View original document

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

measuring Produciwity
2/ t o
In State and Local Government
U.S. Department of Labor
Bureau of Labor Statistics
December 1983
Bulletin 2166







measuring Productivity
O State and Leeal Oewemment
n
U.S Department of Labor
Raymond J. Donovan, Secretary
Bureau of Labor Statistics
Janet L. Norwood, Commissioner
December 1983
Bulletin 2166


For sale by tlie Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402





Library of Congress Cataloging in Publication Data

Fisk, Donald M.
M e a s u r in g p r o d u c t i v i t y i n s t a t e a nd l o c a l g o v e r n ­
m ent o
( B u l l e t i n / U .S . D e p a r tm e n t o f L a b o r , B u re a u o f L a b o r
S t a t i s t i c s ; 2166)
" J a n u a r y 198U ."
Prepared by Donald M. Fisk.
B ib lio g ra p h y : p.
I . Government p r o d u c t i v i t y — U n i t e d S t a t e s — M e a s u r e ­
m e n t.
2 . S t a t e g o v e r n m e n ts .
3 . L o c a l g o v e rn m e n t —
U n ited S t a t e s .
I. T itle .
I I . S e r i e s : B u l l e t i n (U n ite d
S t a t e s . B u re a u o f L a b o r S t a t i s t i c s ) ; 216 6 .
JK 2^80 .L2^F5T

198 U

3 5 0 . 1 ^ 7 ’ 0973

83-600318

Pir@
fae<§

The Joint Economic Committee of the U.S. Con­
gress, the National Research Council of the National
Academy of Sciences, and the General Accounting Of­
fice have all called for further exploration into the
possibility of measuring State and local government
productivity. This bulletin was prepared in response to
the recommendations of these organizations and the
recognition that national productivity indexes are lack­
ing for the 13 percent of the civilian labor force that is
employed by State and local government.
The bulletin reviews past research in the area,
discusses conceptual issues, reviews national data which
could be used to calculate productivity, examines seven
State and local government services, and offers recom­
mendations for future research. While the focus is on
producing national indexes of State and local govern­
ment productivity, the concepts and procedures are
equally valid for individual governments. The study was




carried out during 1980-82. While some of the
statements and data presented have been superseded by
recent events, the basic conclusions remain unchanged.
Donald M. Fisk prepared this bulletin under the
supervision of Jerome A. Mark, Associate Commis­
sioner for Productivity and Technology. Dagmar Horna
assisted in the research and tabulations, and Rita
Walker typed the manuscript. Charles Ardolini, Chief
of the Division of Industry Studies, Office of Produc­
tivity and Technology, and his staff reviewed much of
the research and provided helpful comments. A number
of individuals outside b l s reviewed parts of the
manuscript; their contributions are acknowledged in the
appropriate chapters.
Material in this publication is in the public domain
and, with appropriate credit, may be reproduced
without permission.

O teiniBs
oim

Page
Chapters:
I. Introduction, conclusions, and recommendations................................................................
Types o f m easures..................................................................................................................
Basic measurement issu e s.....................................................................................................
Services exam in ed ..................................................................................................................
Current status..........................................................................................................................
A strategy for development...................................................................................................
II.

1
1
2
3
4
4

Background and uses of productivity measurements............................................................
Research and surveys..............................................................................................................
Decisions and the use of productivity measurements.......................................................
The focus .................................................................................................................................

7
7
9
11

III. Methodological considerations .................................................................................................
The production framework and p rocess............................................................................
The measurement of o u tp u ts...............................................................................................
The measurement of inputs...................................................................................................
Other issues...............................................................................................................................

12
12
14
16
21

IV. Measuring the measurable: Three case stu d ies........................................................................
Electric utilities........................................................................................................................
Institutional considerations...........................................................................................
Research and statistics.....................................................................................................
O utputs...............................................................................................................................
Labor inputs and employee c o sts..................................................................................
Productivity in d ex e s.......................................................................
Suggested research............................................................................................................

25
25
25
26
28
30
32
34

State alcoholic beverage control operations......................................................................
Institutional considerations...........................................................................................
Research and statistics................
O utputs...............................................................................................................................
Labor inputs and employee c o sts..................................................................................
Productivity in d ex e s.......................................................................................................
Conclusions and suggested research ............................................................................

34
34
37
38
40
41
42

Unemployment insurance...................................................................................................
Institutional considerations...................................................................
Research and statistics.....................................................................................................
O utputs...............................................................................................................................
Labor inputs......................................................................................................................
Productivity in d ex e s.......................................................................................................
Suggested research............................................................................................................

42
42
44
45
49
50
51

Thinking about the unmeasured: Four case studies..............................................................
Solid waste collection and d isp o sa l.....................................................................................
Institutional considerations.................................
Research and conceptual is s u e s .....................................................................................
O utputs...............................................................................................................................
Labor inputs......................................................................................................................
Suggested research............................................................................................................

52
52
52
54
55
58
59

V.




IV

Contemts— Goimtimuedi
Page

Chapter V—Continued
Drinking water supply..............................................
Institutional settin g.........................................................................................................
Research .............................................................................................................................
O utputs..............................................................................................................................
Labor inputs......................................................................................................................
Suggested research...........................................................................................................

59
59
60
62
64
64

Mass transit ............................................................................................................................
Institutional settin g.........................................................................................................
Research and conceptual is s u e s ....................................................................................
O utputs..............................................................................................................................
Labor inputs......................................................................................................................
Suggested research...........................................................................................................

65
65
67
68
71
72

The Employment Service .....................................................................................................
Institutional considerations...........................................................................................
Research and conceptual is s u e s ....................................................................................
O utputs..............................................................................................................................
Labor inputs......................................................................................................................
Suggested research...........................................................................................................

73
73
74
76
81
81

1. Basic model of the production process ................................................................................
2. Sophisticated model of the production process .................................................................

12
13

Charts:

Tables:
1. Terminology of government productivity measurement............................................
2. Illustrative matrix for selecting services for computing State and local
government productivity ...............................................................................................
3. State and local government employees by function, October 1977............................
4. Decisions and State and local productivity index requirem ents................................
5. Examples of steps in the production of selected government serv ices.....................
6. State and local government expenditures for salaries and wages as
a percent o f total expenditures by function, fiscal year 1977 ..................................
7. State and local government activities included in three labor
hour m easures...... ...........................................................................................................
8. Estimated use of volunteers by State and local government function .....................
9. Four national surveys used to collect State and local government
employment data..............................................................................................................
Electric utilities:
10. Distribution of kilowatt hours sold, customers served, and plant and
equipment owned by type of utility ownership, 1978 ...............................................
11. Finances of State and local government electric utilities by type of
government, fiscal year 1977 ............................................................. . .........................
12. Weights for calculating output indexes for State and local government
electric utilities by class o f service, 1967,1972, and 1977 ........................................
13. Three output indexes for State and local government electric utilities,
1967-78..............................................................................................................................
14. Two employment measures for local government electric
utilities, 1967-78 .............................................................................................................
15. Two employment indexes for private sector electric utilities, 1967-78.....................
16. Three employment indexes for State and local government
electric utilities, 1967-78.................................................................................................



v

2
5
8
10
13
17
18
19
21

26
27
29
30
31
31
32

C o n ten ts— Coontinued

Page

Electric utilities—Continued
17. Index o f average salaries and wages of local government electric utility
employees, 1967-78.........................................................................................................
18. Indexes o f output, employees, and output per employee for local government
electric utilities, 1967-78.................................................................................................
19. Indexes of output, employees, and output per employee for local and selected
State government electric utilities, 1967-78 ...........................................................
20. Weighted indexes of output, employees, and output per employee for 33
large government electric utilities, 1967-78 ...............................................................
21. Three productivity indexes for State and local government electric utilities,
1967-78.................................................................................................................
22. Average annual rates of change for government and private electric utility
output, labor input, and productivity, 1967-78.........................................................
State alcoholic beverage control operations:
23. Type o f control of alcoholic beverage sales by S t a t e ...................................................
24. Distribution o f alcoholic beverage control revenue and employees by State,
fiscal year 1977..................................................................................................................
25. Forms and functions of State alcoholic beverage control operations.......................
26. Selected data contained in State alcoholic beverage control annual rep o rts...........
27. Three gallonage indexes for alcoholic beverage control operations, 1968-78 .........
28. Three employment indexes for alcoholic beverage control operations, 1967-78 ..
29. Two full-time-equivalent employment indexes for alcoholic beverage control
operations, 1967-78 .......................................................................................................
30. October earnings for alcoholic beverage control personnel, 1967-78 .....................
31. Three productivity indexes for alcoholic beverage control operations, 1967-78 ..
Unemployment insurance:
32. Unemployment insurance programs, 1963-80 .............................................................
33. Unemployment insurance program benefits, selected years, 1965-80 .....................
34. Synopsis o f unemployment insurance activities...........................................................
35. Distribution o f time expended by Unemployment Insurance Service function,
fiscal year 1979............................................
36. Four output indexes for the Unemployment Insurance Service, fiscal years
1963-79.......................
37. Selected quality appraisal measures o f the Unemployment Insurance S ervice___
38. Positions in the State Unemployment Insurance Service,
fiscal years 1963-79................
39. Indexes of output, employee positions, and output per employee position,
Unemployment Insurance Service, fiscal years 1963-79..........................................
Solid waste collection and disposal:
40. Local government solid waste collection and disposal expenditures,
fiscal year 1977........................... •............................................................................ .......
41. Refuse materials by kind, composition, and source.....................................................
42. Studies o f effect of location and frequency o f collection on solid waste
collection costs..................................................................................................................
43. Community surveys o f solid waste collection ...............................................................
44. Local government sanitation employment, 1967-80 ...................................................
Drinking water supply:
45. Community water systems by size o f population, 1979 ...............................................



vi

32
32
33
33
34
34

35
35
36
39
40
41
41
41
42

43
43
45
46
47
49
50
50

53
54
55
58
59
60

C@nt®Bi]ts— ©®ot0
ny©d

Page

Drinking water supply—Continued
46. Finances of government-operated water utilities by type of government,
fiscal year 1977...........................................................................................................
47. Revenue of State and local government-owned water utilities,
fiscal years 1967-80...................................................................
48. Local government water utility employment, 1967-80 ............................................
Mass transit:
49. Finances of government-operated mass transit systems by type of government,
fiscal year 1977...........................................................................................................
50. Transit modes (private and public) ranked by passenger trips, revenue,
and miles, 1980 .........................................................................................................
51. Heavy rail operations in the United States, 1980 ......................................................
52. Examples of relationship of quality and level of service to program change,
labor cost, and two output measures........................................................................
53. Availability of employment data by transit mode ............................................: . . .
The Employment Service:
54. Placements per staff year as shown in the President’s budget,
fiscal years 1974-82...................................................
55. Comparison of Employment Service transactions and individual placements,
fiscal years 1975-80...........................................................................................
56. Agricultural and nonagricultural placement transactions, Employment
Service, fiscal years 1955-79 ...................................................................................
57. Five placement transaction time series, Employment Service, fiscal years
1955-79.......................................................................................................................
58. Missions and measures of performance of the Employment Service as
suggested by C.K. F airchild.....................

61
62
64

66
67
67
71
72

75
78
79
79
80

Appendixes:
A. Number of State and local governments and number of em ployees............................
B. Comparison of Bureau of the Census classification of government functions with
Standard Industrial Classification.....................................................................
C. Large electric u tilities.......................................................................................................

84
87

Bibliography ...................................................................................................................................

88




vii

82




Chapter I. Introduction, Conclusions,
aond ^®©@
mm@
rii(dlatn©sii

State and local government employment and expen­
ditures have increased greatly over the past two decades.
In 1960, these governments employed about 6.1 million
workers, or 8.7 percent of the civilian labor force, and
spent about $46.5 billion on the purchase of goods and
services, or 9.2 percent of the gross national product. By
1980, these governments employed 13.4 million, or 12.5
percent of the civilian labor force, and spent $341.2
billion on goods and services, or 13.0 percent of the
gross national product.
Despite the growth and current importance of State
and local governments, no national index of their pro­
ductivity is maintained such as those calculated for the
private sector or the Federal Government. Nor, ap­
parently, is there a current index for a single State or
local government service.1 State and local government
productivity remains the largest unmeasured sector of
the economy.
Several organizations have recommended, particular­
ly in the 197Q’s, that State and local government pro­
ductivity be measured. The Joint Economic Committee
o f the U.S. Congress, the National Academy of
Sciences, and the General Accounting Office have sug­
gested additional research to measure State and local
government productivity.2 Each group recognized the
problems associated with such an undertaking but
nevertheless believed its importance warranted further
investigation.

literature has expanded, definitions and concepts have
become more complex. Today, public sector productivi­
ty literature variously defines productivity as efficiency,
effectiveness, cost reduction, input-output, manage­
ment improvement, performance measurement,
methods improvement, systems analysis, work measure­
ment, and program evaluation.4 Because o f the confu­
sion over terms, some of the more important definitions
used in this study are presented in table 1. As Jerome A.
Mark notes, however, there is no best or right measure.5
Productivity measurement requires that measures be
shaped for the decision process, which leads to a variety
o f types of measures.
In this bulletin, government productivity measures
are assigned to one of three categories based on the type
o f output measure. These are: (1) Measures which focus
on operational issues, (2) those which focus on
organizational or program outputs, and (3) those which
are concerned with program consequences.
Operational measures are concerned with the internal
workings or efficiency of the organization. Work
measurement, which deals with resource requirements
under a given technology or set of conditions, is a com­
mon operational measure. Intermediate activities, such
as the number of reports produced, number o f audits
completed, or the number of samples tested, and utiliza­
tion measures, such as equipment downtime, are other
types o f operational measures. Each is important for
day-to-day management of government.
The second category of productivity measures, direct
outputs, is the final organizational output divided by
the resources used to produce the output. The direct
output productivity measure is the measure most com­
monly computed for the private sector. Public sector ex­
amples of such measures are the “ tons o f solid waste
collected per employee hour” for sanitation services and
the “revenue gallons of water sold per employee hour”
for water utilities. These measures are also known as
technical efficiency measures. They do not address the
issue o f whether the services should be produced or
relate them to some desired goal.

Types off measures
Considerable confusion has surrounded discussions
of the basic concepts and procedures used in govern­
ment productivity measurement.3 Furthermore, as the
1 Government enterprises, such as electric power utilities and water
supply, are included in the private sector productivity indexes
calculated quarterly by bls because their outputs are sold in the
marketplace. Enterprise services are included as a group; they are not
separated by type of government or service.
2 National Research Council, Measurement and Interpretation o f
Productivity (Washington: National Academy of Sciences, 1979), pp.
9-10; U.S. Congress, Joint Economic Committee, Productivity in the
Federal Government (Washington: Government Printing Office,
1979), p. 7; and General Accounting Office, The Federal Role in Im­
proving Productivity—Is the National Center fo r Productivity and
Quality o f Working Life the Proper Mechanism? May 23, 1978, p. 45.
3 Jerome A. Mark, “ Measuring Productivity in Govern­
ment-Federal, State, and Local,” Public Productivity Review, Mar.
1981, p. 21.



4
Jesse Burkhead and Patrick J. Hennigan, “ Productivity Analysis:
A Search for Definition and Order,” Public Administration Review,
Jan./Feb. 1978, pp. 34-40.
3 The Meaning and Measurement o f Productivity, Bulletin 1714
(Bureau of Labor Statistics, 1971), p. 7.

1

Table 1. Terminology of government productivity measurement
Term
Activity

The desired results of government programs or services such as improved
citizen safety, increased longevity, and
reduced infant mortality. Sometimes
described as impact or outcome.

Effectiveness

Input.............................. The resource used by an agency to produce
a function or service. Examples of inputs
are labor, facilities, equipment, and
materials.

A task performed by an organization to
produce a desired output. Examples in­
clude miles driven, trucks serviced, and
meters read. Sometimes described as
workload.

Consequence

The degree or extent to which program
goals are met, such as percent of
population served or percent of clients
successfully treated.
The ratio of output to inputs such as work
performed per staff hour or downtime
as a percent of total hours. Includes
productivity, unit costs, and technical
efficiency.
A government service such as police, fire,
and education. “Function” and “ serv­
ice” are used synonymously.
A statement which describes what is to be
accomplished by a program, service, or
agency such as “ insure a safe and
secure environment.”

Efficiency

Function

Goal

Impact.

Outcome........................

Short-term impact or consequence of
government action or outputs, such as
increased income which might come as a
result of job training.

O utput.......................... The result of work performed or produced
by an agency. Outputs are what govern­
ment produces. Examples of outputs are
the number of individuals served, gallons
of revenue water delivered, or tons of
trash collected.
Productivity..................

Inverse of the change in resources used per
unit of output.

Service..........................

A government function such as police, fire,
or education. The terms “ service” and
function” are used synonymously.

Social indicator ............

A measure of societal well-being such as
longevity or happiness. These measures
are of interest because they are con­
siderations which governments wish to
promote.

Workload......................

The long-term effect of a program on a
community of its citizens. Impact and
consequence are used synonymously.
See “ Consequence.”

Amount of physical output per unit of in­
put.

Productivity index........

The third category, consequences, addresses the issue
o f a program’s impact on society and whether that pro­
gram makes optimum use o f resources to achieve its
goals. This type of measure is alternatively known as
outcome, impact, effectiveness, and economic efficien­
cy. Examples of these types of measures are “ deaths
prevented per employee hour” for fire departments and
“jobs created per employee hour” for economic
development agencies. Measures such as these focus on
consumers and consumption whereas operational and
direct output productivity measures are concerned with
production relationships.
Each o f these three general types of productivity
measures is important. However, the most common, at
least nationally, is the second type, the direct output or
technical efficiency measure. It is this type of measure
which is most often computed for the private sector and
the one with which this study is primarily concerned.
In many ways, the measurement of productivity in
government and private organizations should be
similar. Both types of organizations produce goods and
services, both compete in the marketplace to purchase
resources, and both use varying combinations of
resources to produce a product or service.
There are important differences, however. Probably
the most important is the absence o f a market and a
market price for most government production. Govern­



Definition

Term

Definition

A measure of the amount of work performed,
usually an intermediate output, such as
the number of miles driven, or number
of machines serviced. See “ Activity.”

ment officials, not the marketplace, decide what to pro­
duce.

Basse msasuromesit tssyes
Specification and measurement o f output are the
most difficult problems in measuring the productivity of
State and local government. The basic measure o f out­
put should be a homogeneous physical unit, with the
unit measure of output related to the unit labor hours
spent in its production. Where a government provides a
single service, as in the case of some of the special
districts—solid waste disposal and drinking water are
examples—the output can be simply a count o f the units
of service. However, most governments produce a
number of heterogeneous services, and it is often dif­
ficult to even identify the basic services.
Furthermore, most government services are compos­
ed of a number of different subservices or products
which also are difficult to identify. For example, the
direct outputs (not consequences) of police and educa­
tion services are not easily specified. For productivity
measurement, it is necessary to specify homogeneous
service outputs.
In addition, if the output index is to reflect trends ac­
curately, the service units must be homogeneous
through time. In many instances, the scope and dimen­
sions of government services are constantly changing.
2

Many transit systems now provide demand response ser­
vice in addition to regularly scheduled bus service, and
some jurisdictions have added the testing o f automobile
emissions to safety inspections. In both cases the service
unit has changed, which requires product differentia­
tion.
Quality and level o f service considerations are also
important for productivity measurement because of
their potential impact on the unit o f service. To
distinguish service or product shifts from changes in
unit labor requirements, or productivity, outputs or in­
puts must be adjusted or a new product identified.
Movement o f solid waste collection from backdoor to
curbside pickup and improvement o f drinking water
quality to conform with environmental standards will
affect unit labor requirements, and reduction o f welfare
error rates may affect unit labor requirements.
Selection o f the proper measure of output requires a
service-by-service and product-by-product approach.
By dividing a service, it is possible to identify
homogeneous outputs with reasonably stable unit labor
requirements. The difficulties with this approach are a
lack o f research to identify the correct units and a lack
o f data to make the calculations.

no national statistics are available for labor hours of
State and local governments; few governments even col­
lect such data. A measure often used as a proxy for the
number o f hours is the number of full-time-equivalent
employees. Most State and local governments maintain
such statistics. Most governments also have statistics on
the number of employees, a measure widely used in the
private sector.
However, none of the sources of national statistics
are entirely satisfactory for computing individual serv­
ice labor productivity indexes. National statistics col­
lected through the Bureau of the Census’ Census of
Governments, the Employment Service’s ES-202
report, and the Bureau o f Labor Statistics’ Current
Employment Statistics survey (CES-790) are not divided
sufficiently to compute labor productivity indexes for
individual services. In a few instances, Federal Govern­
ment programs, such as unemployment insurance and
the Employment Service, and professional interest
groups, such as the American Public Transit Associa­
tion, do collect data on State and local government
employment. However, comparisons of labor data
drawn from these and other sources reveal considerable
discrepancies.
In summary, no single source o f labor data on State
and local government is adequate to compute labor in­
dexes. Most contain errors. Construction of viable labor
indexes, either total or for individual services, will re­
quire detailed comparison and adjustment o f the data.
As with the private sector, cyclical and secular change
can dramatically affect productivity trends. Cyclical
productivity change most often occurs in services such
as unemployment insurance, where inputs cannot be ad­
justed as rapidly as outputs change. Secular trends are
found in services for which a fundamental change oc­
curs over time, such as electric power. Generally speak­
ing, the greater the number of years included in the pro­
ductivity index, the more representative the index will be
o f long-term trends. Data availability will normally set
the outer bounds for the years measured.

Data to calculate aggregate national, State, and local
government output indexes are generally lacking. The
Federal Government collects some data, particularly in
those areas where it has shared responsibilities, such as
unemployment insurance and drinking water, and some
data are collected by national associations and public in­
terest groups. These statistics are often inaccurate and
incomplete. But more often than not, national statistics
are simply unavailable on State and local government
output.
To collect output data through a regular survey of
State and local governments would be extremely expen­
sive. But even if the decision were made to spend the
necessary funds, there is the fundamental question of
what data to collect. For some services, there is a
reasonably good idea of the data needed on type of out­
put and level and quality o f service for productivity
measurement, but for most services this information is
unknown. In short, it makes little sense to establish a
data collection system at this time. However, if a system
is established to collect State and local government out­
put data, input information should be collected
simultaneously.
The most frequently used measure o f input is labor.
Constituting over half o f all State and local government
operating expenditures, labor is important for public
policy considerations, is easy to calculate compared
with other factors of production, and is the most ac­
cessible o f State and local government factor inputs. It
is the measure recommended for State and local govern­
ment productivity calculations.
The preferred labor measure is labor hours. However,



Sen/ices examined
Seven State and local services were selected for this
report from dozens provided by the government. The
more important government services, such as education,
police, and fire, are not included because of conceptual
or data problems.
For three services—electric power, State alcoholic
beverage stores, and unemployment insurance—
illustrative indexes are calculated. For two services
—sanitation and drinking water—productivity was not
calculated because national data are lacking. For the re­
maining two services—transit and the Employment Ser­
vice—productivity indexes were not calculated because
o f unresolved conceptual and data issues; additional
research is being conducted in these two areas. The
results for each service are briefly noted below.
3

calculated “ labor productivity” for several years but
has been criticized for the measure—placements—and
the resulting measurements. Placements measure out­
come more than output. Also, serious questions have
arisen about the accuracy of data used to calculate
“ productivity.”

Electric power. Considerable research has been con­
ducted into private electric power productivity, and con­
siderable data are collected on private and public
utilities. In 1978, about 2,200 State and local govern­
ment electric power utilities employed about 66,000
workers. From 1967 to 1978, labor productivity increas­
ed an average o f 3.0 percent annually, output (kilowatt
hours) 4.0 percent, and labor input 0.9 percent.

Current s ta tu s
These seven services illustrate the problems and op­
portunities that occur in computing State and local
government productivity indexes. The problems are
substantial and include both conceptual and data issues.
However, the difficulties may not be any worse for
calculating State and local government productivity
than for calculating private sector service industry pro­
ductivity.6
Both sectors produce many of the same services.
There are literally dozens o f such services, ranging from
electric power to alcoholic beverage sales to hospitals to
employment counseling. Not every government service
has its private sector counterpart, but most do.
Furthermore, similar economic forces seem to be at
work. For electric power, the productivity trends in the
two sectors are similar; productivity is affected in both
by economies of scale; and the slowdown in productivi­
ty growth is evident in both. For the Unemployment In­
surance Service, the fluctuations in output mirror shifts
in the number of unemployed. Also, as in the private
sector, productivity increases rapidly during periods of
increasing unemployment as work increases more rapid­
ly than the number of workers (the opposite occurs dur­
ing periods of decreasing unemployment). Quality ap­
parently decreases during periods of high workload but
improves as the workload drops. Productivity trends are
dramatically influenced by the beginning and ending i
points of the index as a result of the cyclical fluctua­
tions.
Much of the past discussion on calculating govern­
ment productivity has been entangled in questions of ef­
fectiveness, outcome, and impact. Productivity analysis
in these areas has become entrapped in externalities. As
long as the discussion is restricted to direct outputs, the
solutions are at least as tractable as in the private sector.
This is not to say that productivity can be computed
for every State and local government service. Thorny
problems exist in calculating State and local government
productivity, just as in the private sector. However, it
should be possible to compute State and local govern­
ment labor productivity for many services.

State alcoholic beverage sales. Seventeen State govern­
ments sell alcoholic beverages. The preferred measure
o f output for retail store operations is bottles; for
wholesale operations, cases. A proxy measure, gallons,
was used to calculate government productivity. Between
1967 and 1978, the average annual increase in produc­
tivity was 2.5 percent; in output, 3.1 percent; and in
labor, 0.6 percent.

Unemployment insurance. Considerable data are
routinely collected on this joint Federal-State program.
In 1978, about 48,000 State employees operated the pro­
gram. Two primary measures of output are: (1) Services
to those applying for or drawing benefits and (2) collec­
tion of funds from employers. A weighted program out­
put measure shows that, between 1965 and 1978, labor
productivity increased at an annual rate of 1.7 percent,
output at 7.4 percent, and labor input at 5.7 percent.
Solid waste collection and disposal. The preferred
measure o f output in this area is tonnage. No national
data are routinely collected on outputs, and input data
are insufficient for productivity calculations. Hence, a
productivity index was not calculated. Local govern­
ment had about 128,000 employees in this area in 1980.
Drinking water supply. Revenue gallons is the preferred
measure of water utility output. Most utilities routinely
collect gallonage data and some data are collected na­
tionally, but there are so many questions about the ac­
curacy o f these data that no index was calculated as part
o f this study. In 1980, about 134,000 State and local
employees worked in this area.

Mass transit. Research into mass transit operations and
calculation o f outputs, inputs, and productivity have a
long history. For some time, the proper measure of out­
put has been debated. On the one side are the tradi­
tionalists who favor passenger miles or a similar
measure. On the other side are the transit operators who
favor an availability or capacity measure such as vehicle
miles. Since 1979, national data have been collected for
both output measures, and for labor inputs.

A strategy for developm ent
Development o f State and local government produc­
tivity indexes must proceed service by service. A threestep process is suggested for developing each service in-

Employment Service. Considerable research has been
done and data collected routinely on the Employment
Service ( e s ), a joint Federal-State operation, e s has



6
Jerome A. Mark, “ Measuring Productivity in Service Industries,”
Monthly Labor Review, June 1982, pp. 3-8.

4

the addition o f the Employment Service and the
Unemployment Insurance Service.
Although the classification system is arbitrary and the
assessments illustrative, the matrix demonstrates some
o f the opportunities and problems in computing State
and local government productivity indexes. First, com­
puting a national index will not be easy for any impor­
tant service. Second, national data and/or a tangible set
o f outputs exist for most services.
After the service has been selected, an initial recon­
naissance will be needed to assess the feasibility o f com­
puting an index. This initial review will normally be suf­
ficient to identify which services are good candidates for
calculating productivity. An in-depth investigation, the
next step in the process, will determine whether an index
computation is possible.
Construction of an index requires a detailed review of
research and conceptual issues, development o f a list of
potential output measures, evaluation of each measure,
selection of the recommended measure, search for data
to compute the index, development o f surrogate or
proxy measures if the data are not available, and, final­
ly, documentation of the results. This bulletin il­
lustrates, for seven services, how documentation might
proceed.
Maintenance of an index requires constant vigilance
and analysis. Changes may occur in the quality and level
of service; intermediate outputs may be contracted out,
which would reduce labor input but not output; and
data series may change.

dex: (1) Select the service to be examined, (2) review ex­
isting research and data, and (3) calculate the index and
document the result. Once established, the index needs
to be updated annually.
In selecting the service to be examined, two—often
conflicting—criteria need to be weighed. One is the im­
portance o f the service judged by its cost or the number
o f employees. The other is the difficulty of computing a
viable index. Education, for example, is the most im­
portant service in resources used but is also one o f the
most difficult areas in which to measure outputs.
Alcoholic beverage sales, on the other hand, is one of
the easier services for measuring outputs but is relative­
ly unimportant in terms o f employment. In selecting a
service to be examined, these two criteria need to be
balanced.
A suggested procedure for selecting a service is to
compare the two criteria—importance and dif­
ficulty—using relative rankings such as those shown in
table 2. Importance, for example, can be measured by
the number o f employees. A three-part ranking is used
in the table: (1) 100,000 employees or less, (2) more than
100,000 but less than 500,000 employees, and (3) more
than 500,000 employees. Difficulty can be assessed by
examining the tangible nature of outputs and the
availability o f national output data. A three-part rank­
ing is used: (1) Tangible outputs and national data, (2)
tangible outputs or national data, and (3) neither tangi­
ble outputs nor national data. The services shown in the
matrix are those listed by the Bureau of the Census, with

Table 2. Illustrative matrix for selecting services for computing State and local government productivity
Importance of service
(number of State and local government employees)

Difficulty of
computing productivity
100,000 or less

100,000 to 500,000

Least difficult—has tangible outputs
and national d ata................................ Electric power
Unemployment insurance
Sewerage
Airports
Alcoholic beverage stores
More difficult—has tangible outputs
or national d a ta .................................

Water supply
Mass transit

Employment Service
Aid to families with dependent
children
Food stamps
Libraries
Housing and urban renewal
Water transportation
Gas supply

5

Corrections
Sanitation
Natural resources

Highways
Hospitals
Police
Higher education
Local schools

Fire
Financial administration
Parks and recreation
Health

Most difficult—has neither tangible
outputs nor national d a ta ..................




500,000 or more

General control

After the development of individual indexes, it should
be possible to construct aggregate indexes for functional
segments, such as for the enterprise services. This group
of services warrants early examination.7 This bulletin
examines four such services—alcoholic beverage sales,
electric power, transit, and water supply—which ac­
count for about two-thirds of total salary and wage ex­
penditures of enterprise services. Services which fall into
this category would be among the easiest for computing
productivity. They are already included in the National
Income and Product Accounts and in national private
sector productivity calculations.
Income maintenance programs, such as aid to
families with dependent children, food stamps, and
unemployment insurance also are suitable for produc­
tivity measurement and functional grouping. The con­
ceptual issues are tractable and data are available for the

major ones as a result of the Federal role in funding and
oversight.
Service groupings, such as the enterprise services and
income maintenance programs, could also be used for
comparison and benchmarking. Enterprise group
statistics can be compared with the National Income
and Product Account data. Social insurance programs
(unemployment insurance, the Employment Service,
and other labor programs); utilities (water, gas, electric
power, and transit); and transportation (highways, air,
and water) can be compared with Bureau o f the Census
employment groupings.
By using the building block approach, individual in­
dexes and groups o f indexes might be combined into ap­
propriate functions, such as public works and public
safety. Eventually it may be possible to develop a na­
tional productivity index for State and local govern­
ment, but this is probably many years away.8

7
The Bureau of Economic Analysis of the U.S. Department of
Commerce assigns 15 State and local government services to this
8
A national index does not have to include every service, but, to be
group. These services are largely sold in the marketplace and conse­
representative, 90 percent of the labor input should be included.
quently have a price associated with them.




6

©huptdir IS i®©kgr@yinidl © d U id t
.
fin !
® Produetiwiity Measurements
ff

About §0,000 State and local governments in the
United States serve their citizens through a vast variety
of programs. State employees operate such diverse serv­
ices as hospitals, universities, forests, hatcheries,
prisons, lotteries, alcoholic beverage stores, and grain
elevators. Employees of municipalities, townships,
counties, school districts, and special districts sweep
streets, inspect restaurants, manage golf courses,
operate swimming pools, assess real estate, counsel drug
addicts, put out fires, direct traffic, and teach students.
The number of employees in each of these areas varies
dramatically. There were over 1 million State employees
in higher education in 1977 but less than 16,000 in State
alcoholic beverage stores, and more than 4.8 million in
local schools but less than 100,000 in local libraries
(table 3). Also, the number of paid employees ranged
from -358,497 (New York City) to 0 for some special
districts.
While there is great diversity in State and local
government operations there are, at the same time,
many similarities. Every State operates an employment
service, an unemployment insurance service, a food
stamp program, a university system, a court and penal
system, and a highway system. And most municipalities
operate fire, police, sanitation, library, and recreation
services. Such similarities permit computation of na­
tional indexes for State and local government produc­
tivity.

Research and stireays
State and local government output, employment, and
productivity have been examined by a number of
economists and members of the research community.
Several scholars examined aspects of the subject in the
latter part of the 1960’s, when State and local purchases
and employment began their dramatic growth.
In 1967, William I. Baumol raised the question of
why the quality of life apparently deteriorated in urban
areas when State and local governments spent more and
more resources to solve problems. He concluded that
productivity growth was extremely low or nonexistent,
and furthermore, that little could be done to improve
the situation. The Baumol thesis was based on a twosector conceptual model of the economy, one sector
characterized by high technological inputs and a high
rate of productivity growth and the other having highlabor inputs and a low rate of productivity growth.
State and local government services, according to
Baumol, lay mostly in the second sector.2
Bradford, Malt, and Oates examined the Baumol
thesis from readily available data for health and
hospitals, education, police, and fire. Their conclu­
sions, which were based primarily on unit cost data,
generally substantiated the Baumol thesis, that is, rising
unit costs and decreasing or slowly increasing produc­
tivity.3
Robert M. Spann also examined the Baumol
hypothesis from data for six State and local government
services—police, fire, highways, general control, finan­
cial and administration, and public welfare. His ap­
proach, output measures, and conclusions generally
parallel those of Bradford, Malt, and Oates. However,
he argued that the lack of productivity growth in State
and local government was due to bureaucratic in­
fluences, not simply to the labor-intensive nature of
State and local government services. In coming to this
conclusion, he compared selected private sector produc­
tivity measures with public sector statistics. He conclud­
ed that private sector services exhibited low, but positive

Furthermore, the major services account for most
State and local government employment—education,
hospitals, police, and highway programs alone account
for 70 percent (table 3).
Also, the large jurisdictions employ a significant pro­
portion of all State and local government workers. The
10 largest States account for about 48 percent of all
State employment; the 10 largest special districts, for
about 25 percent of special district employment; the 10
largest municipal governments, about 22 percent of all
municipal employment; the 10 largest school districts,
about 12 percent of total school district employment;
and the 10 largest counties, about 11 percent of all coun­
ty employment.1

2 William J. Baumol, “Macroeconomics of Unbalanced Growth:
The Anatomy of Urban Crisis,” American Economic Review, June
1967, pp. 415-26.
3 D.F. Bradford, R.A. Malt, and W.E. Oates, “ The Rising.Cost
of Local Public Services: Some Evidence and Reflections,” National
Tax Journal, XXII, No. 2 (June 1969), pp. 185-202.

1 See appendix A.



7

Tabte 3. SSat© and local government! employees by function,
October 1077
Function

tivity (efficiency), found that State and local govern­
ment probably had a negative rate of growth in produc­
tivity between 1959 and 1979, the years for which data
were available.6
A number of studies of individual State and local
government functions have focused on a few subject
areas to identify underlying economic relationships
rather than develop specific productivity indexes. Solid
waste collection, mass transit, and water supply, for ex­
ample, have been extensively investigated, as will be
discussed later in this study.
Many areas have had little investigation of the basic
underlying economic relationships and no developmen­
tal work on productivity measurement. Services such as
parks and recreation, elderly day care, animal control,
and general management have been skipped over entire­
ly or examined only superficially.
Although information on productivity measurement
in individual State and local governments is sparse, the
majority of medium- and large-size governments collect
some data. A 1976 survey found that 65 percent of the
cities and 50 percent of the counties collected and used
efficiency-measures.7 Surveys of local governments in
N orth C arolina and the Denver, C olorado,
metropolitan area in 1978 found the figure to be over 85
percent.8 A survey of selected governments, also in
1978, found that 68 percent of the cities and 47 percent
of the counties with productivity improvement pro­
grams had measurement systems. The same survey
found that 79 percent of the States with improvement
programs had some type of measurement system.9 An
examination of State budgets and discussions with State
budget officers in 1975 found frequent use of produc­
tivity measures although few formal systems.1
0
These and other studies suggest the following conclu­
sions for State and local government productivity
measurement. First, State and local governments collect
considerable data although the information is uneven as
to function and government. Some functions,, par­
ticularly those with tangible products and Federal in­
volvement, are well covered at the individual govern­
ment level. Second, many studies cover individual serv­
ice areas. Most are cross-sectional; a few examine time
series data. Third, several studies of State and local
government productivity have been made from readily

Number Percent of Percent of Percent of
(thousands)
total
State total local total

T o ta l......................................
S ta te ..............................
Local..............................

12,765
3,491
9,275

100

Education..............................
Local sch o o ls................
Higher education ..........
Other education............

6,703
4,885
1,724
94

53
38
14
1

43
1
39
3

56
52
4
-

Hospitals................................
P o lic e ....................................
Other and unallocable............
Highways..............................
General control......................
W elfare..................................

1,105
622
595
563
502
369

9
5
5
4
4
3

16
2
6
7
3
5

6
6
5
3
4
2

Local utilities..........................
Water supply..................
Electric power................
T ransit............................
Gas supply ....................

324
121
64
127
13

3
1
1
1

_
-

4
1
1
1

Financial administration........
Fire protection ......................
Corrections............................
Parks and recreation..............
Health....................................

291
270
224
217
212

2
2
2
2
2

Natural resources..................
Sanitation..............................
Social insurance....................
Libraries................................
Housing and urban
development......................
Sewerage..............................
Airports..................................
Liquor stores..........................
Water transportation..............

209
127
117
91

2
1
1
1

90
86
19
16
14

1
1

100
100

-

-

2
3
1
2
1

3
4
3
5

-

3

-

-

1
1

-

1
1
1

-

-

Source: 1977 Census of Governments—Compendium of Public
Employment (Bureau of the Census, 1979), pp. 13-14.

gains in productivity, unlike their public sector counter­
parts, which showed declining productivity.4
Probably the most thorough examination of local
government productivity to date was one undertaken by
Ross and Burkhead. In addition to reviewing the
research of others and setting the conceptual founda­
tions, they attempted to estimate that part of the change
in local government expenditures which was due to
shifts in cost and workload and the part which was due
to shifts in quality and productivity. Their empirical
work examined education, welfare, police, and fire for
several of the larger cities in New York State, and
welfare and education for all local governments in New
York State. Except for fire protection, they found little
evidence of increasing local government productivity.5

6 Charles R. Hulten, Productivity Change in State and Local
Governments (Washington: The Urban Institute, 1981).
7 Rockham Fukuhara, The Status o f Local Government Produc­
tivity (Washington: The International City Management Association,
March 1977).
8 Comparative Performance Measures fo r Municipal Services
(Raleigh: Research Triangle Institute, December 1978); A Demonstra­
tion o f Comparative Productivity Measurement (Denver: Denver
Regional Council of Governments, December 1978).
9 State and Local Government Productivity Improvement: What
is the Federal Role? (Washington: General Accounting Office, Dec. 6,
1978), p. 13.
1 The Status o f Productivity Measurement in State Government:
0
An Initial Examination (Washington: The Urban institute, 1975).

More recently, Charles R. Hulten used the household
production function to examine the change in State and
local government productivity. His model, which cap­
tured the consequences (effectiveness) as well as produc4 Robert M. Spann, “ Rates of Productivity Change and the
Growth of State and Local Expenditures,” in Thomas E. Borcherding, Budgets and Bureaucrats (Durham, N.C.: Duke University
Press, 1977).
5 John P. Ross and Jesse Burkhead, Productivity in the Local
Government Sector (Lexington, Mass.: Lexington Books, 1974).



8

and from 1971 to 1974 changes in labor productivity
were considered in allowing or disallowing proposed
price increases. In 1978, wage increases above the
guideline were allowed if it could be demonstrated that
explicit changes in work practices led to increased pro­
ductivity.1 The general policy for public sector wage in­
5
creases during this period was to permit the public sec­
tor the same average increase as the private sector. This
percent has generally been tied to the estimated increase
in private sector productivity.
A number of studies have sought to measure the im­
pact of productivity on economic growth. Most of these
have been analyses of the private sector.1 Similar re­
6
search on the public sector would assist decisionmarkers
in formulating national policy. Knowledge concerning
the relationships among State and local government
productivity, the factors of production, research and
development, and economies of scale would be helpful.

available time series data. Although ad hoc and
restricted to a few areas, they have generally concluded
that State and local government productivity has re­
mained stagnant or decreased over the past several
decades.

D©oisI©ns and the use ©f productivity
measurements
Although State and local government productivity
should be measured for many reasons, the ultimate
reason is better management. Four types of manage­
ment decisions that could benefit from State and local
government productivity measurement are: Policy for­
mulation at the national level; program management at
the Federal level; policy, planning, and programming in
State and local government; and day-to-day operations
in State and local government. This section discusses the
kinds of decisions to be made, the types of decision­
makers who need the data, and the types of indexes
needed (table 4).

Federal program management. In addition to pro­
viding a valuable tool for formulating national policy,
State and local government productivity indexes could
help Federal program managers who are interested in
State and local government. Federal managers often
have such an interest because (1) State and local govern­
ments often operate Federal programs; (2) the financial
health of these governments directly concerns the
Federal Government and the Nation as a whole.
Most State and local governments operate some
Federal programs. In fiscal 1980, the Federal Govern­
ment transferred approximately $92 billion to about
65,000 State and local governments in direct Federal
aid. In 1978, 492 different grant programs funnelled
money to State and local governments.1 Although the
7
number of grant programs and total dollars funnelled to
State and local government have decreased slightly, they
will remain an important consideration, and the Federal
Government has a legitimate concern as to how effi­
ciently these dollars are spent.
Even if State and local governments did not operate
Federal programs, their productivity and how it is
changing would still concern Federal policymakers
because of their importance to the national economy.
State and local government purchases of goods and serv­
ices accounted for about 13.0 percent of the Nation’s
gross national product or $341.2 billion in 1980. Some
Federal agencies operate programs to improve State and
local government productivity. These programs range
from funding research and development to operating in­
formation clearing houses for technical assistance.
State and local government productivity indexes

National policy formulation. Productivity indexes are
important tools in forecasting national income, projec­
ting national labor demand, formulating national wage
policies, and assessing the sources of national growth.1
1
Projected changes in productivity, for example, are
often used in forecasting gross domestic product and in
setting fiscal and monetary policy. These projections
often include estimated productivity changes in the
private sector, but normally assume no change in the
public sector. The bias that results by assuming no
change in government productivity is not known but
could be substantial.1
2
Labor market projections and analyses often take in­
to account the impact of changes in productivity in the
private sector.1 The demand for labor can be
3
dramatically affected by such changes. State and local
government productivity indexes could help answer
questions about how changes in State and local govern­
ment productivity affect the supply of and demand for
labor in the economy, how the national supply of and
demand for labor affect State and local government
productivity, how geographic shifts of the population
affect State and local government productivity, and
how new technology influences State and local govern­
ment productivity and the demand for labor.
National income policies and guidelines have long
taken into account changes in productivity.1 The 1962
4
Economic Report o f the President laid down guidelines
for noninflationary wage behavior in the private sector,
1 National Research Council, Measurement and Interpretation
1
o f Productivity (Washington: National Academy of Sciences, 1979).
1 Jerome A. Mark, “ Progress in Measuring Productivity in
2
Government,” Monthly Labor Review, December 1972, pp. 3-6.
1 Methodology fo r Projections o f Industry Employment to
3
1990, Bulletin 2036 (Bureau of Labor Statistics, 1980).
1 National Research Council, op. cit., p. 27.
4



1 Economic Report o f the President (Council of Economic Ad­
5
visers, January 1979), p. 81.
1 National Research Council, op. cit., pp. 145-65.
6
1 The Federal Role in the Federal System: The Dynamics of
7
Growth (Washington: Advisory Commission on Intergovernmental
Relations, December 1980), p. 8.

9

Table 4. Decisions and State and local productivity index requirements
Decision area
National policy form ulation........

State and local government
operations ..................................

Aggregate national indexes includ­
ing State and local government

Congress, Federal officials, and
academicians

National indexes including State
and local government by serv­
ice area

Problem identification
Fiscal analysis
Legislative impact

State and local government
officials, Congress, Federal
officials, and academicians

State and local government
indexes by service area

Setting goals and objectives
Estimating resource requirements
Budget justification
Cost reduction
Scheduling and control of
operations
Accountability
Motivation for improvement

State and local government
officials and academicians

Individual State and local
government indexes by serv­
ice area

would help answer such questions as: Where is produc­
tivity lagging? What are the sources of the decline? How
can productivity improve? Is additional research need­
ed? Does productivity vary by geographic area? If so,
why? What is the relationship between productivity im­
provement and the financial strength of State and local
governments? And so forth.
There is considerable discussion of the impact of
Federal legislation and administration on State and
local productivity. In 1979, an estimated 1,259 man­
dates were in effect, of which 223 were direct orders and
1,036 were conditions of aid.1 Anecdotal examples
8
abound. One report concluded that well-executed
changes in the Federal grants program would raise State
and local government productivity.1 At this time, little
9
is known of the impact of Federal legislation and its ad­
ministration on State and local government productivity.
Some Federal agencies are directly involved in State
and local government operations. The U.S. Employ­
ment Service ( e s ) and the U.S. Unemployment In­
surance Service (UIS) are operated by State personnel
but funded by the Federal Government. Budgets to sup­
port these operations are prepared by the U.S. Depart­
ment of Labor and defended before the U.S. Office of
Management and Budget and the Congress. Productivi­
ty measures have played an important role in justifying
and defending these budgets. The es includes produc­
tivity calculations in its annual budget. The uis uses

detailed work standards to justify its requests for funds
to support State staff.
Productivity statistics are also used to allocate funds
to the States. The e s has used productivity as one ex­
plicit variable in its allocation formula, and the UIS had
used projected work and estimated work standards in
making its allocations.
The Federal Government is also directly involved in
the design and organization of some State and local pro­
grams. In addition to the e s and uis, the Federal
Government oversees other programs such as the Work
Incentive Program, Food Stamps, and Aid to Families
with Dependent Children. A better understanding of
productivity would help in questions such as whether to
decentralize or centralize services and how to allocate
resources among various activities such as counseling,
testing, and training.

State and local policy formulation. State and local
government policy formulators also could benefit from
productivity statistics in identifying potential problem
areas, estimating fiscal effects, and assessing the impact
of State and local legislation.
Some States assign operating responsibilities to local
government and provide partial funding, as in the case
of the Aid to Families with Dependent Children and
Food Stamps programs. Many States assign the respon­
sibility for road maintenance to local government along
with receipts from gasoline taxes. States have local
government oversight responsibilities, one phase of
which is fiscal solvency.

1 The Federal Role in the Federal System, p. 4.
8
1 State and Local Government Productivity Improvement: What
9
is the Federal Role? pp. 41-51.



Type of index

Congress, Federal officials, and
academicians

Gross domestic product
forecasting

Federal program formulation . . . . Problem identification
Legislative impact analysis
Budget justification and
allocation
Program and organization
design
State and local government
policy formulation ....................

Decisionmaker and data user

Decision category

10

Budget justification. Projections of resource re­

Productivity measurement can also play an important
role in State fiscal analyses. The passage o f propositions
13 in California and 2-1/2 in Massachusetts led to selec­
tive employment cuts, the net effect of which is widely
debated but unknown. Some States, such as North
Carolina, fund many local services. Accurate produc­
tivity estimates should result in better fiscal estimates
and help conserve scarce resources. Some cities, most
notably New York, have tied increases in pay to in­
creases in productivity. The measures for assessing
change in productivity are generally lacking.
Finally, State legislation affects the productivity of
local governments just as Federal legislation affects the
productivity o f State governments. Most States require
detailed reports from local governments, and some
restrict local government operations. Some States re­
quire analyses of the fiscal impact of any new legisla­
tion. The effect of new legislation on local government
productivity is not generally known.

quirements, including changes in productivity, are
important parts o f budget preparation and
justification. Capital projects are sometimes
justified by their potential positive impact on pro­
ductivity.

Cost reduction. Productivity measurement provides
a base for measuring change and goals.
Scheduling and control o f operations. Productivity
measures provide techniques for scheduling work,
routing crews, monitoring work performed, and
comparing direct labor with indirect.
Accountability. Productivity measurement may
lend credibility to government operations by mak­
ing managers more accountable and giving the
public a tool with which to assess government
operations.

M otivation f o r im provem ent. P rod u ctivity
measurement can also provide documentation for
bonuses, special recognition, group incentives, pro­
motions, and productivity bargaining.

State and local government operations. Productivity
measures and data probably find their greatest use to­
day in the day-to-day operations of State and local
governments. The following issues have been among
those suggested:2
0

Thefoeus

The four decision areas discussed require, in most
cases, very different types o f productivity indexes (table
Setting goals and objectives. A productivity
4). National policy issues normally would require ag­
measurement system lends specificity to a process
gregate national indexes. Federal program formulation
that is usually general in form and substance. Con­
and management and State and local government plann­
crete data also show managers and workers how
ing and programming would benefit from national,
well goals are met.
regional, and State indexes by functional area. State and
local government operations would require functional
Estimating resource requirements. Productivity
indexes by government.
measurement helps managers better estimate their
This study focuses on midlevel decisions, Federal pro­
resource requirements. For example, productivity
gram management, and State and local government pro­
change should be considered in estimating work
gramming. Conceptual issues and data availability dic­
force needs.
tate this approach. As functional indexes are developed,
2
0
“ Implementing a Productivity Program: Points to Consider” it may be possible to calculate aggregate national and in­
(Washington: Joint Financial Management Improvement Program,
dividual government indexes.
March 1977), pp. 20-26.




11

Ghaptnir D L M©1 dl© gS© D G [niiid<§[r§i'8
D
i[h® l® ® @
i®ms

The methodology underlying the measurement of
State and local government productivity is a fundamen­
tal issue on which there is considerable disagreement.
Not even the definitions and terminology are consistent­
ly applied. This chapter discusses the basic conceptual
issues and presents the approach used in succeeding
chapters.

The pr@
dueS!@ framework and process
n
The measurement of government output may be ap­
proached in two ways: One focuses on welfare aspects
and considers utility functions, indifference curves, and
community satisfaction; the other focuses on produc­
tion possibilities and considers production frontiers and
comparisons. This study deals with production pos­
sibilities. This approach assumes that government pro­
duction decisions can be modeled through a production
function framework similar to that used in private sec­
tor productivity analysis. It requires the general iden­
tification of output and inputs but does not require
detailed specification o f a production function.1
The basic conceptual model is the following: Govern­
ment draws on a series of inputs to undertake a series of
activities which result in one or more outputs intended
to produce a series of desirable consequences. Inputs
consist o f labor, capital, and purchased materials. Ac­
tivities are intermediate services or processes. Outputs
are the final goods or services produced by the govern­
ment. Consequences, which are sometimes known as
outcome and impact, are the intended results of govern­
ment action.2 A basic model of the production process
is portrayed in chart l . 3
In its more sophisticated form, the model includes the
citizen, who is a producer as well as a consumer, and en­
vironmental and community conditions which affect
service production techniques. In this model, consumers
and the environmental setting are necessary parts o f the

production process, although their importance will de­
pend on the service. They are likely to be much more im­
portant in education and policing than in water supply,
although even water supply will be affected by these
considerations. One view of the consummate model is
shown in chart 2.
For some government services, such as sanitation, the
model can be applied in a relatively straightforward
manner. Sanitation organizations use laborers, drivers,
trucks, brooms, gas, and uniforms as inputs. These in­
puts are deployed to produce a series o f activities such
as sweeping streets, emptying litter cans, and picking up
discarded furniture. Outputs, in this case, might be the
trash collected. The consequences should be cleaner
streets and neighborhoods, fewer fire and health
hazards, and, presumably, happier citizens. The more
sophisticated model also includes citizen inputs such as
reporting o f missed collections to government
authorities, separating and preparing trash for recycling
and disposal, carrying trash to the curb for pickup, and
environmental concerns such as the topography,
household density, and amount of precipitation.
For police services, inputs would include patrol
officers, police cars, and communication equipment.
Activities would include recruiting and training police
officers, and taking calls from citizens. Outputs might
include the amount of patrolling and the number of ar­
rests. The consequences of these actions should be a
safer community. The sophisticated model would in­
clude citizen behavior, citizen reports to police, and
even citizen neighborhood watch activities.
For some services, there is general agreement as to
what constitutes an activity, an output, and a conse­
quence, but for many services these are not always ob-

1 Reino T. Hjerppe, “ The Measurement of Real Output of Public
Sector Services,” The Review o f Income and Wealth, June 1980, p.
239.
2 For further exposition of this model, see: D.F. Bradford, R.A.
Malt, and W.E. Oates, “The Rising Cost of Local Public Services:
Some Evidence and Reflections” National Tax Journal, Vol. XXII,
No. 2 (June 1969), pp. 185-202; Jesse Burkhead and Patrick J. Hennigan, “Productivity Analysis: A Search for Definition and Order,”
Public Administration Review, Jan./Feb., 1978, pp. 34-40; and Gor­
don P. Whitaker and others, Basic Issues in Police Performance
(Washington: U.S. National Institute of Justice, 1982), pp. 92-123.
3 Hjerppe, “ Measurement of Real Output,” p. 240.




12

puts—or even to draw the line between a consequence
and an output, or between an output and an ac­
tivity—outputs have to be selected service by service and
organization by organization.
The service and the organizational level will affect the
activity, output, and consequence. The output of one
organization may be an activity o f another. Water meter
repair, for example, would be the output of the water
utility repair shop but not o f the water utility. The out­
put of a catalog unit o f a library would not be the out­
put of the library. This is similar to the intermediate
output issue encountered in private organizations. The
outputs o f an organization’s personnel, data processing,
budget, and communications units are inputs for other
parts of the organization.
This study deals with final organizational or govern­
ment output; that is, service provided to the community
and its citizens. It excludes consequences. Specifically,
we are interested in the rate of change of final govern­
ment output and the inputs (primarily labor) which are
used to produce the output.

vious (table 5). For police and fire services, the intended
consequences and activities are reasonably clearcut but
the outputs are not. For electric power, the activities
and outputs are reasonably clearcut but the intended
consequences are not. Public transit officials, for exam­
ple, argue that their job is to provide service to a com­
munity and its residents. Their output is the operating
transit vehicle; the community’s use of the service is a
consequence. This concept of transit service differs from
the approach o f the private sector transit manager, who
would argue that the capacity provided to a city is an ac­
tivity while use of that capacity is the output.4
Although it may not always be easy to define out­

4
Anthony R. Tomazinis, Productivity, Efficiency, and Quality in
Urban Transportation Systems (Lexington, Mass.: D.C. Heath and
Company, 1975); and Gordon J. Fielding and others, Development o f
Performance Indicators fo r Transit (Irvine: University of California,
1977).

Table 5. Examples of steps in the production of selected government services
Service or function

Output

Activity

Corrections .................................... Clothe inmates
Serve meals
Patrol cell blocks

Consequence

House offenders

Reduce crime
Protect society

Education........................................

Conduct classes
Give tests
Serve meals
Operate school buses

Educate students

Increase literacy rate
Reduce unemployment

F ire.................................................

Maintain fire trucks
Train firefighters

Put out fires
Rescue citizens
Inspect property for fire
hazards

Reduce fire losses
Reduce fire deaths

Food stamps.................................... Conduct interviews
Conduct audits

Issue stamps

Increase nutritional level

Library............................................ Shelve books
Catalog books

Circulate books

Increase literacy rate

Repair streets

Reduce traffic deaths
Reduce travel time

Deliver potable water

Improve community health
Generate revenue to support
government

Street and highway maintenance. ..

Maintain trucks
Dispatch trucks

Water supply.................................. Read meters
Repair water mains




13

Tlh@

off outputs

for example, is heavily subsidized, and the subsidies are
adjusted frequently. Thus, physical measures are
preferable, even when measuring enterprise service pro­
ductivity.

The measurement of outputs is the single most
troublesome problem in computing State and local
government productivity. It is often a problem even
after the output has been identified. Some of the dif­
ficulties are specific to the government activity; others
are more general in nature. This section is concerned
with the latter. Issues examined are the unit of measure;
the extent o f coverage; weighting o f outputs; accounting
for quality change; criteria for selecting outputs; and
availability of output data.

The second issue is the degree of coverage within each
function. Most State and local governments produce
multiple services. Some of these are relatively easy to
identify. Many sanitation departments, for example,
sweep streets, pick up trash, and remove abandoned
cars, a set of easily identifiable services. In other cases,
the multiple services are not so easily identified. Some
electric utilities, for example, conduct energy audits and
provide recreation services as well as electric power.
There are two approaches to the construction of out­
put indexes for multiple-service organizations. One is to
identify each organizational product. For sanitation,
this might be trash pickup (measured by tons removed),
street sweeping (measured by curb miles swept), and
abandoned car removal (measured by the number of
cars removed). A separate index could be calculated for
each product or service; these, in turn, could be combin­
ed into a single sanitation index by using input weights.
The other approach is to identify the dominant out­
put. The index for the dominant or primary output
would represent the entire function. This approach is
valid when secondary outputs are unimportant (at least
when the impact on productivity calculation is
marginal) or when growth in uncovered output would
about parallel the growth in covered output. The tons of
trash removed is probably a reasonably good measure
of sanitation service output, since most cities spend
most of their sanitation resources on household collec­
tion. Street sweeping and abandoned car removal are
normally small consumers of resources and may be safe­
ly ignored.
Whether single or multiple products are used to
measure organizational output depends on whether the
sample product is representative of total output, the im­
portance of the sample product, data availability, and
the importance of multiple products to decisionmakers.
The dominant output approach normally is used when
the primary product accounts for at least 90 percent of
the labor input.
The third issue is accounting for only that part of the
output actually produced during the output cycle (e.g.,
year). This is not likely to be as significant an issue for
State and local government as it is in the private sector
where the production of a single item, such as an office
building or an airplane, may take several years to com­
plete. Most State and local government service outputs
are started and completed in the same year. If the pro­
duct is not completed within the year, an estimate must
be made o f what part of the final output is produced in
each year so that outputs and inputs match.
The fourth issue in service output is including only the
work the organization actually does, since many services

Unit o f measure. The basic output measure(s) of an
organization should be a homogeneous physical unit.
Furthermore, whenever possible, the measure should be
related to unit labor hours spent in its production.
Because of the problems in defining and measuring
government output, a series of outputs should be ex­
amined and tested.5
Street cleaning illustrates some of the issues. Two
commonly used measures of street cleaning output are
cubic yards collected and curb miles swept. The curb
miles swept will be about proportional to labor re­
quirements needed for sweeping. Also, the quality of
service—cleanliness—should be related to the output
and input. On the other hand, cubic yards collected will
not be as closely related to labor inputs. In fact, the in­
verse is likely: As streets are swept more frequently, the
cubic yards increase but at a much slower rate than
labor inputs. The result is decreasing productivity. Curb
miles is the preferred measure of output in this case.
In the private sector, when physical data are not
readily available, value data are often used as the
measure of output. In such cases, price changes must be
removed from the value data to obtain an index of real
output. Removing the price change facilitates calcula­
tion of output where the industry produces and sells a
number of products or where the industry lacks a
discrete, tangible product.
In the public sector, the primary problem with using
price-adjusted value as the measure of output is that, in
most cases, market prices are lacking. Without direct
pricing, estimating output in real terms is impossible.6
Exceptions may be “ enterprise” services, such as water
and electric power, which are sold in the marketplace
much as services are sold in the private sector. They ac­
count for about 6 percent of total State and local
government employment. But even for the enterprise
services, value is not always a good measure of output
since prices are administratively determined and many
have little relation to the costs of production. Transit,
5 Jerome A. Mark, “ Industry Indexes of Output Per Man-Hour”
Monthly Labor Review, November 1962, pp. 1269-73.
6 John P. Ross and Jesse Burkhead, Productivity in the Local
Government Sector (Lexington, Mass.: Lexington Books, 1974), p.
35.



14

There is considerable debate as to what changes in
quality mean and how they should be handled
analytically in index construction.9 Of the two general
approaches, one is based on consumption, the other on
production. For consumption, a quality change is
reflected in a change in consumer utility; for produc­
tion, a quality change is reflected in a change in resource
requirements. This study is concerned with the latter.
It is helpful to divide production quality changes into
those which should be reflected in the output indexes
and those which should'not. The latter type includes
those arising from events external to the production
process. For example, waiting time by clients might in­
crease because of an influx of new clients, or mass tran­
sit commuting times might decrease because of the
opening of a new road. Adjustments are not ap­
propriate in such cases.
When production and unit labor requirements
change, adjustments need to be made in the output in­
dex. An example would be a shift from backdoor to
curbside collection of solid waste, which requires
citizens to carry their trash to the street, a task formerly
done by government collectors. In fact, the government
has introduced a new service or changed the level of serv­
ice. The production process has been modified and unit
labor requirements have shifted as a result of this
change in service.
Some classify changes in quantity or volume of pro­
duction as quality shifts. From the standpoint of
citizens, adding branch libraries or recreation centers
may improve quality since citizens will not have to travel
so far to a facility. From a production standpoint, they
are simply an increase in the quantity or volume of pro­
duction.
One way to adjust for a quality shift is to adjust the
index. In the example of the shifting from backdoor to
curbside collection, the input index would be adjusted
to include the work of the citizen (labor hours) in
transporting the trash from the back door to the curb.
The output index would remain the same.
The second approach to a quality shift is to identify
the new service and create a new index. In this case, the
new service would be curbside collection; the old serv­
ice, backdoor collection. The two productivity indexes
would be linked to create a single index.
The impact of a change in quality on productivity
measurement can vary depending on the output measure
chosen. For street cleaning, the more frequent the clean­
ing, the cleaner the streets. If the output measure is curb
miles swept, unit labor requirements will remain about
constant, productivity will remain about constant, and
the quality of service and output will be about propor­
tional. However, if the measure is cubic yards collected,
more frequent cleaning will result in a decrease in cubic
yards collected with each additional cleaning. In other
words, unit labor requirements will increase, productivi­

are purchased from other governments and private con­
tractors. Government, like private industry, is con­
fronted with “ make or buy” decisions. Many com­
munities have contracts with private firms for
household trash collection, day care, mental health,
custodial services, and data processing. Governmentproduced outputs must be separated from contracted
outputs. In most cases, the separation is relatively
straightforward if data are available. Where one service
is provided partly by the government and partly by a
private contractor, separation of output is more dif­
ficult. This mode of operation is becoming increasingly
popular. Even more troublesome is the contracting for
intermediate services such as custodial or data process­
ing. A problem arises when inputs reflect the shift from
public to private or vice versa, but outputs do not. Iden­
tifying such shifts is important for calculating trends.
Output weights. Calculation of a multiple-service out­
put index or a single service with multiple outputs will
require aggregation of individual measurements and in­
dexes. In combining indexes, it is generally preferable to
use weights—labor weights for labor productivity in­
dexes. However, labor weights are not always available,
and surrogates are often used. Several different ap­
proaches have been used in private sector meas­
urement.7
For construction of most State and local government
output indexes, labor weights are not only preferable,
they are necessary. Except for government enterprises,
most government services do not have unit value
statistics.
Quality o f service. Quality change is a major concern in
developing output and productivity indexes. It is par­
ticularly important in measuring government output.
The justification often given for increases in govern­
ment expenditures is that the quality of service has im­
proved—streets are kept cleaner, snow is removed
faster, and police respond more rapidly to calls for
assistance. Conversely, some feel that productivity gains
are made at the expense of quality. One study of New
York City concluded that output (quantity) had increas­
ed but performance (quality) had deteriorated.8 In the
police department, 36 of 37 measures of output quantity
examined increased, but most measures of quality, such
as the proportion of crimes solved, decreased.
7 National Research Council, Measurement and Intepretation o f
Productivity (Washington: National Academy of Sciences, 1979), pp.
68-70.
8 David Greytak, Donald Phares, and Elaine Morely, Municipal
Output and Performance in New York City (Lexington, Mass.: Lex­
ington Books, 1976).
9 See, for example, Frankin M. Fisher and Karl Shell, The
Economic Theory o f Price Indices (New York: Academic Press,
1972), and Jack E. Triplett, “ Robert Gordon’s Approach to Price
Measurement,” BLS Working Paper 101 (Bureau of Labor Statistics,
April 1980).



15

ty will decrease, and quality of service and output will
not be proportional.
Clearly, quality should be examined function by func­
tion. For some functions, the issues and variables,,if not
the solutions, are straightforward, while for others they
are complex and certainly not obvious.
Unfortunately, identifying the crucial quality con­
siderations in State and local government services is not
always easy. Although some research and discussion of
quality and its measurement have taken place over the
past decade, little research has been done on the ab­
solute or relative impact of quality change on produc­
tivity costs and unit labor requirements. Lacking
systematic research, the process has to be ad hoc.
The suggested approach for handling quality in State
and local government productivity measurement is as
follows:

of services or products. The level and quality of serv­
ice can change, since they can be adjusted, but the
basic service must be repetitive.
4. Output data must be accurate and com­
parable. Much output data currently collected, at
least at the national level, is incomplete, inaccurate,
and inconsistent from period to period. Construc­
tion of a viable output index requires accurate,
comparable data. Comparability is more important
than absolute accuracy in preparing a time series.
5. Output calculations should use existing data
and data collection procedures. Two issues are in­
volved here—whether the records exist in State and
local government, and whether a procedure current­
ly exists to collect national data. In either case, ex­
isting data and data collection procedures should be
used whenever possible, as new procedures will like­
ly be costly and time consuming.

1. Identify service output.

6. Outputs should be easily understood. An in­
dex which is simple and easily understood is most
likely to be accepted, supported, and used. Esoteric
measures and complex quality adjustments should
be avoided.

2. List quality considerations for the output
measure.
3. Assess each quality factor for its potential im­
pact on unit labor requirements.
4. Create a quality index time series if the impact
is potentially important.

7. Outputs should be physical measures. The
lack of a market price for most government services
precludes use of a deflation procedure when
physical quantity data are difficult or impossible to
obtain. Even for services that have a market price,
such as government enterprises, physical output
measures are preferable because prices are often
subsidized and set by administrative decree.

5. Track the quality index through time.
6. Adjust the input index or link a new produc­
tivity index with the old index if the quality index
changes.
Criteria fo r selecting outputs. Eight criteria are
presented below for selecting measures of State and
local government output.1 The first four are essential;
0
the last four are desirable.

§. Output units should reflect the labor units
spent in their production. Since unit labor weights
are used in constructing individual service indexes
and functional groupings, the output measure
should reflect base-year unit labor requirements.

1. Outputs must reflect the final product (serv­
ice) o f the organization. To determine productivity,
the output must be the product or service leaving
the organization, not the intermediate products.
Output must reflect the work rather than the conse­
quences of the work.

Availability o f data. There is no single source of na­
tional data on State and local government outputs; data
have to be collected function by function. National data
are available for a few functions, such as electric power
and unemployment insurance.
More data are available at the State and local level.
Many governments routinely prepare statistical tabula­
tions and performance reports from which output in­
dexes can be constructed.

2. Outputs must be measurable. Absolute (car­
dinal) numbers are required. Arguments that
government services cannot be measured usually
fail to distinguish among the measurement of in­
termediate products, final outputs, and conse­
quences of government service. Whether or not a
service can be measured has to be considered func­
tion by function.
3. Outputs must be repetitive. Construction of
an output index requires a repetitive or recurring set
1
0
For a slightly different list, see Brian Usilaner and Edwin Soniat,
“ Productivity Measurement,” in George Washnis, ed., Productivity
Improvement Handbook (New York: John Wiley, 1981), p. 95.



16

Th@ measurement ©f Snpyts
This section discusses the number and type of factor
inputs used to measure productivity, presents the labor
measures most often used, reviews some of the thorny
questions surrounding these measures, presents criteria
which should be used to select inputs, and discusses data
currently collected that might be used to calculate na­
tional labor indexes for State and local government.

Table 6. State and local government expenditures for
salaries and wages as a percent of total expenditures by
function, fiscal year 1977

Number and type o f inputs. Productivity measures are
often characterized by the number and type of inputs. A
common categorization is in terms of single factor or
multifactor productivity.1 In reality, there is a con­
1
tinuum of inputs or factors: The number and type of
resource inputs included should reflect the use to which
the measure is put.
A single factor productivity measure, the most com­
mon type, relates one resource, most often labor, to
output. However, it does not measure the specific con­
tribution of the factor to output. Rather, it expresses the
joint effect of interrelated influences, such as manage­
ment, technology, and regulation, as well as the factor
input, on overall output.
Multifactor productivity relates two or more inputs to
output and also reflects the joint effect of many in­
fluences. However, it eliminates the impact of the
substitution of one factor for another on overall pro­
duction. Examples of multifactor inputs are labor and
capital; labor, capital, and energy; or labor, capital,
energy, and materials.
This study focuses on* single factor productivity,
specifically labor productivity, for several reasons.
First, labor is of primary importance in public policy
issues. Salaries and wages constitute about 40 percent of
all State and local government expenditures and 55 per­
cent of all current expenditures. Its importance varies by
type of service (tab;e 6). Salaries and wages, in fiscal
1977, constituted 75 percent of the expenditures for
police programs but only 9 percent for electric power.
Fringe benefits would raise these proportions. Second,
labor is relatively easy to calculate when compared to
other factors of production. Third, labor data have
been collected for many years and generally are the most
accessible of State and local government factor inputs.
Fourth, labor indexes are calculated for many parts of
the private sector, for some parts of the Federal Govern­
ment, and for some foreign countries. State and local
government labor-based indexes would permit com­
parisons with these other sectors.
Three ways of defining and measuring labor input for
State and local government productivity calculations
are: Number of hours, number of full-time-equivalent
employees, and number of persons. Hours can be fur­
ther divided among hours paid, hours at work, and
hours actually producing output.

Function
All functions

39
75
74
67

Police............................
Fire ..............................
General co n tro l............
Social insurance..........
Financial administration
Education ....................
Corrections..................
Hospitals......................
Libraries ......................
Sanitation ....................

64
61
60
58
52
52

Natural resources....................
T ransit......................................
H e a lth ......................................
Parks and recreation................
Water transport........................
Highways..................................
A irp o rts ....................................
Housing and urban development
Water supply............................
Sewerage ................................

46
43
43
41
28
24
24
24
22
15

Gas ..............
W e lfa re ........
Liquor sales . .
Electric power

12
10

66

9
9

Source: Computed from data taken from 1977 Census of Govern­
ments—Compendium of Government Finances (Bureau of the Census,
1979), pp. 29 and 33.

Number o f hours. The preferred measure is the number
of labor hours. Labor input hours should simply be the
total hours applied during the period for which outputs
are measured.
Ideally, the labor hour measure would be the actual
time worked to produce one or more outputs. The usual
practice, however, because of definitional and data pro1 Multifactor productivity has also been referred to as total factor
1
productivity.



Percent

17

blems, is to use hours paid. That is, labor input normal­
ly includes actual hours worked plus time on the job
which is paid for but not worked, plus paid time off the
job. Hours on the job but not worked might include
coffee breaks, training, and washup time. Hours off the
job but paid might include vacation, holidays, and sick
leave. Calculations using hours paid tend to understate
true productivity. If the difference between hours paid
and hours worked is increasing, labor productivity
trends are understated.
The converse—hours worked but not paid—also
needs to be considered. Many people, including
managers, teachers, and coaches, work hours for which
they are not paid. These hours should be counted too,
for if they are increasing (or decreasing), productivity
trends will be overstated (or understated) if not includ­
ed.
Many practical problems arise in measuring the hours
of State and local government workers. Should standby
hours of police, fire, and public works officials who are
home but subject to call be ihcluded? Sometimes
employees are paid for standby time but more often
they are not. What about employees who are paid by the
task, such as collecting the trash on a specified route?
When they finish the task they are permitted to go
home. Irrespective of the actual hours worked, the
employees are paid for a fixed, previously agreed upon
number of hours. Should the hours worked or hours
paid be included? Should extracurricular activities and

Table 7. State and local government activities included in
three labor hour measures
Activity

Hours
paid

Hours
at work

Hours
worked

Paid vacation ........................
Paid holidays..........................
Paid sick leave ......................
Jury d u ty ................................
Military leave..........................
S tandby'................................
Meal hours2............................
Washuptime..........................
Rest and coffee breaks..........
Union business......................
Training..................................
Production ............................

X
X
X
X
X
X
X
X
X
X
X
X

_
-

the number of hours worked per person decreases. The
greatest divergence between an hours index and an
employee index will probably occur when part-time
employment increases or decreases.

-

X
X
X
X
X
X

Most State and local governments use part-time
employees extensively. In October 1980, they employed
10.3 million full-time and about 3.0 million part-time
workers. Part-time employment made up 23 percent of
the total. Part-time employment varies substantially by
service. In 1980, it was only 3 percent for transit, 5 per­
cent for corrections, and 6 percent for sewerage, but 42
percent for local libraries and 49 percent for higher
education.
Seasonal employment such as for snow removal, leaf
pickup, park maintenance, and swimming pool opera­
tion often can create measurement problems when
calculating an index of the number of employees. The
primary problem is the period of coverage. Employee
counts are commonly published for one date such as
December 31 each year, but an employee count on July
30 or October 30 may be quite different because of
seasonal variations. To overcome the problem of
seasonal employment, the preferred approach would be
to use a weekly or monthly average of the number of
employees to calculate the index.

-

-

-

X

1Some governments pay for standby time, such as to water supply and
electric power employees.
2Some workers, including police officers and firefighters, receive pay for
meal times.

citizen meetings be counted? Teachers often supervise
activities outside school and attend meetings for which
they may not be paid. How should these hours be
counted? Most of these issues relate to a specific govern­
ment function or service and should be addressed in that
context.
The relationships among State and local government
activities and hours paid, hours at work, and hours
worked are shown in table 7.

Comparison o f the three approaches. Whether the three
approaches to labor measurement—hours, years, and
employees—would produce markedly different labor
trends in the public sector is not known. In the private
sector, labor trends for hours and employees differ, but
only slightly over the long run.
Several government occupations, especially fire­
fighting, have moved toward a shorter workweek over
the past decade, and at least one State government, New
Hampshire, has shortened the workweek of its
employees. An employee-hour index would reflect such
changes, but indexes of employee years or number of
employees would not. Whether the shift in the hours
worked is sufficiently large to affect a national produc­
tivity index is not known.
The three indexes did not vary greatly for the services
examined in this study. Data problems caused more
variation than the type of index. However, in some ser­
vices, such as fire protection, the type of labor index
may substantially affect labor trends and productivity
calculations.

Employee years. Government labor indexes often reflect
the number of employee years or the number of fulltime-equivalent employees, since the number of
employee hours is rarely available. An employee year
commonly equals 2,080 hours (40 hours per week times
52 weeks per year) and includes all paid time including
overtime, vacation, holidays, and sick leave. Part-time
employment is usually computed on a full-timeequivalent base, such as two half-time employees equal
one full-time equivalent. Seasonal employment is also
usually computed this way; for example, four summer
employees equal one full-time equivalent. Overtime can
be handled in the same manner but is more often simply
ignored. A full-time-equivalent index could produce a
rate of change exactly the same as an hours index,
depending on the computational rules. In most in­
stances it approximates the hours index.

Number o f employees. An index of the number of
employees is sometimes computed when data are not
readily available to compute an hours or a full-timeequivalent index. This type of index simply counts the
number o f employees who produced the output without
concern for the time each employee worked. It approx­
imates the actual hours expended to produce the out­
puts. An index o f the number of employees will
understate the change in labor input when the number
of hours worked per person increases, such as in over­
time, and will overstate the change in labor input when



Volunteers. Volunteers are used by most governments,
although the extent of use varies considerably.1
2
Volunteers are common in services such as fire, educa­
tion, hospitals, museums, and recreation but are
1
2
Martha A. Shulmann, “Alternative Approaches for Delivering
Public Services,” Urban Data Service Reports, Vol. 14, No. 10
(Washington: International City Management Association, October
1982), pp. 8-9.

18

relatively rare in others such as public utilities and tran­
sit (table 8).
Conceptually, a labor index should include volunteer
participation. Volunteers contribute to output just as do
paid employees. The treatment of volunteer labor
should depend on the projected uses of the productivity
data. Some uses would require that volunteers be
separately identified.
However, the way volunteer labor is accounted for
will, in most cases, be moot. Probably in no more than
four or five State and local government service areas are
volunteers even potentially important. If the ratio be­
tween paid and volunteer labor remains constant in these
areas, their inclusion or exclusion will not affect labor
trends. Further, in the real world of data collection and
measurement, identifying volunteer labor input or out­
put is extremely difficult. Records on volunteers are
almost nonexistent in State and local government. Pro­
ductivity calculations usually include the output
generated by volunteers but not their inputs. However,
the decision will need to be made service by service.

Supposedly, more education will increase police output.
However, an increase in output from additional educa­
tion is not an increase in productivity but an improve­
ment in labor input—i.e., a shift in the composition of
the labor input.
The method generally used to adjust for changes in
labor force composition is pay differentiation. This re­
quires information on the change in pay by occupation
through time, data which are not readily available from
State and local governments. Although often discussed
in private sector productivity measurement, changes in
labor force composition are of secondary concern at this
time for State and local government and are not con­
sidered further.
Criteria fo r selecting inputs. Five criteria are suggested
for identifying input data, specifically labor inputs.
Criteria which are essential for productivity measure­
ment are presented first, followed by those which are
desirable.
1. Inputs must match output. Calculation of pro­
ductivity requires that the resources applied match
the measured organizational output. For organiza­
tions with multiple outputs, like the typical city
government, this will require careful identification
of resources used to produce the outputs.

Changes in work force composition. Labor is often
treated as a homogeneous input although clearly it is
not. Depending on the mix, labor inputs can produce
very different levels of output. If the mix changes, the
level of output can be affected. For example, police
departments increasingly require new recruits to have
some college education. The rationale behind the re­
quirement is the creation of a police force which can
better deal with the public and today’s complex society.

2. Inputs must be measurable. Absolute numbers
are required.
3. Input data must be accurate and comparable.
Much of the labor data collected on State and local
government operations is inaccurate and inconsis­
tent from period to period. Comparability is more
important than absolute accuracy. Data checks and
analysis must be part of the construction of any in­
dex.

Table 8. Estimated use of volunteers by State and local
government function
Use of volunteers

Function
Fire protection
Parks and recreation

1
|

More than 10 percent of
total employed
Less than 10 percent but
more than 1 percent

Education
Libraries

4. Input calculations should use existing data. New
data collection procedures will likely be time con­
suming and costly to develop and maintain, and
burdensome for those providing the data. Existing
data and collection procedures should be used
whenever possible.

Airports
Corrections
Electric power
Financial administration
Gas
General control
Health
Highways
Hospitals
Housing and urban development
Liquor stores
Natural resources
Police
Sanitation
Sewerage
Social insurance
Transit
Water supply
Water transportation
Welfare

*

5. Inputs should be easily understood. General ac­
ceptance, support, and use of an index are more
likely if the construction is straightforward and
easily understood. This is one reason that labor in­
dexes are widely used.

Less than • percent

Availability and accuracy o f labor statistics. Every State
and local government collects labor statistics for use in
its day-to-day operations. Two types of labor measures,
the number of full-time-equivalent employees and the
number of employees, are most common; less common
is the number of hours. Most State and local govern­

/

Source : Based on discussions with State and local government

experts.




19

of government. Third, the employment counts for State
and local government as taken from the c p s are
markedly different from those obtained from other
sources. In short, the CPS has too many major limita­
tions for productivity calculations to be considered fur­
ther in this study.

ments should be able to prepare labor indexes by func­
tion and for the government as a whole.
Preparation of national or regional labor indexes is
not as straightforward. Some labor data are collected
and published by function by trade associations, public
interest groups, and Federal agencies. The International
City Management Association, the American Public
Works Association, the American Water Works
Association, and others routinely collect statistics on
public employment for specific functions and
sometimes for government as a whole. Federal agencies
such as the Department o f Labor, the Department of
Justice, and the Department of Health and Human Serv­
ices often collect data on the number of State and local
government employees for the programs they fund and
coordinate.
In addition, there are four sources of national State
and local government employment statistics: The Cen­
sus of Governments; the Current Population Survey
( c p s ); the Current Employment Statistics survey ( c e s
790); and unemployment insurance reports ( e s 202).
The characteristics of these surveys are discussed below
and summarized in table 9.

Current Employment Statistics survey ( c e s 790). The
CES survey is used to collect data monthly from
establishments in nonagricultural industries and govern­
ment on the number of employees, average hours work­
ed, and average hourly and weekly earnings. About half
o f all State and local government employment is
covered. Employment statistics are broken down into
eight functional areas for State government and into
seven areas for local government. One advantage o f the
CES 790 data for productivity measurement is its
timeliness; preliminary data are published about 3
weeks after reporting. Another advantage is that they
include hours worked. However, the CES survey has
several deficiencies. First, statistics on hours are col­
lected for nonsupervisory employees only. Second,
statistics are not available for many government func­
tions. Third, the sample size does not permit presenta­
tion of data below the national level. Fourth, and most
important, coding of the data by government function is
very poor.

Census o f Governments. The best known and probably
the most widely used national statistics on State and
local government employment are produced by the U.S.
Bureau o f the Census. Sample data are collected and
published annually. Every 5 years (years ending in 2 and
7), the Census Bureau takes a complete enumeration
and publishes the results. Statistics are collected and
published on the number of employees (full time and
part time), number of full-time-equivalent employees by
major function, and salaries and wages.
Several problems arise in using these employment
statistics to calculate government productivity. First,
the statistics are for a single month, October, o f each
year. Second, the functional classification system used
by Census is very broad—e.g., police, fire, employment
security—probably too broad for productivity calcula­
tions. Third, the information is not available until 6 to 9
months after the reference date. Fourth, no information
is collected on hours, whether hours paid, hours work­
ed, or hours at work.

Unemployment insurance reports (e s 202). Since
January 1978, all State and local government employees
have been covered by unemployment insurance. As a
result, State and local governments record monthly
employment and wages and report the data quarterly to
the U.S. Unemployment Insurance Service. Since the e s
202 is linked to financial reports, it should provide the
most accurate statistics available on the number of per­
sons employed by State and local government, by State,
county, and Standard Metropolitan Statistical Area.
Although the e s 202 reports are comprehensive, they
lack detail. The primary problem insofar as productivity
calculations are concerned is the inadequate division by
function—most employees are assigned to the general
government category. Furthermore, no information is
collected on hours and type o f worker. The attraction of
the e s 202 report lies not in its current form but in its
potential if the coding by function were improved.

Current Population Survey ( c p s ) . Data collected by the
Current Population Survey are used primarily to
calculate the monthly employment and unemployment
statistics. A number of other statistics, such as hours
worked and pay, are collected. The two strengths o f the
survey for productivity measurement are its timeliness
and the information on hours worked. However, the
CPS has a number of problems for State and local
government productivity measurement. First, it con­
tains no information on services or functions. Second, it
is impossible to separate the employment data by type




Data accuracy. Data collected by the Census o f Govern­
ments and the unemployment insurance reports were
compared with data obtained through personal contacts
with individual governments for three functions—elec­
tric utilities, drinking water, and State liquor stores.
Considerable variability was evident. The variance
resulted from differences in categorizing functions, in
assignment of work activities to functions, and from
clerical mistakes.
20

Table 9. Four national surveys used to collect State and local government employment data

Census of
Governments

Characteristic
Initial y ea r..........................

Current
Population Survey

Current Employment
Statistics survey

Unemployment
insurance reports

Prior to 1967

Prior to 1967

Prior to 1967

1978 for all government

About 30 major
functions

None

Local—8 functions
State—7 functions

None nationally;
Standard Industrial
Classification for a
few States

States, Standard Metro­
politan Statistical
Areas, cities,
towns, school
districts

National

National

States, Standard
Metropolitan Statistical
Areas, counties;
no city data

Definition of employee. . . .

All paid persons

All employed persons
16 years or older,
paid or unpaid

All paid persons

All paid persons

Frequency of survey..........

Annual—October of
each year

Monthly

Monthly

Monthly—reported
quarterly

Reference period................

Payroll period con­
taining October 15
of each year

Survey week containing
12th of each month

Payroll period contain­
ing 12th of each
month

Payroll period containing
12th of each month

3 weeks

3 weeks (for prelimi­
nary data)

About 9 months

Coverage by government function ................

Coverage by government u n it........................

Timeliness of publication . . About 9 months
Sample size..........................

All governments every
5th year—sample
other years

About 60,000 households About 12,000 govern­
ments

All governments

Nonresponse......................

Unknown

3-5 percent

About 9 percent

None

Benchmark..........................

Quinquennial Census
for sample survey

Decennial Census

Quinquennial Census
of Governments

None

Data collected....................

Number of full- and
part-time employees
Monthly earnings in
October

Employment and hours
at work
Annual earnings in
March
Weekly earnings in
May

Employment and hours
for nonsupervisory
employees, full and
part time

Total employment
Total wages

Method of collection..........

Mail survey (some
interviews)

Household interviews

Mail survey shuttle
form

Byproduct of admini­
strative reporting

Information on employee
quality..............................

No

No

No

No

Information on volunteers..................................

No

No

No

No

Suvervisory vs. nonsupervisory employees . . No

No

Yes

No

Information on parttime employees ..............

No

Yes

No

Yes

Productivity comparisons—levels and trends. Underly­

Three conclusions are to be drawn from this examina­
tion: First, no single data source is likely to be entirely
acceptable. Second, major errors are likely in each data
series. Third, viable labor-based input indexes will re­
quire detailed data comparison and adjustment irresspective o f which data set is used.

ing all productivity measurement is comparison—com­
parison through time, comparison of producing units,
or comparison of producing units through time. Most
private sector productivity measurements are time com­
parisons, such as, “ Productivity increased by x percent
between 1967 and 1982.” Trends are routinely com­
puted for industries, for groups o f industries, for in­
dividual countries, and for groups of countries.
Similarly, State and local government productivity
trends might be calculated and stated as, “ Municipal
electric power productivity increased by x percent bet­
ween 1972 and 1978.” Productivity trends could also be

Other Issues
This section discusses productivity comparisons, fre­
quency of measurement, geographic coverage, period
coverage, service definitions, and the productivity index
itself, important issues which were only briefly touched
on in the preceding review.



21

When national data are not readily available,
representative data may be collected by sampling. The
sample would have to be balanced for size, geographic
distribution, and any other factors which might affect
productivity. Consistency through time, among
jurisdictions, and between inputs and outputs is crucial
in such cases.
The size and location o f a jurisdiction can affect pro­
ductivity. Some government services, such as water,
sewerage, electric power, and refuse collection, benefit
from economies o f scale.1 Other services, such as police
3
and recreation, evidently do not. Garbage collection
and street repair are affected by topography and
climatic conditions.
Productivity measurements are much more useful to
national, State, and local government decisionmakers if
they are available by location. For some services, such
as solid waste collection, no national output data are
readily available; for others, such as electric power, only
national data are available, and for still others, such as
unemployment insurance, both State and national data
are readily available.

computed, for example, for sanitation services in New
York City between 1967 and 1978, or for the Unemploy­
ment Insurance Service in the Southern States between
1963 and 1978.
Some productivity measures focus on absolute levels:
“ Each employee produces on the average x tons o f steel
or y cars.” Local government level measures might be,
“ w tons of trash collected” or “ x miles of street swept
per employee.” Comparisons of levels could be made
between jurisdictions or regions, or with the Nation as a
whole.
Trends and levels are complementary. A city service
might have a low level o f productivity but a high rate of
change, or vice versa. A true picture of productivity re­
quires examination of both levels and trends. However,
the data and analyses required to compute trends are
much less demanding than those required to calculate
absolute levels. This study focuses on productivity
trends.

Frequency o f measurement. Productivity trends are
normally calculated annually, although some national
estimates are produced quarterly. In view o f the cost
and the doubtful benefit of more frequent measure­
ment, State and local governments should probably
focus on annual measurement, at least initially.
Benefits which would accrue from more frequent
measurement depend primarily on the decisions to be
made. Until these decisions can be specified, there is lit­
tle point in developing other than annual measures. Fur­
therm ore, m onthly or quarterly productivity
movements might not even be detectable for most State
and local government services.
In addition, seasonal adjustments may have to be
made if quarterly or monthly calculations are to be
useful. This, of course, requires knowledge about
seasonal fluctuations, which is generally lacking.
Finally, some State and local government services,
such as education, have outputs that require more than
one month or one quarter to produce and, thus, to
measure.

Time period coverage. Two types o f time periods need
to be considered in formulating a productivity index: (1)
Number of years to be covered by the index, and (2)
whether the calendar or fiscal year is used.
The number o f years covered by a productivity index,
and the beginning and ending years, can have a marked
effect on the overall rate of change. Generally, the
longer the time span, the less important the beginning
and ending years. Also, a longer time span would nor­
mally be more representative o f long-term trends.
Cyclical fluctuations can affect long-term productivi­
ty trends. Such fluctuations occur when inputs do not
change as rapidly as outputs. Unemployment Insurance
Service (uis) outputs, for example, parallel the
unemployment cycle, and inputs usually lag behind
changes in outputs, as discussed later in this study. The
result is a productivity index which will shift significant­
ly depending on the years included in the index. To
avoid arbitrary cutoff dates, and to reflect long-term
trends more accurately, average annual growth rates
might be calculated from peak to peak, trough to
trough, or midpoint to midpoint.
Sudden shifts in the economy can also influence
calculations of long-term rates of productivity change.
State and local government electric power productivity
dipped markedly with the increase in energy prices in
1973 just as it did in the private sector.
The second time-related issue is whether the calendar
or fiscal year should be used. Most productivity indexes

Geographic coverage. State and local government as
defined for this study includes the 50 States, the District
o f Columbia, and all cities, counties, special districts,
townships, and school districts. There are, in total,
about 80,000 State and local governments.
Data sources are not consistent in their geographic
coverage. Some sources include trust territories as well
as the 50 States, others include only the larger jurisdic­
tions. Whether trust territories are included is probably
not important since they are such a small part of the
total. Focusing on only the larger jurisdictions,
however, will introduce bias into productivity calcula­
tions if economies of scale are present. Large municipal
electric power utilities, for example, are growing
somewhat faster than smaller utilities.



1
3 William F. Fox, Size Economies in Local Government Serv­
ices: A Review (U.S. Department of Agriculture, Economics,
Statistics, and Cooperative Service, 1980).

22

categories are not widely used in State and local govern­
ment data collection and analysis.

are based on the calendar year. The Federal Govern­
ment’s productivity measurement system, however, uses
the Federal Government’s fiscal year o f October
1-September 30.

Both the sic and Census of Governments classifica­
tion systems should be helpful in structuring productivi­
ty analysis, collecting data, and making comparisons.
However, for some services, neither classification
scheme will be satisfactory, and further specification
will be necessary. Appendix B presents an initial attempt
at a cross-classification o f these two systems.

The question is more complicated for State and local
government. For a single government, or a group of
governments with the same fiscal year, there is no pro­
blem. The Bureau of the Census, for example, asks all
State and local governments to use the July 1-June 30
fiscal year in reporting financial data.
However, State and local government fiscal years
vary. The U.S. Department of Transportation collects
data from over 300 transit systems with fiscal years end­
ing on March 31, April 30, June 30, September 30, and
December 31. As discussed in a later chapter, electric
power data are reported by calendar year; Unemploy­
ment Insurance Service statistics by the Federal fiscal
year.
The closing month of the “ productivity” year is not
important for trend determinations. The same month
should be used each year, and the inputs and outputs
should cover the same period.

The productivity index. This study aims to develop a
procedure for developing two basic types of State and
local government indexes—one, a series o f productivity
indexes for individual government services such as fire,
police, water supply, personnel, and purchasing; and
two, an index of total State and local government pro­
ductivity.
To compute the total State and local government pro­
ductivity index, output indexes for individual functions
must be developed. After these outputs have been
developed, a total national aggregate State and local
government output index may be computed. The total
State and local government productivity index would be
the sum of the final outputs weighted by their inputs.
The index would not include intermediate outputs.
However, all inputs would be included, intermediate as
well as final.

Service specifications and definitions. Definitions of
services unfortunately vary among governments and
through time. Public works, for example, may be
specified as a single unit; may be broken into major
components such as sanitation, water supply, and street
maintenance; or may be divided into subservices such as
solid waste residential collection, street sweeping, street
flushing, and so forth.

The mathematics of the productivity index is simple
and straightforward.1 The index should compare the
4
quantity of service in the current year with the base
year, each weighted by the employee hours expended
per unit produced per unit of labor input in both
periods.

For single measurements or studies, definitions can
usually be adjusted to meet analytic needs and data
availability. For preparation of a national, aggregate
productivity index, a formal classification system is
needed.
Most of the summary data in this study are based on
the classification system and definitions o f the Census
of Governments (see appendix B). This structure has
been used for years by the Bureau of the Census, State
and local governments, and the research community.
However, even this classification system has several
deficiencies for productivity measurement. First, the
service categories are too broad. Second, governments
differ in the manner in which they structure, and thus
report, their operations. The functions assigned to
police departments or even State alcoholic beverage
control agencies vary from jurisdiction to jurisdiction.
The Standard Industrial Classification (sic) system,
in contrast to the Census of Government system, is very
detailed. It includes all goods and services produced by
private and public establishments. State and local
government is a small part of the sic (appendix B lists
State and local government services included), sic




If State and local government produced a single out­
put, the index would simply be the reciprocal of
employee hours spent to produce a unit of output over
the two periods of time or:

1

Lj

Where
Iu = unit employee hour index
Ir

= output per employee hour index

Lj = unit employee hours in current period
Lo = unit employee hours in the base period.

1
4
B ls Handbook o f Methods, Bulletin 2134-1 (Bureau of Labor
Statistics, 1982), p. 102.

23

The same ratio would be used for a single service such as
sewerage or water.
Where more than one service is produced in a func­
tional area, such as sanitation, or where several func­
tional areas are combined, a composite index is re­
quired:

Where
Iu = unit employee hour index
Q0 = quantity in the base period
Lj = labor hours expended in current period
L0 = labor hours expended in the base period.

Q L1

This particular form, the base-period composite, is used
in the following chapters.

I, = ____




24

Chapter 0¥= Msasurimg tlh® Usasuiralbl®"
Thir®© Case Studies

owned utility accounts for about three-fourths of all
production and sales.1 The 250 investor-owned utilities
sell about 78 percent o f the kilowatt hours, serve about
76 percent of the customers, and own about 73 percent
of the Nation’s electric plant and equipment (table 10).
The second type consists of Rural Electrification Ad­
ministration ( r e a ) cooperatives, which expanded
dramatically into the rural areas in the 1930’s under the
sponsorship of the Federal Government. Today, about
900 cooperative systems sell about 7 percent of the Na­
tion’s electricity, serve about 10 percent of the
customers, and own about 7 percent of utility plant and
equipment.
The third type, the government-owned utility, sells
about 16 percent of all kilowatt hours, serves about 14
percent of the customers, and owns about 20 percent of
the plant and equipment. The two basic types of
government-owned utilities are Federal and State/local.
The Federal Government is primarily a generator and
wholesaler of electric power. It produces about 11 per­
cent of the Nation’s power and owns about 8 percent of
the plant and equipment, but serves less than 1 percent
o f final users.
State and local electric power systems include all
government systems other than those operated by the
Federal Government—State, special district, city, and
county operations. The 2,200 State and local systems,
also called municipal systems, account for about 14 per­
cent of the Nation’s customers, 12 percent o f kilowatt
sales, and about 12 percent of electric plant and equip­
ment investment.
In 1977, State and local systems sold about one-third
o f their kilowatts to residential users, one-third to industrial/commercial users, and one-third to other users
such as other public power authorities, railroads, and
highway and street lighting authorities. Generating
capacity was divided among steam (52 percent), nuclear
(8 percent), hydroelectric (33 percent), and internal
combustion engine (7 percent).2
The primary factor input into State and local power
operations is capital, which accounted for about 43 per-

This chapter presents illustrative national productivi­
ty measures for three State and local government serv­
ices—electric power, alcoholic beverage sales, and
unemployment insurance. These three services were
selected for examination because each has a reasonably
well-defined set of outputs, and output and employment
data are available for calculating a productivity index.
Each service is discussed briefly, a synoposis is
presented of past research, potential output measures
and data to calculate output indexes are considered,
labor input data are examined, s'everal productivity in­
dexes are calculated, and suggestions are made for
future research. The specific approach, geographic
area, and time period covered vary by service, depen­
ding upon data availability. Though the following ex­
amples illustrate how national productivity indexes
might be calculated, they should not be taken as
representative of State and local government outputs,
data availability, or productivity.

El@©trSe U tilities
Electric utilities are a good starting place for an in­
vestigation of the feasibility of calculating State and
local government productivity indexes because they are
easily identifiable, they have a measurable set of out­
puts, and the larger ones report annually to the Federal
Government. Moreover, productivity indexes have long
been calculated for private and cooperative utilities so
there is a large analytical and institutional base of
knowledge on which to build.

Institutional oonsiderations
Of the three basic types of electric utilities—private,
cooperative, and government—the private or investor-

N ote : This chapter has benefited from the comments of Edwin

Adelman, Charles Ardolini, Horst Brand, Arthur Herman, Clyde
Huffstutler, James Urisko, and Martin Ziegler of the Bureau of Labor
Statistics; Steven Barsby, Consultant; Susan Clark of the National
League of Cities; John Flynn of the New York State Legislative Staff;
Paul Fry of the American Public Power Association; Harry Hatry of
the Urban Institute; John Humphrey of the Unemployment Insurance
Service; James Jarrett of the Council of State Governments; Ray­
mond Long of the National Association of State Budget Officers;
Gary Marshall of the Distilled Spirits Council of the United States;
and Allan Stevens of the Bureau of the Census.




1 Some authorities divide the utilities into two types—privately and
publicly owned. Cooperatives are considered as privately owned.
2 Statistics o f Publicly Owned Electric Utilities in the United
States-1977 (U.S. Department of Energy, Energy Information Ad­
ministration, 1979), pp. 6-7.

25

Table 10. Distribution of kilowatt hours sold, customers
seresd, and plant and equipment owned by typ® ©f utility
ownership, 1078

The Federal Energy Regulatory Commission (ferc),
the Energy Information Administration (eia) of the
U.S. Department of Energy, and the American Public
Power Association (appa ) include all State and local
government utilities in their statistics, eia , for example,
defines a “ municipal” power utility as “ a city, county,
irrigation district, drainage district, or other political
subdivision or agency of a State competent under the
law thereof to carry on the business of developing,
transmitting or distributing power.” 8
The Standard Industrial Classification (sic) system
also includes all “ companies and systems engaged in the
generation, transmission and/or‘distribution of electric
energy for sale.” 9 This definition is used in the b l s data
and productivity analysis of investor-owned and
cooperative utilities.
The discussion and calculations that follow focus on
State and local government electric utilities whether they
generate, transmit, or distribute power. Utilities in the
U.S. territories are excluded, however. These are includ­
ed in the eia and the appa statistics, but not in those
compiled by Census.
A potential problem, particularly for data collection,
is the utility that covers more than a single service. Com­
bined electric-gas utilities are common in the private sec­
tor, and a separate classification is given to them in the
sic . The problem is likely to be even more common in
local government, where utilities may be combined in a
single agency. In Los Angeles, for example, water and
power are combined; in Memphis, water, light, and gas
are combined.

(Percent)
Kilowatt
hour
sales

Customers
served

Plant and
equipment
owned

T o ta l..........................

100.0

100.0

100.0

Private (investor)....................
Cooperative ( r e a ) ..........................
Government..........................
Federal..........................
State and local ..............

77.6
6.6
15.8
3.6
12.2

76.2
10.2
13.6
(')
13.6

73.3
7.1
19.6
8.0
11.6

Type of
ownership

1 Less than 0.05 percent.
SO URCE:

Public Power, Jan./Feb. 1980, p. D-3.

cent of all utility expenditures in 1977. Salaries and
wages accounted for about 9 percent (16 percent of cur­
rent operating expenditures). Fuel, materials, supplies,
and purchased power accounted for the remaining 48
percent (table 11).
State and local systems are scattered throughout the
United States. The District of Columbia, Hawaii, and
Montana are the only jurisdictions which have no State
or local government power system. For the 48 States
with one or more systems, the average number of
employees was 1,220 and the average revenue was $146
million in 1976. California led the list with 9,600 fulltime-equivalent employees and $1,063 million in
revenue.3
According to the Bureau of the Census, State and
local utilities had gross revenue of $7,142 million in
fiscal year 1977. The American Public Power Associa­
tion (a p p a ) estimates the figure at $7,059 million (ex­
cluding Puerto Rico) for calendar year 1977.4 The 19
largest utilities accounted for 57 percent of total dollar
sales. The 160 largest reporting units accounted for
about 95 percent, according to information published
by the U.S. Department of Energy.5
Kilowatt hour sales to the ultimate customer is the
statistic most often used to measure electric utility out­
put. In 1978, State and local systems sold 234,478
million kilowatt hours.6 The 10 largest systems ac­
counted for about 35 percent of the kilowatt sales and
the 25 largest for about 50 percent. The largest 160 ac­
counted for almost 90 percent.7
The coverage of utility statistics varies. Bureau of the
Census financial and employment statistics include all
State and local government electric power operations;
before 1980 the Census employment statistics included
only local employees.

Research am statistics
id
Research and statistics abound on electric utility
operations. This stems from the public’s interest in utili­
ty regulation and rate setting, the great debates of the
1930’s over public vs. private power, and the more re­
cent interest in the safety of nuclear power and the ef­
fect of acid rain, all issues that lend themselves to
economic analysis. Universities, private consulting
firms, utilities, and government regulators routinely
study the industry.
Productivity measurement has attracted a moderate
amount of interest on the part of economists. Jacob
Gould was one of the first to attempt an electric utility
productivity index.1 Gould calculated productivity in­
0
dexes for 1899-1942 for the total electric utility in­
dustry, public and private. He measured output of
kilowatt hours, both unweighted and weighted by class
of service; and inputs of labor, fuel, and capital.

3 Number of employees from 1977 Census o f G overn­
ments—Compendium o f Public Employment (Bureau of the Census,
1979); revenue from 1977 Census o f Governments—Compendium o f
Government Finances (Bureau of the Census, 1979).
4 Public Power, Jan./Feb. 1980, p. 0 -2 .
5 Publicly Owned Electric Utilities—1977, p.16.
6 Public Power, Jan./Feb. 1980, p. D-2.
7 Publicly Owned Electric Utilities—1977, p. 16.




8 Ibid., p. 1.
9 Standard Industrial Classification Manual (U.S. Office of
Management and Budget, 1972), p. 218.
1 Jacob Martin Gould, Output and Productivity in the Electric and
0
Gas Utilities—1899-1942 (New York: National Bureau of Economic
Research, 1946).

26

Tab!® 11. FSnanees ® Stat© and local government @!©c?rle utilities by type of government, fiscal year 1077
?
(Millions)
Expenditures
Government

Current operations

Revenue
Total

Capital

Interest
on debt

Total

Salaries
and wages

Other

Total ......................................

$7,142

$9,313

$3,167

$856

$5,289

$861

$4,428

States ................................................
Municipalities....................................
Special districts..................................
Townships..........................................
Counties............................................

377
5,353
1,273
127
11

982
5,377
2,804
131
19

630
909
1,619
7
1

175
345
329
2
5

177
4,123
856
121
12

41
587
217
15
2

136
3,536
639
106
10

Note:

Finances (Bureau of the Census, 1979), p. 33.

Because of rounding, detail may not add totals.

So u r c e :

1977 Census of Governments—Compendium of Government

power, and material and supplies.1
7
Howard Axelrod called for the use o f a series of par­
tial productivity ratios targeted to different functions.1
8
He suggested kilowatts per man hour and total capitaliz­
ed cost per kilowatt hour for measuring power genera­
tion productivity, and the number o f bills processed per
man hour and the number of complaints answered per
man hour for assessing customer account productivity.
Research papers by lulo and Pace are also widely
cited in the literature.1 Both authors used econometric
9
techniques to identify the variables most responsible for
differences in utility unit costs. Neither study found that
employee hours play a significant role.
Lastly, J.W. Wilson and Associates conducted a
study for the U.S. National Bureau of Standards which
focused on helping States improve their electric utility
regulatory process.2 As part of the study, they review­
0
ed electric utility productivity measurement and
associated problems.
Several points stand out in the research:

Kendrick and Barzel calculated indexes of private
electric utility productivity. Barzel’s work covered
1929-55 for partial and total factor productivity.1 He
1
derived kilowatt hour class weights through regression
analysis. The Kendrick indexes were for total factor
productivity and covered 1948-69.11
2
3
Dragonette and Jaynes of b l s calculated and publish­
ed an index of labor productivity for the electric and gas
industry in 1965 which covered 1932-64.1 This index of
3
investor-owned and cooperative electric utilities is up­
dated yearly and has recently been divided between gas
and electricity.1 The Bureau also calculates an index of
4
Federal electric power productivity as part of its Federal
productivity measurement program.1 No index is
5
routinely published for State or local government elec­
tric power productivity.
Much of the research has focused on utility regula­
tion. Recently, it has centered on the role productivity
should play in regulation, particularly in rate setting.
Kendrick discussed the issue in several papers.1 Rodney
6
Stevenson considered conceptual issues, including par­
tial and total factor productivity, rates of technological
change, econometric modeling, and management
audits. He also presented a private electric utility pro­
ductivity index. Total factor productivity was estimated
using five factors—capital, labor, fuel, purchased

1. The principal measure of output is the
kilowatt hour, often weighted to reflect the dif­
ferent classes of service.
2. No input measure is dominant. Labor is often
used, apparently because of the difficulty in
measuring capital and fuel, the two factors most
favored in theoretical discussions.

1 Yoram Barzel, “ Productivity in the Electric Power Industry
1
—1929-1955,” Review o f Economics and Statistics, November 1963,
pp. 395-408.
1 John W. Kendrick, Postwar Productivity Trends in the United
2
States (New York: National Bureau of Economic Research, 1973).
1 Joseph E. Dragonette and Philip W. Jaynes, “ Output Per Man3
Hour, Gas and Electric Utilities,” Monthly Labor Review, January
1965, pp. 34-39.
1 Productivity Measures fo r Selected Industries—1954-81, Bulletin
4
2155 (Bureau of Labor Statistics, 1982).
1 Measuring Federal Productivity (U.S. Office of Personnel
5
Management, 1980).
1 John W. Kendrick, “ Efficiency Incentives and Cost Factors in
6
Public Utility Automatic Reserve Adjustment Clauses,” Bell Journal
o f Economics, Spring 1975, pp. 299-313; and “ Some Productivity
Issues in the Regulated Industries,” in Public Utility Productivity,
Waiter L. Balk, ed., (Albany: New York State Department of Public
Services, 1975), pp. 3-9.




3. The research has focused on private utility
measurement, particularly as related to regulatory
issues.
1 Rodney E. Stevenson, “ Regulating for Efficiency in the Public
7
Utility Industry” and “Productivity in the Private Electric Utility In­
dustry,” in Public Utility Productivity.
1 Howard J. Axelrod, “ Measuring Electric Utility Productivity,”
8
in Public Utility Productivity, pp. 57-69.
1 William lulo, Electric Utilities—Costs and Performance
9
(Pullman, Washington: State University Press, 1961); and Joseph D.
Pace, “ Relative Efficiency in the Electric Utility Industry” (Ann Ar­
bor, University of Michigan, 1970).
2 J.W. Wilson and Associates, The Measurement o f Electric Utility
0
Productivity, Vols. I and II (National Bureau of Standards, 1980).

27

Weighting output. Because production costs vary, dif­
ferentiating or segmenting output to account for the dif­
ferent classes of service is common practice, bls, for ex­
ample, uses three basic weighted aggregates—residen­
tial, commercial/industrial, and other—to estimate
labor productivity o f investor-owned u tilities.2
2
Wilson and Associates propose a seven-way break
—residential, commercial, industrial, street and
highway lighting, public railroads and railways, in­
terdepartmental, and sales for resale.2
3
Weights should reflect the input being measured: A
labor productivity measure should use unit labor inputs,
and a capital productivity measure should use capital in­
put weights. However, appropriate statistics are not
always available, and it has become common practice to
use average price per kilowatt hour for each class of serv­
ice (revenue divided by kWh’s sold) as the weight. In­
sofar as labor and capital requirements and costs are
proportional to price differentials, price weights are
useful.
lulo has shown that the relationship between unit cost
and unit revenues is good.2 Both bls and Wilson use
4
unit revenue as weights, and this procedure is suggested
for State and local government output measurement.
Utility weights have been calculated for 1967, 1972,
and 1977 (table 12). For 1967 and 1972, weights were
calculated for residential, commercial/industrial, and
other. For 1977, additional data made it possible to
divide the commercial/industrial field between small
and large producers.

4. Little interest has been shown in State and
local government productivity measurement, and
no productivity index has been calculated.
Although no productivity index exists, considerable
data are available on State and local government electric
power operations. The eia , the appa , the Bureau of the
Census, investment firms, and individual utilities all
publish some data in this area. Statistics on the number
of customers, kilowatt hour sales, revenues, number of
generating stations, miles of transmission lines, plant
cost, and allowances for depreciation and amortization
are routinely collected and published. Data which could
be used to construct a State and local government pro­
ductivity index are reviewed in the following sections.

Outputs
The output measure used most often is kilowatt hours
sold. Other measures are the number of customers,
kilowatt hours generated, percent of capacity used,
generator capacity, dollar sales, and net profit. The
relative strengths and weaknesses of the different
measures are seldom discussed in the literature.
William lulo, one of the few researchers who has ex­
amined the different measures, offers four reasons for
using kilowatt hours:
1. The measure is familiar to industry and the
public and has long been used by both.
2. Data are readily available. All utilities collect
and keep, and most report, statistics on kilowatt
hours sold.

Quality o f service. Quality of service has not been an
issue for most researchers. Whether this lack o f concern
is due to conceptual difficulties, data problems, a feel­
ing that quality is not an important issue, or a combina­
tion of factors is not known. Researchers who have
studied the quality issue have singled out the following
as important:

3. The kilowatt hour is a standard physical unit
which is not affected by price changes.
4. Kilowatt hours are a rough indicator of the in­
dustry’s ability to produce electric energy.
The only argument lulo offers against the use of the
kilowatt hour is that production costs per kilowatt hour
are not always similar though this may be implied.2
1
The previous chapter listed the criteria used in this
study to select State and local government output
measures. The kilowatt hour satisfies the four essential
criteria and three o f the four optional criteria. The only
criterion not met is the one noted by lulo: Kilowatt
hours are not always proportional to the cost (labor
hours) spent in their production.
Production costs vary by class o f service. Capital re­
quirements to construct distribution systems for in­
dustrial users are normally less than those required to
service residential customers per kilowatt hour
delivered. Similarly, the labor required to maintain and
service industrial/commercial distribution is likely to be
less than that required for residential service.

1. Reliability. This factor concerns the number,
length, and duration of supply interruptions. Inter­
ruptions can be caused by factors such as weather,
disaster, lack of equipment, or lack of fuel.
Building redundancy into the system increases
reliability.
2. Voltage. Lack of proper equipment or insuffi­
cient generating capacity may cause voltage fluctua­
tions which result in damage or malfunction of user
equipment. Installing additional equipment can
control voltage fluctuations.
3. Aesthetics. The aesthetic factor most often
discussed is placement of utility lines— above vs.
2 Dragonette and Jaynes, “ Output per Man-Hour,” pp. 34-39.
2
2 Wilson and Associates, Electric Utility Productivity, Vol. II, p. 16.
3
2 lulo, op. cit.
4

2 lulo, Electric Utilities—Costs and Performance, p. 30.
1




28

Tab!© 12. Weigfals for eaSeisSating output indexes for State
and local government electric utilities by class of service,
1967, 1972, and 1977

while inputs include all the labor needed to distribute,
transmit, and generate electricity.

Dollars per kilowatt hour
Class of service

Residential ............................
Commercial/industrial............
Small..............................
Large..............................
Other......................................

1967

1972

$0.0149
.0116
n.a.
n.a.
.0147

$0.0163
.0132
n.a.
n.a.
.0168

Statistics. Statistics on kilowatt hours sold to ultimate
consumers by government utilities have been collected
for many years. Today, eia collects and publishes infor­
mation annually on approximately 160 of the largest
State and local utilities. The appa also collects and
publishes kilowatt hour statistics on State and local
utilities.
The eia and its predecessor organizations, particular­
ly the Federal Power Commission, have published
statistics on publicly owned utilities since 1946.
However, reporting requirements and tabulation pro­
cedures have been modified so that year-to-year sum­
mary comparisons are likely to be misleading. Also, the
number o f utilities reporting each year depends on the
reporting requirements and whether the utilities abide
by such requirements.
The new eia series starting with 1974 includes those
utilities which have consistently filed annual reports and
have annual operating revenue of at least $5 million.
The new series includes abut 160 utilities and reporting
units. (The old series included 511 utilities; the new
series lists the utilities in Tennessee which purchase their
power from the tva under a single heading, “ tva pro­
viders.” )
The eia statistical reports include, in addition to sum­
mary data, details for individual utilities on the number
o f customers, kilowatt hour sales, revenues, production
expenses, assets, liabilities, profit and loss, generation
capacity, and the number o f miles o f transmission lines.
Kilowatt hour sales are divided by class of customer so
that unit revenue* weights may be developed.
Insofar as national output measurement is concerned,
a limitation o f eia statistics is that they do not indicate
the extent of their coverage. But if one accepts the appa
estimates for total kilowatt hour sales by State and local
utilities, the eia statistics account for about 90 percent.
Another limitation is that the sample excludes the
smaller utilities, which may bias productivity calcula­
tions.
The appa collects kilowatt hour sales monthly from
the 29 largest municipal utilities. It uses these statistics,
data collected by eia , and data from the private sector
to estimate the annual sales of kilowatt hours to
ultimate users, appa is the only known source of
published data on total State and local government
sales.
Several problems arise in using appa estimates to
calculate national State and local electric utility output.
First, the figures include kilowatt hours sold in
American Samoa, the Canal Zone, Guam, Puerto Rico,
and the Virgin Islands. Statistics for these five territories
can be subtracted from the total, however, to obtain the
estimated kilowatt sales in the United States. Moreover,

1977
$0.0293
.0253
.0328
.0211
.0345

n.a. = not available.
Computed from Statistics of Publicly Owned Electric Utilities
in the United States, selected issues (U.S. Department of Energy, Informa­
So u r ce:

tion Administration).

below ground. Placing utilities below ground in­
creases initial costs. Its impact on operating costs is
open to debate.
Adjustments for quality have not been attempted in
this study.

Generation vs. sales. Because State and local govern­
ment utilities are not closed systems, a problem may
arise in calculating productivity when State and local
government utilities generate and sell power to nonState and nonlocal utilities, and purchase and distribute
power that other utilities generate and transmit. North
Platte, Nebraska, for example, sold 178 million
kilowatts in 1978 to ultimate consumers but generated
no electricity itself. The New York State Power
Authority, on the other hand, generated 34 billion
kilowatt hours in 1978 but sold only 14 billion to
ultimate consumers; it sold the remaining kilowatt
hours to other utilities.
The difference between generation'and sales to the
ultimate consumer becomes a problem when the ratio
between generation and sales is changing, as it is with
State and local government electric power utilities. In
1967, State and local government utilities generated
101,672 million kilowatts and sold 142,928 million
kilowatts to ultimate consumers, a difference of 29 per­
cent. In 1978, 228,645 million kilowatts were generated
and 234,478 million kilowatts were sold, a difference of
about 2 percent. While growth has been considerable in
both generation and ultimate sales, the gap has been
closing.
Although the growth of State and local government
generating capacity vis-a-vis sales to ultimate customers
has been large, the impact on productivity apparently
has been limited. Labor productivity with, and without,
the incremental change in employees working in genera­
tion differ by less than 1 percent for large utilities, the
only group for which data are readily available. Any
bias as a result of increasing generation is downward.
That is, productivity gains are understated, since output
is calculated as a function of final sales to' customers



29

the Puerto Rico Electric Power Authority is the only
large utility located in the territories, and its output can
be easily removed from the summary statistics, as has
been done in the statistics presented here.
Second, appa statistics are not separated by
class—residential, commercial, highway lighting, and so
forth—or by geographic area. As noted, a class division
is needed to assign unit revenue weights. If that com­
putation is not made, which the preceding discussion
suggested was desirable, this type of information is not
needed. Geographic information is valuable for match­
ing and comparing statistics from various sources.
Third, and more troublesome, the error associated
with the appa estimate is not known since the true
universe is unknown. However, comparison with other
data suggests that they are reasonably good estimates of
the universe.

Table 13. Three output indexes for State and local
government electric utilities, 1967-78
(1977 = 100)
Kilowatt hour sales to ultimate customers
Local
utilities

Large
utilities

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

63.6
69.1
75.5
81.5
87.3
88.3
90.0
91.1
91.5
96.2
100.0
104.3

65.2
71.6
77.8
84.2
90.4
91.6
93.0
93.8
94.6
99.0
100.0
103.3

54.4
57.4
62.5
67.4
70.7
76.2
83.5
83.1
85.1
90.1
100.0
106.2

Average annual percent
change1..............................

4.0

3.7

6.0

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

1Average annual rates of change in this and all subsequent tables are
based on the linear least squares trend of the logarithms of the index
numbers.

Output indexes. Three output indexes are presented here
(table 13). Two are based on appa statistics. One is an
index of local and State kilowatt sales to ultimate
customers with statistics for Puerto Rico removed. The
other reflects only local government electric power
kilowatt sales. In addition to Puerto Rico, sales of the
major State authorities have been removed so that the
index would match Census local government employ­
ment statistics. In addition, a third output index has
been constructed for 33 large State and local govern­
ment electric utilities for which data were readily
available.2 The kWh statistics were taken from Federal
5
Government sources and data provided by the utilities.
The average annual rates of growth for these three in­
dexes are 4.0? 3.7 and 6.0 percent, respectively. The first
two indexes, as expected, are very similar; the large utili­
ty index grows about 50 percent faster than the other
two. These statistics will be used later in this section to
calculate electric power productivity.

S o u r c e : Local and State utilities—January issues of Public Power (Puer­
to Rico statistics are excluded). Local utilities— local and State utility index
with major State utility sales excluded. Large utilities— compiled from
Federal and individual utility sources.

number of employees. A review of a number o f annual
reports did not find any statistics of the number of
hours worked, part-time employment, or seasonal
employment. Discussions with government utility of­
ficials suggest that in most cases data are readily
available on the number o f employees but not on
employee hours.
Appa collects some employee data in its survey of
utility salaries. The salary survey, initiated in 1957, was
conducted every other year until 1977, when it became
an annual survey. The 1979 survey collected informa­
tion from 358 State and local utilities. Data collected in­
clude the number of permanent employees, salaries of
selected officials, and the number o f years since the
nonsupervisory engineers received their degrees. Other
statistics include kWh sales, kWh purchases, and
generating capacity. Data on sales to ultimate con­
sumers are not collected. The appa statistics can help in
data analysis and evaluation but by themselves are not
sufficient to construct a viable labor index.
The Bureau of the Census is the only organization
that routinely collects and publishes time series data on
public power utility employment. It publishes figures on
both total employment and full-time-equivalent
employment. However, several problems arise in using
these data to measure electric utility labor productivity.
First, Census did not include State employees in its
electric utility series until 1980. Since output statistics
include both State and local kilowatt hours, the labor
input series must include both State and local govern­
ments. The 1980 Census figures show 3,000 State power
employees; other data suggest that the figure is closer to

Labor inputs and employee eosts
The previous chapter listed three labor measures to be
used in calculating State and local government labor
productivity: Number of employees, number o f fulltime-equivalent employees, and number of employee
hours. Data to calculate these measures are not always
available.

Sources o f data. The three sources of data on publicly
owned power system employment are: (1) The in­
dividual public power systems, (2) the American Public
Power Association (appa ), and (3) the Bureau o f the
Census.
Although all public power agencies collect and main­
tain data on employment, and some publish them an­
nually, these statistics are mostly simple counts of the
2 See appendix C for a list of the utilities included.
5




Local and
State

Year

30

4,000. The important consideration for trend deter­
minations is relative growth, not absolute numbers.
A second problem is that Census employment is col­
lected for only one month—October—o f each year.
These statistics do not capture seasonal employment. If
the October/seasonal proportion remains constant, the
October statistics will be satisfactory for trend deter­
minations. A set of October/seasonal statistics is needed
to establish whether there is constancy.
A third problem is that Census statistics are available
for only aggregate and full-time-equivalent employ­
ment. No data are collected on the number of hours,
nor are the statistics broken down between operations
and force account (construction) employees.
Another potential problem, discussed briefly earlier,
is the assignment of government personnel to the power
function when, in fact, they work in other or multipleservice areas—e.g., gas, water, and sewerage. Overhead
personnel are a special case. In labor data collected for
this study, however, the inability to allocate personnel
was not a significant problem. Only one of the 33 large
utilities indicated it was difficult to allocate personnel.
Two large utilities not included in the sample also in­
dicated that they were unable to provide the employee
data by function. The extent of the problem for the
smaller utilities is not known, but Census personnel do
not feel that it is significant. Census suggests that when
governments have this type of problem, they should
allocate personnel using revenue figures.
Although statistics are not available to compute an
hours index, other sources suggest that such an index
would parallel the total employee index and the full­
time-equivalent employee index. Census full-timeequivalent and total employment show a high degree of
correlation (table 14). Between 1967 and 1978, both
grew at the same annual rate—0.7 percent. For trend
comparisons, the two measures should not differ a great
deal. Also, private electric utility trends for labor hours

Table 15. Two employment indexes for private sector
electric utilities, 1967-78
(1977 = 100)
Year

Year

Full-timeequivalent
(thousands)

57
56
54
56
58
58
60
60
60
59
58
60

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

82.3
84.1
86.5
89.8
90.8
93.3
96.5
98.9
97.4
98.0
100.0
105.1

82.6
84.6
87.4
90.4
91.2
94.1
98.3
99.7
96.6
97.8
100.0
107.0

Average annual percent
change:
1967-78 ........................
1975-78 ........................

2.1
1.3

2.0
1.3

SO URCE: Donald M. Fisk, "Pilot Study Measures Productivity of State,
Local Electric Utilities,’’ Monthly Labor Review, Dec. 1981, p. 46.

and total employment closely parallel each other during
the time period examined here (table 15).

Employment indexes. Three employment indexes were
computed (table 16). The first, based on Census
statistics, is for total local government electric power
employment. The second index attempts to capture
State as well as local government employment by adding
employment for the four largest State systems. This in­
dex should approximate total State and local govern­
ment electric power employment since local govern­
ments constitute over 90 percent of electric power
employment and the four State systems constitute over
90 percent of State electric power employment. The
third index, based on statistics provided by the utilities,
reflects employment in 33 large, publicly owned
utilities.
The average annual rates o f growth for the three in­
dexes are 0.7, 0.9, and 1.6 percent for the local,
local/selected States, and large utilities, respectively.
The first two indexes are quite close, as expected, but
the third grows at almost twice the rate o f the other two.
These indexes are used to calculate the productivity in­
dexes presented later.

Full-timeequivalent as
a percent of
number of
employees

59
58
55
58
59
59
61
62
62
61
60
62

Employee hours

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

Table 14. Two employment measures for local government
electric utilities, 1967-78
Number of
employees
(thousands)

Employees

96.6
96.6
98.2
96.6
98.3
98.3
98.4
96.8
96.8
96.7
96.7
96.8

Salaries and wages. During 1967-78, employee salaries
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

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

S o u r c e : Employment data from
(Bureau of the Census).




Public Employment,

and wages increased at an average annual rate o f 7.8
percent (table 17). This index, computed from data
taken from Public Employment, an annual publication of
the Bureau of the Census, is for salaries and wages paid
to full-time local government personnel in October of
each year. The criticisms of Census employment
statistics discussed earlier apply equally here-—they in­
clude only one month of data and exclude State employ­
ment. An additional criticism is that fringe benefits are
not included. This exclusion will bias labor cost trends if

annual issues

31

hours and total State and local government employment
result from local operations.
The local government index shows a 51-percent
growth between 1967 and 1978, an annual increase of
3.0 percent (table 18). The average annual increase for
kilowatt hours and employment during 1967-78 was 3.7
and 0.7 percent, respectively. This index is unweighted
since only aggregate kilowatt hour sales are available.
The main question surrounding this index is data ac­
curacy, which was discussed earlier. Kilowatt hour sales
were taken from a p p a statistics, from which State sales
to final customers were removed. IState kWh sales were
taken from State and f p c / f e r c / e i a data, and should
be accurate. The employment statistics were taken from
the Bureau of the Census; the strengths and weaknesses
of that source of data have been noted.

Table 16. Three employment indexes for Slate and local
government electric utilities, 1867=78
(1977 = 100)
Local
utilities

Local and
selected State
utilities

Large
utilities

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

98.1
96.5
91.5
96.5
98.1
98.1
101.5
103.1
103.1
101.5
100.0
103.1

96.2
94.7
90.0
95.0
96.7
96.9
100.3
102.1
102.3
101.1
100.0
103.7

86.7
87.2
89.6
92.5
96.2
96.2
98.0
99.9
99.8
99.7
100.0
103.2

Average annual percent
change ..............................

.7

.9

1.6

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

Local utilities— data to compute index from Public Employ­
annual issues. Local and selected State utilities— data for local
government employees from Public Employment, annual issues; data for
State government employees provided by individual utilities. Large
utilities— data to compute index provided by individual utilities.
So u r c e :

Local and selected State government index. An il­

ment,

lustrative index was also calculated for all local and
selected State government utilities. Twenty-one Stateowned utilities were identified. Four—Power Authority
of the State of New York, South Carolina Public Serv­
ice Authority, Grand River Dam Authority, and Lower
Colorado River Authority—accounted for almost all
employment and sales to ultimate customers. These four
utilities were combined with the local government in­
dexes to arrive at the local and selected State govern­
ment index. This index should include 99 percent o f all
State and local government employment and kilowatt
hour sales to ultimate customers.
The results of these calculations show that local and
selected State government productivity increased by
about 52 percent from 1967 to 1968 (table 19). The
average annual increase in output per employee was 3.0
percent; in kilowatt hours, 4.0 percent; and in
employees, 0.9 percent. The index is unweighted.

Table 17. Index of average salaries and wages of local
government electric utility employees, 1967-78
,(1977 = 100)
Year

Index

1967..............................................................
1968................................................................
1969........................................
1970........................................................
1971................................................................
1972............................................................
1973........................................
1974................................................................
1975................................................................
1976................................................................
1977.................................................................
1978............................................ ..............

48.6
50.3
55 3
58.6
62.1
66.0
74.6
79.3
87.3
96.3
100.0
105.6

Average annual percent change:
1967-78..................................................
1967-72..............................
1973-78..................................................

7.8
6.6
7.5

S o u r c e : D a ta t o c o m p u t e in d e x f r o m

Public Employment, a n n u a l is s u e s .

Table 18. Indexes of output, employees, and output per
employee for local government electric utilities, 1967-78
(1977 = 100)

the ratio of fringe benefits to salaries and wages is
changing through time.

Employees

Output per
employee

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

65.2
71.6
77.8
84.2
90.4
91.6
93.0
93.8
94.6
99.0
100.0
103.3

98.3
96.7
91.6
96.7
98.3
98.3
101.7
103.3
103.3
101.7
100.0
103.0

66.3
74.1
84.9
87.1
92.0
93.2
91.5
90.8
91.5
97.3
100.0
100.0

Average annual percent
change ..............................

Prodyetivifiy Snd@ 3
x@

3.7

.7

3.0

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

Three illustrative productivity indexes for electric
utilities are presented in this section. One covers local
government; another, local and selected State govern­
ment; and the third, large, government-owned utilities.
All thfee draw on the data and investigative approaches
presented in the preceding discussion.

Local government index. Local governments sell most
of the kilowatt hours supplied by government to
ultimate customers in the United States; the States play
a relatively minor role. Although precise statistics are
not available, more than 90 percent of total kilowatt



Output

Year

S o u r c e : O u tp u t, t a b le
c a lc u la t e d .

32

1 3 ; e m p lo y e e s , t a b le

16; o u tp u t

p e r e m p lo y e e ,

Table 19. Indexes of output, employees, and output per
employee for local and selected State government electric
utilities, 1967-78

There is very little difference between the weighted and
unweighted indexes.
The employment data used in the index reflect the
average number of employees. These statistics, for the
most part, were provided by the individual utilities, and
were checked against Census data insofar as possible.

(1977 = 100)
Output

Employees

Output per
employee

......................................
......................................
......................................
......................................
......................................
......................................
......................................
......................................
......................................
......................................
: ....................................
......................................

63.6
69.1
75.5
81.5
87.3
88.3
90.0
91.1
91.5
96.2
100.0
104.3

96.2
94.7
90.0
95.0
96.7
96.9
100.3
102.1
102.3
101.1
100.0
103.7

66.1
73.0
83.9
85.8
90.2
91.1
89.8
89.2
89.4
95.2
100.0
100.7

Average annual percent
change ..............................

4.0

.9

3.0

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

S o u r c e : O u tp u t, t a b le

1 3 ; e m p lo y e e s , t a b le

16; o u t p u t

Comparison o f productivity indexes. The local and
local/State indexes show much the same productivity
growth between 1967 and 1978; both increased at exact­
ly the same average annual rate—3.0 percent (table 21).
Similar trends are to be expected for these two indexes,
since over 90 percent of local/State government electric
utility kWh sales and employment are those of local
government.
The 33 large utilities, which are a subset of the
local/State government index, showed a markedly
faster rate o f productivity increase—4.4 percent—be­
tween 1967 and 1978. This, too, is to be expected since
economies of scale are important in electric power
generation and distribution.
All three indexes showed a dramatic drop in output
and productivity in 1973-74, in response to the oil em­
bargo and the recession. Also, productivity growth
peaked in 1969 in all three indexes.
Although uncertainty surrounds some of the data us­
ed to calculate the indexes, the output index, the three
input indexes, and the three productivity indexes move
together quite well, and any differences, such as with
the large utility index, are easily explained. Thus, even
though the precise increase in local and State govern­
ment electric utility labor productivity over the past
decade may be open to question, the general movement
is clear.

p e r e m p lo y e e ,

c a lc u la t e d .

Table 20. Weighted indexes of output, employees, and
output per employee for 33 large government electric
utilities, 1967-781
(1977 = 100)
Output

Employees

Output per
employee

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

54.4
57.4
62.5
67.4
70.7
76.2
83.5
83.1
85.1
90.1
100.0
106.2

86.7
87.2
89.6
92.5
96.2
96.2
98.0
99.9
99.8
99.7
100.0
103.2

62.7
65.8
69.7
72.8
73.5
79.2
85.1
83.1
85.2
90.3
100.0
102.9

Average annual percent
change ..............................

6.0

1.6

4.4

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

Comparison with private utility index movements. In­
dexes that BLS routinely calculates for investor-owned
and cooperative electric utilities grew somewhat faster
than those for State and local government utilities dur­
ing 1967—78 (table 22.) However, this may be a function
of size. Large public system output and productivity
grew faster than that of private utilities.2
6
Productivity growth slowed in both the public and
private sectors in the early 1970’s. Between 1967 and
1972, private utility growth dropped from 5.5 percent to
2.4 percent. The comparable figures for State and local
governments were from 6.7 percent to 2.8 percent.
In neither government nor private electric utilities did
productivity increase fast enough to offset the increase
in salaries and wages. The result was an increase in
average kWh unit cost, particularly in the latter part of
the period. In State and local governments, unit cost in-

1See appendix C for list of utilities included.
S o u r c e : O u tp u t, t a b le

1 3 ; e m p lo y e e s , t a b le

16; o u tp u t

p e r e m p lo y e e ,

c a lc u la t e d .

Large, government-owned utility index. A productivity
index was also calculated for 33 of the largest publicly
owned utilities. The two State and 31 local government
systems included account for about 45 percent of all
municipal electric utility employment and 55 percent of
all kilowatt hour sales to ultimate customers. They can­
not be considered representative of all governmentowned utilities because of their large size.
Between 1967 and 1978, the index for these 33 utilities
increased by 64 percent, or an annual increase of 4.4
percent (table 20). The average annual increase for
kilowatt hours was 6.0 percent; for employment, 1.6
percent.
The kWh’s were taken from f p c / f e r c / e i a publica­
tions and were weighted for class o f service—residen­
tial, commercial/industrial and other—as discussed.



2
6 The average number of employees in State and local government
utilities was about 30 per system; in the private sector, 435; and in
large public systems, 925.

33

Table 21. Three productivity indexes for State and local
government electric utilities, 1967-78

penditures were estimated at 43 percent. A 1973 study of
investor-owned utilities found labor to be 15 percent
and capital 48 percent of total expenditures.2 For a bet­
7
ter understanding of productivity and cost movements
in the electric power industry, it would be necessary to
examine capital, fuel, and other factor inputs.

(1977 —100)
Local
utilities

Local and
selected State
utilities

Large
utilities

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

66.3
74.1
84.9
87.1
91.9
93.2
91.5
92.8
91.5
97.3
100.0
100.0

66.1
66.4
83.9
85.8
90.2
91.1
89.8
89.2
89.4
95.2
100.0
100.7

62.9
66.0
69.9
73.0
73.6
79.4
85.3
83.3
85.4
90.6
100.0
103.1

Average annual percent
change ..............................

3.0

3.0

4.4

Local

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

Stats Aicoholie Bs^srags
Control Operations
All States and many local governments regulate
alcoholic beverage sales. States license sellers, tax sales,
regulate advertising, set the legal age for purchase of
beverages, and establish the hours and days of sale. In
addition, about one-third of the States operate retail or
wholesale alcoholic beverage stores. These operations
are the focus of this section.
Alcoholic beverage operations have not captured the
attention of productivity measurement specialists,
unlike electric power production. Productivity
measurements do not exist for either the private or
public sector. However, the service is straightforward,
at least on the surface; the outputs are tangible; and
data are routinely collected on many aspects of State
alcoholic beverage operations.

S o u r c e : L o c a l u t ilit ie s , ta b le 1 8 ; lo c a l a n d s e le c t e d S t a t e u t ilit ie s , t a b le 1 9 ;
la r g e u t ilit ie s , t a b le 2 0 .

Table 22. Average annual rates of change for government
and private electric utility output, labor input, and
productivity, 1967-78
(Percent)

Type of utility

Local government utilities. . . .
Local and selected State
government utilities ..........
Large government utilities . . .
Private utilities ......................

Output
(kwh)

Labor input
(number of
employees)

Productivity
(output per
employee)

3.7

0.7

3.0

4.0
6.0
5.8

.9
1.6
2.1

3.0
4.4
3.7

Institutional eonsidorations
States which operate their own stores using govern­
ment employees are known as control or monopoly
States. Those States which use private sellers are known
as license States. There are 18 control States and 33
license States including the District of Columbia (table
23).
Although States are often divided into these two
groups, there is a broad spectrum of institutional ar­
rangements, from almost completely private operation
to total government control and operation. These can be
grouped into the following fairly distinct categories:

SO URCE: Local government, table 1 8 ; local and selected State govern­
ment, table 19; large government, table 20; private, Donald M. Fisk, “ Pilot
Study,” p.46.

creased 7.8 percent annually in 1967-78. Between 1967
and 1972, the rate of increase was 6.6 percent or about
the same as output per employee. But between 1973 and
1978, the average annual increase was 7.5 percent while
the rate of increase in output per employee dropped to
2.8 percent per year.
For private utilities, the average annual increase was
9.4 percent. Between 1967 and 1972, the rate of increase
was 6.7 percent, about the same as government and
slightly ahead of the 5.5-percent increase in output per
employee. However, between 1973 and 1978 the rate of
salary and wage increase jumped to 8.7 percent while
the increase in output per employee dropped to 2.4 per­
cent.

1. Private retail and wholesale operations, in ef­
fect in more than half of the States.
2. Private retail and government wholesale, as in
Mississippi and Wyoming.
3. Private and government (municipal) retail and
private wholesale, as in Minnesota.
4. Government (city and county) retail and
private wholesale, as in North Carolina.

Suggested research

5. Government and private agency retail and
government wholesale, as in Ohio.

Further analysis of State and local government elec­
tric productivity should focus on multifactor produc­
tivity. The Bureau of the Census estimates that in fiscal
year 1977 labor accounted for only 9 percent of State
and local government utility expenditures. Capital ex-

6. Government retail and wholesale,
Alabama and Virginia.

as in

This study focuses on those States included under 2,
2
7 Stevenson, in Public Utility Productivity. These percentages un­
5, and 6 above. North Carolina and Minnesota are not
doubtedly have shifted in recent years as fuel costs have increased.



34

Table 23. Type of control of alcoholic beverage sales by State
S ta te
o p e r a t io n s
S ta te

( c o n t r o l/
m o n o p o ly )

A l a b a m a ..............................................................

-

o p e r a t io n s
S ta te

( lic e n s e )

( c o n t r o l/
m o n o p o ly )

P r iv a te
o p e r a t io n s
( lic e n s e )

X

A l a s k a ....................................................................

S ta te

P r iv a te
o p e r a t io n s

X

A r i z o n a .................................................................

X

A r k a n s a s ...............................................................

X

C a l i f o r n i a ..............................................................

-

X
X

C o l o r a d o ..............................................................
C o n n e c t ic u t .........................................................

“

X

D e la w a r e ..............................................................

“

X

F l o r i d a ....................................................................

~

X

G e o r g i a .................................................................
H a w a i i ....................................................................

—

X

I d a h o .......................................................................

X

-

“
X

—

X

I l l i n o i s ....................................................................
Io w a

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

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

“

X

K e n t u c k y ..............................................................

“

X

L o u is ia n a ..............................................................

X

—

M a r y l a n d ..............................................................

-

X

M a s s a c h u s e t t s ................................................

-

X

M i c h i g a n ...............................................................

X

—

M i n n e s o t a ............................................................

X

X

K ansas

M a in e

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

M is s i s s ip p i............................................................
M is s o u r i.................................................................

N e w H a m p s h ir e
N e w M e x ic o

X

X

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

X

N e w J e r s e y .........................................................
_

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

X

X

X

-

X

N o r t h D a k o t a ......................................................

X
_

-

O h i o ..........................................................................

X

-

O k l a h o m a ............................................................

X

P e n n s y lv a n ia ......................................................

X
-

-

S o u th C a r o l in a ...................................................

-

X

S o u th D a k o t a ......................................................

-

X

N e w Y o r k ...............................................................
N o r t h C a r o l i n a ...................................................

U ta h

X

X

-

X

X

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

X

_

T e n n e s s e e ............................................................
T e x a s .......................................................................

_

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

X

W a s h i n g t o n .........................................................

X

V ir g in ia

X

W i s c o n s i n ............................................................
X

W y o m i n g ...............................................................

discussed because local rather than State personnel
operate stores, and data are not readily available.
The 17 control States (18 including North Carolina)
account for about 30 percent of the U.S. population and
25 percent of the spirits sold in the United States. In
1977, State alcoholic beverage sales totaled over $2
billion.
Five States accounted for 65 percent of the revenue
and 70 percent of the employment in fiscal 1977 (table
24). Pennsylvania alone accounted for 20 percent of the
revenue and 31 percent of the employment.
Most State alcoholic beverage control commissions
are responsible for four functions—wholesale sales,
retail sales, enforcement, and licensing (table 25). Each
o f the 17 control State commissions operates wholesale
or retail alcoholic beverage facilities. Sixteen o f the 17
State commissions license others, such as wineries and
restaurants, to sell alcohol. Fourteen of the 17 are
responsible for enforcement of State alcoholic beverage
laws and regulations; three States assign the respon­
sibility to their departments o f public safety.
This investigation includes wholesale as well as retail
sales. Two o f the 17 States operate only wholesale
operations; the other 15 operate a combination of
wholesale and retail. All o f the 15 State retail operations
sell spirits; 13 of these sell wine as well as spirits, and
5 also sell beer. Spirits account for about 75 percent of all
gallons sold in State stores, wine about 25 percent, and
beer less than 0.01 percent. Ten of the 15 States that
operate retail stores use agency (private) stores to aug­
ment their operations.



_

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

R h o d e I s l a n d ......................................................

X

D i s t r i c t o f C o lu m b ia ........................................

I n d ia n a

X

N e b r a s k a ...............................................................
N evada

The agency arrangement is an important issue in
calculating productivity. Usually, private retail mer­
chants operate agency outlets in addition to their nor­
mal sales. Control States have long used agents to serve
the population o f sparsely populated areas where a
“ full-service” State store could not be justified. Ohio,
for example, permits agents only in towns with a
population o f less than 10,000. Agents are usually paid
Table 24. Distribution of alcoholic beverage control revenue
and employees by State, fiscal year 1977
(Percent)
State

Revenue

Employees

All control States........

100

100

Pennsylvania..........................
Michigan................................
Ohio ......................................
V irginia..................................
Washington............................
O regon..................................
New Ham pshire....................
Alabam a................................
Io w a ......................................
Mississippi ............................

20
15
14
8
8
5
5
5
5
3

31
4
13
11
7
2
3
7
5
1

West Virginia..........................
Montana................................
Utah ......................................
M a in e ....................................
Idaho......................................
Vermont ................................
Wyoming................................

3
2
2
2
1
1
1

6
3
1
2
2
1
1

S
: Computed from 1977 Census of Governments—Compendium
of Government Finances (Bureau of the Census, 1979), p. 39; and 1977
Census of Governments—Compendium of Public Employment (Bureau
ource

of the Census, 1979), pp. 33-83.

35

rant operation of a State store. Licensing and en­
forcement of liquor laws and regulations are handl­
ed by another part of the State government.

a percent of their gross sales, although some States pay
a fixed fee or negotiate a price with the individual mer­
chant. In all cases, prices of alcoholic beverages are set
by the State.
Several States have begun to substitute agents for
State stores. Oregon, for example, has reduced the
number of State stores from 20 in 1976 to 6 in 1980, and
would have used agents entirely except for the interven­
tion of the State legislature. Maine, Montana, and Utah
have also substituted agents for State-operated stores.
Utah uses three different forms of agency ar­
rangements. In the first, agents operate beverage stores
just as they might a State store, but do not hire State
employees. In the second, merchants contract to sell
alcoholic beverages in addition to their regular product
lines. In the third, resort or hotel owners operate an
agency store as a convenience to their guests, usually at
no cost to the State.

Iowa. The Iowa Beverage Control Council operates
approximately 215 State retail stores and a
warehouse. The stores sell spirits, wine, and beer.
The Council also licenses on-premise alcoholic
beverage sales. Enforcement is the responsibility of
the Department of Public Safety.

Maine. The Bureau of Alcoholic Beverages, part of
the Department of Finance and Administration,
operates about 70 State stores and oversees the
operation of about 50 agency stores. Licensing is
part o f the State store operation, but enforcement is
handled by another part of the State government.

Michigan. The Liquor Control Commission o f the

Synopsis o f State operations. Alcoholic beverage opera­
tions vary markedly from State to State. A brief descrip­
tion o f operations in each control State as of 1980
follows.

Department of Commerce oversees the operation of
76 State stores and about 3,300 agency stores. State
stores provide a combination o f wholesale and
retail services; 99 percent of their operation is
wholesale. Licensing and enforcement are also
handled by the Commission.

Alabama. The Alabama Alcoholic Beverage Con­

Mississippi. The Alcoholic Beverage Control Com­

trol Board is responsible for sales, licensing, and en­
forcement within the State. It operates approx­
imately 130 retail stores and a warehouse. Enforce­
ment activities, which require about 100 of the
1,000 State Control Board employees, include drug
abuse as well as alcohol control activities.

mission of the State Tax Commission handles
wholesale spirit and wine sales, and alcoholic
beverage licensing and enforcement. Private stores
handle retail sales.

Montana. The Montana Department o f Revenue
operates approximately 115 retail stores and
oversees the operation of 30 agency stores. The
Department also handles alcoholic beverage licens­
ing and enforcement.

Idaho. The State Liquor Dispensary operates ap­
proximately 90 State stores and one warehouse and
oversees the operation of about 45 agency stores.
Agency stores are licensed to sell spirits and wine,
primarily in locations where sales would not war­

New Hampshire. The New Hampshire Liquor
Commission operates about 70 retail stores and one

Table 25. Forms and functions of State alcoholic beverage control operations
Type of sales

Scope of operations
State

Alabama ......................
Id a ho ............................
Iowa..............................
Maine............................
M ichigan......................
Mississippi....................
Montana ......................
New Hampshire............
O hio..............................
Oregon..........................
Pennsylvania................
Utah..............................
Vermont........................
Virginia..........................
Washington..................
West Virginia................
W yom ing......................

Sales

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

Licensing

Enforce­
ment

X

-

Wholesale
only

Wholesale
and retail

State
store

-

X
X
X
X
X

X
X
X
X
X

-

-

X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X

X

-

-

-

X
X
X
X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X

X

-

-

S
: Distilled Spirits Council of the United States, Summary of State
Laws and Regulations Relating to Distilled Spirits, 1977; and Retail

Agency

X
X
X
X

X
X

X
X

X
X

-

Type of beverage sold
Spirits

Wine

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

X
X
X
X
X
X
X
X

X
X
X
X
X
X
X

Beer

X

X

X
X

X

-

Outlets for the Sale of Distilled Spirits-1978, 1979; National Alcoholic
Yearbook 11th Edition, 1978.

ource




Type of retail sales

Beverage Control Association, Inc.,

36

warehouse. The Commission is responsible for
licensing and enforcement as well as sales. Sale of
alcoholic beverages provides about 18 percent of
total State revenues.

Dakota also permits local jurisdictions to operate their
own stores.

Ohio. The Department of Liquor Control operates

Little research or even descriptive writing has been
done on alcoholic beverage store operations, either
State or private. Academic research has focused almost
entirely on tax issues, such as the incidence o f the liquor
tax and its impact on consumption. The principal
references to State store productivity found during this
investigation were in State annual reports.
Statistics on alcoholic beverage operations are col­
lected under a variety of definitions. The Standard In­
dustrial Classification Manual defines alcoholic
beverage stores as:

Research amd statistics

about 290 retail liquor stores and is responsible for
another 125 agency stores. The Commission is also
responsible for licensing and enforcement.

Oregon. The Oregon Liquor Control Commission
relies heavily on agency stores for retail alcoholic
beverage sales. There are approximately 175 agency
stores and 6 State stores. The Commission is
responsible for licensing and enforcement.
Pennsylvania. The Pennsylvania Liquor Control
Board is responsible for licensing, enforcement,
and sales. Pennsylvania runs the largest State
alcoholic beverage control operation in the United
States, with approximately 725 State retail stores.
The stores sell spirits and wine.

“Establishments primarily engaged in the retail sale of
packaged alcoholic beverages such as ale, beer, wine and
whiskey, for consumption off premises.”2
8
The Bureau of the Census uses two slightly different
definitions for alcoholic beverage operations. For finan­
cial transactions, Census defines a “ liquor store” as an
alcoholic beverage distribution facility “ operated by 17
State governments and by some counties and small
municipalities in a few States.” “ Liquor store” expen­
ditures consist of purchases of “ beverages for resale and
provision and operation o f liquor stores. Excludes ex­
penditure for law enforcement and licensing activities,
which are classed under general expenditures.” 2
9
For employment statistics, Census limits its definition
to the “ administration and operation o f retail liquor
stores operated by State governments.” 3 Statistics col­
0
lected under this heading apparently include wholesale
as well as retail operations; many control States are
unable to separate licensing and enforcement personnel
from State store personnel. The Census employment
statistics do not include any local government person­
nel.3
1
The Distilled Spirits Council o f the United States
( d i s c u s ) collects and publishes statistics by State on
sales of spirits, wine and beer, gallons sold, revenue
generated, specific revenue source, and number of retail
outlets.3 Many statistical series date from the end of
2
Prohibition.
The National Alcoholic Beverage Control Associa­
tion (nabca), the association o f State control officials,
collected and published statistics on store operations in
its annual Yearbook from 1973 and 1977. These includ-

Utah. The Utah Liquor Control Commission
handles sales, licensing, and enforcement. There are
about 30 State stores and 74 agency stores which sell
spirits, wine, and beer.

Vermont. The Department of Liquor Control is
responsible for spirit, wine, and beer sales, as well
as licensing and enforcement. Both State-operated
stores and agency stores are used.
Virginia. The Virginia Alcoholic Beverage Control
Commission is responsible for sales, licensing, and
enforcement. About 260 State stores sell spirits, or
wine and spirits. Wine and beer are also sold
through private distributors.
Washington. The Washington Liquor Control
Board oversees alcoholic beverage sales, licensing,
and enforcement. At one time it operated its own
bottling plant. Today, it operates 6 warehouses and
about 140 retail stores, and monitors the operation
of another 180 agency outlets. These outlets are
managed by State employees—1 per outlet—who
share in the profit from their sales. Beer, wine, and
spirits are sold in State and agency stores.
Wyoming. The Wyoming Liquor Commission ad­
ministers licensing, enforcement, and wholesale
sales. Private vendors handle retail sales.
In addition to these 17 States, four States permit or
require local government sales. North Carolina requires
local government sales if alcohol beverages are sold for
use off-premise. Local governments operate about 225
county stores and 120 city stores in the State.
Maryland, Minnesota, and South Dakota permit
local option. Minnesota has a combination of private
and municipal liquor stores. As of October 1978, there
were 567 private and 113 municipal stores. South



2 Standard Industrial Classification Manual, p. 272.
8
2 Government Finances, p. 630.
9
3 Public Employment, p. 462.
0
3 Rough calculations suggest that local governments have about
1
2,500-3,000 employees in alcoholic beverages sales and operations.
3 See Public Revenues from Alcohol Beverages, Annual
2
Statistical Review, and Retail Outlets fo r the Sale o f Distilled Spirits,
all published annually by the Distilled Spirits Council of the United
States.
37

ed the total number of personnel, enforcement and
licensing facts, the number of State stores and store per­
sonnel, and revenue raised through State stores, n a b c a
ceased collecting and publishing these statistics in 1977.
N a b c a still collects and summarizes each month
statistics on the number of sales by case and by bottle
size in control States, d i s c u s uses these statistics to
compute and publish the number of gallons sold.
In addition, each control State publishes an annual
report on its operations, ranging from a few summary
pages of financial information to detailed reports on all
phases of the State’s operations, including the number
o f employees, gallons and cases sold, and financial in­
formation on each store (table 26). Two States, Idaho
and Michigan, present labor productivity statistics in
their annual reports.

that amount through any mix o f taxes and mark-ups.
Some States depend relatively heavily on mark-ups;
others on taxes. Since mark-ups usually can be changed
administratively whereas tax changes require legislation,
there is a tendency to adjust mark-ups rather than taxes.
Thus, in concept, mark-ups do not serve the same pur­
pose in control States as they do in the private sector.” 3
3

An additional problem, common to most revenue in­
dexes, is that price and revenue usually reflect changes
in input costs, including labor. If deflated revenue is us­
ed as the measure of output, it is necessary to adjust for
change in input price.
Prices of alcoholic beverages vary markedly by pro­
duct, but this variance seems unrelated to labor re­
quirements. Pennsylvania, for example, reported that
wine sales accounted for 40 percent of the gallons sold
in 1972, but only 17 percent o f the revenue generated.
Furthermore, sales have shifted from higher priced to
lower priced distilled spirits and from distilled spirits to
wine over the past several years.3
4
In addition, dollar sales must be adjusted to reflect
price changes, which vary according to whether wine or
spirits are examined. Between 1967 and 1978, wine
prices increased over 75 percent while spirit prices in­
creased less than 25 percent.
In short, dollar sales are a poor measure of State
alcoholic beverage operations for calculating labor pro­
ductivity.

Oytpyts
This section discusses how State alcoholic beverage
store outputs might be measured, and presents several
illustrative indexes for wholesale and retail sales. As in­
dicated earlier, many alcoholic beverage control
authorities are responsible for enforcement and licens­
ing as well as wholesale and retail sales. Enforcement
and licensing, which account for less than 10 percent of
beverage commission employees, would require a dif­
ferent set of output measures; they are not considered in
this review.
This investigation focuses on sales of spirits and
wines. Although several States sell beer, beer sales are
such a small part of State sales—less than 0.01
percent—they can be safely ignored in any output
calculations.
Sales o f alcoholic beverages are commonly measured
in one of five ways—dollars, customers, bottles, cases,
or gallons. A brief discussion of each measure follows.

Customers. The main virtue of using the number of
customers is that it is a physical measure. Also, it is easi­
ly understood and a repetitive unit.
It has two shortcomings, however. First, stores stock
and sell bottles or cases; the number of customers is on­
ly a proxy for the number of bottles or cases sold. Unit
labor requirements more closely correlate to bottles and
cases than to the number of customers. Second, data to
calculate this measure are not readily available, at least
nationally. Only one State report examined during this
investigation included data on the number o f customers.

Dollar sales. Dollar sales figures measure the final
organizational output, and are readily available, easily
understood, and repetitive. Every State collects dollar
sales data, and many publish them annually. Further­
more, the Census Bureau collects these statistics and has
published them annually since 1960, for each State and
in total.
The difficulty with using dollar sales as a measure of
output is that it is not a good measure of* the base-year
unit labor required to move and sell the product. This is
true for retail trade in general, where manufacturing
costs are a large part of the retail cost. It is particularly
true for alcoholic beverages, where taxes and mark-ups
are a significant part of the sales price. One expert
describes the situtation as follows:

Bottles. Probably the best measure of labor effort in
retail alcoholic beverage sales is the number o f bottles
sold. Most retail spirit and wine sales are made by the
bottle, and shelves are replenished bottle by bottle. In
addition, this measure is easily understood.
Several problems arise in using bottles as the measure
of output. First, bottles range from miniatures to 4
liters, and different size bottles require different
amounts of effort to stock and sell. The impact of these
differences is not known but it could be substantial.
Furthermore, the average bottle size has increased over
the past several years. Also, bottle statistics are not
published nationally, and only two States included the

“ Control States adopt a variety o f postures with regard
to earning revenues via taxes or mark-ups, depending on
enabling legislation. States are interested in generating a
certain amount o f revenue per gallon, and can arrive at




3 Personal communication from Steve L. Barsby, dated August
3
11, 1981.
3 Ibid.
4

38

Table 26. Selected data contained in State alcoholic beverage control annual reports
State

Dollar
sales

Alabama......................................................................
Idaho ..........................................................................
Iow a.............................................................................
Maine..........................................................................

X
X
X
X

Michigan ....................................................................

X

Mississippi..................................................................
Montana.......................................................................
New Hampshire..........................................................
O h io .............................................................................
Oregon.........................................................................
Pennsylvania..............................................................
U tah.............................................................................
Vermont.......................................................................
Virginia.........................................................................
W ashington................................................................
West V irginia..............................................................
W yom ing.....................................................................

X
X
X
X
X
X
X
X
X
X
X
X

Gallons
sold

Cases
sold

X

X

-

-

X
X

_

X

X

Employment

X
X
X

Bottles sold

X

Bottles sold
and number
of customers
Bottles sold
and bottles
sold per clerk
_

-

-

-

-

X

-

-

_
_

X

-

-

-

-

-

X
X
X
X

-

-

X
X
X
X

-

-

Other

X

_
X
X
X

-

_

_

_

_
_
_

_
_
_
_
-

Gallonage indexes. Retail gallonage spirit sales are
available by State and year from d i s c u s ; wine gallonage
sales are available by State and year from the Wine In­
stitute. Also, as noted, the majority of States publish
statistics on gallons sold. These three sources of data
have been used to prepare three illustrative output in­
dexes—all States, the five largest States, and wholesaleonly States (table 27).
Total State gallonage sales increased about 40 percent
between 1967 and 1978, an average annual increase of
3.1 percent. Gallons sold rose every year, but the rate of
increase dropped throughout the period: 4.3 percent
from 1967 to 1970, 3.3 percent from 1971 to 1974, and
2.0 percent from 1975 to 1976.
Gallon sales of the five largest control States—Penn­
sy lv a n ia , O h io , V irg in ia , W a sh in g to n , and
Alabama—increased about 30 percent, an average an­
nual increase of 2.3 percent between 1967 and 1978.
Like total sales, the rate of increase decreased
throughout the period.
G a llo n s so ld by the three w h o le sa le -o n ly
States—Michigan, Mississippi, and Wyoming—increas­
ed 52 percent between 1967 and 1978, an annual in­
crease of 3.7 percent.3 Gallons sold increased every year
6
but again the rate of increase decreased over the period.
Agent sales need to be removed from the retail sales
statistics insofar as possible because they create dif­
ficulties in productivity calculation. Ten of the 17 con­
trol States use agents (non-State employees) to sell
alcoholic beverages.
The procedure used to remove agent sales was to
calculate the ratio of dollar sales by agents to total
dollar sales and to apply that ratio to total gallons sold.
In 6 of the 10 States using agents, data were readily
available for this computation. In four States—Idaho,

statistics in their annual reports. However, these data
are available in unpublished form nationally.

Cases. Another commonly used measure o f output is
the number of cases sold. It is also easily understood.
For warehouse operations, it is probably better than
bottles as a measure of output and unit labor re­
quirements.
About half (8 of 17) of the State annual reports in­
cluded statistics on the number of cases sold. No
published national statistics were found but they are
available in unpublished form.
Gallons. Probably the most common physical output
measure is gallons. The trade associations calculate it
annually and more than half (9 of 17) o f the control
States carry the statistic in their annual reports. Com­
parable data are available for a number of years.
Although not as good a measure of unit labor re­
quirements as bottles or cases, it is probably an ade­
quate surrogate.
Preferred measure. Experts suggest that the best output
measure for computing State alcoholic beverage retail
productivity is the number of bottles sold. Probably the
best measure of wholesale operations is the number of
cases.
The most readily available physical statistic, however,
is gallons sold; bottle and case statistics are compiled for
spirits although they are not published. Bottles, cases,
and gallons are highly correlated, and output trends
constructed for these three measures would probably be
similar, at least in the short run. Data are not generally
separated between retail and wholesale movements.3
5

3
5
Statistics for Michigan, which publishes data on all three
measures, show a correlation of .91 between cases and bottles, .89 be­
3
6 Michigan is primarily a wholesale State; less than 1 percent of its
tween bottles and gallons, and .99 between cases and gallons for 1978.
gallonage sales are retail sales.



39

Table 27. Three gallonage indexes for alcoholic beverage
control operations, 1967-78

be satisfactory for trend analysis. A set of ratios is need­
ed to determine this.
The other principal source of employment data is the
State alcoholic beverage agencies themselves. As noted
earlier, all agencies publish annual reports; about half
include statistics on the number of staff by function.
Several include statistics on salaries, the number of
supervisors, and part-time and full-time employment.
However, none provides information on hours worked,
although several provide statistics on full-time
equivalency.
The primary problems in using State agency reports
to calculate a labor index are inconsistency through time
and incomplete information on coverage and methods
of derivation. Individual State data series will appear in
several annual reports, disappear for several years, and
then reappear. Sometimes yearend figures are
presented; other times full-time equivalency is included.

(1977=100)
All States

Five largest
States

Wholesaleonly States

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

73.3
76.3
80.0
82.9
86.1
89.9
94.2
94.6
97.4
98.5
100.0
103.4

97.5
82.8
86.3
87.8
90.2
93.2
97.9
97.2
99.9
99.4
100.0
102.4

69.1
71.8
76.4
80.9
82.7
89.5
92.4
93.9
94.8
97.7
100.0
105.0

Average annual percent
change ..............................

3.1

2.3

3.7

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

S o u r c e : State annual reports, unpublished data provided b y individual
control States, and published reports of Wine Institute and Distilled Spirits
Council of the United States.

Employment indexes. Labor indexes were prepared for
Montana, Utah, and West Virginia—agent sales data
were not readily available for the period examined and
the calculations could not be made. However, these four
States accounted for only about 8 percent of total State
sales; hence, including agent sales for these four States
is unlikely to affect the output index significantly.

the total number of employees, total full-timeequivalent employees, and total hours paid. State and
Census data were used to make the computations. In
each case, State license and enforcement personnel have
been removed from the figures.
The total number of employees, part time and full
time, was taken from unpublished Census Bureau
statistics and adjusted with State data when available
and appropriate. For example, the 8-percent jump in the
Census figure in 1973 was directly attributable to one
State, Pennsylvania. This statistic was at odds with data
provided directly by the State as part of this study, and
the Census statistics were adjusted accordingly. Census
employee counts have been available only since 1970.
Statistics for 1967-69 were taken from State data or
Census full-time-equivalent data when State statistics
were unavailable.
The total number of full-time-equivalent employees
was taken from published and unpublished State data,
augmented when necessary by Census data.
The third set of data is for hours paid for State
alcoholic beverage control employees.3 These data were
7
provided by the individual States or were calculated
from information provided by the State (average hours
worked) and the Bureau of the Census (full-timeequivalent employment). Pennsylvania, which employs
about 30 percent of all State and alcoholic beverage per­
sonnel, provided statistics on hours paid. Ten States
(about 40 percent of the employees) provided full-time
equivalent data and the average hours paid per
employee per year. The remaining six States (about 30
percent of the employees) were unable to provide the re­
quested data so Census data and average workweek
hours were used to estimate total paid hours.

Labor inputs and ©mployse eosts
The number of employees, number of full-timeequivalent employees, and number of hours are the in­
put measures suggested in the previous chapter to
calculate State and local government labor productivity.
The three sources of such data for alcoholic beverage
control are: (1) The Bureau of the Census, (2) individual
State authorities, and (3) the National Alcoholic
Beverage Control Association (nabca). Since NABCAno
longer collects and publishes these data, as noted
earlier, nabca statistics are of use only in constructing
historical time series.
The Census figures are collected annually as part of
the survey of government employment. This survey’s
overall strengths and weaknesses were discussed earlier.
However, the State alcoholic beverage control data have
several specific weaknesses. First, agencies apparently
report all their personnel, which usually include en­
forcement and licensing staff as well as store and
warehouse staff. One State, as noted earlier, assigns
drug control operations to its enforcement staff. To
match employment with sales, enforcement and licens­
ing personnel must be separated, which is not possible
with Census information. Second, State liquor stores
use part-time, intermittent, and seasonal employees ex­
tensively. Census statistics, collected each October, miss
the peak holiday (November and December) sales and
staffing period. If the ratio of October to holidayperiod sales remains constant, the October statistics will



3
7 No State could supply a time series of actual hours worked or
hours at work. Several States had limited recent data.

40

Table 30. October earnings for alcoholic beverage control
personnel, 1967-78

Tabs® 2®. Three employment indexes for aleoholic beverage
control operations, 1967-78
(1977 = 100)

Year
Number of
employees

Full-timeequivalent
employment

Hours paid

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

93.8
96.5
99.6
100.3
100.4
100.5
99.2
98.7
100.9
101.4
100.0
100.3

92.7
94.9
98.3
102.0
100.4
101.6
100.2
99.6
100.8
100.9
100.0
98.5

92.6
94.8
98.1
101.9
100.2
101.4
100.2
99.5
100.7
100.9
100.0
98.5

Average annual percent
change ..............................

.4

.4

.4

Index
(1977 = 100)

Dollars

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

S o u r c e : D o lla r s — P u b lic

annual

is s u e s ;

in d e x — c o m ­

Prodye^svity indexes
This section presents three productivity indexes
calculated from data presented and discussed in the
preceding sections. All use gallons as the output
measure and full-time-equivalent employment as the
measure of labor input. Separate indexes are presented
for the total, for the five largest States, and for the
States which operate wholesale-only operations.
The results of these calculations show that labor pro­
ductivity for total State alcoholic beverage control
operations increased 33 percent between 1967 and 1978,
or 2.7 percent annually (table 31). The rise was fairly
constant throughout the period.
The figures for the five largest States generally
parallel the statistics for all States, which is to be ex­
pected since they account for about 70 percent of all
employment. From 1967 to 1978, large-State productivi­
ty increased 22 percent, or 2.0 percent annually.
Productivity in the three wholesale-only States
—Michigan, Mississippi, and Wyoming—increased at a
much more rapid rate than in the control States as a
whole or in the large States. Labor productivity for the
1967-78 period increased 55 percent for full-timeequivalent employment, or 4.6 percent annually.
None of the three productivity indexes, including the
wholesale-only index, grew as rapidly during this period
as State employee earnings, which increased 7.3 percent

Table 29. Two full-time-equivalent employment indexes for
alcoholic beverage control operations, 1967-78
(1977 = 100)
Five
largest States

Wholesale-only
States

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

92.3
95.5
99.3
103.9
102.7
102.6
99.7
99.4
100.5
100.1
100.0
97.4

106.4
101.5
106.0
108.4
107.2
105.5
105.5
93.7
94.7
94.8
100.0
104.4

Average annual percent
change ..............................

.3

-.8

3
8
A comparison was also made between data presented here and
data without adjustments for enforcement and licensing personnel
and obvious errors. Unadjusted data rose 0.2 percent a year compared
with 0.4 percent presented here. The absolute numbers differed by
about 8 percent.

Bureau of the Census and State data.




E m p lo y m e n t ,

salaries and wages (excluding fringe benefits) was
calculated for 1967-78. These statistics, which are col­
lected and published by Census, are for October of each
year. They show that during 1967-78 salaries and wages
more than doubled, or increased 7.3 percent annually
(table 30).

Salaries and wages. The average annual increase in

So u r c e :

7.3

p u te d .

The total change and the- average annual rate of
change for the 1967-78 period were similar for the three
indexes (table 28). The number of employees increased
6.5 percent, full-time-equivalent employment increased
5.8 percent, and hours paid increased 5.9 percent. The
averge annual change was 0.4 percent for each index.3
8
Full-time-equivalent employment indexes were also
calculated for the five largest States and the wholesaleonly States for the 1967-78 period (table 29). The index
for the five largest States—Pennsylvania, Ohio,
Virginia, Washington, and Alabama—increased a total
of about 5 percent, or 0.3 percent annually. The three
wholesale-only States—Michigan, Mississippi, and
Wyoming—decreased about 2 percent, or 0.8 percent
annually.

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

52.0
53.7
55.4
62.2
61.6
67.7
77.5
83.0
86.2
94.3
100.0
108.6

Average annual percent
change ..............................

S O U R C E : Number of employees— 1967-69, State-provided data or
statistics published by Bureau of the Census; 1970-78, Bureau of the Census
computer printout. Full-time-equivalent employment— State and Census
data. Hours paid— State data or computed from State-provided data
augmented by Census data.

Year

$ 493
510
526
590
585
642
735
788
818
895
949 .
1,031

......................................
......................................
......................................
......................................
......................................
......................................
......................................
.................................... ,.
......................................
......................................
......................................
......................................

41

Table 31. Three productivity indexes for alcoholic beverage
control operations, 1967-78

U n e m p l o y m e n t Sr?suram©@

(1977 -100)
Total

Five largest
States

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

79.1
80.4
81.4
81.3
85.5
88.5
94.0
95.0
96.6
97.6
100.0
105.0

86.1
86.7
86.9
84.5
87.8
90.8
98.2
97.8
99.4
99.3
100.0
105.1

64.9
70.7
72.1
74.6
77.2
84.8
87.6
100.2
100.1
103.1
100.0
100.6

Average annual percent
change
......................

2.7

2.0

The Unemployment Insurance Service (uis) presents a
very different set of productivity measurement issues
from those encountered in electric utility operations or
alcoholic beverage control sales. The uis provides a serv­
ice without a market price, one for which demand fluc­
tuates greatly. It typifies State and local government in­
come maintenance programs in that it covers every State
and is jointly administered by the Federal and State
governments.

Wholesaleonly States

4.6

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

SnstituttosiaS considerations
Unemployment insurance was established by the
Social Security Act of 1935. The intent of the act, in­
sofar as unemployment insurance was concerned, was
to provide financial security for the majority of the Na­
tion’s workers during times of temporary unemploy­
ment.

SO URCE: C o m p u te d f r o m d a t a in t a b le s 2 7 , 2 8 , a n d 2 9 .

annually. The net result was an increase in unit salary
and wage cost.

Coverage. About 97 percent of all workers are covered
by unemployment insurance today. The only employees
not covered are the self-employed, domestic employees
earning less than $1,000 per quarter, and agricultural
workers who work on farms with 10 or fewer employees
which have a payroll of less than $20,000 per quarter.
The number of beneficiaries fluctuates substantially,
depending on economic conditions. In mid-1980, about
4.2 million individuals were drawing unemployment in­
surance benefits.3 In 1968, the figure was 1.1
9
million; in 1976, 3 million. In 1977, total paid
unemployment insurance benefits were about $13
billion; in 1980, they were almost $19 billion.

Conclusions and suggested research
Calculation of a national State alcoholic beverage
store productivity index should be a straightforward
operation. Although no national index exists today,
several States calculate their own labor productivity. All
States collect data that could be used as the base from
which to build a national productivity measure.
The preferred output measure has not been discussed
in the literature. Several State officials have suggested
that the number of bottles is the preferred retail measure
and the number of cases the preferred wholesale meas­
ure. The national data which are most readily available
are the number of gallons sold, which is a good proxy
for bottles and cases. Insofar as inputs are concerned,
Census collects and publishes annually full-timeequivalent employment statistics which can be used to
calculate labor trends.
Several illustrative productivity indexes were
calculated from available data. In general, they track
quite well and match intuitive judgment. The summary
index, which uses gallons as the output and full-timeequivalent employment as the input, shows labor pro­
ductivity increasing at an annual rate of 2.7 percent bet­
ween 1967 and 1978. Gallons sold increased 3.1 percent
annually while employment increased 0.4 percent an­
nually.
Topics for further research have been noted and
discussed. They include substituting bottles and cases
for gallons as the measure of output; separating
wholesale from retail operations; more accurately iden­
tifying and removing agent sales from the totals; further
analyzing the impact of licensing and investigation on
productivity; and including local government alcoholic
beverage operations in the index.



Program description, uis provides coverage through a
series of programs. In 1981, these included the regular
State, Federal, and veterans programs which provide up
to 26 weeks of benefits. States with high unemployment
provide another 13 weeks of benefits through the ex­
tended benefit program.
The individual States fund the regular State program.
The Federal Government covers the cost of the regular
Federal civilian and military programs. The extended
benefit program is financed jointly (50-50) by the
Federal and State governments. Each State sets its own
level of benefits. Each State also sets the waiting period,
if any, the qualifying wage, dependent allowances, and
other requirements.
In addition to the regular unemployment insurance
programs, several special programs provide assistance
under unique situations, such as natural disasters, trade
dislocations, and deregulation. These programs provide
stipends, in addition to normal unemployment benefits,
3
9 “Unemployment Insurance Claims” (Employment and Training
Administration, July 10, 1980).

42

Table 33. Unemployment insurance program benefits,
selected years, 1965-80

for relocation and retraining. They are funded by
general Federal tax receipts.
Uis programs are born in response to economic and
political conditions of the times. Between 1963 and
1980, 10 different uis programs were in operation (table
32). Some, such as the regular State program, operated
throughout the period. Others, such as the temporary
compensation program, operated for only a brief time.

( M illio n s )

Program
T o ta l..........................

1965

1970

1975

1980

$2,283

$4,158

$19,362

$18,790

2,166
50
67

3,847
76
199

13,239
158
387

14,486
129
294

34
1
1

2,492
44
75

1,697
14
30

Regular programs:
State ..............................
Federal..........................
Veterans........................

Size and scope. The size and the scope of the programs
vary substantially. Some are broad-based and operate in
every State. Others, such as the National Redwood Park
program, are tailored to a small geographic area or
clientele.
The importance of a program can vary considerably,
depending on the time period examined (table 33).
Trade readjustment allowance, for example, began as a
small program in the mid-1970’s but expanded rapidly
in the latter part of the decade. In 1975, $38 million of
benefits were paid; in 1980, $2.1 billion were paid. The
program has since been cut substantially.
The regular State program is the backbone of the uis.
It normally accounts for 90-95 percent of all benefit ex­
penditures, although it accounted for only 55 percent in
1976 when a number of temporary programs were in
force.

Extended benefits:
State ..............................
Federal..........................
Veterans........................

-

-

Federal supplemental
benefits:
State ..............................
Federal..........................
Veterans........................
Special unemployment
assistance..........................
Trade readjustment allowa n c e ..................................
Disaster unemployment
assistance..........................

-

-

-

-

2,133
59
55

-

-

-

670

-

-

-

38

2,138

“

3

2

-

SO URCE: U n p u b lis h e d d a t a f r o m U n e m p lo y m e n t I n s u r a n c e S e r v ic e , S e p t.

24, 1981.

uis programs. These include the regular State program,
the Federal employee program, the ex-serviceman pro­
gram, extended benefits, temporary compensation,
Federal supplemental benefits, and special unemploy­
ment assistance. Excluded from this study are special or
nontraditional programs, including disaster unemploy­
ment assistance, trade readjustment allowance, and the
National Redwood Park program, which each require a
unique set of output measures. Also, the railroad

unemployment insurance program, which is sometimes
included in unemployment insurance statistics and
discussions, is not included since it is not financed or ad­
ministered by uis or the State governments.
The regular uis program has two primary activities:
Making payments to unemployed workers, and collec­
ting money from employers. Making payments to
unemployed workers includes activities such as registra­
tion, establishment of eligibility, issuance of checks,
and hearing appeals. Collection o f funds from
employers includes monitoring and auditing employers’
contributions, auditing employers’ books, and captur­
ing money due from delinquent accounts.

Table 32. Unemployment insurance programs, 1963-80

Financing, uis programs are financed through three

Programs studied. This study focuses on the traditional

Program

sources—a Federal payroll tax, a State payroll tax, and
general Federal revenue. The Federal payroll tax, which
is set at 0.7 percent on the first $6,000 o f annual wages
paid to each employee, is levied on the employer. These
funds are used to administer the program, finance ex­
tended benefits, and maintain a reserve from which
States may borrow if their reserves are inadequate to
pay beneficiaries.
A State payroll tax is also assessed against each
employer. These assessments are used to pay regular
benefits. Each State establishes the tax level and the
base against which it is levied. Taxes currently range up
to 7.5 percent on the first $6,000-$10,000 of annual
wages paid to each employee.
The third source of financial support is general taxes.
These are used to finance Federal and veterans
unemployment insurance and special programs, such as

Period

Regular State program..............................

1935 to present

Unemployment compensation for
Federal employees..................................

1955 to present

Unemployment compensation for
ex-servicemen..........................................

1958 to present

Extended benefits......................................

1970 to present

Temporary compensation ........................

1972-73

Disaster unemployment assistance..........

1972 to present

Trade readjustment allowance..................

1972 to present

Federal supplemental benefits..................

1975-78

Special unemployment assistance . . . . . . .

1975-78

National Redwood Park............................

1978-84

S ource : Unemployment Compensation: Final Report, (National
Commission on Unemployment Compensation, 1980).




43

trade readjustment and disaster assistance.
Most uis funds are used to pay beneficiaries. Ad­
ministrative or operating expenditures account for only
5-10 percent of total expenditures, depending on the
year. About 80 percent o f the administrative funds are
for wages, salaries, and fringe benefits.4 The remain­
0
ing 20 percent are for items such as mail, computer lease
and purchase, and building rental and operation.

example, in 1981 an average of 53 minutes was required
to process an initial State claim (uc). An initial Federal
claim ( u c f e ) took twice as long, or 105 minutes. The
overall weighted average was about 60 minutes since the
majority o f claims are State claims.
The time per unit also varies by State. One State pro­
cessed its initial claims on the average in 38 minutes.
Another State required 67 minutes.
Uis uses such statistics, updated annually, to set State
ui budgets and allocate funds.
For the purposes of this study, cost-model dollar and
time expenditures are grouped into three basic func­
tions: Beneficiary payments, finance operations, and
support (table 35). Beneficiary payments, which ac­
count for about 75 percent of the labor time input, in­
clude activities such as initial claims, weeks claimed,
and appeals. Finance or tax functions account for about
17 percent. Support or overhead accounts for about 26
percent o f the labor input.

Definitions. “ Unemployment insurance” is the term
most commonly used when discussing the program.
“ Unemployment compensation” and “ unemployment
assistance” are also used, although in a strict technical
sense the terms are quite different. The three terms are
used interchangeably here.
Neither the Census of Governments nor the Standard
Industrial Classification (sic) system separately iden­
tifies the uis program. Thus, it is impossible to rely on
either of the two systems most often used to collect and
categorize State and local government statistics. Alter­
native sources of data will be discussed later.
The Census of Governments includes the uis program
under the general category of Income Security. The sic
Manual assigns the program to Industry 9441, “ Ad­
ministration of Social, Manpower, and Income
Maintenance Programs.” 4 The program is lumped
1
with equal employment opportunity offices, public
welfare administration, and workers’ compensation of­
fices. The major group title is, “ Administration of
Human Resources Programs.” Local employment ser­
vice offices are assigned to Industry 7361, Employment
Agencies, which are part of Business Services, Major
Group 73.

Other studies. The cost-model work has spawned a
number o f internal uis studies which have particular
relevance for productivity computation.4 Some of
2
the more important findings include:
1. The time to process a claim varied significant­
ly, depending on the workload mix. A veteran or
Federal claim took significantly longer to process
than did a regular intrastate claim.
2. Internal office procedures substantially in­
fluenced unit processing time.
3. Larger States required less unit labor.
4. The more heavily urbanized a State, the more
efficient its uis operations.

R@ areh and sflatisflies
s@

5. The greater the percent of employers (not
employees) added to and subtracted from the rolls,
the greater the unit labor requirements.

Surprisingly little formal research has been published
on unemployment insurance operations in view of the
size of the program and the massive amounts of data
collected and published. Most published research has
focused on actuarial issues, such as financial solvency,
and program issues, such as the impact o f benefit levels
on the willingness to work.

6. Unit labor requirements differed greatly from
State to State for the same functions. For initial
claims, unit labor requirements varied by 368 per­
cent.
7. A number o f procedural changes would im­
prove UIS productivity, including more extensive
use of computers and paying recipients biweekly.

The cost model, uis has funded some research on ad­
ministrative issues. Most notable is the development of
a cost model to establish budget needs, allocate money
to the States, undertake cost comparisons, and identify
cost-effective procedures.
The cost model divides the ui process into six ac­
tivities: Initial claims, weeks claimed, nonmonetary
determinations, appeals, wage records, and tax func­
tions (table 34). The labor required for each activity
varies, sometimes significantly, by workload mix. For

Statistics. The Federal Government, as the primary fund­
ing and coordinating agency,

4
2 “ Development and Utilization of the Cost Model Management
System in the Unemployment Insurance Program” (Unemployment
Insurance Service, May 1979), pp. 40, 76-87; “ Report on the Analysis
of Initial Claims and Wage Record Activities by the Operational Im­
provement and Cost Equalization Project” (Unemployment In­
surance Service, 1979), p. 4; and Millions Can be Saved by Improving
the Productivity o f State and Local Governments Administering
Federal Income Maintenance Assistance Programs (General Accoun­
ting Office, June 5, 1981).

4 Unemployment Compensation: Final Report (National Commis­
0
sion on Unemployment Compensation, 1980), p. 128.
4 Standard Industrial Classification Manual, 1972, p. 340.
1




requires dozens of

44

Table 34. Synopsis of unemployment insurance activities

Activity

Synopsis

Activity

Initial claims.................... The process whereby individuals file applications for unemployment in­
surance. Initial claim activities include
completion and review of the applica­
tion, monetary determinations, and
monetary redeterminations. The
number of initial claims directly
reflects uis coverage and the rate of
unemployment. Between 1963 and
1979, the number of initial claims filed
ranged from 10 million (1969) to 30
million (1975 and 1976).
Weeks claim ed................

Nonmonetary determinations..........................

The number of weeks for which unemploy­
ment insurance payment is requested
by those filing claims. The tasks under
this heading include certification of
eligibility, periodic client interviews to
verify job search, and the processing
and paying of benefits (i.e., writing
checks). The number of weeks claimed
is somewhat larger than the weeks com­
pensated (or checks written) since
about 20 percent of the claims are
disallowed. Between 1963 and 1979,
the number of weeks claimed ranged
from 55 million (1969) to 292 million
(1976).
E xam inations, determ inations, and
redeterminations of whether an in­
dividual is eligible to draw benefits.
Determinations result from a protest
by an interested party, such as a former

routine reports from the States and, in turn, prepares a
number of summary reports. State governments, as
operating agencies, collect statistics from their own of­
fices and contributing employers, many of which they
summarize, analyze, and publish.
The basis for most of these data is the State Employ­
ment Security Agency ( s e s a ) accounting system. Over
40 SESA reports are prepared routinely in five general
categories—time distribution, property, appropriation,
general ledger, and activity expenses.4 In addition to
3
the accounting information and reports, there are about
two dozen program reports and one dozen financial
reports.4
4
Specific output and input statistics are discussed in
the following two sections.

employer or employee, or from new
information such as a new Federal or
State statute or court ruling. Common
reasons for determinations include
refusal to accept an offer of work and
nonavailability for work. There were
about 10 million nonmonetary deter­
minations and redeterminations in 1979.

Appeals..........................

Processes that give the former employee or
employer the right to challenge a claim
ruling. All States have hearing officers or
referees to hear appeals, investigate the
circumstances, and hand down written
decisions. In fiscal 1979, there were
about one million appeals—lower,
higher, intrastate, and interstate.

Wage records................

Processes required to maintain and update
employee earnings files. Computerized
records are maintained by 38 States and
the District of Columbia. Normally, uis
updates wage records quarterly, uis
measures the workload by the number of
updated wage records.

Tax functions................

All activities involved in collecting money
from employers to support uis beneficiary payments. Activities include new
account registry and discovery, accoun­
ting, auditing, and delinquent account
follow-up. The work in this area has
grown steadily since 1963.

The following discussion briefly reviews seven possible
output measures:
Number o f employees covered
Number o f employers covered
Number o f beneficiaries
Number of compensation weeks
Composite benefit/finance index
Composite program index
Composite functional index.

Number o f employees covered. The number of
employees covered is simply a count of the people who
have earned wages in jobs covered by unemployment in­
surance. This type of measure is often used by the
private insurance industry; their measure o f output is
the number of policies sold. The arguments for using
the number of employees covered as the output measure
for uis include the following: (1) It is measurable, (2) it
has been calculated for a number of years, (3) it is a
physical measure, (4) it is easily understood, and (5) it is
supported by good data.
The primary argument against using this measure is
that it is not a measure of the final product of the uis.
There is apparently little correlation between the

Outputs
Selection of the preferred uis output measure is not as
straightforward as it might first appear. No research on
the question was uncovered during this investigation.

4 “ S esa Accounting System Accounting Manual—Report
3
Utilization Guide” (Employment and Training Administration, Oc­
tober 1978).
4 Summary o f Employment Security Statistical Reports-August
4
1977 (Employment and Training Administration, 1977).



Nonmonetary determinations—Continued

Synopsis

45

takes into account the length of time unemployment in­
surance is drawn, as well as the number of people draw­
ing insurance, is the number of compensation or
beneficiary weeks. This measure is the number o f people
drawing unemployment insurance each week during a
given period, such as a month or a year.
Variations o f the number of compensation weeks
which would produce time series that parallel the basic
measure are: (1) Average weekly number of bene­
ficiaries for a year, or the total divided by 52; and (2) the
average weekly insured unemployed, which includes all
persons reporting at least 1 week of unemployment dur­
ing the reporting period. A person may report being
unemployed but be refused unemployment insurance or
drop out of the program before receiving a check.
The number o f compensation weeks is sometimes sug­
gested as the output measure for unemployment in­
surance. It is measurable, repetitive, accurate, com­
parable through time, and is not affected by different
State unemployment insurance levels. Also, it is an easi­
ly understood, unitary measure, and data exist to
calculate an output index for a number of years. The
primary argument against this measure is that compen­
sation is only part of uis operations; the other part,
financing the program, is not covered by this measure.
Despite its faults, the number o f compensation weeks
has been used to calculate an output index. These com­
putations show that in 1963 uis paid about 86 million
weeks of compensation to the unemployed. The figure
dropped to 49 million in 1969, jumped to 260 million in
1976, and declined to 109 million by 1979. The average
annual change between 1963 and 1979 was 7.1 percent.

Table 35. Distribution of time expended by Unemployment
Insurance Service function, fiscal year 1979
Function

Percent

Total....................................................

100

Benefits..........................................................
Initial claims ..........................................
Weeks claimed ......................................
Nonmonetary..........................................
Appeals ...................................................
Wage re c o rd s ........................................
Finance...........................................................
S upport...........................................................

57
17
19
11
8
2
17
26

So u r c e :

Unpublished

d a ta

from Unemployment Insurance Service.

number of persons covered and the resources required
to operate the uis.

Number o f employers covered. The number of
employers covered is simply a count of the businesses,
firms, and organizations which have one or more in­
dividuals who are covered by unemployment insurance.
It could be divided by size of firm, type of firm, or other
characteristics. General Motors, New York City, and
the State of California are examples of covered
employers. Arguments for and against using the number
of employers as the uis output measure are essentially
the same as those noted in the preceding discussion.
In the finance function of uis, the number of
employers and unit labor requirements should be closely
related. This avenue will be explored further in a later
section.

Number o f beneficiaries. The number of beneficiaries is
the number of individuals who draw unemployment
benefits during a given period. The basic measure has
several variations: The number o f different persons (one
person may draw unemployment benefits several times
during a year); the number of different times a person
is assisted during a year; and the number of claims.
Claims and beneficiaries, though closely related, are not
synonymous; individuals can file claims but do not
become beneficiaries until the claim is approved and a
check is written.
All these outputs are related to work performed, are
measurable, repetitive, and easily understood. Also,
data are available to calculate an index.
A criticism of this measure is its failure to take into
account the length of time an individual draws
unemployment insurance, and this time does vary. Dur­
ing periods of high unemployment, the average time in­
creases as tasks associated with the maintenance o f a
person on unemployment compensation (check writing,
recertifications, and appeals) increase. This measure of
output does not consider these additional labor re­
quirements.

Benefit/finance index. A good case can be made for
dividing uis outputs into two distinct parts. One would
focus on service to the unemployed, that is, those apply­
ing for and drawing benefits. The other would focus on
finance operations.
Service to the unemployed is a function o f the number
applying for and drawing ui. One measure of
beneficiary output is the number of compensation
weeks paid, already discussed. A better measure would
take into account the different unit labor requirements
needed to process those applying for and those drawing
ui. Separate indexes have been calculated for each. Be­
tween 1963 and 1979 the average annual percent change
in compensation weeks was 7.1 percent, as already
noted, and for initial claims, 4.6 percent.
Finance operations, the second part, have the objec­
tive of ensuring the integrity of employer tax payments.
These operations are a function of the number of
employers, not the number unemployed. If uis benefits
and operations were funded from general revenue, the
tax collection function and the supporting State staff
would not exist.

Number o f compensation weeks. A measure which



46

The output measure suggested for finance operations
is the number of employers. It is tangible and
straightforward, and accurate data exist to make the
calculations. The measure is not subdivided by size and
type of business, since base-year labor requirements are
not readily available for such a division.
Finance output has increased fairly constantly since
1963. The average annual increase between 1963 and
1979 was 4.8 percent.
The three indexes that make up the benefit/finance
index have been combined using annual labor weights.
This index shows an average annual growth of 5.4 per­
cent between 1963 and 1979.

Table 3®. F@ur output Indexes for th® Unemployment
Insurance SerwSe®, fiscal years 1983=79
(1977 = 100)

Year

Function
(activity)

Benefit/
finance

Program

1963
1964
1965
1966
1967

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

45.3
42.2
35.5
28.0
27.4

48.1
46.7
42.6
38.6
38.8

53.2
50.5
46.6
42.4
43.7

54.3
51.6
47.5
43.2
44.5

1968
1969
1970
1971
1972

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

28.2
26.0
33.9
58.0
61.7

38.6
37.8
42.5
57.5
59.1

43.2
42.0
48.9
62.3
64.7

44.0
42.7
49.8
63.0
65.5

1973
1974
1975
1976
1977

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

46.7
48.9
102.0
137.4
100.0

56.9
60.5
90.5
108.7
100.0

60.2
64.9
105.6
121.2
100.0

60.9
65.7
106.2
119.6
100.0

1978 ......................................
1979 ......................................

68.0
57.8

84.9
80.6

79.4
76.1

80.2
77.4

Average annual percent
change:
1963-79 ........................
1966-78 ........................
1966-79 ........................
1969-79 ........................
1973-79 ........................

Program index. An even better measure of uis output
would be one that maintained the separation between
benefit and finance but took into account the dif­
ferences in program unit labor requirements in the base
year for initial claims, the area where unit labor re­
quirements vary by program. As indicated earlier, in
1981, the time required to process the basic State initial
claim was 53 minutes; the unemployed Federal worker,
105 minutes; and the unemployed veteran, 70 minutes.
An index weighted by prograpi was calculated. Other
than for the initial claim output, the index is the same as
the benefit/finance index presented earlier. That is, the
index has three outputs—initial claims weighted by pro­
gram unit labor requirements, weeks compensated
(unweighted), and tax (unweighted). These three output
indexes are combined by using labor weights to arrive at
a single output index. The average annual increase in the
program index between 1963 and 1979 was 5.3 percent.

7.1
13.3
10.9
10.3
4.7

6.2
9.7
8.7
9.4
6.7

5.4
8.8
7.5

5.3
8.6

7.7

7.6
3.9

3.8

7.4

SO URCE: Computed from unpublished data provided by the Unemploy­
ment Insurance Service.

most rapidly (7.1 percent per year), followed by the
function index (6.2 percent), the benefit/finance index
(5.4 percent), and the program index (5.3 percent) (table
36). In other words, there is very little difference bet­
ween the program index and the benefit/finance index,
and not a great deal of difference between these two and
the function (activity) index. The compensation week
index grew at a higher rate than the other three; because
finance activities are not explicitly considered and no
provision is made for separating intake and continuing
payments, it is not considered further.
The rate of growth of the program, benefit/finance,
and function indexes varies depending on the period ex­
amined. Rates of growth have been calculated for five
different periods. They are:

Function (activity) index. Another approach to measur­
ing uis output is to focus on activities or functions. The
six basic functions were described at some length earlier
in this paper. Separate indexes were calculated for each
function, which have, in turn, been combined by labor
weights to form a single index. The functions were not
weighted by program.
The primary arguments in favor of using a functional
index are its familiarity and the ready availability of the
data. The data should be reasonably accurate and com­
parable through time. The primary argument against
the functional index is that it measures functions and ac­
tivities in several instances, not final outputs.
The functional index shows an average annual rate of
change of 6.2 percent between 1963 and 1979.

1. 1963-79—the entire period,
2. 1966-78—the period covered by the productivity
index presented later in this study,
3. 1966-79—three uis cycles,

Comparison o f output indexes. Seven measures of ui
output have been discussed in this section. Indexes have
been calculated for four measures: Compensation
weeks, functions (activities), benefit/finance, and pro­
gram.
Comparison of trends for the four indexes for
1963-79 shows that the compensation week iijdex grew



Com­
pensa­
tion
weeks

4. 1969-79—two uis cycles, and
5. 1973-79—one uis cycle.
Four points stand out in examining the rates of
change in these different periods. First, the period can
dramatically affect the rate of growth of the three in­
dexes. In fact, the rate of change depends more on the
47

period chosen than the index chosen.
Second, there is very little difference between the rate
of change of the benefit/finance and program indexes,
regardless of the period.
Third, the benefit/finance and program indexes
generally move in concert with the function index. The
function index generally increases at a more rapid rate.
Fourth, the cyclical highs and lows generally fall in
the same years for each of the three indexes.
To summarize, the benefit/finance, program, and
function indexes move in much the same manner, and
there is virtually no difference between the first two.The
program index is preferable from the conceptual stand­
point.

d ire c tly co n cern th is in v e s tig a tio n . O th er
characteristics, such as the error rate, could affect unit
labor requirements through time and thus need to be ex­
amined. Assessment of the impact of quality change on
productivity trends requires two kinds of informa­
tion—a time series on quality and the relationship bet­
ween the quality characteristic and unit base-year labor
requirements.
Since 1975, the uis has measured quality in ten areas,
and has established minimum standards (desired levels
of achievement) for 9 of the 10 (table 37). Data are col­
lected, tabulated, and published by State for 35 dif­
ferent measures. States conduct self-appraisals for two
consecutive years. Every third year, staff from the
Federal Government and other States conduct ap­
praisals. With this information it is possible to trace
(since fiscal 1977) how uis quality is changing by
State.4
5
This study examined trends for 7 of the 35 quality
measures: Timeliness of higher and lower appeals,
promptness of intrastate and interstate payments,
promptness of status determinations, percent delin­
quent employers, and percent employers audited. Data
on appeals are available since the 1930ss; data on the re­
maining areas, since the mid-1970’s.
These seven time series show the following trends:
Two series—timeliness of high-level appeals and of in­
trastate payments—show little change; three series
—timeliness of lower level appeals, promptness of in­
trastate payments, and promptness of status determina­
tions—show quality improvement; and two series—per­
cent deliquent employers and percent employers
audited—show quality deterioration.
Although these statistics suggest no general shift in
uis quality, the data are too incomplete to draw hard
and fast conclusions.
The four time series that include the 1974-75 reces­
sion do suggest that, during that period of increased
workload and increased production, the quality of serv­
ice may have dropped temporarily. Both the timeliness
of appeals (lower and higher) and promptness of
payments (intrastate and interstate) deteriorated during
the recession. After the recession, quality returned to its
previous level,, These brief dips in quality evidently did
not affect long-term productivity trends.
Uis has examined the relationship between quality
and workload on several occasions but has been unable
to associate shifts in quality with changes in unit labor
requirements and productivity. This may be due to in­
sufficient change in quality variables, insufficient data,
inadequate analytic techniques, or simply the absence of
such relationships. The impact on productivity of quali­
ty of uis service remains an area for future discussion
and analysis, particularly as additional time series data

Output data. Although there is no lack of data for com­
puting output indexes, several potential problems exist
in using the data. First, because of the cyclical nature of
uis operations, the period covered by outputs must
match that covered by inputs. The output information is
for the calendar year or the fiscal year, or both, depend­
ing on the source. Because the labor information in this
study is summarized by fiscal year, output information
and indexes are by fiscal year.
A special problem with federally collected fiscal year
data is how to handle the shift in the beginning of the
fiscal year in 1976 from July 1 to October 1. The indexes
presented here simply delete that period from both out­
puts and inputs.
Another potential problem is inconsistent coverage of
the statistics. Some output series include only State pro­
grams, others include State and Federal uis programs,
and some include State, Federal uis, and railroad
unemployment insurance.
Third, most ui statistics, including those presented
here, include the trust territories. An index of “ State
only” uis output could be computed with some addi­
tional effort. The trust territory statistics have not been
removed from the statistical series presented here since
they account for only about 0.05 percent of uis
resources and are unlikely to have much effect on pro­
ductivity calculations.
Quality o f service. The issue of quality of uis service is
important, complex, and often discussed. The uis has
collected statistics on quality for decades and today col­
lects statistics on 35 different quality variables, which it
summarizes and reports annually.
The question for this study is: How does quality af­
fect productivity measurement and, particularly, how
does quality change affect productivity trends? To af­
fect productivity trends, two conditions must exist: (1)
The quality attribute must affect base-year unit labor
weights and (2) quality must be changing. Some quality
characteristics, such as courtesy and helpfulness of uis
staff, are important to uis managers but probably do
not affect unit labor requirements and thus do not



4
5 “ Unemployment Insurance Quality Appraisal Results for FY
1981” (Employment and Training Administration, May 1981).
48

Table 37. Selected quality appraisal measures of the Unemployment Insurance Service

Area

Typical measure

Quality characteristic

Federal
standard

Initial claims ..................................

Performance (interstate, intrastate)
Promptness (interstate, intrastate)

Number of undetected issues per 100 cases
Percent payments within “x” days

Yes
Yes

Weeks claimed................................

Performance (interstate, intrastate)

Percent weeks claimed affected by undetected
issues

Yes

Nonmonetary determinations . . . .

Performance
Promptness

Percent acceptable errors
Percent issues resolved within “x” days

Yes
Yes

Combined wage claims..................

Promptness

Percent forms processed within “x” days

No

Percent appeals scoring 80 or above
Percent discussions issued within “ x” days

Yes
Yes

Appeals............................................ Performance
Promptness
Status determinations....................

Performance
Promptness

Percent acceptable cases
Percent determinations within 180 days

Yes
Yes

Employer accounts........................

Promptness

Percent of monies deposited within “x” days

Yes

Field audits......................................

Penetration

Percent employers audited

Yes

Report delinquency........................

Promptness

Percent reports delinquent

Yes

Collections......................................

Performance/promptness

Percent delinquent employers with payments with­
in 150 days

Yes

S ource : Adapted from data in “ Unemployment Insurance Quality
Appraisal Results for FY 1980” (Employment and Training Ad­
ministration, May 1980).

are collected and national indexes are calculated.
Indexes in this paper have not been adjusted to ac­
count for quality shifts. However, future adjustments
may be needed, particularly in fraud identification and
control, for which the uis has created separate ad­
ministrative units. Two quality measures—initial claims
performance and weeks claimed performance—should
reflect the increased emphasis that the U IS is giving to
this area.

counts are made of the total number of employees.
The employee figures used here are the number of uis
positions reported by State labor officials to the U.S.
Department of Labor. The statistics relate to positions
used, not positions budgeted. The trend computed from
the number of State positions should approximate an
hours trend if it were possible to compute that index.
However, it probably would differ from the trend of the
total number of employees since States use many inter­
mittent employees.

Labor inputs
The uis program is extremely labor intensive. About
80 percent of all administrative funds go to pay

Data on positions are available by State, by function,
and for the total uis since 1963. They show that the total
number of positions increased from about 35,000 in
1963 to 45,000 in 1979, or 27 percent (table 38). The
numbers fluctuate considerably, ranging from 26,567 in
1967 to 57,321 in 1976. The average annual rate of
change between 1963 and 1979 was 3.7 percent.
The next section o f this study examines three time
periods: 1963-79, 1966-78, and 1972-79. The average
annual rate of growth of positions in these three periods
was 3.7, 6.6, and 5.6 percent, respectively.
These statistics cover the basic unemployment in­
surance program, including the regular State program,
Federal and veterans programs, extended benefits, sup­
plemental benefits, special unemployment assistance,
and temporary compensation. Positions for trade read­
justment allowance and disaster unemployment
assistance have been removed from the totals since these
two programs are not included in the outputs. The
number of positions dealing with disaster is extremely

employee salaries and benefits. All operating personnel
are State employees; no local employees are involved.
About 200 uis Federal employees oversee and coor­
dinate State activities, but they are not included in this
study. In any case, they constitute less than 0.05 percent
o f all uis employment.
Three labor measures are recommended for
calculating State and local government labor produc­
tivity—all employees, all employee hours, and number
of full-time-equivalent employees (see chapter III).
The only nationwide uis employee statistic routinely
collected is the number of positions—analogous to the
number of full-time-equivalent employees. The Federal
Government uses these statistics, which are available by
activity (e.g., initial claims, appeals, and tax) by State,
to budget for uis programs, allocate funds to the States,
and account for funds allocated. Statistics on hours
worked are collected by State but not for the Nation. No



49

Table 38. Positions in the State Unemployment Insurance
Service, fiscal years 1963-79
Year

N um ber

In selecting time periods, it is preferable to focus on
complete cycles, that is, from trough to trough or peak
to peak or midpoint to midpoint. The 1963-79 troughs
were 1966, 1969, 1972, and 1978; the peaks were 1964,
1967, 1971, and 1975. The period examined here is
1966-78, which includes three cycles.
Between 1966 and 1978, outputs increased at an
average annual rate o f 8.6 percent; inputs increased at a
rate of 6.6 percent; and output per employee at 1.9 per­
cent.

In d e x
(1 9 7 7 -1 0 0 )

1963

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

3 5 ,1 4 6

1964

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

3 2 ,9 4 6

6 3 .7
5 9 .7

1965

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

3 0 ,9 2 1

56 .1

1966

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

2 8 ,4 8 4

5 1 .6

1967

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

2 6 ,5 6 7

4 8 .2

1968

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

2 7 ,4 8 3

4 9 .8

1969

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

2 7 ,0 6 5

4 9 .1
5 1 .7

1970

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

2 8 ,4 8 9

1971

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

3 2 ,7 2 0

5 9 .3

1972

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

3 7 ,7 9 9

6 8 .5

1973

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

3 5 ,1 3 7

6 3 .7

1974

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

3 2 ,7 1 1

5 9 .3

1975

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

4 4 ,5 2 8

8 0 .7

1976

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

5 7 ,3 2 1

1 0 3 .9

1977

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

5 5 ,1 5 2

1 0 0 .0

1978

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

4 8 ,2 0 5

8 7 .4

1979

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

4 4 ,5 4 6

8 0 .8

Output per employee (productivity) increases rapidly
during periods of increasing .work for uis (higher
unemployment) and falls during periods of decreasing
work (falling unemployment). For example, during the
1974-75 recession, output per employee increased 34
percent. In the subsequent two years it fell 26 percent.
Apparently it is difficult to add staff as rapidly as the
work increases. Conversely, management is reluctant to
reduce staffing as rapidly as the work diminishes. This
pattern is like that found in the private sector.
Two other points should be noted: For each of the
three cycles, the relative increase in output was larger
than the relative increase in employees—hence the in­
crease in productivity. In addition, the longer the period
examined, the more stable the index and the less impor­
tant the individual cyclical fluctuations. As additional
years are added to the index, the results should become
more and more stable; however, results will be influenc­
ed by the magnitude of the fluctuations.

A v e ra g e a n n u a l p e rc e n t
change:
1 9 6 3 -7 9

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

1 9 6 6 -7 8

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

-

6 .6

1 9 7 2 -7 9

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

-

5 .6

3 .7

N o t e : E x c lu d e s d is a s t e r a n d t r a d e r e a d ju s t m e n t p e r s o n n e l.
SO URCE: Unpublished data from Unemployment Insurance Service except
for fiscal 1975, which was taken from the fiscal 1977 Federal Budget, Appen­
dix, p. 517.

small, usually two or three per year, and would not
make any marked difference in productivity calcula­
tions. However, the positions supporting trade readjust­
ment could have a significant effect if included. In the
mid-1970’s, there were fewer than 100 such positions; in
1980, the figure had climbed to almost 2,000. With 1981
Federal legislation, the number has dropped dramatically.

Table 39. Indexes of output, employee positions, and
output per employee position, Unemployment Insurance
Service, fiscal years 1963-79
(1977 =-100)

Productivity indexes
This section presents national uis productivity trends
drawing on the output and input indexes from preceding
sections. Productivity trends for six States are compared
with the national index. (All average annual rates of
change are based on the linear least squares trend of the
logarithms of the index numbers.)

Employee
positions

Output per
employee
position

1963
1964
1965
1966
1967

54.3
51.6
47.5
43.2
44.5

63.7
59.7
56.1
51.6
48.2

85.2
86.4
84.7
83.7
92.3

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

44.0
42.7
49.8
63.0
65.5

49.8
49.1
51.7
59.3
68.5

88.4
87.0
96.3
106.2
95.6

1973
1974
1975
1976

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

60.9
65.7
106.2
119.6

63.7
59.3
80.7
103.9

95.6
110.8
131.6
115.1

1977 ......................................
1978 ......................................
1979 ......................................

100.0
80.2
77.4

100.0
87.4
80.8

100.0
91.8
95.8

Average annual percent
change:
1963-79 ..........................
1966-78 ..........................

least from the conceptual standpoint, is the programbased index. Between 1963 and 1979, program output
increased at an average annual rate of 5.3 percent. Dur­
ing the same period, employee positions increased at an
annual rate of 3.7 percent and output per employee, or
productivity, increased at a rate of 1.5 percent (table
39). The average annual rate of change is extremely sen­
sitive to the period examined, as has been noted earlier.
For example, productivity increased at an annual
average rate of 7.2 percent between 1972 and 1976 but
decreased at a rate of 0.7 percent between 1972 and
1979.

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

1968
1969
1970
1971
1972

National index. The preferred measure of output, at




Output

Year

5.3
8.6

3.7
6.6

1.5
1.9

So u r c e :

50

Tables

36

and

38.

State trends. Additional insights can be gleaned by ex­

national index presented earlier by the same procedures
used to compute the six State indexes. That is, the new
or truncated national index uses the shortened time
period (1972-79), the benefit/finance output measure,
and the 1972 base year.
The new national index does mirror the national in­
dex presented earlier: The average annual trend in out­
put per employee is exactly the same for the two indexes
(-0.7 percent) for the 1972-79 period even though in­
dividual years do not match precisely.
In four States—Virginia, Kentucky, Illinois, and
Alabama—productivity grew faster than in the total
uis. In two States—Pennsylvania and Texas—produc­
tivity declined. Both outputs and employees increased in
the total uis and in each of the six States. Outputs in­
creased 4.9 percent and employees 5.6 percent.
Employee growth was smaller for the Nation than in
any of the six States examined, and output was smaller
than in five o f the six States.
These statistics are presented for illustrative purposes
only. The truncated national index is not representative
of long-term trends. Also, it is doubtful that the State
trends presented here are representative of long-term
State trends.

amining individual State trends, which often move quite
differently from national trends. Some differences are
due to economic conditions and institutional situations,
but many reflect management processes and pro­
cedures. Management processes probably account for
the greatest differences in uis productivity.
Productivity trends have been computed for six States
selected to display a range of institutional arrangements
and productivity trends-—Alabama, Illinois, Kentucky,
Pennsylvania, Texas, and Virginia. They are il­
lustrative, not representative, of the States.
The State indexes differ from the national index in
several respects. First, labor data to compute the pro­
gram index were not readily available, so the
benefit/finance measure, which parallels the program
index quite closely, was used for output. Second, the
base index year was shifted to 1972 for ease of com­
parison. Third, and most important, the time period
covered was restricted to 1972-79. State data were not
readily available before fiscal 1972.
The results of these computations show that the
average annual change in output per employee ranged
from 4.7 percent for Virginia to -3.1 percent for Texas,
as follows:

Sugg©st@d research

Annual percent
change, 1972-79
Virginia......................................
Kentucky..................................
Illinois........................................
Alabama....................................
Pennsylvania............................
Texas..........................................

A number of extensions to the work presented here
are possible and desirable. Probably the most useful ex­
tension would be to compute trends for each State by
methods similar to those presented for the six States.
This would provide each State a means with which to
compare its progress. The uis routinely collects data
which could be used to make such calculations.
Other potentially useful extensions would be the
preparation of an index of the number of uis employees
and comparisons of the index of positions with an index
of hours paid. Additional research is needed into quality
factors and their possible impact on unit labor re­
quirements, particularly as additional quality data
become available. Also, it should be possible with little
additional effort to separate trust territory statistics
from State statistics. Finally, the uis is a service which
lends itself to the computation of absolute levels of pro­
ductivity in contrast to the productivity trends examined
here. The feasibility and implications of making such
calculations should be addressed.

+ 4.7
+ 2.6
+1.9
+1.2
- 1.7
- 3.1

These trends generally support the conclusions

drawn from the national UIS productivity data. First,
those States with the most rapid growth in output have
the highest rate of productivity growth. Second, the in­
crease in employee input generally follows the increase
in output. And third, the employee growth peak does
not generally reach the peak in output. The exception in
this case was Texas.

Comparison o f State and national indexes. A national
productivity index can be useful as a norm or bench­
mark for assessing State productivity progress. To
make the comparison, it was necessary to recompute the




51

Chapter V. Thinking Afb®ut
the Unmeasured: Four
Gas® Studies

Institutional ©©nsId@
rati<o>ms

This chapter examines the possibility o f calculating
productivity indexes for four services—solid waste col­
lection and disposal, drinking water supply, mass tran­
sit, and the Employment Service. Each service presents
a different set of issues and problems. Some are simple
problems of data availability, some are conceptual prob­
lems, and some combine conceptual and data prob­
lems. In no case has a productivity index been cal­
culated.
These four services were selected because of past con­
ceptual and analytical work in the area. Each illustrates
some of the problems encountered in measuring State
and local government productivity, but should not be
construed as being representative of all the problems
that might be encountered.
The investigative approach for each service is similar
to that used in the preceding chapter: Institutional con­
siderations and past research are reviewed; possible
measures o f service output are examined, the preferred
measure is selected, and data needed to compute the
measure are reviewed; labor input issues and data are
discussed; and recommendations for future research
and data collection are presented.

Although the Federal Government and the States set
standards, local government is responsible for day-today operations o f solid waste collection and disposal.
Most local governments are deeply involved as service
providers or as regulators. Private firms and voluntary
organizations are also active in the solid waste field.
Few cities are served by a single provider. One investiga­
tion identified 12 different types of solid waste collec­
tion, ranging from total private to total public opera­
tions.1 A combination of providers is the norm for most
jurisdictions; for example, private service for commer­
cial establishments, government service for private
residences, and recycling by voluntary associations.
Local governments collect solid waste from about
two-thirds of the Nation’s residents. They serve the
residents o f most medium- and large-size cities.
Private firms serve the other one-third o f the Nation’s
residents, but two-thirds o f the Nation’s cities; private
companies serve most small cities.2 According to a 1966
survey, private firms collected about 36 percent o f the
Nation’s residential solid waste.3
Voluntary associations, such as the Boy Scouts and
neighborhood recycling centers, are primarily concern­
ed with resale items, such as newspapers, bottles, and
aluminum cans. These organizations, which play a
relatively minor role in this field, are not considered fur­
ther.

Solid W aste C ollection and Disposal
Sanitation service, particularly solid waste collection,
is the government service which is used most often to il­
lustrate productivity m easurement issues. Its
straightforward nature, tangible outputs, private sector
counterparts, and past research make it a useful exam­
ple for analysis and examination. Even though the
analytic and institutional base of knowledge underlying
the service is fairly large, no productivity index, or even
time series data with which to build one, has been pro­
duced.

Expenditures. This discussion focuses on local govern­
ment, which spent $2.3 billion on solid waste collection
and disposal operations or about 1 percent o f all State
and local government expenditures in 1977 (table 40).
Approximately 133,000 local workers or 1 percent of all
State and local government employees were in this field.
For the average city, solid waste collection and disposal
constitute about 9 percent of its budget.4
1 E. S. Savas, The Organization and Efficiency o f Solid Waste
Collection (Lexington, Mass.: Lexington Books, 1977), p. 29.
2 Ibid., pp. 63-64.
3 R. J. Black and others, 1968 National Survey o f Community
Solid Waste Practices: An Interim Report (Cincinnati: U.S. Bureau of
Solid Waste Management, 1968).
4 E. S. Savas and Barbara J. Stevens, “ Solid Waste Collection,”
in George J. Washnis, ed., Productivity Improvement Handbook
(New York: John Wiley, 1981), p. 592.

N o te : This chapter has profited from the comments of Charles
Ardolini, Gary Burdette, Richard Carnes, and Arthur Young of the
Bureau of Labor Statistics: Robert Clark of the Environmental Pro­
tection Agency; George Craft of the American Water Works Associa­
tion; John Greiner and Harry Hatry of The Urban Institute; Norman
Paulhus of the Department of Transportation; and Allan Stevens of
the Bureau of the Census.




52

Table 40. Local government solid waste collection and
disposal expenditures, fiscal year 1977

Solid waste services, as a group, are fairly labor inten­
sive. Salaries and wages account for about 52 percent of
all sanitation expenditures and 58 percent o f current
operating expenditures (table 40). Inclusion of fringe
benefits would raise the latter figure to about 65 per­
cent. A 1970 study found that labor-related costs ac­
counted for 88 percent of all solid waste expenditures.5
A 1975 study of Connecticut cities found that 73 percent
of collection costs were for labor.6 A more recent study
estimated salaries, wages, and fringe benefits at 69 per­
cent.7
Solid waste operations are commonly divided be­
tween collection and disposal. One 1968 study estimated
that 85 percent o f funding supported collection, 15 per­
cent supported disposal. Another study, at about the
same time, estimated the ratio at 80-20.8 A breakdown
o f Detroit’s solid waste expenditures in 1969 showed
that 69 percent went for residential collection, 6 percent
for street cleaning, 4 percent for snow removal, and 22
percent for disposal operations.
For most governments, solid waste collection is
residential collection. Collection may also encompass
street sweeping, dead animal pickup, abandoned vehicle
removal, leaf collection, snow removal, and bulky waste
pickup (e.g., stoves and refrigerators). Some govern­
ments also collect commercial and industrial waste. At
one time most jurisdictions collected garbage and trash
separately; today they are usually combined.
Solid waste disposal is primarily disposal at a sanitary
landfill or a dump. A 1966 survey found that cities with
a population o f over 25,000 used landfill mostly (79 per­
cent), incineration some of the time (20 percent), and
composting infrequently (1 percent). A 1967 survey
found that 92 percent o f the solid waste was disposed of
at landfills and 8 percent was burned at incinerators.9 A
1974 survey found that landfills accounted for 85 per­
cent of the disposal operations.1 Resource recovery and
0
recycling are much discussed but apparently play a small
role in municipal disposal operations.

Type of expenditure
Total....................................................

$2,374

Current operation ..........................................
Salaries and w ages................................
O th e r.......................................................
Capital outlay...................................................
Construction..........................................
O th e r.......................................................

2,126
1,239
888
247
129
119

Note : Because of rounding, detail does not add to totals.

1977 Census of Governments—Compendium of Government
Finances (Bureau of the Census, 1979), p. 30.
Source :

which are used by others include the following:
The Bureau o f the Census uses the term “ sanitation
services,” which it defines as follows:
“Refuse collection and disposal, operation of sanitary
landfills and street cleaning activities.” 1
1
Snow and ice removal are not included.

The Standard Industrial Classification Manual
assigns solid waste systems to one o f two codes, depen­
ding on whether disposal is part o f t h e process. Code
4212 is used for collection without storage; code 4953 is
for refuse with disposal. The precise definitions are the
following:1
2
Code 4212 Local trucking without storage:
“Establishments primarily engaged in furnishing
trucking or transfer services without storage, in a
single municipality, contiguous municipalities, or a
municipality and its suburban areas.”
Code 4953 Refuse systems:
“Establishments primarily engaged in the collection
and disposal of refuse by processing or destruction. In­
cludes:
Acid waste, collection and disposal
Ashes, collection and disposal
Dead animal disposal
Dumps, operation of
Garbage: collecting, destroying, and
processing
Incinerator operation
Radioactive waste materials, disposal
Refuse systems
Rubbish collection and disposal
Street refuse systems
Waste materials, disposal at sea.”

Definitions. A variety o f terms are used to describe
sanitary services, including solid waste, refuse, trash,
rubbish, and garbage. Solid waste is used most often to­
day. This report uses them interchangeably. Definitions

5 National Center for Resource Recovery, Municipal Solid Waste
Collection (Lexington, Mass.: Lexington Books, 1973), p. 42.
6 Peter Kemper and John M. Quigley, The Economics o f Refuse
Collection (Cambridge, Mass.: Ballinger Publishing Company, 1976),
p. 109.
7 Columbia University Graduate School of Business, Evaluating
Residential Refuse Collection Costs: A Framework fo r Local Govern­
ment (Washington: Public Technology, Inc., 1978), p. 18.
8 Black, 1968 National Survey.
9 Fred N. Rubel, Incineration o f Solid Wastes (Park Ridge, N.J.:
Noyes Data Corporation, 1974).
1 Eileen Berenyi, “ Solid Waste Disposal,” In George J. Washnis,
0
ed., Productivity Improvement Handbook (New York: John Wiley,
1981), p. 629.



Dollars (millions)

The American Public Works Association ( a p w a )
defines solid waste as garbage, rubbish, ashes, bulky
wastes, street refuse, dead animals, abandoned vehicles,
1 1977 Census o f Governments—Compendium o f Public Employ­
1
ment (Bureau of the Census, 1979), p. 461.
1 Standard Industrial Classification Manual (Office of Manage­
2
ment and Budget, 1972), pp. 224 and 238.
53

Table 41. Refuse materials by kind, composition, and source
Kind of refuse

Composition

Garbage..................

Wastes from preparation, cooking, and serving of food; market
wastes from handling, storage, and sale of produce

R ubbish.................

Combustible: Paper, cartons, boxes, barrels, wood, excelsior, tree
branches, yard trimmings, wood furniture, bedding, dunnage

Source

Households, restaurants, institutions,
stores, markets

Noncombustible: Metal cans, metal furniture, dirt, glass, crock­
ery, materials
Ashes .....................

Residue from fires used for cooking and heating and from on-site
incineration

Street refuse............

Sweepings, dirt, leaves, catch basin dirt, contents of litter re­
ceptacles

Dead animals..........

Cats, dogs, hordes, cows

Abandoned vehicles

Unwanted cars and trucks left on public property

Industrial wastes . .

Food processing wastes, boiler house cinders, lumber scraps,
metal scraps, shavings

Factories, power plants

Demolition wastes.

Lumber, pipes, brick masonry, and other construction materials
from razed buildings and other structures

Demolition sites to be used for new
buildings, renewal projects, express­
ways

Construction wastes

Scrap lumber, pipe, other construction materials

New construction, remodeling

Special wastes ........

Hazardous solids and liquids: Explosives, pathological wastes,
radioactive materials

Households, hotels, hospitals, institu­
tions, stores, industry

Sewage treatment .

Solids from coarse screening and from grit chambers; septic tank
sludge

Sewage treatment plants; septic tanks

Streets, sidewalks, alleys, vacant lots

Source : American Public Works Association, Municipal Refuse
Disposal (Chicago: Public Administration Service, 1970), p. 13.

construction and demolition wastes, industrial refuse,
special wastes, animal and agricultural wastes, and
sewage treatment residues (table 41.)
The following discussion includes all solid waste col­
lection activities—whether identified as trash or gar­
bage, residential or commercial, street sweeping, dead
animal removal, or leaf pickup—and all government
disposal operations. The crucial factor for this study is
that local government personnel provide the service.

amount of analysis of recycling and resource recovery
issues.
Research on collection falls into three general
categories—technical, geographic, and policy. Many
studies have dealt with technical issues such as routing
procedures, track size, crew size, aging of equipment,
use of transfer stations, and crew incentive systems and
their relationship to unit costs and productivity. These
factors are not the concern of this study although they
obviously affect labor productivity.
Geographic and community considerations, such as
precipitation, topography, collection density, and com­
position of solid waste also affect collection productivi­
ty. For example, studies show that costs for cities which
have hilly terrain average about 15 percent more than
costs for cities which are relatively flat.1 Manage­
4
ment has little or no control over geographic factors,
which are not considered further here.

R®s®ar©Bi and ©@
ffi©®pty©l ossy®s
Considerable research has been done on sanitation
services, particularly solid waste collection. Much of it
focuses on technical issues such as crew size, truck
routing, work rales, vector control, and equipmentlabor mix. Also, a modest amount of research has been
done on broad technical and economic issues such as
cost functions, quality, effectiveness, and economies of
scale.1
3
By contrast, little research has been done on the
economics of disposal. Most disposal research has
focused on environmental concerns, with a modest

1
4
Johns Hopkins University, Mathematical Modeling o f Solid
Waste Collection Policies, Public Health Publication 2030 (U.S.
Public Health Service, 1970); Ronald A. Perkins, “ Satellite Vehicle
Systems for Solid Waste Collection, Evaluation and Application”
(Cincinnati: U.S. Environmental Protection Agency, 1971); and
University of California, An Analysis o f Refuse Collection and
Sanitary Landfill Disposal (Berkeley: University of California, 1952).

1 Savas, Organization and Efficiency, pp. 107-110.
3



54

Local government officials control policy issues, such
as the level of service, which are subject to modifica­
tion, vary from jurisdiction to jurisdiction, and affect
unit labor requirements. The point and frequency of
pickup are two important policy issues which affect out­
put and need to be considered in calculating productivi­
ty. Shifts from curb or alley to backyard pickup can in­
crease collection costs from 25 to 100 percent, according
to studies (table 42). Shifting pickup of residential solid
waste from once to twice a week will increase costs from
19 percent to 74 percent.
Economies of scale of operations can be important in
constructing a productivity index. Evidence is conflict­
ing for solid waste. Several studies have not found any
economies of scale1 others have found some.1 One of
3;
6
the most recent studies found increasing economies
(decreasing unit costs) in cities with a population of
under 20,000; cities with a population of 20,000-50,000
probably had decreasing costs, although the evidence
was not conclusive; and cities with a population of more
than 50,000 showed no evidence of changing unit cost.1
7
A review of six different studies in this area concluded
that economies occur in the smaller cities but not in
medium-size or larger jurisdictions.1
8
There is general agreement that major economies of
scale exist in landfill operations, although few studies
have specifically addressed the issue. One found large
decreases in unit costs until the 100,000 tons per year
figure was reached (residential population of about
100,000).1 Beyond this point, the costs continued to
9
decrease but at a lesser rate.
The amount of economic research on solid waste has
declined rapidly since the mid=197Q5 today, little is be­
s;
ing done outside environmental concerns. However, the
analytic groundwork has been well laid for productivity
calculations.
Several points stand out in the research:
1. The measures of output used most frequently are
tons, cubic yards, and residents served.
2. The input measures most often used are labor
hours or number of employees.
3. Almost all quantitative research has been cross
sectional.
4. Calculation of a solid waste productivity index
needs to take into account level-of-service issues,
such as frequency of service and point of collec­
tion. These areas have experienced major 1
9
7
6
*
3
1 Werner Z. Hirsch, “ Cost Functions of an Urban Government
3
Service: Refuse Collection,” Review o f Economics and Statistics,
February 1965, pp. 87-92.
1 Dennis Young, How Shall We' Collect the Garbage?
6
(Washington: The Urban Institute, 1972).
1 Columbia University, Evaluating Collection Costs, p. 58.
7
1 William F. Fox, Size Economies in Local Government Services: A
0
Review (U.S. Department of Agriculture, 1980), p. 24.
1 Thomas I. Sorg and H. Lanier Hickman, Sanitary Landfill Facts
9
(U.S. Department of Health, Education, and Welfare, 1970).




55

Table 42. Studies ©f effect of location and frequency of
collection on solid waste collection costs
Investigator or author

Basis of
comparison

Change in cost
(percent)

Location: Shift from
curb/alley to backdoor
or backdoor to curb/alley
Hirsch....................................
University of California..........
Citizens Budget ( N Y C ) .................
Rawn......................................
Ralph Stone............................
Flintoff/M illard........................
Clark/Gillean..........................
Kemper/Quigley....................
Savas ....................................

Statistical
Statistical and
time/motion
study
?
?
Statistical
?
Statistical
Statistical
Statistical

60
26-48
37
30-45
100
40
25-38

Statistical
?
Statistical and
simulation
Time/motion
Statistical
Statistical

67
50
50
74
50
50
19-25

100
65

Frequency: Shift from
once to twice weekly or
twice to once weekly
University of California..........
Los Angeles County ..............
Perkins (dollars) ....................
(hours)......................
Rogers/Bellenger..................
Kemper/Quigley....................
Savas ....................................

S o u r c e : C o m p ile d f r o m r e f e r e n c e s lis te d in b ib lio g r a p h y .

shifts over the past two decades, and such shifts
are likely to continue.

Outputs
P ro d u c tiv ity m easures sh o u ld re fle c t an
organization’s physical output. The measure most often
used to compute solid waste output is tons collected and
disposed. Other measures frequently used are cubic
yards collected and disposed and residential or
household units served. Private collectors often base
their charges on the number of containers emptied. The
relative strengths and weaknesses of the different
measures are seldom discussed. The following examines
these issues, quality considerations, and the data which
could be used to calculate a national index.
Solid waste collection. Tonnage is the measure most
often used to track and analyze solid waste collection
output. This is a good measure of work performed, par­
ticularly in residential and commercial collection, and it
is familiar to the industry. Also, it is a physical unit not
affected by price level changes. Most large governments
keep tonnage statistics.
Two reasons sometimes given for not using tonnage
as the measure of output are: 1) It is not a particular­
ly good measure of work performed for some of the
secondary sanitary services such as street sweeping,
snow removal, leaf collection, and dead animal pickup.
These services, however, account for a small portion of
sanitation resources in most cities. 2) Tonnage statistics

are not collected by all governments, particularly
smaller ones.
Cubic yards is often used to measure output in
jurisdictions which do not weigh their refuse. Like ton­
nage, cubic yards is not affected by price change.
The major argument against using cubic yards as an
output measure is its instability or variability. That is,
most trash is compacted as it is collected, and the rates
of compaction differ markedly from jurisdiction to
jurisdiction, and even from truck to truck. Uncom­
pacted household wastes usually weigh 200-250 pounds
per cubic yard. Wastes compacted in a collection vehicle
usually weigh: 550-750 pounds per cubic yard, and land­
fill wastes are usually compacted at a density of
750-1,000 pounds per cubic yard.2 Compaction rates
0
directly affect resource requirements—fewer trips are
made to the landfill and thus fewer labor hours and
other resources are required. Furthermore, compaction
rates have improved over time, which would bias output
trends. The problem is further compounded by second­
ary services such as street sweeping and dead animal
pickup. In short, solid waste collection experts fed that
cubic yards is not a very good measure of output for
productivity calculations.
The number o f residential units served is used by
many jurisdictions as a measure of output. It is a
physical unit unaffected by price change, and data to
support this measure are normally available and should
be quite accurate. However, it is not a very good
measure of the work performed since volume and
weight differences are not reflected. Refuse generation
varies considerably by household, and the underlying
variables are not stable through time. That is, the
number of individuals per residential unit has decreased
while the amount of refuse generated per person has in­
creased. Also, some sanitation departments serve com­
mercial and industrial businesses as well as residential
units. Nor can the measure be used to track secondary
sanitation services such as street cleaning, leaf removal,
and dead animal pickup.
Finally, the number o f containers emptied is
sometimes used as a measure of output, particularly by
private collectors. However, it is not a very good
measure for local government. Containers vary in size
and weight, and average container size and weight have
changed through time. Also, most jurisdictions lack
statistics on the number of containers.

Evidence is skimpy on the level and rate of change of
solid waste collection. Cross-sectional studies conducted
over the past two decades suggest that the frequency of
service has remained fairly stable but pickup has shifted
from yard to alley or street. Calculations of productivity
trends need to take such shifts into consideration.
Two ways to handle a shift from backyard to curbside
collection are: 1) To include total resource inputs, not
just those provided by government. The time residents
take in setting out the solid waste and carrying back the
cans would need to be included in the computations. 2)
To differentiate the services, in this example backyard
and curbside pickup, by using labor weights or by parti­
tioning the data. This approach is recommended here.
Quality factors. Quality factors, such as missed col­
lections, spillage, collector noise, and damage to
residential property, are sometimes considered in pro­
ductivity discussions. As noted earlier, this study in­
cludes only attributes which affect unit labor re­
quirements. Since solid waste collection quality factors
probably would have little impact, no adjustment is sug­
gested here.

Level o f service. Level-of-service factors, such as
point of residential collection pickup and frequency of
pickup, can markedly affect unit costs and unit labor re­
quirements (table 42). If levels remain constant through
time, the point becomes moot for productivity trends. If
the levels are changing, they must be considered.
2
0
John Reindi, “ Interrelationships Within the Solid Wastes
System” Solid Waste Management, April 1977, p. 23.




56

Secondary services. In addition to traditional residen­
tial service, some sanitation departments remove aban­
doned automobiles, empty city park refuse cans, pick
up dead animals, collect leaves, and sweep streets. To
measure output, these services may be handled in
several ways.
One approach is to remove secondary service outputs
and inputs from productivity calculations. That is, the
tons of refuse from parks and tons of sweepings from
streets would not be included in outputs, and in­
dividuals who performed these services would not be in­
cluded in inputs. However, data are rarely available to
make such calculations.
Another way to handle secondary services is simply to
include them as part of the primary service. That is, no
distinction would be made between tons collected from
the park, on the street, or from residential units, and no
distinction would be made in labor requirements. The
rationale behind this approach is that most secondary
services account for only a small proportion of sanita­
tion department resources and are unlikely to affect
overall productivity irrespective of how they are handl­
ed. The argument against this approach is that some
secondary services, such as street cleaning, are impor­
tant in some cities and their productivity may change at
a different rate from that of residential collection. Also,
tonnage and cubic yards are not very good measures of
many secondary service outputs.
The preferred approach in measuring secondary serv­
ice productivity is to develop separate output and input
measures for each service, and to combine them with
proper weights to calculate the solid waste collection in-

minimally acceptable set. Both use tonnage as the basic
recommended measure o f output. The idealized set of
measures is presented first. Tasks as well as measures
are listed.
Collection
Residential—tons collected (small container
collection)
Once-a-week pickup
Backdoor
Curbside/alley
Two or more pickups per week
Backdoor
Curbside/alley
Commercial and school—tons collected (large
container pickup)
Qnce-a-week pickup
Two or more pickups per week
Secondary services
Abandoned car removal—cars removed
Bulk collection—tons collected
Dead animal removal—carcasses removed
Leaf removal—cubic yards removed
Street cleaning—curb miles cleaned
Disposal
Landfill—tons buried
Resource recovery—tons recovered
Other—tons disposed

dex. Examples of secondary services and suggested out­
puts are:

Abandoned cars - number of cars towed
number of carcasses collected
Dead animals
number of cubic yards
Leaves
removed
number of cans emptied
Refuse cans
Street cleaning - number of curb miles swept
The primary problem with this approach is data
availability.
Finally, some jurisdictions use sanitation equipment
and personnel for services normally assigned to other
areas. An example is snow removal, often included as
part of street operations. In computing a sanitation pro­
ductivity index, these services, both outputs and inputs,
should be removed from the calculations. If they con­
stitute a small part of total sanitation resources, they
can be ignored.

Solid waste disposal The factors that are most often us­
ed to measure disposal outputs are weight (tonnage) and
volume (cubic yards). The arguments advanced for and
against these measures are essentially the same as those
discussed earlier. The measure that is suggested for use
is weight (tonnage). To quote one expert:
“ In many respects weight data are of limited value; after
all, when we design a landfill we are concerned about
volume, not weight. However, weight is an easier
measurement to take since volume measures will depend
upon how much the wastes have been compressed, and
whether or not it is a dense material, such as foundry
sand, or a very light material, such as waste foam
rubber.” 2
1

Data are not always collected and maintained by local
government in the detail needed to compute the ideal­
ized measures. Furthermore, for reasons discussed
earlier, idealized measures may not yield a more ac­
curate productivity index than a less comprehensive set
of measures and data. Therefore, a set of minimum
measures- is presented below for which data should be
readily available in most local governments. All services
are included, but without the detailed product division
presented in the idealized measures.

Most solid waste is buried. Although precise figures
are lacking, about 80 percent of municipal solid waste is
disposed of in landfills. The remaining 20 percent goes
to resource recovery operations such as composting,
materials recovery, and energy recovery, or is dumped
into the ocean. Because unit labor requirements vary by
disposal process, and the mix between the various pro­
cesses may be changing, separation of different pro­
cesses is important in computing productivity trends.
Because disposal tonnage does not often equal collec­
tion tonnage, the two should be measured separately. In
most communities, the private collector dumps at the
community disposal site. In others, the sanitation
department may dump at private sites or at sites
operated by other governments. Regional disposal sites
are quite common.

Collection
Pickup—tons collected
Once-a-week pickup
Backdoor
Curbside/alley
Two or more pickups per week
Backdoor
Curbside/alley
Secondary services—tons collected
Disposal
Landfill—tons buried
All other—tons disposed
For most Jurisdictions, government collection ton­
nage and government disposal tonnage will differ
because of the use of multiple collectors—private and
nonprofit as well as government—and multiple disposal

Recommended measures. Two sets o f output measures
are presented here, one an idealized set and one a
2 Reindi, “ Interrelationships,” p. 23.
1



57

Table 43. Community surveys of solid waste collection

site operators—private and government. Rarely does a
government collect all the solid waste that is dumped at
its disposal site. Some governments do not even operate
disposal sites, and some that operate disposal sites do
not collect any solid waste with their own employees.
Because of the use of multiple suppliers, tonnage must
be separated between collection and disposal. Also,
statistics should indicate the point of pickup, frequency
o f collection, and type of secondary service.

Survey

Massachusetts Institute of
Technology..................................
Municipal In d e x..............................
American Public Works
Association (apwa) .......................

Number of
communities
responding

Response
rate
(percent)

?
?

1902
1929

161
667

Ap w a ................................................

1939
1955
1964
1968
1969
1973
1973

190
908
995
6,259
234
1,630
661

31
51

International City Management
Association ................................
Columbia University........................

1974
1975

1,092
1,377

48
100

Ap w a ................................................
Ap w a ................................................
Public Health Service......................
Ralph S to n e ....................................
Public Works ................................

Availability o f data. As noted earlier, data needed to
calculate a national index are not routinely collected. A
number of special studies have collected cross-sectional
data, some of which have focused on a small geographic
area; others have sampled jurisdictions throughout the
United States.
At least 11 surveys have been made of community
solid waste collection practices since 1900 (table 43).
The largest of these, at least in number of communities
contacted, was the 1968 Public Health Service survey.
This survey collected data from over 6,000 communities
located in 38 States and covered about 46 percent o f the
U.S. population. Data were collected on many aspects
o f solid waste activities including total tonnage or cubic
yards collected and disposed, number of abandoned
automobiles removed, and type of disposal activity.
Three States were asked to assess the quality of the in­
formation collected. They gave good marks to institu­
tional information but poor marks to tonnage and yard­
age statistics.
Probably the most comprehensive solid waste collec­
tion survey was one in 1975 by Columbia University,
which was a stratified random sample of over 1,300
jurisdictions. The sample was carefully chosen and the
survey rigorously administered. Data were collected
from each government on institutional arrangements;
type of financing; contractual forms; outputs, including
tons and cubic yards; and inputs, including costs, labor,
and type of equipment.
Most of the surveys leave much to be desired. The
response rates are poor, and definitions and cut-off
levels are unclear.2 None of the surveys, including the
2
Columbia study, collected time series data.
While national time series data are lacking to com­
pute a solid waste output index, most large and
medium-size jurisdictions do collect and record tonnage
data.2 Some communities, particularly the smaller com­
3
munities, do not weigh their solid waste, but many keep
statistics on the estimated number of cubic yards or the
number of trucks unloaded, statistics which can be used
to estimate tonnage. Although the percentage of local
governments that weigh solid waste is not known, the

?

38
32
?
?

Source : E.S. Savas, The Organization and Efficiency of Solid Waste
Collection (Lexington, Mass.: Lexington Books, 1977), p. 36.

consensus among experts is that most large governments
weight their waste, and hence that the major portion of
local government waste is weighted.

Labor inputs
Three labor measuures are suggested for calculating
local government labor productivity: Number of
employees, number of full-time-equivalent employees,
and employee hours. For sanitation, these data should
be available by function and subfunction—collection,
disposal, abandoned car removal, and so forth.
Only two sources of time series data exist on local
government sanitation employees: (1) The Bureau o f the
Census and (2) the individual municipalities.
The Census Bureau is the only organization that
routinely collects and publishes national time series data
on sanitation employment (table 44). These data show
128.000 local government employees in 1980, of whom
119.000 worked full time and 9,000 part time. Full-timeequivalent employees numbered 121,000. The peak
employment year was 1977.
A number of problems, in addition to those noted in
chapter III, result from using Census data. First, the
data are not separated by function or subfunction or
between primary and secondary tasks. Labor data need
to be divided at least between collection and disposal;
separation by frequency of collection and point of
pickup, and between landfill and other techniques of
disposal, also would be helpful.
Nor are Census data available for hours worked,
hours at work, or hours paid. Hours worked is impor­
tant for solid waste collection because some govern­
ments permit employees to go home as soon as they
have completed their assigned task. The task system
enables many workers to receive pay for 8 hours but to
work less time. If program operations remain constant
through time, then the task system would have no effect
on productivity trends—that is, the relationship bet­
ween the time paid and the time worked would remain

2 Savas, Organization and Efficiency, p. 36.
2
2 Harry P. Hatry and Donald M. Fisk, The Challenge o f Produc­
3
tivity Diversity; Measuring Solid Waste Collection Productivity, Part
II (Washington: The Urban Institute, 1972), pp. 11-12.



Date

58

Table 44. Local government sanitation employment, 1967-80

trends, can provide an important additional dimension
to productivity measurement. With a unitary measure
o f output, such as tons, computation o f productivity
levels should be possible, and would be helpful for in­
dividual government managers.

(Thousands)

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980

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

SO URCE:

Total employment

Full-time-equivalent
employment

125
125
127
133
128
125
130
130
128
127
133
131
132
128

Year

118
118
120
125
121
120
124
124
121
121
127
124
123
121

Drinking W ater Sypply
Water supply, like solid waste collection and disposal,
is a service with a tangible set of outputs and a welldefined set of inputs. The research community has ex­
amined this service in some depth and, unlike solid
waste services, some time series data have been col­
lected.

Institutional setting

P u b lic E m p lo y m e n t, a n n u a l is s u e s ( B u r e a u o f t h e C e n s u s ) .

Of the several hundred thousand public water systems
in the United States,2 about 60,000 are community
5
systems which serve 25 or more year-round residents or
which have 15 or more connections serving permanent
residents.2 Community water systems serve most of the
6
American public, an estimated 192 million in 1976.2
7
Most water utilities (76 percent) serve fewer than
1,000 people (table 45). However, the large systems pro­
vide most o f the drinking water in the United States.
Eight percent of the systems serve about 80 percent of
the population; at the other end of the scale, 80 percent
o f the systems serve 8 percent of the population.2
8
More systems are publicly owned than privately own­
ed (56 percent vs. 44 percent). Publicly operated systems
serve more than four-fifths o f the population (84 per­
cent vs. 16 percent).2 Publicly owned systems tend to be
9
located in large metropolitan areas; privately owned
systems in rural areas. Public systems tend to provide
more extensive water treatment than private systems,
which often provide no treatment.
Water utilities operated by local governments in every
State and the District of Columbia have over 120,000
employees. The average number of public employees
per State is about 2,200. California leads the list with
almost 15,000 employees and revenue of $766 million
(fiscal 1977). Vermont is last with 139 employees and $6
million in revenue.
In fiscal 1977, government water utilities spent about
$6.4 billion, or 2 percent of all State and local govern­
ment expenditures.3 Salaries and wages of water utility
0

constant from year to year. Such information is not
available.
Another problem with Census statistics is that they
are collected only for October of each year. Seasonal
help may be an important factor in sanitation services.
The seasonal variance in labor employed is not known
but one study estimated a 20-percent average variance in
tonnage collected.2 A variance of this magnitude would
4
suggest the need for additional labor input during in­
creased tonnage periods. If the October/seasonal pro­
portion remains constant, the October statistics would
be satisfactory for the determination of the trends. A set
of October/seasonal statistics is needed to determine
whether there is constancy.
The only other regular source of local government
sanitation employment statistics is local government
itself. Some sanitation agencies publish annual reports
which include employment statistics, and most govern­
ments include employment statistics in their annual
budgets, usually divided between collection and disposal
and often between primary and secondary collection
services.

Suggested research
Data collection is the next step in computing a sanita­
tion productivity index. Two separate approaches are
possible: 1) A census of all jurisdictions, and 2) a
stratified, random sample which might include all large
governments and a sample of medium-size and small
governments. The latter approach would be less expen­
sive and would still produce satisfactory information
for a national index. It would not permit examination of
small geographic areas (depending on the sample size).
Also, a procedure would have to be developed to collect
the data regularly if the index was to be kept current.
Analysis of productivity levels, as contrasted with

2 Small System Water Treatment Symposium (U.S. Environ­
5
mental Protection Agency, September 1979), p. 2.
2 The number of community water systems is open to debate. In
6
1975, the figure was estimated at 40,000; in 1976, at 35,000; in 1979 at
58,000; and in 1980 at 61,000. epa staff suggest that the most recent
figures may have some double counting.
2 Temple, Barker, and Sloane, Inc., Survey o f Operating and
7
Financial Characteristics o f Community Water Systems (U.S. En­
vironmental Protection Agency, April 1977), p. II-4.
2 Ibid.
8
2 Ibid., p. II-5.
9
3 1977 Census o f Goverments—Compendium o f Government
0
Finances (Bureau of the Census, 1979).

2 National Center for Resource Recovery, Municipal Solid Waste
4
Collection, p. 13.



59

Tab!® 45. Community water systems by size o
ff
population, 1979
Number of
systems

Percent of total

T o ta l..........................

58,379

100

Less than 1 0 0 ........................
100-999 ................................
1,000-4,999 ..........................
5,000-9,999 ..........................
10,000-100,000 ....................
More than 100,000 ................

21,468
22,907
9,221
1,915
2,599
269

The Standard Industrial Classification Manual
assigns water supply to Industry Group 494 and In­
dustry 4941. The definition presented in the Manual for
4941 is the following:

37
39
16
3
4
1

Population

So u r c e :

Small System Water Treatment Symposium,

“Establishments primarily engaged in distributing
water for sale for domestic, commercial, and industrial
use. Systems distributing water primarily for irrigation
service are classified in Industry 4971.” 3
4
Agencies that administer water quality, including
regulation, research, and planning, are “
-assigned to code
9511.

(U.S. En­

vironmental Protection Agency, Sept. 1979), p. 1.

workers accounted for about I percent of all State and
local government salary and wage expenditures; capital
expenditures accounted for about 4.5 percent of all
State and local government capital expenditures.
The primary factor input into water systems operated
by State and local governments is capital, which ac­
counts for 32 percent of all water supply expenditures
(fiscal 1977). Salaries and wages (without fringe
benefits) account for 22 percent (39 percent of current
operating expenditures).3 Interest on debt, materials
1
and supplies, energy, and purchased water account for
the remaining 46 percent (table 46).
Utility operations are frequently broken into four
functions—acquisition, treatment, distribution, and
overhead. Acquisition, normally a small part of the cost
of water, includes all operations before treatment, such
as storage and transport to the treatment facility. Treat­
ment, as the word implies, Includes any purification of
water before distribution, a relatively small part of utili­
ty operations. Distribution includes all operations after
treatment such as storage and transmission of water to
the ultimate customer. Overhead includes all ad­
ministrative and customer services required to manage a
utility. Overhead and distribution account for the
largest portion of utility expenditures.
Until 1980, Census public employment statistics
defined water supply as “ local government activities
associated with the production or acquisition and
distribution of water to the public.” 3 In other words,
2
employment statistics included only local government
employees. The Census finance statistics, on the other
hand, included State as well as local government expendditures.3 Exclusion of State employees before 1980 is
3
probably not important since they accounted for less
than 0.05 percent of total State and local government
salaries and wages for water supply workers (table 46).
Only three StMes—Massachusetts, Nevada, and New
Hampshire—operate utilities supplying drinking water.

Research
Considerable research has been done on drinking
water supply. Although most of this research has dealt
with the environment and .public health, economic
issues also have been investigated. Economic research
helps answer questions important for productivity
calculation, such as relationships among the factors of
production, the relationship between costs of produc­
tion and pricing, the role of size in production costs, the
proper measure of output, and the effect of quality
change on costs.
Several investigators have done research on produc­
tion and cost functions, but the results have not been en­
tirely satisfactory. In the 1960’s, Ford and Warford attemped to derive a total cost function for the water sup­
ply industry of England and Wales.3 Hines attempted a
5
similar analysis with data from Wisconsin, and God­
dard did likewise with data from Cincinnati.3 Although
6
the industry product was relatively homogeneous, Ford
and Warford noted that production conditions were
quite dissimilar. They concluded that either the dif­
ferences in technology or sources swamped the indepen­
dent variables or production costs had to be broken
down further.
Studies of individual parts of the production process
have explained somewhat more successfully cost varia­
tions among systems. Studies by Orlob and Lindorf,
Koenig, and Hinomoto were all reasonably successful in
explaining the differences in treatment costs.3 Orlob
7
and Lindorf studied cost of operation as a function of
average daily treatment (in millions of gallons) for 32
treatment facilities. Koenig expanded the methodology
of Orlob and Lindorf to examine specific treatment
costs. His correlation coefficient for the average cost
relationship was .77. Hinomoto, using Koenig’s data,
3 Standard Industrial Classification Manual, p. 238.
4
3 J.L. Ford and J.J. Warford, “ Cost Functions for the Water In­
5
dustry,” Journal o f Industrial Economics, November 1969, pp. 53-63.
3 Lawrence G. Hines, “The Long Run Cost Function of Water
6
Production for Selected Wisconsin Communities” Land Economics,
Vol. 45, February 1969, pp. 133-40; Haynes G. Goddard and others,
Planning Water Supply: Cost-Rate Differentials and Plumbing Per­
mits (Cincinnati: U.S. Environmental Protection Agency, 1978).
3 See Goddard, Planning Water Supply, pp. 20-27, for a descrip­
7
tion of these studies.

3 An e p a study of 12 large water utilities found that labor costs ac­
1
counted for 42 percent of the utilities’ operating cost. See James I.
Gillean and others, The Cost o f Water Supply and Water Utility
Management, Contract Report 68-03-2071 (Cincinnati: (U.S. En­
vironmental Protection Agency, 1977), p. 9.
3 Public Employment, p. 462.
2
3 Government Finances, p. 10.
3



60

Table 46. Finances of government-operated water utilities by type of government, fiscal year 1977
(Millions)
Expenditures
Government

Current operations

Revenue
Total

Capital

Interest
on debt

Total

Salaries
and wages

Other

Total ......................................

$4,995

$6,381

$2,047

$786

$3,547

$1,395

$2,152

States ................................................
Municipalities....................................
Special districts..................................
Townships..........................................
Counties............................................

6
3,823
772
139
256

21
4,306
1,413
187
453

3
1,261
518
44
221

6
475
206
20
79

13
2,570
688
124
153

8
1,085
194
47
63

5
1,485
494
77
90

Note: Because of rounding, detail may not add to totals.

Source: Government

Finances, p.33.

estimated unit and total costs for a plant of a given
capacity operating at a given rate. His estimating equa­
tions used seven resource inputs: Capital, chemicals,
pumping energy, heating energy, labor, maintenance
and repair, and miscellaneous. None of these three
studies explicitly considered economies o f scale which
were evident in the derived equations.
A number of studies have been done on economies of
scale in drinking water supply. A study by Clark of
water treatment costs notes that the unit cost water
treatment curve changes dramatically between the
ranges o f 0 - 2.5 million gallons per day (MGD), 2.5 - 20
m g d , and more than 20 m g d .38 Another study by Clark
found that in 42 municipal utilities large-system unit
costs were about half those of small systems. Although
statistics were not presented for the relationship be­
tween labor and output, he noted that they would
parallel the decreasing cost relationship.3
9
An e p a study of 70 investor-owned utilities showed
economies of scale as measured by unit costs and the
number of employees for the small and medium-size,
but not large systems.4
0
William Fox examined five studies of economies of
scale in water utilities.4 Each showed increasing
1
economies. Two studies used the quantity of water sold
as the measure of output. Another used the number of
users. The other two studies used both quantity and
customers/population.
The water supply industry is extremely capital inten­
sive as measured by the investment per dollar of
revenue. According to one study, water utilities require
$6-10 of capital investment for each dollar of revenue.

Comparable figures for airlines, railroads, telephone
companies, and electric utilities are $1, $2, $3, and $4
respectively.4
2
Although there is no national index of water supply
productivity, Clark has calculated labor efficiency
measures for a sample of 12 utilities.4 Using revenue
3
gallons as the output measure, he found that over a
10-year period labor hours per million gallons decreased
(an increase in labor efficiency) but not sufficiently to
offset increasing labor costs. The net result was that
dollar cost per million gallons increased.
Water quality has long been an important issue, and
the amount of capital investment is partly a function of
the treatment required to ensure a safe water supply.
Operating cost, including labor input, is also a function
of the amount of treatment required . The exact relation­
ships depend on a number of factors such as the type
and amount of pollutants in the influent, difficulty in
removing the pollutants, alternative sources of supply,
and available technologies to remove the pollutants.
One analysis found that removing five different types of
pollutants increased unit operating costs from 1 to 700
percent depending on the process used.4 The impor­
4
tance of labor was not specifically considered. In
another study, examination of an exchange process to
remove nitrates from a utility’s water supply showed
that labor cost accounted for about half the increase in
annual operating cost.4 A study o f 67 water utilities in
5
the Cincinnati area showed that increasing the quality of
water to a “ good” level could substantially affect the
costs of production for many utilities, particularly
smaller ones.4
6

3 Robert M. Clark, “ Small Water Systems: Role of Technology,’’
8
Journal o f the Environmental Engineering Division, American Socie­
ty of Civil Engineers, February 1980, pp. 19-35.
3 Robert M. Clark, “ The Safe Drinking Water Act: Its Implica­
9
tions for Planning,” in David Holz and Scott Sebastian eds.,
Municipal Water Systems: A Challenge fo r Urban Resource Manage­
ment (Bloomington: Indiana University Press, 1978), pp. 117-37.
4 “ Comparisons of Cost, Manpower Utilization, and Flow in
0
Operation and Maintenance of Investor Owned Water Companies and
Municipal Waste Water Systems” (U.S. Environmental Protection
Agency, June 5, 1979), unpublished report.
4 Fox, Size Economies, pp. 26-30.
1

4 Temple, Barker, and Sloane, Community Water Systems, p. II-8.
2
4 Robert M. Clark, “ Labor Wage Rates, Productivity, and the
3
Cost of Water Supply” Journal o f the American Water Works
Association, July 1979, pp. 364-68.
4 Clark, “ Small Water Systems,” pp. 26-27.
4
4 Robert M. Clark, “Water Supply Regionalization: A Critical
5
Evaluation,” Journal o f the Water Resources Planning and Manage­
ment Division, American Society of Civil Engineers, September 1979,
p. 286.
4 Robert M. Clark and Haynes G. Goddard, “ Cost and Quality of
6
Water Supply” Journal o f the American Water Works Association,
January 1977, pp. 13-15.




61

Several important conclusions for productivity
measurement emerge from the research:
1. Production conditions are dissimilar.
2. There are marked economies of scale in the
production process, particularly in treatment.
3. The industry is capital intensive, but labor
plays an important role.
4. The function has a relatively homogeneous
product, although water quality factors differ­
entiate the product.

O u tp u ts
Four common measures o f water utility output, one
monetary and three physical, are: Sales adjusted for
price changes, number of customers, number of connec­
tions, and number of gallons. The strengths and
weaknesses of each measure are briefly enumerated
below.

Dollar sales. The primary virtue of dollar sales is its
ready availability. It is the only output measure for
which national data are available for a number of years
(table 47). The Bureau of the Census collects and
reports these data as part of its annual survey of govern­
ment finances.
A primary problem in using dollar sales as a measure
of output is adjusting for price level changes. The price
deflator most often used is the joint water-sewerage
deflator of the Consumer Price Index, which is not en­
tirely satisfactory for adjusting drinking water sales.
First, the CPI reflects changes in sewerage as well as
water supply and it is not known whether water and
sewerage price changes move at the same rate. Second,
governments turn to price increases to discourage the
use of water and to increase utility profits. Whether
deflated revenue any longer adequately reflects changes
in physical output is debatable. Third, no adjustments
are made for quality changes for either water or
sewerage, and both have improved greatly since the Safe

(Millions)

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980

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

Unadjusted
$2,187
2,313
2,464
2,687
2,980
3,165
3,463
3,712
4,142
4,463
4,995
5,512
6,250
6,766

Deflated1
$4,575
4,617
4,606
4,665
4,671
4,781
4,954
5,016
5,095
4,948
4,995
4,952
5,365
5,426

Statistical Abstract of the United States,

purchased by consumers. Revenue gallons are similar
but not identical to gallons treated, gallons pumped, or
gallons produced, which are sometimes used as output
measures. Revenue gallons do not include water
pumped but lost through leaky water mains and open
hydrants. Revenue water reflects actual sales.

selected years

(Bureau of the Census).




number of people served and assumes that the average
quantity of water used per person remains constant.
Research shows a good correlation between the water
used and the number of people served. Population
figures are often used to plan the amount of water a
community will need. This statistic is kept or can readily
be produced by most water utilities (number of residen­
tial connections times average household population).
Most national surveys of water utilities collect these
data.
There are two arguments against using this as a
measure o f output. First, most water utilities serve com­
mercial and industrial users as well as residential
customers. If the community has a large number of
nonresidential users, or a few that consume a large
quantity of water, population is not a particularly good
measure o f output. A recent study estimated that
residential units made up 90 percent of the billings of
U.S. water systems but accounted for only 60 percent of
the water delivered. If the proportion of residential to
nonresidential units remained constant through time,
the measure would be satisfactory for trend determina­
tions, but the ratio is not known. Considerable pressure
has been exerted in recent years for industrial and com­
mercial users to restrict their use of water through
recycling.
The second argument against this measure of output
is that sometimes population does not correlate well,
particularly over the short run, with the quantity of
water used. Variations in temperature and rainfall will
affect water usage, and a population measure will not
reflect these changes.

Revenue gallons. This measure is the number of gallons

1Water/sewerage index of the Consumer Price Index used to deflate
revenue dollars; 1977 ^base year.
So u r ce:

Population served. This measure simply counts the

Connections. The number of connections is the number
of residential, industrial, and commercial hookups. It is
a surrogate measure for the water produced and
delivered to users. This measure of output, which has
many of the same strengths and weaknesses as the
population measure, has one distinct advantage: Almost
every water system should have accurate data readily
available on the number o f connections. This informa­
tion is certainly more accurate than a population count,
although population is probably a better estimator of
water use.

Table 47. Revenue of State and local government-owned
water utilities, fiscal years 1967-80
Year

Drinking Water Act was passed in 1974. In short,
physical outputs are preferred for productivity calcula­
tions.

62

Another reason for focusing on sales to final
customers is interutility water sales. The magnitude of
resales is not known, but focusing on sales to ultimate
users minimizes the problem of double counting in a na­
tional index. This approach is similar to that used in
electric utilities as discussed in the previous chapter.
The primary argument against using this measure is
data availability. There are no comprehensive national
statistics and, in some cases, no individual utility
statistics on revenue gallons. Furthermore, some water
systems do not collect statistics on the quantity of
revenue water sold. New York City, for example, which
does not meter most of its sales, has no record of the
number of gallons purchased by its customers. Even
those cities which have a policy of metering, such as
Boston, New Orleans, and Washington, D.C., do not
meter all sales.4 However, most large and medium-size
7
water systems do meter most sales.
The use of gallons as the output measure also implies
that production input requirements (unit labor re­
quirements) vary by gallon when, in fact, they do not.
Water treatment has economies of scale, particularly for
small and medium-size utilities. Most water systems set
graduated rates based upon the amount of water used—
the greater the amount of water purchased, the cheaper
the unit price.
An analogous situation exists in electric power pro­
duction and sales—the greater the amount of power us­
ed, the cheaper the unit price. Electric power productivi­
ty calculations often take into account decreasing pro­
duction costs (unit labor requirements) by separating
users into classes and weighting output accordingly.

Most jurisdictions, particularly medium-and largesize ones, carefully monitor and control water quality.
Most State governments, with the assistance of the En­
vironmental Protection Agency, collect and tabulate
statistics on how well drinking water quality standards
are met. This information is available by system, by
State, and nationally since 1978. e p a estimates that
13,600 water systems, private and public, do not meet
one or more drinking water quality standards.4 Small
8
systems account for the majority of the problems.
The important issue insofar as productivity indexes
are concerned is how quality has changed through time.
Additional research is needed into the area, particularly
the effect on unit labor requirements. As a first step,
utilities should be divided into two groups: Those
meeting and those not meeting e p a standards. Labor
weights could be used to combine the indexes for a na­
tional index.

Recommended measure. The recommended output
measure to compute a national water supply productivi­
ty index is revenue gallons, the measure used by most
economic analysts and water utility managers. The out­
put should be weighted by type of customer— residen­
tial, commercial, and industrial—since unit labor re­
quirements vary by type. Unit labor requirements are
the preferred weight, but price may be a satisfactory
surrogate. Data and outputs should be separated be­
tween those which meet e p a standards and those which
do not. If sample data are used, outputs should be
weighted by size of utility—e.g., small (less than 2.5
million gallons per day), medium (2.5 - 20.0 million
gallons per day), and large (more than 20 million gallons
per day).

Quality considerations. Drinking water quality concerns
every water utility. Of the dozens of water quality at­
tributes, some affect consumer utility, some affect pro­
duction costs. This discussion deals with water pressure
and with health.
Adequate water pressure is an important considera­
tion in drinking water supply. Inadequate pressure can
result in minor inconveniences, such as improperly
working washing machines and dishwashers, or major
effects, such as backsiphonage, contamination of the
water supply, and reduced firefighting capability. A
number of factors affect water pressure. Most impor­
tant are the design and construction of the system.
Operation and maintenance of the system are less im­
portant and less significant for labor productivity and
are not considered further here.
The health issue is often raised in discussions of
drinking water supply. Dozens o f pollutants can affect
water quality and community health, and their removal
can increase production costs, as noted above.

Availability o f data. No comprehensive set of statistics
is available to compute a revenue gallon output index
for State and local government water utilities at this
time. Some data may be obtained from several ad hoc
surveys and two ongoing data collection activities.
e p a ’s Federal Regional Data System, the largest
ongoing data collection system, collects data annually.
This system, which was initiated in fiscal 1978, monitors
compliance with Federal drinking water standards.
However, data are also collected on operations, in­
cluding gallonage, population served, number of
meters, and type of customers served. Whether' the
gallonage data are revenue water, treated water,
pumped water, or some other gallonage measure is not
known. Most likely they are a combination of these.
Possibly there is some double counting. As e p a works
with the States to improve data and reporting, this
system should be more useful in calculating water sup­
ply output, e p a does not collect input data, unfor­
tunately.

4 Additional Federal A id fo r Urban Water Distribution Systems
7
Should Wait Until Needs are Clearly Established' (General Account­
ing Office, November 24, 1980), p. 30.




4
8
States’ Compliance Lacking in Meeting Safe Drinking Water
Standards (General Accounting Office, March 3,1982), p. 11.
63

Probably the most accurate data on community water
supply operations are available from those e p a collects
from 78 utilities for research purposes. The data are col­
lected in the field, and the time period covered ranges
from 6 to 15 years, depending on the utility. The data,
which reflect fiscal years, include revenue gallons, labor
hours, costs of operation, population served, and
population density. These sample data, however, may
not be representative of U.S. community drinking water
systems. For example, all 78 utilities meet e p a quality
standards. Data from the e p a 78-system series can be
useful, but they are not sufficient for computing a pro­
ductivity index.
e p a also collected data on operations from 1,000
utilities in 1976. This Community Water System data
base includes statistics on population served, connec­
tions by type, sources of water, treatment, gallons
delivered, and revenue for one year. Although the data
are useful for cross-sectional analysis, they are of no use
in computing a revenue gallons output index.
Revenue gallonage statistics are available from
several of the American Water Works Association
( a w w a ) surveys. However, a w w a statistics cover only
selected years, they are a sample of the large water supp­
ly utilities, and the number of utilities varies from sam­
ple to sample. Although the a w w a statistics can help in
preparing an output index, they are not sufficient.

Labor inputs
Chapter III discussed three labor measures for
calculating State and local government productivity:
Number of employees, number of full-time-equivalent
employees, and number of employee hours.
The foui; sources of data on local government water
system employment are: (I) The Bureau of the Census,
(2) the American Water Works Association, (3) e p a ,
and (4) the individual utilities.
The Bureau of the Census is the only organization
that regularly collects and publishes a comprehensive set
of time series data on employment in local water
utilities. The data show that these utilities had about
134,000 employees in 1980, a 17-percent increase since
1967 (table 48). About 11 percent were part-time
employees in 1980. Census statistics do not include State
employees until 1980, which is probably not a limitation
for calculating a national index because of the small
number of State employees—fewer than 1,000. A more
serious problem with the Census statistics is that only
aggregates are collected and presented.
No information is collected on force account (con­
struction) employees. The number of such workers, and
the extent to which the number has changed during the
past decade, is not known. This would be an important
issue for measuring labor productivity trends if the ratio
between construction and operating employees had
changed substantially. Also, Census statistics are not



64

Table 48. Local government water u tility employment,
1967-80
(T h o u s a n d s )

Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980

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

Total employment

Full-time-equivalent
employment

115
114
116
117
116
114
121
129
129
129
131
134
132
134

108
107
108
110
108
108
113
115
115
115
118
121
120
122

S o u r c e : P u b lic E m p lo y m e n t, a n n u a l is s u e s .

divided by type or size of utility, which is useful in
.developing and applying unit labor weights.
Finally, the assignment of government personnel to
water supply operations when they work in other
areas—e.g., sewerage, gas, and/or electric power—is a
potential problem. The Bureau of the Census suggests
that governments allocate personnel among functions
using revenue figures. Statistics from larger electric
power utilities, as discussed in the preceding chapter,
suggest that this may not be a problem, particularly for
trend calculations.
The only other time series on employment is e p a ’ s
78-utility sample. It includes detailed statistics for each
utility on the number of personnel and hours paid, and
includes output as well as employment. The primary
deficiency of the information, insofar as this study is
concerned, is that it may not be representative of the
universe. Also, data collection is not timely; some data
are collected 4 or 5 years after the fact. However, this
information, collected through field visits, is probably
the most accurate available.
The American W ater Works Association also
periodically collects statistics on employment, wages,
salaries, and fringe benefits. The problem with the
a w w a statistics is their sporadic nature and inconsistent
sample. Although these statistics may help in analyzing
employment data, they are not sufficient to compute a
labor index.
The only other sources of personnel statistics are the
individual utilities.

Suggested research
An accurate, representative set of output and input
data is the first requirement for calculating an index of
productivity for State and local water supply. Analysis
and research should also be conducted in other areas.
Quality of service, particularly water quality and its ef­
fect on unit labor requirements, warrants further in-

vestigation. There is some research in the area but much
remains to be done.
Another area for research is multifactor productivity.
The Bureau of the Census estimates that salaries and
wages (without fringe benefits) accounted for only 22
percent of State and local government water supply
utility expenditures in fiscal 1977. Capital accounted for
32 percent, and interest on debt, materials and supplies,
energy, and purchased water the remaining 46 percent.
To better analyze shifts in productivity, all major factor
inputs and their interrelationships need to be examined.
Analysis of levels of productivity, as opposed to trend
analysis presented here, can provide an important addi­
tional dimension. With a unitary measure of output,
such as revenue gallons, computation of productivity
levels should be possible. Comparisons among jurisdic­
tions that take into account differences in law,
geography, topography, hydrology, weather, popula­
tion size and density, location of water, and water quali­
ty will be more difficult to prepare.

H ass Tramsot
Although mass transit, like sanitation and drinking
water services, has been the subject of considerable
research and analysis, no national productivity index is
currently calculated. Unlike the other two services, data
are routinely collected on outputs and inputs, and com­
putation o f a productivity index should be relatively
straightforward. The problem, as with so many State
and local government services, is to know what measure
of output to use.

Institutional sotting
Interest in measuring State and local government
transit productivity is a fairly recent phenomenon. Until
World War II, most transit systems were privately own­
ed. In 1950, only 26 systems or less than 2 percent of all
transit systems in the United States were publicly own­
ed. By 1980, 55 percent of the systems were public.
More importantly, public systems carried the vast ma­
jority of passengers. The private systems were mostly
small operations. State and local government systems
delivered 94 percent of the industry’s passenger trips,
operated 93 percent of the vehicle miles, and owned or
leased 90 percent of the vehicles.4
9
The shift from private to public systems has been par­
ticularly rapid since the mid-1960’s. In 1968, 65 percent
of the industry’s passenger trips were public; in 1978, 93
percent. In 1968, 56 percent of all vehicle miles were
public; in 1980, 93 percent. In 1968, 55 percent of all
vehicles were publicly owned; in 1980, 90 percent. This
4 American Public Transit Association, Transit Fact Book—1981
9
( apt a , 1981), p, 43.




shift from private to public systems continues, although
obviously not at the same rate of speed.
Even today, institutional arrangements are mixed,
varying from total private to total public, including:
1. Private ownership and operation, such as the
South Carolina Electric and Gas Company, which
operates the buses in Charleston, South Carolina.
2. Private ownership and operation with a govern­
ment subsidy, such as the Southern Pacific
commuter operation to and from San Francisco.
3. Joint ownership with private operation. An ex­
ample of this type is the Massachusetts Bay
Transportation Authority, which owns the
commuter cars while the Boston and Maine
Corporation, a private corporation, provides the
tracks and service.
4. Public ownership with private operation, such as
the commuter railroads in and around Chicago.
5. Public ownership, private management with
government employees. An example is the a t e
Management and Service Company of Cincinnati,
which manages 49 municipal bus systems through­
out the United States. Approximately 125 systems
of this type operate in the United States today.
6. Public ownership and management, as in
Washington, D.C.
The following discussion focuses on types (5) and (6),
those in which government employees operate the
system.
State and local government transit systems operate in
44 States. New York, with over 39,000 State and local
government employees and almost $2 billion in expen­
ditures (fiscal 1977), is the most deeply involved. The
large systems are heavily concentrated in urban areas,
particularly in the Northeast. The large systems
dominate production; the 13 largest systems, for exam­
ple, account for about 65 percent of all transit employ­
ment.
In fiscal 1977, State and local transit systems spent
approximately $5.6 billion, or 1.7 percent of all State
and local government expenditures.50 Their expen­
ditures for salaries and wages accounted for about 2
percent of all State and local salary and wage expen­
ditures; their capital expenditures, for about 3.7 percent
of State and local capital expenditures.
Passenger fares cover about one-third o f all transit
expenditures. Transfer payments and gasoline, sales,
and property taxes support the other two-thirds. In
fiscal 1977, the Federal Government provided $1.9
billion in transfer payments for mass transit, 24 percent
of all State and local government expenditures in this
area.
The primary factor input into transit operations is
labor. Salaries and wages account for 43 percent of all
expenditures (65 percent of current operating expen5 Government Finances, pp. 29 and 33.
0

65

Table 49. Finances of government-operated mass transit systems by type of government, fiscal year 1977
(Millions)
Expenditures
Government

Current operations

Revenue
Total

Capital

Total ......................................

$1,991

$5,598

$1,681

S ta te s ................................................
Municipalities....................................
Special districts..................................
Townships..........................................
C ounties............................................

308
941
703
1
38

628
1,811
3,052
107

109
433
1,124
15

Interest
on debt

$250
_

89
157
- ■
2

Total

Salaries
and wages

Other

$3,668

$2,391

$1,277

520
1,288
1,770
89

290
966
1,074
51

230
322
696
38

Source : G o v e r n m e n t F in a n c e s , p. 33.

Dash = zero or rounds to zero.
Note : Because of rounding, detail may not add to totals.

Baltimore and Washington, with State of Maryland
subsidies. Since these were not operated by government
employees, they are not included in this study.
Only nine streetcar systems were still operating in the
United States in December 1980. These systems
operated about 1,000 cars which traveled about 380
million passenger miles.
Five trolley coaches using 825 cars traveled almost
200 million passenger miles during 1980.
Sixteen ferry boat establishments which operated 68
boats traveled almost 340 billion passenger miles in
1980. Eight of the 16 systems were publicly owned and
operated.
During 1980, one publicly owned and operated cable
car system and five inclined planes were in operation.
None are sufficiently important to warrant special in­
vestigation in national productivity calculations.
Paratransit includes services such as dial-a-ride,
subscription bus service, airport limousines, and jitneys.
They provide variable-route or demand response ser­
vice. The latter type of service and costs of production
are very different from the fixed route and time opera­
tions. Paratransit operations are not included in this
discussion.
The focus here is on regularly scheduled transit opera­
tions. No distinction is made between “ transit” and
“ mass transit.”
The Standard Industrial Classification Manual
defines transit (SIC 4111) as follows:5
2

ditures) (table 49). Inclusion of fringe benefits would
further increase this percentage. Capital expenditures
account for 30 percent, and interest, fuel, materials,
supplies, and the like account for the remaining 27 per­
cent.
The Bureau of the Census reported 87,000 public
transit employees in 1968. By 1978, the figure had in­
creased to 125,000 and by 1980 to 172,000, or about 1
percent of all State and local government employees.
Although some of this increase reflects redefinition of
terms and new data collection procedures, employment
growth has been real and rapid. The vast majority of the
172,000 State and local,government employees work for
local government (92 percent), and most are full-time
employees (97 percent).5
1
Public transit can be divided into nine different types
(modes): Bus, heavy rail such as subways, commuter
rail, light rail such as streetcars, trolley coach, urban
ferry boat, cable car, inclined plane, and paratransit.
The first two modes account for aboilt 95 percent of all
passenger trips.
The motor bus was the most important mode in 1980
in the number of passenger trips (67 percent), operating
revenue (60 percent), and vehicle miles (74 percent)
(table 50). About 60,000 urban buses and vans were in
use in the United States in 1980. These vehicles logged
over 5.7 billion passenger trips and approximately 23
billion passenger miles. Over 85 percent of the fleet of
urban buses were either owned or leased by government.
Heavy rail or subway is the next most important
mode of public transit. In 1980, 11 heavy rail systems,
all publicly owned and operated, logged 2.3 billion
passenger trips for an estimated 10.6 billion passenger
miles (table 51).
Eighteen commuter railroads logged about 285
million passenger trips and 5.9 billion passenger miles in
1980. All were government owned or received govern­
ment subsidies. Amtrak, for example, provided service
for Los Angeles, San Diego, and Washington, D.C.,
and the Baltimore and Ohio provided service to

“Establishments primarily engaged in furnishing local
and suburban mass passenger transportation over
regular routes and on regular schedules. Such
transportation may involve use of one or more modes
of transportation. Establishments primarily engaged in
furnishing passenger transportation by automobile or
bus to, from or between airports or rail terminals over
regular routes are included in this industry.”
The Bureau of the Census and the American Public
Transit Association use analogous definitions.5
3
5 Standard Industrial Classification Manual, p. 221.
2
5 Public Employment, p. 462.
3

5 Public Employment in 1980 (Bureau of the Census, 1981).
1




66

Table 50. Transit modes (private and public) ranked by
passenger trips, revenue, and miles, 1980
(Percent)
Passenger
trips

Operating
revenue

Vehicle
miles

T o ta l..............................

100

100

100

Motor b u s ..............................
Heavy rail (subway)................
Commuter ra il........................
Light rail (streetcar)................
Trolley coach ........................
Ferryboat..............................
Cable c a r................................
Inclined pla n e ........................
Paratransit ............................

67
27
3
1

60
23
14
1

74

1

1

1

1

1

-

Mode

17

7
1

-

-

-

-

-

-

-

-

-

Dash = not reported.
So u r c e :

Transit Fact Book— 1981,

(American Public Transit Associa­

tion), pp. 40-41.

Table 51. Heavy rail operations in the United States, 1980
Authority

Location

Chicago Transit Authority......................
Greater Cleveland Regional Transit
Authority.......................................
Massachusetts Bay Transportation
Authority........................................
Metropolitan Atlanta Rapid Transit
Authority........................................
Municipality of Metropolitan Seattle . . . .
New York City Transit Authority............
Port Authority Trans-Hudson Cor­
poration .................................................
Port Authority Transit Corporation
of Pennsylvania and New Jersey..........
San Francisco Bay Area Rapid Transit
District....................................................
Southeastern Pennsylvania Trans­
portation Authority ..............................
Washington Metropolitan Area
Transit A uthority..................................
So

u r c e

:

Transit Fact Book — 1981,

Chicago
Cleveland
Boston
Atlanta
Seattle
Brooklyn
New York
Camden
Oakland
Philadelphia
Washington

p . 16.

Research and conceptual issues
The measurement of transportation productivity has
a long history. The Bureau of Labor Statistics has
calculated private-sector indexes for railroads, intercity
trucking, intercity buses, air transportation, and
petroleum pipelines for a number of years.5 These
4
measure final output by revenue traffic units. Two out­
put measures are used for railroads (sic 401). One,
revenue traffic units, the preferred measure, is a
weighted composite of freight ton-miles and passenger
miles. Freight ton-miles are adjusted for changes in
commodity mix. Freight and passenger miles are com­
bined using unit revenues. The second measure, car-mile

productivity, is an aggregate of loaded and empty car
miles for both freight and passenger service. For inter­
city trucking (sic 213), the output measure is tonmiles for six classes o f service weighted by the number
of employees. As for railroads, outputs are adjusted for
changes in commodity mix. The ouput measures for in­
tercity buses (Sic 4111, 4131, 414) are passenger miles
and deflated freight revenue. The air transport industry
(Sic 4511) output is measured by revenue passenger
miles, freight ton-miles, express ton-miles and mail tonmiles. They are combined using unit revenue weights.
Lastly, petroleum pipeline (sic 4612 and 4613) output
reflects barrel miles (a barrel mile is one barrel of
petroleum moved one mile).
A number of studies have been made of private sector
transportation productivity. John W. Kendrick com­
pleted a detailed study of U.S. air, pipeline, waterway,
intercity bus, intercity motor trucking, and local
passenger transit productivity in 1966.5 He developed
5
indexes for outputs, inputs, and partial and total factor
productivity from 1948 to 1964. The measure for
passenger output was either the number o f passengers or
passenger miles. The measure for employee input was
hours worked, estimated from data on the number of
employees and average hours worked per employee.
Deakin and Seward completed a study of United
Kingdom bus, railway, truck, boat, airport, and port
productivity in 1969. Indexes were calculated for out­
puts, labor and capital inputs, and productivity change.
Outputs were measured in either ton-miles or passenger
miles. For each mode, outputs were separated by type of
transport and weighted. The labor input measure was
adjusted for age and sex to account for composition
changes.5
6
Scheppach and Woehlcke published a study in 1975
which demonstrated how productivity measures might
be used in transportation regulation. Three
modes—rail, air, and trucking—were reviewed. The
output measures were traditional ones. For rail
passenger service they used the number of passengers,
passenger miles, and passenger revuenue.5
7
The shift from private to public mass transit has been
accompanied by a number of studies of public transit
productivity, efficiency, effectiveness, and impact. One
study by Anthony R. Tomazinis addressed performance
measurement from the perspective of four different
potential actors—operator, user, society, and govern­
ment.5 Dozens of measures were examined. Important
8

5 John W. Kendrick, “ Productivity Trends in U.S. Transportation
5
Industries,” unpublished paper, 1966, as cited in Scheppach and
Woehlcke, Transportation Productivity (Lexington, Mass.: Lexington
Books, 1975).
5 B.M. Deakin and T. Seward, Productivity in Transportation
6
(Cambridge, England: Cambridge University Press, 1969).
5 Raymond C. Scheppach, Jr., and L. Carl Woehlcke, Transporta­
7
tion Productivity (Lexington, Mass.: Lexington Books, 1975).
5 Anthony R. Tomazinis, Productivity, Efficiency, and Quality in
8
5
4
See, for example, Productivity Measures for Selected Industries, Urban Transportation Systems (Lexington, Mass.: D.C. Heath and
Company, 1975).
1954-81, Bulletin 2155 (Bureau of Labor Statistics, 1982).




67

Outputs

conclusions were that performance should be divided
between efficiency and effectiveness, and that produc­
tivity should be measured by efficiency. Efficiency is
concerned with what is provided, and effectiveness is
concerned with what is consumed. Services provided in­
clude vehicle hours, vehicle miles, and seat miles; ser­
vices consumed include the number of passengers and
passenger miles.
Gordon Fielding, among others, used the efficiency
and effectiveness concept to measure transit system pro­
ductivity.5 Fielding defined efficiency as “ doing things
9
right” and effectiveness as “ doing the right things.” He
examined a series of measures, collected data from a
number of transit systems, and compared the systems.
His efficiency measures were divided into three
types— capital utilization, operating expense per pro­
duced output, and labor productivity. The labor pro­
ductivity measure recommended by Fielding was
revenue vehicle hours per employee.
A study of the performance of urban bus systems by
Sinha and Jukins examined five labor productivity
measures: Vehicle miles per employee, vehicle miles per
driver, vehicle miles per driver hour, vehicle hours per
employee, and vehicle hours per driver.60 Vehicle miles
per employee was their recommended measure.
Another study, of bus operations in different systems
in 1960-74, found that the number of buses and
employees remained relatively constant over the time
period while patronage decreased. Almost every pro­
ductivity index examined showed a decline.6
1
These are only a few of the many studies of public
sector transit productivity, some of which are analytic,
some conceptual, and some simply descriptive. Govern­
ment officials have generally followed the recommenda­
tions of these studies and separated efficiency and effec­

tiveness

m easurem ent.6 Consequently,
2

a

Physical quantities are the preferred measure o f out­
put for government productivity calculations. Dozens
of such measures exist for public transit, and, as noted
earlier, there is considerable debate over which measure
best describes transit output. The measures most often
recommended include:
1.
2.
3.
4.
5.
6.

Analysts o f private sector transit productivity tend to
focus on the last three measures in assessing outputs.
Public sector transit analysts and managers use all
measures but tend to focus on the first three for produc­
tivity calculations. Strengths and weaknesses of each of
these measures are analyzed at length in the literature.6
3
Each measure is briefly reviewed here.

Revenue vehicle hours. Revenue vehicle hours (rvh ) is
the single output measure that most transit managers
prefer for productivity analysis. Revenue refers to the
hours a vehicle is in service and capable of generating
revenue rather than the amount of revenue actually
generated, r v h does not include the hours spent travel­
ing to and from storage facilities, other deadhead travel,
and layover time. Revenue vehicle hours is a measure of
transit availability for a community. A bus could travel
a route for 8 hours without any passengers but still
generate 8 revenue hours.
Arguments in favor of this measure are:
1. It is a good measure of the costs of production.
A recent study showed that 54 percent of the varia­
tion in operating costs among transit systems was
explained by this factor.6
4
2. It is a good surrogate measure for the service
provided to a community. If the hours of service
are extended, revenue vehicle hours increase. If
service is cut, they decrease.
3. It is a measure over which managers have good
control.
4. It encourages reduction of nonproductive use of
vehicles such as deadhead and layover time.
5. Most transit systems have information to calculate
this measure, although it has not been collected
nationally until recently.

real

dichotomy has developed between private and public
sector transit productivity measurement over the past
decade. Private sector measurement has focused on
final outputs, such as passenger miles, while public sec­
tor measurement has focused on capacity measures,
such as vehicle hours or vehicle miles. Capacity
measures are thought to capture, at least partially,
social objectives reflected in numerous administrative
and legal requirements.
5 See, for example, Gordon J. Fielding, Roy E. Glauthier, and
9
Charles A Love, Development o f Performance Indicators fo r Transit
(Irvine: University of Califorina, Institute of Transportation Studies,
1977); and G.J. Fielding and Roy E. Glauthier, Distribution and
Allocation o f Transit Subsidies in California (Irvine: University of
California, September 1976).
6 Kumares C. Sinha and David P. Jukins, Definition and Measure­
0
ment o f Urban Transit Performance (West Lafayette, Ind.: Purdue
University, December 1978).
6 Wells Research Company, Trends in Bus Transit Operations,
1
1960-74 (U.S. Department of Transportation, 1977).
6 Transit System Productivity (Washington: Urban Consortium,
2
1976, revised 1978); Proceedings o f the First National Conference on
Transit Performance, (Washington: Public Technology, Inc., 1978);
and Eckart Bennewitz, “ Mass Transit,” in George J. Washnis, ed.,
Productivity Improvement Handbook, pp. 771-72.



Revenue vehicle hours
Vehicle miles
Seat miles
Number of passengers
Passenger miles
Passenger revenue.

The principal argument against using revenue vehicle
hour is that it is a measure of capacity rather than use.
An increase in revenue vehicle hours does not necessari­
ly lead to an increase in the number of passengers car6 For example, see Fielding, Performance Indicators.
3
6 Ibid., p. 12.
4
68

ried or revenue collected. It is not a measure of final
output.

this measure provides no information about the length
of the ride.

Vehicle miles. This is a measure of total distance travel­
ed by revenue vehicles. Some vehicle mile measures in­
clude deadhead and revenue miles; others focus only on
revenue vehicle miles. The arguments for and against
using vehicle miles as an output measure are similar to
those already discussed under revenue vehicle hours. In
addition, revenue vehicle hours explained 54 percent of
the variation in operating cost; vehicle miles explained
only 28 percent.6
5

Passenger miles. Passenger miles is probably the most
widely used physical output measure of private sector
passenger transportation productivity. This measure is
better than a simple count of the number of passengers
since it takes into account the length of the ride.
Passenger miles are normally defined as the number of
miles traveled by all paying passengers in a set time
period. One passenger traveling one mile is one
passenger mile. The private sector studies of b l s , Ken­
drick, Deakin and Seward, and Scheppach and
Woehlcke cited earlier all use passenger miles in one
form or another.
Three arguments are normally advanced against
passenger miles as the measure of output:

Seat miles. A measure closely related to vehicle miles is
the number of seat miles, which are the total vehicle
miles on passenger-carrying routes multiplied by the
seating capacity. Seat miles is a better measure of transit
capacity than vehicle miles, although it ignores standee
capacity, which is important in many communities with
heavy rash hour traffic. Like revenue vehicle hours,
data should be readily available in most transit systems
to calculate this measure. However, no national time
series data are available.

1. As with the “ number of passengers” measure,
it does not reflect many mandated operations.
2. Studies show that passenger miles are not closely
correlated with the costs of production or unit
labor hour requirements.
3. Data to calculate passenger miles are not generally
available, are difficult to collect, and are not very
accurate when available. A primary problem is
estimating how far passengers ride in fixed-fare
systems. If the length of the average passenger
trip remains constant through time, which is prob­
ably a reasonable assumption in the short run,
then passenger miles and number of passengers
would result in the same index.

Number o f passengers. A measure of transit use, in con­
trast to capacity provided, is the number of passengers.
The basic measure is a simple count of all passengers
using a transit system. The measure is sometimes divid­
ed by type of passenger—paying, nonpaying, school
child, reduced fare, elderly, and so forth. A special issue
is how to count passenger transfers. Some systems count
transfers as additional passengers; others do not.
The basic strength of the measure is its focus on
usage, or final output. As Fielding noted, “ for the
typical transit system, increased patronage from one
year to the next is much more significant than any other
financial or operating statistic.” 66 Another study noted
that “ service performance must ultimately be measured
by the number of riders attracted.” 6 Most transit
7
systems keep statistics on the number of passengers. The
American Public Transit Association has national
statistics on total passengers, and revenue passengers
for private and public systems.
Two basic arguments are advanced against using total
passengers as the measure of transit output. First, it
does not consider the numerous legal and administrative
mandates such as assisting the handicapped, serving
low-income riders, reducing air pollution, and conserv­
ing energy, a point often raised by transit officials and
academicians.6 This argument applies equally to all
8
consumption-based measures of transit output. Second,

Passenger revenue. This measure is the total revenue col­
lected from passengers. Passenger revenue is sometimes
known as “ farebox revenue.” This measure is available
in every system that collects fares, and it is available na­
tionally. For those systems that cover costs through
fares, it reflects transit usage.
For several reasons, passenger revenue is not a par­
ticularly good measure of output for productivity
measurement. First, passenger revenue makes up only
about one-third of total national transit revenue (sub­
sidies make up the difference). Second, it has been
decreasing as a percent of total revenue for a number of
years. Furthermore, nonpaying passengers are often an
important user group. Every system has some, and some
systems have many. A few systems charge no fares
whatsoever. Also, farebox revenue is a function of ad­
ministered fares, which may not relate‘directly to the
cost of providing the service and to unit labor re­
quirements.

6 Ibid.
5
6 Ibid., p. 32.
6
6 Massachusetts Bay Transportation Authority and Tidewater
7
Transportation District Commission, Bus Service Evaluation Pro­
cedures: A Review, 1979, p. 23.
6 John R. Meyer and Jose A. Gomez-Ibanez, Measurement and
8
Analysis o f Productivity in Transportation Industries (Cambridge,
Mass.: Harvard University, 1975).



Preferred measures. The line has been drawn over the
past decade between measuring public and private ser­
vice output. Private output is measured by the tradi­
tional “ passenger mile.” Public sector transit service
69

has focused on multiple measures. Transit managers
and many researchers prefer capacity measures such as
revenue vehicle hours for calculating productivity. The
strengths and weaknesses of the two approaches are well
documented.
Output and productivity trends vary depending on
which approach is used.6 Actually, the two approaches
9
are complementary—one focuses on use, the other on
availability. Both approaches should be used as part of
future research in this area.
The preferred measures are passenger miles and rev­
enue vehicle hours. The two measures should be
calculated and weighted by transit mode. Nine modal
divisions, noted earlier, are preferable but two divisions
(bus and rapid rail) should be adequate for national out­
put indexes.

justments are not needed as long as quality and level of
service remain constant or approximately constant.
These attributes need to be followed through time.
Six quality and level of service attributes are often
cited in the literature. Table 52 presents examples of
how they might change and their likely impact on total
and unit labor requirements for two output measures.
Frequency of service seems to affect unit labor re­
quirements the most. At the very least, it needs to be
tracked through time. Three levels are suggested: (1)
Rush hour only, (2) rush hour and limited service up to
12 hours per day, and (3) 12 hours or more o f service per
day.

Availability o f data. Until recently, the only regular
source of national transit statistics has been the
American Public Transit Association ( a p t a ). a p t a
has regularly collected a wealth o f statistics from transit
operators. It publishes the Transit Fact Book and the
Operating Statistics Report annually. Some of the data
series go back to the turn of the century.
The a p t a statistics cover the entire transit field,
private and public. They include outputs, as noted
earlier, including revenue and passenger miles, and in­
formation is available on individual systems and for the
total industry for the major transit modes.
The primary problem with the a p t a statistics for
calculating government output is the difficulty in
separating private from public operations and the im­
possibility of assessing the error associated with the
statistics, a p t a statistics are provided by a p t a members
and reporting is strictly voluntary. Less than half the
transit operators are a p t a members and some of these
do not report. The statistical bias associated with the
data is not known.
The availability of transit data has improved greatly
with the implementation of Project f a r e (Uniform
Financial Accounting and Reporting Elements), com­
monly known as Section 15.70 Section 15 reports provide
detailed statistics on all transit systems, private and
public, that receive Federal financial support, which in­
cludes most systems in the United States. The initial
reports were submitted in 1979, and annual reporting is
required. Statistical reports are available on outputs, in­
puts, operations, community characteristics, personnel,
finances, and so forth, by mode, by system, and by
geographic area. Statistics are available on revenue vehi­
cle hours and passenger miles, the two measures recom­
mended in this report for further examination. The
primary problem with Section 15 data is its newness.
In addition to these national data bases, every transit
system collects and maintains output statistics, and
many also publish them. Some States, including
California and Michigan, collect data from individual

Quality and level o f service. Quality and level of service
are important considerations in every transit operation.
Travel time, reliability, comfort, and frequency of
operation are all important dimensions of transit out­
put. Chapter III suggested that such factors need to be
explicitly considered and adjustments made whenever
they affect base-year unit labor requirements. Unit
labor requirements will be affected by three factors: (1)
The relationship of quality and level of service to total
labor requirements; (2) the unit output measure; and (3)
the magnitude of the shifts.
Quality and level of service must be taken into ac­
count when a change markedly affects labor re­
quirements. This is not to say that the attribute is unim­
portant when it does not affect labor requirements, only
that it need not be considered in productivity calcula­
tions. For example, employee courtesy is an important
quality attribute, to which the public and transit
authorities alike are sensitive. To improve employee
courtesy, transit managers sponsor “ Driver of the
Month” awards and courtesy training, and, when all
else fails, disciplinary action. Important as these pro­
grams might be to a transit manager, they are relatively
unimportant for unit labor requirements and need not
be considered in transit productivity calculations.
The second factor to be considered is the output
measure used to calculate productivity. The quality and
level of service attribute can affect unit labor re­
quirements for some measures but not others. For ex­
ample, frequency of service plays a major role in unit
labor requirements. Systems which provide 24-hour ser­
vice usually have very different unit labor requirements
from those which provide only rush hour service.
Changing the level of service will likely have a major ef­
fect on unit labor requirements for output measures
such as passenger miles, but will have little effect on the
revenue vehicle hour measure.
The third point is that, for trend computation, ad­

7
0
“Urban Mass Transportation Industry Uniform System of Ac­
counts and Records and Reporting System” (U.S. Department of
Transportation, Urban Mass Transit Administration, 1977).

6 Ibid., p. 24.
9



70

Table 52. Examples of relationship of quality and level of service to program change, labor requirements, and two output
measures

Attribute

Illustrative
program change

Impact on total
labor requirements

Impact on unit labor
requirements of two
output measures
Revenue
vehicle hours

Passenger
miles

Increased employee training

Small

Small

Small

Increased monitoring of complaints

Small

Small

Small

Driver-of-the-month awards

Small

Small

Small

Increased safety training

Small

Small

Small

Increased maintenance

Small

Small

Small

Increased monitoring and discipline

Small

Small

Small

Increased maintenance

Small

Small

Small

Increased monitoring of routes

Small

Small

Small

Replace old equipment

Small

Small (probably
reduce)

Small (probably
reduce)

Cut number of stops

Small

Small

Small

Add express buses/trains

Large

Small

Unknown

Comfort (such as adequacy of heat)

Increased repair of equipment

Small

Small

Small

Frequency of service (such as rush
hour and non-rush hour)

Shift equipment

Small

Small

Add equipment

Large

Add service hours

Large

Unknown (pro­
bably small)
Small

Unknown (probably
small)
Unknown (probably
large)
Unknown (probably
large)

Employee courtesy (such as operator
demeanor)

Safety (such as number of accidents)

Reliability (such as percent runs completed)

Travel time (such as average commuting time)

operators and prepare regular transit operating reports.
Although no single source of statistics is available to
calculate a mass transit output index, Section 15, apta ,
and individual system data should provide sufficient
data to build a national index, and Section 15 data
should be adequate to update the index once calculated.

worked in transit systems. The number of full-timeequivalent employees was 169,000.
Chapter III listed three labor measures to be used in
calculating State and local government produc­
tivity—all employees, all employee hours, and the
number of full-time-equivalent employees. The three
sources o f yearly time series data on the total number of
public transit employees are the Bureau of the Census,
the American Public Transit Association, and the U.S.
Department o f Transportation (Section 15 reports).
The Bureau of the Census reports total employment,
full time and part time, and full-time-equivalent
employment. Census statistics have several problems for
calculating a labor index. First, Census included only
local government employment until 1980, when State
figures were added. Second, data are not separated by
mode of transportation. Information by mode is
necessary because of the apparent differences in produc­
tivity and the shift in employment among modes—e.g.,
fewer buses and more subways.
Like the Bureau of the Census, the American Public
Transit Association has collected and published
statistics on transit employment for years. The Census

Labor inputs
Transit operations are labor intensive. Salaries and
wages made up 43 percent o f all transit expenditures in
fiscal 1977 and 65 percent of current operating expen­
ditures.7 When fringe benefits are added, the figures
1
rise substantially. One study calculated that salaries,
wages, and benefits accounted for §2-87 percent of
operating expenditures.72 Another study found that they
accounted for 73 percent of operating expenditures.7 In
3
1980, according to the Bureau of the Census,- 159,000
local and 13,000 State employees, a total of 172,000,
7 Government Finances, p. 33.
1
7 U.S. Urban Mass Transit Administration, “ Transit Operating
2
Performance and the Impact of the Section 5 Program,’’ November
1976, p. 23.
7 American Public Transit Association, Transit Fact Book—1981,
3
p. 48.



71

distinction is made between government and private
employees.
None of the three sources collects employee hours.
Because of overtime, these could be important in
calculating transit productivity. Split shifts and over­
time are common among transit employees and the pro­
blems they create for transit operations and staffing are
widely discussed. Although hourly figures are not
available, full-time-equivalent employment statistics of
Section 15 reports should be good surrogate measures.
So far as is known, none of the three sources includes
employment on commuter rail systems. However, this
information is reported to the Interstate Commerce
Commission and also to the Department of Transporta­
tion, which plans to publish it in the future.
Table 53 summarizes the availability of employment
data.

and APT A statistics vary dramatically, particularly in the
early years. Several reasons account for the differences.
First, a p t a figures include both public and private
employment; Census figures, only public employment.
Second, a p t a figures reflect estimated employment for
six transit modes—bus, subway, trolley, streetcar, in­
clined plane, and cable car (commuter rail, ferry boat,
and automated guide way are excluded). Census
statistics, on the other hand, reflect employment in tran­
sit agencies as defined by the responding government.
Third, a p t a figures include Puerto Rico; Census does
not. Neither Census nor a p t a separates employment by
mode.
The Section 15 reporting system provides some data
not available from the other two systems. It provides in­
formation on the number of full-time:equivalent
employees by mode and by system. It also separates
employees by class (maintenance, transport, and general
administration) and collects information on salaries,
wages, and benefits by mode.

Suggested research
The next step should be to attempt to calculate a
series of productivity indexes using the available sum­
mary data, augmented, where necessary, with individual
system data. Initial calculations should focus on the two
recommended measures, passenger miles and revenue
vehicle hours, for bus and rapid rail. Level of service
and quality need to be addressed. In addition to na­
tional trends, it should be possible to construct trends
by geographic area, by size of system, and by mode.
Statistics on the level of productivity would provide
important additional information, and with a unitary

Section 15 reports have five problems for productivity
calculations. First, the few years of data available are
useful for benchmarking but are not sufficient to com­
pute a labor index. Second, statistics are not collected
on the number of employees or employee hours.
However, an index of employee hours should parallel
the full-time-equivalent index. Third, Section 15 reports
do not include all public transit systems, but they do
cover about 95 percent. Fourth, the reports include
private as well as government-owned systems. Fifth, no
Table 53. Availability off employment data by transit mod®

Labor coverage
Mode

Number of employees (full time,
part time, seasonal)

Full-time-equivalent

Hours

Multimode...................................... Full time and part time available
from Census for October each
year. No statistics available on
seasonal workers.

Available from Census for Octtober each year. Section 15
provides information; see
form 404.

Information not currently avail­
able. Some operator statistics
collected by American Public
Transit Association (apta). N o
plans to collect data.

B u s .................................................

Information not currently avail­
able by mode.

Section 15 provides informa­
tion; see form 404.

Some operator statistics collected
by apta. No plans to collect ad­
ditional data.

Heavy rail.......................................

Information not currently avail­
able by mode.

Section 15 provides informa­
tion; see form 404.

Operator statistics collected by
apata. No plans to collect ad­
ditional data.

Commuter rail................................

Information reported to Inter­
state Commerce Commission
(icc) and Department of
Transportation (dot ).

Can be computed from inform­
ation reported to icc and

Information reported to icc and

Information not currently avail­
able by mode.

Section 15 provides inform­
ation; see form 404.

Other:
Light rail
Trolley
Ferry boat
Cable car
Inclined plane




DOT.

DOT.

72

Information not currently avail­
able. No plans to collect data.

output measure, such as revenue hours, it should be
possible to compute productivity levels with little addi­
tional work. Meaningful comparisons among jurisdic­
tions will be much more difficult because of differences
in laws, topography, weather, population density, road
networks, and congestion.
Data are available to permit examination of multifac­
tor productivity. However, this will require considerable
additional work, and probably is not warranted given
the importance of labor.

This study focuses on traditional services, which in­
clude:
1. Interviewing jobseekers
2. Identifying job openings
3. Matching job applications with openings and
referring qualified applicants to employers
4. Counseling applicants
5. Testing applicants
6. Preparing and distributing labor market
information.

T b & EmpBoyment S s rw fe

An interview, the first step for a client entering the ES
process, sheds light on a client’s job skills, knowledge,
and interests. In fiscal 1979, there were over 15.5 million
new and renewal applications.
Identification of job openings or job development,
the next step, is an employer-focused function. The
primary es activity is contacting employers to obtain job
listings. These contacts consist of personal visits,
telephone calls, mail, and promotional activities. Job
listings are entered into computerized job banks daily.
In fiscal 1979, more than 1.8 million employers were
contacted and 9.0 million job openings were listed.
The next step in the process is job matching and refer­
ral to employers. In fiscal 1979, ES made 8.2 million
referrals which resulted in 4.5 million placements. ES
centers are installing computers throughout the country
to assist in this process.
Counseling is available to those who need assistance
in choosing a field of work, who wish to change their
occupation, or who have difficulty in holding a job. In
fiscal 1979, over 1 million applicants received job
counseling.
Applicants who do not have a trade or occupation, or
who wish to change occupations, may take general ap­
titude, specific aptitude, or general interest tests, es,
which tested about 800,000 applicants in fiscal 1979, has
an active program to develop and revalidate tests.
Labor market information is needed to support many
es activities. Information is routinely developed on the
number of job openings and the characteristics and
number of workers seeking jobs. The national es office
develops and maintains aids such as the Dictionary of
Occupational Titles, Handbook o f Occupational
Keywords, and special handbooks such as the Health
Careers Guidebook.
A 1974 study estimated the es staff time spent on each

Federal and State governments have been concerned
for a long time about measuring Employment Service
(es) productivity. Like sanitation, drinking water, and
mass transit, considerable research has been conducted
and much information has been collected. Unlike the
other service areas, “ productivity” has been routinely
calculated for the ES. However, there is considerable
unhappiness with this work and the resulting productivi­
ty measurements.

Institutional considgrations
ES was first established during World War I to recruit
defense workers.7 The Service languished during the
4
192Q’s but was given new life by the Wagner-Peyser Act
of 1933 and the Social Security Act of 1935. The Social
Security Act, which established the unemployment in­
surance program, called for a work test as a condition
for receiving unemployment payments, and ES was
assigned responsibility for administering that test. ES
was also deeply involved in operating referral service for
work relief programs. In the 194Q’s and 1950’s, ES first
recruited and referred workers to the defense program,
and later helped veterans and defense workers return to
civilian employment. The emphasis of es shifted in the
196Q’s to assisting the disadvantaged and administering
registration required of welfare recipients. The 1970’s
saw a return to more traditional labor exchange opera­
tions, although the registration function (work tests) re­
mained.
e s is a joint Federal-State operation; local govern­
ment is not involved. The Federal Government is
responsible for setting procedures, standards, and
guidelines to operate the system. The States operate the
service. In 1980, about 2,600 es offices and 30,000 State
employees worked on traditional ES activities authorized
by the Wagner-Peyser Act. An additional 14,000 State
employees worked on es responsibilities under afdc
and food stamp programs, and 6,000 other State
employees worked under other Federal labor
contracts.7 More recently, these numbers have been
5
reduced as programs have been terminated.

of four functions—intake, counseling, referral, and
labor market information—as follows:7
6
7
6
Neil S. Weiner, John H. Powel, and C. Michael Rahm, The
United States Employment Service: A Conceptual Model o f Outputs,
Values and Illustrative Estimations, Vol. II, (Arlington: Boeing Com­
puter Services, 1976), p. B-9-13.

7 Employment and Training Report o f the President, 1977, p. 71.
4
7 Ibid., p. 73.
5



73

Function

Percent o f staff time

T otal....................................................
Referral ......................................................
Intake.........................................................
Counseling.................................................
Labor market information........................

sonnel services such as space, utilities, computers, and
travel. In fiscal 1980, about $742 million went to sup­
port about 25,000 basic program staff; $44 million went
to support about 2,400 individuals in the AS&T function;
and about $127 million went for nonpersonnel services.
Labor market information services received about $32
million from several sources; special projects, enforce­
ment, and worker protection received $26 million.8
1
Labor, the primary factor input into ES programs, con­
sumes about 85 percent of the budget.8
2
For several years, es has routinely calculated
placements per staff year, a measure of productivity.
Recently, these statistics have been included in the Presi­
dent’s annual budget (table 54). The basis for these
calculations is not known but it certainly varies from
year to year. The figure for fiscal year 1978, for exam­
ple, is 201, 238, or 265 depending on whether the figure
is taken from the 1979, 1980, or 1981 budget.
es uses two basic data systems to track its operations
today: One collects program information; the other col­
lects resource and cost accounting information. The
Employment Security Automated Reporting System
(esars) has collected basic program information since
the early 1970’s. The State Employment Security Agen­
cy (sesa) Accounting System has collected resource data
since 1970. Both systems collect data from local es of­
fices. Both program and resource data have been
available since the late 1930’s but not in the depth of
coverage that is available from the esars and sesa Ac­
counting Systems.
The Bureau of the Census does not separately identify
es operations in its statistics, es operations are included
under the general heading of Employment Security,
which includes all State labor activities.
The sic Manual assigns ES operations to Industry
7361, Employment Agencies. This category includes:

100
61
25
8
6

An analysis which followed a sample of applicants
through the various activities in fiscal 1974 showed that,
for every 100 applicants,
44 dropped out after intake;
4 dropped out after counseling/testing;
2 entered training;
21 dropped out after job referral; and
29 found jobs.7
7
In other words, about 29 percent of the referrals led
to jobs. The stability of these proportions through time
is not known but fiscal 1979 showed exactly the same
placement rate.7
8
In addition to the traditional labor market activities,
es is responsible for enforcement of work test re­
quirements and three compliance activities. Enforce­
ment activities are the reqistration and monitoring of
unemployment insurance and food stamp and welfare
recipient activities. The three compliance activities are
immigration service certification, farm worker and
wage standards certification, and worker complaint
referral.7 Immigration certification requires that es at­
9
test that immigrants do not take jobs which unemployed
Americans could fill and that they are paid the prevail­
ing wage for that job. ES local offices take the applica­
tions; regional offices determine employability. Less
than 2 percent of the es budget is allocated to this activi­
ty.
The es also certifies that the employer provides ade­
quate housing and pays prevailing local wages to foreign
and migrant workers. These activities are now a
“ relatively insignificant proportion of the ES
budget.” 8
0
The third activity—worker complaints—involves for­
warding complaints that arise in the workplace, such as
working conditions, pay, and discrimination, to other
government agencies. This is a minor activity of ES.
Funding for the traditional es program comes from a
tax on employers (Federal Unemployment Tax
Act— futa). Funding for nontraditional es programs,
such as compliance and enforcement, comes from
general tax revenue.
The Department of Labor allocates funds to in­
dividual States to support es program staff; ad­
ministrative, supervisory, and technical (as& ) staff;
t
labor market information; special projects; and nonper-

“ Establishments primarily engaged in providing em ploy­
ment services except theatrical employment agencies (In­
dustry 7922) and m otion picture casting bureaus (Industry
7819). Establishments classified here may assist either
employers or those seeking em ploym ent.” 8
3

State administrative offices apparently are assigned to
Industry 9441, Administration of Social, Manpower,
and Income Maintenance programs.
•Most States and the Federal Government currently
refer to local ES operations as the Job Service.

Research and conceptual issues
Considerable research has been done on es opera­
tions and management, including productivity measure­
ment. A recent study reviewed 27 different papers which
8 Charles K. Fairchild, A Performance and Needs Based
1
Methodology fo r Allocating Employment Service Grants, (Cam­
bridge, Mass.: Abt Associates, 1980), pp. 12-13.
8 Ibid., p. 39.
2
8 Standard Industrial Classification Manual, p. 304.
3

7 Ibid., p. 32.
7
7 Report o f the President, 1980, p. 58.
8
7 Weiner and others, A Model o f Outputs, pp. 62-64.
9
8 Ibid., p. 64.
0




74

Table 54. Placements per staff year as shown in the
President’s budget, fiscal years 1974-82

Jacob Benus, focused on the duration o f unemployment
and on earnings.8 His initial conceptual study was
8
followed by a pilot study.

Fiscal year of President’s budget
measured

1977

1974 ....................
1975 ....................
1976 ....................
1977 ....................
1978 ....................
1979 ....................
1980 ....................
1981....................
1982 ....................

225
211
208
224
-

1978

1979

1980

1981

1982

Service category productivity. Service category produc­
tivity focuses on the efficiency with which specific es
services, such as counseling, testing, and job develop­
ment, are handled.
Fred Englander undertook three studies for the New
Jersey es to identify the net influence of es on job
placements. One studied the factors affecting the
number of job openings.89 Another looked at the impact
o f external, internal, and demographic considerations
on placements. The third examined the method es used
to allocate funds to State employment services.
A study by Neil Weiner used the general productivity
function to estimate the social return of ES.90 Weiner
divided es direct outputs between the labor market and
compliance requirements. Time accounting statistics
were used to estimate the resource flows.
Mark Chadwin and his associates used an institu­
tional approach to examine service category productivi­
ty.9 They studied the organizational characteristics
1
which produced good performance in State es agencies.
Their productivity measure was individuals placed (not
transactions) per staff year.

_
217
225
-

225
238
241
-

265
’229
’238
-

201
’207
’ 190
'297
-

207
’ 186
’ 177
’212

1 Estimate.
Source: Appendix to President's budget, selected years.

examined and discussed ES productivity.8 Most were
4
published between 1975 and 1978. The authors assigned
the 27 papers to one of four categories—organizational
productivity, service category productivity, client effec­
tiveness, and labor market effectiveness. In other
words, two categories dealt with efficiency and two with
effectiveness. Although this report focuses on efficiency
(productivity), the papers dealing with effectiveness are
briefly noted first.

Labor market effectiveness. Labor market effectiveness
is defined as the impact of es on the economy as a
whole, such as its effect on the gross national product
and the unemployment rate. Basil Moore of the U.S.
Bureau of the Budget examined this issue in 1966.8
5
Although his purpose was to develop an analytical
framework, he examined data and made simple calcula­
tions. His basic output measure was the difference bet­
ween the length of unemployment of those registered
with ES and the length for all unemployed persons.
Ten years later, Donald Frey completed a conceptual
examination of the same general issue.8 8 Factors in­
6
7
vestigated were the duration of unemployment, job turn­
over, and deterrent effects of es operations. Potential
data sources were reviewed but no attempt was made to
calculate the effect of es.

Charles K. Fairchild examined the development of
performance standards, both for placement and
placement-support functions.92 He developed suggested
standards and procedures for evaluating output and
allocating funds to the States and within the States to ES
offices. Four output measures were considered: In­
dividuals placed per staff year, placements per staff
year, percent of openings filled, and percent of ap­
plicants placed. Fairchild concluded that placements per
staff year was the preferred output measure for produc­
tivity calculations.

Organizational productivity. Three organizational or
composite productivity studies were identified and
reviewed in the 27-study review. One was a General Ac­
counting Office review o f ES operations in 1978 which

Client effectiveness. Client effectiveness, the second
category of studies examined in the 27 papers, focused
on how clients themselves, rather than the Nation as a
whole, benefited from es operations.
A study by Arnold Katz examined the impact of es
labor exchange programs on applicant unemployment

8 Jacob Benus and others, Use o f an Experimental Design in
8
Assessing the Impact o f the United States Employment Service (Menlo
Park, Calif.: Stanford Research Institute, September 1976).
8 Fred Englander, “Factors Affecting the Receipt of Job Openings9
by the State Employment Service Agencies and by Local Employment
Service Offices in New Jersey;” “The Impact of Demographic, Inter­
nal and External Factors on the Placement Performance and Staffing
of the Local Offices of the New Jersey Employment Service;” and
“The Impact of Demographic, Internal and External Factors on the
Placement Performance of State Employment Service Agencies”
(Trenton, N.J.: Department of Labor and Industry, May 1977).
9 Weiner and others, A Model o f Outputs.
0
9 Mark L. Chadwin and others, The Employment Service: An In­
1
stitutional Analysis (Employment and Training Administration,
1977).

8
4

David W. Stevens and others, “ Specification and Measurement
use s ,” draft report (Employment Service,
December 1980).
0 Basil Moore, “A Benefit-Cost Analysis of the United States
5
Employment Service” (Bureau of the Budget, November 1966).
3 Donald E. Frey, A Methodology fo r Measuring the Impact o f the
6
United States Employment Service (Winston-Salem: Wake Forest
University, 1976).
9 Charles K. Fairchild, Development o f Performance Standards
2
8 Arnold Katz, Exploratory Measures o f Labor Market Influences
7
fo r Job Placement and Support Services o f the Public Employment
o f the Employment Service (University of Pittsburgh, 1978).
Service (New York: E.F. Shelley and Company, August 1975).
o f Productivity in the




75

measures of performance for four of the missions. Sec­
ond, he examined two production functions for the
basic labor exchange mission. Both used total in­
dividuals placed as the output measure. Staff years was
the primary factor input.
External and internal factors were examined. The
significant independent variables in the first production
function were: Staff years, civilian labor force, popula­
tion density, number of new and renewal applicants,
youth as a percent o f all applicants, claimants as a per­
cent of all applicants, the year, and the es region. Fairchild analyzed 5 years of ES data, State by State. His
equations explained 97.5 percent of the variance o f “ in­
dividuals placed,” the dependent variable. An
estimating equation was also developed from 1979 data.
This equation dropped three variables: Number of new
and renewal applicants, the year, and the es region.
The research on es productivity is extensive, as this
brief review indicates, but several, basic conclusions
emerge:
1. Most productivity research has used placements as
the preferred measure of output. However, there
are a number of variations on the basic theme.
2. Production function research, using the number of
placements as the output, has been fairly suc­
cessful in explaining variance about the depen­
dent variable.
3. Externalities play an important role in exploring
placement variation.

examined both placement and job development func­
tions.93 The primary performance statistics were
placements made and jobs filled. The accuracy of place­
ment statistics also was considered. The findings will be
discussed later.
Another study, by C. Meike, focused on productivity
o f individual State es agencies.94 It identified outputs,
developed quantification procedures, and combined the
answers into an overall measure of productivity. The
report used four categories of outputs: Applicant pro­
cess, employer process, placement process, and man­
dated outputs. A value-added approach was proposed.
The third study, by H. Kaitz, examined relationships
among internal factors, external factors, and output.9
5
Placements per staff year was his preferred measure of
productivity.
Two other research reports not covered in the Stevens
draft report bear mentioning. One is a 1979 investiga­
tion of es productivity by Thorpe and Toikka which us­
ed a production function approach.9 Three slightly dif­
6
ferent measures of output were examined: Placements
of 3 to 150 days, called permanent placements; total
placements less subsidized (ceta) placements; and total
placements (agricultural and nonagricultural). The last
measure was dropped because the regression results
were so “ poor.”
Thorpe and Toikka tested a number of independent
variables. Significant variables were the number of ES
staff years, the size of the civilian labor force, the per­
cent of workers unionized in nonagricultural industries,
the percent of applicants who were unemployment in­
surance claimants, and the percent of applicants who
were economically disadvantaged. Coefficients of all
the independent variables had the expected sign. The
unadjusted 'coefficient of determination for the
Thorpe/Toikka equations ranged from .93 to .96. The
coefficients ranged from .92 to .94 for two
variables—staff years and civilian labor force. The data
used by Thorpe and Toikka were Title III grants for
1977 only.
A Fairchild study examined the feasibility of restruc­
turing the es grants mechanism using performance and
needs as the criteria for distributing funds to the
States.97 His investigation is of interest for this study for
two reasons. First, he separated ES goals into five mis­
sions or parts—basic labor exchange, supplemental ser­
vices, employer technical services, labor market infor­
mation, and compliance/enforcement—and assigned

Outputs
Satisfactory measurement of output is the fundamen­
tal issue in calculating es productivity. Two basic out­
puts—placements and services—are examined here.
In many respects, the issues and concerns that sur­
round the measurement of ES outputs are similar to
those that are found in private sector employment
measurement: Both government and private firms
counsel, test, and assist individuals in obtaining employ­
ment (placements). There is one important difference,
however. Although a bundle of services may translate
into dollar output for private firms, es output requires a
physical measure.9
8

Placements. The number of placements is the measure
almost always used today for es outputs, es has col­
lected statistics on the number of placements since 1938;
it tracks the number of placements monthly; and it in­
cludes the number o f placements as part of its annual
budget justification.
For es records, a placement occurs each time an

9 The Employment Service—Problems and Opportunities fo r Im­
3
provement (General Accounting Office, 1977).
9 C. Meike and others, sesa Productivity Measurement System
4
(Vienna, Va.: Analytic Systems, September 24, 1976).
9 Employment Service Performance Handbook fo r Local Offices
5
(Rockville, Md.: westat , Inc., 1979).
9 Charles O. Thorpe, Jr. and Richard S. Toikka, Determinants o f
6
State Employment Service Productivity (Washington: The Urban In­
stitute, March 1979).
9 Fairchild, Methodology fo r Allocating Employment Service
7
Grants.



9
8
Employment services in the private sector generally are priced
in two ways: (1) Payment is contingent on the individual being placed
in a job; without placement there is no payment. (2) Payment is made
for a bundle of services such as testing, counseling, referrals, resume
preparation, and the like. The price in this case is for a bundle of ser­
vices and is not contingent on job placement.
76

employer hires an applicant who is referred. For a place­
ment to be recorded, five steps must take place:
1. A job order form must be prepared before
referral;
2. Arrangements must be made with the employer for
the referral of an individual or individuals;
3. The individual must not have been specifically
requested by the employer;
4. A reliable source, preferably the employer, must
verify that the individual entered on the job; and
5 . The placement must be recorded on appropriate es
form s."

According to John P. Campbell:
“ The State’s unemployment rate consistently yields
the highest correlation (negative) with placements. Ad­
ding a measure of new hires and a measure of the
percentage of the work force in lower level jobs boosts
the total variance accounted for to 40 to 45 percent of
the total variance in placements. Adding certain addi­
tional independent variables to the equation con­
sistently increases the variance accounted for to 60 to
70 percent of the total.” 1 1
0

Studies by Thorpe/Toikka and Fairchild, discussed
earlier, found external variables to be crucially impor­
tant in explaining placement variance. The only ES input
that was statistically significant was the amount of labor
input.

The basic argument in favor of using placements to
measure output is that placing individuals in jobs is the
role of es . Counseling, operation of the job bank,
registering workers, and preparing labor market
analyses all support the basic service of getting jobs for
people.
Furthermore, placements are measurable, repetitive,
and easily understood. They are physical measures, and
data to make the measurements are readily available.
The three principal arguments against the use of
placements are: (1) ES has responsibilities in addition to
its labor market mission; (2) externalities have a major
effect on placements; and (3) much of the placement
data are questionable. This last point will be discussed
later in this section.
Regarding the first point, es has been assigned addi­
tional missions from time to time. In 1977, it was in­
volved in administering 21 laws, 11 Executive Orders,
and 14 agreements with Federal agencies.9 0 These
1
9
0
responsibilities ran the gamut from certifying aliens for
work in the United States to checking the adequacy of
housing for migrant v/orkers.
Fairchild, as noted, identified five separate ES mis­
sions or functions. Although Fairchild did not estimate
es resources for each mission, labor exchange is clearly
the largest.
A more troublesome issue than multiple es missions
in using placements as the output measure is the exter­
nality issue, or the problem of separating placements
from other considerations. When the economy is boom­
ing and employment is high, placing individuals in jobs
is relatively easy, but in a declining economy with high
unemployment, placing individuals is difficult. Other
considerations, such as the skills requested, availability
of a skilled work force, and size of the labor market,
also influence the placement rate, es placements are
probably influenced more by factors external to es
operations, particularly the state of the economy, than
by internal considerations. For this reason, placements
are not a very good measure of organizational outputs.

Variations on the theme. Part of the controversy sur­
rounding the use of placements as the es output measure
stems from the different ways in which the term is used.
es makes three important distinctions: Transactions vs.
individuals; agricultural vs. nonagricultural; and length
of time the placement lasts—3 days or less, more than 3
but less than 150 days, and 150 days or more. Each is
briefly reviewed below.
Transactions vs. individuals. Simply stated, a place­
ment transaction takes place when an applicant is hired.
The placement may last from 1 hour to many years; one
individual can be placed several times each year.
Researchers and some es administrators feel that the
number of individuals placed is a better measure than
the number of transactions for two reasons: First, the
time required to place an individual is more stable than
the time required to complete a transaction. That is, a
new applicant requires considerable time; a registered
applicant taking an intermittent job requires little time.
The primary effort is initial registration, counseling,
and testing. No research has been found during this
study which shed light on the time required for an in­
dividual placement vis-a-vis a transaction placement.
Second, focusing on individuals encourages longer
term placements and benefits the client. This thesis,
though valid, is not a productivity argument and will
not be pursued further.
es believes that both individual and transaction
placements are im portant, and both should be
calculated and used. Comparison of the two statistics,
which es has published since 1975, suggests that they
move together (table 55). Between 1975 and 1980, in­
dividual placements rose by 15.7 percent while transac­
tion placements increased by 15.1 percent. The ratio of
individual to transaction placements varied between 67
and 71 percent between 1975 and 1980. Thus, in
calculating productivity trends, it may not make a great
deal of difference which measure is used.

9
9
Glossary o f Program Terms and Definitions (Employment and
Training Administration, 1978).
11
0
John P. Campbell, “ Comments,” in Stevens, “ Specification
10 General Accounting Office, The Employment Service, p. 2.
0
and Measurement o f Productivity in the uses , ” p. 188.



77

Table 55. Comparison of Employment Service transactions
and individual placements, fiscal years 1975-S0

Year

Transactions

1975..............................
1976..............................
1977..............................
1978..............................
1979..............................
1980..............................

4,670,610
4,936,222
5,283,715
6,015,728
6,206,674
'5,376,871

Individual
placements

3,137,542
3,367,007
3,732,152
4,213,423
4,180,635
'3,630,930

“ crew” placement requires less labor input than an in­
dividual placement.

Individual
placments as
a percent of
transactions

Number o f days per placement. Since the mid-1970’s,
has collected and published placement statistics by
job duration—less than 3 days, 3 to 150 days, and more
than 150 days. This division is one attempt to identify
the quality of placement, the rationale being the longer
the placement, the better off the client. In other words,
this is a client effectiveness issue.
The question for productivity measurement is
whether unit labor requirements vary significantly by
placement duration. If they do, some type of division or
weighting is needed. No research was uncovered which
shed light on this subject. The issue is not considered
further here although the subject needs further
research.1 3
0
es

67
68
71
70
67
68

1 Excludes New Jersey and Puerto Rico.
Source: Employment Service staff.

Agricultural vs. nonagricultural. es has traditionally
divided its transaction placements between agricultural
and nonagricultural; at one time the ES budgeted for
these separately. In I960, agricultural placements ac­
counted for 62 percent of total e s placements. By 1970,
placements were evenly divided, but by 1979,
agricultural placements made up only 6 percent of the
total (table 56).
When agricultural placements were an important part
of its work, e s divided them into four types—crew,
pool, individual, and other. Crew placements resulted
from interview and selection of a crew leader to recruit
individual workers. Pool placements were made from
applicants who gathered at an assembly point each day.
No interviews were conducted and the work period was
1 day. Individual placements were those which placed a
single individual in a farm job, a process that was
similar to nonagricultural placement. Other placements
were primarily family groups hired for no longer than 1
day. Most agricultural placements were of the crew and
pool types.
Several factors were responsible for the decline in
agricultural placements. First, demand lessened as
mechanization increased. Second, the Bracero Program
was terminated; this program brought Mexican labor to
American farms. Third, the Judge Richey decision in
1974 required that agricultural workers be provided
with the same services as other workers. Fourth, the
Farm Service Bureau was merged into the e s in the
1960’s.
Today no distinction is made and no productivity
calculated for the different types of agricultural place­
ment. However, anyone preparing a productivity index
that covers the period when agricultural placements
were an important part of e s ’s work dearly needs to
consider the issue of agricultural placements.
The 1968 Senate budget hearings noted that
agricultural placements per staff year for three States
were in the thousands; nonagricultural placements were
in the hundreds. Mississippi reported 7,896 agricultural
placements per e s employee.1 2 Clearly a “ pool” or
0

Placement data. Transaction placement data are
available from 1938 on and can be found in a number of
places. The data series do not always agree, particularly
in recent years, but these differences are usually ex­
plainable, as discussed later. Five series were identified
during this investigation: Selected Department of Labor
annual reports before 1979, the Department' of Labor
annual report for 1979, Historical Statistics, budget
data, and Employment Security Automatic Reporting
System statistics (table 57).
A more serious concern is the accuracy of the place­
ment data. The General Accounting Office (GAO) in
1975 conducted the best known of the various studies on
the issue. This study examined statistics collected in
the field and also reported on 1974 and 1975 studies of
the same issue by the Department of Labor. The 1974
Labor study found in a sample or four States that 15
percent of the claimed placements did not take place.
The 1975 Labor study found in a sample of five States
that between 14 and 20 percent of the claimed
placements in seven e s offices did not take place. The
g a o found in its own sample that 44 percent of those
recorded as placed in a job claimed that they did not get
a job or obtained it through some mechanisn other than
the e s referral.1 4 Although the Department of Labor
0
disagreed with g a o methodology and the magnitude of
the error identified, it did not dispute the fact that
claimed placements often did not take place.
Four interrelated types of error occur in the data
series: Coverage errors, definitional errors, data collec­
tion errors, and fraud. Probably the largest error comes
from definitional issues, although no study has examin­
ed the relative importance of each type of error.

13 Job duration statistics as reported by the es are “ expected
0
duration” of the job as reported by the employer, not actual duration
12
0
Labor-Health, Education, Welfare Appropriation fo r Fiscal on the job. Research shows expected duration statistics to be extreme­
ly inaccurate. See “ Michigan Placement Follow-up Demonstration
Year 1968 (U.S. Senate, Subcommittee of the Committee on Ap­
Project” (Silver Spring, Md.: Macro Systems, Inc., March 17, 1981).
propriations, 90th Congress, 1st session, 1967, Government Printing
14 General Accounting Office, The Employment Service, p.ll.
0
Office) p. 121.



78

Table Si. Agricultural and nonagricuSSural placement
transactions, Employment Service, fiscal years 1955-791

Some types of coverage error already have been
discussed. Program coverage is probably the most im­
portant. Traditional and nontraditional program
placements are sometimes commingled. For example,
placement data series include placements of food stamp
and a f d c recipients as well as traditional placements
before 1971. Any productivity series covering these
earlier years needs to make allowance for this change.1 5
0
The time period covered is another potential source of
error. Placement data are reported and published mon­
thly and summarized by calendar and fiscal year. At
times the period being examined is not made clear.
Geographic coverage also may be inconsistent in some
data series.
The second reason for data error, probably the largest
single reason and major issue for this study, is the
definitional issue: What constitutes a placement? As
noted earlier, e s has very explicit rales, but they are not
always followed despite the validation routines.
The third type of error is in data collection. Problems
do occur in keypunch, clerical, and computer opera­
tions. For example, the State of Washington was not in­
cluded in national statistics in 1973 because the data
were lost. The magnitude of this type of error is not
known but probably is not large. Where such error ex­
ists, it probably does not bias transaction statistics.
The fourth reason for error is outright fraud. Because
of the pressure to produce placements, some individuals
have falsified records, the most well-known case occuring in West Virginia. Although newsworthy, fraud is
probably not an important source of error in placement
statistics, according to ES staff.
The relevant issue for productivity analysis is not
placement error or even bias, but how the statistics vary
through time. A constant bias—30 percent underrepor­
ting, for example—would have no major effect on out­
put trends. However, If the magnitude of the bias
changed, the output trend would be affected.
The ability to identify and correct error varies by type
of error. Coverage error is probably the easiest to treat.
Elimination of this error requires, first and foremost, a
good understanding of e s programs and data series.
Definitional error is probably the most difficult one
to deal with. Studies show that it can be significant,
although discussions with e s staff suggest that the data
are becoming more accurate and less biased. If so, the
change in placements and productivity will be
understated, but whether the error and bias are chang­
ing and, if changing, by how much, is not known.
Data collection errors are probably not large and pro­
bably do not produce bias. The fourth type of error,
fraud, is probably not important, as noted.
e s has recently embarked on a major program to
validate placement and other e s a r s data. One of the

(Thousands)
Year

Agricultural

Nonagricultural

1955
1956
1957
1958
1959

.......................... , ..........
......................................
......................................
......................................
......................................

8,992
9,249
9,002
8,710
9,615

5,536
6,174
5,976
5,236
5,704

1960
1961
1962
1963
1964
1965
1966
1967
1968
1969

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

9,747
9,004
9,029
7,924
7,125
6,047
4,339
4,113
4,573
4,865

6,083
5,591
6,506
6,632
6,454
6,330
6,587
6,142
5,760
5,524

1970
1971
1972
1973
1974
1975
1976
1977
1978
1979

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

4,550
3,264
2,715
2,105
1,758
1,498
594
314
340
363

4,604
3,597
3,610
4,517
4,913
4,274
4,641
4,970
5,675
5,844

' Includes Puerto Rico.
1955-76, selected annual reports of Department of Labor;
1977-79, Employment Service staff.
SO URCE:

Table 57. Five placement transaction time series,
Employment Service, fiscal years 1955-79
(Thousands)

Year

Depart­
ment of
Labor
annual
reports
(selected
issues)

Depart­
ment of
Labor
annual
report
for
1979

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

14,528
15,422
14,978
13,946
15,319

1960 ....................
1961....................
1962 ....................
1963 ....................
1964 ....................

15,830
14,595
15,534
14,556
13,579

_

1965
1966
1967
1968
1969

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

12,377
10,926
10,255
10,332
10,389

1970 ....................
1971....................
1972 ....................
1973 ....................
1974 ....................
1975
1976
1977
1978
1979

1955
1956
1957
1958
1959

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

-

14,528
15,422
14,960
13,946
15,319

-

-

-

-

-

-

-

-

15,830
14,595
15,534
14,556
13,579

_

_

-

-

-

-

_
10,892
10,255
10,332
10,389

12,377
10,926

-

-

10,476
10,323
10,337

-

9,154
6,860
6,325
6,622
6,672

9,154
6,860
6,325
6,622
6,672

-

9,144
6,128
6,325
6,738
6,127

5,159

5,772
5,234
3,808
6,015
-

5,772
6,918
5,932
6,632
6,755

5,662
4,645
5,250
6,015
6,207

4,671
4,936
5,284
6,016
6,207




-

Historical
President's Employ­
Statistics
budget
ment'
of Employ­
and ap­
Security
ment
pendixes, Automated
Security
selected Reporting
Activities,
System
issues
1938-66

-

-

-

_
-

-

-

-

-

-

-

15
0
U.S. Congress, House of Representatives, Committee on
Government Operations, Operation o f the U.S. Employment Service,
(Government Printing Office, 1976), p. 305.
79

mischief. Additional investigation is needed; the con­
cluding section of this discussion suggests a four­
pronged approach.

tasks is validation of placement statistics in selected
States. Although the study is not complete, results thus
far suggest major errors (overstatements of the number
of placements) similar to those noted here.1 6
0

Quality and level o f service. All parts of ES operations,
and all measurements of productivity, are heavily en­
twined in quality and level o f service considerations. As
noted in other parts of this report, the focus here is on
how quality and level of service issues affect unit labor
requirements and productivity. Many, such as the
helpfulness of employees, are important from the
citizen’s standpoint but have little or no impact on unit
labor requirements.
Many of the quality and level of service factors which
could affect unit labor requirements and productivity
already have been identified. These include the distinc­
tion between agricultural and nonagricultural
placements, the four types of agricultural placements
used in the 1960’s, and the difference between short­
term or intermittent and long-term placements. On the
other hand, services (counseling and testing, for exam­
ple) and the process used to produce these services vary
by jurisdiction.

Services and activities. Another type of measure which
has been used to measure e s performance is the service
unit or activity measure. This includes factors such as
counseling, testing, employer visits, referrals, and the
like. During the 1950’s and early 1960’s, ES budgeted
and accounted for funds based on the services to ap­
plicants and employers.
e s service or activity measures have several attractive
attributes. They are physical, measurable, repetitive,
and, for.many, data are available since 1936. Most im­
portant, they are measures o f work performed which
are largely unaffected by external forces such as the
strength of the economy and the labor market.
Although activity data are readily available, their
completeness and accuracy are not known. Many data
problems noted earlier for placements may apply equal­
ly here.
However, the principal argument against using ser­
vice counts is that they are measures of work perform­
ed, not outputs. Knowing the number of people inter­
viewed, number of follow-ups, and number of file
checks does not tell much about the basic output of the
ES—placing people in jobs. This reasoning, of course,
led to the use of placements as the measure of output.

Table 58. Missions and measures of performance ©f the
Employment Service as suggested by C.K. Fairchild
Mission
Basic labor exchange......................

Mission-based outputs. Neither services provided nor
placements made are entirely satisfactory measures of
es output. Placements are outcomes; services are work
performed or activities.
Fairchild recently examined es missions (goals) and
suggested a series of measures for four missions (table
58). For compliance and enforcement, the number of
registered recipients; for employer technical services,
the number of employers assisted and visited; for sup­
plemental services, applicants registered and number
served; for basic labor exchange, placements. Except
for the basic labor exchange, his recommendations are
measures of output. Futhermore, they are physical,
measurable, and repetitive, and data should be readily
available. They could be combined into a single index by
using labor weights.
The problem remains of how to measure basic labor
exchange outputs. The basic e s role in the labor e x ­
change is to refer individuals. Statistics are available
on the number of referrals but this type of output
measure opens the door to numerous types of statistical

Individuals placed
Total
From target groups
Type of job placements
Nonagriculture
High wage
Long term
By occupation
Placement transactions
Openings filled by type of job
Performance increase on
measures listed above

Supplemental services.............

Applicants registered
Total
Target group
Number served
Any service
Counseling, testing
Referral to job
Referral to other service

Employment technical services . ..

Number of employers assisted
Number of employer visits

Compliance and enforcement . . . .

16 In a recent survey, The Urban Institute, using intensive
0
telephone follow-ups of employers and employees, found the
overstatement to be only 2-5 percent, a dramatic improvement over
the figures noted here (Personal correspondence from John Greiner,
September 30, 1982.)




Measure of performance

80

Number of UI, food stamp, and
welfare recipients registered
with Employment Service

Labor market information............

None proposed

S ource : Charles K. Fairchild, A Performance and Needs Based
Methodology fo r Allocating Employment Service Grants: Final
Report (Cambridge, Mass.: ABT Associates, 1980), p. 9.

gram, such as food stamps or unemployment insurance,
and another part o f the day on another program.
Position data are readily available by State and for
the total e s . They are of sufficient detail for use in com­
puting labor requirements for any of the three output
measures—placements, service and activity, or mis­
sion—discussed in the preceding section.
A potentially confounding issue is the use of
volunteer and “ free” staff. Some e s offices evidently
use many retired aides, Public Service employees (in
past years), work experience interns, and others not
charged under the s e s a accounting system.10 Their
7
overall importance to e s and to ES productivity calcula­
tions is not known. This area needs investigation if
an e s productivity index is to be calculated.

Output indexes need to take into account level and
quality of service when they affect unit labor re­
quirements. However, the information is relatively
sparse. No quality rating system exists such as for the
Unemployment Insurance Service. Nor has research
been done at the national level. Calculation of an es
productivity index requires a systematic examination o f
the effect on unit labor requirements of changes in
quality and level of service.

Labor inputs
Labor dominates es resource inputs. As noted earlier,

about 85 percent o f the ES budget goes for labor.
Although Federal employees provide oversight, State
government employees manage and operate the pro­
grams. No local government employees are involved.
This discussion is restricted to those State employees
who work in the traditional ES program; other sources
fund ES-related operations, such as food stamp and
a f d c certification. Thirty thousand State positions were
funded from 1960 to 1980 to support the traditional
program.
The three labor measures recommended for considera­
tion in calculating State and local government produc­
tivity in chapter III were all employees, all employee
hours, and number of full-time-equivalent employees.
The best source o f es labor data is the State Employ­
ment Security Agency (sesa) accounting system, the
system used for internal office operations as well as
State and Federal management. This system was
automated in 1978 and reports are currently generated
daily, weekly, monthly, quarterly, and yearly, accor­
ding to the need.
Insofar as labor statistics are concerned, data are col­
lected on the hours worked, number of persons, and the
number of positions (full-time-equivalent employment).
However, the only statistics available nationally are the
number of positions.
Statistics on the number of hours are summarized by
each State and could probably be obtained from each
State. But since the number of positions is computed
from the number of hours, any position index would be
identical to the hours index.
Data on the total number of employees are more dif­
ficult to generate; no national or State count is
available. Each State has a master file o f employees
from which these data could probably be summarized;
the effort required to make such a count is not known.
However, such a count would not be very meaningful
since ES personnel often work part of a day on one pro­




Suggested research
The primary problem in measuring e s productivity
is specification o f output. The research strategy sug­
gested is a detailed examination and comparison of four
different approaches. One would focus on placements
to measure output with adjustment for changes in the
level and quality o f service. The second approach,
which has been thoroughly investigated, would focus on
placements with adjustments for externalities. The third
approach would attempt to compute a weighted activity
index. Finally, research should pursue the mission out­
put measures along the lines suggested by Fairchild. In
each case, accuracy and data verification need to be
considered carefully to avoid problems noted earlier.
Suggesting additional research for an area that has
already been examined on numerous occasions may
seem peculiar. However, as a recent review of the
research noted, most research has been narrowly focus­
ed.18 Several approaches should be compared to
0
resolve the problems.
If the output issue can be solved, or at least resolved,
then computing a national e s productivity index and in­
dividual State indexes should be relatively straightfor­
ward. Also, computation o f absolute levels of produc­
tivity, as contrasted with productivity trends, may be
feasible. Such computations can provide additional in­
sight into how productivity varies and how it can be im­
proved. Finally, this discussion has focused primarily
on traditional labor programs, which in the past have
employed about two-thirds of the e s staff. Other e s pro­
grams, such as welfare certification, warrant investiga­
tion too.
17 Personal correspondence from John Greiner, September 30,
0
1982.
18 Stevens and others, “ Specification and Measurement of Pro­
0
ductivity in the uses . ”

81

AppsndlSx
Wymfe@r ®f Stat® and!
L@ea! Governments and!
Numb@r of Employ®®®

reflect the way the data are presented by the Bureau of
the Census, the source of the data.
Often, a few large jurisdictions account for a sizable
proportion of the total employment for that level of
government. Ten States, for example, account for
almost half of all State government employment.

Or e difficulty in collecting data from State and local
governments is the number o f governments that must be
contacted. Tables in this appendix show, for each level
of government, the number of governments and the
number of employees. Data are presented separately for
States, municipalities, school districts, counties, special
districts, and townships. The class sizes in each table

Table A-1. State governments and State government employment, 1979
(Cumulative)
Employees

Percent of total
employment

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

1,860,239
2,735,609
3,338,541
3,668,875

48
71
86
95

Total, 50 ................................

3,869,282

100

States
10
20
30
40

S o u r c e : C o m p u t e d f r o m d a t a in

Public Employment in 1979 ( B u r e a u

o f t h e C e n s u s , 1 9 8 0 ), t a b le 8 , p. 1 4 .

Table A-2. Municipal governments and municipal government employment, 1977
(Cumulative)
Percent of total
employment

Municipalities

Employees

6 ........................................................
24 ......................................................
46 .............................. ........................
64 ......................................................
163 ....................................................
908 ....................................................

522,088
820,905
988,173
1,089,363
1,516,887
2,024,355

20
31
38
42
58
77

Total, 18,878 ..........................

2,623,271

100

S o u r c e : Computed from data in 1977 Census
(Bureau of the Census, 1979), table 18. p. 343.

of Governments—Compendium of Public Employment

Table A-3. School systems and school employment, 1977
(Cumulative)
Districts and
systems

Employees

10 ......................................................
20 ......................................................
40 ......................................................
618 ....................................................
1,639 ..................................................
3,533 ..................................................

622,046
886,520
1,199,377
2,302,840
3,178,251
4,004,456

12
17
23
44
61
76

Total, 16,489 ..........................

5,242,028

100

S o u r c e : C o m p u te d f r o m d a t a in




Public Employment, t a b le
82

Percent of total
employment

2 1 , p. 4 0 5 .

Table A-4. County governments and county government employment, 1977
(Cumulative)
Counties
10
20

Percent of total
employment

Employees
198,063
303,325
449,189
866,167
1,043,746
1,285,601
1,500,970

40 . .
181 .
343 .
679
1,274
Total, 3,040
S o u r c e : C o m p u te d f r o m d a t a in

17
26
49
59
73
85

1,761,242

. .
. .

100

Public Employment, t a b le

11

17, p, 3 1 8 .

Table A-5. Special districts and special district employment, 1977
(Cumulative)
Districts

Employees

Percent of total
employment

10 ......................................................
20 ......................................................
40 ......................................................
499 ....................................................
910 ....................................................
1,724 ..................................................
2,550 ..................................................

98,555
138,274
182,831
274,257
309,746
343,797
359,309

25
34
46
68
77
86
89

Total, 26,010..........................

401,880

100

SOURCE: Computed from data in

Public Employment, table 22, p. 423.

Table A-6. Township governments and township employment, 1975
(Cumulative)
Townships

Employees

Percent of total
employment

961 ....................................................
1,831..................................................

213,260
262,187

59
73

Total, 16,827 ..........................

360,763

100

So u r c e :




Computed from data in

Public Employment, table 19, p. 344.

83

Appendix B. ComparosQoi ©
fi
® tlh© Oiiniiys Classification ©f
(f
Government Fmetfens w
itth)
Standlard! Industrial Classification

Government function
(Bureau of the Census)

SIC industry

Description

Airports...................................

Operation and support of publicly operated
airport facilities.

4582
4583

Airports and flying fields
Airport terminal services

Corrections.............................

Activities pertaining to the confinement and
correction of adults and minors convicted
of criminal offenses. Pardon, probation,
and parole activities are also included.

8361
8399

Juvenile correctional homes
Social services—parole, proba­
tion
Correctional institutions

9223

Local government activities associated with
the production or acquisition and distribu­
tion of electric power to individual con­
sumers.

4911

Generation, transmission, or
distribution

Financial administration........ Activities concerned with tax assessment and
collection, custody and disbursement of
funds, debt management, administration of
trust funds, budgeting, and other govern­
mentwide financial management activities.
This function is not applied to school dis­
trict or special district government.

9311

Tax assessors, budget agencies

Fire protection.........................

Local government fire protection and pre­
vention activities plus any ambulance, res­
cue, or other auxiliary services provided by
the fire protection agency.

9224

Fire departments, fire prevent­
io n offices

Gas supply...............................

Local government activities associated with
the acquisition and distribution of gas sup­
plies to individual consumers.

4924

Natural gas distribution

General control.......................

Judicial, legislative, and governmentwide
administrative agencies of governments.
Includes planning and zoning activities,
central personnel, and administrative serv­
ices, the office of chief executive, legis­
lative activities, and court and courtrelated activities. This function is not ap­
plied to school district or special district
government.

7374
9111
9121
9131

Data processing
Executive offices
Legislative assemblies
Executive and legislative offices
combined
Personnel agencies and person­
nel boards
Courts
Legal counsel and prosecution
Public safety not elsewhere clas­
sified
Zoning boards

Electric pow er.........................




9199
9211
9222
9229
9532

84

Appendix B. Comparison of Bureau of the Census Glassification of Government Functions with
Standard Industrial Classification=Continued
Government function
(Bureau of the Census)
Health

Description

SIC industry

Administration of public health programs,
community and visiting nurse services, im­
munization programs, drug abuse rehabili­
tation programs, health and food inspection
activities, operation of outpatient clinics,
and environmental pollution control activi­
ties.

8081
9431
9641

Outpatient care facilities
Public health agencies, environ­
mental health and immuni­
zation programs
Food inspection

Higher education

State and local government degree-granting
institutions which provide academic training
above grade 12.

8221
8222

Colleges
Juinor colleges

Highways

Activities associated with the maintenance
and operation of streets, roads, sidewalks,
bridges, tunnels, toll roads, and ferries.
Also includes snow activities.

1611

Highways and street constructruction
Bridge, tunnel, and elevated
highway construction
Ferries
Operation of toll roads and
bridges

1622
4452
4784

Government-operated medical care facilities
which provide inpatient care.

8062
8063

General medical hospitals
Mental hospitals

Housing and urban renewal. . . The operation of housing and redevelopment
projects and other activities to promote
or aid housing and community revewal.

6513

Operators of apartment build­
ings
Housing agencies
Community development agen­
cies

Hospitals

9531
9532

Libraries

Libraries operated by local governments for
use by the general public. School and law
libraries are excluded; they are included in
the “local schools” or “higher education”
and “ general control” categories, respec­
tively.

8231

Libraries

Liquor stores

Administration and operation of retail liquor
stores operated by State governments.

5182
5921

Liquor—wholesale
Liquor and beer—retail

Local schools

All activities associated with the operation of
public elementary and secondary schools
and locally operated vocational-technical
schools. Special education programs opera­
ted by elementary and secondary school
systems are also included, as are all ancil­
lary services associated with the operation of
schools, such as pupil transportation and
food service.

8211

Elementary schools, secondary
schools, and vocational high
schools
School buses

Natural resources

Other education




Activities primarily concerned with the conser­
vation and development of natural resources
—forest fire prevention and control, irri­
gation, drainage, land and forest reclama­
tion, fish and game preservation and con­
trol, soil conservation, forestry, agricultural
aids and research, agriculture development
and inspection, and mineral resources acti­
vities.
State government activities relating to the su­
pervision and regulation of public and pri­
vate elementary and secondary schools;
programs and institutions for the training
of blind, deaf, and other handicapped per­
sons; and vocational rehabilitation pro­
grams.

85

4151

0851
0921
0971
1629
4971
9512
9631
9641
8249
9411

Forest management and serv­
ices
Fish hatcheries
Operation of game preserves
Flood control projects
Irrigation system operations
Soil conservation
Irrigation districts
Agriculture extension services
Vocational schools other than
high schools
Administration of educational
programs

Appendix B. Comparison of Bureau of the Census Classification of Government Functions with
Standard Industrial Classification—Continued
Government function
(Bureau of the Census)

Description

SIC industry

Parks and recreation................ Local government activities which include
the operation and maintenance of parks,
playgrounds, swimming pools, public beach­
es, auditoriums, public golf courses, muse­
ums, marinas, botanical gardens, and zo­
ological parks. State government park and
recreation activities are included in the “na­
tural resources” function.

0782
4469
7992
7999
8411
8421

All activities concerned with the enforcement
of law and order, including coroners’ of­
fices, police training academies, investiga­
tion bureaus, and local jails, “lockups,” or
other detention facilities not intended to
serve as correctional facilities.

9221
9223

Police departments
Jails

Sanitation other than sewerage. Refuse collection and disposal, operation of
sanitary landfills, and street cleaning ac­
tivities.

4212
4953
4959

Garbage and refuse collection
Refuse systems
Sanitary services not elsewhere
classified
Sanitary engineering agencies
Sanitary districts

Police protection.....................

9512

9511
9631
Sewerage .................................

Provision, maintenance, and operation of
sanitary and storm sewer systems and sew­
age disposal and treatment facilities.

1623
1629
4952
7699
9511

Lawn and garden services
Marinas
Golf courses
Swimming pools and beaches
Museums, noncommercial
Botanical gardens and zoolog­
ical gardens
Recreational program admini­
stration

Sewerage collection and disposal
line construction
Sewage treatment plant con­
struction
Sewerage systems
Sewer cleaning and rodding
Sanitary engineering agencies

Social insurance.......................

Administration and conduct of social insur­
ance programs. For State governments and
the government of the District of Columbia,
these activities include unemployment com­
pensation and worker compensation pro­
grams, and work/study programs.

7361
8331
9441

Employment agencies
Job training
Unemployment insurance of­
fices

Transit.....................................

Activities relating to the operation and main­
tenance of public mass transit systems (e.g.,
bus, subway, surface rail, and street rail­
road systems). Elementary and secondary
school transportation systems are included
in the “local schools” function.

4111

Busline operation and subway
operation
Local passenger transportation
not elsewhere classified

4119

Local government activities associated with
the production or acquisition of water and
distribution to the public.

4941

Water supply

Water transportation.............. The provision, operation, and support of
canals and other waterways, harbors, docks,
wharves, and other related marine terminal
facilities.

4463
4464

Docks, terminal operation
Canal operation

Welfare.....................................

8321
8351
8361
8399

Public welfare services
Day care services
Residential care
Social service information ex­
changes
Public welfare administration

Water supply...........................

Activities such as the administration of va­
rious public assistance programs for the
needy, operation of homes for the elderly,
indigent care institutions, and programs
which provide payments for medical care
and other services for the needy. Health
care and hospital services provided directly
by a government, however, are included in
the “ health” and “hospitals” functions
rather than here.

So u r c e : 1977 Census o f Governments—Compendium o f
Public Employment (Bureau of the Census, 1979), pp. 459-62,




9441

and Standard Industrial Classification Manual, 1972 (Office
of Management and Budget, 1972).
86

Appendix C Isrg® Electric UtSiutnss
L

Data for the following State and local government utilities were combined to compute the pro­
ductivity index for “ large utilities” in chapter IV. Utilities are ranked in order of size (kilowatt
hour sales to ultimate customers).

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.

Los Angeles Water and Power Department, Calif.
New York Power Authority, N.Y.
Memphis Light, Gas, and Water, Tenn.
Salt River Power, Ariz.
Nashville Electric Service, Tenn.
Seattle City Light, Wash.
San Antonio City Public Service, Tex.
Jacksonville Electric Authority, Fla.
Sacramento Municipal District, Calif.

18.
19.
20.
21.
22.
23.
24.
25.
26.

Eugene Water and Electric Board, Oreg.
Orlando Utilities Commission, Fla.
Nebraska Public Power District, Nebr.
Lansing Board of Water and Light, Mich.
City of Colorado Springs, Colo.
Lincoln Electric System, Nebr.
Santa Clara Municipal Electric Department, Calif.
Imperial Irrigation District, Calif.
Grays Harbor County, Wash.

Chattanooga Electric Power, Tenn.
Omaha Public Power District, Nebr.
Snohomich County Public Utility District, Wash.
Tacoma City Light, Wash.
Austin Electric Department, Tex.
Cowlitz County Public Utility District, Wash.
South Carolina Public Service Authority, S.C.
Clark County Public Utility District, Wash.

27.
28.
29.
30.
31.
32.
33.

Garland Power and Light Department, Tex.
Grant County Public Utility District, Wash.
Benton County Public Utility District, Wash.
Modesto Irrigation District, Calif.
Central Lincoln Public Utility District, Oreg.
Tallahassee Electric Department, Fla.
Lakeland Electric Department, Fla.




87

@©neral

Mark, Jerome A ., “ Measuring Productivity In Service
Industries,” Monthly Labor Review , June 1982,
pp. 3-8.

A Demonstration o f Comparative Productivity Mea­
surement. Denver: Denver Regional Council of
Governments, December 1978.

Mark, Jerome A ., “ Measuring Productivity in Govern­
ment—Federal, State, and Local,” Public Product­
ivity Review, March 1981, p. 21.

Baumol, William J., “ Macroeconomics o f Unbalanced
Growth: The Anatomy of Urban Crisis,” American
Economic Review, June 1967, pp. 415-26.

Mark, Jerome A ., “ Progress in Measuring Productivity
in Government,” Monthly Labor Review , De­
cember 1972, pp. 3-6.

Bradford, D.F.; Malt, R.A.; and Oates, W .E., “ The
Rising Cost of Local Public Services: Some
Evidence and Reflections,” National Tax Journal,
XXII, No. 2 (June 1969), pp. 185-202.

Mundell, Marvin E., Measuring and Enhancing the

Productivity o f Service and Government Organiza­
tions. Hong Kong: Nordica International Limited,

Burkhead, Jesse and Hennigan, Patrick J., “ Product­
ivity Analysis: A Search for Definition and Order,”
Public Administration Review, Jan./Feb. 1978, pp.
34-40.

1975.
National Research Council, Measurement and Inter­
pretation o f Productivity. Washington: National
Academy of Sciences, 1979.

Comparative Performance Measures for Municipal Serv­
ices, five volumes. Raleigh, N. C.: Research

“ New York State Survey: Draft Summary and Conclu­
sions.” New York: Council on Municipal Perfor­
mance, November 1978.

Triangle Institute, December 1978.

Federal Actions to Support State and Local Govern­
ment Productivity Improvement. Washington: Na­

Ridley, Clarence and Simon, Herbert, Measuring
Municipal Activities. Chicago: International City
Manager’s Association (now International City
Management Association), 1937.

tional Productivity Council, August 1979.
Fukuhara, Rockham, The Status o f Local Government
Productivity. Washington: The International City
Management Association, March 1977.

Ross, John P. and Burkhead, Jesse, Productivity in the
Local Government Sector. Lexington, Mass.: Lex­
ington Books, 1974.

Hall, John R., Factors Related to Local Government
Use o f Performance Measurement. Washington:
The Urban Institute, 1978.

Siegfried, John J., “ Public Sector Productivity,” A t­
lanta Economic Review, Vol. 27, No. 5, Sept./Oct.
1977, pp. 29-34.

Hamilton, Edward K., “ Productivity: The New York
City Approach,” Public Administration Review,
N ov./D ec. 1972, pp. 784-95.

Spann, Robert M., “ Rates of Productivity Change and
the Growth o f State and Local Expenditures,” in
Thomas E. Borcherding, ed., Budgets and
Bureaucrats. Durham: Duke University Press,
1977.

Hatry, Harry and others, The Challenge o f Productivity
Diversity. Washington: The Urban Institute, 1972.
Hayes, Frederick O’R, Productivity in Local Govern­
ment. Lexington, Mass.: Lexington Books, 1977.

State and Local Government Productivity Improve­
ment: What is the Federal Role? Washington: U.S.

Hulten, Charles R., Productivity Change in State and
Local Governments. Washington: The Urban In­
stitute, 1981.

General Accounting Office, Dec. 6, 1978.
Stein, Herbert and Mark, Jerome A ., The Meaning and
Measurement o f Productivity, Bulletin 1714. Wash­
ington: Bureau of Labor Statistics, 1971.

“ Implementing a Productivity Program: Points to Con­
sider.” Washington: Joint Financial Management
Improvement Program, March 1977.

The Federal Role in Improving Productivity—Is the Na­
tional Center for Productivity and Quality o f
Working Life the Proper Mechanism? Washington:

Improving Productivity in State and Local Govern­
ment. New York: Committee for Economic

U.S. General Accounting Office, May 1978.

Development, March 1976.



88

The Status o f Productivity Measurement in State
Government: An Initial Examination. Washington:

Mark, Jerome A ., “ Industry Indexes o f Output Per
Man-Hour,” Monthly Labor Review, November
1962, pp. 1269-73.

The Urban Institute, 1975.
U.S. Bureau of Labor Statistics, Methodology fo r Pro­
jections o f Industry Employment to 1990, Bulletin
2036. Washington: U.S. Government Printing Of­
fice, 1980.

National Research Council, Measurement and Inter­
pretation o f Productivity. Washington: National
Academy of Sciences, 1979.
Ross, John P. and Burkhead, Jesse, Productivity in the
Local Government Sector. Lexington, Mass.: Lex­
ington Books, 1974.

U.S. Congress, Joint Economic Committee, Producti­
vity in the Federal Government. Washington: U.S.
Government Printing Office, 1979, p. 7.

U.S. Department of Housing and Urban Development,

Shulmann, Martha A ., “ Alternative Approaches for
Delivering Public Services,” Urban Data Service
Reports, Vol. 14, No. 10. Washington: Interna­
tional City Management Association, October
1982.

Productivity Improvement for State and Local
Government. Washington: U.S. Government Print­

The Federal Role in the Federal System: The Dynamics
o f Growth . Washington: Advisory Commission on

U.S. Council of Economic Advisers, Economic Report
o f the President. Washington: U.S. Government
Printing Office, January 1979.

ing Office, 1981.

Intergovernmental Relations, December 1980, p. 8.

U.S. National Center for Productivity and Quality of
Working Life, Final Report. Washington: U.S.
Government Printing Office, 1978.

Tomazinis, Anthony R., Productivity, Efficiency, and
Quality in Urban Transportation Systems. Lex­
ington, Mass.: D.C. Heath and Company, 1975.

U.S. Office of Management and Budget, Special

Triplett, Jack E., “ Robert Gordon’s Approach to Price
Measurement.” BLS Working Paper No. 101.
Washington: Bureau of Labor Statistics, April
1980.

Analyses: Budget o f the United States Government,
Fiscal Year 1979. Washington: U.S. Government
Printing Office, 1979.

Methodology

U.S. Bureau o f Labor Statistics, BLS Handbook o f
Methods, Bulletin 2134-1. Washington: U.S.
Government Printing Office, 1982.

Bradford, D.F.; Malt, R.A.; and Oates, W.E., “ The
Rising Cost of Local Public Services: Some
Evidence and Reflections,” National Tax Journal,
Vol. XXII, No. 2 (June 1969), pp. 185-202.

U.S. Bureau of the Budget, Measuring Productivity o f
Federal Government Organizations. Washington:
U.S. Government Printing Office, 1964.

Burkhead, Jesse and Hennigan, Patrick J., “ Produc­
tivity Analysis: A Search for Definition and
Order,” Public Administration Review , Jan./Feb.
1978, p. 34-40.

Usilaner, Brian and Soniat, Edwin, “ Productivity
Measurement,” in George Washnis, ed., Product­
ivity Improvement Handbook. New York: John
Wiley, 1981.

Fielding, Gordon J. and others, Development o f Per­
formance Indicators for Transit. Irvine: University
of California, 1977.

Whitaker, Gordon P.and others, Basic Issues in Police
Performance. Washington: U.S. National Institute
o f Justice, 1982.

Fisher, Franklin M. and Shell, Karl, The Economic
Theory o f Price Indices. New York: Academic
Press, 1972.

Eleetrie power

Fisk, Donald M., “ Pilot Study Measures State, Local
Electric Utilities,” Monthly Labor Review, Decem­
ber 1981, pp. 45-7.

Axelrod, Howard J., “ Measuring Electric Utility Pro­
ductivity,” in Walter L. Balk, ed., Public Utility
Productivity . Albany: New York State Department
of Public Services, 1975.

Fox, William, Size Economies in Local Government
Services: A Review. Washington: U.S. Department
of Agriculture, Economics, Statistics, and Cooper­
ative Service, 1980.

Barzel, Yoram, “ Productivity in the Electric Power In­
dustry— 1929-1955,” Review o f Economics and
Statistics, November 1963, pp. 395-408.

Greytak, David; Phares, Donald; and Morely, Elaine,

Capron, William M., Technology Change in Regulated
Industries. Washington: The Brookings Institution,
1971.

Municipal Output and Performance in New York
City. Lexington, Mass.: Lexington Books, 1976.
Hjerppe, Reino T., “ The Measurement of Real Output
o f Public Sector Services,” The Review o f Income
and Wealth, June 1980, pp. 237-50.



Directory o f Electric Utilities, annual. New York:
McGraw-Hill Publishing Company.
89

Oragonette, Joseph E. and Jaynes, Philip W., “ Output
per Man-Hour, Gas and Electric Utilities,” Monthly
Labor Review, January 1965, pp. 34-39.

U.S. Bureau of the Census, 1977 Census o f Govern­

ments—Compendium

o f Public Employment.

Washington: U.S. Government Printing Office,
1979.

Electric Power and Government Policy. New York:
Twentieth Century Fund, 1948.
Gould, Jacob Martin, Output and Productivity in the
Electric and Gas Utilities— 1899-1942. New York:
National Bureau of Economic Research, 1946.

U.S. Bureau of Labor Statistics, Productivity Measures
for Selected Industries— 1954-81, Bulletin 2155.
Washington: U.S. Government Printing Office,
1982.

“ Indexes of Output Per Man-Hour, Gas and Electric
Utilities Industry, 1932-1962.” Unpublished study.
Washington: Bureau of Labor Statistics, April
1964.

U.S. Department of Energy, Energy Information Ad­
ministration, Statistics o f Publicly Owned Electric
Utilities in the United States, annual report.
Washington: U.S. Department of Energy.

M o , William, Electric Utilities—Costs and Perform­
ance. Pullman, Washington: State University Press,
1961.

U.S. Office of Management and Budget, Standard In­
dustrial Classification Manual. Washington: U.S.
Government Printing Office, 1972.

Kendrick, John W., “ Efficiency Incentives and Cost
Factors in Public Utility Automatic Reserve Adjust­
ment Clauses,” Bell Journal o f Economics, Spring
1975, pp. 299-313.

Wilson, J.W. and Associates, The Measurement o f
Electric Utility Productivity, Vols. I and II.
Washington: National Bureau of Standards, 1980.

Alcoholic beverages

Kendrick, John W., Postwar Productivity Trends in the
United States. New York: National Bureau of
Economic Research, 1973.

Annual Statistical Review, annual issues. Washington:
Distilled Spirits Council of the United States.
Barker, Twiky W., Jr., “ State Liquor Monopolies in
the United States.” Thesis submitted to the Univer­
sity of Illinois, 1955.

Kendrick, John W., “ Some Productivity Issues in
Regulated Industries,” in Walter L. Balk, ed.,
Public Utility Productivity. Albany: New York
State Department of Public Services, 1975.

Ohio Department of Liquor Control, Annual Report
published yearly.

Pace, Joseph B ., “ Relative Efficiency in the Electric
Utility Industry.” Thesis submitted to the Universi­
ty of Michigan, 1970.

“ Preliminary Evaluation of the Feasibility of Convert­
ing the Pennsylvania Liquor Board’s Operations to
Licensed Private Enterprise.” Report by Laventhoe, Krikstein, Horwath and Horwath to Penn­
sylvania Liquor Control Board, Sept. 15, 1973.

Public Power. January/February issue each year con­
tains statistical information and directory.
Smith, J. Edward, The Measurement o f Electric Utility
Efficiency. Washington: National Association of
Regulatory Utility Commissioners, 1975.

Public Revenues From Alcohol Beverages, annual is­
sues. Washington: Distilled Spirits Council of the
United States.

Smyth, David, J., “ Short-Run Employment Fluctua­
tions in Public Utilities.” Thesis submitted to the
Claremont Graduate School, Claremont, Calif.

Retail Outlets for the Sale o f Distilled Spirits— 1978.
Washington: Distilled Spirits Council of the United
States, 1979.

Stevenson, Rodney E., “ Productivity in the Private
Electric Utility Industry,” in Walter L. Balk, ed.,
Public Utility Productivity. Albany: New York
State Department o f Public Services, 1975.

Revenue Generating Capacity o f Pennsylvania’s Liquor
Control System. Office of the Budget, State of Penn­
sylvania, March 1978.

Stevenson, Rodney E., “ Regulating for Efficiency in
the Public Utility Industry,” in Walter L. Balk, ed.,
Public Utility Productivity. Albany: New York
State Department o f Public Services, 1975.

State of Maine, Bureau o f Alcoholic Beverages, Annual
Report published yearly.

Summary o f State Laws and Regulations Relating to
Distilled Spirits. Washington: Distilled Spirits

Turvey, Ralph, Optimal Pricing and Investment in Elec­
tricity Supply. Cambridge, Mass.: The MIT Press,
1968.
U.S. Bureau o f the Census, 1977 Census o f Govern­

Council of the United States, 1977.
Vermont Liquor Control
published yearly.

ments—Compendium o f Government Finances.

Virginia Department of Alcoholic Beverage Control,
Annual Report published yearly.

Washington: U.S. Government Printing Office,
1979.



Board, Annual Report

90

West Virginia Alcohol Beverage Control Commission,
Annual Report published yearly.

Millions Can be Saved by Improving the Productivity o f
State and Local Governments Administering Fed­
eral Income Maintenance Assistance Programs.

U.S. Bureau of the Census, 1977 Census o f Govern­

Washington: U.S. General Accounting Office, June
5, 1981.

ments—Compendium o f Government Finances.
Washington: U.S. Government Printing Office,
1979.

Oregon Employment Division, 41st Annual Report.
Salem: State of Oregon, 1979.

U.S. Bureau of the Census, 1977 Census o f Govern­

ments—Compendium

o f Public Employment.

Oregon Employment Division, Financing the Oregon
Unemployment Insurance Program. Salem: State
of Oregon, 1979.

Washington: U.S. Government Printing Office,
1979.
U.S. Bureau of the Census, Public Employment, annual
issues. Washington: U.S. Government Printing Of­
fice.

“ Report on the Analysis of Initial Claims and Wage
Record Activities by the Operational Improvement
and Cost Equalization Project.” Washington: U.S.
Unemployment Insurance Service, 1979.

U.S. Bureau of the Census, State Government Finances,
annual issues. Washington: U.S. Government Print­
ing Office.

“ SESA Accounting System Accounting Manual—
Report Utilization Guide.” Washington: U.S.
Employment and Training Administration, Oc­
tober 1978.

U.S. Office of Management and Budget, Standard In­
dustrial Classification Manual. Washington: U.S.
Government Printing Office, 1972.

Unemployment Compensation: Final Report, National

Yearbook,

annual issues. Washington: National
Alcoholic Beverage Control Association, Inc.

Commission on Unemployment Compensation,
1980, p. 128.

Unemployment Compensation: Studies and Research,
ym pl®yinnie[nit imsyran©®
i©m
Basic Structure o f A Federal-State Unemployment In­
surance Program and Related Supporting Provi­
sions. Arlington, Va: National Commission on

three volumes. Washington: National Commission
on Unemployment Compensation, July 1980.
“ Unemployment Insurance Claims,” Washington: U.S.
Employment and Training Administration, July 10,
1980.

Unemployment Compensation, November 1979.
“ Clarification of UI Budget Workload Definitions.”
Washington: U.S. Unemployment Insurance
Service, Feb. 1, 1980.

Unemployment Insurance—Inequities and Work Disin­
centives. Washington: U.S. General Accounting

“ Development and Utilization of the Cost Model
Management System in the Unemployment In­
surance Program.” Washington: U.S. Employment
and Training Administration, May 1979.

Unemployment Insurance—Need to Reduce Unequal
Treatment o f Claimants and Improve Benefit Pay­
ment Controls and Tax Collections. Washington:

Office, 1979.

U.S. General Accounting Office, 1978.

Diefenbach, Donald L., Financing America’s Unem­
ployment Compensation Program. Washington:
U.S. Government Printing Office, 1979.

Unemployment Insurance Quality Appraisal Results.
Washington: U.S. Employment and Training Ad­
ministration. Report published annually.

Employment and Training Report o f the President, an­
nual reports. Washington: U.S. Government Prin­
ting Office.

U.S. Bureau of the Census, 1977 Census o f Govern­

43rd Annual Report. Oklahoma City: Oklahoma

Washington: U.S. Government Printing Office,
1979.

ments—Compendium o f Government Finances.

Employment Security Commission, 1980.

U.S. Bureau of the Census, Statistical Abstract o f the
United States, annual issues. Washington: U.S.
Government Printing Office.

Handbook o f Unemployment Insurance Financial Data
1938-1976. Washington: U.S. Employment and
Training Administration, 1978.

U.S. Employment and Training Administration, Sum­

Holen, Arlene and Horowitz, Stanley A ., The Effect of

mary o f Employment Security Statistical Reports.

Unemployment Insurance and Eligibility Enforce­
ment on Unemployment. Arlington, Va: Public

.Washington: U.S. Government Printing Office,
1977.

Research Institute, April 1974.



91

Fox, William F., Size Economies in Local Government
Services: A Review. Washington: U.S. Department
of Agriculture, 1980.

Manpower Administration, Handbook o f
Unemployment Insurance Financial Data—1938
-1970. Washington: U.S. Department of Labor,

U .S.

1971.

Gueron, Judith M., “ Economies of Solid Waste Handl­
ing and Government Intervention,” in Selma
Mushkin, ed., Public Prices for Public Products.
Washington: The Urban Institute, 1972.

U.S. Manpower Administration, Summary Tables o f

Unemployment Insurance: Program Statistics
1970-1971. Washington: U.S. Government Print­
ing Office, 1973.

Hatry, Harry P. and Fisk, Donald M., The Challenge of

U.S. Office of Management and Budget, Standard In­
dustrial Classification Manual. Washington: U.S.
Govenment Printing Office, 1972.

Productivity Diversity; Measuring Solid Waste Col­
lection Productivity, (Part II). Washington: The
Urban Institute, 1972.

U.S. Unemployment Insurance Service, Twenty Years
o f Unemployment Insurance in the USA— 19351955. Washington: U.S. Government Printing Of­
fice, 1955.

Hatry, Harry P. and others, How Effective Are Your
Community Services? Washington: The Urban In­
stitute, 1980.
Hirsch, Werner Z., “ Cost Functions of an Urban
Government Service: Refuse Collection,” Review
o f Economics and Statistics, February 1965, pp.
87-92.

Virginia Employment Commission, Annual Report for
the Year-1977. Richmond: State of Virginia, 1978.

Solid west® ©©liaetiom and disposal
American Public Works Association, Municipal Refuse
Disposal. Chicago: Public Administration Service,

Johns Hopkins University, Mathematical Modeling o f
Solid Waste Collection Policies. U.S. Public Health
Service, Public Health Publication No. 2030.
Washington: U.S. Government Printing Office,
1970.

1970.
American Public Works Association, Solid Waste Col­
lection Practices. Chicago: Public Administration
Service, 1975.

Kemper, Peter and Quigley, John M., The Economics
o f Refuse Collection. Cambridge, Mass.: Ballinger
Publishing Company, 1976.

Berenyi, Eileen, “ Solid Waste Disposal,” in George J.
Washnis, ed., Productivity Improvement Hand­
book. New York: John Wiley, 1981.

Los Angeles County, “ A Report to the Directors o f the
County Sanitation Districts.” Los Angeles, 1955.

Black, R.J. and others, 1968 National Survey o f Com­

munity Solid Waste Practices; An Interim Report.

Mark, David, Evaluation Policy—Related Research in

Cincinnati: U.S. Bureau o f Solid Waste Manage­
ment, 1968.

the Field o f Municipal Solid Waste Management.
Cambridge, Mass.:
Technology, 1974.

Clark, Robert M., “ Measures of Efficiency in Solid
Waste Collection,” Journal o f Environmental
Engineering Division, American Society of Civil
Engineers, August 1973, pp. 447-59.

National Center for Resource Recovery, Municipal
Solid Waste Management. Lexington, Mass.: Lex­
ington Books, 1973.

Clark, Robert M. and Gillean, James I., “ Solid Waste
Collection: A Case Study,” Operations Research
Quarterly, Vol. 28, No. 4, pp. 795-806.

Perkins, Ronald A ., “ Satellite Vehicle Systems for
Solid Waste Collection, Evaluation and Applica­
tion.” Cincinnati: U.S. Environmental Protection
Agency, 1971.

Citizens Budget Commission, “ Reducing Refuse Col­
lection Costs in New York City.” New York:
Citizens Budget Commission, 1972.

Ralph Stone and Company, A Study o f Solid Waste

Systems Comparing One-Man with Multi-man
Crews; Final Report. Public Health Service Pub­

Columbia University Graduate School of Business,

Evaluating Residential Refuse Collection Costs: A
Framework for Local Government. Washington:

lication No. 1892. Washington: U.S. Government
Printing Office, 1969.

Public Technology, Inc., 1978.

Rawn, A.M ., “ Report Upon the Collection and
Disposal of Refuse in County Sanitation Districts.”
Los Angeles, 1951.

Feldman, Stephen L., “ Waste Collection Services: A
Survey o f Costs and Pricing,” in Selma Mushkin,
ed., Public Prices for Public Products. Wash­
ington: The Urban Institute, 1972.

Reindi, John, “ Interrelationships Within the Solid
Wastes System,” Solid Waste Management, April
1977, pp. 22-23 and 54-59.

Flintoff, Frank and Millard, Ronald, Public Cleansing.
London: McClaren and Sons, 1969.



Massachusetts Institute of

92

Rogers, Peter A. and Bellenger, Geoffrey J., “ Fly and
Economic Evaluation of Urban Refuse Systems,”
California Vector Views, May 1967.

Clark, Robert M., “ Cost and Pricing Relationships in
Water Supply,” Proceedings o f the American
Society o f Civil Engineers, Vol. 102, No. EE2, April
1979, pp. 361-73.

Rubel, Fred N ., Incineration o f Solid Wastes. Park
Ridge N.J.: Noyes Data Corporation, 1974.

Clark, Robert M., “ Labor Wage Rates, Productivity,
and the Cost o f Water Supply,” Journal o f the
American Water Works Association, July 1979, pp.
364-68.

Savas, E.S., The Organization and Efficiency o f Solid
Waste Collection. Lexington, Mass.: Lexington
Books, 1977.

Clark, Robert M., “ Small Water Systems: Role of
Technology,” Journal o f the Environmental
Engineering Division, American Society of Civil
Engineers, Vol. 106, No. EE1, Proceedings Paper
15181, February 1980, pp. 19-35.

Savas, E.S. and Stevens, Barbara J., “ Solid Waste Col­
lection,” in George J. Washnis, ed., Productivity
Improvement Handbook. New York: John Wiley,
1981.
Solid Waste Management Advisory Group, Oppor­

tunities for Improving Productivity in Solid Waste
Collection. Washington: National Commission on

Clark, Robert M ., “ The Safe Drinking Water Act: Its
Implications for Planning,” in David Holz and
Scott Sebastian, eds., Municipal Water Systems: A

Productivity, 1973.

Challenge fo r

Sorg, Thomas J. and Hickman, H. Lanier, Sanitary
Landfill Facts. Washington: U.S. Department of
Health, Education, and Welfare, 1970.

Clark, Robert M., “ Water Supply Regionalization: A
Critical Evaluation,” Journal o f the Water

U.S. Bureau o f the Census, “ Classification Manual,
Government Finances,” revised March 1977, un­
published report.

Resources Planning and Management Division,
American Society o f Civil Engineers, Proceedings
Paper 14842, September 1979, pp. 279-94.

U.S. Bureau o f the Census, 1977 Census o f Govern­

ments—Compendium o f Government Finances.

Clark, Robert M. and Goddard, Haynes G., “ Cost and
Quality o f Water Supply,” Journal o f the
American Water Works Association, Vol. 69,
January 1977, pp. 13-15,

Washington: U.S. Government Printing Office,
1979.
U.S. Bureau o f the Census, 1977 Census o f Govern­
ments—Compendium o f Public Employment. Wash­
ington: U.S. Government Printing Office, 1979.

Clark, Robert M. and Stevie, Richard G. “ Meeting the
Drinking Water Standards: The Price o f Regula­
tion,” in Clifford S. Russell, ed., Safe Drinking
Water: Current and Future Problems. Washington:
Resources for the Future, 1978.

U.S. Council on Environmental Quality, Environmen­
tal Quality— 1977. Washington: U.S. Government
Printing Qffice, 1977.

Clark, Robert M. and others, The Cost o f Removing

U.S. Council on Environmental Quality, Environmen­
tal Quality— 1979. Washington: U.S. Government
Printing Office, 1979.

Chloroform and Other Trihalomethanes from
Drinking Water Supplies. Cincinnati: U.S. En­
vironmental Protection Agency, 1977.

U.S. Office o f Management and Budget, Standard In­
dustrial Classification Manual. Washington: U.S.
Government Printing Office, 1972.

Clark, Robert M.; Gillean, James I.; and Adams, Kyle,

Renovated Wastewater as a Supplementary Source
fo r Municipal Water Supply: An Economic Evalua­
tion. Cincinnati: U.S. Environmental Protection

University of California, An Analysis o f Refuse Collec­
tion and Sanitary Landfill Disposal, Bulletin No. 8,
Series 37. Berkeley: University of California,
December 1952.

Agency, 1976.
Clark, Robert M., Machisko, John A. and Stevie,
Richard G., “ Cost of Water Supply: Selected Case
Studies,” Journal o f the Environmental Engineer­
ing Division, American Society o f Civil Engineers,
Proceedings Paper 14365, February 1979, pp.
89-100.

Weiss, Samuel, Sanitary Landfill Technology. Park
Ridge, N.J.: Noyes Data Corporation, 1974.
Young, Dennis, How Shall We Collect the Garbage?
Washington: The Urban Institute, 1972.

Drinking water supply

Clark, Robert M.; Stevie, Richard G.; and Trygg,
Gregory D., “ An Analysis of Municipal Water
Supply Costs,” Journal o f the American Water
Works Association, Vol. 70, October 1978, pp.
543-47.

Additional Federal A id fo r Urban Water Distribution
Systems Should Wait Until Needs are Clearly
Established. Washington: U.S. General Accounting
Office, Nov. 24, 1980.



Urban Resource Management.

Bloomington: Indiana University Press, 1978.

93

States’ Compliance Lacking in Meeting Safe Drinking
Water Standards. U.S. General Accounting Office,

Clark, Robert M.; Stevie, Richard G.; and Trygg,
Gregory D., The Cost o f Municipal Water Supply:
A Case Study . Cincinnati: U.S. Environmental
Protection Agency, 1976.

Mar. 3, 1982.
Stevie, Richard G.; Clark, Robert M.; and Adams, Jef­
frey Q., Managing Small Water Systems: A Cost
Study, Vols. I and II. Cincinnati: U.S. En­
vironmental Protection Agency, 1979.

Commissioner o f Labor, Fourteenth Annual Report
— 1899. Washington: U.S. Government Printing
Office, 1900.

Temple, Barker, and Sloane, Inc., Survey o f Operating

“ Comparisons of Cost, Manpower Utilization, and
Flow in Operation and Maintenance of Investor
Owned Water Companies and Municipal Waste
Water Systems.” U.S. Environmental Protection
Agency, June 5, 1979, unpublished report.

and Financial Characteristics o f Community Water
Systems. Washington: U.S. Environmental Protec­
tion Agency, 1977.
Thomas, John R., Statistical Summary o f Water Supply
and Treatment Practices in the United States, PHS
Publication No. 301. Washington: U.S. Govern­
ment Printing Office, 1953.

Ford, J.L. and Warford, J.J., “ Cost Functions for the
Water Industry,” The Journal o f Industrial
Economics, November 1969, pp. 53-63.

U.S. Bureau of the Census, 1977 Census o f Govern­

ments—Compendium o f Government Finances.

Fox, William F., Size Economies in Local Government
Services: A Review. Washington: U.S. Department
of Agriculture, 1980.

Washington: U.S. Government Printing Office,
1979.
U.S. Bureau of the Census, 1977 Census o f Govern­
ments—Compendium o f Public Employment .
Washington: U.S. Government Printing Office,
1979.

Gillean, James I. and others, The Cost o f Water Supply
and Water Utility Management, Vol. II. Cincinnati:
U.S. Environmental Protection Agency, 1977.
Glass, Andrew C. and Jenkins, Kenneth H ., Municipal

U.S. Bureau of the Census, Government Finances, an­
nual issues. Washington: U.S. Government Prin­
ting Office.

Water Facilities in the United States, January 1,
1958, PHS Publication 1039. Washington: U.S.
Government Printing Office, 1963.

U.S. Bureau of the Census, Public Employment, annual
issue. Washington: U.S. Government Printing Of­
fice.

Goddard, Haynes G.; Stevie, Richard G.; and Trygg,
Gregory B ., Planning Water Supply: Cost—Rate
Differentials and Plumbing Permits. Cincinnati:
U.S. Environmental Protection Agency, 1978.

U.S. Bureau of the Census, Statistical Abstract o f the
United States, selected years. Washington: U.S.
Government Printing Office.

Gumerman, Robert C., and others, Estimating Water
Treatment Costs, four volumes. Cincinnati: U.S.
Environmental Protection Agency, 1979.

U.S. Office of Management and Budget, Standard In­
dustrial Classification Manual. Washington: U.S.
Government Printing Office, 1972.

Hanke, Steve H., “ Pricing Urban Water,” in Selma
Mushkin, ed., Public Prices for Public Products.
Washington: The Urban Institute, 1972.

BMass transit
American Public Transit Association, “ Monthly Tran­
sit Ridership,” monthly, Washington.
American Public Transit Association, Transit Fact
Book, annual, Washington.

Hatry, Harry P., and others, Efficiency Measurement

fo r Local Government Services: Some Initial Sug­
gestions. Washington: The Urban Institute, April
1978.

American

Hines, Lawrence G., “ The Long Run Cost Function of
Water Production for Selected Wisconsin Com­
munities,” Land Economics, Vol. 45, February
1969, pp. 133-40.

Public

Transit

A ssociation,

Transit

Operating Report, annual, Washington.
Bennewitz, Eckart, “ Mass Transit,” in George J.
Washnis, ed., Productivity Improvement Hand­
book. New York: John Wiley, 1981.

Operating Data for Water Utilities— 1976. Denver:

B.M. and Seward, T., Productivity in
Transportation. Cambridge, England: Cambridge

Deakin,

American Water Works Association, 1979.

Operating Data fo r Water Utilities—1970 and 1965.

University Press, 1969.

Denver: American Water Works Association, 1970.

Fielding, Gordon I., and Glauthier, Roy E., Distribu­

Small System Water Treatment Symposium. Wash­

tion and Allocation o f Transit Subsidies in Califor­
nia. Irvine: University of California, September

ington: U.S. Environmental Protection Agency,
1979.



1976.
94

Fielding, Gordon J. and Glauthier, Roy E., Obstacles to

U.S. Bureau o f the Census, Public Employment, annual
reports. Washington: U.S. Government Printing
Office.
U.S. Bureau of Labor Statistics, Employment and
Earnings, monthly periodical. Washington: U.S.
Government Printing Office.

Comparative Evaluation o f Transit Performance.
Irvine: University of California,
Transportation Studies, 1977.

Institute

of

Glauthier, Roy E., Evaluating the Performance o f Rail
Transit: B A R T . Irvine: University o f California, In
stitute of Transportation Studies, 1977.
Lieb,

U.S. Bureau o f Labor Statistics, Productivity Measures
for Selected Industries, 1954-81, Bulletin 2155.
Washington: U.S. Government Printing Office,
1982.

Massachusetts Bay Transportation Authority and
Tidewater Transportation District Commission,

Bus Service Evaluation Procedures: A Review,

U.S. Congress, Joint Economic Committee, Productivi­
ty in Urban Transit. Hearings, 93rd Congress,
Second Session, July 7, 1974.

1979.

U.S. Department of Transportation, A Directory o f

Robert C., Labor in the Transit Indus­
try. Washington: U.S. Department of Transporta­
tion, 1976.

Regularly Scheduled, Fixed Route, Local Public
Transportation Service. Washington: U.S. Depart­

Meyer, John R., and Gomez-Ibanez, Jose A ., Measure­

ment and A nalysis o f Productivity in Transporta­
tion Industries. Cambridge, Mass.: Harvard

ment of Transportation, July 1979.
U.S. Department of Transportation, Urban Mass
Transportation Administration, “ Comparing the
Efficiency of Privately- and Publicly-Owned Bus
Systems.” Draft paper prepared by Cindy Bur­
bank. Washington: 1976.

University, Department o f City and Regional Plan­
ning, 1975.

Proceedings o f the First National Conference on Transit
Performance. Washington: Public Technology,
Inc., 1978.

U.S. Department of Transportation, National Trans­
portation Statistics, annual issues. Washington:
U.S. Government Printing Office.

Roess, Roger P ., “ Criteria for Measuring Rail Transit
Efficiency,” Proceedings o f the Special Conference
on Urban Transportation Efficiency. New York,
July 26-27, 1976. New York: American Society of
Civil Engineers, 1977.

Wells Research Company, Trends in Bus Transit Opera­
tions, 1960-74. Washington: U.S. Department of
Transportation, 1977.

Scheppach, Raymond C., Jr., and Woehlcke, L. Carl,
Transportation Productivity. Lexington, Mass.:
Lexington Books, 1975.

The Employment Service

Transit Performance.

Adams, Leonard P., The Public Employment Service in
Transition, 1933-1968. Ithaca, N.Y.: Cornell
University, 1969.

Three volumes. West Lafayette, Ind.: Purdue
University, 1978.

Allegations About the Ohio Bureau o f Employment
Service’s Operations in Cleveland, Ohio. Wash­

Tomazinis, Anthony R., Productivity, Efficiency, and
Quality in Urban Transportation Systems. Lex­
ington, Mass.: D.C. Heath and Co., 1975.
“ Transit Operating Performance and the Impact of the
Section 5 program.” Washington: U.S. Urban
Mass Transit Administration, November 1976.
Transit System Productivity. Washington: Urban Con­
sortium 1976, revised 1978.

ington: U.S. General Accounting Office, Aug. 27,
1980.

Sinha, K.C. and Jukins, David P., A Definition and

Measurement o f Urban

Bain, Trevor, Labor Market Analysis: A Review and

Analysis o f Manpower Research and Development.
New York: Center for Policy Research, Inc., June
1975.
Benus, Jacob, and others, Use o f an Experimental

Design in Assessing the Impact o f the United States
Employment Service. Menlo Park, Calif.: Stand-

Urban Mass Transportation Industry Uniform System
o f Accounts and Records and Reporting System.

ford Research Institute, September 1976.
Borus, M.E. and Tass, W .R., Measuring the Impact o f
Manpower Programs: A Primer. Ann Arbor: In­
stitute for Labor and Industrial Relations, 1970.

U.S. Department of Transportation, Urban Mass
Transit Administration, 1977.
U.S. Bureau of the Census, 1977 Census o f Govern­

ments—Compendium o f Government Finances.

Chadwin, Mark L., and others, The Employment Serv­
ice: An Institutional Analysis. Employment and
Training Administration. Washington: U .S.
Government Printing Office, 1977.

Washington: U.S. Government Printing Office,
1979.
U.S. Bureau of the Census, 1977 Census o f Govern­

o f Public Employment.

Employment and Training Report o f the President, an­

Washington: U.S. Government Printing Office,
1979.

ments—Compendium

nual reports. Washington: U.S. Government Print­
ing Office.




95

Employment Service Needs to Emphasize Equal Oppor­
tunity in Job Referrals. Washington: U.S. General

“ Michigan Placement Follow-up Demonstration Pro­
ject.” Silver Spring, Maryland: Macro Systems,
Inc., Mar. 19, 1981.
Moore, Basil, “ A Benefit-Cost Analysis of the United
States Employment Service.” Bureau o f the
Budget, November 1966.

Accounting Office, Sept. 17, 1980.

Employment Service Performance Handbook for Local
Offices. Rockville, Md.: WESTAT, Inc., 1979.
ESARS Handbook. Washington: U.S. Employment and
Training Administration, November 1979 revision.

State Job Placement Programs. Albany: Legislative

Fairchild, Charles K., A Performance and Needs Based

Commission on Expenditure Review, December
1975.

Methodology fo r Allocating Employment Service
Grants. Cambridge, Mass.: Abt Associates, Inc.,

Stevens, David W. and others, “ Specification and
Measurement of Productivity in the USES,” draft
report. Washington: U.S. Employment Service,
December 1980.

April, 1980.
Fairchild, Charles K., Development o f Performance

Standards for Job Placement and Support Services
o f the Public Employment Service, four volumes.

Summary o f Employment Security Statistical Reports.

New York: E.F. Shelley and Company, Inc., 1975.

Washington: U.S. Employment and Training Ad­
ministration, August 1977.

Frey, Donald E., A Methodology fo r Measuring the Im­

pact o f the United States Employment Service.

The Employment Service—Problems and Opportunities
fo r Improvement. U.S. General Accounting Office,

Winston Salem, N.C.: Wake Forest University,
1976.

1977.

Glossary o f Program Terms and Definitions. U.S.

Thorpe, Charles O., Jr. and Toikka, Richard S., Deter­

Employment and Training Administration, 1978.

minants o f State Employment Service Productivity.

Historical Statistics o f Employment Security Acti­
vities— 1938-66. Washington: U.S. Bureau of

Washington: The Urban Institute, March 1979.
U.S. Congress, House of Representatives, Committee
on Government Operations, Operation o f the U.S.
Employment Service. Washington: U.S. Govern­
ment Printing Office, 1976.

Employment Security, 1968.
Jones, Melvin and Toikka, Richard S., A Review o f

Previous Research on the Determinants o f Employ­
ment Service Productivity and Proposed Exten­
sions. Washington: The Urban Institute, May 5,

U.S. Congress, Senate, Subcommittee of the Committee
on A ppropriations, Labor — Health, Education,
Welfare Appropriation fo r Fiscal Year 1968, 90th Con­
gress, 1st Session. Washington: U.S. Government Print­
ing Office, 1967.

1978.
Katz, Arnold, Explanatory Measures o f Labor Market
Influences o f the Employment Service. University
o f Pittsburgh, 1978.
Levitan, Sar A. and Mangum, Garth L., Federal Train­
ing and Work Programs in the Sixties. Ann Arbor:
Institute of Labor and Industrial Relations, 1969.

Weiner, Neil S., Powel, John H ., and Rahm, C.
Michael, The United States Employment Service: A

Conceptual Model o f Outputs, Values and Il­
lustrative Estimations, Vols. I and II. Arlington,

Meike, C., and others, SESA Productivity Measurement
System. Vienna, Va.: Analytic Systems, Sept. 24,
1976.




Va., Boeing Computer Services, Inc., December
1976.

96