasunic. 1982. Evaluating the 1980 projections of occupational
employment. Monthly Labor Review 105(July): 22-30.
Eck, A. 1984. New occupational separation data improve estimates of job replacement
needs. Monthly Labor Review 107(March):3-10.
Goldstein, H. 1983. The accuracy and utilization of occupational forecasting. In Respon-
siveness of Training Institutions to Changing Labor Market Demands, R. E. Taylor,
H. Rosen, and F. C. Pratzner, eds. Columbus, Ohio: The National Center for Research
in Vocational Education.
Representative terms from entire chapter:
occupational employment
