G. Donald Wood, Jr., Bureau of Labor Statistics

   This is the first of two papers on the                       The method of computing change had little
Employment Cost Index (ECI).     The ECI is the             impact on the measures of change of compensation
newest Bureau of Labor Statistics (BLS) series              for all workers--the change in the AHC was with-
on wages and compensation (wages plus the em-               in 0.1 percentage point of the change in the ECI
ployer's expenditures for employee benefits).               compensation index.    The concept of pay used
The ECI is a fixed employment weighted Laspeyres            made a difference of 2.2 percentage points--the
index. This paper compares the ECI with related             change in the ECI compensation index for all
series, and explains why the series change rela-            workers increased 28.3 percent while the ECI
tive to one another as they do. The purpose is              wage index for all workers increased 26.1 per-
to help users of the BLS data understand the re-            cent.    The group of workers covered made a dif-
lationships between, and the uses of, the limited           ference of 0.5 percentage point--the change in
data available on the change in wages and compen-           the ECI wage index for production workers was
sation.   The second paper discusses the survey             26.6 percent, as compared to the 26.1 percent
design and the estimation of the ECI.                       change in the ECI wage index for all workers.
   The first comparisons are between the ECI                The method of computing change had a large impact
and series that compute the change in the simple            when only production workers are considered--the
average.    These are the Average Hourly Earnings           change in the AHE of 22.5 percent was 4. I per-
(AHE) and the Average Hourly Compensation (AHC).            centage points less than the change in the ECI
The AHE is total wages and salaries paid to pro-            wage index for production workers.
duction workers in manufacturing, mining and con-               The differences between the ECI indexes in
struction and nonsupervisory workers in the rest            Table i are the result of different rates of
of private nonfarm industry divided by hours paid.          change for different components of costs, or
The AHC is total compensation of all persons in             different worker groups.    The compensation index
the private business sector divided by the hours            increased more than the wage index because the
of all persons.    Comparisons are also made with           index of benefit cost increased more than the
the earnings data from the Current Population               index of wages.    Similarly, the ECI wage index
Survey (CPS), and the Hourly Earnings Index (HEI).          for all workers increased less than the ECI wage
The last two series use more complex measures of            index for production workers because the wage in-
computing change than the simple average, i/                dex for nonproduction workers increased less than
The Change in the ECI and Simple Averages                   the index for production workers.     These rela-
   The ECI, AHC and AHE differ in three                     tionships are known, although estimates of the
respects •                                                  index of benefit cost and the wage index for non-
           i. The method used to compute change--           production workers are not available, because the
               the change in the simple average or          change in an aggregate index is a weighted aver-
               the change in a fixed weight index.          age of the change in the component series.
           2. The concept of pay used--wages or                 The differences between the change in the ECI
               compensation.                                indexes and the corresponding average series in
          3. The workers covered--all workers in            Table i are the result of compositional shifts
               the private nonfarm economy or only          within the workforce among industries and occu-
               production and nonsupervisory                 pations with different levels of pay (wages or
               workers.                                     compensation).    Average pay changes when the rate
   The series are arranged in Table I so that               of pay for a specific job changes, or when there
adjacent series differ in only one of the three             are compositional shifts in employment between
ways.   The ECI (PROD) is an index from the ECI             occupations and industries with different pay
series specifically prepared for this paper.     It          levels.   The ECI index holds the composition
includes the same group of workers as those in-             among jobs fixed and the ECI index changes only
cluded in the AHE.     (Workers included in the AHE         when the rates of pay for specific jobs change.
are called production workers and those excluded             Shifts from higher to lower pay jobs will cause
from the AHE, but included in the ECI are called             the change in the average to increase less than
nonproduction workers).     The period is from               the index because after the shift there are rela-
December 1979, when the ECI compensation index               tively fewer high pay jobs included in the aver-
begins, until December 1982.                                age than before the shift.
Table i: Comparisons of the Change in the ECI                    Compositional shifts had little impact on the
            Wage and Compensation Indexes and the           change in average pay when compositional shifts
            Change in the AHE and AHC.                      among all workers are included--the change in the
                                                            AHC was equal to the change in the ECI compensa-
         (December 1979-December 1982)                       tion index.   Compositional shifts had a large im-
    series                      Percent change               pact on the change in average pay when only com-
                                                             positioanl shifts among production workers are
All private nonfarm workers                                  included--the change in the AHE was 4. i percent-
     AHC                            28.2                     age points less than the change in the ECI wage
     ECI:  Compensation             28.3                     index for production workers.    The different im-
     ECI : Wages                    26. I                    pact of compositional shifts of the workforce
                                                             within all workers and within production workers
 Production workers                                          is the result of cyclical shifts in employment
     ECI (PROD) : Wages             26.6                     that occurred from December 1979 to December 1982.
     AHE                            22.5                         December 1979 to December 1982 was a period of

economic contraction. 2/ During the contraction             ing on the industry, ranged from 3.7 to 8.0
there were shifts in relative employment from high          percent.
to low pay industries.   An indication of the mag-             When all compositional shifts are included,
nitude of the industrial shifts that occurred is            the shifts from low compensation production work-
given by examining the unemployment rate by indus-          ers to high compensation nonproduction workers
trial groups in December 1982.   Unemployment rates         offset the industrial shifts from high to low
in high pay industries such as construction,                compensation industries.     The change in the AHC
mining and durable manufacturing were 22.0, 18.1,           is equal to the change in the ECI compensation
and 17. i percent.  Unemployment in low pay indus-          index.
tries such as finance, and services was 7.9 per-               Note that the change in the average wage of
cent.   The relative shifts in the employment of            all workers depends on relative shifts in employ-
production workers from high wage to low wage in-           ment from low wage production to high wage non-
dustries caused the AHE to increase less than the           production workers, as well as on the change in
ECI (PROD).                                                 the average wage for each group.     The change in
   It is likely that the relative shifts in                 average wage of all workers is not a weighted
employment from high to low wage industries also            average of the change in the average wage of the
cause the change in average earnings for nonpro-            component series.   It is possible that over the
duction workers to increase less than the wage              period December 1979 to December 1982 the change
index for nonproduction workers.   This cannot be           in the average wage for all workers was larger
verified since neither the change in average earn-          than the change in the average wage of either
ings nor the wage index for nonproduction workers           production or nonproduction workers.
is available.                                                  Additional comparisons between the change in
   During periods of economic contraction,                  the ECI wage index and the change in earnings
individual establishments attempt to maintain               for occupational groups are made when data from
their skilled, high pay, workforce.   This results          the CPS are examined.    Both the ECI and CPS give
in a shift in relative employment from production           estimates for the Major Occupational Groups.
workers, which are typically low pay occupations,           Earnings from the Current Population Survey
to nonproduction workers, which are typically high             The CPS measures the median rather than the
pay occupations. 3/ An indication of the magni-             mean.   It uses weekly wage and salary earnings
tude of the occupational shifts that occurred is            rather than hourly earnings.     It covers all
given by examining the unemployment rate by occu-           civilian workers rather than only workers in the
pational groups in December 1982.   Unemployment            private nonfarm economy.     These differences cer-
rates are for the Major Occupational Groups as de-          tainly influence the change in the CPS, but it
fined by the 1970 Census.   These groups do not             is clear from the data in Table 2 that the dif-
closely correspond to the occupations included in           ferences between the CPS and the ECI wage in-
production workers.   Blue-collar and service               dexes are dominated by shifts in composition of
workers who are entirely included in production             the workforce among jobs with different wage and
workers had unemployment rates of 16.3 and 12.2             salary levels.
percent.   Managers and administrators who are en-             For every occupational group, with the excep-
tirely excluded from production workers had an un-          tion of sales workers, the change in the CPS was
employment rate of 4 percent.   The unemployment            less than the change in the ECI.     The differences
rates for other white-collar workers, that are in-          for blue-collar and service worker occupations
cluded or excluded from production workers depend-          ranged from -i0.0 t o - 6 . 1 percentage points.

Table 2.    Comparison of the Change in the ECI Wage and Compensation Index and the Change in the CPS.
                                       (December 1 9 7 9 - December 1982)

                                          Change                   Change
           Series                           in                       in                  Difference
                                           CPS                       ECI

All occupational groups
Wages ...............................      25.4                     26.1                   -0.7

Wages by occupational group

  Service workers   ...................    20.5                     30.5                  -i0.0

    Laborers ........................      14.6                     23.4                   -8.8
    Craft and kindred ...............      20.0                     26.7                   -6.7
    Operatives exc. transport .......      20.0                     26.2                   -6.2
    Transport operatives ............      15.2                     21.3                   -6. i

    Managers and administrators .....      18.6                     23.8                   -5.2
    Clerical workers ................      25.6                     27.2                   -1.6
    Professional and technical ......      29.4                     30.9                   -1.5
    Sales ...........................      24.4                     19.4                    5.0

The differences for white-collar workers, exclud-                   of the time period selected is shown in Chart I,
ing sales workers, ranged f r o m - 5 . 2 t o - 1 . 5 per-          that plots the change in the ECI (PROD) and the
centage points.   The -4. i percentage point dif-                   AHE for the year ended in the quarter shown.
ference for production workers found by compar-                         During the period of falling unemployment the
ing the AHE and ECI (PROD).                                         change in the AHE is greater than the change in
falls above the differences for the blue-collar                     the ECI (PROD), while the reverse is true during
workers and in the low range of the white-collar                    the contraction.    It would be easy to select a
workers.   Production workers include all of the                    period where the change in the AHE was equal to
blue-collar and service occupations, exclude                        the ECI (PROD). Whatever period studied, how-
managers and administrators and include some of                     ever, the difference would be the result of the
the workers in the remaining t h r e e white-collar                 shifts in the composition of the workforce that
groups.                                                             occurred over the period.
    The differences between the CPS and the ECI                     The Use of Various Series 4/
wage index for occupational groups are large com-                       The reasons why the series change differently
pared to the -0.7 percentage point difference be-                   are also the reasons why different series must be
tween the aggregate CPS and ECI wage index.                         estimated.    Wages and compensation are costs to
Shifts in relative employment from high to low-                     the employers of labor, and income to the employ-
er wage industries cause the change in the CPS to                   ees. The change in labor costs of production are
be less than the change in the ECI for the indi-                    not necessarily the change in average labor in-
vidual occupational groups.     Shifts in relative                  come.    For a given group of workers, the change
employment from lower to higher wage occupational                   in the labor cost of production differs from the
groups offset most of the industrial shifts with-                   change in labor income when there are shifts be-
in occupational groups, and the change in the                       tween jobs with different wage or compensation
aggregate CPS is relatively close to the change                     levels.
in the aggregate ECI wage index.                                        An increase in the wage paid for a particular
    The 0.7 percentage point difference between                     job (e.g., higher wages for laborers) increases
the change in the aggregate CPS and the change in                   the average wage earnings or income of employed
the aggregate ECI wage index is large when com-                     workers.    It also increases the labor cost of
pared to the 0. I percentage point difference be-                   producing a given amount of output, changes the
tween the AHC and the ECI compensation index.                       relative cost of employing different kinds of
High wage jobs are usually high compensation                        labor, and changes the cost of employing labor
jobs.   Shifts in the composition of the workforce                  relative to other factors of production.     A rela-
would be expected to cause the difference between                   tive shift in employment from low wage to high
the AHC and the ECI compensation index to be                        wage jobs (e.g., an increase in unemployment of
roughly the same as the difference between the                       low wage laborers) also increases the average
change in the CPS and the ECI wage index.             It is         wage income of workers.     Such a shift does not,
possible that employment shifts could cause the                     however, change the cost of producing a given
change in average wages to differ from the ECI                      amount of output or the cost of employing
wage index by 0.7 percentage points and cause the                   different kinds of labor for any producer.     Each
change in the AHC to differ from the compensation                   employer of labor pays each kind of labor the
index by 0. i percentage point.      It is likely that              same amount after the shift as before.
other differences in the CPS series, such as the                        The ECI should be used when the change in
use of the median rather than the mean, and                         wages or compensation as a cost of employing
weekly rather than hourly earnings, account for                      labor is wanted.   The ECI should be used in ana-
some of the 0.7 percentage point difference be-                      lyzing the change in wages or compensation on
tween the change in the CPS and ECI wage index.                     output prices, the allocation of labor between
    The series discussed, the AHE, AHC, CPS and                     industries and occupations, or the distribution
the ECI index, are highly complementary.             A more         of income between factors of production.
comprehensive picture of the change in labor mar-                       The HEI measures neither the change in wages
kets is obtained by examining the series jointly,                   as income nor as costs.     The use of the HEI, and
than can be obtained from any single series con-                    the justification for publishing it, requires a
sidered separately.   The final series compared,                    historical digression.    The HEI was issued in
the HEI, does not contribute much additional                         1971 as an attempt to approximate the change in
information.                                                         labor cost until the ECI could be developed.
The Hourly Earnings Index                                           When it was issued the BLS stated:     "Even this
    The HEI forms an index by weighting simple                      measure [HEI] is defective in not adjusting for
average wages for detailed industries from the                      employment shifts among establishments within an
AHE by hours paid in the base year.         The HEI re-              industry or shifts in the occupational composi-
moves some of the impact of between establishment                   tion of employment within an establishment.     Un-
shifts.   Over the period shifts were from high to                  til a general wage index [ECI] is developed---ad-
 low wage establishments and the HEI increased                      justed average hourly earnings [HEll is the best
25.4 percent; 1.9 percentage points more than the                   comprehensive measure." 5/ The HEI meets the
AHE.   Shifts between establishments still had a                    needs of those who require a measure of the trend
greater impact on the HEI than employment shifts                     in wages that includes the years before 1975
within establishments, and the HEI increased 1.2                     (when the ECI was released), but do not require
percentage points less than the ECI (PROD).                          that the measure relate to any particular
The Time Period Compared                                            economic concept.
    The direction and magnitude of the differences                   Summary
between the series depend on the time period.                           There are many uses for wage and compensation
Had another time period been used, the relation-                    data.    The uses require measures that differ in
ships would be quite different.        The importance                the concept of pay---wages or compensation----the

           Employment Cost Index (Production) and Average Hourly Earnings
           changes for 12 month periods ending March, June, September and December

             Ernp!°ymenf- Index(Pr°ducfi°n) I                                               I





                                                                    [ Slowly 1                    [ Rapidly 1
                 .~         J '   Falling                    ~           rising   I---IP~I-I          rising
                 ~,      = l unemployment              ~'.          i unem-       I     -       -.i u n e m _        i.-

              1976       1977           1978          1979              1980          1981                1982             1985
   O   _

                 Source: Bureau of Labor Statistics
workers included, and the method used to compute                         ment Continued to Rise in 1982 as Recession
change.  The method of computing change deter-                           Deepened," Monthly Labor Review, February
mines if the measure is one of changing labor                            1983.
cost, or changing labor earnings.   The change in                  3/    For a discussion of pay levels by industry
labor cost relative to the change in average                             and occupation, see:   Profile of Occupational
earnings depends on employment shifts within a                           Pay: A Chartbook, U.S, Department of Labor,
group of workers.                                                        2037.
   The actual differences depend on the period                     4/    For additional discussion of the uses of the
covered.  Had a diSferent period been selected                           series, see: N. J. Samuels, "Developing a
the relationships would be different.   The change                       General Wage Index," Monthly Labor Review,
in the AHE could be greater than or equal to the                         March 1971; T. W. Gavett, '~easures of Change
change in the ECI (PROD). The AHC could differ                           in Real Wages and Earnings," Monthly Labor
substantially from the ECI compensation index.                           Review,   February 1972; V. J. Sheifer, "Em-
But whatever the magnitude or the direction of the                       ployment Cost Index: A measure of change in
diiferences, they would be determined by the                             the price of labor," Monthly Labor Review,
shifts in employment that occurred over the                              July 1975; J. E. Triplett, "Concepts of Qua-
period.                                                                  lity in Input and Output Measures:   A Commen-
          FOOTNOTES                                                      tary on An Old Debate," The U.S. National
I/ Additional information on all series used,                            Income and Product Accounts:   Selected Topics
    except the CPS data, can be found in: BLS                            (Murray Foss, ed.); J. E. Triplett, "A
    Handbook o5 Methods, Vol. I., U.S. Depart-                           Conceptual Model for Labor Market Data," pre-
    ment o5 Labor, Bureau of Labor Statistics,                           pared for the Session on Model Structure,
    December 1982, Bulletin 2134-1.   Information                        Data Bases, and Policy Issues, 95th Annual
    on the CPS data can be found in: Technical                           Meeting of American Economic Association
    Description o5 the Quarterly Data on Weekly                           (1981); J. E. Triplett, "An Essay on Labor
    Earnings from the Current Population Survey,                         Cost," BLS Working Paper 130, June 1981.
    U.S. Department of Labor, Bureau of Labor                      5/    T. W. Gavett, "Measures of Change in Real
    Statistics, January 1982, Bulletin 2113.                             Wages and Earnings," Monthly Labor Review,
2/ M. A. Urquhart and M. A. Hewson, "Unemploy-                           February 1972.


To top