View original document

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

Employment in the frrst half:
robust recovery continues
Employment surpasses the levels
posted before 1981-82 recession
as rebound completes sixth quarter;
June jobless rate of 7.1 percent
is 3 .6 points below recession high
RICHARD M. DEVENS, JR.

Employment grew throughout the first half of 1984. as a
very strong cyclical recovery continued through its fifth and
sixth quarters. Unemployment. after posting declines in January and February, was essentially unchanged through April
before dropping sharply in May and June. In June, the
overall unemployment rate (including the resident military
in the labor force) was 7.0 percent. and the unemployment
rate for civilian workers was 7.1 percent.
By June, total employment. as measured by the household
survey, 1 and non farm payroll employment, as measured the
establishment survey, 2 had surpassed the levels registered
before the recession began in July 1981.·' The unemployment rates had returned to prerecession levels, having fallen
3.6 percentage points from their highest point.
This article will briefly describe seasonally adjusted labor
force data for the first 6 months of 1984. examine the recovery in employment in comparison to earlier cycles. and
discuss those industries where lingering problems of unemployment and slow recovery are concentrated.

Women lead in job gains
Total civilian employment grew strongly in the first quarter
of the year gaining about one-and-a-quarter million. The

Richard M. Devens, Jr., is an economist in the Division of Employment
arid Unemployment Analysis, Office of Employment and Unemployment
Statistics, Bureau of Labor Statistics.

second quarter's gain was even stronger-nearly one-anda-half-million. (See table I.)
The job gains in the first half occurred disproportionately
among women:
Percent
Percent of' change
1!f' December
December-June
employment
Total . . . . . . . . . . . . . . . . . .
100.0
100.0
Men .. .. .. .. .. .. .. .. .
53.1
46.7
Women . . . . . . . . . . . . . .
40. 7
46.4
6.3
6.8
Teenagers .. .. .. .. .. ..
This was in contrast to the first year of recovery, when
men accounted for 55 percent of the employment growth.
The proportion of men with jobs (the employment-population ratio) rose about one-and-a-half points to 72. 5 percent
during the first year of recovery and increased an additional
full percentage point in the next 6 months. Women had a
somewhat smaller increase in their employment-population
ratio in 1983, but in the first half of 1984 their ratio rose
by more than a full percentage point to 50.5 percent.
Employment among blacks grew nearly three times as
quickly as among whites during the first half of the year6.1 versus 2.1 percent-but the employment-population ratio for black workers. 52.6 percent. was still more than 8
percentage points lower than that for whites. Employment
gains among blacks were confined almost entirely among
women. as employment of black men grew only intermittently over the first half.
3

MONTHLY LABOR REVIEW August 1984 • Employment in the First Half
Table 1. Selected quarterly labor force Indicators, seasonally adJusted, 1982 to date
[Numbers in thousands]
CbaracterlsHc

1982

1983

IV

1984

II

Ill

IV

I

II

112,607
64.1
103,740
59.1
8,866
7.9

113,642
64.5
105,146
59.7
8,496
7.5

Total
Civilian labor force . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . ...
Participation rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Employed
. . . . . . . . . . . . . . • . . . . . . . . . . ..... .
Employment-population ratio . . . . . . . . . . . . . . . . . ..... .
Unemployed .................................... .
Unemployment rate . .. .. ...................... ..

110,829
64.0
99,054
57.2
11,775
10.6

110,700
63.8
99,214
57.2
11,486
10.4

111,277
64.0
100,037
57.5
11,240
10.1

112,057
64.2
101,528
58.2
10,529
9.4

112,012
64.0
102,506
58.6
9,507
8.5

Men, 2D ,ears and over
Civilian labor force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Participation rate ............................... .
Employed ..................................... .
Employment-population ratio ....................... .
Unemployed ............•..............•...•.....
Unemployment rate . . . . . ....................... .

58,335
78.7
52,537
70.9
5,798
9.9

58,208
78.2
52,563
70.6
5,645
9.7

58,634
78.5
53,095
71.1
5,539

59,017
78.4
54,418
72.2
4,599
7.8

59,360
78.3

9.4

58,983
78.6
53,839
71.8
5,144
8.7

59,584
78.3
55,680
73.2
3,904
6.6

Women, 20 years anll over
Civilian labor force ...........•...............•......
Participalionrate ............................... .
Employed
...........•...•............•.
Employment-population ratio
.................
Unemployed . . . . . . . . . • . . . . . . . . . ................ .
Unemployment rate ............................ .

44,053
52.9
40,108
48.1
3,945
9.0

44,247
52.9
40,313
48.2
3,934
8.9

44,442
53.0
40,654
48.5
3,788
8.5

44,868
53.3
41,324
49.1
3,545
7.9

44,971
53.2
41,717
49.3
3,254
7.2

45,232
53.2
42,084

46,009
54.0
42,920
50.3
3,088
6.7

8,441
54.1
41.1
2,032
24.1

8,245
53.3
6,338
41.0
1,907
23.1

8,201
53.4
6,288
41.0
1,912
23.3

8,206
54.0
6,366
41.9

8,024
53.2
6,370
42.3
1,654
20.6

8,014
53.7
6,446
43.2
19.6

8,049
54.5
6,545
44.3
1,503
18.7

96,521
64.4
87,368
58.3
9,153
9.5

96,263
64.1
87,459
58.2
8,804
9.1

96,719
64.2
88,231
58.6

97,541
64.5
90,353
59.7
7,187

8.8

97,420
64.5
89,485
59.3
7,935
8.1

7.4

98,135
64.5
91,478
60.1
6,657
6.8

98,706
64.8
92,378
60.7
6,328
6.4

11,498
61.4
9,133
48.8
2,364
20.6

11,559
61.5
9,226
49.1
2,333
20.2

11,671
61.8
9,287
49.2
2,384
20.4

11,728
61.8
9,452
49.8
2,277
19.4

11,613
60.9
9,531
50.0
2,081
17.9

11,803
61.4
9,854
51.3
1,949
16.5

11,968
62.0
10,065
52.1
1,903
15.9

5,968
63.6
5,052
53.9
916
15.3

6,019
63.9
5,083
54.0

6,167
63.5
5,293
54.5
874
14.2

6,146

6,221
64.0
5,467
56.3
754
12.1

6,371
64.2
5,677
57.2
694
10.9

6,336
63.5
5,660
56.7
676
10.7

Teenagers, 16-19 years
Civilian labor force .. .. .. .. . .. .................... ..
Participation rate . . . . . . . . . . . . .. . . .. . . .. . . . . . .. . ..
Employed .................................... .
Employment-population ratio . . . .. . .. .. .. . . . .. . . .. . ..
Unemployed .................................... .
Unemployment rate ......•.......•...•...........

6,409

1,840
22.2

55,211
72.9
4,149
7.0

49.5

3,149
7.0

1,568

Whitt

Civilian labor force ..........•.......................
Participation rate .. . . . . . . . . .. . . . • . . . . .. . . . . .. . . ..
Employed ............•..•...................•..
Employment-population ratio ...........•............
Unemployed .............•.......................
Unemployment rate ............................. .

8,488

Black
Civilian labor force ................................. .
Participation rate . . . .. . .. .. . .. .. .............. ..
Employed ..................................... .
Employment-population ratio ...•..•................•
Unemployed •...........................•........
Unemployment rate ............................. .

Hispanic
Civman labor force ............................•.....
Participation rate . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . .
Employed . . . . . . • . . . . . . . • . . . . . . ............... .
Employment-population ratio
. . . . . . . . . . . . . . . . . ..
Unemployed ..............................•......
Unemployment rate ............................ .

Hispanic workers experienced a 2.3-percent rise in employment during the first half, but, because their population
was rising faster than the average, their employment-topopulation ratio showed little improvement.
The civilian worker unemployment rate fell in January
and February and then held at 7.8 percent before dropping
sharply in May and June. The rate stood at 7 .1 percent in
June. (See chart l.) The unemployment rate for men, which
had been higher, on average, than that for women during
1982 and 1983, dipped under the women's rate in March
1984, and in June stood at 6.3 percent, 0.1 point below the
rate for women. The rate for men typically has been somewhat below that for women; one result of the recession was
4

936
15.6

63.5
5,360
55.4
785
12.8

a reversal of this pattern because of the disproportionate
impact of the downturn on industries and occupations with
higher concentrations of male employees.

Minority jobless rates remain high
The black unemployment rate, which averaged 16.2 percent, was more than twice the rate for whites throughout
the 6-month period. The rate for black teenagers averaged
43 .4 percent during the first half. Hispanic unemployment
did fall below double digit levels and averaged 10.8 percent.
The median duration of unemployment dropped by more
than a week, as the number of long-term (27 weeks or more)
jobless was reduced by 475,000. The number unemployed

less than 5 weeks was little changed, presumably reflecting
the increased number of new entrants and reentrants in the
labor force. The two entrant categories made up about 40
percent of total unemployment in June, while workers on
layoff were about 14 percent.
The labor force, which had grown more slowly during
the first year of the latest recovery than in the upturns of
the 1970's and 1980-81, started to pick up the pace of its
growth in February. Over-the-year labor force growth spurted
over the 2-million-a-year mark at that time and had risen to
2.5 million by the second quarter. Increases in the labor
force were driven by the resurgence of the labor force participation rate of women, which had dipped in late 1983.
By March 1984, the rate for women regained its previous
high of 53.4 percent and continued to rise in subsequent
months to 54.2 percent. By contrast, labor force participation by men stayed within a tenth of a percentage point
of the 78.3 percent it stood at the beginning of the year.
Among teenagers, a slight rise in the participation rate was
offset by a decline in population, yielding virtually no net
change in the teenage labor force.

Discouraged workers decline in number
Discouraged workers-persons not in the labor force who
want to work but do not think they can get a job-numbered
1.3 million in the second quarter of 1984, down by 160,000
from the fourth quarter of 1983 and 520,000 from the fourth
quarter of 1982, when the recession was at its worst. Persons
at work on nonfarm part-time schedules for economic reasons, another supplemental measure of labor market problems, fell slightly over the course of the first half to 5.5
million. The number of these workers, sometimes called
involuntary part-timers, has fallen by more than a million
over the course of the recovery.
Nonfarm payroll employment, as measured by the establishment survey, rose 2.0 million between December 1983
and June 1984. The goods-producing sector generated about
40 percent of the new payroll slots-a proportion somewhat
above the roughly 30 percent of total nonagricultural jobs
accounted for by the sector at the end of 1983. Goodsproducing employment had been most severely affected by
the 1981-82 recession, and its high share of job growth
reflected continued recovery from those very low levels.
Construction and durable goods manufacturing were the
growth leaders through midyear in the goods sector. The
service-producing sector, which grew rapidly during the
half, was paced by the services industry, particularly such
industries as business and health services.
Average weekly hours for production or nonsupervisory
workers in private nonagricultural establishments reached
35.4 in April, the longest average workweek recorded since
early 1981. The manufacturing workweek, which is in some
ways a more sensitive indicator of the labor market, averaged 40.8 hours in the first half. This indicator, which has
been on a long-run downtrend, thus approached levels that

prevailed in the mid-1960s. At midyear, the manufacturing
workweek was 40.6 hours, still quite high by recent standards.

Recovery unusually strong
In many ways the current business recovery has been the
strongest since the cycles of the mid- and late l950's. In
terms of absolute growth in employment, the recovery was
a record setter as early as its second quarter (the second 3
months of 1983). By the second quarter of 1984, total employment growth, as measured by the household survey,
had reached a phenomenal 6.1 million. The reduction in
unemployment, again in absolute terms, was almost as dramatic as the rise in employment. It took four quartersuntil the last quarter of 1983-for the cumulative reduction
in joblessness to exceed previous declines, but by the end
of the first half of 1984 the number of unemployed had
dropped by nearly 3.3 million.
When these developments are more properly analyzed in
terms of percentage changes, the narrative is almost as impressive, reflecting an employment recovery stronger than
any in 30 years. The quarterly percent change in employment has been at or above the average for all previous post-

Chart 1. The employment picture,
December through June
Million

Employment increased substantially .•..

106

105
104
103

102
Percent

8.5
8.0

and the unemployment rate fell sharply ...

E

7.5
7.0
Million
114.0

even as the labor force expanded

113.5

113. 0
112. 5
112.0
Dec. Jan. Feb. Mar. Apr. May June
NOTE: Seasonally adjusted data for civilian workers ,

5

MONTHLY LABOR REVIEW August 1984 • Employment in the First Half
World War II upturns. (See table 2.) During the fifth and
sixth quarters of recovery, the cumulative percent change
in employment firmly established this as the strongest cyclical upswing in the series since the recovery from the
1953-54 recession.
On the unemployment side, there was a slower start. It
was not until the fourth quarter of the recovery that quarterly
changes were measurably higher than the average of previous recoveries, and it was not until the fifth quarter of
recovery (the first quarter of 1984) that cumulative percent
declines in unemployment approached the drops that occurred in the 1958 recovery. With continued strength in the
economy, however, unemployment had declined by nearly
32 percent by the end of the first half of 1984. This was
the strongest cyclical decline in this series for any post-1950
recovery. Moreover. the rate of unemployment has had the
largest cumulative drop recorded over a similar period since
the recovery from the recession of 1948-49.
Nonagricultural payroll employment followed a different
growth pattern than total employment in the household survey. Payroll job growth was less than the postwar recovery's
average for the first three quarters of the latest upturn and
drew level with the average during the last quarter of 1983
and the first half of I 984.

Sectoral imbalances in cycle
As the recovery from the 1981-82 recession completed
its fifth and sixth quarters by mid-1984, a clearer perspective
on the intersectoral imbalances that marked the cyclical

episode could be obtained. At the most aggregate level, the
goods-producing sector accounted for about 28 percent of
payroll employment at the beginning of the recession, fell
to 26 percent at the trough, and by June 1984 was back to
27 percent, reflecting the sector's higher rate of job gain in
recovery. Rates of unemployment in the broad categories
have also reflected uneven experiences of economic fluctuations. The rate of unemployment for wage and salary
workers in the private nonfarm goods sector rose from 8.8
percent at the prerecession peak to 16.0 percent at the
trough before recovering to the starting point at 8. 7 percent
in mid-1984. In the service-producing sector, there was less
cyclical volatility, as the unemployment rate was 6.4 percent
at its trough, reached a high of 9.2 percent and by the first
half of 1984 had fallen to 6.1 percent.
It is intuitive, and to some extent correct, to interpret
these developments in terms of a "the-farther-they-fall-thehigher-they-bounce" analogy. Such an analogy, however,
masks severe problems of intrasectoral imbalance that become visible at the next finer level of statistical detail. Confining analysis to the level of detail that is published monthly
under the Bureau of Labor Statistics quality standards for
seasonally adjusted data, one finds only two industries appearing on both lists of the 10 industries with the largest
percentage reductions in employment during the recession
and the IO with the largest percentage increases from the
trough through June 1984. (See table 3.)
The motor vehicles industry suffered substantial declines
in employment and followed the ·'bounce'' analogy quite

Table 2. Quarterly changes In employment and unemployment during business cycle recoveries
(In percent)
2nd Quarter
3rd Quarter
1st 011arter
Cllange Cumulatlve Change Cumulative Change CUmulatlvt

Reces1lan Ctraugh)

1948-49
1953-54
1957-58
1960-61
1969-70
1973-75
1981-82

Employment1
(July 1949)
(July 1954)
(Apr. 1958)
(Apr. 1961)
(Sep. 1970) ...
(Mar. 1975) .......
(Dec. 1982) ... ..

Unemployment1
(Oct. 1949) . . . . . . .
(Sep. 1954)
(July 1958) . . . . ...
(May 1961)
. . . ..
(Dec. 1970) .
1969-70
{June 1975) . , ....
1973-75
(Dec. 1982) . ' .. ...
1981-82

1948-49
1953-54
1957-58
1960-61

Nonlann employment2
(Oct. 1949) . ' . .
1948-49
(Aug. 1954) ...
1953-54
(June 1958) .
1957-58
...
1960-61
(Feb. 1961)
(Nov. 1970) ...
1969-70
(Apr. 1975) . . . .. .
1973-75
(Dec. 1982) ' , , ...
1981-82
,

1Current

0.6
.5
.4

.1
0
0

.2

0.6
.5
.4
.1
0
0

.2

6

0.5
1.5
1.2
.6
.1
.9
1.0

1.8
1.4
.7

-20.0
-20.9
-13.6
.5
2.3
-5.7
-4.5

-16.2
-6.1
-8.6
-90
2.1

33
1.8
1.8
1.4
9
1.6
1.0

2.9
1.7
1.7
.7
.3
1.2

-8.7
10.3
.4

-8.7
-10.3

-3.3
2.1
-3.6
-2.5

-3.3
2.1
-3.6
-2.5

-12.3
-11.9
-13.2
-8.7
.2
·-2.3
·-2.1

0.8
.7
.7
.4
.4

0.8
.1
.7
.4
.4
1
.1

2.5
1.1
1.2
.9
5
.9
.8

.7

.1

- .4

Population Survey (household survey).
2Current Employment Statistics (establishment survey).
Nore: Troughs are series specific.

0.0
.9
.8
.5
.1
.9
.8

.6

.3
.4
1.5

-5.7

-6.3

.9

4111 Quarter
Change

5th Quarter
&lb Quarter
Cumulatlve Change Cumulatlve Cllange CUmulallve

2.4
2.9
1.9
1.2
.4
1.3
2.5

1.2
1.8
1.3
.3
.7

3.6
4.7
3.2
1.5

1.4

2.7
3.5

-33.0
-25.7
-21.0
-19.7
4.4
-11.2
-10.6

-9.7
-4.7
-11.4
-2.2
.4
-1.7
-9.7

-39.5
-29.5
-30.1
-21.4

6.4
3.5
3.6
2.1
1.2
2.9
1.9

1.6

1.0

1.0
1.5

.6
.6
1.8
1.4

0.3
.9
.2
.5
.9
1.0
1.2

3.9
5.6
3.4
1.9
2.1
3.8
4.7

.6
1.4

4.4
6.1
3.5
2.1
3.3
4.5
6.2

-12.7
-19.3

-16.6
3.4
3.9
1.5
-1.9
3.2
-6.7

-49.5
-26.8
-27.3
-19.1
2.9
-10.0
-24.7

-11.9
4.0
6.2
-.5
-1.2
.9
-4.2

-55.5
-.29.7
-22.8
-20.6
16.3
-9.1
-27.8

8.1
4.6
5.1
2.8
1.8
3.7
3.4

1.8
1.1
0
1.0
1.2
.6
1.2

10.0
5.8
5.1

0.8
1.0
.4
.5
1.1
.6
1.0

10.9
6.9
5.5

1.1

4.8

3.8
3.0
4,2
4.6

0.4
.4
.1

.1
1.2

4.3

4.1
4.8
5.7

Table 3. Industries with large losses In employment, large
Increases, and high rebounds, November 1982 through
June 1984
lndustry 1

Ratl.o. ol
jobs
gained
to jobs
lost

Percent
Ion
In
recession

Percent
gain
In
recovery

-12.4
-13.1
-27.6
-33.1
-16.3
-18.6
-14.4
-20.3
-13.1
-15.0

--4.9
9.0
8.5
1.5
9.9
8.5
17.1
34.1
4.7
--2.4

-8.5
-12.0
-12.2
-10.4
-7.2
-14.4
-20.3
-9.4
0.8

13.9
18.8
19.0
14.4
14.8
17.1
34.1
17.6
21.0

1.50
1.38
1.36
1.25
1.90
1.01
1.34
1.70

-12.0
-12.2
-10.4
-7.2
-20.3
-9.4
-0.6
-4.2
-1.6
-2.8

18.8
19.0
14.4
14.8
34.1
17.6
3.5
5.7
7.2
8.1

1.38
1.36
1.25
1.90
1.34
1.70
5.85
1.28
4.36
2.82

Job reductions

Oil and gas extraction
Stone, clay, and glass .
Primary metals .
Blast furnace and basic steel
Fabricated metals
Machinery, except electrical
Transportation equipment .
Motor vehicles and equipment
Textile mill products
Leather and leather products .

(2)
.60
.22
.03
.51
.37
1.01
1.34
.31
(2)

Job gains

Construction
General building contractors
Lumber and wood products
Furniture and fixtures
Electrical and electronic equipment .
Transportation equipment .
Motor vehicles and equipment
Rubber and miscellaneous plastics .
Business services .

(3)

Job reslllency

General building contractors
Lumber and wood products
Furniture and fixtures
Electrical and electronic
Motor vehicles .
Rubber and miscellaneous plastics .
Wholesale trade (nondurable)
General merchandise stores
Auto dealers and service stations .
Real estate
1Ranked

by Standard Industrial Classification.
showed no employment gain following the recession.
31ndustry incurred no employment loss during the recession.
21ndustry

well by having the strongest recovery among heavy losers.
The industry's cyclical pattern also made it the sole large
loser on the list of high rebounders-industries whose re-

coveries, measured as a percent of jobs lost in the recession,
were strongest. However, it should be noted that employment in the auto industry is still lower than in early I 979,
the time of record employment in that industry.
The figures also indicate that the blast furnace and basic
steel industry was the most seriously affected by the recession and that it had regained less than 2 percent of its lost
jobs by June. The primary metals industry-which includes
basic steel-joins seven other goods-producing industries
and two service-producing in a "low rebounders" group .
Five of these-mining, petroleum and coal products, leather
goods, public utilities, and local government-are industries that actually lost jobs between the end of the recession
and midyear 1984.
One characteristic that distinguished the low-rebound from
the high-rebound groups was the timing of job gains. Two
of the low-rebound industries-tobacco and chemicalshad job losses as of December 1983, measured from the
November 1982 trough. Instruments and nonelectrical machines had achieved barely half of their eventual rebound
in the first 13 months. (Primary metals, the group's exception, had made virtually all of its weak rebound by December.) By contrast, the high-rebound group tended to have
about two-thirds of its rebound completed by the 13th month.
The first half of 1984 saw the Nation complete two more
quarters of recovery from a severe recession. The gains in
employment were substantial; indeed, they set records for
postwar upturns. The number of unemployed had dropped
from a recession high of 11. 9 million to 8. I million at the
end of the first half. The rate of unemployment in June was
high by historical standards. There were also industries that
had not seen as full a recovery as the overall economy.
Thus, while the preponderance of the news was good during
the first half of the year. there was still basis for continued
concern.
O

--FOOTNOTES--

1
The Current Population Survey (CPS) is conducted monthly by the
Census Bureau on behalf of the Bureau of Labor Statistics. The survey is
conducted among a scientifically selected sample of about 60,000 households and provides information on labor force. employment. and unemployment by a variety of demographic and economic characteristics.
~ Data from the Current Employment Statistics (CES) program are collected from the payroll records of nearly 200.000 nonagricultural estab-

lishments by the Bureau of Labor Statistics in cooperation with State
agencies. This survey provides estimates of the number of persons on
payrolls of businesses. their average hours. and their average hourly and
weekly earnings.
'The identification of turning points in the business cycle is, by general
consensus of the economics profession, carried out by the National Bureau
of Economic Research, a private institution based in Cambridge. Mass.

7

Discouraged workers: how strong
are their links to the job market?
More than half of the discouraged workers
have not looked for work in more than a year;
while some of them return to work or resume
job search, the majority seldom test the job market
PAUL

0.

FLAIM

In line with the cyclical ups and downs in the number of
unemployed, the number of discouraged workers, that is,
persons who report they want to work but are not looking
for a job because they think they could not find one, has
also exhibited large swings over the last decade and a half.
For example, during the early 1970's, when the number of
unemployed fluctuated in the 4- to 5-million range, the
number of discouraged workers oscillated between 600,000
and 800,000. When the number of unemployed climbed
past the IO-million mark, as it did in the 1982-83 period,
the number of discouraged workers rose to the 1.6- to 1.8million range.
Given the fairly strong cyclical sensitivity in the number
of discouraged workers, 1 one might conclude that they have
strong links to the job market, that they test it periodically,
and that they are ready to jump back into it if they believe
jobs are available. However, an indepth look at available
data on the behavior of discouraged workers leads to a quite
different conclusion. While some of them may, indeed, keep
Paul 0. Flaim is chief of the Division of Data Development and Users'
Services, Office of Employment and Unemployment Statistics, Bureau of
Labor Statistics.

8

a close eye on the job market, the majority appear to have
few, if any, concrete contacts with it. For example, of the
discouraged workers interviewed over the late 1970's and
early 1980's, two-thirds or more reported that generally
more than a year had gone by since they last held a job.
And special surveys conducted over this period showed that
less than half of them had made any jobseeking efforts
during the year preceding their interview. More importantly,
according to a special study of data for the 1976-77 and
the 1982-83 periods, only a minority of these persons reentered the job market in the 1-year period following their
original classification as discouraged.

Little recent work experience
Discouraged workers can be divided into three roughly
equal groups in terms of their recent work history. As shown
in table 1, about one-third reported, for the 1979-83 period,
that they have either never worked at all or that more than
5 years have gone by since they last held a job. Another
0ne-third report that their last job dates back from l to 5
years. Thus, only one-third are found to have held a job in
the 1-year period preceding the interview in which they are
identified as discouraged workers.

Another surprising finding emerges when one examines
the cyclical changes in these three groups of the discouraged
worker population over the 1979-83 period. Given the sharp
cutbacks in employment in various industries over this period, one would expect that most of the increase in the
number of discouraged workers would have been accounted
for by persons with fairly recent work experience-that is,
persons who lost jobs and quickly lost hope of finding new
ones. However, such was not the case. The largest increases
were posted by the two groups of discouraged with the least
or most remote work experience. And among the discouraged with the most recent experience (those who had worked
during the previous 12 months) only a little more than half
cited economic problems as the main reason for leaving the
last job.
Of course. even if certain discouraged workers have no
recent work experience. they can still be sincere in reporting
that they want a job and in perceiving that their search for
one would be futile. For example. a detailed breakdown of
the data in table I shows that a majority of the discouraged
with no previous work experience w,hatsoever are youths
who would apparently like to land their first job. Given the
very high rates of unemployment among youths in recent
years. it is not surprising that some. although desirous of
work. were not confident enough of their prospects to initiate
(or resume) the job search process.
The same detailed data also show a large concentration
of women 25 to 55 years among the discouraged whose last
job dates back more than 5 years earlier. Although we know
little about the work history of these women. we suspect
they may have left the labor force during their childbearing
and childrearing years. They would "now" like to rejoin
the labor force but may be deterred by their belief that they
could not find a suitable job.
So. by itself. the fact that many discouraged workers have
little or no recent work experience does not allow us to
questiorr their desire to work. But there is other evidencenamely. the fact that they seldom test the job marketwhich leads us to question at least the "intensity" of their
desire for jobs.

Table 1. Discouraged workers by when last worked and,
tor those who worked the previous year, reasons tor
leaving last job, 1979-83
[Numbers ,n thousands[

-

When last worked and
reason for leaving last job
Total

Never worked
Last worked more than 5 years ago
Last worked 1 to 5 years ago
Worked last year
Left 10b because of
School. family
Health
Retirement
Economic problems
Other reasons

1979

1980

1981

1982

1983

766

993

1 103

1.567

1.641

101
158
251
255

155
217
288
334

141
221
366
375

223
339
536
469

229
332
625
454

40
16
8
125
67

54
10
8
180
82

63
15
11
202
83

62
12
17
268
109

57
10
16
280
92

Few job search efforts
It is generally assumed that a worker becomes discouraged over job prospects after failing in repeated efforts to
find work. Indeed, discouraged workers are popularly described as persons who "have simply given up the search
for work." But, again, the data do not conform to such
description.
While discouraged workers are not questioned regularly
as to when they last looked for work, such questions are
asked in special surveys. In these surveys, less than half of
the discouraged report having tested the job market over the
preceding year.
For example, in a special supplement to the Current Population Survey conducted both in September and October
of 1978, 2 the persons identified as discouraged workers were
asked, among other things, when they had last looked for
work. The findings are summarized in the following tabulation which shows the percent of discouraged workers interviewed in September and October 1978 who had recently
searched for work:
Searched j<Jr ll'ork
d11ri11K prel'ious-

Total discouraged .............
Reason:
Job-market factors ........
Personal factors ..........

12

3
months

6
months

months

34.2

39.8

44.2

40.4
18.3

48.3
18.3

52.4
23.7

Of all discouraged workers. one-third reported that they
had tested the job market over the previous 3 months: 40
percent had done so over the previous 6 months. and 44
percent had looked for work at anytime during the previous
year.' The proportion with any job search efforts was particularly low-about one-fifth-among those persons citing
"personal factors" (age problems. skill or education deficiencies. or other personal handicaps) as the reason for their
discouragement.
Roughly the same results were obtained from the Methods
Development Survey. a small experimental survey conducted over several years by the Bureau of the Census. 4
This survey also revealed that about two-fifths of the discouraged workers had tested the job market in the 6-month
period preceding their interview.

Post-interview behavior
Given the structure of the Current Population Surveyin which the households falling in the sample are interviewed in the same calendar months for 2 consecutive yearsit is possible. through computer matching of individual reports. to determine what the persons who were classified as
discouraged workers in the initial year of interviewing were
doing I year later. and whether they had worked during the
intervening I-year period. Such matching. which had pre9

MONTHLY LABOR REVIEW August 1984 • Discouraged Workers
viously been done with data for the 1976-77 period. when
both unemployment and the number of discouraged workers
were relatively low, was recently repeated with data for the
1982-83 period, when the ranks of unemployed and discouraged were much greater. Both sets of data indicate little
labor force activity over these periods among persons who,
in the initial year, were classified as discouraged workers.
As shown in table 2, of the roughly I. 7 million persons
classified as discouraged in the second half of 1982. about
one-fifth were employ~d I year later. nearly one-fifth were
looking for work, while the remainder were still out of the
labor force. And of those not in the labor force. only about
one-fourth still reported themselves as discouraged.
Persons who, in the second half of 1982. attributed their
discouragement to job-market factors (··could not find job··
or "think no job available") were somewhat more likely to
be either employed or still discournged I year later than
were those who had attributed their discouragement to personal factors (skill problems. age problems. or other personal handicaps). Likewise, among those who had been
discouraged in 1982. the men were more likely to be in the
labor force I year later than were the women. But even
among these men. more than one-half were still out of the
labor force when interviewed in 1983.
Some of the persons who were discouraged in the second
half of 1982 and still inactive in the second half of 1983
did have temporary employnlent in between. But the proportions were small-one-tenth for those still out of the
labor force the following year. and less than one-fourth for
those who were then looking for work.
These findings are in line with those obtained from the
tracking of discouraged workers over the 1976- 77 period.
Although the economic situation was then more favorable
than over the 1982-83 period. it did not result in many
discouraged workers returning to the job market. Only 20
percent of those who had been discouraged in 1976 were
employed I year later and only 15 percent were looking for
work. The balance. about two-thirds. were still out of the
labor force. although a small fraction of them reported they
had worked during the I -year period between the two interviews.~

:Factors affecting reentry
Recent work experience appears to be the main fm:tor
affecting the probability of reentry into the labor force among
discouraged workers. Although only a small proportion of
the discouraged return to the job market. those who report
in the initial interview that they had worked during the
previous 1.2 months-and who generally make up about
one-third of the total-are far more likely to be employed
I year later than arc those with more remote work experiem:e.
For example. of all persons classified us disrouraged in
the second half of 1982. the proprn1ion employed in the
second half of 1983 wus 33 percent for those who had held
lO

Table 2. Discouraged workers in the second half of 1982
distributed by their labor force status In the second haH of
1983, and by original reason for discouragement
Status In 11cond hall of 1983
Reason for
discouragement

Number

Not in labor force

Un(lllousands) Total EmDisployed employed Total
couraged Other

Total ..
Personal factors .
Job-market factors , .
, ,
Men .......
Personal factors .
Job-market factors ..
,

Women ....
Personal factors ...
Job-market factors ..

1.693
442

1,251

577
161
416
1,116
281
835

100.0 19.9
100.0 12.6
100.0 22.8

17.8
19.7
17,0

62.3
67.7
60.1

17.0
10.6
19.5

45.4
57.2
40.6

100.0 25.2
100.0 217
100.0 26.7

20.7
20.9
20.6

54.1
57.4
52.7

18.6
13.0
21.0

35.5
44.4
317

100.0 17.4
100.0 7.5
100.0 21.1

16.4
19.0
15.4

66.1
73.5
63.5

16.2
9,3
18 9

63.7
44.6

49.9

a job in the 12-month period preceding their initial interview. For those with more remote work experience (or none
at all), the proportion who actually had a job in 1983 was
only 15 percent.
Relative to the dominant role of recent employment. other
variables on which data are gathered through the Current
Population Survey appear to have much less influence on
the probability of labor force reentry for discouraged workers. For example. among those who had been discouraged
in 1982 but were found to be employed in 1983. 87 percent
had reported in their initial interview that they planned to
seek work over the next 12 months. But even among those
still out of the labor force in 1983. the proportion that had
reported in 1982 that they were planning to seek work was
also quite high-73 percent. So. it appears that the alleged
intention of a discouraged worker to enter the labor force
is a very weak indicator of his or her future labor force
status.

Weak link to job market
In conclusion. several sets of data-covering periods of
relatively low as well as very high unemployment-show
that a large proportion of persons classified as discouraged
workers in the Current Population Survey have rare contacts
with the job market. For many. the last job dates back many
years. More than half report no jobseeking efforts in the
year preceding the interview. And those without any recent
work experience when first identified as discouraged appear
quite unlikely to make any subsequent forays into the job
market. This indicates that many of the discouraged. although expressing their desire for a job and their intention
to look for one. find it very difficult to translate their sentiments into concrete and productive jobseeking efforts.
The evidence presented here supports the present practice
of not including discouraged workers in the labor force. It
also supports the recommendation made by the National
Commission on Employment and Unemployment Statistics
that the measurement of discouraged workers should be
limited to those who have made some efforts to find a job
during the preceding 6-month period. 6
D

--FOOTNOTES-1
The cyclical sensitivity of the discouraged workers' series has been
examined by Paul 0. Flaim in "Discouraged workers and changes in
unemployment," Monthly Labor Review, March 1973. pp. 95-103, as
well as by Carol M. Ondeck, "Discouraged workers' link to jobless rate
reaffirmed," Monthly labor Review, October 1978, pp. 40-~2.

2 The

Current Population Survey has a sample of about 60,000 house•
holds, and is conducted monthly by the Bureau of the Census for the
Bureau of Labor Statistics.
3 For a more detailed description of these numbers, see Harvey R. Hamel,
"Two-fifths of discouraged sought work during prior 6•month period,"
Monthly labor Review, March 1979, pp. 58-60.
4
The experimental survey is the "Methods Development Survey" con-

ducted by the Census Bureau to test new approaches that might be introduced into the Current Population Survey. The experimental survey was
begun in May 1978, but the special questions on discouraged workers were
introduced in "Phase II," which began in December 1979.
5
See Barbara Cottman Job, "How likely are individuals to enter the
labor force?" Monthly Labor Review, September 1979, pp. 28-34.
6
National Commission on Employment and Unemployment Statistics,
Counting the Labor Force (Washington, Government Printing Office, 1979).
For a summary of the commission's recommendations, see Rohen L. Stein,
"National Commission recommends changes in labor force statistics,"
Monthly Labor Review, April 1980, pp. 11-21.

Stress and satisfaction
It has been shown that man has to contend both with a physical component of his working environment and a psychosocial component; that
conditions in either or both may be unacceptably stressful; and that work,
whether physical or skilled, may itself constitute an unacceptable stress.
One must also recognize that, in the terms defined, stress is always present
to a greater or lesser degree and that, paradoxically, the total absence of
apparent stress becomes in itself a stress. Thus, on the one hand, stress
can be considered as a load, increasing to an overload, arising from addition
to the man-machine-environment complex of qualities which are undesirable from the human point of view, such as intolerable working conditions,
harsh supervision, or unreasonable working hours. On the other hand,
removal of desirable attributes by, for example, the creation of a stultifying
environment, with reduced stimulation and inherently boring work, can
act as a kind of negative loading which can be equally stressful. The stress
experienced by an individual lies somewhere on the continuum between
that arising from removal of desirable qualities and that arising from the
addition of undesirable qualities. Thus, there is some point where his stress
level can be optimum.
-T. M. FRASER

Human Stress, Work and Job Satisfaction:
A Critical Approach (Washington, International
Labor Office, 1983), p. 55.

11

An evaluation of BLS' projections

of 1980 industry employment
Employment was underestimated in projections
made in 1970, 1973, and 1976; estimates
of labor force growth and unemployment
turned out to be offsetting factors
JOHN TSCHETTER

The Bureau of Labor Statistics periodically publishes projections of gross national product (GNP) and output and
employment by industry. These projections provide a framework for BLS' occupational projection program as well as
for employment analysis of energy. housing. transportation,
and defense issues. This article is a final step in the pro-jection process-evaluation of the projections of the 1980
economy. Evaluation is an important part of the projection
program. for only after the projected period has run its
course can we quantify the limitations of our pn~jected data.
BLS published three projections of the 1980 economy. 1
Those published in April 1970 underestimated employment
(including military) in 1980 by 4.0 percent: those published
in December 1973 underestimated employment by .9 percent: and those published in 1976 underestimated employment by 1.4 percent. These errors were kept modest by
offsetting estimates: for example. an underestimate of labor
force growth was offset by an underestimate of the unemployment rate. The 1980 recession slightly increased the
gap between projected and actual employment.
For the three projections, the absolute difference between
the projected and actual trends by industry was 1. 9 percentage points per year. The absolute difference in the number of projected and actual jobs was 90,000 or 15 percent,
per industry. The larger differences, for the most part, occurred among the smaller industries in terms of employment,
John Tschetter is an economist in the Office of Economic Growth and
Employment Projections, Bureau of Labor Statistics. Howard N Fullerton,
an economist in the same office, assisted in the preparation of this article.

12

BLS accurately projected one-third to one-half of the fastestgrowing industries.
Among major industries. projected employment levels in
State and local government and manufacturing were consistently higher than actual levels: employment in the other
major industries was usually lower. The errors for manufacturing partly reflect the effect of the unanticipated 1980
recession on durable goods industries. (The projections are
intended to capture longer term secular trends. rather than
business cycle changes.) In addition. other factors such as
trade issues and their impact on manufacturing industries
were not sufficiently anticipated. As a consequence of the
errors for the major industries. each of the projections slighdy
underestimated the long-term shift from goods- to serviceproducing industries.
How good were the employment projections when compared to alternative employment projections and projecting
techniques'? The errors in BLS' projections were the same
size or magnitude as the errors of projections developed by
two private organizations. And BLS' projections. which reflect models and judgments. performed better than two simpler models.
There are five components of the 1980 projections: labor
force. aggregate or macroeconomic activity. industry outputs. industry employment. and occupational employment.
The labor force and occupational employment projections
have been evaluated. 2 This article evaluates the projections
of 1980 aggregate economic activity and industry output
and employment. It discusses errors in the employment projections and calculates the pan which can be attributed to

the 1980 recession. It also examines the effects of industry
employment projections on occupational employment projections. Finally, the sources of errors in the employment
projections are determined.

Evaluation complicated by revisions
The 1970 projections of the U.S. economy in 1980 estimated industry employment trends over the 1968-80 period; the 1973 projections estimated trends over the 197280 period; and the 1976 projections estimated trends over
the 1973-80 period. Projected employment trends are based
on assumptions about labor force growth, unemployment
rates, and the adjustment between the number of employed
persons and the number of jobs. 3 One assumption is that
the economy will expand steadily toward full employment.
In 1970, employment projections assumed a 1980 economy
near full employment; in 1973 and 1976, they assumed a
point on a path towards full employment by 1985. Projected
trends in industry employment are based on assumptions of
total employment, level and distribution of the gross national
product, labor productivity by industry, and an input-output
matrix.
To emphasize the uncertainty of projections, BLS has traditionally developed scenarios which cover alternative assumptions about employment and GNP levels. The projections
reviewed here are the middle or base scenario. While the
differences at the macro or GNP level among the scenarios
were moderately broad in terms of percentage and dollar
amounts, the differences in terms of trends were narrow.
This also applies for industry employment.
This evaluation is complicated by revisions in the series
which were projected. For example, the definitions and
methods for structuring the industries have changed twice.
The 1970 projections reflected the 1958 Standard Industrial
Classification (SIC); the 1973 projections reflected the 1967
sic; the 1976 projections reflect the 1972 sic. This and other
revisions mean that the projected values, as originally published, cannot be directly compared to current data. For this
evaluation, the projected trends are applied to the revised
historical data series to obtain projected 1980 values which
are consistent across the three projections. In essence, the
base for each projection has been revised to reflect data
revisions. The projected trends are unchanged.

Total employment underestimated
BLS underestimated total employment growth in each of
the three projections by .2 to .4 percentage points per year.
The following tabulation shows projected and actual annual
growth rates in total employment for the three projections: 4

Year
published
1970 ...............
1973 ...............
1976 ...............

Period
covered Projected Actual Difference
-0.3
1.8
1968-80
1.5
-.2
1972-80
2.0
2.2
-.4
1973-80
1.8
2.2

In the 1970 projection, BLS expected total employment to
grow 1.5 percent per year over the 1968-80 period; employment actually grew 1.8 percent per year, a difference
of 0.3 percent.
The difference between projected and actual trends reflects some offsetting estimates. BLS consistently underestimated labor force growth during the 1970's, especially the
trends in participation rates for women. 5 However, the low
labor force estimates were offset by estimated unemployment rates which were l. 7 to 3.2 percentage points lower
than actual rates. There was an upward trend in unemployment throughout the 1970's, and the rate did not return to
the relatively low 1973 level following the 1973-75 recession, despite uninterrupted growth over the 1975-79 period.
The labor force underestimate was further offset by an
overestimate of the adjustment between the number of employed persons and the number of jobs. Because a person
can have two or more jobs, the number of jobs in the economy exceeds the number of persons employed.
The 1970 projections put the number of jobs in 1980 at
101.7 million, compared with the actual number of 105.9
million, a difference of 4.2 million jobs. The 1973 projections estimated the number of jobs would be 104. 9 million;
the 1976 projections, 104.4 million.
One trend that has characterized employment over the
past several decades is the movement from the goods-producing sector (agriculture, mining, construction, and manufacturing industries) to the service-producing sector
(transportation, communication, public utilities, finance, trade,
other services, and government industries excluding military). The projections slightly underestimated this shift. In
the 1970 and 1973 projections, the service-producing sector
was projected to account for 69.6 percent of all civilian jobs
in 1980, and in 1976 projections, 69.9 percent. In 1980,
70.8 percent of all civilian jobs were in the service-producing sector. The difference for the most part can be attributed to the unanticipated 1980 recession.

Industry differences modest
At the industry level, the differences between actual and
projected trends were usually modest. (See table 1.) For the
1970 projections, industry employment was expected to grow
an average of .86 percent per year over the 1968-80 period;
the actual growth was 1.08 percent per year, a difference
of .22 percentage points per year. The following shows the
mean projected and actual employment trends and differences by industry for the private economy, except households, for the three projections: 6
Year
ProDifAbsolute difference
published jected Actual ference Unweighted Weighted Squared
1970 ...... 1.08 0.86 -0.22
l.30
1.02
1.81
1973 ...... 2.31 2.07
- .24
2.73
2.05
3.59

1976 ...... l.64 1.34 - .30
1.50
1.18
2.07
For the three projections, the difference between projected
and actual trends was less than 2 percentage points per year
13

MONTHLY LABOR REVIEW August 1984 • Evaluating Projections of Industry Employment
Table 1. Employment In 1980, projected and actual trends by Industry for the private economy, except households
[Average annual rate of change)

1981-811 period
Industry

~

Actual

1972-80 period
DH·

lerence

Livestock and livestock products . . . . . . . . . .
Crops and other agricultural products
Forestry and fishery products . . . . . . . . . . . .
Agriculture: !orestry, and fishery services . . . .
Iron ore mining . . . . . . . . . . . . . . . . . . . . .
Copper ore mining . . . . . . . . . . . . . . . . . . .
Other nonferrous metal ore mining . . . . . . . .
Coal mining . . . . . . . . . . . . . . . . . . . . . . .
Oil and gas extraction . . . . . . . . . . . . . . . . .
Stone and clay mining and quarrying . . . . . . .

-2.6
-2.6
1.0
1.0
-0.9
0.4
0.4
- 3.0
-0.9
0.9

-1.8
-1.8
3.9
3.6

.7

4.2
-.1

Chemical and fertilizer mining . . . . . . . . . . . .
Construction . . . . . . . . . . . . . . . . . . . . . . .
Complete guided missiles and space vehicles . .
Other ordnance . . . . . . . . . . . . . . . . . . . . .
Food products............. . . . . . . . . .
Tobacco manufacturing . . . . . . . . . . . . . . . .
Fabric, yarn, and thread mills . . . . . . . . . . . .
Miscellaneous textiles and floor coverings . . . .
Hosiery and knit goods . . . . . . . . . . . . . . . .

0.9
2.5
- 2. 7
-2.7
-0.1
-2.2
-1.0
-0.8
-0.8

.7

-.1

-~~ ··························

1~

Miscellaneous fabricated textile products . . . .
Logging, sawmills, and planing mills . . . . . . .
Millwork, plywood, and other wood products
Household furniture . . . . . . . . . . . . . . . . . .
Other furniture . . . . . . . . . . . . . . . . . . . . .
Paper products . . . . . . . . . . . . . . . . . . . . .
Paperboard . . . . . . . . . . . . . . . . . . . . . . . .
Publishing . . . . . . . . . . . . . . . . . . . . . . . .
Printing . . . . . . . . . . . . . . . . . . . . . . . . . .

1.0
.0
.0
1.7
3.0
1. 3
.8
1.2
1.2

Chemical products . . . . . . . . . . . . . . . . . . .
Agricultural chemicals . . . . . . . . . . . . . . . . .
Plastic materials and synthetic rubber . . . . . .
Synthetic fibers . . . . . . . . . . . . . . . . . . . . .
Drugs . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cleaning and toilet preparations . . . . . . . . . .
Paint . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Petroleum products . . . . . . . . . . . . . . . . . .
Rubber products . . . . . . . . . . . . . . . . . . . .
Plastic products ......... , . . . . . . . . . .

0.4
0.4
1.9
1.9
1.9
1.9
0.5
-1. 7
2.5
2.5

Leather, footwear, and leather products . . . . .
Glass . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cement, clay, and concrete products . . . . . . .
Miscellaneous stone and clay products . . . . . .
Blast furnaces and basic steel products
Iron and steel foundries and forgings . . . . . . .
Primary copper and copper products . . . . . . .
Primary aluminum and aluminum products . . .
Other primary nonferrous metals and products
Metal containers . . . . . . . . . . . . . . . . . . . .

-2.3
1.2
2.0
2.0
- .8
- .8
1.8
1.8
1.8
.1

Heating apparatus and plumbing fixtures . . . .
Fabricated structural metal products . . . . . . .
Screw machine products . . . . . . . . . . . . . . .
Other fabricated metal products . . . . . . . . . .
Engines, turbines, and generators . . . . . . . . .
Farm machinery . . . . . . . . . . . . . . . . . . . .
Construction, mining, and oilfield machinery . .
Material handling equipment . . . . . . . . . . . .
Metal working machinery . . . . . . . . . . . . . . .
Special industry machinery . . . . . . . . . . . . .

1.6
1.6

General industrial machinery . . . . . . . . . . . .
Machine shop products . . . . . . . . . . . . . . . .
Computers and peripheral equipment . . . . . . .
Typewriters and other office equipment . . . . .
Service industry machines . . . . . . . . . . . . . .
Electric transmission equipment . . . . . . . . . .
Electrical industrial apparatus . . . . . . . . . . . .
Household appliances ................ ,
Electric lighting and wiring . . . . . . . . . . . . . .
Radio and television sets . . . . . . . . . . . . . . .

1.0
4.1
4.1
4.1
2.3
1.1
1.1
1.3
1.5
.9

Telephone and telegraph apparatus . . . . . . . .
Other electronic communication equipment . . .
Electronic components ... •. . . . . . . . . . . . .
Other electrical machinery . . . . . . . . . . . . . .
Motor vehicles . . . . . . . . . . . . . . . . . . . . .
Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . .
Ship and boat building and repair . . . . . . . . .
Railroad and other transportation equipment . .
Miscellaneous transportation equipment . . . . .
Scientific and controlling instruments . . . . . . .

.9
.9
2.1
.5
.2
-1.0
.9
.9

14

.8
1.0
.7
1.4
1.1
2. 8
.7
1. 7

.9

1.6

-.1

1.9
1.9
5.7
3.3

0.8

.8

3.0

2.6

.8
1.5
1.5
8.7

2.6

.1

-5.1
-5.1

-2.4
-2.4
-.3
.6
-.5

-.4

-1.6
-1.5
-.5

Pro11cled
In 1973

Actual

-4.4
-5.3
1.2
1.5
2.6
1.8
3.0

.1
-1.0
1.2
1.3
1.4
3.7
-2.0
.1
-.7

.6

.3
.3

.7
.3
1.8

.9

-2.3
-1.4
.9

2.4
-.7

.9

.9

1.0
2.8

-.5
-1.0
-.4

.0

1.5
.2

-.8
1.3
1.3

.5

.5
-1.1
-1.1
2.1
2.1

-.6
.8
2.5
2.5

-3.4

-1.6
-1.5
-1.1
-1.6
.1
.1

.1
.1
-3.0
-3.0

.2
.2

-1.1
2.5

-.0
-.0

2.6
2.0
3.7
2.0
2.2

2.7
1.8
4.8

lerence

Projected
In 1978

-4.8

-0.3

5.8
.1

-3.4

.5
1.3

6.7
3.0
-2.8

8.5
4.7
9.1

2.6
5.2
5.3

9.4
2.5
.0
-1.0
-2.2
-3.6

-2.1
-1.5
.2

.4
.3
-1.5
1.6

.4

- .7
2.6
4.7

.1

2.8
2.7
-1.1

1.5

-.0

2.0
3.3
4.0
4.1
4.6
1.2

2.0
2.3

1.5
-.4
1.5
2.0
2.4
3.5

-.0
-1.2
.5
1.4
1.0
2.4

1.1
1.1

-1.7

2.2
3.4

.4
-1.4

.2
-.9
-.9
-.9
-.3

3.2
3.2
3.2
2.1
.9

.9
-.6
.6

-.9

-.9
-.9
2.8
2.8
.5
-2.0
1.8
1.8
1.8

1.1

-.2
-.2
-1.9
-.9
-1.7
-1.7
-1.7
.6
2.3
.3
-1.0

.9
.9
.9

-.5

.7
2.8
-1.5

-4.3
-1.0
-1.6
-4.4
.6
2.5

.2

.0
.6
.6
2.3
1.2
1.9

1.6

.6

.3

.8
2.8
1.1
2.1

-5.6
-3.7
-.9
-.5
-2.3
1.2

1.1

.6

1.1
-1.7
-.4

1.7
1.7

-.7

1.1

-1.1
-2.0

1.1

-.9

-.9

1.7
.5
-.7
1.8
2.2

1.1
-.1

-3.0
.6
.4

1.9
2.7
-1.5
-2.3
2.7
1.9
-.2
.7
-.9
4.3

1.2
1.9
4.3
-3.4

-2.3
.1

-.4

.6
.8
-1.5

.2
-2.7
.2

-.6

-.o

-.3
.0

.6
-1.6

-.3

.0

.7
-3.3

.7

2.7

-1.3

-.5
-5.5

4.1
1.7

-.4
2.0
2.4

2.3

8.0
12.2

3.2
3.1

.1
.7
1.6
1.2

-.o

-1.1
-3.9

2.0
2.0
2.0
3.8
2.5

2.6

5.9
1.4
.8
1.8
3.3
1.6
2.5

.6

-3.4

5.2
-1.1

-4.7

-4.3
-3.6

-2.6

3.9

2.9

.2

3.3

6.9
.6
-7.9

-.6
.8

-12.8

9.0
-8.2
5.3
-12.9

1.9
-1.8
1.9
3.1
4.1
-.4

3.7

-.3

2.1

3.9

.0

.8
-1.2

-1.0
5.7
2.2
2.1
-2.1

-5.6
, 10.5
-.9
6.5

2.7
3.2

-1.5

.2
.4

3.3

2.3
1.5
7.2
1.2
.1

2.8
1.9

-.8
-.9

4.4

1.2

4.2
1.8
-.0
-2.6

.3
.7

1.3

-.6

1.2

6.0
10.8

.6

1.2
.8

.4

.6

5.7

1.6
.9

3.4
2.5

1.0
2.8
3.9

2.7
.3

-.3

-3.5

.9

5.4

.9

-.7

-1.8

.7
5.6
1.1

-2.8

4.6

3.0

-2.0

.6

6.1

4.2
2.8
.7

6.9

-1.3
.9

-3.2
-2.1
1.0

2.9
1.4
2.9
4.9
4.6
3.0

10.0
4.0
2.8
5.7
2.3

.1

3.1
3.1
1.3
3.6

.8
1.2

-2.4

.9
.3

.4
.4
-1.2

6.0

1.0

3.2

-4.3

-4.7
-1.3
-1.3
-2.9
-2.0

1.2

-1.8
3.0
6.2
1.5
2.6
-4.9

-1.7

-2.0

-1.9

-5.2
.1
7.2
2.8
.3
-3.9
10.0
5.7

Allsolute
difference

-.0
-.7
-1.3

-2.9

-3.7

.8

D11·

lerence

-2.6

-2.1

.5

Actual

-.3

5.3

1.3

1.3

.0
1.0
.7
- .1
-.6

-4.9

-.5

-2.2

.3

2.7
3.3
-.1

-2.5
3.2

-1.7
2.4

.7
.0

.4

4.6
-.1

3.1

1.4
.5
1.8
4.1

-3.7
-1.7

-.9
-.9

4.0
3.9
5.7

1.3

-2.3
1.0

3.6
3.3
1.9

1.5

-0.1
2.9
1.7
2.1

.8

5.2
.4
-4.6
5.5
4.6
10.1
1.4

-2.9
1.1

1.1
2.2
-.8
-2.1
2.7
2.8
-.5

-1.1
-.4
-1.5
-1.5
-.1
-.1
-1.5
-1.5
-1.5
-1.3

.4
.4

1973-80 period

D11·

2.5

2.3
1.2
.9
3.0
.4
1.0

1.0
.8
4.3
3.4

.6
.5
1.6
1.3
2.5
.8

1.7
2.0

1.2
1.2

1.5
.7

-.9
1.7
2.5
-1.6

-2.4

2.2

1.7
-3.4

1.3

.4

.7

1.6
.2
1.4
2.2

2.4
3.7
1.3

2.7

1.2

2.0

4.5
6.4
1.9

-.9

-.4

.7
.3

3.0
1.5

.1
.4

1.1

-.2
-1.2
.1
.1
-1.2
1.5
.4
.5

.2
.1

1.3

-1.5
.4
-3.3

-2.0
-3.0
-2.1
-.7

-.1
2.7

-.3
3.3

4.8
2.8
.0

3.9

2.5
2.0

2.7

-2.2
-.6

1.0

1.9

.8

.9
1.8
.6
-1.1

3.5

-.5

-5.0
2.4

-4.6
.3

.5
1.1
1.5

.6
.6

.8

2.7
1.2
1.2
.8

.1
1.1
3.0
2.4
2.0
1.6

2.8
3.8
1.3

3.3
3.5

3.4
2.2
6.1
.4

Table 1. Continued-Employment In 1980, projected and actual trends by Industry for the private economy, except households

[Average annual rate of change]

1968-80 period
Industry

Projected

In 1970
Medical and dental instruments .........
Optical and ophthalmic equipment .
Photographic equipment and supplies .....
Miscellaneous manufactured products .....
Railroad transportation
.... .
Local transit and intercity buses . . . . .. .
Truck transportation
...............
Water transportation .................
Air transportation ...................
Other transportation . . . . . . . .
. ....
Communications, except radio and television

.

Radio and television broadcasting ........
Electric utilities ....................
Gas utilities . . . . .
. ..... .
Water and sanitary services ............
Wholesale trade . . . . . . . . . .........
Retail trade .......................
Finance . . . . . . . . . . . . ...........
Insurance ........................
Other real estate
.............
Hotels and lodging places .............
Other personal services ...............

.
.

Miscellaneous business services .........
Advertising ............... .
Miscellaneous professional services .......
Automobile repair . . . . . . . ..........
Motion pictures ....................
Other amusements ..................
Health services, excluding hospitals .......
Hospitals ........................
Educational services .................
Nonprofit organizations ...............

.

NOTE:

.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

Actual

1972-80 period
Dif-

Projected

ference

In 1973

.9
.9

2.7
2.7

1.7
1.7

4.2

.9
.9

2.7
.4

4.0
.3

.6
.6
.6

1.1

1.7
-.5
.4

1.1
1.1
1.1

.4
.4
.4
.4
.4

1.7

.8

3.7
1.7

.6
.6

1.7
1.7

1.7
1.7

2.8
2.8
3.2
3.2
4.2
1.6
1.6

.6

.6
.6
.9

2.1

.6

1.9
1.9
1.2
1.8
1.8
4.1
4.1
4.1
1.7

2.1
2.1
3.2
3.2
3.2
3.2

1.1

1.1

5.9
5.9
5.9
3.9
3.9
3.9
3.9
3.9
3.9
3.9

Dif-

Projected

ference

In 1976

6.5
3.6

2.4
2.9

1.3

-2.7
-2.3

.7

-2.9
1.0
3.1

-2.0
-.6
2.5
2.2

.2
4.0

2.8
5.7

1.3
1.6

9.3
4.4

1.6
1.1

1.2

5.9
5.0

1.1
1.1

.8
4.1

1.2
1.2

1.3
1.3
2.9
-.2
-.2

1.8
1.8
1.8
2.2
1.8
1.8

1973-80 period

Actual

.2
2.0
2.1

1.4

5.4
3.9
2.6

4.1

5.5

2.0

3.5
6.5
2.8

1.3
2.4
1.6

6.6
2.1
3.6
3.4

.3

3.3

1.8

8.9
5.7
7.3
4.9
1.0
4.4

.7

4.9

5.5

.7
,7
,7

5.3
1.9

3.6

2.9

3.3

4.1

2.3
1.5
-.8

2.5
1.8
7.9

Actual

Dllterence

4.2
1.3
2.0
1.1

6.4
4.1
1.4

2.2
2.8

-.5

2.5

.2

-.9

1.7
1.2

-2.2

-1.1

.4

- .2
2.3

1.2
-.6
1.2

1.1
-1.0
1.8

2.0

.4
3.1

1.4
13

4.5

6.5
1.3

-.1

2.9

1.4

4.6
4.8
.6

2.2

4.4

2.2

2.0

2.7

.7

.1

.7
-1.7
1.3

-.6

1.3
1.9
.6
1.4

4.1

2.4

1.8
2.2
3.7

3.1
2.9
4.1

1.5

2.3

5.2
.3
.2

1.8
2.3

2.6
3.8
3.4

.1

1.1

2.3
3.7
3.7
1.5
.7
1.1

4.8
.8
5.0

7.1
2.7

.5
-1.7
2.1
,3'

Absolute
difference

.7

.4

3
2.1
1.1
1.0

2.3

2.3

5.0

1.9
1.0
2.7

.1
1.8

2.5

2.4

4.7
5.9

2.9
-1.1

6.9
4.6
2.9
2.4

6.0

4.1
4.1

- .5

2.4

-.1

1.1

2.1

1.3
.8
.8
1.5

1.2
4.3
1.3
2.8
2.2
,8
1.4
1.4
.8
1.0

1.0
3.4
,6
.5

2.1
2.4
2.2
2.1
1.6
1.9
.8
1.0
1.3
.4

The actual trends are least squares growth rates; the projected trends are compound interest rates.

for two-thirds of the 293 industries (71 trends for the 1970
projections and 111 trends each for the I 973 and 1976
projections). Percentage differences, however, are not the
appropriate statistic for evaluating projections because they
allow positive differences to offset the negative differences.
The unweighted absolute difference, which looks at the differences without regard to positive or negative signs, indicates that the projected trends differed by 1.3 percentage
points per year for the 1970 projections.
Another way to evaluate the projection errors is to weight
the differences between projected and actual trends by the
employment size of each industry, that is, the weighted
absolute difference. This procedure reveals that the larger
differences occurred in the smaller industries, as the weighted
differences are smaller than the unweighted differences.
A final way to evaluate the projections is to fault a projection for particularly large errors in individual industries,
the root mean squared difference. The projections contain
numerous large differences between actual and projected

These data suggest that the 1970 projections were the
most accurate, even though the projected levels (at least for
the total economy) were off by a larger margin than the
1973 and 1976 projections. The absolute differences, whether
unweighted, weighted, or squared, were smallest for the
1970 projection. In terms of employment levels, the absolute
difference was 149,000 jobs, or 15.2 percent of 1980 employment per industry, for the 1970 projections; 81,000
jobs, or 17 .0 percent, for the 1973 projections; and 62,000,
or 12.9 percent, for the 1976 projections.

trends for individual industries. This is apparent in the pre-

industries which, in tum, reflect the effects of the 1980

ceding tabulation-the squared differences are considerably
larger than the absolute differences. The largest differences
between actual and projected trends occurred in copper ore
mining, plastic materials, synthetic fibers, metal stamping,
and other transportation equipment industries, all of which
are small in terms of employment.

recession, the 1978-79 surge in oil prices, and an overestimate of domestic auto sales. These and other factors caused
employment in motor vehicles alone to decline 20 percent
between 1979 and 1980. Projected employment for 1980 in
motor vehicles was overestimated an average of 22 percent
in each of the three projections. Durable manufacturing

Major industry employment. Employment growth projections in government and manufacturing were consistently
overestimated, while employment growth in the other industries was usually underestimated. (See table 2.) The
overestimation of State and local government employment
reflects the cutbacks in government programs in the late
1970's. The high estimates for manufacturing reflect, for
the most part, overestimates of production for durable goods

15

MONTHLY LABOR REVIEW August 1984 • Evaluating Projections of Industry Employment

employment declined .6 percentage points during the 197980 period. Clearly, the recession increased the projections
errors.
Within manufacturing, projected employment in the hightech industries differed from actual employment by - l . 7
percent for the 1970 projections, 3. 6 percent for the 1973
projections, and - 3 .1 percent for the 1976 projections.
Manufacturing high-tech industries include those with a greater
proportion of technology-oriented workers than the average
for manufacturing and a ratio of research and development
expenditures to sales near or above the average for all industries. 7 The projection errors for these industries were
less than the errors for manufacturing as a whole.
The low estimates of jobs in trade and services in each
of the three projections reflect greater than expected declines
in the average workweek and less than expected gains in
labor productivity. Again, the errors for some industries are
magnified by the employment shifts that occurred between
1979 and 1980.
Industry rankings. How well did BLS project the relative
growth rates of individual industries? With each projection,
BLS attempted to characterize the fastest growing industries.
In terms of employment, BLS correctly ranked 7 of the 17
fastest growing industries in the l 970 projections; l l of 27
in the 1973 projections; and 15 of 27 in the 1976 projections.
In 1970, BLS projected that employment in office computing and accounting machines, business services, and
medical and educational services would grow the fastest of
all industries in the private sector. These were among the
fastest growing industries. As projected, employment in
office machines grew 5.0 percent per year over the 196880 period. Optical equipment and coal mining were two of
the fastest growing industries; BLS projected them to be
among the slowest.
We can examine the ability to project relative growth
Table 2.

rates across industries by calculating the correlation between
actual and projected trends. If our projections were perfect,
then the projected trends would explain 100 percent of the
variation in the actual trends-perfect correlation. The projected trends accounted for only 28 percent of the variation
in the actual trends in the 1970 projections; 33 percent in
1973; and 15 percent in 1976.
We can also examine the ability to project relative employment levels-the correlation between actual and projected 1980 employment levels. Here, the projected levels
explained more than 90 percent of the variation for each
projection. These differences in the explanatory power of
trends versus levels is to be expected because trends are
considerably more volatile in the long run.

Recession affects industry employment
The fact that BLS did not anticipate the 1980 recession
increased the difference between projected and actual trends
by 2 to 5 percentage points per industry. The projections
were not intended to be forecasts of a specific year, but
rather estimates of what the economy might look like as it
moves along a steady medium-term growth path toward full
employment. By emphasizing 1980, it appears that BLS
overestimated the medium-term trends for some industries,
for example, the auto industry where employment was expected to grow .4 percent per year over the 1968-80 period.
Auto employment declined .8 percent per year over the
1968-80 period, but grew l.4 percent per year over the
1968-78 period.
We illustrate the effects of the recession by calculating
"projections" of the 1978 and 1979 economies. The calculation applies the projected 1968-80 industry employment
trends of the 1970 · projections to the 1968-78 period to
obtain an estimate of 1978 employment, and to the l 96879 period io obtain an estimate of 1979 employment. The
following tabulation compares the mean absolute percent

Employment In 1980 In major Industries, projected and actual levels

(Numbers in thousands\

Projected In-

Industry

Total employment
Government .
Federal .... . . ........
Civilian .
Military
State and local . .

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

... . .
. ' ....
..... ' ..

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

.. .

'

.... . ' ..
. .....
.
Private
Agriculture . . . ' ....
. .... ....
Mining
....
... ' .. . ' .. . . ' . .
....
Construction
. .... ' . . .
.. .
Durable manufacturing . ....
Nondurable manufacturing
. ..
Transportation ..
. ..
.....
Communication
. . . ... . . .
Public utilities
Trade ............ ...
.. . . .
Finance and real estate
Services, except househo.lds· : · ·
Households . . .
Other gave rnment enterprises ..

16

.

.

. ..

. .
. .

.

. ..
. ..

.

.

. . . . ...
. . ' ..
.. . ..
. ..

..... . .
.

..

1970

1973

1976

101,725

104,944

104,399

19,203
5,647
2,184
3,463
13,556
82,522
2,664
451
5,546
13,167
8,974
3,085
1,116
812
19,594
4,536
18,491
2,770
1,316

18,647
4,893
2,055
2,838
13,754
86,297
2,186
537
5,286
13,757
9,294
3,278
1,304
863
21,156
5,334
20,048
1,825
1,429

18,899
5,105
2,142
2,963
13,794
85,500
2,589
599
5,384
13,167
8,753
3,037
1,318
919
21,541
5,407
19,867
1,291
1,655

Actual

Percent differences

1970

1973

1976

105,920

-4.0

-0.9

-1.4

17,914
5,126
2,207
2,919
12,788
88,006
2,860
723
5,865
12,423
8,250
3,250
1,362
966
22,493
5,702
21,097
1,598
1,501

7.2
10.2
-1.0
18.6
6.0
-6.2
-6.9
-37.6
-5.4
6.0
8.8
-5.1
-18.1
-15.9
-12.9
-20.4
-12.4
73.3
-12.3

4.1
-4.5
-6.9
-2.8
7.6
-1.9
-23.6
-25.7
-9.9
10.7
12.7
0.9
-4.3
-10.7
-5.9
-6.5
-5.0
14.2
-4.8

5.5
-0.4
-2.9
1.5
7.9
-2.8
-9.5
-17.2
-8.2
6.0
6.1
-6.6
-3.2
-4.9
-4.2
-5.2
-5.8
-19.2
10.3

errors in the employment projections of the 1978 and 1979
economies with those of the 1980 economy:
Year published
/980
1970 ................ 15.3
1973 ................ 17.0
1976 ................ 12.7

/979
14.0
12.7
9.8

/978
13.4
11.4
8.2

Impact on occupational employment. As noted earlier, industry employment estimates, along with a projected industry-occupation matrix, are the basis of occupational
employment estimates. 8 However, only the industry employment projection published in 1970 was used in the occupational employment projections. The 1973 and 1976
industry employment projections were used to estimate 1985
occupational employment.
To isolate the effect of industry estimates on occupational
projections, the industry estimates for I980 are multiplied
by actual 1980 industry staffing patterns. This yields a projection of occupations which is then compared to actual
employment by occupation.
Industry employment estimates caused some sizable errors in the projections of occupational employment in the
1970 projections. For example, professional and technical
workers' share of employment would have been overestimated by 2.8 percentage points, the hypothetical share of
19. I percent compared with the actual share of 16. 3 percent.
To a large extent, the error reflected the overestimate of
State and local government. especially employment of
teachers. However, other elements in the occupational projections offset estimates of industry employment because
the projected share of professional occupations underestimated the actual share by .8 percentage points. In the 1970
projections, the share of service occupations would have
been slightly overestimated because of the industry projections.
The following tabulation shows the mean percent error
in 1980 occupational employment projections attributed to
1970 industry employment estimates ( 1973 and 1976 estimates were not used to develop 1980 occupational employment. but are calculated here to show their implications):
Year published

Error
1970......................... 6.0
1973 ......................... -0.3
1976 ......................... -1.2

Absolute
error
12.4
4.8
4.1

Unexpected structural changes
During the 1970' s. several events substantially affected
the structure of the U.S. economy: the increases in energy
prices in 1974 and the 1979-80 period, and the increases
in food prices in response to the 1973 Russian wheat deal
and to weather conditions in 1978. These events were not
anticipated by BLS. They affected the performance and structure of the economy in several ways. The higher energy
prices. for example. were partly the cause for a considerable
deceleration in labor productivity growth. The economy was

also affected by new fiscal and monetary initiatives to control inflation.
Between 1970 and I 980, the labor force grew by 23. 7
million persons, compared with 13.4 million between 1960
and 1970, a difference of 10.3 million. The magnitude of
the 1970-80 increase was not fully anticipated by BLS and
caused a number of problems for the projections. For example, one factor in the slowdown in labor productivity was
the number of inexperienced workers entering the labor
force in the 1970's. Also, demand for several industries
grew more rapidly than anticipated (the fast-food restaurants
component of retail trade, for example).

Source of the errors
Were our errors caused by erroneous assumptions, by
incorrectly specified models, or by other factors? In essence,
were we right for the wrong reason or wrong for the right
reason? Knowing the source of our errors may help improve
future projections and will also highlight the imprecise nature of the projections. So far, the discussion has focused
on industry employment, one of the end products of the
projection process. Reaching this result involved (I) assumptions about future trends in the labor force, unemployment rate, aggregate labor productivity, and other
variables and (2) a model which depicted the structure of
the U.S. economy. The errors in projecting industry employment could have occurred because of incorrect assumptions, incorrectly specified models, random errors, or
a combination of these factors. 9

Employment. A first step in our projection methodology
is the derivation of total employment. This begins with a
projection of the labor force. The labor force, when combined with an assumed unemployment rate among civilian
workers and an assumed level of Armed Forces, yields the
number of employed persons. This number is then adjusted
for dual jobholders and other factors to achieve a projected
estimate of the number of jobs in the economy.
To determine the source of the error attributed to each
component of the employment estimate. we calculated a
series of hypothetical employment levels. For the error caused
by the labor force assumption alone, we projected total
employment as if the correct unemployment rate, Armed
Forces, and other factors were known. A comparison of this
hypothetical employment with actual 1980 employment gives
a measure of the effect of the projected labor force estimate.
For the 1970 projections, if the projected civilian labor force
had been the only error, then the projected employment
would have been 8,641,000 jobs below actual employment.
If the projected unemployment rate had been the only error.
then the projected employment would have been 3,506,000
jobs above the actual employment. Thus, for the 1970 projections as well as the other projections. these two variables
were offsetting. (See table 3.)
Supply

GNP.

Another step in the projection methodology
17

MONTHLY LABOR REVIEW August 1984 • Evaluating Projections of Industry Employment
is the estimates of supply GNP. These estimates begin with
the derivation of employment from assumptions about the
labor force and the number of unemployed persons for the
target year. Employment was combined with projected annual hours per employee to provide an estimate of total
annual hours paid. This figure, multiplied by output per
hour-aggregate labor productivity-yielded an estimate of
potential GNP. Because this estimate was derived from economic resources, it is called "supply GNP." As seen in the
following tabulation, BLS consistently overprojected the 1980
supply GNP:
GNP ( 1972

Year published
1970 ................... .
1973 ................... .
1976 ................... .
Actual 1980 .......... .

Projected
$1,729.2
1,718.9
1.607.7
1.474.0

dollars in billions)
Percent difference
17.3
16.6
9.1

To isolate the error which would be attributed to each component of GNP, we calculated a series of hypothetical GNP's.
For the error caused by the labor force assumption, we
constructed a projected supply GNP as if the correct labor
productivity, number of unemployed, and other factors were
known. A comparison of the hypothetical and actual 1980
GNP is the measure of the effect of the projected labor force
estimate.
For the 1970 projections, projected supply GNP was $255.2
million greater than actual GNP for the year 1980. (See table
4.) If the projected labor force had been the only error, then
the GNP estimate would have been $125.4 billion below the
actual figure. If the nonfarm labor productivity estimate had
been the only error, then the projected GNP would have been
$250. 8 billion too high. Because of offsetting errors in projections of the labor force. unemployment. average workweek, and other factors. the labor productivity error was
nearly the same as the total GNP error.
The largest source of error in the three projections of GNP
stemmed from overestimation of private labor productivity.
The 1973 projection assumed a considerable acceleration in
labor productivity, compared with its postwar growth. The
1970 and 1976 projections embodied only modest changes,
compared with past trends. In fact. a large deceleration in

Table 3. Factored errors In computation of total
employment
1970 pn,jecllons 1973 projections 1976 projections
MIiiions Percent Mllllons Percent MIiiions Percent

n,m
Total error . . . . . . .

. .

.

Error due to:
Civilian labor force .
Unemployment level
Armed forces ..
Adjustment factor . .
Interaction .......

4.195

100.0

976

100.0

1,521

100.0

-8,641 -206.0 -6,364 -652 0 -4,752 -312.4
3,506 359.2
3.506
83.6
2.725 179.2
581
13.8
-150 -15.4
58
3.8
16.2 2.525
678
258.7
699
46.0
-319
-7.6
-493 -50.5
-251 -16.5

NorE: Data reflect the calculation of total employment (jobs concept) with the projected
value of an individual variable and the actual value for all other variables in the employment
equation.

18

Table 4. Factored errors In computation of supply gross
national product
[Billions of 1972 dollars!

1970 projectlDIII 1973 pn,jecllDnl 1978 pn,jlcllDnl
Bllllon1 Perun! Bllllon1 Percent BIiiions Percent

Item

Total error ..........
Error due to:
Labor force . . . . . .
Unemployment level.
Adjustment factor ..
Federal government
employment ....
State and local
government
employment . . . .
Agriculture
employment ....
Agriculture
workweek ......
Nonagriculture
workweek ......
Agriculture
productivity ....
Nonagriculture
productivity ....
Interaction .......

$255.2
-125.4
56.t
16.8

100.0

$133.7

100.0

-49.1 -100.9 -41.2
53.4
22.0
22.0
13.8
6.6
33.7

-74.2
42.4
15.6

-55.5
31.7
11.7

100.0

$244.9

-6.6

-2.6

-.8

-.3

-1.4

-1.0

-5.9

-2.3

-5.4

-2.2

-9.1

-6.8

.6

.2

-3.0

-1.2

3.3

2.5

-1.6

-1.2

1.0

.4

-1.0

- .4

72.1

28.3

27.8

11.4

28.2

21.1
5.7
97.8
-5.8

9.5

3.7

10.5

4.3

7.6

250.8
-14.1

98.3
-5.5

226.4
3.8

92.4
1.6

130.7
-7.8

NOTE: Data reflect the calculation of supply GNP with the projected value of an individual
variable and the actual value for all other variables in the supply GNP equation.

labor productivity trends occurred during the 197 5-79
period.

Industry outputs. For all three projections, the absolute
difference between projected and actual industry output trends
was 2.68 percentage points per year per industry. ln onethird of the estimates, the difference between actual and
projected trends was less than 2 percentage points per year.
The absolute. unweighted, weighted. and squared differences were smallest for the 1970 projections:
Year
published
1970 . . . .
1973....
1976....

Projected
4.21
5.40
3.83

D(f
Absolute d(tl'erence
Actual ference Unll'eit?hted Weit?hted Squared
2.59 1.62
1.87
1.36
2.30
2.64 2.75
3.41
2.48
4.05
2.60 1.22
2.58
1.86
3.43

The largest overestimates of output usually occurred in
construction and durable manufacturing industries, reflecting the effects of the 1980 recession. Residential investment
expenditures dropped over the 1979-80 period and as a
result, construction output was overestimated by 30 percent
or more. During the 1970's, increases in the exploration for
oil and investment expenditures for commercial office buildings minimized errors in estimating construction activity.
Auto production was overestimated by more than 40 percent in each projection. Problems in the auto industry affected the steel, tire, and other supplying industries. The
influx of foreign steel and autos into the domestic market,
the 1980 recession, and energy-related problems were not
anticipated. The errors in estimating construction activity
affected the estimates of the cement and heating and plumbing industries. However, these errors offset underestimates
in some industries such as the optical and ophthalmic equip-

ment, computers and peripheral equipment, and electronic
equipment industries.
Industries with the largest projection errors included other
transportation equipment (motor homes, bicycles), copper
ore mining, other nonferrous ore mining, tires and inner
tubes, and primary copper products. These are small industries in terms of output.
GNP components.

The components of GNP-consumption,
investment, foreign trade, and government-were more indicative of 1979 than 1980. The difference, of course, is
because of the 1980 recession. The share of investment in
the 1980 GNP was overestimated by I . I percentage points
in the 1970 projections and 3.0 percentage points in the
1973 projections. (See table 5.) During the 1980 recession,
residential investment declined $11. 9 billion (1972 dollars)
from 1979's level, or 20 percent. The change in business
inventories dropped from a $7. 3-billion increase in 1979 to
a $5-billion decrease in 1980. If 1979 had been the target
year of the projections, the investment errors would have
only been .2 to 1.1 percentage points. Producer durable
equipment's share of GNP was also overestimated in the
three projections.
The errors in estimating consumption· s share of 1980 GNP
ranged from - I . 7 to - . 5 percentage points. If 1979 had
been the target year, the errors would have been slightly
smaller, - . I to - I. 3. For all three projections, consumption was expected to grow at about the same yearly rate as
total GNP, and this occurred. The most difficult component
of consumption to estimate was purchases of consumer
durables. In the 1970 and 1973 projections, consumer durables were expected to grow slightly slower than total consumption; the reverse occurred. Expenditures for consumer
nondurables were expected to grow modestly slower than
GNP; this pattern occurred. Expenditures for consumer services were expected to grow either at the same rate or
Table 5.

Percent distribution of demand gross national

product in 1980, projected and actual
Item

Projections
published In -

Actual

1970

1973

1976

1979

1980

100.0

100.0

100.0

100.0

100.0

61.4
9.2
24.1
28.1
17.1
4.2
7.3
4.1
1.5

62.6
(,)
(1)
(1)

62.7
9.9
23.9
28.9

63.1
9.3
24.1
29.7

Investment .
Nonresidential structures
Producer durable equipment.
Residential structures .
Inventories .

61 4
8.4
23.3
29.7
16.2
3.9
6.8
4.3
12

16.6
3.8
7.6
3.7
1.5

16.0
3.3
8.2
4.0
.5

14.1
3.3
8.0
3.2
- 3

New exports .
Exports .
Imports

.8
6.1
-5.3

.8
8.3
-7.5

1.4
8.5
- 7.2

2.5
9.9
- 7.4

3.4
10.8
-7.4

Government purchases
Federal.
State and local.

21.6
7.9
13.7

20.7
7.2
13.5

20.6
6.8
12.6

18.8
6.9
11.9

19.3
7.2
12.1

Gross national product.
Consumption
Durable goods .
Nondurable goods .
Services.

1Not

available.

slightly faster than total consumption; in reality, they grew
faster.
The three projections underestimated the export share of
1980 GNP by 2.3 to 4.7 percentage points. BLS analysts did
not anticipate the surge in the export of food and feed grains,
capital goods (except autos), and services. Even if 1979 had
been the target year, exports would have been underestimated.
The import share of GNP was reasonably accurate for the
1973 and 1976 projections, but not for the 1970 projection.
The 1970 projection was based on import growth .5 percentage points per year slower than GNP; it grew 2.5 percentage points per year faster.
Estimates of Federal Government purchases were reasonable for the three projections. State and local government
expenditures, however, were overestimated, reflecting unanticipated budgetary problems facing State and local government in response to tax amendments, such as "Proposition
13" in California and "Proposition Two and One-half" in
Massachusetts, and to the 1980 recession.

Iso/ating output errors.

Estimated output reflects several
factors-level and distribution of real GNP, projected bridge
tables, and projected input-output coefficients. The bridge
table converts the broad final demand categories, such as
consumption expenditures for durable goods, to the industries producing the items in the categories, such as electrical
appliances. The input-output coefficients represent purchasing patterns of businesses and technologies and innovations in producing goods and services. Both the bridge
and input-output tables embody assumptions concerning energy, computers, business services, and other products and
technologies.
Because of changes in input-output definitions and other
factors, it is not possible to show the projection errors for
the bridge tables and input-output coefficients. Nor is it
possible to estimate the effect that projected final demand
distribution, input-output coefficients, and bridge tables had
on the projected output trends. The combined projection
errors for these three factors increased the absolute errors
of the output projections by the same magnitude as the errors
in the projected level of GNP.
Isolating the error which would be attributed to two components involves constructing two hypothetical projections
of outputs. For the effect of the GNP level, we constructed
industry output levels which combined projected GNP and
actual industry distributions. A comparison of these hypothetical outputs with actual outputs is a measure of the effect
of projected GNP level. For the effect of final demand, inputoutput coefficients, and bridge tables, we constructed industry outputs which combined the actual GNP and the projected distribution of industry outputs. A comparison of
these second hypothetical outputs with actual output levels
is a measure of the impact of final demand and other factors.
The errors attributable to the projected distribution of
19

MONTHLY LABOR REVIEW August 1984 • Evaluating Projections of Industry Employment
outputs were small, 4 to 7 percent. However, the absolute
error attributable to the projected distribution of outputs is
nearly the same as that attributable to the projected GNP
level for the 1970 and 1973 projections.
The following tabulation shows the effects of GNP and
other factors on output estimates in private industries, except
households:
Error due to
distribution
Outeut errors Error due to GNP
Absolute
Absolute
Year
Absolute
published Percent percent Percent percent Percent percent

1970
1973
1976

28.9
29.2
18.7

32.4
33.8
25.1

20.9
22.3
11.8

20.9
22.3
11.8

4.5
5.9
6.5

17.7
19.5
18.4

Employment/output ratios. For the three projections, the
absolute difference between projected and actual labor productivity trends was 2.38 percentage points per year per
industry. In more than half of the estimates, the difference
between the actual and projected trends was less than 2
percentage points per year. The 1970 projections were the
most accurate of the three, with the lowest absolute differences, whether unweighted, weighted, or squared. Unlike
employment and output, the larger difference did not always
occur in the smaller industries in terms of employment. The
following tabulation shows projected and actual employment/output trends by industry for the private economy,
except households:
Absolute d(fference
Year
Propublished jeered Actual

Un-

D/f

ference weighted Weighted Sr111ared

1970 .... -2.92 -1.66
1973 .... -3.84 - .97
1976 .... -2.36 -.94

- 1.27
-2.87
-1.42

1.27
2.94
2.38

1.50
3.76
I. 96

2.0 I
3.76
3.26

Analysis of industry employment errors
Projected outputs times projected employment/output ratios yields projected industry employment. There are sufficient data to identify the errors for four factors-the level
and distribution of both GNP and labor productivity. (See
table 6.) The distribution of output includes the effects of
the final demand distributions. bridge tables. and inputTable 6. Factor analysis for Industry employment
Effect of projectedYear
published

Percent
1970 .
1973 .
1976 .
Absolute percent
1970 .
1973 .
1976 .

Projection
error

Output

Productivity

Level

Dlstrlbutlon

Level

Dlslrt·
butlon

0.2
5.5
2.3

20.9
22.5
11.8

4.5
5.9
6.5

-19.8
-17.9
-13.0

1.4
3.1
2.1

15.3
17.0
12.9

20.9
22.3
11.8

17.7
19.5
18.4

19.8
17.9
13.0

15.7
18.5
13.6

NOTE: For the 1970 projections. these data are the mean values for 71 industries: for
the 1973 and 1976 projections. 111 industries.

20

output coefficients. The distribution of labor productivity
r.cflects the estimated relative growth trends of labor productivity.
The data in table 6 highlight that aggregate errors in the
GNP and labor productivity levels are nearly offsetting at the
industry level. The distribution of industry outputs and labor
productivity increased the employment errors. However, the
errors resulting from the distributions of outputs and labor
productivity are about the same as the errors resulting from
the aggregate assumptions.

Alternative projections
Were BLs' projections significantly less accurate than those
of other analysts? If so, then more radical remedies and
significant chances for improvement exist. The difference
between projected and actual trends for employment were
about the same for BLS and other medium-term forecasts of
employment. BLS underestimated total employment by .9
percent in its 1973 projections and had an absolute difference
per private" industry of 10.6 percent when weighted for size
of the individual industry. In 1974, Clopper Almon of the
University of Maryland underestimated total employment
in 1980 by 3.0 percent and had an absolute difference per
private industry of 11.6 percent when weighted for industry
size. 10
In 1976, BLS underestimated total employment by 1.4
percent, and had an absolute error per private industry of
8.1 percent when weighted for industry size. In its I 976
projections, Chase Econometrics underestimated total employment by 4.2 percent and had an absolute error per private industry of 8.3 percent. 11
BLS calculation of industry projection errors is based on
111 observations; both Almon's and Chase Econometrics'
errors are based on 44 observations. Almon's and Chase
Econometrics' estimates are for full-time equivalent jobs;
BLS' are for jobs regardless of the number of hours worked.
This distinction might affect the comparison if the workweek
differed among the projections. Since the projections cited
here, Chase Econometrics, Almon, and BLS have extensively revised and expanded their models.
Like BLs', Almon's and Chase Econometrics· projections
of industry employment were based on a series of econometric and input-output models as well as judgments. However, specifications of the respective projection models differ.
The similarity in the aggregate projection error may not be
surprising because BLS' labor force projections were used
by both Almon and Chase Econometrics. All three assumed
the economy would move steadily toward full employment
and thus did not anticipate the 1980 recession. 12 The differences in total employment between BLS and the other
forecasters reflect the targeted levels of unemployment and
the adjustments between the number of employed persons
and the number of jobs.

Simpler techniques.

BLS'

projections are better than either

a simple extrapolation of past trends in industry employment
or a simple regression equation when forecasting. The following tabulation shows the absolute percent errors in industry employment projections of the 1978, 1979, and 1980
economy: 13
Year
published

Period
covered

1970 ........•.
1973 ..........
1976 ..........

1978
1979
1980

Proiections based on
GNP regressions
BLS
model Projected Actual
15.3
17.0
12.7

24.3
22.7
16.7

25.5
16.9
11.7

Time
trend
36.0
21.8
14.1

projections are based on a series of econometric and
input-output models plus judgments. One might substitute
either a time trend or a regression equation approach. A
regression approach might relate an industry's employment
to trends in GNP and the unemployment rate. A time trend
would extrapolate past trends in industry employment forward to some target year. These two alternatives are certainly naive approaches, yet they provide a useful upper
bound to acceptable projection errors.
In the preceding tabulation, two projections of 1980 employment are made with the regression technique. One uses
actual GNP and unemployment rate values; the other uses
BLS' projected 1980 GNP and the unemployment rate values.
The difference between the two projections illustrates the
effect of the aggregate errors. BLS could have correctly projected the GNP and unemployment rate but used a simple
regression model. The accuracy of this combination would
have been about the same as BLS' projections over a relatively short period, but less accurate than BLS' projections
over a longer period.
BLS'

Past evaluations, future benefits
BLS has now evaluated five industry employment projections: one each of the 1970 and 1975 economy. and three

of the 1980 economy. 14 When the time span of each projection is considered, the magnitude of the projection errors
has remained about the same across the five projections, as
shown in the following tabulation:
Year
published
1966
1973
1970
1973
1976

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

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

Year
projected
.......
.......
.......
.......
.......

1970
1975
1980
1980
1980

Absolute percent error
Unweighted Weighted
10.3
14.8
15.3
17.0
12.9

8.1
8.0
12.5
I0.6
8.1

Since the 1980 projections, BLS has expanded the industrial
detail and other aspects of the projection process. These
expansions may or may not lead to more accurate projections. One characteristic of any projection-economic, demographic, or other-is that small groups or industries are
not likely to be as accurately projected as large groups. 15
This raises the difficult question of the appropriate level of
detail for a projection. From the point of view of accuracy
alone, greater detail may impair the projection, yet the interaction of detailed industry groups may be one of the most
valued characteristics of the projection. Although the detailed industry projections may have greater errors, their use
may lead to more accurate aggregate projections.
Since these I 980 projections, BLS has also replaced its
macroeconometric model with one developed and maintained by a private company. The new model is much broader
and more detailed than the models used in the projections
evaluated here. This should foster a better understanding of
the interaction of many economic trends. However, projections must at some place in their structure hold change
constant, whether it is the elasticities of substitution between
income and expenditures or the concept of full employment.
And yet the structure of the economy is continuously changing. This has the effect of underestimating the degree of
change. 10
D

--FOOTNOTES-1
The initial projections of the 1980 economy were described in · "The
U.S. economy in 1980: a preview of Bl.s projections.·· Mo111h/_r Labor
Re,·iell'. April 1970. pp. 3-34. and in Pa11ern.1· of U.S. t:rnno111ic Grml'th.
Bulletin 1672 (Bureau of Labor Statistics. 1970). The second projections
of the 1980 economy were described in ··Projections of GNP. income.
output. and employment.·· Mo111h/_1· Labor RerieH·. December 1973. pp. 2742: The U.S. Eco11//111_1· in /9/i5. Bulletin 1809 (Bureau of Labor Statistics.
1974): and The Stmcture of the U.S. Ec//1101111· in /9/iO a11d /9/i5. Bulletin
1831 (Bureau of Labor Statistics. 1975). The third projections were described in Ronald E. Kutscher. ··Revised GNP projections to 1980 and
1985: an overview ... Monthl_r Labor Rel'iell'. March 1976. pp. 3-8: Charles
T. Bowman and Terry H. Morlan. ··Revised projections of the U.S. economy to 1980 and 1985 ... Mo111hl_r Labor Rnie,r. March 1976. pp. 9-21:
Thomas J. Mooney and John H. Tschetter. ··Re~iscd projections to 1985. ··
Month/_r Labor Rel'ie,r. November 1976. pp. 3-9: and Max L. Carey.
··Revised occupational projeclions to 1985. ·· Molllhlr Labor Rnie,r. Nowmbcr 1976. pp. 10-22.
·

~sec Howard N Fullerton. Jr.. ··How accurate were the 1980 labor force
projections'?·· M()llfhl_r La/,()r 'Re,·ie,r. July 1982. pp. 15-21: and Max

Carey and Kevin Kasunic. ··Evaluating the 1980 projections of occupational employment.·· Momh/_r Labor Re,·ieH·. July .1982. pp. 22-30.
'The methodology for the 1970 projections is described in Patterns (If
U.S. t:co110111ic Groll'th. Bulletin 1672 (Bureau of Labor Statistics. 1970:
for the 1973 and 1976 projections. The Structure of the U.S. Econom,· in
19/iO and /9X5. Bulletin 1831 (Bureau of Labor Statistics. 1975).
'In this tabulation and throughout the article. the actual rates are based
on least squares growth rates: projected trends are based on compound
interest rates.
; See Fullerton. "How accurate were lhe 1980 labor force projection~'?'·
0

In this tabulation and throughout the article. for the 1970 projections.
the data are the mean values for 71 industries: for the 1973 and 1976
projections. 111 industries.
7
Richard W. Riche. Daniel E. Hecker. and John U. Burgan. ·•High
technology today and tomorrow: a small slice of the employment pie.··
Mo11thl_,. Labor Rnie,r. November 1983. pp. 50-58. The authors give
three delinitions of high technology: this evaluation uses the third.

21

MONTHLY LABOR REVIEW August 1984 • Evaluating Projections of Industry Employment
x For an evaluation of the occupational projections to I980. sec Carey
and Kasunic. "Evaluating the 1980 projections of occupational employment."
''The ideal way to analyze the sources of error would be to have macro
and input-output models plus 1980 data which arc consistent with the
models and data used in each of the projections. Then one could. in turn.
examine the effect of each model and assumption. However. this is not
feasible becau,e of changes in the definitions and concepts of input-output
tables. revisions in National Income Accounts. and changes in the system
for classifying industries. The di,cussion in this section is an approximation
of the ideal. As will be apparent. assumption and modeling errors appear
to be about e4ual in magnitude and usually are offsetting.
The projections assume there arc no changes in the income and price
elasticities of the foreca,ting model. Such an assumption is obviously
unrealistic as the forecast horizon lengthens and for changes during a
projected period which exceed the changes during a historical period. The
unanticipated events discussed earlier are indicative of these probkms.
'"Clopper Almon. Margaret B. Buckler. Lawrence M. Horwitz. and
Thomas C. Reimbold. /91'i5: /111ai1u/11.1'/1T.fi1rernst.1· o(lhe American ern110111r (Lexington. Mass .. Lexington Boob. 1974).

11
Chase Econometrics, U.S. macroeconomic long-term.fiJrecasts . .fimrth
quarter 1976.

12

Almon and others. /9i'i5. p. 5.

One regression uses actual GNP and the unemployment rate to project
employment: the other uses projected GNP and the unemployment rate.
1.1

1
• For evaluations of the production and employment projections for the
1975 employment and output projection. see Paul T. Christy and Karen
1. Horowitz. "An evaluation of BLS projections of 1975 production and
employment." Mo111hlr Labor Rn·iell'. August 1979. pp. 8-19: for the
1970 projections. sec Valerie A. Pcrsonick and Robert A. Sylvester. "Evaluation of !!LS economic and employment projections." Mo111hlr Labor
Rel'ie11·. August 1976. pp. 13-26.

"Henri Theil. Applied t,'rnnometric Forecastin)i (Chicago. Rand-McNally
and Co .. 1966).
"' Jacob Mincer and Victor Zarnowitz. "The Evaluation of Economic
Forecasts." in Jacob Mincer. ed .. Economic Forecast.,· and Expectations:
Ana/rses of' f'orecasting Beh111·ior and Perfimnance (New York. National
Bureau of Economic Research, 1969). pp. 3-46.

A note on communications
The Monthly labor Review welcomes communications that supplement,
challenge, or expand on research published in its pages. To be considered
for publication, communications should be factual and analytical, not polemical in tone. Communications should be addressed to the Editor-inChief, Monthly labor Review, Bureau of Labor Statistics, U.S. Department
of Labor, Washington, D.C. 20212.

22

Research
Summaries
Are unions facing a crisis?

labor officials are divided
BRIAN HESHIZER AND HARRY GRAHAM

Union officials are concerned about the future of the labor
movement, according to results of a recent survey. They
recognize that they face a period marked by economic,
technological, social, and political changes which do not
augur well for organized labor. While the leaders who responded to the survey do not ex.press panic over this prospect, they definitely see a period of difficulty ahead.
In 1963, a survey was sent to all national and international
union presidents and union research and education directors
listed in the Bureau of Labor Statistics directory of labor
unions. 1 Of the 339 questionnaires sent, 85 responses were
obtained, a response rate of 25 percent. In 1983, the same
questions were asked of presidents and research directors
of unions and employee associations listed in the 1979 Bureau of Labor Statistics directory of labor organizations. Of
the 212 questionnaires sent, 79 usable responses were received, a response rate of 37 percent. 2
Responding organizations represented unions and employee associations covering the broad spectrum of the labor
movement. The size of responding unions ranged from several thousands to hundreds of thousands in membership.
Most of the respondents were from traditional strongholds
in manufacturing and nonmanufacturing and from the public
sector, although unions in other sectors of the economy were
represented as well. Employee associations in the respondent group came from States with strong union movements
and with public sector bargaining laws. The size and industry distribution of respondents indicate that the sample
is representative of labor organizations.
Responses to both the 1983 and 1963 surveys are shown
in table I. Using paired comparison t-tests, statistically significant differences at the .05 level, or lower, of significance
were found for several questions, indicating a shifting of
opinion among union officials on some important issues. In
1983, 62.7 percent of the union officials surveyed believed
members do not know what their union does for them, up
from 53.6 percent in 1963. Interestingly, a study by Thomas
Brian Heshizer is an assistant professor and Harry Graham is a professor
of management and labor relations, Cleveland State University.

Kochan shows that union members expect their union to
perform at a higher level than what they perceive their
unions to be providing. 3 Whether this reflects unrealistic
expectations by union members is speculative, but union
officials might interpret those results in that fashion.
Union officials perceive a general weakening of labor's
power, compared with 20 years ago, saying that labor's
social impact and collective bargaining power is weaker
today. More of the current respondents strongly believe that
economic considerations have limited their ability to improve the well-being of members. The effect of general
economic conditions is also registered on the question of
the employer's ability to pay. In the current survey, a significantly higher proportion of union officials recognize that
this factor should be taken into account in bargaining.
Union officials believe that opportunities for advancement
in union hierarchy are better today than they were in the
earlier survey. They also believe the AFL-CIO should not
coordinate activities such as organizing to any greater extent
than it already does. (Only 15 percent agreed or strongly
agreed that the '' Federation should have rnore authority over
organizing," a big change from 1963 when 25 percent
agreed.) 4 They say labor is better prepared today to meet
the problems posed by automation and economic change.
In both surveys, a large majority of respondents agreed
that the labor movement has suffered from a lack of vitality
in recent years. However, union officials do not see this as
emanating from union leadership. In 1983, a larger proportion believed that such values as dedication and idealism
are widely held, but still a sizable minority (37 percent in
1983 and 44 percent in 1963) believed such values are not
widely held among union officials.
The responses indicate that union officials see the barriers
to union growth as coming from outside the labor movement. A smaller proportion in I 983 believed that priority
should be given to organizing white-collar workers over
blue-collar workers. 5 Little change occurred in the beliefs
of union officials on the need to establish links between
organized labor and nonlabor reform groups. The entreaties
of commentators from outside the labor movement who have
called for re-establishing and forging new connections with
nonlabor reform groups apparently have not shifted the views
of union officials. 6 Nor was there any change in attitudes
about borrowing ideas from foreign labor movements. Only
25 percent of the current respondents disagreed that unions
were doing all they could to bring blacks into the movement,
23

MONTHLY LABOR REVIEW August 1984 • Research Summaries

Table 1. Union offlclals' attitudes about the labor movement
Survey
date

Question

...... ' .........

1983
1963
1983
Need for formal opposition within unions ...
1963
1983
Lack of vitality in labor movement ........
1963
1983
Need for government involvement in internal union affairs ....
1963
1983
Leader-held values of self-sacrifice, idealism, and dedication ....
1963
1983
Less upward mobility in union hierarchy today
1963
1983 ..
More stress on organizing white-collar workers
1963
1983 . .
Future of labor movement is secure ...........
1963
Internal problems are weakening the ability of labor union growth ..
1983
1963
Most important force behind social legislation
1983
1963
1983
Federation should have more authority over organizing
1963
Labor's collective bargaining power is weaker today .
1983
1963
Too much political involvement, put more stress on collective
1983
1963
bargaining
Structure not adequate to meet challenge of robotics, automation ..... 1983
1963
1983 ....
Disregard economic situation of company in bargaining
1963
1983
Not enough influence on foreign policy
1963
Should borrow more from European labor unions
... 1983
1963
Closer ties with nonunion reform groups .
1983
1963
Unions doing all they can to bring blacks into the ranks
1983
1963
Economic conditions weakening ability to get better wages
and benefits
1983
1963

Members do not understand what union does

1Attitudes were scaled from 1 to 5, with "strongly agree" equaling 1 and "strongly
disagree," 5. The mean is the average value for responses to the question.

compared with 40 percent in 1963.
About one-fourth of the current officials were confident
about the security and status of the labor movement, compared with slightly more than one-fifth in I963. Nonetheless, a majority of both current and past respondents disagreed
that the "future of labor movement is secure," indicating,
perhaps, that organized labor does not feel accepted in this
country. However, with the difficulties caused by a weak
economy and an increase in employer opposition to unions,
. the extent of agreement with the statement when compared
with conditions at the time of the 1963 survey could indicate
a more self-confident labor movement.

Specific problems
Respondents were asked to comment on several questions
on labor's problems and their causes. The responses are
presented in table 2.
In 1983, 51 percent of the respondents believed there was
a crisis in the American labor movement. The most frequently identified problems causing the crisis were union
policies and structure, "antilabor" government policies,

24

Strongly
agree

Agree

Neither
agree nor
disagree

Disagree

Strongly
disagree

Mean1

29.4
10.7
5.1
8.3
29.1
27.4
7.6
10.7
10.1
9.5
3.8
9.4
10.1
25.9
6.3
2.4
7.6
9.8
26.9
38.6
2.6
8.6
11.4
12.2
8.9
6.0
15.2
16.9
2.6
3.7
8.9
10.8
7.6
4.8
5.1
6.1
13.9
12.9

33.3
42.9
22.8
21.4
44.3
36.9
32.9
37.6
38.0
33.3
19.2
30.6
25.3
27.1
20.3
18.3
20.3
17.1
42.3
45.8
12.7
25.9
59.5
39.0
7.6
7.2
26.6
37.3
10.3
25.6
39.2
51.8
17.7
19.3
40.1
42.7
32.9
34.1

10.3
9.5
32.9
16.7
8.9
6.0
10.1
3.6
15.2
13.1
16.7
10.6
43.0
13.0
17.7
12.2
12.7
8.5
16.7
6.0
25.3
13.6
12.7
11.0
10.1
3.6
12.7
19.3
19.2
13.4
30.4
8.4
38.0
22.9
27.8
20.7
27.8
11.8

23.1
26.2
20.3
39.3
12.7
23.8
27.8
24.7
24.1
33.3
47.4
36.5
12.7
31.8
36.7
48.8
34.2
42.7
9.0
9.6
38.0
34.6
12.7
32.9
34.2
37.3
36.7
24.1
50.0
47.6
17.7
20.5
26.6
38.6
17.7
23.2
17.7
28.2

3.8
10.7
19.0
14.3
5.1
6.0
21.6
23.5
12.7
10.7
12.8
13.0
8.9
2.4
19.0
18.3
25.3
22.0
5.1
0.0
21.5
17.3
3.8
4.9
39.2
45.8
8.9
2.4
18.0
9.8
3.8
8.4
10.1
14.5
8.9
7.3
7.6
13.0

22.38
22.83
3.25
3.30
2.20
2.40
3.23
3.13
2.91
2.98
23.45
23.16
2.85
2.58
3.42
3.63
3.49
3.50
22.23
21.87
23.63
23.26
22.38
22.79
3.87
4.10
22.97
22.58
23.71
23.34
2.68
2.64
3.14
3.39
2.85
2.83
2.72
2.94

32.9
10.7

60.8
54.8

0.0
10.7

5.1
19.0

1.3
4.8

21.81
22.52

2Mean significantly different at .05 level or below.

and labor's public image. Only a few mentioned the economy and union leadership. Automation and unemployment
were not even mentioned, unlike in 1963 when half of the
respondents said these were the main problems. One respondent commented, "We live in an anti-union environment ... a period of extreme uncertainty politically and
economically'' which has hurt the labor movement. A union
president said unions were often perceived as ''standing in
the way of progress," and employers have used this to
weaken unions. Another remarked that business "refuses
to accept labor as a partner,'' unlike the situation in other
Western industrialized countries, echoing a comment made
in the 1963 survey.
Membership, bureaucracy, and leadership apathy have
made it difficult for labor to respond positively to an economic situation that has eroded union strength in basic industries. While several respondents called for expansion of
membership in the growing service sector and among whitecollar workers, their comments evinced little in the way of
optimism. A union official remarked that union membership
has declined because "it [labor] did too good a job ofraising

the standard of living of its members [who] are now complacent. In raising the standard of living for its members,
other segments of society have been pulled along [and] these
segments see no reason now to unionize."
The internal causes of organized labor's problems were
identified as leadership and union policies and ideology.
Respondents cited arrogance, inability to prepare successors, dogmatism, adherence to outdated ideas, and shortsightedness as leadership problems. The overall tone of this
line of criticism is illustrated by this comment: "[the labor]
movement has lost its role as a cause for many [leaders]
and is simply a job. Many are more interested in holding
union office for money and power and not to effect significant change. Union leadership worries about their reelection more than anything."
Several respondents criticized union policy, or the ''lack
of philosophy, "as an internal cause of labor's problems.
One noted that organized labor "has not been able to persuade the majority of workers of the worth of unions." The
unions emphasize "the more, more, more philosophy instead of planning for the future.'' Too often, the unions
come across to the public with an attitude of ''to hell with
the consumer. We want what we want or we'll cripple the
economy.'' Another believed that corruption still tainted the
labor movement and that "unions need to purge corruption
with the same effort they fight arbitrary management.'' Yet,
as one official wrote, even when labor has made "substantial
inroads into solving the problem of ... corruption," the
public perception remains negative.
In the 1963 survey, respondents listed three main external
causes of labor's problems: antiunion propaganda, unsympathetic government policies, and technological change and
unemployment. Respondents in the current survey view the
external causes as emanating from similar sources with some
differences, though, in emphasis. Those who see government policy as an external cause mention the inability of
the National Labor Relations Board to get compliance for
some of its orders, the "hostile" administration of the National Labor Relations Act, the Railway Labor Act, and the
Table 2. Respondents Hating specific problems In labor
movement, 1983
Hem
Percent
11 there

I

crl1l1 In the labor movement?

Agreeing

51

Whit .,. the problem• CIUllng Ille crl1l1?

Government policies and legislation
Labor's public image ........... .
Union structure and administrative policies
Union leadership
Economy

39
21
47
11
18

What are the Internal causn ol organized labor problems?
Leadership problems .
Structural problems ..
Policies and ideology

45
15
39

What are Ille ellemal cauus ol organized labor problems?
Economic conditions/changes .
Government policy
Management hostility

39
25
36

Landrum-Griffin Act, and the proliferation of State and local
laws that hamper labor's effort to organize and represent
workers.
Current respondents see economic conditions and managerial attitudes often acting jointly to trouble the labor
movement. Several commented that "there is . . . in the
establishment . . . a concentrated effort to downgrade unions''
by taking every opportunity to create a ''public image of
. . . unions as corrupt manipulators who steal . . . dues and
cause all of a company's problems .... " They believe
that employers have taken advantage of the weakened economy, especially in basic industries, to close unionized plants
and move elsewhere. The activities of antiunion groups
spreading propaganda against unions and the use of union
busters by management have made organizing and maintaining existing bargaining units more difficult.
The economy, one union official said, has served as a
battering ram that companies have used to break collective
bargaining relationships. That along with the transition to
a service economy has "eliminated thousands of traditional
union jobs." The exasperations of many respondents were
summarized by one official: ''The unions are blamed for
productivity problems-why doesn't anyone ... chide the
corporations for failing to modernize instead of paying stock
dividends."
All the respondents who answered the question on external causes saw such causes as serious threats to the labor
movement. One referred to the conjoining of these forces
as a "debacle" for the labor movement. The broader implication of the weakening of organized labor is summed
up in this comment: "[Unionism] has been the underpinning
of middle-class achievements. We seem to currently be
moving to a bipolar structure which will weaken further the
middle class as changes occur in the economy. The effect
of this ... is yet to be seen ... but is frightening." D
--FOOTNOTES-' Bureau of Labor Statistics, Directory of National and International
Labor Unions in the United States. /96/, Bulletin 1320 (Washington, U.S.
Government Printing Office, 1962). For an account. of the 1963 survey,
see Solomon Barkin and Albert A. Blum, "Is There a Crisis in the American Trade Union Movement?-The Trade Unionists' Views," The Annals
of the American Academy of Political and Social Science, November 1963,
pp. 16-24.
2
Bureau of Labor Statistics, Directory of National Unions and Employee
Associations, 1979, Bulletin 2079 (Washington, U.S. Government Printing
Office, 1980). The post office was unable to deliver questionnaires to 30
labor organizations. This reduced the sample to 106 unions and associations.
'See Thomas Kochan, "How American workers view labor unions,"
Monthly Labor Review, April 1979, pp. 28-30.
4
See Derek Bok and John Dunlop, Labor and the American Community
(New York, Simon and Schuster, 1970), pp. 194-96 for an analysis of
the problems of federation-sponsored organizing which supports this conclusion.

~The means for the statement, "more stress on organizing white-collar
workers," were almost significant at the . 10 level; the calculated t value
for the means was .13.
6

Bok and Dunlop, Labor, pp. 31-34; and H.W. Benson, "Labor Leaders, Intellectuals, and Freedom in the Unions," Dissent, vol. 20, Spring
1973, pp. 206-19.

25

MONTHLY LABOR REVIEW August 1984 • Research Summaries

Preferences of temporary workers:
time, variety, and flexibility
MARTIN

J.

GANNON

Part-time work-defined as less than 35 hours per weekis becoming increasingly important in the United States.
Before World War II, only a negligible number of workers
were classified as part time. Since then, the proportion of
the civilian work force classified as part time has hovered
around 18 percent. During economic upturns, this percentage tends to decrease, as many individuals desiring fulltime employment are forced to work part time during recessionary periods. 1 Still, the percentage varies by only a
few points from 1946 to I 983, as the majority of part-time
workers do not want full-time jobs.
Within the part-time work force, temporary help constitutes a significant subgroup. Firms in the temporary help
industry, such as Manpower and Kelly Services, send out
their employees to complete assignments in various organizations. Afterwards, the employees return to the temporary help firms until additional assignments materialize.
Hence, the workers are employees of the temporary help
firms and not of the companies where they work. In 1956,
there were only about 20,000 employees in this industry. 2
Today it is estimated that from 2 million to 3 million workers
are employed as temporaries at some time-often for only
a few hours, but more frequently for several days over a
period of 3 or 4 months-during each year. 3 The number
of temporary employees will probably increase substantially, because the industry provides job opportunities that
do not require a full-time work commitment and, at the
same time, helps businesses to solve many staffing problems, such as the need for additional workers during busy
periods.
It should be emphasized that it is somewhat difficult to
classify temporary help as either full time or part time,
because many are seeking a full-time position, but only for
a short while. However, the vast majority of these workers
are employed less than 35 hours per week, as the temporary
help firms typically do not have enough work to provide
full-time employment opportunities. 4

Areas of study
This study focuses on two aspects of temporary help. The
first is the specific time preferences of temporary employees,
that is, when do they want to work. In contrast to the few
previous studies, 5 it provides a relatively exhaustive analysis
of these time preferences: days of the week, time of the
day, and time of the year.
A second area concerns the relative importance (to the
Martin J. Gannon is a professor. Faculty of Organizational Behavior and
Industrial Relations. College of Business and Management. University of
Maryland.

26

employee) of flexibility in hours of work versus variety in
the work or frequency in changing assignments. Some authors have argued that temporary employment is particularly
attractive because it allows for variety in work. 6 However,
the counterargument can also be made, that is, that the
predominant reason for seeking this form of work is flexibility in scheduling hours, especially for working wives who
may view work as subordinate to familial activities. Previous research suggests that flexibility and variety are independent dimensions or reasons for desiring temporary
employment. 7
Finally, it should be noted that the temporary help industry is generally considered to consist of three major sectors, and the percentage of employment in each of these
sectors is estimated to be about 65 percent in the clerical/
secretarial area, 30 percent in the industrial area, and 5
percent in the technical/professional area. 8 Previous studies
rarely, if ever, go beyond a comparison of employees in
the clerical/secretarial area and the industrial area. This study
cuts across the three sectors, by focusing on the relationship
between skill level and the issues of variety/flexibility and
time preferences among employees in the medical temporary
help area, which is the fastest growing segment of the temporary help market.

Method of analysis
The study took place in a large, national firm that hires
more than 50,000 health-care temporary employees each
year. Four groups of workers were selected for intensive
study: registered nurses, licensed practical nurses/licensed
visiting nurses, nurses' aides, and homemakers. These groups
were chosen because they represent the major occupations
of the firm. More importantly, the skill level of each group
is very distinct and decreases in the following order: registered nurses, licensed practical nurses/licensed
visiting nurses, nurses' aides, and homemakers. Hence, it
was possible to study the relationship between skill level
and time preferences of the employees, all of whom were
women.
Questionnaires were sent to 1,393 employees and the
overall response rate was 79 percent, or I, IO l respondents. 9
The following tabulation shows the distribution of questioimaires among the occupations and the corresponding
response rates (in percent):
Received
questionnaire
Registered nurses . . . . . . . . . . . . . . . .
340
Licensed practical nurses/
licensed visiting nurses . . . . . . . .
275
Nurses' aides . . . . . . . . . . . . . . . . . . . . 517
Homemakers . . . . . . . . . . . . . . . . . . . .
261

ReJponse
rate
77.0

80.5
79.0
80.5

To analyze the issue of time preferences, the respondents
were asked to provide specific information on several aspects of their work preferences. For example, the respondents indicated whether they preferred to work some days

of the week rather than others. Three answers were possible: yes, definitely; yes, but my preference depends on
such factors as family responsibilities and time of the year;
and no, I have no preferences.
The respondents were also asked to indicate -which days
of the week they preferred to work on a regular basis; they
could select as many days as they desired. Hence, it is
possible to analyze time preferences by day of the week and
by total number of days per week the employees preferred
to work.
The respondents were then requested to indicate which
times of the day they preferred to work-morning, afternoon, evening, and night. Again, there was no limit on the
number of categories that could be selected. As in the case
of the days in the week, it is possible to analyze both the
actual times of day and the total number of times per day
that they preferred to work. In addition, the respondents
were asked to indicate whether they wanted to work during
a specific time or times of the day, and three responses were
possible: yes, definitely on a regular basis; yes, but my
preference depends on such factors as family responsibilities
and time of the year; and no, I have no preferences.
To determine why individuals wish to become temporary
help employees, the respondent was asked to identify her
most important reason for choosing to work for the firm.
The following choices were provided: (I) variety in work,
that is, frequent changes in assignment; (2) a stopgap measure until I can obtain a permanent job; (3) freedom to
schedule my work in a flexible manner; (4) employment
during school vacations; and (5) other.
Chi square was used to analyze the relationships. This
statistical test measures whether two discrete variables are
independent of or related to one another.

Survey results
There was a significant relationship between the respondents' skill levels and their preferences to work some days
of the week over others. (See table l.) In particular, 55.0
percent of the registered nurses and 44.2 percent of the
licensed practical nurses/licensed visiting nurses, but only
30.8 percent of the nurses' aides and 27.0 percent of the
homemakers indicated that their preference depends on such
factors as family responsibilities and time of year.
Table I also profiles the specific days of the week that
the respondents desired work. Because the respondents were
allowed to check as many days as desired, it was not possible
to use chi square. However. 31. 1 percent of the registered
nurses, 32.2 percent of the licensed practical nurses/licensed

visiting nurses, 22.8 percent of the nurses' aides. and 14.4
percent of the homemakers preferred Sunday. Thus, skill
level was positively associated with the desire to work Sundays.
An important relationship was also established between
skill level and the total number of days that the respondents
preferred to work each week. (See table l.) Only 44.2

percent of the registered nurses and 56. l percent of the
licensed practical nurses desired 5 days or more per week,
while 70.9 percent of the nurses' aides and 66.3 percent of
the homemakers were of a similar persuasion.
In addition, table l indicates that there was a significant
correlation between skill level and the preference to work
a particular time or times of the day (morning, afternoon,
evening, or night). First, as skill level rose, there was an
increase in the desire to work during a certain part of the
day, and the preference depended on such factors as family
responsibilities and the time of year. More specifically, as
skill levels rose, so did the preference to work in the evening
and at night. (Again, because the respondent could check
as many times as she preferred, it was impossible to compute
chi square.)
The relationship between skill level and the total number
of preferences for a particular time or times of working
during the day (morning, afternoon, evening, or night) is
significant only at the . IO level. Still a significant proportion
of all four work groups, regardless of skill level, prefer to
work only during one time of the day (morning, afternoon,
evening, or night).
Table 2 contains information on the issue of variety and
flexibility. Only 16.6 percent of the entire sample cited
Table 1. Relationship between skill level and preferences
of temporary help employees
Skill

lev■ I

(high to lowl

Reglstered
nurses

Licensed practical
nurses/licensed
vlstlng nurses

Nurses·
aides

Homemakers

31.0

38.1

50.1

44.5

55.0
14.0

44.2
17.7

30.8
19.1

27.0
28.5

Specific days preferred on
regular basis:
Saturday.
Sunday ..
Monday
Tuesday
Wednesday.
Thursday .
Friday .

33.2
31.1
62.1
65.8
67.9
65.3
58.9

28.1
32.2
74.9
70.8
77.8
74.3
66.1

29.9
22.8
79.2
80.6
80.3
78.9
78.9

23.0
14.4
79.2
84.8
77.5
84.3
74.7

Total number of days each
week preferred to work: 2
One.
Two.
Three
Four .
Five.
Six
Seven .

3.2
21.1
15.3
16.3
37.9
3.7
2.6

2.3
8.8
20.5
12.3
44.4
7.0
4.7

1.4
7.7
11.7
8.3
60.3
8.3
2.3

2.8
9.6
9.0
12.4
57.9
4.5
3.9

Preference to work a
certain time of day: 3
Yes, on a regular basis

56.9

61.6

57.8

54.5

37.7
5.4
64.4
54.1
35.1
28.2

32.9
5.6
58.5
47.5
35.0
31.5

29.6
12.6
69.5
49.9,
26.3
22.7

25.7
19.8
82.2
58.9
21.7
15.0

Preferences

Preference to work some
days over others: 1
Yes, definitely
Yes, but depending on
family responsibilities
and time of year .
No preference .

Yes. but depending on

family responsibilities
and time of year .
No preference .
Morning .
Afternoon .
Evening
Night .
1Chi

2Chi

square= 61.70 (p,;; 001).
square = 80.71 (p,;; .001).

3Chi square = 37.79 (p,;; .001).

27

MONTHLY LABOR REVIEW August 1984 • Research Summaries
variety as the most important reason for becoming a temporary help employee, while 60.2 percent chose freedom
to schedule work in a flexible manner. In addition, there
was a significant relationship between skill level and the
most important reason for working as this type of employee
(p :,;;; .001). The two groups highest in skill level, registered
nurses and licensed practical.nurses/licensed visiting nurses,
cited freedom to schedule work flexibly much more frequently than did the two groups lowest in skill level, nurses'
aides and homemakers. The opposite pattern emerged on
the dimension of variety in work, that is, the two groups
lowest in skill level cited this reason much more frequently
than did the two groups highest in skill level.

Conclusions
Previous research has suggested that temporary help firms
experience great difficulty obtaining employees during vacation periods. 10 This study confirms and extends this generalization to indicate that this difficulty will be exacerbated
at particular times of each day and each week as skill level
rises.
This finding is important in view of the fact that the
technical/professional sector of the temporary help industry
possesses great potential for expansion, and that industry
needs a great number of highly skilled and educated workers. However, because the higher-skilled workers are less
available than the lower-skilled workers, there will probably
be a great amount of unmet demand in the marketplace.
As expected, the most unpopular times of the day to work
are in the evening and at night, and on weekends. It is
during such times that many temporary help firms must deny
customer requests for workers. 11 Hence, such firms may not
be able to expand into new markets because of the limited
availability of employees.
Table 2. The relationship between skill level and the most
Important reason for working at this temporary help firm
[In percent]

Reason

(1) Variety in work,
that is,
frequent
changes in
assignment ..
(2) A stopgap
measure until
I can obtain a
permanent
job ...
(3) Freedom to
schedule my
work in a
flexible
manner .
(4) Employment
during school
vacation. ..
(5) Other.
NOTE: Chi square

Entire
sample

Reglstered
nurses

licensed
practical
nurses/IIcensed visitIng nurses

Nurses·
aides

Homemakers

16.6

8.8

13.1

21.5

20.6

8.2

9.2

8.0

7.2

9.0

60.2

70.3

65.3

55.9

50.8

1.0
14.1

.8
10.9

0
13.6

.8
14.6

2.6
16.9

Why are the higher-skilled employees less available? Previous studies have shown that the rate of moonlighting among
the more skilled workers is significantly greater than among
those of lower skill. 12 In effect, many of these workers
appear to be using temporary help employment as a second
job. Another possible reason for limited availability of highskill workers may be that they possess greater financial
resources than those having lower skills and hence do not
need temporary work as much.
The present study also clarifies the concept that the temporary help employee is seeking a full-time job, but only
for a short period of time. 13 The majority want to work 8
hours per day. However, a significant minority of these
workers desired work for only one time of the day, regardless of skill level. Thus, it appears that many of these employees are seeking employment for a short period of time,
but employment involving only 4 hours per day.
Finally, the research indicates that flexibility in scheduling is a much more important source of motivating individuals to apply to a temporary help firm than is variety,
at least in terms of frequencies. The study also shows that,
the higher the skill level, the greater the probability of citing
flexibility in scheduling as the most important reason for
becoming a temporary help employee.
D
--FOOTNOTES-' Robert Bednarzik, "Short workweeks during economic downturns,"
Monthly labor Review, June 1983, pp. 3-11.
2 Mack Moore, The Role of Temporary Help Services in The Clerical
Labor Market, Ph.D. diss. (Madison, University of Wisconsin, 1963).
3 Martin J. Gannon, "An Analysis of the Temporary Help Industry,"
Labor Market Intermediaries, Special Report No. 22 (Washington, National Commission for Manpower Policy, March 1978), pp. 195-255.
4
Gannon, "An Analysis of the Temporary Help Industry."
5
W. Albeda and G. Veldkamp, eds., Temporary Work in Modern Society. Part 2: Temporary Work within a Socio-Economic Framework (The
Netherlands, Kluwer, 1978).
6
Genmaine Greer, The Female Eunuch (New York, McGraw-Hill, 1970).
See also Alvin Toffler, Future Shock (New York, Random House, 1970).
7
Richard Leone and Donald Burke, Women Returning to Work and Their
fllleraction With a Temporary Help Service (Springfield, Va., National
Technical Infonmation Service, 1976).
"Gannon, "An Analysis of the Temporary Help Industry."
9
This response rate was substantially higher than that reported in most
previous studies, possibly because a dollar was attached to each questionnaire.
HlLeone and Burke, Women Returning to Work.
11
Informal interviews with executives in this industry confinm this trend.
12
Gannon, "An Analysis of Temporary Help."
13
Moore, The Role of Temporary Help Services; and Leone and Burke,
Women Returning to Work.

Pay gains tempered

in basic steel mills
NORMA

~

40.26 (p.;; .001).

W. CARLSON

The gain in steelworkers' pay lagged behind that of all
workers in the durable goods manufacturing industries, according to a Bureau of Labor Statistics occupational wage

28

- - - - - - - - - - - - -----·

MONTHLY LABOR REVIEW August 1984 • Research Summaries
variety as the most important reason for becoming a temporary help employee, while 60.2 percent chose freedom
to schedule work in a flexible manner. In addition, there
was a significant relationship between skill level and the
most important reason for working as this type of employee
(p :,;;; .001). The two groups highest in skill level, registered
nurses and licensed practical.nurses/licensed visiting nurses,
cited freedom to schedule work flexibly much more frequently than did the two groups lowest in skill level, nurses'
aides and homemakers. The opposite pattern emerged on
the dimension of variety in work, that is, the two groups
lowest in skill level cited this reason much more frequently
than did the two groups highest in skill level.

Conclusions
Previous research has suggested that temporary help firms
experience great difficulty obtaining employees during vacation periods. 10 This study confirms and extends this generalization to indicate that this difficulty will be exacerbated
at particular times of each day and each week as skill level
rises.
This finding is important in view of the fact that the
technical/professional sector of the temporary help industry
possesses great potential for expansion, and that industry
needs a great number of highly skilled and educated workers. However, because the higher-skilled workers are less
available than the lower-skilled workers, there will probably
be a great amount of unmet demand in the marketplace.
As expected, the most unpopular times of the day to work
are in the evening and at night, and on weekends. It is
during such times that many temporary help firms must deny
customer requests for workers. 11 Hence, such firms may not
be able to expand into new markets because of the limited
availability of employees.
Table 2. The relationship between skill level and the most
Important reason for working at this temporary help firm
[In percent]

Reason

(1) Variety in work,
that is,
frequent
changes in
assignment ..
(2) A stopgap
measure until
I can obtain a
permanent
job ...
(3) Freedom to
schedule my
work in a
flexible
manner .
(4) Employment
during school
vacation. ..
(5) Other.
NOTE: Chi square

Entire
sample

Reglstered
nurses

licensed
practical
nurses/IIcensed visitIng nurses

Nurses·
aides

Homemakers

16.6

8.8

13.1

21.5

20.6

8.2

9.2

8.0

7.2

9.0

60.2

70.3

65.3

55.9

50.8

1.0
14.1

.8
10.9

0
13.6

.8
14.6

2.6
16.9

Why are the higher-skilled employees less available? Previous studies have shown that the rate of moonlighting among
the more skilled workers is significantly greater than among
those of lower skill. 12 In effect, many of these workers
appear to be using temporary help employment as a second
job. Another possible reason for limited availability of highskill workers may be that they possess greater financial
resources than those having lower skills and hence do not
need temporary work as much.
The present study also clarifies the concept that the temporary help employee is seeking a full-time job, but only
for a short period of time. 13 The majority want to work 8
hours per day. However, a significant minority of these
workers desired work for only one time of the day, regardless of skill level. Thus, it appears that many of these employees are seeking employment for a short period of time,
but employment involving only 4 hours per day.
Finally, the research indicates that flexibility in scheduling is a much more important source of motivating individuals to apply to a temporary help firm than is variety,
at least in terms of frequencies. The study also shows that,
the higher the skill level, the greater the probability of citing
flexibility in scheduling as the most important reason for
becoming a temporary help employee.
D
--FOOTNOTES-' Robert Bednarzik, "Short workweeks during economic downturns,"
Monthly labor Review, June 1983, pp. 3-11.
2 Mack Moore, The Role of Temporary Help Services in The Clerical
Labor Market, Ph.D. diss. (Madison, University of Wisconsin, 1963).
3 Martin J. Gannon, "An Analysis of the Temporary Help Industry,"
Labor Market Intermediaries, Special Report No. 22 (Washington, National Commission for Manpower Policy, March 1978), pp. 195-255.
4
Gannon, "An Analysis of the Temporary Help Industry."
5
W. Albeda and G. Veldkamp, eds., Temporary Work in Modern Society. Part 2: Temporary Work within a Socio-Economic Framework (The
Netherlands, Kluwer, 1978).
6
Genmaine Greer, The Female Eunuch (New York, McGraw-Hill, 1970).
See also Alvin Toffler, Future Shock (New York, Random House, 1970).
7
Richard Leone and Donald Burke, Women Returning to Work and Their
fllleraction With a Temporary Help Service (Springfield, Va., National
Technical Infonmation Service, 1976).
"Gannon, "An Analysis of the Temporary Help Industry."
9
This response rate was substantially higher than that reported in most
previous studies, possibly because a dollar was attached to each questionnaire.
HlLeone and Burke, Women Returning to Work.
11
Informal interviews with executives in this industry confinm this trend.
12
Gannon, "An Analysis of Temporary Help."
13
Moore, The Role of Temporary Help Services; and Leone and Burke,
Women Returning to Work.

Pay gains tempered

in basic steel mills
NORMA

~

40.26 (p.;; .001).

W. CARLSON

The gain in steelworkers' pay lagged behind that of all
workers in the durable goods manufacturing industries, according to a Bureau of Labor Statistics occupational wage

28

- - - - - - - - - - - - -----·

Table 1. Number of production and related workers and average straight-time hourly earnings, 1 by selected characteristics,
basic Iron and steel mllls, United States and raglons,2 August 1983

UnHI~ Stites
Number

Nortlleall
Number

of

Average
llourly

woran

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

South
Number

of

Average
llotirly

earnings

worllen

184,078

$11.87

Size of community:
Metropolitan areas 3. . . . . . . . . . ..............
Nonmetropolitan areas ......... . ' ...........

171,637
12,441

Size of establishment:
100-999 employees .......................
1,000-2.499 employees ....................
2,500 employees or more ...................

North Central
Number

of

Average
hourly

11rnlng1

worken

48,388

$11.71

11.96
10.65

41,876
6,512

46,746
29,225
108,107

10.37
12.26
12.41

Size of company:
100-9,999 steel industry employees ............
10,000 or more steel industry employees .........

82,131
101,947

Job and pay system:
Common job and pay system4 ............... '
Not under common job and pay system ... .......
Labor-management contract coverage:
Establishments withMajority of workers covered ........ ...... . .
None or minority of workers covered. . . . . . . ....

Cll1ract1r!lllc

All production workers .....

WIii
Number

of

Average
hourly

of

Average
hourly

earnings

worllen

worllen

worllen

earnings

32,265

$11.24

92,848

$12.03

10,577

$13.05

11.88
10.61

26,336
5,929

11.36
10.68

92,848

12.03

10,577

13.05

14,705
14,941
18,742

10.85
12.00
12.17

9.71

10.59
12.54
12.22

-

12.74

13,104
12,136
67,608

-

15,370

11.36
12.28

24,505
23,883

11.51
11.92

14.786
17,479

9.51
12.70

35,504
57,344

11.66
12.26

7,336

13.12

90,286
93,792

12.13
11.62

31,557
16,831

11.86
11.44

22,268

10.98

45,491
47,357

12.33
11.75

7,336

13.12

169,010
15,068

12.06
9.75

47,034

11.64

23.084
9,181

12.13
8.99

89,623
3,225

12.10
10.15

9,269

13.54

1Excludes

premium pay for overtime and for work on weekends, holidays, and late shifts.
2The regions are defined as follows: Northeast-Connecticut. Maine. Massachusetts,
New Hampshire. New Jersey, New York. Pennsylvania, Rhode Island. and Vermont; South-Alabama. Arkansas. Delaware. District of Columbia, Florida, Georgia, Kentucky, Louisiana. Maryland. Mississippi, North Carolina. Oklahoma, South Carolina. Tennessee,
Texas, Virginia. and West Virginia; North Central-Illinois, Indiana, Iowa. Kansas. Michigan. Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wiscon,in:
and West-Arizona. California, Colorado, Idaho, Montana, Nevada, New Mexico. Oregon, Utah, Washington, and Wyoming. Alaska and Hawaii were not included in the study.

survey conducted in August 1983. At that time, production
and related workers in basic iron and steel mills averaged
$11.87 an hour-43 percent above the $8.32 recorded in
February 1978. 1 During this same period, the Bureau's Employment Cost Index of wages and salaries in durable goods
industries rose 50 percent.
Moreover, the Bureau's Employment and Earnings series
reported that average hourly earnings for steelworkers declined in 1983 for the first time in 20 years-by about 4
percent for the year. 2 This decrease ended a trend that began
in 1970 with average annual increases in gross earnings in
basic iron and steel mills exceeding those in all manufacturing industries and in durable goods production.
These developments reflect the bargaining in the spring
of 1983 between the union's Basic Steel Industry Conference and the seven Coordinating Committee Steel Companies. Settlements called for reduced pay and benefits in
exchange for improvements in job security and increased
aid to laid-off workers, as well as for capital improvements
to existing facilities. About six-tenths of the workers surveyed were affected by the $1.25-an-hour cut in regular pay
plus elimination of a cost-of-living allowance of 6 cents.
For workers who were paid on an incentive basis, the pay
reduction amounted to slightly over $1.31. 3

Norma W. Carlson is an economist in the Division of Occupational Pay
and Employee Benefit Levels, Bureau of Labor Statistics.

-

-

15,851

-

-

-

-

-

-

-

-

-

-

-

-

-

-

3Standard

metropolitan statistical areas as defined by the U.S. Department of Commerce
through October 1979.
4 1ncludes mills in common job evaluation ana pay systems. that is. with the same $9.495
minimum hourly wage and 14. 7 cents-per-hour increment between job classes.
Nore: Dashes indicate that no data were reported or that data do not meet publication
criteria.

Survey findings
Variation in regional pay patterns. Steelworkers in the
North Central States made up one-half of the employees
surveyed and averaged $12.03 an hour. (See table I.) Western mills recorded the highest pay level, $13.05 an hour,
but accounted for less than one-tenth of the work force.
Earnings averaged $11. 71 an hour in the Northeast and
$11.24 in the South. Although southern mill workers had
the lowest regional hourly average, workers in large establishments and companies and in unionized plants fared better
than those in the Northeast and North Central States, with
an average pay advantage of 4 percent.
Occupational earnings. In 1983, separate wage data were
developed for 62 occupations covering slightly more than
one-third of the production workers surveyed. To facilitate
analysis, the jobs were divided into two groups. In the first
group, job classifications were limited to selected production
departments; the second group comprised jobs that cut across
departments.

Average hourly earnings of the first job group ranged
from $ l 5 .45 for continuous billet or slab casters to $8. 81
for cut-off machine operators in tube finish mills. Wire
drawers in rod and wire mills, the largest occupation surveyed with over 1,500 workers, averaged $10.37. Job classifications with at least 500 workers included first helpers
at electric furnaces ($14.60), keeper helpers in blast furnaces
29

MONTHLY LABOR REVIEW August 1984 • Research Summaries
($11.91), and cut-off machine operators.
Pay levels in the second group ranged from $13.56 an
hour for bricklayers to $9.50 for laborers. The largest group
studied-12,000 millwrights-averaged $12.72. Jobs with
at least 4,000 workers included laborers ($9.50) and motor
inspectors ($12.92).

Incentive workers predominant. Almost four-fifths of the
· steelworkers surveyed received pay based on wage incentives. This proportion was higher in establishments using
the common job evaluation and pay system (nine-tenths)
than fo mills with other types of formal job evaluation systems (two-thirds). The predominance of incentive workers
is traceable to the design of the pay system which provides
for direct, indirect, and secondary indirect incentives. The
three types are differentiated by the extent to which a worker,
alone or as part of a crew, can affect or control the rate of
output or the utilization of equipment. For example, furnace
operators are direct workers, while millwrights assigned to
specific departments are indirect. Maintenance workers and
general laborers not assigned by department are secondary
indirect employees. 4
Employee benefits. Virtually all of the workers were in
establishments providing paid holidays and vacations and
various health and insurance benefits. The most common
provisions were 10 paid holidays annually and I week of
paid vacation after I year of service, 2 weeks after 3 years,
3 weeks after 10 years, and 4 weeks a:fter 25 years. Almost
all production workers were eligible for life insurance, sickness and accident insurance, hospitalization, surgical and
basic and major medical insurance, and retirement pension
plans. Supplemental unemployment benefits, dental insurance, and retirement severance plans applied to four-fifths
of the workers or more. Most of the health, insurance, and
retirement plans were paid for entirely by the employer.
A summary report, Basic Iron and Steel Mills, August
1983 (Summary 84-6) is available from the Bureau or any
of its regional offices. A comprehensive bulletin is scheduled for publication later this year.
D
--FOOTNOTES-1
Earnings data e)lclude premium pay for overtime and for work on
weekends, holidays, and late shifts. Incentive payments, such as those
resulting from piecework or production bonus systems and cost-of-living
adjustments, were included as part of the workers' regular pay. For a report
on the earlier survey. see Industry Wage Sun•ey: Basic Iron and Steel,
1978-/979. BLS Bulletin 2064 (Bureau of Labor Statistics. 1980). The
1978 survey was limited to establishments with 250 workers or more; the
1983 study covered establishments with 100 workers or more.
Both surveys included establishments employing workers engaged-primarily in manufacturing steel products classified in the following industries
as defined in the 1972 Standard Industrial Classification Manual (SIC) of
the U.S. Office of Management and Budget: (I) Blast furnaces (excluding
merchant coke ovens). steelwork. and rolling mills (part of sic 3312);
(2) steel wire drawing and steel nails and spikes (SIC 3315); (3) cold rolled
steel sheet. strip. and bars (SIC 3316); and (4) steel pipe and tubes (SIC
3317). Excluded from the surveys were merchant coke ovens (part of SIC
3312). electrometallurgical products (SIC 3313), establishments producing
steel solely for use by their parent company and not classified in the steel

30

industry, and separate auxiliary units such as central offices and warehouses.
2 The Bureau's Employment and Earnings series reports gross earnings
which include premium pay for overtime, holidays, vacations, and sick
leave paid directly to the employee.
3
The concessions included elimination of a cost-of-living allowance of
6 cents accumulated since November 1982. Cost-of-living adjustments
were suspended until at least August 1984, a paid holiday was dropped,
and other benefits were reduced. One week of regular paid vacation was
eliminated for employees eligible for at least 2 weeks' vacation in 1983,
but the week was restored in 1984. The extended vacation plan was discontinued in 1983 and vacation bonuses were eliminated in 1984. In return,
steel firms agreed to invest the savings in capital improvements and to
increase financing of the Supplemental Unemployment Benefit fund. For
details, see Current Wage Developments, March 1983, pp. 1-2.
4
For a detailed discussion of the incentive pay system in the industry,
see Joseph Bush, "Incentive pay patterns in the steel industry," Monthly
Labor Review, August 1974, pp. 75-77.

Area occupational pay
in auto dealer repair shops
Occupational pay varies widely in automobile dealer repair
shops, according to a Bureau of Labor Statistics' survey.
The November 1982 study covered eight occupations in 24
metropolitan areas and found that diverse employee skills,
extensive use of incentive pay plans, and pay differences
among individual shops contributed to the wide range of
earnings.
Journeyman auto mechanics, who repair, rebuild, or overhaul major assemblies of cars and light trucks, averaged
from $14.52 an hour in San Francisco to $8.59 in Birmingham. 1 (See table I.) Most commonly, journeyman mechanics averaged 20 to 40 percent more than automotive
service mechanics in the same metropolitan area. Average
earnings of service mechanics, who perform minor repairs
and tuneups, ranged from $10.99 an hour in Dallas-Fort
Worth to $6.80 in Boston; in most areas, however, averages
were between $7 and $9 an hour.
Among the jobs studied, body repairers or painters had
the highest average in 20 of the 24 areas studied. Averages
for painters ranged from over $15 an hour in three areasDenver-Boulder ($16.49), Chicago ($15.61), and Kansas
City ($15.59)-to $8.60 in New York and $8.12 in Memphis. For body repairers, who repair bodies and body parts
of automotive vehicles, hourly averages ranged from $14. 68
in San Francisco to $9 in Indianapolis. Painters typically
averaged 8 to 14 percent more than body repairers in the
12 areas where painters held the wage advantage. When
body repairers held the edge in an area, their wage advantage
was usually 11 percent or less.
Lubricators and new-car get-ready workers, usually the
lowest paid, averaged between $5 and $8 in a majority of
the areas. Service salesworkers, who examine automobiles
to determine the need for and cost of repairs, averaged more
than $9 an hour in most of the areas surveyed. Their highest
hourly average was found in San Francisco-Oakland ($13.58)

typically paid on an incentive basis. In nearly all areas, for
example, earnings of the highest paid journeyman mechanic
exceeded those of the lowest paid by at least $9 an hour.
In San Francisco, where virtually all journeyman mechanics
were time rated, their earnings fell within a comparatively
narrow range-$13 to $16.50.
The dispersion of individual earnings resulted more from
disparate pay levels among establishments than from pay
differences within establishments. For example, the earnings of the highest paid body repairer rarely exceeded those
of the lowest paid by more than $6 an hour within individual
establishments. However, earnings of the highest paid body
repairer in an area exceeded those of the lowest paid by at
least $14 an hour in nearly all areas. As a result of the wide
dispersion of earnings within an occupation, there was a
considerable overlapping of individual workers' earnings

and their lowest in Pittsburgh ($7. 72). Service salesworkers
averaged 15 to 30 percent more than parts clerks in each
area but Houston, where parts clerks held a slight edge$12.27 to $12.16.
In the six occupational classifications for which data are
available for all areas, 2 pay levels in November 1982 were
most often highest in San Francisco-Oakland and lowest in
Birmingham and Pittsburgh. The interarea spread in average
earnings, however, differed considerably by occupation. For
example, new-car get-ready workers in San Francisco-Oakland averaged 90 percent more than their counterparts in
Washington, D.C., whereas the spread between these two
areas was 33 percent for journeyman automotive mechanics,
14 percent for body repairers, and 2 percent for painters.
Within the same area and occupation, individual earnings
were widely dispersed, especially when the occupation was

Table 1. Number of workers and average straight-time hourly earnings 1 In selected occupations In auto dealer repair shops,
24 areas,2 November 1982
Body repairers
Area

Northeast
Boston .......... ..
Nassau-Suffolk ..... .
New York ...... .
Philadelphia ..... .
Pittsburgh .... .

Lubricators

Mechanics,
automotive,
journeymen

Mechanics,
automotive,
service

116
121
253
376
170

$6.88
7.91
7.42

88

$6.80
7.94
8.85
7.01
6.91

10.95
8.59
11.65
12.74
8.65
11.36
10.90

190
90
771
114
76
67
264

9.47
8.53
10.99
10.06
7.04
8.17
7.50

128
40
212
419
60
99
370

6.65
4.66
9.49
7.48
4.97
7.15
5.15

83
14
191
175
41
57
170

11.70
10.62
11.94
13.70
8.12
11.43
14.54

1,997
1,292
403
561
487
822
893

12.34
12.24
8.97
10.66
11.42
12.00
11.70

151
514
90
44
79
72

7.21
8.68
7.42
7.86
7.49

256
129
55
54
85
67
125

6.62
6.85
7.24
8.28
5.87
8.45
10.16

93
301
58
43
22
117
13

15.61
12.48
12.07
15.59
11.59
13.49
12.71

10.60

711

11.74

107

10.43

93

7.41

55

194
48
59

10.18
6.21
5.58

3,023
534
508

12.39
9.85
10.87

579
124
79

9.73
8.84
8.13

377

66
71

8.02
7.09
5.56

163

9.60

1.742

14.52

24

10.82

102

9.78

S5.32
6.02
7.08
5.50
4.46

1,069
663
1,058
1,544
883

$10.13
11.96
11.39
10.10
10.05

100
342
679
583

464

60
7
64
27
48

South
Atlanta ...
Birmingham ....... .
Dallas-Fort Worth ..
Houston .
Memphis
Miami .......... .
Washington ..

240
86
481
571
96
171
580

12.06
9.50
13.23
13.55
12.09
12.51
12.90

29
17
47
65
18
19
12

7.46
5.44
8.24
8.50
5.53
6.49
3.74

810
224
702
1.310
234
436
1,801

North Central
Chicago ....... .
Detroit ........ .
Indianapolis.
Kansas City .
Milwaukee.
. ...
Minneapolis-St Paul .
St. Louis

783
756
204
281
283
314
465

13.67
12.53
9.00
12.53
11.68
12.92
12.98

33
107
22
43
59
122

7.67
7.04
6.05
9.61
6.44
8.05
9.79

193

13.71

30

817
188
123

12.48
11.49
10.78

501

14.68

1Excludes

Painters

Parts clerks

Service
salesworkers

Number Average Number Average Number Average Number Average Number Average Number Average Number Average Number Average
al
hourty
DI
hourty
DI
hourly
al
hourty
ol
hourly
ol
hourly
DI
hourly
al
hourly
workers earnings workers earnings workers earnings workers earnings workers earnings workers earnings workers earnings workers earnings

$10.32
10.20
10.46
9.84
10.15

West
Denver-Boulder .
Los AngelesLong Beach .
Phoenix ..
Portland .........
San FranciscoOakland .....

New-car
get-ready
workers

408
101
236

366

36

premium pay for overtime and for work on weekends, holidays, and late shifts.
2The areas used in this survey are defined as follows: NORTHEAST: Boston-Suffolk
County, 16 communities in Essex County, 34 in Middlesex County, 26 in Norfolk County,
and 12 in Plymouth County, Mass.: Nassau-Suffolk-Nassau and Suffolk Counties, N.Y.:
New York-New York City (Bronx, Kings, New York. Queens, and Richmond Counties and
Putnam, Rockland. and Westchester Counties, N.Y.. and Bergen County. N.J .. Philadephia-Bucks, Chester, Delaware. Montgomery, and Philadelphia Counties. Pa.; and Burlington, Camden, and Gloucester Counties, N.J.; and Pittsburgh-Allegheny, Beaver,
Washington, and Westmoreland Counties. Pa.; SOUTH: Atlanta-Butts. Cherokee. Clayton. Cobb, Dekalb, Douglas, Fayette, Forsyth, Fulton. Gwinnett. Henry, Newton. Paulding,
Rockdale, and Walton Counties, Ga.: Birmingham-Jellerson, St. Clair. Shelby. and Walker
Counties, Ala. Dallas-Fort Worth-Collin, Dallas. Denton. Ellis. Hood. Johnson. Kaufman.
Parker, Rockwall, Tarrant, and Wise Counties. Tex.: Houston-Brazoria. Fort Bend, Harris.
Liberty, Montgomery, and Waller Counties, Tex.: Memphis-Shelby and Tipton Counties.
Tenn.; Crittenden County, Ark.: and DeSoto County, Miss: Miami-Dade County. Fla.:
and Washington-The District of Columbia: Charles. Montgomery. and Prince Georges
Counties, Md.: and Alexandria, Fairfax, and Falls Church Cities and Arlington. Fairfax,

8.77

5.77

$11.69
9.34
8.60
9.78

$7.10
8.29
9.13
6.69
6.12

241
193
391
422
193

$8.46
10.21
9.49
8.20

8.12
8.60
10.56
12.27
8.47
8.11

7.70

204
75
348
293
60
132
424

10.32
10.74
12.30
12.16
8.74
10.42
9.60

738
536
192
193
166
295
310

7.29
6 83
7 00
8 19
6 62
8.69
10.47

510
408
91
146
142
159
190

9.35
7.86
8.07
11.44
8.44
10.50
10.95

16.49

323

8.31

207

10.98

287
89
38

13.76
11.27
11.60

1,374
224
208

9.62
8.67
7.93

815
144
123

12.70
10.99
9.69

124

14.76

509

11.90

383

13.58

7
28
38
131

5.61

357
263
529
591
317
320
114
565
614
155
201
630

7.72

Loudoun, and Prince William Counties. Va.; NORTH CENTRAL: Chicago-Cook, DuPage,
Kane. Lake. McHenry, and Will Counties, Ill.: Detro,t-Lapeer, Livingston, Macomb, Oakland. St. Clair, and Wayne Counties, Mich.; Indianapolis-Boone, Hamilton, Hancock.
Henricks, Johnson, Manon, Morgan, and Shelby Counties, Ind.; Kansas City-Cass, Clay,
Jackson, Platte, and Ray Count"les. Mo : and Johnson and Wyandotte Counties, Kans.:
Milwaukee-Milwaukee. Ozaukee. Washington, and Waukesha Counties, Wis.: Minneapolis-St._ Paul-Anoka, Carver, Chisago, Dakota, Hennepin, Ramsey, Scott, Washington,
and Wright Counties. Minn.: and St. Croix County. Wis.; and St. Louis-St. Louis City:
Franklin, Jefferson, St. Charles. and St. Louis Counties. Mo.; and Clinton, Madison,
Monroe, and St. Clair Counties, 111. WEST: Denver-Boulder-Adams. Arapahoe, Boulder. Denver. Douglas, Gilpin, and Jefferson Counties, Colo : Los Angeles-Long BeachLos Angeles County, Calif.: Phoemx-Mancopa County, Ariz.: Port/and-Clackamas, Mull··
nomah. and Washington Counties. Oreg.: and Clark County, Wash.; and San FranciscoOakland-Alameda, Contra Costa. Marin. San Francisco. and San Mateo Counties, Calif.
NOTE:

Dashes indicate no data reported or data that do not meet publication criteria.

31

MONTHLY LABOR REVIEW August 1984 • Research Summaries
even among jobs with substantially different pay averages.
Incentive pay systems, most commonly flat-rate hours
plans, determined the earnings for just over one-half of the
91,680 service workers covered by the study. 3 Under flatrate hours plans, which applied to three-tenths of the workers, pay is computed by multiplying the number of flat-rate
hours predetermined for each task by an established hourly
rate. Group bonus and commission plans together covered
one-seventh of the service workers. Other incentive systems
in auto dealer repair shops include individual bonus plans
and flat-rate percent plans. In. the latter, workers receive a
stipulated proportion (most often 50 percent) of the labor
cost charged to the customer. These flat-rate percentage
plans applied to fewer than one-tenth of the workers.
Slightly more than two-fifths of the service workers were
paid time rates in November 1982, typically under informal
plans providing individual rates in specified occupations.
Formal time-rated plans providing single rates for specified
jobs within establishments were more common than the
informal plans in eight areas, including San Francisco; there,
single-rate plans applied to four-fifths of the workers.
Paid holidays were provided to at least nine-tenths of the
workers in all areas except Denver-Boulder, where the proportion was about seven-tenths. Holiday provisions, however, varied widely by area. In seven areas (Boston, Chicago,
Minneapolis, Nassau-Suffolk, New York, San Francisco,
and St. Louis), at least two-thirds of the workers received
9 holidays or more annually; in most southern areas, provisions for more than 5 days were rare.
Incentive workers, particularly those paid under flat-rate
systems, may receive holiday pay which differs from their
usual pay. About one-third of the incentive workers were
granted holiday pay which was substantially less than their
usual pay. Most of the remainder received holiday pay that
equaled, or approached, their regular pay. A few incentive
workers received holiday pay that. was greater than their
regular pay.
Virtually all nonsupervisory service workers were in shops
providing paid vacations after qualifying periods of service.
Although vacation provisions varied substantially among the
areas, typical provisions were I week of pay after I year
of service and 2 weeks after 2 years. Provisions for at least
3 weeks of vacation pay, generally after IO to 15 years of
service, were more common in the Northeastern and North
Central areas than in the other two regions. Only in Chicago,
Minneapolis, St. Louis, and San Francisco were a majority
of the workers covered by 4-week plans.
Almost all service workers were in establishments providing hospitalization, surgical, basic medical, and major
medical insurance for which employers paid at least part of
the cost. Provisions for life insurance covered nine-tenths
of the workers; accidental death and dismemberment insurance, four-fifths; and short-term protection against sickness or accident, two-thirds. As with the other elements of
this survey, incidence of certain health and insurance plans
32

varied widely by area.
Retirement pension plans (other than social security) applied to at least 90 percent of the workers in MinneapolisSt. Paul, St. Louis, and San Francisco. Elsewhere pension
plans covered a majority of the workers in eight areas and
typically from one-fourth to one-third in the remaining 13,
principally in the South.
The 3,363 auto dealers within the scope of the surveythose with at least 20 workers-employed 173,682 workers
in November I 982. Included were the repair departments
of establishments engaged primarily in selling new, or new
and used, automobiles. Dealerships primarily selling trucks
and used cars, and general automobile repair shops, were
not included. In the 24 areas combined, executive, supervisory, and office personnel made up 24 percent of the work
force; auto salesworkers made up 19 percent, and the nonsupervisory service workers accounted for 57 percent.
One-third of the areas accounted for about three-fifths of
the 91,680 nonsupervisory service workers: The Los Angeles-Long Beach area had the largest number (10,083),
followed by Washington (8,024), Chicago (7,080), Houston
(6,107), Philadelphia (5,924), Detroit (5,623), Dallas-Fort
Worth (5,557), and San Francisco (4,579). In the remaining
16 areas, employment ranged from 3,898 in New York to
approximately 1,000 in Birmingham.
Slightly more than one-fifth of the nonsupervisory service
workers were covered by labor-management agreements.
The proportion was about nine-tenths in San Francisco and
St. Louis; between three-fifths and four-fifths in Chicago,
Minneapolis, Nassau-Suffolk, and New York; nearly twofifths in Kansas City; and one-fourth or less in Boston,
Detroit, Milwaukee, Philadelphia, and Pittsburgh. In the
remaining 12 areas, primarily in the South and West, no
establishment visited reported a majority of its nonsupervisory service workers under union contracts. The major
unions in the industry were the International Association of
Machinists and Aerospace Workers (AFL-cto) and the International Brotherhood of Teamsters, Chauffeurs, Warehousemen, and Helpers of America (Ind.). In a few areas,
both of these unions had bargaining agreements with the
same establishment.
A comprehensive report on the survey findings, Industry
Wage Survey: Auto Dealer Repair Shops. November /982
(Bulletin 2198), is for sale by the Government Printing
Office, or by any of the Bureau's regional offices.
D

--FOOTNOTES-1
Earnings data exclude premium pay for overtime and for work on
weekends, holidays, and late shifts.
2
Data did not meet publication criteria for automotive service mechanics
in St. Louis and for painters in Pittsburgh.
3
These "nonsupervisory service workers" included working supervisors
and nonsupervisory workers in all departments except the office and auto
sales departments. Included are workers in departments such as repair,
service, and parts.

New Jersey trends
in high tech employment
The State of New Jersey has consistently been among the
leading centers of high technology industry in the Nation.
A recent State study, employing a broad definition of high
technology, found that employment in New Jersey's high
tech industries rose 3.1 percent annually between 1975 and
1980, compared to a 2.0-percent increase for all other private nonfarm industries. However, because the national rate
of growth in high tech jobs was 4. 7 percent per year over
the same period, the State's share of the U.S. total actually
declined from 4.6 percent in 1975 to 4.2 percent by 1980.
Nearly 224,000 persons were employed in New Jersey's
high tech industries in 1980, about 31,000 more than in
1975. These workers, who accounted for I of every 11
private nonfarm jobs in the State, were distributed among
four broad components: manufacturing (69 percent); communications (23 percent); computers and data processing
(7 percent); and research ( I percent). The employment performance of the four components was mixed over the study
period, with sizable annual increases in computers and data
processing and in communications, slower growth in manufacturing, and large absolute declines in the research area.
Manufacturing was the largest component, accounting for
70 percent (155,559) of New Jersey's high technology jobs
in 1980. Although the 2.0-percent annual employment growth
in the State's high tech manufacturing industries over the
study period was modest, it outpaced the 1.2-percent increase recorded for traditional manufacturing, with the result
that the high tech share of the State's total manufacturing
employment grew from 18. 7 percent in I 975 to 19.3 percent
by 1980. The drug industry was the largest high tech manufacturing employer with 32,679 workers in 1980, reflecting
annual growth of 3. 3 percent since 1975. Other numerically
important three-digit sic industries and their 1975-80 compound annual rates of growth:
1980

Communications equipment ............... .
Electronic components ..
Electrical lighting equipment ............... .
Computer machinery ... .
Surgical instruments ... .
Control instruments .... .

employment
31.042

Annual rate
o( Rrowth
0.9

18.363
11.311

3.9
0.4

9.944
9.230
6,970

5.5
-0.6
9.3

Among nonmanufacturing industries. the second largest
component of New Jersey's high technology sector was
communications, with more than 50.000 employees in 1980
and growth of 5.8 percent per year. 1975-80. Telephone
communications accounted for the bulk (44,644) of the
workers in 1980, after 5 years of increase at a 4. 9-percent

annual rate. Pulling up the average growth rate for the communications component were the small but rapidly growing
telegraph communication and communications services industries, which recorded gains of 18.5 percent and 16.3
percent per year over the study period.
The computer and data processing component of the State's
high tech sector posted a hefty 8. 9-percent yearly rise between I 975 and 1980, employing 15, 157 workers in the
latter year. In sharp contrast was the performance of the
research component, which consisted of research and development laboratories and noncommercial educational, scientific, and research organizations. Employment in R&D labs
fell by 8.2 percent annually to 1,089 workers by I 980;
noncommercial organizations lost jobs at a 12.3-percent
rate, and employed only 524 persons Statewide at the end
of the study period. However, the declines noted in the
research component should be interpreted with caution, because employment in research units that are divisions of
larger firms is often reported under the sic code of the parent
company and cannot be broken out separately for statistical
analysis.
The study, based on information from the Census Bureau's County Business Patterns, also compared the employment performance of New Jersey and 15 other States
with large high tech sectors. Among the salient findings
from this portion of the analysis:
• New Jersey ranked seventh of 16 in terms of 1980 high
tech jobs-behind California, New York, Illinois, Texas,
Massachusetts, and Pennsylvania.
• Declining employment shares in high tech manufacturing
between 1975 and 1980 were observed in States whose
economies have traditionally been manufacturing based,
such as Connecticut, New York, Pennsylvania, Ohio,
Illinois, and New Jersey. There thus appears to be a link
between the health of a State's overall manufacturing
sector and its share of high tech employment. New Jersey
ranked seventh among the States in terms of such employment in 1980.
• In terms of 1980 employment, New Jersey ranked eighth
in the communications component, eleventh in independent noncommercial scientific and research organizations,
and twelfth in research and development laboratories. The
State's highest ranking-fifth-was in computer and data
processing services.
High tech employment trends over the study period are also
presented for each State by major industry component.
New Jersey's High Technology Economy: A Profile of Recent Developments and Comparative Performance was prepared by Theodore A. Minde of the Office of Economic
Research, New Jersey Department of Commerce and Economic Development (Trenton, 1983).
D

33

Foreign Labor
Developments

Lifetime employment in Japan:
three models of the concept
KAZUTOSHI KOSHIRO

As the Japanese economy overcame the adverse effects of
two oil crises, admiration for its management emerged in
foreign countries. The success of Japan's economy tended
to promote myths about the "lifetime employment" practices of Japanese firms. Following is a brief look at three
models of the lifetime employment system. 1
Lifetime employment is a long-established practice in large
Japanese firms. However, it is a "gentlemen's agreement"
and is not guaranteed by statute or collective bargaining agreement. 2 The recent concept of lifetime employment is described
as follows:
Workers become employed right after their graduation from
school with a particular company. The employer will not lay
off his workers if possible even in the course of depression. The
employee in turn will not quit his job at this company but tend
to continue working there until he reaches his retirement age. ·1

This definition reflects the concept of lifetime employment which prevailed during years of high economic growth
which began about 1955. It also reflects the social ideas
generated by labor unions' resistance to mass dismissals
during the preceding decade, as well as court decisions to
restrict employers' right to dismissal due to business difficulties. This concept differs from the prototype of the lifetime employment system originated in large firms around
1910. Three differences, in particular, should be noted.
First, in the prewar period, there had been a considerable
number of job changes by workers prior to their entering
large firms. Most workers usually established a career after
finishing compulsory military service, not right after graduation from school. Therefore, many tended to change jobs
during the first 10 years after school. After World War II,
because of the abolition of military service and continuous
growth of large companies, employment practices changed
so as to recruit a new work force mostly from recent gradKazutoshi Koshiro is a professor. Faculty of Economics. Yokohama National University. Yokohama. Japan.

34

uates. However, in the I 960's through the early I 970's,
many fast-growing companies faced labor shortages due to
a continuous increase of output. Occasionally, these companies would recruit temporary workers until they could fill
vacancies with regular workers from the ranks of new school
graduates. A number of these temporary workers had opportunities to be retained as "halfway" regular employees
if they could demonstrate good performance and efficiency.
Most of the "halfway" workers came from rural areas.
Their pay was less than that of standard regular workers,
even if they had equal capabilities and skills, because they
had fewer years of service with the company. However,
over time the wage differentials between the two groups
could be reduced because of accumulated merit ratings.
Second, there was intense competition among workers
with many years of service in the same company for promotions and wage increases. Lifetime employment and
seniority-based wages do not exclude competition among
workers, although since the war, the labor movement has
endeavored to control the wage differentials resulting from
merit rating. Even blue-collar workers arc rated by merit at
least three times a year. Promotion is determined according
to the results of such accumulated merit ratings. To maintain
the fairness and continuity of the merit rating system, management keeps detailed records of the personal history of
each worker. At one time, these records were kept in both
the worker's ledger and the wage ledger. Now they arc
stored in a computerized data base.
Third, the prototype stationary model of lifetime employment assumes that a certain percentage of workers will
voluntarily quit their jobs as a result of competition among
work groups.

The stationary model. Under the stationary model, lifetime employment is a system of highly developed internal
labor markets. 4 It consists of a web of administrative rules
for pricing labor and allocating the labor force within a firm.
It is characterized by specialization of labor, on-the-job
training, and a body of firm-specific customs.
The stationary model is illustrated below. For simplification, the maximum length of service was limited to 10
years.

Number of
employees
1
2
3
4

5
6
7
8
9
10

55

Years of
service
10

Wage
rate
10

9

9

8
7
6
5
4
3
2
I

8
7
6
5
4
3
2
I

Total
wages
10
18
24
28
30
30
28
24

Separation
rates
1/2
1/3
1/4
1/5
1/6
1/7
)/8
1/9

18

1/10

10

0

220

)9.3

New workers are recruited only at the bottom of hierarchy
(usually from among new school graduates). To maintain
the hierarchy, it is implicitly expected that a worker will
quit voluntarily as a result of failure to compete successfully
with fellow workers of the same generation and tenure.
Theoretically, this should be the least efficient worker in
the group. Each remaining worker can then receive a wage
increase of one grade and can be promoted to higher positions. The wage fund can be maintained at the stationary
level of 220 in spite of wage increases for remaining workers. Thus, the average wage rate can be maintained at the
constant level of 4. Because one worker from each generation quits, the total annual rate of separation is 19. 3 percent
in this model. (If the maximum length of service is extended
to 30, the average separation rate becomes 10 percent.) In
other words, all workers who are hired after school cannot
necessarily continue their employment until the age of mandatory retirement, contrary to the usual definition of lifetime
employment cited above.

The growth model. For this model, the concept of stationary lifetime employment is modified. The organizational
growth of a company makes it possible for all workers to
expect to remain employed and be promoted each year until
retirement. One of the basic characteristics of the growth
model is its strong dependence on organizational growth,
which in turn requires the expansion of market shares. The
larger the market share of a company, the greater the opportunities for organizational growth which guarantees employment security and improvement of wages and other
conditions of work. In this sense, Japanese firms tend to
have stronger impetus for organizational growth, rather than
increased rate of return on investment.
The stagnation model. After the oil crisis in 1973-74,
most of the major firms changed their employment strategies
to adjust to new market situations. They reduced employment by various measures: cutting overtime, laying off temporary workers, stopping new recruitment, not filling
vacancies, and transferring workers to other shops or plants
within their company as well as to related companies or
subsidiaries. Some deeply depressed industries, such as
shipbuilding and petrochemicals, promoted voluntary sep-

aration by offering severance payments. For example, more
than 10,000 workers left Mitsubishi Heavy Industries Corporation during the years following the first oil crisis. Parttime workers with lower labor costs were recruited to fill
the vacancies. As a result, organizational hierarchies tended
to shrink, illustrating the stagnation model.
D
--FOOTNOTES-1 This report is excerpted from Kazutoshi Koshiro, ""Personnel Planning,
Technological Changes, and Outsourcing in the Japanese Automobile Industry," a paper prepared for the Workshop on Industrial Relations and
Industrial Change in the World Automobile Industry, Brussels, February
16-18, 1983. The workshop was part of an international joint project on
the future of the automobile. The paper. Discussion Paper Series 83-3,
May 1983, is available from the Center for International Trade Studies,
Faculty of Economics, Yokohama National University, Yokohama, 240
Japan.
2 The civil law requires an "unavoidable reason" to terminate an employment contract without notice. The labor standards law introduced an
even tighter restriction-it permits dismissal without notice only when
there is an "inevitable cause." There are no laws requiring a reason for
dismissals with notice. However, legal theory has established some very
strict rules concerning dismissal with notice. See T.A. Hanami, Labour
Law and Industrial Relations in Japan (The Netherlands, Kluwer-Deventer, 1979), p. 82.
'Kazuo Koike, "Nihonteki Koyo Kanko" ["Japanese Employment
Practices") in Toyokeizai Shinposha, ed., Keizaigaku Daijiten [Encyclopedia of Economics] (Tokyo, Toyokeizai Shimpo Sha, 1980), Vol. II, pp.
100-08.
4
Peter B. Doeringer and Michael J. Piore, Internal Labor Markets and
Manpower Analysis (Lexington, Mass., D.C. Heath and Co., 1971).

Robots are a big success
at auto plant in Japan
KAZUTOSHI KOSHIRO

In 1971 , robots were first introduced in a plant at X Motor
Co. in Japan. 1 During the latter half of the 1970's, the
number of robots at the plant increased dramatically; by
1981, the company had 730 robots. Most of them (90 percent) perform welding operations in the body assembly shops.
The company also uses robots for painting, and is considering robots for battery and spare-tire loading. Other automation, such as computer-aided design and manufacturing,
transfer machines, and automobile loaders are widely used
by the company.
The robots were obtained largely to do heavy, hazardous,
and monotonous work for which very few workers were
available during the period of high output growth. Because

robots are adaptable and can simultaneously work on different models of cars, the company believed they would
improve product quality and save energy and space.
Automation at X Motor Co. has contributed to improved
product quality by decreasing human error and increasing
mechanical reliability. The company's output increased 186
percent between 1970 and 1980, and productivity increased

35

MONTHLY LABOR REVIEW August 1984 • Foreign labor Developments
139 percent. However, it is difficult to isolate the impact of
robots on productivity because other factors such as rationalization of the production process, automation, improved
equipment, and efforts by quality control circles also contributed to improved product quality and output. Over the
1970-80 period, expenditures for capital equipment decreased because the flexibility of robots allowed mixed production, and because of the extended life of robots.
The investment for robots is returned within 2 years. For
example, cost of a welding robot is about 12 million yen,
whereas the average annual wage for a welder is 5 million
yen. However, the average value of depreciation per employee was 13 million yen in fiscal 1981. Total depreciable
assets were 331,310 million yen, of which the value of the
robots, 8,760 million yen, represents only 2.6 percent.
Each of the 730 robots at X Motor Co. replaces 0. 7
worker. Because the plant has two shifts, one robot replaces
1.4 workers. Therefore, 1,022 workers or 1.8 percent of
the company's total employment have theoretically been
replaced by robots.
When the robots were first introduced, maintenance and
operating workers were sent to robot manufacturers for technical instruction and training. Thus, these workers were able
to program the robots. Although the number of workers at
X Motor Co's body assembly shop decreased by 4 percent
because ot the introduction of robots, there were about the
same percentage of retirements, so few, if any, workers
needed to be transferred. About 100 workers were moved
to new assembly lines which required the use of robots.
The welding robots improved product quality and reduced
the price of automobiles, causing an increased demand for
automobiles. In turn, employment in the body assembly
shops increased to some extent, especially in the more skilled
jobs such as operating, maintaining, and programming robots. Work injuries decreased and job satisfaction was enhanced as workers were relieved from noise, oscillation,
and other job hazards.
Prior to the introduction of robots and other automation,
X Motor Co. consulted with trade unions at the Central
Labor-Management Consultative Council on long-term production and investment plans and matters related to technological changes. The Council's subcommittees are
responsible for discussing problems relating to production,
technology, overtime, transfer, improvement of work environment, health and safety conditions, and other matters
which might arise during the introduction of automation.
Each month, a plant's managers and union representatives
can consult with the subcommittees on any of these matters.
In Japan, a trade union is organized on a company-bycompany basis. X Motor Trade Union, an affiliate of the
Federation of Japan Automobile Workers' Union (JAW, Jidosha Roren), organizes all of the plants of X Motor Co.
At each plant, there is a local branch of X Motor Trade
Union with several full-time officers. 2 Shop stewards or

36

chief stewards meet and negotiate with section chiefs (who
are union members) or foremen on working conditions within
that workshop, consulting the Central Labor-Management
Consultative Council if necessary. Transfers to another
workshop are negotiated between union branch officers and
plant management.
Labor-management relations in the automobile industry
have been cooperative and harmonious since the collapse
of militant left-wing unionism in 1953. Like other unions,
the X Motor Trade Union has been pursuing "3-P movements" (productivity, progress, and participation) and generally has been positive about the introduction of automation.
In recent years, however, the Federation of Japan's Automobile Workers' Union began a campaign to get X Motor
Co. to sign an agreement covering the introduction of new
technology. In 1982, the union submitted to management a
proposal containing the following requirements.
• Consultation with the Federation of Japan Automobile
Workers prior to the introduction of new technologies.
• No layoffs resulting from the introduction of robots.
• No demotions or wage reductions from the introduction
of robots.
• Education and retraining for affected workers prior to, as
well as after, the introduction of robots.
• A fair distribution of the fruits of increased productivity
which results from the introduction of robots.
The Federation of Japan Automobile Workers demands
that it be consulted even at the initial stage of planning new
technologies. It contends that this proposal is not new, but
merely a reflection of long established labor-management
practices at X Motor Co. Although management had some
misgivings, it signed a new contract in March of 1983 covering the i~troduction of new technologies based largely on
the union's proposals.
The government is also taking a cautious approach toward
robots, partly because some industrial accidents occurred
while workers were programming the robots. Also, the Ministry of Labor is concerned that employment may be adversely affected if the economy continues to stagnate. D
--FOOTNOTES-1This report is excerpted from Kazutoshi Koshiro, •·•Personnel Planning,
Technological Changes, and Outsourcing in the Japanese Automobile Industry," a paper prepared for the Workshop on Industrial Relations and
Industrial Change in the World Automobile Industry, Brussels, February
16- I 8, 1983. The workshop was part of an international joint project on
the future of the automobile. The paper, Discussion Paper Series 83-3,
May 1983, is available from the Center for International Trade Studies,
Faculty of Economics, Yokohama National University, Yokohama, 240
Japan.
2
See Kazutoshi Koshiro, "Industrial Relations in the Japanese Automobile Industry," a paper presented at the Workshop on the Future of the
Automobile, Wissenschaftszentrum, Berlin, March 1982. This paper, Discussion Paper Series No. 82-5, August 1982, is available from the Center
for International Trade Studies, Faculty of Economics, Yokohama National
University, Yokohama, 240 Japan. See also, Kazutoshi Koshiro, "Personnel Planning."