"i
I
I

:CES

%C‘

R:

”RENT P

'5
am»
hi

mm

m
mew

Ekm‘M

v
11;. v,.;,
IY‘Lh-n‘. .2.

2,41":

in”?
bu‘rfll

’4

  
  

ll/glllg/MIBl/llllIll/lll/llllllll m y
2423123; a 2m

Um my

   

This is to certify that the

thesis entitled

COST CONTROL OF EDP SERVICES:
A COST ACCOUNTING MODEL AND

EXAMINATION.OF CURRENT PRACTICES
' presented by

Evan Williams Richards, Jr.

has been accepted towards fulfillment
of the requirements for

Ph.D . . Accounting
degree in

 

H

\ \‘ Majdr professor

 

Date W S?\c“lb

0-7639

:‘er
-I.L_.J;._nl_

 

 

 

 

 

    

12-

0 ~ " ’
tacotmtl' . r K 3- an. Hui." t: ._r
.. “:1

1:;th ”I I‘hz'tl- ' V " 00 I' ~ ‘ .;:|‘!.. x‘" [of ’j.-v_,lr ‘ .
. , - fi «I M ‘

I'm! a“. . .':' 3"“) ' Jm. (\2‘ a 2 4M 4: ultra 514'“ l'

I ' . " ‘ , D
““01. LI" 1" I a“ ' ' ‘
.. O '
.t- :. ad, 'I ‘16P. " .

x0 propour . Tl'n'i- U; a ' :LI " y v‘n v
.

. \V
' ébntrul 1nd,. up; we , . .21» m‘ .i. 'hit' . n uthI’IEAm‘
it" ‘
“$81 c.’am-ds-..".-' vmfluf' . ml |n": fie.
' .‘ r, \
voyreueih avidwuzc n22, ."-"n;_g ' .- m 2‘ .11? um. az'.,.._-w‘n1{ng
and :‘ncir alarm-tn» isms :u .w-‘sttcl 64:; mt? area. " "

  
 

lo amt inpz'ovacau in Current 3;va an!!! ‘Uut,201_PIOCI‘.u9. W:- 2
| .y—“q

5 15111:) and can mean. ' I.

 

. . ,

..\&“ accmmtmg system m st 5 3-1» new m Gevoiopadl . 1’ I; k
lanai cost accounting littx‘vuu. .n‘s We! em“ are}. : :4“
_ ng -

*vfl of con: collection and IuYMQ'L‘R’

   

 

 

 

Noll“

 

 

.x.

.3

 

G/asy

ABSTRACT

COST CONTROL OF EDP SERVICES:
A COST ACCOUNTING MODEL AND

EXAMINATION OF CURRENT PRACTICES

By

Evan Williams Richards, Jr.

The phenomenal growth of electronic data processing (EDP) in the
last two decades has been accompanied by increased management and control
problems. Many statements in the literature imply or state directly that
the cost accounting systems in use are often faulty and partially to blame
for the lack of control. However, the comments have been quite general,
and little is known about the specifics of actual cost accounting systems
used to control the EDP area.

The purpose of this study is threefold:

1. To propose a model of an EDP cost accounting system for
management control including the components of that system and determine
the individual components' viability in actual practice.

2. To present evidence regarding the use of EDP cost accounting
system components and their alternative forms to control the EDP area.

3. To suggest improvements in current EDP cost control practices.

A model cost accounting system for the EDP area was developed
based on traditional cost accounting literature. The model components are:

1. Low level of cost collection and aggregation.

2. Responsibility and cost centers.

B

I ' I
"F'

DI

 

- .
u-----

 

 

’.

Evan Williams Richards, Jr.

3. Budgeting and reporting system.

4. Output or activity measurement and performance criteria.

5. Long-range financial planning.

6. Tracking users' EDP—related costs.

7. Charging system for users.

Six field studies were conducted to examine current EDP cost
accounting practices and to determine the extent to which the model was
present both as a whole and as individual components. Firms with a
relatively high degree of EDP sophistication were chosen. Two larger
studies by the General Accounting Office (GAO) and the Cost Accounting
Standards Board (CASB) were used to provide summary and supporting
data on cost accounting practices for a much larger cross section of
firms.

The proposed EDP cost control model was found to exist in a
reasonably complete form in one case study firm, and nearly so in the
remaining firms. Most components found considerable support on an
individual basis.

The level of cost collection and aggregation in the case study
firms was about as expected. Most firms used essentially the same
chart of accounts in the EDP area as they did in other areas of the
firms. The excellent use made of responsibility and cost centers
reflected very well-organized EDP areas, and all case study firms were
above the observer's expectations in this area based on the revelant
EDP and cost accounting literature.

A11 firms had standard budgeting and reporting practices for
the EDP area which were part of an overall corporate procedure. Flexible

budgeting was found to be in use in one firm, but was a highly inappropriate

-- c
'.—-"
..
..
.
,- .w

__o
be a

 

 

#.- C'l

“as- o..-

.n «v--'
.-

o. ...;»5

 

‘ V .-
m... :x
at”-
fiav'
-
O
....' a-
.- ‘ ,
‘-
.. .

a
"’vo .
~.,.‘

 

».
. v u

\ ‘p
‘H .- T
n

'-

‘ on.

4..‘.
“a-
.
u.‘ _—
e.._‘
..-
o
>_‘ :
-.~

 

 

 

'-
.q

a
\
wc.

 

.—.~.— -- ——.

Evan Williams Richards, Jr.

technique of which the EDP managers had taken advantage. Although

monthly cost/budget reports were used in both the systems design and computer
operations areas, the reports had greater usage as a control device in

the computer operations area.

The use of an IBM software package, System Management Facility
(SMF), was widespread, but in some firms its use was limited to providing
inputs into the charging algorithm rather than also providing utilization
and other data for managing the data center. All case study firms had
fairly sophisticated work measurement and reporting systems for the data
entry area.

The next two areas——1ong-range financial planning and tracking
users' EDP—related costs--were the model components that had the weakest
support from the case studies. Two firms had little in the way of formal
plans for the EDP area beyond one year. Even the firm that had the most
complete long-range planning system could point to little in the way of
resultant, concrete benefits. In some cases the lack of formal long-range
planning in the EDP area was due to inadequate planning at the corporate
1eve1—-goals, objectives, and strategies of the corporation were not
outlined first.

One firm made a very complete analysis of users' EDP-related
costs in its annual summary of EDP costs, and three firms made minor
efforts to include such costs. In general, EDP personnel were more
responsive to the concept of tracking users' EDP-related costs than to
long-range financial planning, and consider the former to be a useful
idea.

All firms charged users for computer operations costs, and four

charged for systems development costs. There were considerable variations

a ":
-1
.- '
a
, o
‘ c. ‘
._-

\-

-‘o.
\v ‘
~-.~

"<.I.

'e....

 

53-“:

‘~«.

 

 

Evan Williams Richards, Jr.

in the determination and usage made of specific rates. In summary, the
model found considerable support on a component by component basis.

Suggested improvements in observed EDP cost accounting practices
include:

1. Examine carefully for adequate detail the chart of accounts
used to control the EDP area.

2. Examine the use made of SMF or similar data if available.

3. Consider the trade—offs made in cutting EDP costs and increasing
users' EDP-related costs.

4. Consider the behavioral effects of the current charging
practices in both the systems and computer operations areas.

5. Have top level management examine closely the need for long-
range planning at both the corporate and EDP levels.

Questions raised by the study which seem fruitful ground for
further research include:

1. The usefulness of long-range planning in the EDP area.

2. The effect of industry practices on a firm's EDP cost control
techniques and chargeout practices.

3. User reaction to complex chargeout algorithms.

4. The need for and most effective structure of a priority
setting device.

5. EDP capacity measurement.

COST CONTROL OF EDP SERVICES:

A COST ACCOUNTING MODEL AND

EXAMINATION OF CURRENT PRACTICES

By

Evan Williams Richards, Jr.

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Accounting and
Financial Administration

1976

 

taxi:
stream-h? '

 

 

 

V
“I

~9-‘..',,‘

"'-A‘.-d.
.

.5.

 

~\L":b
.N.“ '3».

ACKNOWLEDGMENTS

   
  
  
  
  
  
  
  

T
lVaberger, chairman; Gardner Jones; and John Visonhaler are

.acknowledged. Special thanks are due Dr. Sollenberger for
End patience in discouraging circumstances.

, m tine and effort spent by many individuals in the firms

V‘ V #138,1er the study is deeply appreciated.

:E.Pinally, a special thanks goes to my wife, Evonne, for her

”V support and encouragement throughout my graduate work, and

fifinding the fine line between encouragement and nagging at the

y.
. . .
"|'l «- I
, .

410:. .

f m

6.": ‘ ’

”‘Tlx

il‘LChwpi‘ . v‘ ~

5'.“ 'm ‘3: c' ~' m .2.
r {"1672 1 . , . . . . .
moment» ‘i (0‘: 3“"! um: 55%;: . . . . . . '
{I 2 ‘ i ‘ J .
3 . .
‘3 A a

 

... .

 

 

LIST OF

LIST OF

CHAPTER

I.

II.

III.

TABLE OF CONTENTS

TABLES . . . . . . . . . . . . . . .

FIGURES . . . . . . . . . . . . . .

STATEMENT OF PROBLEM, METHODOLOGY, AND LIMITATIONS

Introduction . . . . . . . . . .
Statement of Problem . . . . . . . .
Purpose of the Research . . . . . . .
Methodology . . . . . . . . . .
Limitations of the Study . . . . . .

Nature of the Propositions . . . . .
External Validity . . . . . . . .
Management Involvement . . . . . .
Organization of the Study . . . . . . .

SURVEY OF THE LITERATURE

Introduction . . . . . . . .
Background and EDP Cost Control Studies . .
Books . . . . . . . . . . .
Journal Articles . . . . . . .
Background on Transfer Pricing . . .

EDP Transfer Pricing Methods and Literature .
Why Charge? . . . . . . . . . .
Specific Charging Methods . . . . . . .
Average Full Cost . . .
Approximating Market Price-—Cost Plus .
Standard Full Costs . . . . . . .
Marginal Costs . . . . . . . . .
Turney's Model . . . . . . . . .
Value-based Allocation . . . . . .
Axelrod' 3 Model . . . . .
Nunamaker and Whinston' 3 Model . . .
Flexible Pricing . . . . . . . .
Priority Pricing . . . . . . . .
Multiprogramming . . . . . . . .
Chapter Summary . . . . . . . . . .

EDP COST SYSTEM FOR MANAGEMENT CONTROL . . . .
Introduction . . . . . . . . . .

Components of the EDP Cost System . . . .

iv

Page
. vii
viii
. l
. l
. 2
. 7
. 8
. ll
. ll
. 12
. 12
. l3
. 15
. 16
. l6
. 22
. 34
. 39
. 39
. 45
. 45
. 47
. 47
. 51
. 52
. 53

53
. 55
. 56
. 56
. 57
. 57
. 59
. 59
. 60

 

CHAPTER

IV.

V.

Low Level of Cost Collection and Aggregation . .

Responsibility and Cost Centers . . .
Controllable and Noncontrollable Costs . .
Budgeting and Reporting System . . .
Output or Activity Measurement and Performance
Criteria . . . . . . . . . .
Long-range Financial Planning . . . . . . .
Tracking User' 5 EDP-related Costs . . . . .
Charging System for Users . . . . . .
Charging for Computer Operations . . .
Billing in Terms of Computer Activity Units
Priority Pricing . . . . . . .
A Word of Caution . . . . . . .
Charging for Systems Services . . . . .
Resource Usage Basis . . . . . . . .
Lump Sum Negotiated Amount . . . . . .
Chapter Summary . . . . . . . . . . .

RESEARCH FINDINGS--CURRENT EDP COST CONTROL PRACTICES

Introduction . . . . . . . . . . . .
Centralization vs. Decentralization of EDP Resources
Level of Cost Collection and Aggregation . . .
Responsibility and Cost Centers . . . . .

Controllable/Noncontrollab1e Costs . . .
Budgeting and Reporting System . . . . . .
Output or Activity Measurement . . . . . .

Computer Operations . . . . . . .
Systems Design and Programming . . . .
Long-range Planning . . . . . . . . .
Tracking Users' EDP-related Costs . . . . .
Charging System for Users . . . . . . . .
Firm A . . . . . . . . . . .
Firm B . . . . . . . . . . . . .
Firm C . . . . . . . . . . . . .
Firm D . . . . . . . . . . . . .
Firm E . . . . . . . . . . . . .
Firm F . . . . . . . . . . .

Support from GAO and CASB Findings . . .
Chapter Summary . . . . . . . . . . .

CONCLUSIONS AND NEW DIRECTIONS . . . . . . . .

Introduction . . . . . . . . . . . .
Case Study Conclusions . . . . . . .
Presence of the Entire Model . . . . .
Individual Model Components . . . . . .
Suggested Improvements in Observed EDP Cost
Accounting Practices . . . . . . .
New Directions in Future Research . . . . .
Chapter Summary . . . . . . . . . . .

97

103
108
112
115
120
121
124
128
132
134
134
138
141
144
147
149
153
160

162

162
162
168
168

175
177
179

e 0 1 2 5
8 8 9 1 1
a 1 1 2 2
P
I I I I
I I I I
I I I I
I I I I
o o o o
o I o o
o o a o
I I o I
I I I I
o I O o
o o o o
I I I I

 

.Va-h
. I

 

 

a
-
0‘).

'..-
NI

LIST OF TABLES

Page

  
  
   
   
  
 
  
  
  
  

GAO end CASB findings-firms charging users . . . . . 154
GAD and CASB findings--D. P. budget vs. charging practices 155

GAD findings--objectives of and criteria used in
designing charging systems . . . . . . . . . 156

LF’
CASB findings-considerations in charging users . . . 157
:GAO findings-charging methods used . . . . . . . 157
1 GAO ‘findings--types of cost accounting systems used . . 157

QASB f1nd1ngs-—inc1usion of hardware/software costs in
" charging rates . . . . . . . . . . . . . 159

Gflfi‘and CASB findings-disposition of variances . . . 159

GAD S CASB findings—-circumstances resulting in
' premium/discount rates . . . . . . . . . 160

GAO findings--EDP cost as a percentage of sales . . . 167

 

.\
.J

‘Q’.
‘ O

IQ‘
.. "

("1

(-n

LIST OF FIGURES

Page

   
   
   
   
    
  
  
   
   
  
  
  

2 Helpful questions in determining a charge-out policy . . 42
_ A; ’EDP organization chart-firm A . . . . . . . . . 109
VEDP organization chart—firm B before change . . . . 110
VEDP organization chart-firm B after change . . . . . 111

;E*'”Pidwchart of'the budgeting process-—firm D . . . . . 117

‘"Vflillab1e items and charging algorithm-—firm A . . . . 137
fibbpsrative charging rates—-firm B . . . . . . . . 139
(t‘L ‘Billable items and charging rates-firm C . . . . . 143
-‘: 1. ‘Summary of case study findings . . . . . . . . . 163

. / onPastor's used to evaluate case study firms in model
6‘11}. : cmponept areas I a o o o o n o o s 165.166

'IVC Illustration of calculation of burden amounts . . . . 218_

"2 u .. .a -.
. V ' - A x '4' . r ‘
1 ’3" V3,“! Th1! Cl. ’. - ' ; , . . .
'.'V d A.“

. ' ‘ \' , \ >

3.5nohitfl“ 1r, " . ', _-.. . .- _. .. J Q
J- kV‘IV':fi .. 1"

'V t "PEEL“ y ‘4. ' V '. “an i . - an . . " . L: .3135;

-‘ 56179" ‘ ‘
=4 Nu. I‘m. , .qw.‘ i f‘wv. . 1. L. .~ -;.~f!v;.‘?s'-.:ru5‘.

,1: :1.

   

dorm “ h’” ‘r .s:- ‘1 (hi‘ ..z_ “ -; . =. -*9«.. a; :. v
Secs mum} ;, 1:12.; faint-.2. are .12. ' r '- ‘u u- 11”»xlztivs‘t:v.'

.-
£9311!“ we “ *1 A i
' I - ’0-. I 4
VV‘DLV‘ V1.11, .‘ . :“ +32.
r‘Y—V». ‘ ‘ ‘ -'L
w. A - i ‘
4 ,s , .- q

 

*- ....—

CHAPTER I

STATEMENT OF PROBLEM, METHODOLOGY,

AND LIMITATIONS

Introduction

This dissertation examines cost accounting systems used to control
and manage the EDP* area of business firms. Costs have often not been
well controlled in this area, and many statements in the literature imply
or state directly that the cost accounting systems in use are often
faulty and partially to blame for the lack of control.

This study develops a cost control model for the EDP area from
traditional cost accounting literature and examines the extent to which
the model was present in existing control systems of selected business
firms. Cost accounting has been refined to a relatively fine art in many
manufacturing settings, yet this state has not been reached in the EDP
area. This study attempts to refine the state of cost accounting and
control systems for the EDP area.

This chapter presents literature support for the existence of a
control problem in the EDP area. The purpose of the study and the
research methodology are then outlined, and the limitations of the study
are delineated. The chapter concludes with an overview of the remaining

four chapters.

 

*

The term "EDP" is used in this study to represent computer
operations, systems design, programming, and other related and supportive
groups within a business firm.

‘,»a

.‘a9

t...‘

 

 

Statement of Problem
The phenomenal growth of electronic data processing (EDP) in the
last two decades has been accompanied by increased management and control
problems. The number of computers used in business applications has
burgeoned from zero prior to 1953 to roughly 80,000 by the end of 1975.
Costs have soared correspondingly. Yet control in general and accounting
controls in particular seem to have been often neglected.

Present management practice in controlling data processing
costs does not seem to have kept pace with progress in the
other aspects of data processing. Thinking in regard to

cost control generally appears more appropriate to the
beginning days of computer use for commercial data processing
in the early 1950's than to current requirements. [74, p. 269]

Many authors seem to agree about the lack of cost control in the EDP area.
Some example comments follow:

There are five basic reasons why computer installations are
not paying for themselves. . . 2. Costs are not being
controlled within the computer department. [158, p. 48]

In too many instances, management has not provided the same
direction and control in this important area of operations

[EDP] as it has in other areas. Perhaps this is due to the
mystique of the computer, the relative newness of EDP, or a
difficulty in communicating with EDP people. [100, p. 8]

Computer—system costing as a relatively new discipline has
suffered from a general shortage of historical data. . . .
Available data have suffered from serious qualitative
shortcomings. . . Many cost reports have not been structured

to provide the kinds of data needed for computerized information-
systems-cost analysis. [49, p. 20]

Cost overruns came to be expected from the EDP function.
Overruns were ually the worst in system development where
overexpenditures of 100 percent were common. The processing
of regular EDP jobs and the production of user outputs,
however, also were marked by major cost excesses. [162, p. 72]

Fairly soon after a company begins to use data processing,
it often finds that the department is expanding. In many
cases, this growth is much more visible in terms of larger
staff and greater costs than in terms of increased service
to the company. [85, p. 72]

Most [data processing] installations are running less efficiently
than they could, and few data processing departments are subject
to the normal rules of assessment that are applied to other cost
centers in the corporate environment. Despite its cost, the
computer for some reason has been given an almost invoilable
status-—perhaps because too few people have the combination of

DP knowledge and accoutancy acumen to prepare cost efficiency
analyses on computer systems. Or perhaps the DP department, for
so long isolated from the rest of the company, is reluctant to
have outsider criticize its operations. [57, p. 103]

From the viewpoint of the executive vice president, "The EDP
manager always waffles around when he has to explain his budget.‘
From the viewpoint of the EDP manager, 'The executive vice
president never seems to understand why this department needs

a lot of money.‘ [58, p. 76]

In most companies, EDP management practice has developed in
isolation from the organization's other management systems and
functions. . . . Management practices are not well formulated
nor as well implemented in this portion of the business as in
others; and senior management must take care to ensure that the
characteristically "messy" EDP department is as effectively
controlled and planned as the old (and presumeably neater)
regular departments of the company. [106, p. 132]

Cost overruns have been so common in connection with both systems
projects and EDP operations that many management people have come
to assume a sort of overbudget inevitability where EDP is
concerned. [44, p. 89]

These types of comments are not new. As early as 1958 John Diebold
was saying:
Caught with their budgets down and with the responsibility of
enforcing cost reductions on the rest of the organization, many
controllers are red-faced about the cash drain of their own
computer centers. [73, p. 92]
An initial study done for the National Association of Accountants listed
the following among its conclusions:
In most of the companies studied, management of the computer
resource is charactierized by isolation. There is little
attempt to apply broadly applicable managerial knowledge and
techniques to those areas within the computer operation where
they might be relevant. [24, p. 144]

An article summarizing the results of a study conducted for the

National Aeronautics and Space Administration concludes as follows:

‘
I.

U.

Q..- 'P
.. at: DC

(3'an

unfit-1D?

[me bro;

CC[Porgy

0‘ 9001-

A computerized technical information center offers a new and
interesting challenge to the accounting profession. In a

field characterized by a serious accounting deficiency, a
[computer] center exemplifies the need for dependable financial
information. . . . While current applications of managerial
accounting procedures are sparse, preliminary investigations
indicate that such techniques are feasible and necessary

in a center's operation. [72, pp. 609, 610]

One of the most recent articles is also among the most scathing
in its remarks:

If I were backed to the wall for a one—sentence statement of
executive responsibility in the EDP area, I'd say: Put a stop
to the biggest ripoff that has ever been perpetrated upon
business, industry and government over the past 20 years.

A statement like that merits a bit of elaboration. What do I
mean by a 20—year ripoff? For 20 years executives have had
more and more expensive machinery sold or leased to them to
the point where today computing machines are costing big money,
and still running inefficiently.

For 20 years executives have employed expensive people--computer
people-—and they are still working far below achievable
productivity——in an age and in a country where productivity

is the key to survival.

They are still being led into attempts to computerize which

turn out to be outright failures, squandering huge sums of money.
They are still being led into EDP developments that fall short of
their promised benefits. They are still being promised services
in the sales pitch which are not forthcoming in practice.

EDP developments are still being completed much later than
promised, at a much higher cost than portrayed at the outset. . .
The whole situation is now reaching serious proportions. Recent
studies show that EDP costs are averaging one percent of company
costs and are reaching four percent in some cases. This
inefficient body of activities is speading into more parts of
each organization every day, and it has more power within an
organization than most think. [95]

The broadside continues, but space considerations are a limiting factor.

In a recent dissertation on EDP transfer pricing, the original
intent was to investigate the EDP cost data of "four or five medium-to—large
corporations." Only one company's data was eventually examined "because

of poor accounting records and inconsistency of definition." [149, p. 67]

~-

‘ - I

"n’!

v-..~4

*-~o.

"'-¥.

(—v
.11
(1‘

The poor quality of the data and the difficulty in collecting

usable information cannot be overstressed. Few companies

appear to utilize transfer pricing for their MIS services.

Those that do seem to rely on rather inadequate and elementary

cost accounting systems. This presented data to the researcher

that were inconsistent, incomplete and often poorly defined.

Many of the cost categories that were defined were affected

by more than one variable and it was impossible to isolate

the effects of just one of those variables. [149, p. 67-68]

Two alternative conclusions could be implied from such statements:

1. Because of the highly specialized and technical nature of the
EDP function, traditional cost accounting systems are of minimal use in
controlling EDP costs and operations.

2. Many firms have just not made the required effort to apply
cost control techniques used elsewhere in the organization to the EDP
area.

Most authors, of course, have held that the latter is true, but
few of them have present any evidence of firms that g9 effectively employ
a cost accounting system to control the EDP area.

The approaches to solving this control problem have varied widely.
Much of the emphasis in the literature of the area has been placed on the
transfer prices of computer services to internal users within the firm.
Many authors have evidently felt that if user departments could only be
charged, then costs of the EDP area would be automatically controlled.
Other authors have proposed specific ways to cut costs here and there.
Some have stressed that the EDP area should be controlled much the same
as other areas of the firm.

Underlying many of the comments on the problem has often been
the implicit assumption that an adequate cost accounting system exists

for the EDP area and that the data produced by such are accurate and useful.

As the last quoted statement shows, this is not always the case.

.
e N
‘-'.¢L ~
.

iiaract
:0 van
'aTC'VaI
'3 the
mafia:

Fear P6

"Hts.

The question might be raised at this point: "What is so special
about the EDP area? Isn't cost control important for all areas of the
firm? Why pick on this one?" The following reasons could be given for
concentrating on this area:

1. Because the EDP area services many users throughout a firm,
the degree to which it is well—controlled and efficient influences the
operation of the whole organization.

If the data processing department falls short of its responsi—

bilities, or if its operation is indifferently managed and

consequently inefficient, every part of the organization will
suffer. . . . Efficiency in the data processing department

is less an ideal than a practical necessity. The department

is responsible, not just for its own operational efficiency

but for the efficiency of most other departments in the

organization, in that it provides the information on which

these departments rely for their operation. Without reliable
information they cannot function efficiently; therefore, the

DP department must be operated with extra special precision.

The introduction of computer systems into an organization

creates a situation in which tremendous reliance is placed

upon one department and fundamentally a very high standard

is demanded from it. If inefficient performance is tolerated

there, it will not just inconvience those who rely on it, but

it could seriously affect the functioning of activities

throughout the firm. [147, pp. 92-93]

2. The computer resource has a unique set of supply and demand
characteristics. 0n the supply side we see (1) a high ratio of fixed
to variable costs, (2) great potential for economies of scale in computer
hardware, and (3) incremental capacity usually acquired in large blocks.
On the demand side we see (1) a rapid growth in demand and increasingly
complex user needs, (2) cyclical processing—-month1y and end-of—fiscal—
year peaks, and (3) highly variable user priorities and turnaround require—
ments. [107, pp. 471-473]

3. There are constant changes in available hardware and

technology, often with implications for the nature of costs as well as

for cost control.

u;

 

 

n:

b
h

E

4. EDP centers are typically managed by technicians, often with
little or no business orientation or training.

5. EDP costs are often quite large with respect to other service
departments in a firm. [71, p. 6] In many installation these costs are
growing at an annual rate of around 25 percent. [137, p. 74; 51, p. 29]
Thus the problem of controlling EDP costs seems unique and important enough

to warrant special attention.

Purpose of the Research

The purpose of this study is threefold:

1. To propose a model of an EDP cost accounting system for
management control including the components of that system and to
determine the individual components' viability in actual practice.

2. To present evidence regarding the use of EDP cost accounting
system components and their alternative forms to control the EDP area.

3. To suggest improvements in current EDP cost control practices.

The outline of the model cost system is presented below. Further
development and explanation of the model components is reserved for
Chapter Three. .
The components of the model cost system are:

1. Low level of cost collection and aggregation.

2. Responsibility and cost centers.

3. Budgeting and reporting system.

4. Output or activity measurement and performance criteria.
5. Long-range financial planning.

6. Tracking users' EDP~re1ated costs.

7. Charging system for users.

C
'J
(x.

u 'AAI':
.. . u
35-... .k».

In
should idea
Kiel. Du:
:suc'n de
G? situati

Md mega

I“ ’
LJ

q

.
h

“
r1

, tzen:
ietemine
illocation
the organi
this would
Th'

tlc'n wouh

In determing the model's viability in actual practice we are
really looking at two propositions:

l. The model EDP cost system provides an effective means of
controlling EDP costs and is present (as a whole system) in existing
organization practices.

2. All of the postulated components mentioned are needed in a
reasonably complete cost control system and are individually present
in existing EDP cost accounting and control systems.

Number one focuses on the existence or nonexistence of a reasonably
complete EDP cost system. Number two examines the viability fo the indi—
vidual proposed cost control components. If one or more of the elements
is not found in existing practices, its usefulness for EDP control

might be questioned.

Methodology

In proposing a control model of any sort, the research methodology
should ideally include the following steps. (1) Develop the conceptual
model. During this step no attempt should be made to specify the model
in such detail that it could be immediately applied to a specific company
or situation. This degree of specificity with the EDP cost control model
would negate the advantage of general applicability to many corporate EDP
departments. (2) Determine the economic validity of the model. To fully
determine the proposed model's effects on decision making and resource
allocation, it must be implemented and its effects observed. (3) Observe
the organizational and behavioral effects of the proposed model. Again,
this would require implementation of the model.

Thus, to accomplish (2) and (3) one or more firms must be found

which would be willing to implement an untested, conceptual model over

«ne‘e‘

O
; ,...~—

.9:
.oou-b-s-v“

. . .
1 ...9.
1-... d.

.tp-o-~~

» - F‘v '

.'I‘bbh‘
.J

C‘r- -
mt“?-
“ s.
.‘__.
~‘y '
..~“ 1

3Pear

.0116Ct

a period of time. This is wishful thinking. No firm, under normal
circumstances would be willing to let the cost accounting system of a
multi—million dollar service department be changed at will. Thus other
strategies must be used.

There has been considerable precedent set for the use of field
studies or case studies as a research methodology in the area of EDP
control and management, possibly for the reason given above. Starting
with the NAA survey by Churchill, Kempster, and Uretsky in 1969 and
continuing with Sollenberger's NAA study in 1971, and Turney's and
Heitger's dissertations, both in 1972, field studies have been used
almost exclusively. [24, pp. 4—5; 142, pp. 4-5; 149, pp. 67-68; 71,
pp. 12-13] The number of firms used varied from one to eighteen.

Field studies, based mainly on interviews, have a number of
strengths and weaknesses. [5, pp. 223-224] The advantages include:

1. Realism.

2. The ability in investigate a large number of variables and
their interaction in a complex setting.

3. The ability to explore in depth those characteristics which
appear of value.

4. The ability to clarify intent and/or meaning of questions
asked of the respondent.

5. Interviews are extremely well suited for gathering behavioral
data, e.g., opinions, future intentions, and attitudes.

The disadvantages include:

1. The inability to separate and manipulate variables.

2. Great quantities of time and high costs may be involved in

collecting such data.

V
Air“ 9"
' d ..
elk-o ah.

i:':.. u}...

"““?’:x
no.2».u

firmtux

hdustl‘j

10

3. Due to its ex post facto nature, causal relationships are

 

"weaker" than in controlled studies such as laboratory experiments.

4. Face-to-face stresses may cause biased or untruthful data.

5. Faulty interviewer and/or interviewee perception are possible.

In order to determine to some degree the viability of the proposed
model and its components, the case study method was chosen. It would have
been much easier to stop at the end of the first step (see p. 8) of model
construction as others have done. However, the lack of published data
regarding actual EDP cost control practices and the need for such data,
coupled with the above—mentioned advantages of case studies, provided
the impetus for an attempt at determining the validity of the model.
The case studies could also provide evidence regarding the question of
the degree to which cost accounting systems are useful in controlling
EDP costs and the adequacy of several actual systems.

The field studies included visits to six firms. At least two
and usually three or more people were interviewed. In most firms, the
persons interviewed included the Director of Corpoate Information Systems
and managers from computer operations and systems development. Time
spent at each firm ranged from one half day to two days. Notes were
taken and used as the basis to write a summary of each visit as soon as
possible after each interview. Although the interviews were not highly
structured, thereby allowing perusal of topics of interest to those being
interviewed, certain structured data were obtained from all firms.

The firms that were visited were all large firms within their

industries. A breakdown of firms by product class or industry follows:

 

 

 

.
on. ~
bean

¢>A

C?

v'F-

C'gra

abut

~V

_-

11

Classification Number of Firms

1. Heavy manufacturing and materials

conversion 2

2. Chemicals and pharmaceuticals 2

3. Utilities 1
4. Electronics _1_
Total __6_

Limitations of the Study

The limitations of the study will be discussed in three parts:
(1) the nature of thepropositionsmade, (2) external validity, and (3)
the overall question of upper management involvement in the control of

the EDP area.

Nature of the Propositions

A statistical or "testable" hypothesis is a statement about one
or more specific parameters of a population. [5, p. 210] The propositions
stated on page eight are not statements about specific parameters of a
opoulation and, as such, are not statistically "testable". It it were
possible to chose a random sample of "large" firms, divide the sample
into treatment and control groups, implement the proposed cost model in
the treatment group firms, and compare results between the two groups
over a series of time periods, then specific hypothesescouLdbe stated
about certain parameters (e.g., total EDP costs) of the two groups.
However, as already stated, this is impossible; no firm would consider
it. Thus our propositions remain working hypotheses, not statistical

hypotheses.

so -" f

. ...u.
—

\*~.«'r
u... 4L1
AH}.

1‘

._:‘E f

o
'v
.g. —j
I. "l
u
1‘ .
a

a ._
ll,

12

External Validity

The concept of external validity relates to the relative strength
of the generalizations made from the sample and applied to the population.
The six firms chosen were not selected randomly. Generalizations from
such a small sample to all business firms would be without merit even if
the firms were selected randomly. To improve the degree of external
validity the results of two other studies are included in the chapter
dealing with the analysis of the field study firms. One of these studies
was a very large scale field study conducted by the General Accounting
Office (GAO). Approximately 50 computer installations were visited,
including 22 business firms, and numerous interviews were conducted.

The other study was a questionnaire survey of intrafirm EDP pricing
practices conductedbytfluaCostAccounting Standards Board (CASB).
Summarized data from both studies will be used to give a broader

basis for a discussion of overall corporate EDP cost control policies
and will give some idea of how "typical" the six field study firms are

of the corporate population.

Management Involvement

This study does not purport to solve the overall problem of
controlling the EDP area within business firms. Many authors have
felt that the apparent lack of control in many firms was the result
of upper management's hesitancy to appraise EDP expenditures. [See 137;
85; 73; 106; and especially 94] Surely upper management involvement
is crucial if costs are to be controlled. Yet the first step in this
direction is the designing of an EDP cost accounting system. The
accounting system will not guarantee that costs are being controlled——

that is mans ement's responsibilit . Yet, without an adequate cost
3 Y

i

a"
17 “a...
"-

.-
lu' "
I. I .u
i

A w
“’v-\.4..

-~-U‘—J
'

2‘5“

“‘5\.»

P
up-..

”dyser

13

accounting system, control would be difficult if not impossible. In this
study the groundwork for controlling EDP costs is formulated by designing
a cost accounting system and to some degree determing in actual practice
the viability of its components. Implementation of the system in specific
corporate circumstances would require consideration of numerous factors
such as corporate size, corporate experience with EDP, corporate goals

and objectives, and specific objectives for the EDP area.

Organization of the Study

The study is organized into six chapters. Chapter One
introduces the problem and the need for the dissertion. The research
methodology and accompanying strengths and limitations are discussed.
Chapter Two reviews the literature pertinent to the topic. The discussion
will be divided into three general areas for convenience:

1. Literature regarding the overall management of the EDP area
which has significance for the control of such costs. This is a broad
category and contains some background studies.

2. A brief background on trasfer pricing literature in general.

3. Literature which applies transfer pricing specifically to
the EDP area. Proposed models of EDP transfer pricing methods will be
examined. Studies which mention transfer pricing, but which are more
concerned with the overall control process will be discussed in the first
of these three areas.

In Chapter Three the cost accounting model for EDP will be
developed. Traditional cost accounting concepts and literature will be
applied to the EDP area based on the background of Chapter Two. Chapter
Four contains the findings of the field studies. The GAO and CASB

Studies will be reviewed for a larger picture of EDP cost accounting

Iv:"" '-
0L:

a."'

p.-

“I:

 

14

practices. The findings will be examined for support or nonsupport of
(l) the model and components and of (2) various ideas in the literature
regarding actual practices. Chapter Five summarizes the findings,
paying particular attention to how EDP cost systems may be improved.

Suggestions for future research are included.

CHAPTER II
SURVEY OF THE LITERATURE

Introduction

The amount of literature devoted to computers and related topics
has paralleled their increased usage in the last two (plus) decades. The
majority of the literature has dealt with such topics as the changes (real
and imagined) brought on by automation, new use or applications of the
computer, or technical problems related to hardware and/or software.
Only a relatively small, but growing, portion of the literature deals
with the control of the EDP resource.

Much of this control literature deals with the acquisition
decision, cost/benefit analysis, or the transfer pricing of EDP services.
Although recognizing the extreme importance of the acquisition decision
and the role that accurate cost data must play, this study is more
concerned with the cost control of ongoing operations. This study is also
concerned with providing accurate cost data for use in cost/benefit analyses,
but the estimating of benefits is beyond its scope. Much of the cost/benefit
literature has dealt chiefly with the problems of measuring benefits and
has (implicitly) assumed that accurate cost estimates are easily
forthcoming from existing accounting data. [For example see 65.] Thus,
exCIUding the transfer pricing articles, only a very small portion of
the current literature relates to controlling ongoing computer operations

and peripheral services.

15

*6-

rv‘
v A.
..». -O
...
s...

a-."‘
aiaueh

.“
“n."-
. __...-.

. L: o a c 1-
«A. G . v . fit {A «A. n3
.: pla «\— a .. 8L v a
a , u u v n 2.. . .Ih '78 :1 u‘v
~ g .. o a a u — um.» - on 5.1%
. . . . e .0. a V. .iv . n

V.
E .u

l6

Charging for EDP services used (transfer pricing) is really an
internal marketing mechanism; it affects the volume and "product" mix.
However, it is not incorrect to think of it as a long—run cost control
mechanism. Controlling corporate usage of computer services directly
affects the required investment in hardware and personnel and therefore
the amount of costs incurred. Since so much of the EDP control literature
is devoted to the specifics of this topic we will separate it from the
more general literature and provide a brief background/introduction
with a summary of the overall transfer pricing literature. Within the
first of the three literature review areas, outlined in the previous
chapter, books will be discussed first, followed by journal articles

and other sources.

Background and EDP Cost Control Studies

Excluding transfer pricing articles, only a relatively small
portion of the EDP literature relates to cost control of ongoing computer
operations and peripheral services. The National Association of Accountants
has sponsored two studies which, although not dealing specifically with
control of ongoing operations, provide much helpful background information

and discuss EDP control problems, particularly in the development stage.

Books

The first of the two NAA studies is Computer—based Information
Systems for Management: A Survey, published in 1969. [24] Although

Primarily concerned with determining existing computer applications, the
study is helpful in several ways. (1) It emphasizes the importance of
8Pecifying objectives or goals for the EDP organization within a firm.

“eitger later made this a focal point of his dissertation on transfer

II A

.......

n. m (“I (V rv

..~ Em:
”’C, .A‘
m. .u]

l7

pricing. (2) It identified a divergence of typical EDP—related
activities that is likely to cause control problems.

To the extent the computer resource is devoted to processing

the already programmed activities and to automating structured

tasks in a relatively straightforward manner, managing the

computer is similar to managing a factory production or
assembly operation. To the extent the computer resource is
devoted to developing new applications in the unstructured

or managerial areas, its management problems are more like

those of an R & D operation. [24, p. 125]

Several authors have since commented on the similarities between ongoing
computer operations and the production process and have reasoned that
cost controls similar to those used in manufacturing activities would
work well in at least some of the EDP functions.

(3) The study identified a need for continuing management
involvement, not only in overseeing the development of new applications,
but in the control of ongoing operations. The EDP area is too important
to be left "to the nonmanagerial ministrations of technically oriented
people." [24, p. 144] Again, several authors have since stressed the
same point. (4) Although a case study approach was not new, this study
and its successor used such on a large scale, setting considerable
precedent for research in this area.

The second study in the NAA series is Sollenberger's Management
Control of Information Systems Development. This study will be referenced
again in later chapters, but a summary of its major conclusions is in
order at this point. Sollenberger's research, and therefore his
conclusions, dealt primarily with the general problem of controlling
information systems development, but they by necessity overlapped into
the area of ongoing operations. His eight general conclusions [142, pp.

8‘9] follow:

 

 

.— 4

.-n—...ol

.
.‘P

.so

4...!-

-e~‘u.

 

 

 

-
v

a
a
\
*-
a
\
w...
-
‘
u
u
a

 

 

18

(l) The key ingredient to successfully managing the information
systems development activities is proven managerial skill rather than
computer expertise. Others have agreed wholeheartedly with this finding.
[For example see 73, p. 95 and 31, p. 5]

(2) The systems function should be highly user-oriented.

(3) Sollenberger concurred with the earlier NAA study in that a
strong degree of management involvement at every level in each systems
endeavor is desirable. Interestingly enough, he found that this involve-
ment was broader than generally implied in the literature.

(4) "Clear-cut assignment of responsibility is critical to the
management of information systems development."

(5) Open communication between users and analysts is essential.

(6) The overall control philosophy of the company was important
for its effect in setting plans and measuring performance in the systems
area.

(7) "A strong application of economic evaluation appears needed
in order to justify continued large expenditures and to allocate use of
scarce systems resources among competing demands for its services." This
conclusion is discussed further in the paragraph following the eighth
conclusion.

(8) The organizational location of the systems function was not
a major issue. This result seems strange considering the many comments
regarding the desirability and even the necessity of placing the EDP
organization at as high a level as possible within the organization.

[For examples see 123, pp. 22-23; 73, pp. 93-94; and 98, p. 252] Again,
the (relatively) many years of EDP experience these firms have had might

be a causal factor for this conclusion. Sollenberger goes on to state

..u ' S
,4. 6 ‘
.
a O
,...
:_.: ”1
I

w

_..._. 0...!- -
tl: 5U:

. 77A P
.l-sbl L

"”P‘ A” "
$~‘~.\rw.—L
.
“A. 50“-
... J . .'
.._.'. ':
.uo‘a,’ _

ifije:t I

.
"v-v‘ -.

runes1

n . v
.

'x"'"r~

"“ “‘11
\—

19

that "a sufficient authority level and adequate independence were
frequently cited as criteria for the appropriate vantage point."

Sollenberger's seventh conclusion and his comments in this area
have the most import to the present study. His Chapter Seven discusses
control techniques used by systems management. These included: (1) policy
development and documentation, (2) a time recording and reporting system
for both projects and individual personnel (an integral part of number
four), (3) a system service request and related control unit, (4) a
project reporting system, (5) cost assignment——charging for system
services, and (6) post-implementation evaluation of systems output.
Although means and degree of implementation differed in the first four
areas, there was not the wide variation in opinion found in the fifth
area. Sollenberger lists a number of advantages and disadvantages of
transfer pricing of systems services. These will be discussed later in
this chapter. A particular strong point of his Chapter Seven is a series
of flowcharts showing the interrelationships between items (3), (4), and
(5) mentioned above. [142, pp. 120, 124, 131]

We now turn to other books before examining journal articles in
this first area. A surprising number of books about planning, organizing,
and managing the EDP area contain nothing about cost controls or only
passing mention of the topic. [See 161; 123; 19; 66; 103; 160; 16; and
70] Several books were found that did allude to means of controlling
.EDP costs. These will be discussed in increasing order of importance,
starting with those that spent the least time on the subject.

Albrecht's Organization and Management of Information Processing
SzStems relies heavily on the budget as the principal medium of cost

ccnitro. Typical information system "divisional plans" (budgets) are

20

presented and idscussed; later they are broken into "departmental plans."
He briefly mentions planning for individual projects but there is no
description of how a project reporting system might work. His example
schedules of budgets and monthly cost reports are one of the few spots

in the literature where a possible chart of accounts for the EDP area is
presented. [3, pp. 158, 302] However, there is no discussion of a chart
of accounts as such.

Tomlin's Managing the Introduction of Computer Systems briefly
touches project reporting in the third chapter, butuwstof the discussion
of cost control is found in the fifth chapter entitled "Departmental
Control". Tomlin defines control as "a cycle of events which occur to
regulate something which is dynamic" [147, p. 95] and states that:

Control in the data processing department operates to ensure

that the work performed is of an acceptable quality, is

produced in an acceptable time, and at an acceptable cost.

In addition to the control of work, the DP manager must also

control the resources (principally men and machines) to ensure

that they execute the work efficiently to meet the three

criteria of quality, time, and cost. [147, p. 94]

Much of the ensuing discussion is devoted to splitting EDP control into

"administrative, operations,’ and "systems" control. His orientation
to cost control is essentiallyoneofdeterminingapplication development
costs and thereafter ongoing Operating costs. [147, p. 114] His
discussion of a charging policy includes rates for (1) systems work,
(2) programming, (3) data preparation, and (4) computer processing.

In Managing the EDP Function a chapter is devoted to EDP cost
management and control. [44; Chapter Six entitled "EDP Cost Management"]

Ditri, £5 31. begin their discussion by enumerating four categories of

costs: [44, p. 91]

sliver? c

._.... § ,4
...:3:8.

=;;-L UI— ‘
L...“ mt:
siizts are

n,.;~"1
at. ,,;u‘
.

1.25:5 of

Shilar h
iiSCUSgec'

.

:2 finale;
E'liature

1' “Child

ax vari.

21

(1) Direct variable costs—-"these are direct payroll, travel,
living, and supplies expenses which relate specifically to production and

delivery of products and vary with work volume."

An example cited is
increased keypunching costs because of a pre-Christmas rush.

(2) Variable overhead costs-—"these are production—type costs
which vary with work volumes." The wages of supervisors on overtime
shifts are an example of such.

(3) Programmed costs-—"these are planned costs incurred for a
particular time period as a result of a specific management decision."
Costs of systems projects are cited as an example.

(4) Standby costs--these are the fixed costs of doing business.
They include "the prime shift rental of computer equipment, supervisory
salaries, and wages of key personnel."

This categorization of costs is interesting because of a somewhat
similar breakdown used by Turney. [149, p. 117] His study will be
discussed later in this chapter be cause it is oriented mainly toward
transfer pricing. Ditri's cost scheme results from the continuing use of
an analogy likening the EDP function to a manufacturing business-in—
miniature. While his categories are cited in traditional cost literature,
it would seem difficult to draw distinct lines between direct variable
and variable overhead costs and between programmed and standby costs in
an EDP setting.

He subdivides total EDP—related costs in (1) developmental costs
and (2) production costs. Developmental costs include all programmed
costs for system projects and some standby costs of the EDP function and
organization as a whole. Production costs include all four previously

mentioned categories. The discussion of cost management per se is very

"Ho

. u
0";"4 a

y... .bu

:45: 36C

' ' .1 -
V ';-~f . v
v. aoebga

-

..
r “A

~s‘.‘ .Lg‘

"""."':v
--~~.Ck.u.|

H':“’\V
‘eo--.\"

.

1-3:], a

I‘u...‘
whet,

-‘-‘

-t...E [h

slut 300‘

R this

75’3": b
"eel.er

22

limited and consists chiefly of statements to the effect that costs

must be closely associated with scheduling and performance reporting
techniques. [44, p. 14] A brief discussion of cost reduction includes
several examples of opportunities for improvement such as centralization
of facilities and consolidation of reports.

Although Ditri £3.31. touch briefly on the topic of charging
internal users for EDP services they do not seem to strongly favor this
procedure. [44, pp. 95—96] More emphasis is placed on holding the EDP
director responsible for budgeting and expenditures.

A chapter in Information Systems Administration by McFarlan,
Nolan, and Norton is entitled "Management Control of the Computer Resource."
However, most of the discussion centers around charging for EDP services.
Hence this book will be discussed in part three of the present chapter.
The book The Economics of Computers by Sharpe will also be discussed later
in this chapter since the section of major interest deals with pricing

computer services.

Journal Articles

Journal articles (excluding transfer pricing articles) regarding
the control of EDP costs are now examined. The first article discussed
is a background article providing an excellent overview of the typical
development of the EDP area within a firm.

Gibson and Nolan have suggested "four distinct stages in the growth
of all EDP facilities, each with its distinctive applications, its rewards
and traumata, and its managerial problems." [58, p. 76] The four stages
provide an excellent background for discussing other EDP cost control
articles and the EDP transfer pricing literature.

Stage One is the "initiation stage." The EDP facility is often

23

located--possibly unwisely—-in the accounting department. The first
applications are cost-reduction, accounting-type applications such as
payroll, acc0unts receivable, and accounts payable. There is a loose
budget and controls in general are notably lacking. Project priorities
are often assigned by FIFO; there is no chargeout mechanism.

Stage Two is the "expansion stage." There is a proliferation of
applications in all functional areas; e.g., cash flow, general ledger,
budgeting, capital budgeting, forecasting, personnel inventory, order

processing, sales, and inventory control. This period is again typified

’by a loose budget and lax controls, often intended to engender new

applications development. Upper management is often sales-oriented
with respect to the EDP resources. The EDP manager is moved up in the
organization, but in the functional area of the first applications.
The end of Stage Two is often marked by a sudden, top—management
awareness of the runaway computer budget and Stage Three begins
with a crash effort to find out what is going on.

Gibson and NolanentitleStage Three the "formalization stage]'
but a more fitting title would be the "control stage." There is often
a moratorium on new applications; those that are developed are control—
oriented, such as purchasing control and scheduling. A proliferation of
controls is instituted to control the runaway budget. Project management
and reporting are begun. There is strong budgetary planning for hardware
facilities and new applications. A steering committee is set up to ensure
that priorities are properly ordered.

Stage Four is the "maturity stage." EDP is set up as a separate
functional area, with the EDP manager taking on a high level position.

New applications are data-base oriented: on-line customer query system,

24

on—line source data entry, on—line personnel query, simulation models,

and financial planning models. There is a refinement of the management
control system with the elimination of ineffective control techniques

and further development of others. Instead of concern for merely

charging out EDP costs to recover these costs, there is emphasis on
pricing of computer services for engendering effective use of the computer.

These four stages are useful in explaining some of the discrepancies
in the literature and in particular, statements made about actual
practices by business firms. A person observing firms in stages one and
two would arrive at considerably different conclusions about the control
policies of "typical" firms than would a person observing firms in stages
three and four. An example of this will be cited in Chapter Four.

The present study develops an accounting-type control model which
may be applied (in differing levels of detail) to firms in all four
stages. However, case study firms are examined which are well into
stage three or into stage four to observe the controls in use and determine
to some extent the viability of the model and its components.

In a 1970 article, Homes outlines "twelve areas to investigate
for better MIS:"

1. Top management's involvement with the system.

2. Management's ability to organize the MIS function.

3. The use of a master-plan.

4. The attention given to human relations between functions

involved.

5. Management's ability to identify its information needs.

6. Management's ability to apply judgment to information.

7. The condition of basic accounting, cost and control systems.

8. The degree of confidence generated by accuracy at the input

level.

9. The frequency of irrelevant or outdated data provided.

10. The competence of systems technicians and their grasp of
management problems.

11. The justification for projects undertaken.
12. Reliance on equipment vendors. [79, p. 443]

25

Number three and seven of the above have the most import to the present
study. In discussing the use of a master plan, Homes states, "Setting
over—all company objectives and planning to meet those objectives still
leaves much to be desired in most companies." He outlines the proper use
of long—range planning as follows:

Planning must start by setting long—term company objectives

followed with a detailed profit plan. Information needs

within the plan must then be determined and provided for in

the creation of the data bank. All separate subsystems

should be coordinated into one integrated system. Planning

should cover the gathering, transmission, and dissemination

of data as well as its processing. Never should individual

major applications be started without a complete integrated

plan. Above all, the matters of systems concepts, goals,

and long-range planning must not be left to the discretion

of the te chnical staff, but assumed entirely by top management.

[79, p. 446]

"The Value of Information" by Rudolph Hirsch is primarily a
discussion of when information value can and cannot be calculated, and
what determines information costs. [75, pp. 41-45] However, in
concluding the article he discusses three ways of controlling the cost
of information processing equipment, i.e., the computer. (1) Use a
positive reports control scheme. Omit issuing one cycle of a report
suspected to be obsolete; if there areno protests, obsolescence is
confirmed. (2) "Counteract as far as possible a well-defined executive
tendency to use as much and as advanced equipment as possible. The
fascination of novelty the desire to appear progressive and the fact that
using computers has its status value all play their part. Those
responsible for the choice of equipment should avoid novelty for its own
sake." [75, p. 44] (It is interesting to note than even as late as 1972,

Heitger found one of the corporate EDP objectives to be: "Give the

appearance of being progressive." [71, p. 91]) (3) Use a chargeOut

26

scheme, including not only computer running time, but also the initial
system design, programming effort and program testing time. Hirsch
says this would be an incentive to recipients of information not to
request lowhvalue information, and to discontinue it when its value
decreases. However, he recommends that such a system "not be used in
the early days of computer usage in the organization . . . since the
effect could then be to discourage worthwhile computer projects."

In a 1970 article, Hirsch considers the EDP cost control problem
further. [74] The major thrust of this article is toward a detailed
monitoring of the system and programming effort and of actual computer
usage by jobs. This information is used to produce a "Project Control
Report" and a "Programmer Performance Analysis." Examples of such are
given in the article.

Hirsch briefly discusses a cost charge-out method which "has
worked well for a large corporate computer user."

When a department requests computer service, the data

processing department estflmates the costs of that service

and quotes a standard cost to the requesting department.

If the latter accepts that cost, it will be charged that amount

and none other, regardless of the costs actually incurred.

This method has a double advantage: it not only permits

'customer' departments to use data processing services with

(budgetary) confidence, but also enables top management to

monitor the efficiency of the data processing department

simply by analyzing its 'profitability.‘ [74, pp. 270-271]
This method may work well for one company, but it has some problems,
depending on how "profitability" is interpeted. For example, if it
means departmental "profit" as a percentage of "sales," the EDP depart-
ment might be motivated to reduce its size and to quote very high "prices"

and accept only a small number of high-margin jobs. This may or may

not be conducive to achieving overall corporate goals.

___A'K‘

27

Logue describes seventeen practical suggestions for reducing
computer costs. [100] Some of the numbered suggestions include: (9)
"determine whether key punching is the best method," (11) eliminate
unnecessary applications," (12) "ensure that all reports are being used,"
and (16) "consider other suppliers for peripheral devices." [100, pp.
13-16] Number fifteen was "install standards and work measurement,"
but the discussion was oriented toward keypunching, not systems and
programming tasks. [100, p. 16] The first two suggestions were "cost-
justify present equipment" and "consider the cost of alternate methods."
[100, p. 9] Both of these suggestions assume the existence of an adequate
cost accounting system for EDP. Little attention is given to improving
the quality of the cost accounting system and the way it is used.

Logue's last suggestion is "centralize multiple installations." [100, p. 17]
On this point he is in agreementwithseveral other authors. [See 53; 82;
86; and 159]

The book by Ditri 22 21* has already been mentioned. In a 1971
article Ditri and Wood present many similar ideas. [102] They break
EDP costs into the four types previously discussed. EDP is again likened
to a production department with similar controls. They conclude that "the
basic division of the management process into the elements of planning,
implementation, and control applies to EDP as it does anywhere else."
[162, p. 83]

In a 1972 article Brabb and Grosso emphasize two ways of
centralling EDP costs. [18] The first is a project reporting scheme
involving many of the ideas previously discussed. The second is the use
Of Facilities Management (FM), i.e., contracting an outside firm to

flanage and operate the EDP service center. Among the advantages cited

28

for FM is a potential for considerable cost savings. Other advantages
and some definite disadvantages are discussed. The authors briefly touch
on cost/benefit analysis, but, as in most other articles, the emphasis

is on the problem of determing benefits, not costs.

In "Cost Control for Computers," Hendrick Smith examines the
similarity of basic flows and controls for production areas and EDP
operations. [137] His analogy starts with raw materials (input data
and files) and follows through to the finished product (processed data
and information). Some of his cost control comments parallel those of
Logue, e.g., use alternative suppliers for peripheral equipment. [107,
p. 77]

Smith emphasizes the use of budgets. "Hard-nosed budgeting may
be the greatest stimulus for achieving EDP cost savings." [137, p. 78]
Two rules should be satisfied: (1) responsibility for cost incurrence
must be clearly identified and (2) the budgeting process must be simple.
In keeping with the first rule, Smith states that controllable and
noncontrollable costs should be segregated. He stratifies a "typical
EDP budget" in order from the most controllable elements (system
development salries and outside fees) to the least controllable elements
(administration and allocations). At no point however, does he draw a
cutoff line between "controllable" and "noncontrollable" costs. He was
probably wise in not doing so because it would be easy to take exception
wherever it was drawn; but the fact that he didn't draw such indicates
the difficulties a firm would probably have in doing so. This problem
will be discussed further in the next chapter.

Smith stresses that costs by themselves are not enough; the EDP

manager must have reports highlighting the underlying reasons for the

 

29

cost performance achieved. These include:

(1) Peak and average utilization reports for major equipment
components.

(2) Efficiency reports detailing operating time by major
applications, project schedule accomplishments, and data input performance.

(3) Cost exception reports. [137, pp. 78-79]

He briefly mentions the same type of charging system found in
Hirsch's 1970 article. Overall he seems to come out in favor of charging
users, but he is in favor of a simple algorithm.

I have seen charging schemes that account not only for

computer time but also for tapes maintained, forms consumed,

disk files occupied, and even communications channels used.

Frankly, simpler approaches satisfy budgeting objectives as

well if not better. [137, p. 77]

The point is debatable. Although many authors suggest simplicity as one
of the criteria for evaluating a charging algorithm, fairness and impar-
tiality are also suggested. It is doubtful that in this day of multi-
programming and many peripherals that a single basis for charging users
could be equitable from a company-wide point of view.

Riley and Williams approach the cost control problem from the
standpoint of providing reasonably accurate predictions to users of the
cost of obtaining desired information. They reason that this will:

1. Protect the most inexperienced users from very costly,

naive errors.

2. Provide more experienced users with enough information

to make intelligent comparisons between the value of
information and the cost of obtaining it.

3. Provide upper level management with improved means of

supervising system usage. [124, p. 26]
Throughout their discussion of cost prediction they seem to assume the

existence of a highly organized data base of past costs for various

activities. This assumption might hold for some firms, but definitely

30

will not hold for all, e.g., the stage one and stage two firms.

Riley and Williams indicate the need for incorporating desired
:urnaround times into the cost estimate. Their time classes are (a) crash,
(b) normal, (c) overnight, and (d) weekly. There is a premium placed
in jobs in the first category and appropriate discounts on jobs in the
Last two.

Basically their ideas are sound, but there needs to be more of a
liscussion of the classification of past costs and how these could be
lsed to predict future costs.

Witham's [158] approach to EDP cost control includes the following
deas: (l) utilize a service center for initial applications and in place
>f very small centers; (2) develop standards for input, operations, pro—
:ramming, etc. and evaluate performances by comparing with these standards;
:3) implement those jobs which have the highest potential cost savings
for the firm (e.g., inventory models and production scheduling), not
[ust those jobs where there are clerical cost savings; (4) hire a data
trocessing manager who knows something about business; and (5) get
mnagement involved in taking an active part in developing computer
Iystems and seeing that the computer is used for "profitable applications."

It is interesting to notice the different emphases regarding
:he control of EDP costs. Smith is very much budget—oriented; Witham
loes not even mention budgeting. It is possible that Smith is (unknowingly)
)riented toward stage one and two firms whose budgeting systems are very
.ax and need to be improved. Witham could be writing about those stage
:hree and four firms whose budgeting systems are already very tight and

effective.

31

Although the 1974 article, "Managing the Data Processing
Manager," by G. Hunter Jones is not primarily oriented toward cost
control, its conclusions reinforce certain ideas mentioned earlier.

Stress the business aspects of the data processing function

rather than technical issues. . . . Require a realistic

assessment of the cost-effectiveness of any proposed new
application and a reevaluation of anticipated costs and
benefits at key milestones to insure that the early indications
of the project's value have been borne out. . . . Require
technical management to provide understandable explanations

of what the department is doing and why, how it relates to

company goals, what alternatives have evaluated and why

they were discarded. [85, p. 75]

Jones placed particular emphasis on project scheduling and control.

A 1970 article by Donald L. BLack described two forms, the
computer alteration request form and the project/alteration cost-out
sheet, that can be used tOcontrolthe costs of changing existing
computer programs. [14] The first form is required whenever a modifi-
cation is to be made to computer production programs. In all except
error correction situations, the first form must be accompanied by a
cost-savings analysis supporting the requested change. After the second
form is completed by the appropriate systems personnel the estimated
cost is compared with the projected savings and the alteration request
is accepted or rejected correspondingly. The request may also be
rejected by any of the involved or affected department users. This
system has been implemented by a large insurance company and there were
plans for expanded usage of the project/alteration cost-out sheet "as
the kick-off document for a computerized system which will report on
the status of present alterations and projects, show programmer

workload, and forecast the workload over a six month period with

notice being drawn to overloads, etc." [14, p. 103]

32

In "Allocating the Corporate Information Processing Resource"
Ted Coe presents a framework which focuses on "what is gained from the
system" rather than "what the system costs." [27, p. 22] He states that
the framework "focuses on benfits because of the rather extensive work
which has already appeared on cost measurement." [27, p. 20] It would
be very helpful if he cited the references where this "extensive work"
has taken place. As mentioned earlier in this chapter, most cost/benefit
studies tend to concentrate on determining benefits, not costs. [For
example see 93] Coe himself states that in firms he observed, the

justification of new projects on a cost/benefit basis "was hampered

‘by an inability to define benefits, and in some cases, 59 define costs."

[27, p. 19] (Italics added.)

Coe proposes three categories of benefits which need to be
examined for each potential system project: (1) direct cost displacement,
(2) indirect cost and revenue changes, and (3) benefits in "key-result
areas." The third category is really just intangible benefits, and
the gist of the article is that intangible benefits should be considered
when deciding on new system projects, something that others have been
saying for some time.

One of the most interesting—-and probably controversial-—articles
in the area of EDP costs is "EDP: a720 Year Ripoff!" by Harry Larson.
[95] Some of Larson's abrasive comments about "expensive EDP people
and machines" were cited in the first chapter. Larson believes that
top managements have ignored their responsibility in the EDP area:

I've seen the ablest and toughest of executives insist on

increased productivity by a plant manager, lean on accounting

for improved performance and lay it on purchasing in no

uncertain terms to cut its staff. But when these same executives

turn to EDP they stumble to an uncertain halt, baffled by the
snow job and the blizzard of computer jargon. [95]

33

He recommends, as others have before him, that the same fundamentals of
management brought to bear elsewhere in the firm be applied to the EDP
area. He knows that problems will be encountered.

EDP people are not, in general, imbued with a commitment to

hit all three of the fundamental goals; namely schedule,

budget and performance requirements. There are EDP people

who have never seen an EDP development come on schedule, so

why should they believe it can be done—-all their experience

says otherwise. Many EDP people simply assume deep down that

hitting all three can't be done. On the contrary——it can be

done. The executive's responsibility? Insist on hitting all

three! Call a halt to the claimed immunity from normal manage-

ment controls. [95]

Some of Larson's other suggestions made in the article contain
the following points; (1) emphasize user involvement, especially in EDP
development; (2) don't be afraid to fire incompetent EDP personnel;

(3) arrange more contact between EDP managers and personnel and top
management; and (4) charge EDP suppliers (or reduce payments) with the
costs of constant changes in computer operating systems. Although Larson's
article is very critical of past EDP efficiency and effectiveness, it is

at least as critical--if not more so--of top management who have benignly
neglected the EDP area. Truly, the blame does belong in both camps.

Part of the solution to many of the problems Larson mentions may
lie in broadening the knowledge and interest of EDP personnel in the
business operations of their firm. This is the thrust of a 1975 article
by Copley. [31] While addressing other EDP personnel, he states that
"the commitment to learning the business and direction of our organization
must be just as concerted as the commitment was to pursuing data processing
knowledge." [31, p. 5]

In an article entitled "Data Processing Managers Need to Know

Accounting," Martin [101] goes a step further:

34

Required, in successively greater amounts, is the knowledge and
ability to apply conventional management techniques, previously
felt dispensable since the data processing manager was thought
to hold a unique managerial position. [101, p. 26]
Martin goes on to discuss three general areas of accounting--taxation
accounting, managerial accounting, and financial accounting--with which
an EDP manager should be familiar. He concludes that managerial accounting

is the most important for the data processing manager's needs, "although

the other areas are influential to a lesser degree." [101, p. 28]

Background on Transfer Pricing

 

We now briefly review transfer pricing literature in general as
a background for discussing transfer pricing in the EDP environment.
[For summaries and syntheses of previous writings in this area the
following are recommended: 2; 8; 145; and 149.] This portion will be
purposely brief and devoid of the usual graphs and charts that accompany
such discussions. The interested reader should consult the referenced
works for greater details and for additional references for further study.

Transfer prices are the intra—company charges at which goods and
services are "sold" by one organizational unit to another. [61, pp. 435-
436] A complete history of the literature of transfer pricing is neither
possible or needed, but a brief summary will serve to introduce specific
models in the area of EDP services.

Major works on transfer pricing began to appear in the mid-50's.
Paul Cook and Joel Dean each highlighted problems that arise when transfers
occur within a decentralized firm. [29; 37] Each author described and
analyzed the major types of systems being used and reached essentially
the same conclusion: that divisions should operate as profit centers

using market-based prices for intracorporate transfers.

35

Hirshleifer [76; 78] used a rigorous economic analysis of the
problems of transfer pricing and concluded that Cook and Dean's market-
price approach could be used only where the commodity being transferred
was produced in a purely competitive market. Imperfectly competitive
markets implied the use of marginal cost or some price between it and
the market price. It should be noted that Hirshleifer's solutions assume
technological or "cost independence"--the level of operations in one
division will not affect the cost function of the other. This assumption
is obviously invalid when discussing transfer pricing of EDP services
where the services are often used ("bought") to reduce the buyer's costs.

Most of the transfer pricing articles in the 1960's dealt with the
effects of intracorporate pricing on divisional performance, evaluation,
and profit measurement. Among the exceptions was Dupoch and Drake's [47]
proposal of using marginal costs plus "shadow prices" when market prices
were not available and there existed alternative uses of fixed facilities.
Recently Onsi [118] has used a short example to illustrate the problems
in transfer pricing when such opportunity costs are not considered.

One of the better theoretical and applied treatments of transfer

pricing is found in Solomon's Divisional Performance, Management, and

 

Control. [145] His analysis brought together much of Hirshleifer's work
and discussed a few additional elementary cases. Horngren [81] and
Shillinglaw [133] have brief but excellent discussions of transfer
pricing in their respective cost accounting texts.

Gordon [62] has proposed a cost-based transfer pricing system to
allocate resources in a socialist economy that may also be used in the
administration of a large, decentralized firm in a capitalist economy.

This system would probably be very expensive for a single firm to use,

36

and Gordon himself has indicated that "the simpler systems currently in
use are adequate for the decentralization problems such firms encounter."
[62, p. 427]

Some of the types of transfer prices that have been proposed
and/or used are now briefly summarized. Market prices are generally
held to be ideal if the market for the intermediate product is perfectly
competitive, in which case the market price is the opportunity cost to
the firm of not selling units of the product on the external market.
In the absence of perfectly competitive markets, which are seldom
present, there seems to be at least academic agreement on Hirschleifer's
range--marginal cost or some point between it and market price. Marginal
cost is considered correct if short-run capacity is not a limiting factor.
If capacity is limited, the opportunity cost of the firm as a whole of
producing for internal transfers should be added to marginal cost to
arrive at the transfer price. (These opportunity costs are usually the
"shadow prices" from a linear programming algorithm.) Most of the above
methods can in some way (at least partially)txatraced back to Hirschleifer's
analysis and therefore rely on his assumptions of technological independence.

Although market-based transfer prices have been used in some firms,
cost-based prices (usually full cost or full cost plus a markup) and
negotiated prices have been widely used. Negotiated prices have been
criticized for their artificial nature and limited use for decision making
and evaluation purpose. [149, p. 57] The power of the divisional
negotiating committee may be unbalanced and/or the division may come to
regard the negotiation per se as the most important determinant of their
profit. However, a recent examination of transfer prices from a behavioral

context leaned heavily in favor of negotiated prices. [156]

37

Full cost (and cost plus) has been widely criticized because
its usage can easily lead to suboptimal decisions. The use of actual
costs may pass on the inefficiencies of one division to another; full
standard costs may minimize this problem, but suboptimal decisions may
still result. [For a very simple example see 81, p. 741]
Solomons proposes an interesting method where no usable market
price exists and the supplying division can meet all probable requirements.
His solution attempts to retain the decision-making advantages
of marginal cost and yet provide for recovery of fixed cost.
The latter makes the solution more long-term, and satisfies
administrative requirements for recovery of all costs. Solomons
claims, in addition, that it permits more accurate performance
evaluation.
His solution consists of dividing up costs into two parts.
The marginal costs (approximated by variable costs) become the
transfer price and represent the product cost. Fixed costs are
recovered as period costs through a lump sum charge based on
expected usage of of capacity. The lump sum charge is calculated
from budgetary expectations and is not altered if there is a
volume variance. This ensures that only the variable (or
marginal) costs enter the short-run decision models as inputs.
It is a useful modification to the marginal cost method in
situations where marginal cost is constant or declining.
In addition the lump-sum charge for ces the buying division
to consider the long-run costs of utilizing the selling division's
capacity. [149, p. 61]
Turney's dissertation describes how this method might be applied to the
EDP area. [149, pp. 62, 166-172] There are some disadvantages to
this method, but they will be considered later in the discussion of
Turney's work.
Part of the transfer pricing problem is the variety of requirements
that a transfer pricing system must meet, namely: (1) goal congruence--the
Intansfer pricing system must motivate profit center managers to pursue

‘flleir own self-interest in a manner which is conducive to the success of

the company as a whole; (2) performance evaluation--the transfer price

38

must allow central management to evaluate as accurately as possible the
performance of the profit centers in terms of their separate contributions
to corporate profits; and (3) autonomy--if top management wants to preserve
a decentralized structure, it must allow the subunits to make decisions

as decentralized entities. [81, pp. 697, 730; 126, pp. 99-101] It is

not hard to see that these requirements may often be conflicting.

In a recent attempt to synthesize various approaches to transfer
pricing, Abdel-khalik and Lusk suggest that possibly this second require-
ment should be dropped or eliminated. In summary they stated: (1) "per-
formance evaluation does not have to be a function of profits,"

(2) "transfer pricing may blur the evaluation perspective when the
evaluation of performance is strictly profit oriented," and (3) "the
same degree of control and evaluation attempted through transfer pricing
models may be effected through the setting of standards of divisional
performance and the evaluation of deviations from these standards."

[2, p. 23]

Although Abdel-khalik and Lusk stress the limitations of transfer
pricing for performance evaluation, they do not imply that an intrafirm
price mechanism is useless as a tool for allocating scarce resources.
Most attempts to apply transfer pricing to EDP services have had
efficient resource allocation as a primary goal. [For example see 106,
p. 125; 154, p. 15] Before examining various proposed EDP transfer
Pricing methods and models, let us examine some of the advantages and
disadvantages of charging users and criteria that haVe been proposed for
(Evaluating EDP transfer prices. Also discussed early in this section
are some more general articles on transfer prcing EDP services which do

not promote a specific method.

Why7Charge?

39

EDP Transfer Pricing Literature

 

In his NAA study, Sollenberger lists a series of advantages and

disadvantages of charging users for systems services. This list also

summarizes much of a related article discussing the same area. [142]

Because any attempt to further summarize or condense this list would

result in the loss of key ideas, it is quoted in full here:

The pricing system can be beneficial to the systems area.
Support among the firms in the study usually was related to
several of the following points:

1.

The need for user involvement cannot be overemphasized
since these people must function with the system. By
charges directly to their budgets, their interest and
support may be easier to obtain and maintain.

Allocation of computer resources throughout the firm
might be aided if each user has to have the support of

his superiors before systems and data processing costs are
incurred for his benefit.

As part of internal control, the matching of men and
machine hours to a budget can help control performance

in areas which have had budget and schedule problems.

The charging system can also provide a means of

justifying substantial increases in personnel and dollars.
If user areas are able to show the need for additional
services and are willing to pay for these items, budgetary
approval for more men and equipment may be more easily
obtained.

It forces the information services manager to provide
quality services, which will generate demand in other
units of the firm at the prices indicated.

Dependence on the transfer pricing device can also have
undesirable ramifications on systems development and computer
services. Criticism of charging schemes were typified by one
or several of the following points.

1.

The use of a pricing device as the sole allocating tool

in this area of high demand for computerization is not
effective. In some cases managers of these services failed
to meet the responsibility of allocating the scarce resources
which were assigned to them and have used the charging
mechanism as a poor substitute.

The user departments may use a narrow outlook when
examining their data handling problems and fail to see

the company-wide view of of the value of information.

Thus, it is more difficult to obtain broad support of

the concept of integrated systems if charges for this

work are absorbed by individual units.

40

The pricing system may force those areas in greatest
need and least able to buy the services to forego the
services and use less desirable alternatives.

Certain specialized services and capacity to meet peak
needs may become unavailable due to the undesirability
of high rates compared to external suppliers if a full
cost transfer price is used.

As a side effect, the implied precision of cost figures
common in many project justifications gives rise to the
desire for precise benefit measures for return-on-
investment analysis, which have not been proven.

The problems of cost assignment and allocation may be
so severe that the worth of the billing system may be
impaired by arbitrary allocations, may be made meaningless
because of noncontrollability of costs or may be too
expensive to justify its existence. [142, pp. 129-130]

More firms charged for ongoing computer jobs than for systems

development work, and there was more of a conflict in opinion over

whether the latter area should be included in an EDP pricing system.

Since Sollenberger's study is essentially descriptive in nature, he

does not prOpose a specific model for charging either systems or operations

costs to users. He does cite some factors for consideration in each area.

In charging for the systems design area, the following points should be

considered.

1.

The validity of a full-cost charging system where all
systems costs are billed out is highly questionable in
view of nonproductive and nontracealbe costs present.
Flexibility must be provided to all for differences in
project scope, user involvement and sponsoring operating
areas.

Preliminary and economic feasibility studies are probably
'in the company interest' expenditures and not specifically
chargeable; also certain projects are of such general interest
and affect so many managerial areas, that costs for these
might be borne by the corporate systems unit. The charges
for systems development work should probably be based on
work done once the project is formally approved.

The charge should be against the sponsoring manager.

The charging system should be simple and easily understood.
[142, p. 132]

The following desirable attributes of a charging system for

computer operations were listed by one firm in his study:

'41

1. Understandability and ease in administration are prime
factors in managerial acceptance and use.

2. The charge should be equitable from a company-wide point
of view, provide consistent charges for given work,
consider priority graduations and represent the cost of
facilities utilized.

3. The charging procedure should be stable over a period of
time to allow for valid basis of evaluating implemented
systems which were developed under certain operating
cost assumptions.

4. In all probability the accounting system used will be
neither a full nor marginal cost approach and will try
to allocate large fixed costs (salaries and hardware costs)
over many variable activities (orders, employees, etc.).

5. The algorithm should be a simple as possible considering
the above constraints. [142, pp. 133-134; see also 143, p. 28]

Sollenberger presents an example of a charging algorithm used by
one firm which detailed the specific computer resources used (e.g., CPU
time, I/O time, disc and tape storage) and multiplied these by constant
coefficients to develop a total charge. [142, p. 135] He briefly
criticizes the algorithm for its complexity, but adds that complexity
may be necessary in a multiprogramming/muliprocessing environment.

Heitger used a deductive approach in reasoning from corporate
EDP objectives to criteria for an EDP transfer pricing system. However,
since his conclusions were quite specific, his work will be discussed
later with specific transfer pricing models and methods.

In a chapter entitled "Management Control of the Computer Resource"

in Information System Administration, McFarland_g£.§l. state that a good

 

management control system must possess three important attributes:

1. It must have mechanisms to monitor information on the key
aspects of employment of computer resources necessary for
evaluation.

2. It must have mechanisms for communicating the information
to decision-makers.

3. It must have mechanisms to motivate decision-makers to take
action which facilites organizational goal realization.
[107, p. 467]

They discuss briefly the inital resource commitment to the EDP area and

42

the control of systems development and computer opeations. The unique
supply and demand characteristics of computer systems are described.
(These were cited in Chapter One, see page 6.) However, most of the
chapter is devoted to approaches to charging users for EDP resources.

The concept of charging users for the computer resource is

the core backbone of virtually all organizations' approaches

to the management control problem. [107, p. 473]
McFarland g£_§l. then discuss non-charge-out systems, full charge-out
systems, and partial charge-out systems, relying heavily on case study
examples. They conclude the chapter with a conceptual framework for
controlling the computer resource and with a checklist of questions that
should be asked before making a charge-out/non-charge-out decision.
[107, pp. 482-484] A 1973 article by Dearden is essentially the same
as the above-mentioned chapter. [40] However, the questions at the

end of the article are better organized and answers are presented in a

charge-out[non—charge-out spectrum. Some examples are: [40, p. 78]

Spectrum
U u U
a s s
o .4 o o
w m m o
00 +400 on
I u u u .4 u
o m u m .4 m
0.: m.c 3.:
20 940 mu
Are users knowledgeable about the costs
and limitations of computers? No Yes
Are users highly susceptible to
"overselling" of the computer resource? Yes No
Is the company's operating philosophy
one of decentralization? If so, do the
management control system and the loca-
tion of the resource reflect this fact? No Yes
Do complex priorities make it difficult
to generate needed management
information on schedule? No Yes
Is it necessary to monitor and control
EDP management closely? No Yes

Figure 2—1. Helpful questions in determining a charge-out policy.

43

A 1973 article by McFarlan stresses two key structures that
management needs to control the EDP area.

The first is a financial reporting system that allows
it [management] to do the following things:
--Review the department's performance on a periodic basis.
--Compare the department's development against formal
plans for it.
--Check the functioning of the department's project
control systems.
The second is a structure that links the responsiblity for
various departmental decisions to the operations of users—-
ordinarily other company departments. Generally this structure
is a procedure to account for EDP expenses, either on a chargeout
or an overhead basis. [106, p. 133]

There is an excellent discussion of some advantages and disadvantages
of chargeout and overhead accounting for EDP costs. McFarlan argues
quite strongly for the use of chargeout systems. He does not discuss
the way the transfer price would be determined.

Although many authors favor EDP transfer pricing, few trace
a transfer price back to its origin--the EDP cost system. Anderson is
an exception. In a 1974 article he states:

Implementation of a direct charge system requires the
development of an integrated, comprehensive cost accounting
system to supply the necessary charges. [7, p. 29]

He goes on to state some desirable attributes of an EDP cost system:

1. Comprehensive in scope--a11 relevant cost centers should

be included.

Costs should be categorized by project.

3. The user charge should contain an adequate but not excessive
amount of detail regarding resources used by individual
systems (adequate detailed data ought to be readily avail-
able to permit research regarding special problems or
questions that may arise).

4. The cost algorithm should be reasonably simple and easy to
understand to encourage user acceptance and comprehension.

5. User charges should reasonably and equitably reflect the
costs of resources actually commanded by the user.

6. The charges should be stable over time.

N
o

44

7. There should be flexibility in dealing with special situations;
some costs may need need to be absorbed by the EDP area (e.g.,
costs for projects with substantial long-run benefit to the
company but with no immediate sponsor).

8. Economy of operation--costs associated with collecting data

and generating reports should be minimized. [7, pp. 29-30]
Anderson gives examples of project cost reports, a charging algorithm
based on detailed resource measurements, and a charging algorithm
based on the user's primary activity unit. He favors the latter type
of algorithm but believes that the detailed approach would probably
provide more useful information for evaluating system design alternatives
within the EDP area itself. [7, p. 31]

In a 1975 article, John B. Wallace argues that in deciding
whether or not to bill for computer services consideration should be
given as to how decisions are currently made in the organization. [154]
If transfer pricing is already used effectively in a decentralized
environment of "strong, mature managers," then billing is likely to be
effective. If departmental budgets are controlled centrally and "there
is a strong tendency to hide one's expenses in the budget of other
departments," billing users for EDP costs should be avoided. [154, p. 17]

While Wallace's article is quite wide-ranging and a review of
all the points made therein is impossible, it does contain an interesting
list of dicisions which nine data center managers said utilize charging
data on the cost of systems development and/or the cost of computer
operations. These decisions are:

1. Application selection.

2. Project control.

3. Equipment utilization.

4. Computer resource development decision (disc vs. core

storage, etc.)

5. Centralization/decentralization of systems develOpment

and/or computer operations.

6. Make-or-buy; internal vs. external processing.
7. User evaluation decisions. [154, p. 14]

.u'
9..

L)

45

Although a complete ranking of decisions was not given, charging data
were judged most useful by data center managers in the first three

decision areas.

Specific Charging Methods

 

Specific methods of determining EDP transfer prices that have
been used in the past or that have been proposed as models are now
examined. The discussion will proceed along the following lines.

1. Full Costing Approaches
A. Average Full Cost
B. Cost-plus
C. Standard Full Costs
II. Marginal Costs
111. Other Models
A. Turney's Model
B. Value-based Allocation
C. Axelrod's Model
D. Nunamaker and Whinston's Model
E. Flexible Pricing
IV. Other Considerations
A. Priority Pricing
B. Multiprogramming

Average Full Cost

Some of the earlier methods of developing unit costs and of
pricing EDP services came to be used because of government regulations
essentially requiring an average cost method. [88] University computer
centers, because of substantial government contracts and grants, were
particularly affected by this regulation. The price paid by the govern-
ment was established ex post, following an audit of cost and utilization
data for each accounting period. "Provisional prices," based on

estimates of utilization, were used in the interim.

Th:
:i their Cc
2 £11125:
3 unit 0
231.353' t

A]
ggputer 1
gamine
{ESOQICES
[be life

jainishi

:enter ‘
H

were h
9 tion:
A volur

in do

unless
time,
] ,

.0. pr

has no

under

46

There were great incentives for universities to restrict the use
of their computers, even at times when utilization was low. Any increase
in utilization of time otherwise idle reduced the computed average cost
per unit of time and therefore the price paid by government projects for
computer time.

Although most corporations do not sell extensive amounts of
computer time to the government, the use of ex post average full cost to
determine unit costs causes other problems in the allocation of computer
resources. Smidt [136] has suggested that a typical demand pattern over
the life of a computer is excess capacity when acquired, gradually
diminishing until demand exceeds capacity later in its life. If this be
the case, the use of average unit cost results in exorbitant charges
early in the system's life, counteracting possible attempts to encourage
usgae of the new facility. Late in the system's life, when demand needs
to be limited, average costing would result in relatively low charges.

Sollenberger cites an example of one special purpose data
center which used an average full cost approach to charging. Its rates
"were high because of the special services and applications . . .
performed-~too high to attract routine high volume applications. . .

A volume decline and rate increase sequence followed which finally ended
in closing the data center." [143, p. 40]
A further criticism of the average cost pricing method is that,
'"114333 it is coupled with various rates for different classes of computer
tilnee, it fails to distinguish and discriminate between high priority and
1"‘°‘ priority users. A user having a great need for immediate turnaround
lléi§3 no way of purchasing that priority service; another user who is not

ll .
tlci(er such time pressure has no incentive to purchase off—prime-time SETVlce-

1.... ,'
L..»K.

* a

330:1

COSt

47

In a university setting, a student given a fixed budget for computing
time might wish to "buy" more hours of lowbquality service to stretch his

budget.

Approximating;Market Price--Costep1us

 

Krasney has proposed a cost-plus transfer price to approximate
an outside market price which is usually not available. [94] Standard
costs would be determined based on expected activity levels. A "profit
margin" would be added based on the firm's average or expected rate or
possibly the average ROR of comparable computer service firms. Jobs
are submitted to the EDP department which accepts them if estimated
benefits exceed the estimated variable costs of the job.

Revenue to the computer center would be determined by multiplying
the actual quantity of computer service provided by their respective
substitute market rates. The expense charged to the user is some
portion of the corresponding revenue recognized by the EDP department.
The exact amount is determined by the ratio of the estimated benefits
(submitted by the user) and the EDP department's revenue. The maximum
charge to the user would be the computer installation's revenue.

Heitger has raised a number of timely questions about Krasney's
Plan. [71, pp. 52—53] In particular, users would be strongly motivated
‘13 estimate benefits at just greater than variable costs, minimizing the

aHRDunt of expenses charged to their department and misusing one of the

Very reasons for transfer pricing.

Sisififlpdard Full Costs

 

The use of a practical capacity or standard volume as a unit

Cost: standard for allocation purposes in conjunction with full costing

1
several
mjectl‘
at more

are:

Given t
[142, F

Pricing

Heitge
establ
EXPECt

SYStem

 

48

eliminates some of the problems of actual full costing. Heitger [71]
has advocated standard full cost as the best transfer pricing method in
several cases. He reached this conclusion by examining corporate EDP
objectivaes and then establishing transfer pricing criteria to meet one
or more of the objectives. The five corporate EDP objectives he discovered
were:

1. Automate all jobs that can be justified on the basis

of cost reduction.

2. Automate all jobs which will reduce the amount of
peripheral minutiae with which management must deal.

3. Maximize the utilization of EDP resources.
4. Facilitate the creation of change.
5. Give the appearance of being progressive. [71, p. 108]

Given the first two objectives, Heiger (relying heavily on Sollenberger
[142, pp. 133-134])develops the following criteria for an EDP transfer
pricing system:
1. The charge should be based on a predictable rate that is not
affected by the volume of activity of the EDP installation.

2. The charge should be based on actual use.

3. The charge should be levied against the person having
decision authority over the job.

4. The charging algorithm should be sufficiently detailed to
approximate closely the resources used.

5. The elements of the charge should be clearly identified and
as understandable as possible, so the users will be aware
of the cost of the resources they are using.

6. The charge should not be expensive to administer. [71, pp.
109-110]

Heitger concludes that this set of criteria can best be met be
establishing a transfer pricing scheme based on standard costs at the
exDected level of operations envisioned by management when acquiring the

£33’E3tems resources currently available.

. A ‘

if :01.

acme
set of

ashor:

a: best

whet:
full 0
includ
cost c
be add
diVisj
[fans]
will 1

PD. 4-

49

For firms having the objective of maximum utilization of EDP
resources, Heitger replaces the first criterion in the above list with
the following two while the remaining five are the same:

1. The pricing scheme should encourage the utilization of
unused systems resources.

2. The charge should not result in an incremental loss to
the firm. [71, pp. 113-114]

He concludes that a marginal cost transfer price best satisfies this
set of criteria. Heitger contends that this method can only function on
a short run basis. Most of the system managers he interviewed felt that
a marginal cost charge would swamp their installation with job requests
in a very short time.
Objectives four and five were questionable corporate objectives
at best and provided no basis for solving the transfer pricing problem.
In one of the earliest studies comparing EDP transfer pricing
methods, Wormley examined marginal cost, actual full cost, and standard
full cost. [163] Wormley rejects marginal costs for a number of reasons
including: (1) the difficulties of determing and equalizing the marginal
cost of the transferor division and the marginal revenue (which, it might
be added, would often include intangible benefits) of the transferee
division, (2) most EDP costs are fixed costs, and (3) marginal cost
transfer prices will drive a given system to capacity very quickly and
‘Will fail to force an economic utilization of capacity increases. [164,
PP. 4-5]
Wormley noted three objections ot actual cost transfer prices:
(1) since the price must be calculated after usage has occurred the cost
I)€313 computer hour is not known to a user when he needs the data and can

(:(3tltrol his utilization of the resource; (2) the unit cost charged is

.-. ...,”
,5 but-

3;
E '.
C

11% 33
23a: 1

..p-‘
1....“

n-Q
«[112
ml;

stand.

50

is unknown when budgets are prepared; and (3) the EDP center is not
encouraged to control its own spending. [164, p. 5]

Standard full cost is advocated primarily becuase it overcomes
the above three objections to actual costs. In addition, Wormley says
that it is "relevant, understandable and simple." He does not discuss
which particular measure of EDP capacity or utilization should be used.

Sobczak [139] does not specifically advocate "standard full
costs" but rather "cost reflective pricing." However, since he advocates
"actual costs," but with the rates based on final expected level of
utilization, his rates may be, for all practical purpose, considered
standard cost rates.

Sobczak argues that two conditions are necessary for a firm to
efficiently use its computer resource. (1) "Cost reflective" rates must
be available and charged to users. (2) Users must react to these costs
in an economic fashion. [139, p. 63] His emphasis on user reaction is
important in evaluating all types of transfer pricing systems.

If users do not respond in a cost sensitive manner--in other

words, the prices set on DP services are not a factor in their

decision-making process-the DP accounting system has not
provided any benefit. In fact, the firm will probably incur

a loss since there is usually substantial overhead

associated with establishing and maintaining an internal

transfer payment system. The point is that distributing

costs among the user is not 'good in itself,‘ and the user

must react in order for benefit to be derived. [139, p. 62]

MO8t of Sobczak's economic analysis is based directly on Sharpe's

el‘fiélmples. [131, pp. 442-446] It is interesting, however, that
Sharpe concludes in this particular example that marginal cost is the
aPDropriate transfer price; but Sobczak favors standard full cost.
E;()I>czak seems to be mistaken in his analysis of the behavior of EDP

(:(353ts. For example, he states that "the majority of costs in most

Q 0
c)‘llllputer centers are usually of a variable of semivariable nature."

¥

.o-jl I

.
pnrqo I
5:54

:he-

1351.

fi-tv-s
“Wk.

..

\-- a .

“Airy:
O

“a '4'!
315:5!

51

[139, p. 63] Contrast this with Dearden and Nolan's statement: "The
ratio of fixed to variable costs is high." [40, p. 69; see also 107, p.
471; and 149, p. 132] If Sobczak views fixed costs as negligible,

then his standard full cost would be very close to Sharpe's marginal
cost.

Chapter Eleven of Sharpe's The Economics of Computers is

 

devoted to pricing computer services with emphasis on internal pricing.
Sharpe discusses several different situations. His first example [131,
pp. 442-445] is of a marginal cost transfer price, but this was later
discarded when the assumptions underlying the method were stated and
found to be often unrealistic. In the case of a firm having fully-
utilized, fixed capacity, he concludes that the appropriate transfer
price "may be more or less than average cost." [131, p. 455] This
leaves the reader somewhat in the dark as to which way to proceed.
Moreover, it would be extremely difficult (and costly) for a firm

to develop the total value, marginal value, and marginal cost curves

on which Sharpe's analysis is based.

Marginal Costs

 

In the absence of an outside market for the transferred commodity,
many economists have argued for a transfer price based on marginal cost.
Ho'wever, few have strongly advocated it as a hard and fast rule for EDP
transfer pricing. Heitger indicated that marginal cost should be used

iiE; (and that is a very big "if") one of the corporate EDP objectives is
to lIlaximize the utilization of EDP resources. [71, p. 114] Sharpe's
initial internal pricing example concluded that marginal cost was an

appropriate price. [131, pp. 442-445] However, this example assumed

k

(\b
...

1‘4

-. .
... \u
l‘ I

“H 5|.

"'1“

52

"an equilibrium point can be found at a utilization for which marginal
cost is defined. Marginal cost is not defined at points at which the
total cost curve exhibits a kink [e.g., increments of fixed costs],
and it is clearly not defined when utilization reaches capacity."
[131, pp. 200-204]

' a marginal cost

In the tradition of "Solomon's Solution,'
transfer price which is accompanied by a fixed—cost charge has been
proposed for the EDP area. Since this differs from a strict marginal

cost price, it will be discussed in the next section.

Turney's Model

 

Turney [149] has concluded that marginal cost (approximated by
variable costs) should be the basis for transfer pricing EDP services.
However, users would also be charged fixed costs in lump-sum amounts.

He breaks down EDP costs into five categories and suggest cost assign-
ment for each as follows:

1. Development Costs (systems design and analysis; programming);

allocation based on negotiations between the user and the

EDP department.

2. Incremental Costs (program maintenance and conversion costs);
allocation again based on negotiation.

3. Capacity Costs (hardware; some software and personnel costs);
allocation based on expectations of user's share of
practical capacity; fixed for the budget period regardless
of actual usage.

4. Variable Costs (data entry; supplies; some personnel costs);
charge users standard variable cost per unit.

5. Congestion Costs (opportunity costs of excessive turnaround;
a short-run capacity cost); charge users based on actual
usage of computer resources and the desired priority.

All of the above allocations are based on standard costs and users are

not charged for inefficiencies in the EDP department.

¥

.q.
...; ~¢

33;..A-

4
4
4

cu
'a
4

/ :

m.
‘u

/_.

53

Possibly the greatest problem with Turney's approach lies in his
handling of the capacity costs. Although this is a lump-sum charge to
users, he stresses that it should be itemized according to the different
resources utilized by the user. This entails keeping track by users of
such items as CPU time peripheral processing time, core used, etc. The
user would not be immediately charged for his actual usage of these
resources, but the amounts used in one budget period would be the major
input into his lump-sum charge for the next. This time-lag in charging
users seems to be undesirable if these charges are to be used at all in
the performance evaluation of either the user or the EDP center.

Another problem with Turney's analysis if that the costs he
includes in variable costs are a poor approximation of marginal costs.
He apparently ignors the step-wise increases in hardware costs as the
"practical capacity" of a given system is reached and new components

must be added to extend capacity.

Value-based Allocation
Another method proposed by Sharpe is an attempt to directly
maximize the value of the computer installation to the firm. [131, pp.
469-480] Users must be willing to describe in dollars the value associated
With the completion of a job at various times of the day. These values
are assumed to be the same as those for the firm as a whole. Linear
Programing is used to determine the timing (or scheduling) of each

1‘31) and the "cost" (including opportunity costs) it should be charged.

f325§ilrod's Model

 

Axelrod [11] has modeleda control and information system which

71E: claims efficiently allocates computing resources assigned by an

.m’ c

..u-

.4.
.r

...u

54

organzation to its internal users by means of maximizing a general value
function. His model affects the allocation of computing resources by
means of the following controls:
1. The long-term assignment of the computing requirements
of internal users to either internal or external
computer facilities.
2. The medium-term dynamic control of the level of user
demand for services of the organization's internal computer

facility.

3. The short—term scheduling of jobs that have entered the
internal computer.

Number one is solved by the setting of appropriate transfer prices by
a centralized control unit. An iterative procedure based on changing
submission times and turnaround times resolves number two. Number
three requires the use of various queueing techniques such as FIFO
and SJF (shortest job first).

Axelrod's model can be criticized on several points. (1) his
heavy reliance on "central control" would seem to invalidate the model
in decentralized circumstances where divisional (or departmental) managers
are encouraged to make their own decisions about their utilization of
computer resources. (2) The costs of implementing his model (and other
models attempting to incorporate various opportunity costs--e.g., Sharpe's
"value-based allocation") seem very high on an intuitive basis. It is
possible to imagine further theoretical extensions that consider more
factors than submission and turnaround times (e.g., accuracy of processing,
volume of transactions processed, etc.) with a great growth in complexity
and administrative overhead. The examples used in his thesis are quite
small. Applying his iterative control model to a medium or large size

data processing center would seem prohibitive in terms of both time and

n.

f)
O
.'“1

55

costs. (3) It would be difficult to incorporate Axelrod's transfer
prices into a performance evaluation scheme (of users) based on either
costs or profits. The transfer prices do not at all reflect the cost
of resources being used. Also, because of the lack of cost information
about internal computer services, it would be difficult for users to

perform cost/benefit analyses for various applications.

Nunamaker and Whinston's Model

Nunamaker and Whinston [117] have proposed a cost allocation
procedure base in part on the user's cost of obtaining the service from
an alternative (external) source. In many respects this method is
very similar to that recently proposed by Mbriety. [109] Nunamaker and
Whinston apportion the cost savings of having a central computer service
versus the use of alternative, external sources to the various users
based on the probability of a given user incurring the initial fixed
costs of the inhouse center (i.e., being the first job on the system).

Their method implicitly assumes that the total cost of the
in-house system is less than the total alternative cost of acquiring
those services. Charges exactly equal total actual costs; in fact,
the authors apparently considered a transfer pricing approach (rather
than a cost allocation procedure) but rejected it because it did not
guarantee that the full costs of a facility would be covered. However,
Sollenberger [142, p. 134], Sharpe [131, pp. 453, 455], and others
have contended that full cost allocation is not necessary to efficiently
allocate the computer resource.

Nunamaker and Whinston's method would be extremely costly and time

consuming if a large number of users shared a central facility. There

ser a
tuna:
gr: .

ZCEI‘.

3.55311

Elle-

01"

56

would have to considerable effort expended on determining external prices
for all potential jobs and in recalculating the cost allocations every

time a new job is added.

Flexible Pricing

 

Smidt [135] has proposed a flexible pricing scheme where the
user may specify either the maximum price he wishes to pay or the maximum
turnaround time he can tolerate. Sharpe [131] has pointed out, that
given the known parameter (price or turnaround time), the other would be
uncertain. In fact, in a highly congested system there is no real
assurance that the maximum turnaround time constraint could be met at
any price. The major problem with flexible pricing schemes appears to
be the uncertainty resulting from them. Especially in the case where
turnaround is the specified parameter it seems rather pointless to be
constantly varying the price of EDP services when most corporate users
cannot easily adjust their demand for that resource. Sollenberger [142,
p. 134] has pointed out that one of the desirable attributes of a
charging algorithm is stability over time. Otherwise there is little
valid basis for evaluating implemented systems which were developed

under certain cost assumptions.

Priority Pricing_

Sharpe [131, p. 464] and others have suggested that a priority
pricing scheme can easily be incorporated into another method. The basic
idea is to adjust prices so that computer time is cheaper during off-
peak hours. This could be done by means of discounts for jobs run during
evening or weekend shifts, or by premiums on prime time applications,

or both. Sharpe suggests that an alternative is to maintain equal prices

w...”
:3ng

tom

Var:

3'3'8 1

ati<

0:361

CW

57

at all times thereby letting the increased turnaround time be a "price"
that peak-load users would have to pay. The problem with this proposal
is that there is no way for users who demand fast turnaround at peak-
1oad times to buy it at any price. Users should be allowed, at least
in a limited way, the option of paying higher prices for faster

turnaround and/or daytime (peak-load) service.

Multiprggramming

Perhaps no other technical innovation has resulted in such a
spate of transfer pricing related articles as has multiprogramming. While
initially there was some diversity of views regarding approaches to this
problem [See 71, pp. 53-55, 139-146; see also 42; 43; and 138], continuing
advances in systems software and in internal monitoring packages have made
it feasible to measure the usage of individual computer resources (CPU,
disc drives, tapes drives, I/O equipment, etc.) for each job. More
recent articles touching on this problem have agreed that job costs should
be broken down by resource usage and that this is feasible with current
technology. [17; 68; 157] Example algorithms may be found in each of

the articles cited.

Chapter Summary

In this chapter we have reviewed the EDP cost control literature,
touched briefly on transfer pricing literature in general, and examined
various EDP transfer pricing methods and models. A few summary points
may be drawn where several authors are in substantial agreement. (1) The
systems development area is more difficult to control than computer oper-
ations. Emphasis should be placed on a project reporting system. (2) The
operations area (and to some extent the system design area) can be

controlled using many of the same techniques used elsewhere in an

.
1ofi:“‘
“cw-A

v...

F
s lc‘

C35:

inte

to

the

58

organization. In many instances, upper management, bewildered by the
complex technology, has been hesistant to apply these tools to the EDP
area. (3) There is general agreement that a pricing mechanism is an
effective means of controlling computer usage if handled properly. There
is less agreement about the effectiveness of a pricing mechanism as the
sole means of allocating scarce resources. Most authors have spent
more time on "what pricing method?" rather than asking whether pricing
should be used at all. There is some agreement that transfer pricing
could be effective not only in controlling computer operations but
also in the system development area. Recommendations have usually leaned
toward standard costs or a contracted amount.

Although continuing to grow, the proportion of EDP literature
related to cost control is still quite small. Much of the approach has
been indirect rather than direct-~charge users for services to shift the
costs to other parts of the organization. Although Heitger was particularly
interested in transfer pricing, his comment holds for all control aspects
of the EDP area: "A topic as important as this should certainly have a
more comprehensive evaluation than the literature showed." [71, p. 57]

One thing is clear from the literature. No one as yet has attempted
to crystallize a cost accounting model to control the EDP area. This is

the subject of the next chapter.

CHAPTER III

EDP COST SYSTEM FOR MANAGEMENT CONTROL

Introduction

 

The need for a model EDP cost accounting system (not just a
transfer pricing system) has been implied many times in the literature.
In his conclusions, Turney stated explicitly:

Some companies that were looked at possessed only the rudiments

of a cost accounting system for MIS. Those companies that did

have a cost accounting system were still uncertain as to its

correct form. [149, p. 192]

This chapter outlines a model cost accounting system for the EDP area.

In a sense this job is difficult for the model is constructed for a
hypothetical firm. In order to have a very broad (and possibly vague)
model which will cover a very large number of real-world firms, the

number of assumptions made about our hypothetical firm would have to be
greatly limited. Conversely, a great number of assumptions about the
hypothetical firm can be postulated with a resultant, detailed model which
applies only to an extremely limited number of firms.

Hopefully, the middle path has been chosen. The assumptions
made below do restrict the number of firms to which the model will apply.
However, they do permit a good deal more of specificity regarding the
model's components. In a few instances the discussion of the model may

deal with alternatives that arise from relaxing one or more assumptions.

In general we are assuming:

59

CG]:

,
"i

3.131;

6O

(1) profit-oriented firms; governmental and non-profit
organizations are excluded; however, some of the
conclusions may apply to these types of organizations;

(2) medium to large firms; firms spending at least $1 million
a year on EDP activities;

*
(3) some degree of centralization of EDP facilities;

(4) multiple users are served by the EDP facilities within
the firm;

(5) several years experience with EDP; users are relatively
knowledgeable about EDP technology.

The model will be based in general upon traditional cost accounting
literature. As noted in the previous chapter, several authors have
concluded that the techniques used to control other areas of the firm
should be applied to the EDP area. Thus, most of the following concepts
(with the possible exception of number six) have been used for years

elsewhere in business firms.

Components of the EDP Cost System

 

The components of the cost system are:

(1) Low level of cost collection and aggregation.

(2) Responsibility and cost centers.

(3) Budgeting and reporting system.

(4) Output or activity measurement and performance criteria.
(5) Long-range financial planning.

(6) Tracking users' EDP-related costs.

(7) Charging system for users.

 

*This assumption appears well-grounded. Several authors have
commented on the growing trend to centralize corporate computer facilities
and predict that it will continue. [53; 82; 86; 159] Complete centrali-
zation of EDP facilities is not assumed, but where an alternative, such
as "distributive processing," is used, this is presumeably done on an
integrated basis.

61

Number one through four of the above are also found in Sollenberger [140]
and part of the discussion that follows is based upon ideas from that
paper. Following the discussion of each component, an itemization of

the factors used for evaluating the results of the field studies is

given. These factors are summarized in Figure 5-1 on pages 165-166.

Low Level of Cost Collection and Aggregation

 

Shillinglaw [133] has suggested that the typical set of cost
accounts is three-dimensional; i.e., costs may be classified in three
ways:

(1) By organizational unit.

(2) By descriptive element.

(3) By product, project, program or service.

Applying this to the EDP area we may say that costs are often first
classified by organizational units or subunits--computer operations,
systems design, Operating systems support, data entry, etc. This

aspect is emphasized in the next section where responsiblity and cost
centers are discussed. Classification by descriptive element (or

natural element) is exemplified by a breakdown such as salaries, equipment
rental, supplies, and so forth. Possible weaknesses in the typical chart
of accounts for the EDP area and the resulting implications of such

have, amazingly enough, not been discussed in the published literature.
Albrecht has come the closest by presenting an example of a monthly cost
report for a computer operations area. [3, p. 302] However, there are
several weaknesses in his example. Nearly 95% of the costs in his
example are carried in two accounts--sa1ries and equipment rental. Yet
some of the other accounts detail seemingly trivial information-~e.g.,

"Telephone, long distance" - $6.00; "Supper money" - $36.00 (respectively

62

.0001467 and .0008797 of total department costs). If this level of detail
is needed in these accounts, surely more detail could be provided in

place of the two accounts in which 95% of the department's costs are
found!

Part of the purpose of descriptive classifications within depart-
ments is to aid in the control of costs. Otherwise, each department
would be given a single figure of its total costs for a period. Equipment
rental (or depreciation) expense should be split up by major types, e.g.,
mainframes, disc and tape units, printers and card readers, communications
equipment, and special purpose equipment such as plotters. A breakdown
by these major types would seem much more useful for cost control purposes
than knowing that $6.00 was spent on long distance calls. For example,
several authors have commented that peripheral devices and equipment are
typically at least 20% lower from independent suppliers than from mainframe
manufacturers. [137, p. 77; 100, pp. 16-17] A single equipment rental
account is hardly the place to turn for the costs relevant to this
decision. Also, if an alternative supplier is chosen, the cost savings
resulting from that decision would be much easier to track if equipment
costs are split by major types.

The same concept holds true for personnel costs. They might be
broken down further by type of activity or by shifts. Personnel costs
play a very important part in Shillinglaw's third classification--by
product, project, program or service. Although they may account for
30—50% of computer operations costs, they frequently run to 901 of
other EDP cost centers. [149, p. 78; 3, p. 302] Thus a breakdown by
projects is extremely important, especially in the systems design and

systems software development areas. Most authors discussing control of

63

the EDP area touch on project management as being a very important control
tool. Several books have considered the subject. Sollenberger [142, pp.
120, 124, 131] presents an excellent set of flowcharts showing the inter—
relationships between the system service request sequence, the project
reporting system,:uxithe charging system for system development and
operations. Shaw and Atkins' entire book is devoted to system project
management. [132]

It is not our purpose here to detail a project reporting and
management system. That would vary widely from firm to firm and would
be a full-scale study in itself. Rather, we wish to construct the overall
model cost accounting system so that it is compatible with and emphasizes
a project reporting system. This means that personnel time and costs
in particular, must be initially recorded at a detailed, project-by-project
level.

A project reporting system affects the cost accounting system in
at least two other ways. (1) Not all project costs are incurred within
the EDP area. Therefore, the chart of accounts for users must provide some
way of recording EDP-related costs incurred in those areas, and if
necessary, a way of reporting them by projects. This topic will be
further discussed under component six of the model. (2) Some record
must be kept of computer time and resources used by the systems development
area. This should be separable by projects and preferably translated
into dollar amounts so that total project costs can be accumulated. The
requisite charging system to accomplish this will be discussed under
component seven.

It is desirable that costs be classified in other ways than the

three mentioned above. Sollenberger has pointed out that most cost

64

accounting texts contain at least the following classifications: [140,
p. 7; see also 81 and 133]

1. Variable and fixed costs.

2. Incremental and full costs.

3. Planned and actual costs.

4. Direct, assignable, and pro-rated costs.

5. Controllable and noncontrollable costs.

The first two classifications are closely related. The fixed/var—
iable dichotomy assumes some type of activity or volume base. This is not
easy to define even in many manufacturing operations where certain costs
are expected to vary directly with units produced. However, in an age
of time-sharing and multiprogramming it is extremely difficult, if not
impossible, to pinpoint a single adequate activity base for the EDP area
or even computer operations. CPU time might be proposed, but a job
that is heavily I/O oriented would use little of that resource while being
expensive in terms of cards (or other input) or paper (output). Sollenberger
states that the initial cost classification must commonly be made "without
reference to an activity base." [140, p. 8] Turney also concluded, after
examining several firms, that it was impossible "to determine the activity
bases for the variable costs." [149, p. 192] (This is an interesting
comment, coming from an author who bases his transfer pricing system on
variable costs to approximate marginal costs!)

Therefore, although it is desirable to identify variable and fixed
costs in the chart of accounts, it is seldom done for practical reasons.
Most EDP costs are either fixed or step-fixed in nature in the short run.
Examples cited in the literature as variable costs include data preparation,

supplies, operator overtime, and extra shift rental. [149, p. 117]

<::‘
re
ar

31

CC

'3)

65

Li considers "supplies" such as cards, paper, tape, and power to be
partly "committed costs" and partly variable costs. [97, p. 179] He
gives no practical way of distinguishing between what is "variable"
or "committed." In the only empirical study attempting to separate
variable costs, these were found to be less than 5% of total EDP
costs. [149, p. 110]

Multiple regression te chniques have been successfully used to
estimate the cost and time necessary to develop computer programs. [55]
This technique could conceivably be used to determine which factors are
the most important determinants of total EDP costs, thereby revealing
potential bases for use in analyzing variable costs. However, it is
doubtful--with fixed costs running approximately 95% of the total--that
a sufficient range of observations could be found to effectively use
multiple regression analysis. One of the requirements of this technique
is aptly stated by Benston:

The observations on cost and output should cover as wide a

range as possible. If there is very little variations from

period to period in cost and output, the functional relationship
between the two cannot be estimated effectively by regression

analysis. [12, p. 663]

Closely related to the idea of variable costs--and perhaps more
relevant to the EDP area--is the concept of incremental costs. These
are generally defined as "the difference in total cost between two
alternatives" [81, p. 947] and may be thought of as "the net added
costs if a change is made." [140, p. 9] Ideally, the chart of accounts
should be so constructed that the incremental costs of running a new
job or starting a new project may be estimated from past data.

However, difficulties still abound. If excess computer capacity

exists, the incremental cost of a new job X may be only the data prep

.FL

q:

CEH

66

costs and supplies related to it. However, it may bring the computer
system to "capacity" such that, if new job Y is to be run, new disc
storage units must be added to the system and a third shift of operators
must be hired. Are these fixed costs an incremental cost of job X or
job Y? Should they be split between the two jobs? Between all jobs on
the system? Because of difficulties such as these, we may conclude that,
conceptually, variable/incremental costs should be distinguished in the
accounts, but that because most EDP costs come in lumps of fixed or
step-fixed costs, it is impractical to do so to any great degree. At
best, the examples cited on page 64 could be noted as variable for
control purposes.

Three other classifications were mentioned on page 64. The
planned vs. actual distinction will be discussed in part three of this
model under budgeting. The last two groups on page 64 are closely
related. Although they have implications for the chart of accounts, it
is perhaps more logical to discuss them in the context of responsibility
and cost centers--the next section of the model.

The factors used to evaluate case study firms in the area of
cost collection and aggregation are:

1. The degree of classification and detail by descriptive
element-particularly in the areas of personnel and
equipment costs.

2. The degree of classification and detail by specific
program or project--the existence of a project reporting
system and detailed charges by individual programs.

Responsibility_and Cost Centers
A responsibility accounting system recognizes various decision

centers throughout an organization and traces costs (and revenues) to the

 

P!

St

I ‘

’1
..

J

01

SE

67

individual managers who are primarily responsible for the costs in
question. [81, p. 158] The system is based to a large degree upon the
formal organization chart or structure of the company. Within the EDP
organization, various areas such as systems development and computer
operations should be set up as responsibility centers with the area
manager accountable for all controllable costs. These costs would be
aggregated at the information systems manager level and added to costs
he alone controls. Multiple cost centers can be created within a
responsibility center. Cost centers for an area such as computer
operations might include key-punching, data communications, etc.

Two questions may be discussed at this point. Why should the
"responsiblity center" or "service center" form of organization and
reporting be chosen over that of a "profit center?" What are "controlle
able" and "noncontrollable" costs in an EDP environment?

In contrast to a responsibility center--for which costs alone are
usually accumulated--a profit center is "a segment of a business that
is responsible for both revenue and expenses." [81, p. 950] (An
investment center is held responsible for invested capital used in
operations and is often evaluated on a "return on investment" measure.
The tern "profit center" is often indiscriminately used to describe both
profit egg investment centers.) The question of which type of organization
structure is best (or preferable) for the EDP area is inevitably linked
with the transfer pricing question. In general, there are at least three
organizational possibilities for the EDP area:

(1) A strict responsibility center with no transfer pricing for
services rendered.

(2) A profit center with transfer prices including a profit margin.

:sei

«,4-
vii

a;;r
or b
idle

the

“W

are

VEry

68

(3) A modified responsibility center where transfer prices are
used (in general, cost-based), but where performance evaluation is based
on the control of costs, the quantity and quality of services rendered,
and a multitude of other factors including (but without primary emphasis
on) the difference between "revenues" and costs. This latter number may
well be negative in many cases; i.e., billings should not necessarily
cover all costs.

Li [97, pp. 182-183] suggests that alternative one is particularly
appropriate (1) where computer services are used mainly by top management
or by one operating department or (2) where the system has significant
idle capacity. Krasney [94] suggests the second alternative but modifies
the term to "pseudo-profit center." Sobczak [139, pp. 61-62] is not
directly addressing Krasney's article, but he clearly chows that an
attempt to maximize a computer center's revenue by charging prices that
are not cost reflective is detrimental to the firm as a whole. Other
authors have implied similar ideas, and the computer center/profit center
concept is generally rejected except where significant amounts of services
are sold to users outside the firm.

Since we are not considering the case of a dedicated facility,
significant idle capacity, or the sale of services to external users, the
third alternative best fits the assumptions made at the beginning of this
chapter. Further discussion of the transfer pricing issue is postponed

until later in this chapter.

antrollable and Noncontrollable Costs

We turn now to a discussion of "controllable" and "noncontrollable"

costs. Any complete discussion of this topic must by its very nature be

very "situation specific." Horngren defines a controllable cost as "a

5.4..

C.)

69

cost which may be directly regulated at a given level of managerial
authority, either in the short run or in the long run." [81, p. 944]

Thus it is difficult, if not impossible, to generalize which EDP costs

are "controllable" or not. We must be specific as to (1) firm, (2) level
of managerial authority, and (3) short run or long run time frame. Further
complicating the decision is the fact that there is not a clear-cut
dichotomy between controllable and noncontrollable costs. Rather there

is a spectrum of varying degree of "controllability." This model will not
attempt a breakdown of EDP costs between "controllable" and "noncontrol-
lable" elements when such a decision is so situation specific. Rather

we will look at three further cost classification schemes which may help
us implement the controllable/noncontrollable classification of costs in

a particular situation.

The first of these three classifications is that of direct,
assignable, and pro-rated (mentioned earlier on page 64). Direct costs
are costs which are specifically traceable to individual jobs, products,
or services, to a manger, or to a distinct organizational unit. Indirect
costs are costs which are common to two or more cost objects. Often
incurred in lumps, they have varying degrees of traceability to specific
cost objects. Someqimes indirect costs are further classified as assign-
able or pro-rated. Assignable costs are indirect costs for which there
is "a reasonably clear, precise, and easily determined method of allocating
the cost." [140, p. 13] Allocation may be based on the benefits received
or on the amount of resources used. Prorated costs are indirect costs
which can be allocated only on an arbitary basis which has little logical
or theoretical support. Having limited usefulness for decision making,

such costs are probably included on many cost reports because of the

3::

.N.
I .

In

7O

account's preoccupation with full cost allocation.

Keeping in mind that the directlassignable/prorated classification
is a continuum, not a series of neat boxes, we may make some generalizations
about its relationship to the controllable/noncontrollable classification.
In general, direct costs are usually controllable cost, at least in the
long run. Wages and salaries of operators are direct costs to computer
operations. So are equipment rental and depreciation costs, but the
latter are probably not immediately controllable in the short run.
Assignable costs have controllable elements in them, particularly when
allocated on a resource usage basis. Data entry costs may be an assignable
cost to an EDP services user. It is doubtful that prorated costs would
be considered controllable except in unusual circumstances. If reported
on a manager's responsibility report, they should be shown for information
purposes and not for control or evaluation of his performance. [140, p. 14]

Another classification of costs that may be helpful in looking
at the controllability question, particularly in the EDP environment, is
that of managed, committed, and variable costs.

A committed cost represents the aftereffect of a decision or

commitment made in the past. For example, once the computer

planning committee has signed a three-year lease for a computer
installation, the rental becomes a committed cost during the
three-year period. A managed cost is a cost incurred at the
discretion of management at the policy-making level; it differs
from a committed cost in that a specific decision authorizing

the managed cost is needed for each budget period. A proposal

to increase personnel in the computer operations group or to

increase their salaries are examples of managed costs. . . .

A variable cost is one that tends to vary proportionately with

activity. The more hours the computer installation is in

operation, for example, the more punched cards will be consumed.

[97, p. 178]

In effect, what this classification amounts to is a fixed/variable

distinction with fixed costs separated into managed or committed costs.

.\

71

However, since most EDP costs are fixed or step-fixed, this further
subdivision of fixed costs makes sense. The problem lies in making a
useful distinction between committed and managed. Yes, the signing

of an equipment lease for some time period does commit the lessee to
some costs. In addition to the rental payments there exist at least
implied commitments to operator and supervisory salaries as well as
systems design and programming salaries. Does this mean that all costs
associated with the systems design area are committed costs? Probably
not. At least some costs would be discretionary or managed. But
determing the cutoff point would be difficult if not impossible.

Even though it would be difficult to implement, this cost
classification scheme does point out the fact that a manager may not have
control over (and therefore should not be held responsible for) all
direct costs to his area. Hardware costs are direct to the computer
Operations manager, but it is doubtful that he is the one who committed
the firm to that particular configuration.

A final classification scheme is really but a refinement of the
previously discussed one. Wood and Ditri cite four categories of EDP
costs:

1. Direct variable costs are those which vary directly with

production volumes.

2. Variable overhead costs are the production costs which vary
with changes in work volumes but which are not specifically
related to a single product or output.

3. Programmed costs are planned costs incureed as a result of
management decision.

4. Standby costs are . . . the fixed costs of doing business.
[162, p. 76]

Programmed costs are the "managed" costs of the previous classification;

Standby costs are the "committed" costs. Variable costs have been

Classified into direct and overhead. Examples of direct variable costs

 

.....
It;

‘0.

SO

t:

72

would be salaries of keypunch operators who prepare input for EDP jobs

and costs of printout forms. An example of a variable overhead cost

would be the addition of another sift in the computer operations area.
Wood and Ditri make an unnecessarily fine distinction between the salaries
of supervisors on another shift, which they say represents variable
overhead costs, and the salaries of computer operations personnel, which
they say are direct variable costs. [162, p. 76] It is doubtful, that

in an age of multiprocessing/multiprogramming, the salaries of computer
operators can be directly linked to volume

In sumarizing the controllable/noncontrollable distinction the
following points may be made:

1. All costs are controllable at some level of management over
some time period.

2. An objective of a cost accounting system is to identify costs
as controllable/noncontrollable at the lowest possible level of management
where the costs can be directly influenced by the manager's own actions
and decisions.

3. The controllable/nontrollable classification, while necessary
to a good responsibility accounting system, is not a clear split; rather
it is a continuum. The three classifications discussed may help in solving
the problem but do not provide all the answers. In general, variable and
managed costs will be subject to more control in the short run that
committed costs.

4. The actual classification of a particular cost as controllable/
noncontrollable is very situation specific. Hence, a general model cannot
attempt to provide even a suggested list of controllable and noncontrollable
costs. It would vary considerably by asking "by whom?" and "over what time

frame?"

 

 

Std

COT:

$111

to

P0;

73

The factors used to evaluate case study firms in this area are:

1. Agreement of responsibility centers with formal organization
chart.

2. Homogeneous activities within cost centers.

3. Responsibility centers used-~not profit or investment
centers.

4. Attempt made to separate controllable from noncontrollable
costs.

Budgeting and Reporting System

 

Budgeting is standard operating procedure with most firms and
thus most EDP centers. However, budgeting has been far from EDP's tour
d3 force. A perusal of the quotes cited in the first chapter of this
study indicates the extent of this problem. Cost overruns have become so
common that many managers have come to expect them from the EDP function.

Accountants have generally considered the following list to be
the underlying behavioral assumptions in the budgeting process:

1. The budget should be set at a reasonably attainable level.

2. Managers should participate in the development of budgets

for their own functions in the organization.

3. Managers should operate on the principle of management by

exception.

4. Personnel should be charged or credited only for items

within their control.

5. Dimensions of performance that cannot be conviently

measured in monetary terms are outside the budgetary

domain. [125, p. 675]
The extent to which these assumptions are valid has been widely discussed
in the accounting literature. It is not our purpose here to survey and
summarize this literature in detail. Each assumption and its relationship
to budgeting in the EDP area could well be a dissertation in in itself.
Rather, we wish to make a few statements that may help explain the lack

of success that the EDP area has had with the budgeting process, thereby

pointing the way for improvement in the future.

, . ... .
3: .P9 .. . a» b s :0 J... AU '6...

ft
1'6

74

Past studies have shown that failure in reaching budget goals
generally lowers the level of aspiration. [125, p. 676] This may well
be the case in the EDP area; but if so, it is at least partly the fault
of those who have attempted to inaugurate and expand the use of EDP

facilities in a firm. In order to "get a foot in the door, they have
often quoted very low cost estimates, hoping that management will approve
the new addition or expansion. Then, they reason, when management sees
all the tremendous resultant benefits, they will be glad they approved,

in spite of the costs turning out a little higher than estimated. A cycle
of not living within cost estimates is set up and perpetuated by continual
failures. This cycle can be broken by: (1) a hard line approach on the
part of top management (as Larson says, they have put up with the "snow
job" too long [95]) and (2) by realistic time and cost estimates by EDP
personnel.

Findings in the area of participative budgeting are ambiguous at
best. It is quite possible, that while participation may enhance satis-
faction, it does not necessarily increase production. [125, p. 679]

Management by exception is the practice, by a manager, "of
focusing his attention mainly on significant deviations from expected
results." [81, p. 947] However, it is often only the unfavorable
variances which attract a manager's attention.

The response to favorable deviations not requiring corrective

actions often seems to be weaker than that to unfavorable

deviations. As a result, subordinates may be led to view the
system as punitive rather than informative. . . . This suggests
that effort should be made to emphasize positive as well as

negative aspects of performance to provide "positive reinforcement.
[125, p. 680]

These ideas apply in particular to a project management system within the

EDP area. Project managers who are successful in meeting time and cost

 

is

:8]

co:
051
for

3a:

75

estimates should be given "positive reinforcement."

The controllability criterion is one of the most crucial to a
responsibility accounting and budgeting system. As mentioned in the
previous section, it is also very difficult to apply in actual practice
because of the fixed and step-fixed nature of many EDP costs. Difficulty
in application does not nullify its usefulness, and efforts should be made
to distinguish controllable costs on a manager's cost report from those
that are reported for "information only" (if indeed the latter are
reported at all).

Obviously there are many dimensions of performance evaluation in
an EDP environment that are not measurable in monetary terms. This study
is concerned with the single dimension that can be measured in monetary
terms.

The degree to which a budgeting system depends on the two previous
components cannot be overemphasized. The initial budgeting process is
often based largely upon past history and past costs (adjusted, of course,
for expected future changes). If costs have not been classified in the
ways suggested under part one of this model, the budgeting process is made
much more difficult. This is particularly true of a project management
and reporting system. If there has not been experience with similar type
projects and cost accumulation at the project level, then estimates of
future project costs are likely to be in error.

While estimates are the primary tool during the early phases

of the project, in the final analysis the effectiveness of cost

control is limited by the quality, timeliness, and completeness

of the cost accounting system used. Without accurate, dependable

cost reports, estimates and standards cannot be evaluated and

improved, nor can management be warned when significant cost

variances occur. Good project cost accounting will show up
weak spots, stimulate cost reductions through improved methods,

76

and provide objective evaluation of efficiency. To assure

this occurs, all pertinent expenditures must be captured,

properly identified, and posted to the correct budget centers

and account categories. [23, p. 395]

It may also be pointed out that while departmental and firm-wide
budgeting is usually an annual process, budgeting (estimating time and
costs) for systems projects is a continual process occurring year around.
The annual budgeting process seeks to determine what costs should be in
the systems (and other) area(s), considering those projects that lie
before it. Project budgeting is a more specific effort to relate those
costs to output. Project budgeting and reporting can generally be a
much more effective cost control tool (in those areas where it can be
used such as systems design and programming) than the typical monthly
departmental budget report. The latter may be completely in line with
expectations; i.e., no new personnel have been hired and costs have been
incurred as planned. But a look at project report for the area may show
that many are behind schedule and have greatly exceeded initial cost
estimates. In areas where project management is not generally feasible,
such as computer operations, emphasis should be placed on the monthly
budget report for control purposes.

The factors used to evaluate case study firms in this area are:

1. The existence of an annual budgeting procedure.

2. The use of participative budgeting.

3. Monthly cost/budgeting reports with monthly and year-to—date
cost/budget comparisons.

4. The degree to which managers must "live within" their budgeted
amounts.

5. The generation of data to allow continuous cost/benefit
analysis for ongoing and projected systems.

77

Output or Activity Measurement and Performance Criteria

 

In some EDP areas it is probably easier to hold costs constant
while increasing output or performance, then it is to cut costs without
damaging performance levels. For example, a given number of operators
are required to operate a certain computer system. Attempts to cut costs
by reducing the number of operators may be very detrimental to overall
production. Yet efforts to increase the operators' efficiency and
productivity may meet with relative success. Accurate measures of output
or activity are needed, but these alone are not sufficient. Performance
criteria or goals should be set to evaluate actual output.

Measures are needed which will report on (1) human output,

(2) machine output, (3) machine turnaround and service level performance,
and (4) capacity available, used and unused for all EDP resources. [140,
p. 18] Human output measurement will be especially important in such
areas as systems development, programming, and data preparation. The
latter area's output is perhaps the easiest to measure with some unit

such as keystrokes per some time period (minute/hour/day) which is reported
for all workers. Even a programmer's output may be measured in lines of
code per time period although considerable amounts of influential,
intangible factors must be recognized. Human output in the systems

design area is the most difficult to measure but probably the most important
of the three mentioned. Some measure of output could be obtained from
project reports by examing total estimated hours, actual hours to

date, and the estimated completion percentage for all projects for which

a given person is responsible.

The latter three measure are often the "outputs of highly technical
machine dependent systems which have few uniformly applied efficiency

benchmarks."

78

These generate different results depending on, among others, the
machine, physical set up, local interpretation, and different
levels of detail recorded. Also, since the majority of internal
accounting systems are vendor provided, inherent biases exist.
Frequently, great detail and exactness obscure the need for

basic and fundamental output measures which can be matched against

cost. [140, p. 19]

There is really no other practical source for most of these measures, but
their limitations should not be forgotten.

One possible change is in the way item two, machine output, is
usually reported. Often this item is indistinguishable from item four--
resources used. Rather than measuring output in what is really input
terms, it has been suggested that it be measured in terms of CAU's,
computer activity units. [7, p. 31] These could conceivably be
translated into a cost per invoice processed, a cost per order processed,
etc.; cost measurements that users of EDP services and upper management
may find easier to relate to and understand. These latter measurements
do have limitations for internal (EDP) managerial assessments, particularly
in the area of cost planning and controls.

The factors used to evaluate case study firms in this area are:

l. The existence of system software or other method for

measuring individual resource utilization in the hardware

area.

2. The use of the utilization data to manage and control computer
operations, not just to provide input into a charging scheme.

3. The existence of work measurement and reporting systems
for the data entry area.

4. A reporting system for human output in the systems design
and programming areas.

Long-range Financial Planning

 

Most companies have implicit, if not explicit, long-range goals

and Objectives. Long-range planning for the firm as a whole is made

79

toward these objectives. The corporate EDP area is not an independent
organization attempting to chart its own course. It must serve the
corporate interests and goals. Long-range planning for the EDP area is
therefore highly dependent upon that of the firm as a whole. In
general the time horizon of many firms' planning is at least five years;
therefore planning for the EDP area should extend to at least this

time frame.

Very little is found in the EDP literature regarding long-range
planning. Probably this is because of the inherent difficulties of
attempting such activities in this area. Continuous advances in technology
have resulted in continuing changes in hardware, software, and approaches
to problems. Often these changes may well have greatly limited the accuracy
of any long-range planning done in the past. However it must be remembered
that it is not the accuracy of long-range planning which is most important,
but rather its effect on current actions.

Planning can be justified only by its ability to assist

anagement in its current actions. Long-range planning

permits management to view the probable chain of future

actions that will follow as a logical consequence of the

current decision, the first link in the chain. By this

means, current alternatives can be evaluated in the light

of how well they will fit into and affect the coordinated

structure of future developments that are embodied in the

plans. [133, p. 336]

Long-range planning for the EDP area will include estimates of
personnel, hardware, and facilities and translation of these into dollar
terms. However, even more important than (and hopefully the determinant
of) the above items will be the long-range projections of major new projects
and applications designed to further corporate goals and objectives.

It is desirable that discounted cash flow (DCF) techniques such

as time-adjusted rate of return (ROR) and excess (or net) present value

80

be used to evaluate long run investments in both hardware and projects.

The literature is replete with suggestions on how to measure benefits for
such analyses, particularly intangible benefits. We have observed in
Chapter One that costs are seldom accurately estimated. This would

suggest that a specific decision be subjected to a good deal of sensitivity
analysis before conclusions are drawn.

Even more difficult than the initial do/don't decision is the
attempt to determine if a given ROR is being achieved. It is conceivable
that this might be attempted for the EDP area as a whole. However, it
is doubtful whether the EDP area even meets the criteria for a profit
center, much less an investment center. (See pages 67-68.) The EDP
manager is (usually) not given a free hand in the control of (1) his
asset base, (2) his price/quantity mix to internal users, or (3) sales
made to external users. These facts, coupled with the measurement
problems of determing actual intangible (and even tangible) benefits,
are sufficient to question the validity of ROR analysis for the EDP area.
The same types of considerations limit the the measurement of actual ROR's
on projects attempted. We are not saying that DCF techniques are inap-
priate to the EDP area, but that EDP acquisitions and projects should be
subject to the same scrutiny as other acquistions and projects, including
screening through the firm's capital expenditure evaluation techniques.
lf_the firms uses DCF techniques, they should be applied here as elsewhere
in the firm. However, recognition must be made of the difficulties in
comparing actual ROR with projected ROR for performance evaluation purposes
because of the limitations of our present measurement system.

The factors used to evaluate case study firms in this area are:

1. The existence of reasonably complete long-range goals and
plans at the corporate level.

81

2. The existence of longerange goals and plans for future projects
in the EDP area.

3. The existence of long-range plans for personnel and hardware,
and projected dollar costs.

4. Hardware acquisition decisions made on a basis consistent
with other capital budgeting decisions in the firm.

5. Use of a life cycle, cost/benefit approach to system project
decisions.

Tracking Users' EDP-related Costs

 

This is the sole model component not taken from traditional cost
accounting literature. Although the need for such a procedure has been
implied in the literature, it has not been suggested explicitly. By
"tracking users' EDP-related costs" we mean that the users' accounting
systems should be so designed that the total amounts of the EDP-related
costs are readily available. These costs might include any data preparation
done by the user, computer terminals he owns and maintains, and liaison
personnel between the use and EDP operations. Without the data it is
nearly impossible to arrive at a total cost of data processing activites
for the firm. This type of data is also highly relevant to cost/benefit
analyses of future applications. Another important reason to monitor
these costs is because a centralized computer center might cut back on
the quality of its services to users (in an attempt to reduce its own
costs) and in so doing might send users' EDP-related costs soaring.

Perhaps the single most important reason for this component is
that it facilitates a company-wide budgeting of data processing related
costs. In so doing, the duplication of facilities and computer capacity
will hopefully be avoided. In one large firm over one hundred mini-computers

had been acquired by users without any knowledge at the corporate level

82

that this was taking place. [106, p. 137] In another organization,
utilization of a central computer facility declined because its research
laboratories had acquired several large pieces of industrial testing
equipment that contained their own built-in minicomputers. Simpler
(and cheaper) machines could have been acquired and used in combination
with the company's central EDP facility. The author concludes that "such
questionable and unnecessary duplication of facilities is far from
infrequent." [106, pp. 132-133] A careful tracking of users' EDP-related
costs would have given warning that something was afoot in both cases.

The factors used to evaluate case study firms in this area are:

1. Annual summary of data processing costs for the firm is
prepared.

2. EDP personnel approve or are otherwise aware of hardware
acquisitions by other functional areas in the firm.

3. An effort is made to collect and aggregate EDP-related
costs in user areas and report these on a systematic basis.
Charging System for Users
Because of the differences between computer operations (production
oriented) and systems design (similar to R & D), these areas will be

discussed separately. We begin with that of computer operations.

Charging for Computer Operations

As we saw in the previous chapter, there is general agreement in
the literature that in a decentralized firm, the benefits of a transfer
pricing system for computer operations outweigh the disadvantages or costs.
There is, however, wide disagreement as to the best or most appropriate
transfer pricing method.

Conceptually, most economists agree that marginal cost (or marginal

.1:

......

 

1“

al-
A“

 

\

83

cost plus opportunity costs) is the best transfer price for the case
where an outside parket price is not available. Yet it was noted in the
previous chapter (page 35) that one of the assumptions on which this
analysis is built is that of technological or "cost independence"—-the
level of operations in one division will not affect the cost function of
the other. Although this is obviously invalid for the EDP area, it is
not clear what effect the relaxing of this assumption would (or should)
have on the transfer price.

The preponderance of fixed and step-fixed costs in the EDP area
compounds the problem. The step-fixed costs cause the marginal cost
curve to be discontinuous with resultant problems in pricing at those
spots. The total amount of marginal (approximated by variable) costs
as a percentage of total costs is so small (one estimate is 5% [149, p.
110]) that many managers feel that a marginal cost charge "would swamp
their installation with job requests in a very short time." [71, p. 114]
In the one study where marginal costs were seriously proposed as the
appropriate EDP transfer price, it was also suggested that fixed costs
be transferred to users in lump sum amounts based on past usage. [149,
pp. 169-170, 172] It was noted that this time lag in charging users
seems to be undersirable, particularly if the charges were to be used
at all in the performance evaluation of either the user of the EDP center.

As mentioned in the previous chapter, actual full cost is
riddled with many planning and control problems. In fact, actual cost
schemes of any type (marginal or full costing) should be ruled out
because they transfer any inefficiencies in the EDP area to users. While
standard full costs do not transfer inefficiencies in one department to

another, if based on expected usage (especially if that is low compared

84

to machine capacity), they have many of the faults of actual full costs.

We noted earlier that many jobs, once on the system are difficult
to remove. [See 71, p. 80] What is needed is a measure of lBEEZEEB
incremental costs to the firm for each application. To determine this
would be impossible. How does one decide which job or application should
bear the costs of new disc and core storage or possibly of a new mainframe?
Which specific application should bear the incremental costs of the
second and then the third operator shifts? Which single application is
so fundamental to the firm's continued existence that it should bear the
costs of computer security measures? We could go on and on. Obviously,
some allocation is necessary. The costs incurred in lumps in the data
center cannot be neatly charged out in such lumps (unless it is a dedicated
facility--in which case transfer pricing is not an issue). Yet, these
lumps cannot be ignored. It is truly the ongoing processing of computer
jobs that causes these costs to exist, not in an hour—to—hour or even
day-to-day sense, but in a cumulative, long-run sense.

The transfer pricing method that comes the closest to measuring
.all the long-run costs of a job is standard full cost based on a "practical
capacity"* level. Thus early users of a given system are not penalized
by initially high rates. Assuming economies of scale, as additional
computing capacity is added the rate decreases, reflecting lower long-run
incremental costs. Note that this measurement base will rarely "recover"
100% of computer operations costs through the charging algorithm. In

particular, when a system is new, only a small portion of capacity will

 

*Horngren defines practical capacity as "the maximum level at
which the plant or department can realistically operate most efficiently,
that is ideal capacity less allowances for unavoidable operating inter-
ruptions. Also called practical attainable capacity." [81, pp. 949-950]

85

often be utilized and therefore only a small portion of total cost
"recovered." It is conceivable that Operations right at the level of
practical capacity would recover 100% of related costs (if the EDP area
incurred costs in line with the standards). However, if this were the
case, the company is probably not far away from needing a larger hardware
configuration. After it is installed, rates will probably drop because
of the additional capacity (assuming economies of scale), utilization
will be less than practical capacity, and again the charging system will
not "recover" all costs. Therefore, what is expected is a sizeable
"volume variance" unless operations are near or at practical capacity.
This is really a cost of providing additional capacity for growth which
is not being utilized (yet) and therefore should not be charged to users.
The suggestion of using standard costs based on a practical capacity
level is not new. It has been previously recommended by Heitger [71, p. 100]
who probably adapted it from Wormley [163]. However, it does not solve
all questions of implementation. What measure or measures should be
used to determine EDP resource usage? Should users be charged for resources
used on a detailed resource by resource basis? If so, how should costs
be accumulated for and overhead allocated to these resources?
Two approaches to these questions are proposed, based on the
experience with and knowledge of EDP that the average user of data
processing services has. If he is relatively savvy in such affairs, we

suggest a detailed billing Of actual computer resources used.

Billing for Resources Used
This suggestion is not new. In fact, there even seems to be a
concensus in recent literature that this is the way to go. As to the

appropriate measures used determine EDP resource usage, these would be

86

highly dependent upon the specific configuration used. A typical example
found in the literature includes the following breakdown: CPU, memory,

disc drives, tape drives, printer and readers. [17, p. 161] It is probably
not desirable to have the printer(s) and reader(s) lumped into a single
average unit charge. Some jobs may have high input and low output levels
and vice versa.

There are at least two general approaches to determining unit
costs or coefficients used in the charging algorithm. One involves
tracing as many costs as possible directly to the resource "centers" and
determining unit costs based on practical capacity. Then the overhead
and manpower costs that cannot be directly traced in the first step are
stated as a percentage Of traceable costs. All unit costs are then multiplied
by this percentage tO arrive at the coefficients in the charging algorithm.
[157, pp. 61-62]

Another approach is to attempt to directly allocate overhead to the
hardware areas-~CPU, memory, disc drives, etc. For example, building
expenses could conceivably be allocated by the proportionate amount of
square feet used by each hardware component. Electricity and possibly
air conditioning could be allocated on the basis of kilowatts used. It
is possible that even manpower expenses could be allocated on the relative
proportion Of time devoted to each type of hardware. [17, pp. 161-163]
After all costs are allocated, then the cost coefficients of the charging
algorithm are determined based on practical capacity.

Although both methods will charge out approximately the same
amount of costs to users, the latter approach is preferable if the
allocations are not arbitrary. For example, if some of the manpower

expenses can be directly associated with the printer (including bursting

87

and output distribution), these should be a direct determinant of that
cost coefficient, not part of an overhead percentage.

Although the charging algorithm should, in general, be stable
over time, it should be adjusted as soon as feasible for cost reductions
of hardware components and other major cost changes. [71, p. 112] For
example, if a firm decides to lease its disc drives from an independent
supplier, thereby dramatically reducing the cost Of disc storage relative
to tape, this change should be reflected in the charging algorithm. A manager
may then weigh the cost savings of changing to disc storage from tape
against the one-time cost of the systems and programming effort necessary
to effect the change (assuming this cost estimate is provided by the
systems staff).

Although concern has been expressed in the literature about the
loss in productive capacity because of the detailed measurement of resource
utilization and the related charging algorithm, this concern seems to have
greatly diminished in recent years. It is a well known fact that most
business computers are often I/O bound (i.e., input and output are the
constraining facts--they cannot "keep up" with the central processor(s)).
Although there were initial efforts to use hardware monitors or to "sample"
the resources used by a given application over several runs, these measures
have generally been scrapped in favor of detailed measurement by means of

systems software of computer resources consumed on each run of an application.

Billing in Terms of Computer Activity Units

Billing users for detailed resources consumed is preferable to
the following method even if users are not well versed in EDP terms and
technology. Possibly an in-house educational program could be utilized

to improve their knowledge of this area and the workings of the charging

is a

SURE

2885

Proc

ESE!

Crar

‘U
”U
,4.

/

are

jot

88

system in particular. However, if this is not feasible, and users are
highly resistant to a detailed algorithm which they cannot understand,
a simpler billing method suggested by Anderson can be used. [7, p. 31]

Under this method a single "pseudo-measure" of computer activity
is multiplied by an overall cost rate for the particular system on which
the job is run. Thus the charge is determined: charge = c - CAU's, where
"c" is the overall cost rate and "CAU" represents Computer Activity Unit--
a composite measure of computer activity, based in part on memory requested,
CPU time, and records transferred on different devices. [7, p. 31] This
is a much simpler measure for a user to understand, although it sacrifices
some accuracy and equitability.

An extension of this method is to bill users in terms of their work
measurement units. Thus a rate could be developed in terms of invoices
processed, customers billed, file inquiries or updates made, etc. Thus the
user would know that his charges would vary directly with the number of
transactions he has processed. Budgeting for EDP costs in his budget
would be greatly simplified. However, it is doubtful whether this method

could be truely cost reflective over a wide range of volumes and varieties

of measurements.

Priority Pricing

 

Where prime shift time is limited and demand exceeds supply, there
are opportunity costs of running any job-the slower turnaround on other
jobs in the system or in a queue. Even if it is not feasible to directly
measure these costs, it may be desirable to charge varying rates for
different classes of service. Thus a user who desires a fast response
time and is willing to pay a premium for it may do so. A user satisfied

with overnight turnaround may be given a lower rate.

axon

89

A simple way to use priority pricing is to determine charges in
accord with one of the usual methods, and then multiply that charge by
a given percentage depending on the class of service. A high priority
job may be billed at 150% or "normal" rates while an "over-the-weekend job"
may be charged only 50% of such rates. These percentages may then be
adjusted over time to equalize the number of applications run at various

times and to give improved service to all users.

A Word of Caution

 

It may have been implied that a chargeout scheme is the only way
to go, but this is not necessarily the case. Some general observations
about chargeout schemes will be made which will hopefully put them in more
of a perspective.

First, our restrictions and assumptions made at the first of this
chapter are quite limiting. Our hypothetical firm is not intended to
reflect the case of every firm. Care must be taken in applying the
conclusions reached to a particular firm or subset of such.

Second, as Gibson and Nolan have pointed out, it is usually at the
third stage of EDP growth that the chargeout scheme is introduced. "This
stage frequently includes the first formalization of management reporting
systems for computer operations, a new chargeout system, and the
establishment of elaborate and cumbrous quality control measures." [58,
p. 83] As noted earlier, the first stage of EDP growth consist mostly of
cost-reduction, accounting-type applications such as payroll, accounts
receivable, accounts payable, and billings. [58, p. 77] There is really
no need for a charging scheme at this point. Since most of the applications
are accounting oriented, the EDP department may indeed be located within

this area.

 

2:

H1
Eu.

"H
ht.

,

90

In stage two there is a proliferation of applications in all
functional areas. It is probably the tremendous increase of costs at
this stage, coupled with an ever-enlarging number of users that brings
on the stringent controls of stage three including the charging scheme.
It may be suggested that some, if not most, of the control measures of
stage three should be applied at the beginning of stage two, as the
computer center begins to serve all functional areas. The knowledge that
they would be charged on a continuing basis for their applications, might
cause more managers to carefully weigh the long-run costs and benefits of
possible applications. Hopefully the growth in stage two would be more
orderly and costs would not soar at quite the same rate. The emphasis
on control would be more even through time, instead of the sudden stress
on that aspect in stage three.

Although the stage hypothesis of Gibson and Nolan probably applies
to many firms, not all will fit neatly within its confines. In an
earlier article, Dearden and Nolan suggested a set of questions that
should be asked before deciding to use a chargeout scheme (see page 42).
[40, p. 78] The company's operating philosophy with respect to centrali-
zation/decentralization is particularly important. There is no point in
charging a manager in a highly centralized firm for costs that he has
little or no control over.

Finally, the limitations of the charging scheme for short-run
analysis should be readily apparent. It does not attempt to measure the
short run incremental or variable cost of a particular job. If there is
a particular "one-time" need for a job by a user, this job might be

charged only the short-run incremental costs related to it-—systems and

V

»-I‘

a:

C

F. a
3U

‘ .
a .v

361":
CEC
the
SL1:
Ea

91

programming costs, data preparation, supplies, additional operator costs
(if any), and additional lease costs (if there are such past a given

number of hours).

Charging for Systems Services

 

It makes little sense to attempt to control one large group of
EDP costs (computer Operations) by charging users while they may Obtain
"free" systems and programming services. Although charging for systems
services is not discussed nearly as much in the literature as charging
for computer Operations, it has generally found support wherever mentioned.
[See 7; 10; 14; 22; 51; 94; and 154]

The term "system services" or "systems development" generally
encompasses three types of activity: (1) new systems development (including
the programming effort), (2) maintenance programming, and (3) system
maintenance. [71, pp. 124-125] New systems development work is generally
traceable to one or more requesting departments and should be charged to
such although there are exceptions to this rule as we shall see. Some
maintenance programming is done at the user's request and should be charged
to him. Other maintenance programming is required because of changes in
the physical operating system, particularly when converting to a completely
new hardware system. It is doubtful whether these costs can be considered
controllable by the user and whether he should be charged for such.
Finally, the costs of systems maintenance on the operating system should
probably be considered part of computer operation's overhead and recovered

through that charging algorithm.

92

Resource Usage Basis

 

There are two general approaches to charging for system develop-
ment costs. One entails charging users for time spent on their particular
project. Under this method different cost rates would be set for different
classes of personnel; e.g., junior analysts or programmers would be "charged
out" at lower rates than senior analysts or programmers, etc. Anderson
suggests that this could be easily done by setting individual billing
rates as a percentage of gross pay. [7, p. 30] He contends that hourly
rates should be set at a level which will permit recovery of indirect
personnel costs such as vacation pay, training and development, sick
leave, and supervision.

There are problems with this method however. First it is in the
systems development area that cost estimates are Often exceeded. Users
could be given a cost estimate and, if charged for actual time spent on
that project, could be billed for costs well in excess of the intial
estimate. Second, it is doubtful that many indirect costs should be
included in the rate. There is a great difference in chargin for systems

/
services and computer operations. Systems development projects are "one-
time" efforts; they are not "locked into" the computer system for a long
period Of time as are ongoing applications. Therefore all we need

necessarily be concerned with is short-run incremental costs.

Lump Sum Negotiated Amount

The other approach to charging for systems services is to use a
lump sum negotiated amount (or possibly even the lower of the negotiated
amount and the actual costs incurred). The following comment is typical

of several.

93

The charge for development costs will be contractual based on

negotiations between the user and M18. Estimates of the

development costs will be prepared by M18 and presented to

the user for approval. Once the contractual price has been

decided upon MIS must perform the systems and programming at

this price; if there are favorable or unfavorable variances

from this price they must be borne by M18. [149, pp. 166-167]
Although this approach is heavily dependent upon the ability of the systems
area personnel to accurately and impartially provide cost estimates to all
users, it iSprobabLy perferable to charging on the basis of actual
manhours spent on a project. Users will be more likely to undertake new
system projects if they are assured of a maximum contractual price that
they will be charged. Care must be taken so that the cost estimates
provided to users will cover only the expected incremental costs of the
proposed project. In other words, the systems area (as in computer opera-
tions) would not be expected to "recover" 100% of its costs by charging users.

Several comments which would apply to both approahces to charging
in the systems development area are in order. First, "preliminary and
economic feasibility studie are probably 'in the company interest'
expenditures and not specifically chargeable. . . . The charges for
systems development work should probably be based on work done once the
project is formally approved." [142, p. 132] Second, there is the problem
of projects (and resultant ongoing computer applications) undertaken for
the benfit of two or more sponsoring departments. How should costs be
allocated to these departments? The first step should be to trace as
many costs as possible directly to the departments involved. One sugges-
tion as to how to handle the remainder is to allocate it based on "the
degree of involvement of each user. Thus a user who is expected to

utilize sixty percent of the transaction volume of the project should

carry sixty percent of the joint development costs." [149, p. 167]

94

If all else fails it is conceivable that the joint costs could be
allocated to users on a benefits received basis. However, what constitutes
benefits and how to measure such has long been discussed in the EDP liter—
ature. This approach is likely to result in a quite arbitrary allocation,
but it may be the best possible under the circumstances.

Third, the further a firm gets involved in data-base oriented
systems, the more certain projects will cut across many user areas.
Sponsorship by a few areas would be unlikely because the benefits accrue
to so many users. Thus the systems development area (and later the computer
operations) should have allocations in their own budgets for projects that
in the general corporate interests and for which the allocation Of costs
to users is not feasible or is extremely arbitrary.

Finally, regardless of the charging method for systems design
there are likely to be problems over who should bear the costs of changes
or improvements in the project after it has been approved. If the changes
are requested by the user, the additional costs caused by the changes
should be charged to him. If the changes are initiated by the systems
area, it should probably bear the costs of such.

A similar problem is encountered in the computer operations area
regarding the costs of rerunning jobs. Ideally, costs of reruns caused
by users should be charged to them, while the costs of all other reruns
should be absorbed by computer operations. [7, p. 81] However, in both
of the above situations it is difficult to be equitable without being
very costly. The costs of personnel time consumed in presenting such
situations to some type of review committed or impartial individual
probably exceed the costs of the changes or reruns in most situations.

Anderson's suggestion that a general rule of thum be established, and that

95

serious exception be handled individually is probably wise. Thus the
systems area should bear the costs of all changes in systems projects unless
they are exceptional in amount and caused directly by user request. Like-
wise, rerun costs should be absorbed by the computer operations area.

The factors used to evaluate case study firms in the area of a
charging system for users are:

1. A charge to users is made for EDP resources used in the
computer Operations area.

a. The charging system is based on standard full costs
and practical capacity--thus no attempt is made to
charge out all of computer operations costs.

b. Charges are based on a detailed measurement of EDP
resources consumed.

c. Charging rates are not changed more often than once
a year except in unusual circumstances.

d. Consideration is given to priority pricing schemes.

2. A charge to users is made for systems design and programming
services.

a. A lump-sum amount is negotiated for larger jobs.
b. No attempt made to charge out all systems design
costs--e.g., no charges are made for preliminary and

feasibility studies.

c. Charges are made for user-initiated maintenance program-
ming work.

d. Satisfactory resolving of problems in charging multiple
sponsors of large-scale, integrated systems.

3. A mechanism exists for resolving user complaints in both
charging areas.

Chapter Summary

 

Although transfer pricing of EDP services has been widely
discussed in the literature, there has been little effort to view it in

the broader context of an EDP cost accounting system. In this chapter,

96

a model cost accounting system for the EDP area was constructed. A set
of assumptions was made about a hypothetical firm, and seven components
of the model were outlined and discussed.

Ideally it would be desirable to implement the model in several
randomly chosen firms, thereby making some of the more general ideas very
situation specific and providing a sample form which to make some
generalizations about the population. As discussed in chapter one, this
approach is not feasible. Rather, in the following chapter the control
system in several firms is examined and particular attention is paid to
the existence or nonexistence (or degree of such) of the seven model
components. Data from much larger surveys will also be utilized to provide

a broader basis for generalizations.

CHAPTER IV

RESEARCH FINDINGS--CURRENT EDP COST CONTROL PRACTICES

Introduction

 

Very little data has been published on the methods and practices
used to control the overall EDP area. The Churchill, gt a1. study [24]
was primarily a survey of computer applications in various functional
areas. Sollenberger's NAA study [142] was control oriented, but it
dealt chiefly with the systems design and development area. Turney [149]
had a great deal to say about faulty cost control practices in the EDP
area, but his study was based on observations in only one firm. A
published four way panel discussion on controlling EDP costs gives but
a little insight into some of the approaches used by larger firms. [15]

Little empirical data is available on corporate transfer pricing
practices in general. [81, p. 740; see also 8] Martin [102] examined
two firms and in general found "inadequate" cost-based transfer prices.
Larson [96] conducted "in-depth interviews" at eight firms and found most
firms advocating market prices but using prices set by top management
action ranging from cost to market-based. Arpan [8] has examined transfer
pricing in multinational firms where custom duties, currency devaluations,
and changing tax laws are important variables.

Knowledge of transfer pricing practices in the area of EDP services
is also sketchy. Heitger [71, p. 84] found all the firms in his study

to be using some form of transfer pricing for computer services. ‘A priori

97

98

standard full cost was the method he found most widely used. [71, p. 92]
Turney stated that "few companies appear to utilize transfer pricing for
their MIS services." [149, p. 68] He found most companies that did have
a transfer pricing system to be using an SE past full cost charge. [149,
pp. 139-140] McFarlan, gt El- cite "several informal surveys" when they
state that approximately one third of the EDP management control systems
"are based on non-charge-out, one third are based on partial-charge-out,
and one third are based on full-charge-out." [107, p. 473]

Thus, little is known about corporate EDP transfer pricing policies
and about the underlying cost accounting system used to track and control
EDP costs. This chapter is based on the EDP cost accounting systems of
the six case study firms. The data is supplemented by the results of two
much larger surveys by the GAO and the CASB. After an initial examination
of the centralization/decentralization issue and other background informa—
tion, the discussion will follow the outline of the seven components of

the model described in the previous chapter.

Centralization vs. Decentralization of EDP Resources

Two of the case study firms were quite decentralized in their
overall organization structure, and their EDP areas paralleled this
pattern. Although there were "corporate data processing centers" or the
equivalents in both firms, there was not the massive centralization of
hardware and processing activities that has been occasionally predicted
in the literature.

In firm C*, someone in the controller's office had begun a policy

of centralizing computer hardware over 14 years ago. At first it was mostly

 

*
Case study firms were promised anonymity and will be identified
the letters A through F.

99

accounting and clerical type Operations with a fair amount of EAM
equipment also being used. Gradually it grew in size and in importance

to the various operating divisions of the firm. Problems arose among the
various users as to who should get first priority in systems projects and
how computer job scheduling should be organized. The more vocal people--
those complaining the loudest--usually got their way, at the deterioration
Of service to other, less vocal, users. Finally the corporation president
grew tired of hearing V.P.'s scrapping among themselves about computer
services, and he had the centralized center split up into several

smaller centers, each under divisional control.

This sudden shift toward decentralization seems to have worked
fairly well. Immediately after the split, the same amount of work was
being handled by the now "decentralized" divisional data centers
with no apparent increase in hardware or personnel. (The centers are
"decentralized" only in a corporate sense. There is still considerable
centralization of data processing activities within the divisions.

Several divisions have two IBM 370-145'3 or larger systems at their
divisional data centers. As some measure of the large size of this

firm, there are roughly two hundred computer systems in use with nearly
one fourth of these being IBM 360's or 370's.) In the seven years

since that split-up occurred the company has stuck to a general policy

of decentralizing computer services and placing them under line (operating)
authority with some staff support and guidance.

There remain mixed feelings in the firm about whether a central-
ization policy could have worked. One manager believes that centralization
would have worked if the centralized corporate D.P. center would have had

a "higher quality" management in dealing with the disputing divisions about

100

priorities. Another manager seemed to think that centralization would
never have worked and that the divisional V.P.'s would have continued to
gripe until their EDP services were directly under their control.

In any case, it seems in retrospect that there was not a
tremendous amount of economies of scale attributable to the centralized
center before the split-up. Although there is no longer a centralized
corporate D.P. center serving all divisional users, certain activities
are centralized at a Finance/Personnel data processing center. A rough
breakdown of jobs is: personnel--20—25%; controller activities--20-25%;
and treasurer/finance activities--50-60%. All hourly and salaried personnel
of the corpoation are paid through this center and personnel records are
kept here.

In firm A the situation is much the same. However, rather than
attempting a course of centralization of resources (as in firm C), this
company has long maintained a philosophy of EDP decentralization. There
are currently 18 data centers which serve various marketing groups and
production facilities. There is a large corporate data center at the
company's headquarters which serves all users' needs in that locale.
However, recently the marketing group was given the OK to establish
their own data center near the company headquarters. The corporate data
center people are somewhat chagrined but can do nothing. It again (as in
firm C) seems to be the result Of an intense desire by a functional area
to control their own data processing activities, especially job scheduling.
The reasoning goes somewhat along these lines: if we are "paying" (through
intra-firm charges) for all these data processing activities, why not pay
directly (but perhaps a little more) and have these under our direct

control.

101

Another similar incident occurred in firm F. A certain division
was allowed by tOp management to install an IBM System 32 for A/R work.
The corporate computer center people "bent over backwards" trying to get
the division's business on this application, even to the point of charging
only 50% of its normal rates. However, because the division wanted line
control of its own system it was allowed to purchase the System 32.
Computer operations personnel felt that this was a terrible waste of
manpower and hardware (not to say dollars and cents) because the corporate
data center could easily handle the job.

Firm D is really a division of an extremely large company. This
division was larger than several other of the case study firms. Its
parent company allows and encourages very decentralized operations (at
least at a divisional level) so that for most purposestimzdivision could
be considered a company in its own right.

In keeping with this corporate philosophy, the division has a
very free hand in planning and operating its data processing activities.
Most Of these activities are centralized in one large divisional D.P.
center. Thus while the parent company has a philosophy of decentralizing
EDP operations (in keeping with its operating philosophy), these may be,
in fact, highly centralized at the divisional level.

The other three firms had a policy of centralizing EDP operations.
All three had a high level corporate data processing center. There was
some usage of smaller hardware configurations (e.g., IBM System 3's) at
branch offices, but often these were used for RJE (remote job entry)
purposes.

In analyzing the centralization/decentralization issue in terms

of computer hardware the author noticed one interesting fact. At major

102

data processing centers throughout all firms there were usually two large
capacity systems, often the same model. None of the companies visited
had 9953 than two large scale systems in the same data center (although
occasionally some Older and smaller hardware models were still around).

A configuration that turned up quite Often was that of two IBM 370-158's.
Also found were two Burrough's B 6700's, two IBM 370—145's and two 370-
135's.

In firm C there had apparently been considerably more than two
large capacity systems in the corporate data center before the turn
toward decentralization of hardware took place. However, several years
after the split-up there were no more than two large systems at any
divisional data center. Even the smaller firms with centralization
policies had no more than two large configurations in their corporate
data centers.

Although no conclusive evidence can be given showing why this
particular phenomenon was encountered so often, several plausible reasons
can be advanced. First, although economies of scale are possible by
advancing to larger and larger hardware configurations, there is really
no large savings to be made by aggregating many of these large configurations
in a single data processing center. This is particularly true when the
desires of various users to control their own D.P. activities are considered.
Second, the use of two large systems in a single D.P. center is much
"safer" than a single extremely large system, although the latter might
be less costly in terms of hardware and personnel. If one system "goes
down" it does not completely interrupt the operations of the data center,
and the more critical applications may continue to be processed. Finally,
operators and other personnel can more easily work with either system if

both are identical.

103

In summary, it was found that the massive centralization of
hardware and personnel predicted in the literature did not exist. True,
the medium to medium large size firms did have highly centralized data
processing facilities. Yet, the extremely large firms did not centralize
EDP operations at the corporate level, but more typically at the divisional
level if at all. It is difficult to say whether this decentralization was
caused by (1) the sheer size and geographic dispersion of these firms and
their division, (2) their stronger emphasis on decentralization in their

operating philosophy, or (3) a combination of these factors.

Level of Cost Collection and Aggregation

In this section two of the ways in which firms classified EDP
costs are examined--(l) by descriptive element and (2) by project. A
third way in which costs may be classified--by organizational unit--is the
subject of the following section.

The chart of accounts for the EDP area varied little between
firms. The various sets of accounts covering the computer operations
area were quite similar to the example given by Albrecht. [3, p. 302]
A typical list from case study firm F is presented below:

Salaries

Salaries--Premium Pay
Salaries--Overtime

Fringe Benefits

Supplies

Office Service Expense
Supplies--Postage

Repairs and Improvements
Repairs--Outside Maintenance Contract
Professional Services-~Software Expense
Training Expense

Equipment Rental

Equipment Rental-~Property Taxes
Travel

Telephone, Telegraph & Teletype
Telephone--WATS Expense
Depreciation--Equipment

104

Materials Loss

License Expense

Miscellaneous

Intra-divisional Work

Corporate Information Systems

Computer Systems Research
Total Direct Expenses

Net Prorates

Work Order Credit

Net Adjustments to Prior Years Expense
Total Operating Expenses

Usually, the chart of accounts for the systems design area varied slightly
in that there was no need to detail equipment rental and depreciation
expenses. There was, however, generally provision made for charging

the systems area with its share of computer operations charges. (The
reverse was also typically true. Notice the "Corporate Information
Systems" account in the previous example.) The following is an example
set of a chart of accounts for the systems area in firm A:

Salaries

Other Salaries and Wages--Indirect
Departmental Supplies

Printing Material

Telephone--Loca1

Telegraph & Long Distance Telephone
Purchased Supplies

Entertainment

Travel Expense--Indirect

Purchased Services

Company Paid Tuition

Recruiting Expense

Association & Membership Dues

Other Technical and Education Expense
Periodical Subscriptions--Technical Publications
Other Expense

Other Redistributed Expense

Data Processing Expense

Graphic Arts and Reproducing

Several comments may be made about the accounts used to collect
and classify EDP costs. (1) Often the classifications had evolved
through time, and there was little apparent effort made to structure the

Classifications to the needs of EDP managers. In several firms the

105

observation was made that this was essentially the same set of accounts
used throughout the company in other divisions and departments. Surely
the EDP area differs enough from other activities of a firm to deserve
planning and forethought in regards to the set of accounts used to
control its operations.

(2) In reviewing Albrecht's example cost report [3, p. 302] it
was mentioned earlier (see pp. 61—62) that some Of his accounts detail
seemingly trivial information while two accounts--salaries and equipment--
carry nearly 95% of his total costs without being further subdivided. The
degree to which this was true varied between firms. In firms E and F all
salaries in the computer operations area were lumped together in a single
account. In firm B salaries in the operations area were further divided
as follows:

Administrative and Supervisory Personnel

Employees Engaged in Processing, Control, & Scheduling

Employees Engaged in Operations Analysis

Employees Engaged in Data Entry Operations

However this same firm had only two hardware-related accounts:

Computer Equipment Rentals, Maintenance, and Extra Shift Charges
Data Entry Equipment Rentals

This was 100% more than several other firms where a single account was
used. A typical title was Rent--Machinery and Miscellaneous. Most firms
had a separate account for computer software purchased although this was
sometimes included in a general supplies account.

(3) It appeared that an increasing portion of the overall EDP
dollar was going toward personnel costs. In the case study firms this
percentage varied between 50% and 60%. Approximately 30% to 40% of the
EDP dollar was spent on equipment and equipment maintenance. The

remainder went for supplies, occupancy, and miscellaneous expenses.

These pet
for exam
mmlySt
share of
installa
costs it
systems
fiwevet
iollowi
system
furthe'

Specif

tool 1
were 2
TWO 01
seemet
Systet
POSsit

Contrt

SYSten
21 out
and re
theSe a
includ,

in the;

106

These percentages were very much in line with those reported by others.
[For example, see 149, p. 79] In the systems development area approxi-
mately 80% to 90% of the costs incurred were personnel costs. A large
share of the remainder was intra-firm D.P. charges due to the testing and
installation of new applications. With this high percentage of personnel
costs it would seem that a single Salaries and Wages account for the
systems area would not be sufficient to adequately control these costs.
However, all case study firms further subdivided this account in the
following ways. (1) All case study firms utilized a project reporting
system which related personnel costs to specific projects. (2) The firms
further subdivided the systems area so that salaries and other costs in
specific subgroups could be identified.

In all firms the project reporting system was held to be thg key
tool in the control of systems design costs. Generally two types of costs
were accumulated for each project--personnel costs and computer run charges.
Two of the firms that did not charge users for systems development effort
seemed to have better than average project reporting systems. Since
systems services were essentially a free good in these firms, it is
possible that a more complete project reporting system was needed to
control costs in this area.

The high degree of dependence on project reporting to control
systems development costs is supported by GAO data. This study found that
21 out Of 22 commercial firms visited employed a time or cost accounting
and reporting system for systems development activities. Two thirds Of
these associated dollar costs with personnel usage. Seventeen firms
included charges for computer hardware and peripheral equipment usage

in their project reporting system. Only five included charges for

107

supporting services such as keypunching. Thus personnel hours and costs
and direct computer charges appear to be the main elements of project
progress reports.

(4). Several case study firms made no effort to capture and
report occupancy costs as a data processing cost in either the systems
or computer operations area. In the GAO study only 12 out of 22 firms
accumulated facitilies rental as a specific D.P. cost, and only 14 of 22
did so with utility costs.

Although all case study firms accumulated and classified
certain supplies (tapes, "floppy" discs, computer paper, etc.) as a
data processing cost apparently this is not a standard operating
procedure for all firms. Over one third of the commercial organizations
visited in the GAO study (8 out of 22) did Egg accumulate these costs as
"data processing expenses." It is not clear exactly how they were handled,
but evidently they were included in some type of general corporate
supplies expense account.

Although occupancy, utilities, supplies, and other overhead costs
are not very "material" when considered separately and compared against
personnel and hardware costs, these smaller costs may exceed 12% of total
EDP costs. [149, p. 110] A breakdown for one firm as a percentage of
total EDP costs is as follows: supplies-—4.9%; occupancy-~3.9%;
overhead--3.9%. [149, p. 110] A firm attempting to aggregate "total
EDP costs for the firm" for either year-to-year comparative purposes or
as a percentage of sales to compare with industry averages would risk
inappropriate comparisons unless these types of costs were included in

its totals.

108

Responsibility and Cost Centers

 

All case study firms used some type of responsibility accounting
system. Not only did all firms split EDP costs up by at least two
functional areas--computer operations and systems design--but all firms
used additional functional classifications (e.g., data entry activities,
telecommunications, etc.) and/or further subdivided their responsibility
centers into subgroups (cost centers).

As an example, a partial organization chart from firm A is presented
in Figure 4-1. Firm B was visited shortly after a reorganization of the
EDP area. Figure 4-2 is the organization chart before the change, and
Figure 4-3 is after the change. The changes made in this firm seem
indicative of the trend in several firms--subdivision of large groups in
the EDP area, i.e., greater numbers of smaller groups or subgroups.

The responsibility accounting systems used directly paralleled
the formal organization charts. For example, in firm A (see Figure 4-1),
the Director of Resource Management Systems--Marketing received a monthly
cost report summarizing costs for his "division" in addition to four cost
reports for each of the departments under his control. Each of the depart-
ment managers received the cost report covering his area. The Director
Of Corporate Management Systems received a cost report for each of the
four "divisions" under his control (the same report as received at the
"division" director's level) in addition to receiving a summary report
of all costs under his control.

In none of the firms was there any indication of a desire to turn

to a '

'profit center" or "investment center" orientation rather than a
responsibility center structure. In all firms the EDP area was considered

a service activity, not a profit-oriented group.

109

.< aufiwtluugo dowumuuammuo mam ..Te snow:—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mzuemum 35%
.382sz 5.2.5»
a £22092 ”5:55
”.5552: -2354
seems mama «Sana
$03598 Same; Sofia egg
x328 3.3 .I. II
35280 J usage. 520E <53
a «memo
as: .5; mane; $853
a Essen 823.255 Ema—32¢: 228...:
Egg g8: 3:938
gems «ESE Ems Ema
Eng 2nd: I tame; 0325:.—
mmazuu 3.3 «was magmas. unname—
mfiezuo “.5983: SE a on:
$23 u SSE . 995$ cg”
5.2928 .5: 8:8me 8: uuéomnx Ease;

 

 

 

 

 

 

 

 

 

 

_ _ a _
e

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmmwé o ”.595 m @295 < $95 wZOHHHmOh 3.“.me “9:9,:
.595 EaHmmxm .59anme EnHmB—m mmflam Egg—2: SHE mama”: xuflomfizoo
mums NUH> NUH> NUH> mu—rfio ho MOHUNHHQ
_
mucouuooua oua> o>«u=uouu
unsounoum

vuuon 0:» mo :eahunno
uloho gut—BBB

muggy—Ha mo 28

 

 

 

110

 

 

 

HmMA<z<
HQNHOmm
«Oqum

 

02H22¢moomm
a mzuhmwm
mnmoomm
mMZOHmDU

_

 

.Owomno ouomon m angulluuoso ooaumufiomwuo mom

I.

 

2Hz=<mocmm NUZ<2EHZH¢S
mzmamwm H>HH w onmZMHNm

mzufimwm

IduamHZHzn<

 

 

 

 

 

 

 

MOH<2HQ§OO
hhdH

J

 

ozuzz<am
mzmamwm
mmabmzoo

_ _ 7mm?

 

 

x w

MNHDAZOU

 

 

— _ monouuua

 

 

— azaamumma

* .....

 

 

 

.Nuq «names
escapee muo~>~um
mwmum ozmmmmoamm ozmmmuoomm
amasmzco «essence
_ [It
uzmmmmoomm
mzuemwm
«essence

 

 

 

 

a .

 

F

 

 

_

mNOHhmum
MMHDmZOO
mo moaommHn

a

 

 

 

 

 

111

.owcmno momma m aufimlluumno aowumufiamwuo mam .mte ouawfim

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SHAH:
Mafia—Hanan

maouhomm 2U
4.22m . $8.52qu
348mg E8
Banana; 3
onSnHfimS ofifiaufi mzofléumo
453.2: _ session
, .538.— - t a —
38.858 nzwozozczuezumg Ha: _ 3:33; gouge—mo
Sad: 85% 8:3 «mango
no; 23

. Hugh
mug—"83m ”Ugh—hug —[
a mZOHHSHAmfi

 

 

 

 

 

 

 

 

gage

onus—~52» _ y L .—. 53
megs. _

a 28255 , 332.3 , 33.32 _
$955 3:503. . $3.558

t 8802 . $8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mus—8am 2.8% Enigma zones—$293 ufimmuoofi
«985 «8658 9855 82 9:51: guru
E880 «ESE Ease
.- 4 H 1 I F
- H .
.8028
5533

 

_

muquflum
$99500
ho QOBUHGHG

[J

 

 

112

The GAO study indicates that a great majority of commercial
organizations use various types of cost centers or "cost pools" to
collect and summarize EDP costs. Although the following categories are
not mutually exclusive, they represent the GAO's breakdown of the 22
commercial firms visited in that study.

Types of cost pools used:

Work functions 13
Organizational units 3
Major projects or programs -
Individual products or service 1
Other 4
Firms not using any form of
cost centers/cost pools __1
22

The breakdown is somewhat confusing. For example, organizational units
(number two above) may be organized on the basis of work functions
(number one above) and may be in essence the same breakdown of cost

centers .

Controllable/Noncontrollable Costs

In a good responsibility accounting system a manager is held
accountable only for those costs over which he can exercise some degree
of control. Some of the difficulties of determining controllable and
noncontrollable costs were discussed in the previous chapter. Although
in the case study firms attempts were made to include only controllable
costs in department managers' reports, this effort was more successful
in the systems design area than in the computer Operations area.

Most expenses included in the systems design area cost reports
can be considered direct and therefore controllable expenses to that area
and its manager. In the example on pages 103-104 the final group of
three expenses may be considered "assignable" costs. It was suggested

in the previous chapter that "assignable costs have controllable elements

113

in them, particularly when based on resource usage." [See page 70.]
All three of these assignable costs in the example were based on resource
usage and were considered controllable costs. Thus in terms of the
directlassignable/prorated classification the costs included in the
example on pages 103-104 are all controllable by the area manager. This
was the typical pattern in the case study firms.
If the managed/committedIvariable classification is applied to the
systems design area the picture is not as clear. First, variable
costs in this area are negligible in the short run. Second, a breaks
down between "managed" and "committed" costs in difficult and very
situation specific. For example, a high level corporate committee
decision may have committed the firm to the design and installation of
certain systems with at least an implied commitment of a certain group
of analysts and programmers. The costs relating to the other analysts
and programmers may be considered managed costs and could conceivably
vary from one budget period to another. The difficulty lies in separating
"committed" and "managed" costs. There is also the fact that the area
manager will be held responsible for the productivity of the analysts and
programmers whose salaries constitue "committed" costs. Problems such
as these are probably very key reasons why project reporting systems
were found to be much more widely used and valued as control tools in
the systems design area than the monthly department cost reports,
although the latter were also found in every case study firm.

In the computer operations area most costs fell into the "direct"
Category of the direct[assignable/prorated classification. Two firms
subtotaled the direct expenses before including certain miscellaneous

and prorated expenses. Titles used were "Total Direct Expenses" and

114

"Total Controllable Expenses." Items following these subtotals included

"net prorates,' work order credits,’ and "net adjustments to prior
years expenses."

As mentioned in the previous chapter (page 70) it is difficult
to imagine that fill the direct costs of the computer operations area
can be directly controlled by the manager of that area. Some costs--such
as long run lease commitments for hardware and certain Operator costs--
are committed costs resulting from past decisions. Other costs-~e.g.,
short run lease changes in the number of peripherals leased from
independent suppliers--are probably managed costs.

Determining which costs are or are not committed costs is difficult.
However, one thing is claer. Managers of computer operations areas are
entrusted with a very expensive and important set of resources, and they
should be held responsible for the efficient and effective use of the
resources under their control. This will generally entail an ongoing
comparison of costs with outputs or results. The measuring of outputs
will be discussed later in this chapter. Suffice it to say at this
point that the monthly cost reports appeared to be of considerable more
importance in the evaluation of computer operations than of the systems
design area. This is probably true because of certain dissimilarities
between the two areas, i.e., the computer operations area is considerably
more "manufacturing" or "production" oriented than the systems design area.

One further point was made by those interviewed in the case
study firms. To a large degree the costs of both computer operations
and systems design are dependent on the needs and demands of the users

of these areas. The computer operations area in the short run may serve

as an example. If suddenly users submit a new rash of applications or

very
EH8 C
shi.
capa

10M

unde

data

OVE

bud

the

th.

bu

Co

Op

PI‘.

C01

115

very high transaction volumes for existing applications, it is possible
that costs in this area will rise (e.g., Operator overtime, excess

shift rental charges) unless there is considerable slack or unused

capacity in current operations. This again points out the fact that the
monthly cost reports produced by a responsibility accounting system are not
sufficient for the evaluation of performance in the EDP area. The
underlying causes of the cost variations must be understood. Output

data must be available to compare with costs on an ongoing basis.

Budgeting and Reporting System

 

All case study firms prepared annual budgets in the EDP area
by responsibility centers. This was done in all firms as part of an
overall annual corporate budgeting procedure. All firms used "participative

"set" (or suggested)

budgeting," at least in theory. Department managers
their own budgets but Often encountered difficulties in getting them
approved. Cases were cited where approved budgets went all the way to

a top-ranking administrative committee only to have company—wide cuts

made. The budgets then filtered back down the organization structure

until adequate cuts were made at the department level.

One case study firm used flexible budgeting procedures in preparing
the monthly budget reports. Therefore, at the time of setting the annual
budget for the coming year, costs had to be broken into fixed and variable
components. Both the directorcfifthe EDP area and the manager of computer
operations in this firm thought that flexible budgeting was an inappro—
priate technique to use in the EDP area where most of the costs were

completly fixed in the short run. However they were both learning to

"play the game." They had discovered that if they classified more types

of c
cate
have
clas

This

The:
ass

had

vhe

ill
to
vol

att

ii:

te

bu

te
19

f0

811:

116

of costs or a greater percentage of a certain cost in the "variable"
category (regardless of its "true" nature), they would be more apt to
have a favorable budget variance. For example, in the 1975 budget they
classified 10% of computer operators salaries as a "variable" cost.
This percentage was being raised to 60% for the 1976 budget, regardless
of the fact that they considered these costs to be fixed in nature.
Their increases in the costs classified as variable were based on the
assumption that volume would continue to increase in the future as it
had in the past.

Flexible budgeting is a valuable control technique in situations
where a sizeable portion of the costs incurred in a given cost center
vary directly with some measure of volume. However, the above example
illustrates the extent to which the technique can be misused if applied
to an area which has predominately fixed costs and steadily rising
volume. CASB data confirms that few firms (only 29 out of 393--or 7%)
attempt to separate fixed and variable costs in the EDP area.

On page 117 is a flowchart of the budgeting process supplied by
firm D. The term "index" or index budget" refers to the index volume
budget, a flexible budget based on the most likely volume levels. The
term "standard budget" refers to the standard volume budget, a flexible
budget based on the actual volume levels incurred. The budget is
prepared several months before the new budget year begins. Thus the
term "1975 outlook (7 + 5 months)" refers to the seven months of actual
1975 data and 5 months of projected data which are available at the time
fo the 1976 budget preparation.

Note that under the "consolidation" heading, total D.P. costs are

summarized by "D.P. Burden Center." This is not the same as the set of

117

\7

 

 

 

 

 

.Q aunIImmoooua wowuowpsn ozu mo uumsozoam

 

 

 

 

 

 

 

 

 

 

 

 

 

 

     
     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.qus enemas

 

 

 

 

moa<zuqmooo
amoeba
t1 A<Huanzxoo
: = mdwuooeamdu V aw
: = wuuusuuauaonz saunaavuooo Avoxuu «any unquoaon
= : oou>uom J he venomoum .uasuaoun .oouuaaom osma .N
MOHEUNOOU Howgfl .UQOD Ofldflm SgHu—ao msmd now—Sufi chad / HOuQ—ufivuoou
n.0moa Accuses n + av nouquavuooo comm
.o> oomvaa mama sooauao mama so ouuueoum he wnwuowvso
usuaoo uovuo
daemon .m.n an IIL uuuauuoeov up compound loam uo>~oosonu an
unusuuoaon no no unaoooo as. an uaoauuoeoe an [\I.vouueoue anaconda: .H
once an vouceoum v ”mud” to:
nozHVAHDOI nymphet
ascendancy Auouvsn .ozuwaaso.
ouuuno noouwa oboe «coca .m.n no
huollsmv «suck I
— _ vuuvasum ruuuusoo .oasfiHUue
_ . Kenna [1" 33.233. 0;." .c
locum .vuovsdum whoa .u» nsaa
.33“; ...-n: 383. e.- 138 .. 2 a nudes—EA. _
cannons coon so sounding» «coo» adduuoxv n a
cannon noouua mo coin-cough” one o u no A «coca
.uaoe assuage so .ooam- Ava « a gas 61L .uaaqaua> .vanau
Canon—Iv 8.1.33 .5 I'll. luau 953.239 120.338 .53
undamooouh Quad 05a.— LEIMI Lx mug: notion—dd! chad .n
a zo~a<z<qum uoaeoo lawn
O1 . d 5.395 353.33
at ”coca .oanuauo> .uua .uuauz «mafia»
nouaa whoa .voxam so «unauuom
caduceum chad cannuluoa no“ we Hugo»
novaH mama nouns case see .uaaaauu> .uuuaa
soon Annuaoa n + my pudendum onaa o>aua mama «sad .nom
unease unannououm xoo~uno «sou manna mead ouwAHOOquom .uosouvdo: uomuuh .N
noon sou» mama finances n + NV cacao
.nom I «sou .uon gunman oocoexu Ixooauso nuoa uuoucaum announces
.ueov noun-no A: unannououn auovuo new so "Acumen no
u eunuoum Hooquunwm case came unaouua one ha cannouae sonny whoa .mu huesuouh .H
tomcat Iceman loommn
manmnzun 4
III I II 11477
zo~a<uoqq< one<nHSOmzou ozHameoo mmua<au o<ezmm mzanxums muam<a<m
«m36m24x
o u a a n r a a H m H u o a a n u m z m m x m o 2 H m m u o o m m < H < a

unencu-
vou soonooucaua
Mao wannabe .s

oowuuaooueoa .o

undue
Iloouoeuo ooanunnuu

new sauna .n

ouououn oow>uoo
ouauuo an. engages;
r. nowuouu.«aque< .c

mmos use a Adana:

and» won a meadow
ooouaoaaa
nua>aa we once .n

«cacao!»

11 non unseen.

haw-oh uo ouoluu-n .~

A5

 

 

fiwcuOOu undlo>ounlu .A

yea-ouvuoou
cocoon» Annouefilou
Iona ao«»nuuoan~

 

 

rgfi‘i

resp
used
the

the

c'na;
hea:
basu

rat

fir
ho]
mm

Th:

th
of

Ch

In
fi

in

118

responsibility centers being used, but is rather a group of cost centers
used to accumulate costs annually to aid in the setting of rates used in
the charging algorithm. Further discussion of these cost centers and
the way rates are determined is deferred to a later section of this
chapter. Notice however, in the flowchart under the allocation

heading, that the breakdown of 1976 D.P. expense by departments is not
based directly on their expected usage and planned chargeout rates, but
rather on the historical prorate percentage of the past 13 months.

The budgeting:h1user departments of D.P. expenses charged to
these departments was viewed by at least three case study firms as a
very important step in the control of EDP costs. Personnel in these
firms felt that it was much easier to control EDP costs by reducing (or
holding stable) user demand for EDP services. This was done by careful
monitoring and control of EDP costs in the user departments' budgets.
This approach was deemed a considerable improvement over attempts to
control EDP costs merely through that area's budget without regard to
the possibilities of increasing demand for services by users. Control
of EDP costs through users' budgets is possible only where some type of
chargeout or transfer pricing machanism is present.

There was little variation in the format of monthly reports
making cost/budget comparisons. At least five columns of numbers were
generally present: (1) current month actual, (2) current month budget,
(3) current month variance, (4) annual budget, and (5) year-to-date.

In several firms there was a column for the previous (actual) year-to—date
figures. Often, a budgeted and actual departmental head count was

included in the monthly reports.

119

The degree to which the budgetary constraints were enforced seemed
highly dependent on whether the EDP group in question was a corporate
staff group or was under divisional (or line) control. Across firms it
was observed that EDP groups under line control and authority had to
"live within their budgets" to a greater degree then did staff organiza-
tions. Although staff groups were expected to meet their budget, there
was no constant pressure from a divisional controller (with an eye to his
division's "bottom line") to keep costs exactly in line. In at least
two firms, there were direct remarks made that there was a more lenient
policy toward cost overruns in the staff D.P. center and systems group
than in similar groups under line authority.

In all EDP areas, whether they were line or staff, there was
considerable flexibility regarding the composition of costs as long as
the total was within the budgeted amount. Several managers commented
that if their costs exceeded budget estimates in some areas they could
often reduce or postpone costs in other areas. Thus, even in line
organizations, the budget was not a tool to force rigid conformity to a
prescribed plan, but rather a means of seeing that total costs were
controlled.

In only one firm did the D.P. personnel think that top management
had been "niggardly" (their term) regarding budgetary support of the EDP
area. This was probably due in part to the deteriorating financial
condition of this firm in the late 1960's, but D.P. personnel felt that
management had been "niggardly" on top of that. This condition was
somewhat alleviated when this firm was acquired as a subsidiary of a
larger firm in 1971. Conditions have steadily "improved" and currently
the firm budgets nearly 6% of sales for EDP costs (the highest percentage

among the case study firms).

in mi<

tiona
the c
rise
desig

hardu

repox

f011<

intc

desj

120

Most companies had no provision for revising the annual budget
in midyear. Those that did have such a provision actually used it only
in extraordinary circumstances. It was usually done because of organiza-
tional changes, and typically consisted of redistributing the budget to
the changed organizational units. One other circumstance that might give
rise to a budget revision was the top-level approval of a new systems
design project that entailed additional commitments of personnel and
hardware. However, on the whole, upward budget revisions were uncommon
in all firms.

The GAO study provided little information on the budgeting and
reporting procedures of the 22 commercial firms that were visited. The
following outline summarizes its findings in this area.

Commercial installation
a. Cost used as a control measure.
b. Periodic reporting of actual cost compared to budgets.
c. The degree of detail included in the reports varies
with the management levels of the users.

1. Very detailed reports are generated for local
managers of data processing installations and
slightly less detailed reports are generated
for users or requestors of the services provided.

2. Lesser degree of detail is provided to accounting
personnel.

3. Broad summary type reports are being generated
for use by top management.

d. Top management is frequently active during planning and
budgeting stages for projects.

Output or Activity Measurement
The discussion of output or resource measurements will be divided

into two areas—computer operations (including data preparation) and systems

design and programming.

121

Computer Operations

 

In the computer operations area there was a relatively high
degree of similarity in what resources were tracked. Five of the six
case study firms used an IBM system software package called System
Management Facility (SMF). SMF is designed to provide, among other
things, very detailed measurements of computer resources used (e.g., CPU
time and core, tape, and disc usage). The most obvious use of SMF data
is for individual job costing and accounting systems. However, the
usefulness of SMF data does not end there. Currach and Morino [35]

suggest the following breakdown for further uses of SMF data:

Workload analysis and classification
Trend analysis of resource utilization
Standards establishment and enforcement
Identification of resource wastage

. Detection of exceptional conditions

. Other factors affecting system performance

O‘U‘IDUJNH

Although some firms made use of SMF data for other than job costing and
charging purposes, there appeared to be much room for improvements,
particularly in light of the Currah and Morino paper. Firm C, which
made better use of such data than most, required that utilization data
from all computers used in the company be provided to the Management
Services Office (MSO), a high-level staff group which included the
systems area. In all the large-scale configurations in the firm this
data was provided by the SMF package in use. Computers with low or
declining utilization were subject to close scrutiny and possible removal.
However, another group which had access to this data and probably
made more effective use of it was a five-man group of EDP audit specialists
located in the Internal Auditing area. The head of the EDP audit group
had recently stopped an expansion project for a certain D.P. center which

had already reached the president's desk. Based partly on utilization

122

data, a decision was finally made not only to stop the expansion, but to
"reverse" the center to an RJE installation. (Evidently the data center
manager had been "empire building," to use the auditor's words.)

Although the capabilities of the SMF package seemed to be
underused, they may be put to better use in the future. For example,
in firm E measurements of computer usage were based on wall clock time
starting in 1971 (before that, usage was evidently not tracked at all).
However in 1975 this firm began to implement RAS (Resource Accounting
System). Initially this is a job costing procedure relying heavily
on SMF data. However, it is expected that greater use will be made of
SMF data along the lines suggested by Currah and Marina as the system
is developed further.

The firm having the Burroughs 6700's also used a resource utili-
zation package similar in purpose to IBM's SMF. This firm was also
beginning to make wider use of this package and was planning on following
up its Resource Utilization System with a Data Center Information Reporting
System. Extensive daily and monthly reports were already in use at the
corporate data center and were being implemented throughout the firm's
other data centers.

Because the resource accounting systems in use were such an
integral part of the charging algorithms used, further discussion of
specific resources tracked will be delayed to that section of this
chapter. In summary it may be said that SMF and similar type data were
not fully utilized in the management, control, and planning of data
center operations, although several firms were beginning to make much

greater use of this data.

123

All case study firms used some type of work measurement system to
report on personnel productivity in the data entry area. Usually this
as measured in terms of strokes perhour, cost per line (a standard number
of strokes), or a similar measurement. Firm E had recently implemented
a Data Entry Statistical System (DESS) as part of its overall Resource
Accounting System. DESS accounted for all data entry (chiefly keypunch)
activity. Among the reports generated were daily summaries by operator
and by job. The daily report on an operator's activities had the following
column headings;

Job

Description

Operator (name and number printed only on the lst and total
* lines of the report)

Strokes

Hours* (in hundreths)
Strokes/Hour
Cost/Hour (a constant)
Cost*

Cards

*
Operator Errors

The statistics for each job worked on during a day occupied a separate
line on the report. The starred columns were totaled and an average
number of strokes per hour for the day was computed. Nonproductive
time periods, such as idle time and lunch and breaks were added to pro-
ductive hours for a grand total hours worked.

As was expected, there was considerable variance in performance,
even between different jobs punched by the same operator in one day.
Strokes per hour (by jobs) ranged from a low of 2,400 on a very small
jobs to 18,431 on larger jobs. Small jobs probably required as much
set up time as larger jobs and therefore resulted in lower averages.
Also, some jobs are inherently more difficult to punch and can be expected

to take greater amounts of time.

124

Another daily report provided by DESS was the "Keypunch Statistical
Analysis Report by Job. Column headings included:

Job

Description

Operator

Strokes

Hours

Strokes/Hour

Cost/Hour

Cost

Cards

Standard Rate
Thus the data for a given job was broken down by operators that punched
that job and then totals and averages were summarized. Although DESS
was in Operation in Firm E, it had not been fully tied in with the
charging system, but plans had been made to do so.

One thing that was expected, but was not found, was specific
standards for data entry personnel. Rather than setting specific goals,
the case study firms monitored output individually and compared it with
that of one's co—workers. Evidently the knowledge that they were being

monitored and the ability to check on their own progress was enough

motivation to increase or maintain performance in this important area.

Systems Design and Programming

 

Monthly cost reports, by themselves, are of little use in
controlling and managing the systems design and programming area. Since
most costs in these areas are fixed personnel costs, there is little
fluctuation in amounts from month to month. More important is the
attempt to measure and quantify human productivity and progress made
on the projects under way.

Thus, all case study firms had detailed project reporting systems.

Although these differed in detail, there were many points in common.

125

All had some type of time recording system as a basic input into the
project reporting system. All had similar objectives:

1. To allow management, both within and without the EDP area,
and systems personnel to observe and evaluate the progress made on the
various projects under way.

2. To build a data base of historical data in order to improve
time and cost estimation on future projects.

Three of the case study firms charged for systems work and a fourth used
a memo billing procedure. For these firms the project reporting system
had a third objective:

3. To provide input data into the charging algorithm, resulting
in charges to users of systems services.

The one report common to all case study firms was the project
status report. Called by different titles in different firms, this
report was generally issued weekly. In its simpler forms this report
was concerned chiefly with manhours, not dollar costs. For example, the
report used in firm D contained the following information on each project:

Request #

Job Code

Project Name

Requesting Group (or Department)

Requested by (individual's name)

Programmer (Analyst)

Actual Hours (itemized by programmers)

Estimated Total Hours

Total Actual Hours

Percentage Complete

Date Received

Last Activity (date)

Estimated Completion Date

In more elaborate reports, there was a breakdown of analysts; time, a

recording of computer charges by projects, and a reporting of total dollar

126

costs. In firm A the following items were reported on a monthly summary
report:

System Name and Number

Year-to—date Distribution of Analyst Time (in man-months)

Functional Analysis
Technical Design and Specifications
Programming
Documentation
Supervision
Training
Other
Year-to-date Man Months
Total
Budgeted
Year-to—date Computer Charges
Total
Budgeted
Total Year-to—date Annual Budget
Total Year-to-date Actual Costs (includes computer charges and
analysts salaries)
Usually at least one other report was generated--a breakdown of analyst/
programmer efforts by individuals rather than by projects. Thus it was
possible to tell which projects an individual was working with, and what
types (and amounts) of work he had recently done on those projects.

Most managers were enthusiastic about their project reporting
systems as a control device, but occasionally some trouble spots appeared.
For example, Firm D let the analysts involved with a specifc project
estimate the percentage of completion. (Who should know better than they?)
Therefore, in the early stages of many projects, the percentage of
completion seemed to correlated nicely with the ratio of actual hours to
estimated hours. But when the percentage of completion neared the 90—99Z
level, the number of actual hours used continued to rise with very little
change in the completion percentage. More than one project was noted on

this firm's project status report that was supposedly 99% complete and

already the actual hours worked were more than 350% of the original time

127

estimates. Doubtless, the projects were less than 99% complete because
the revised estimated completion dates were still some time in the future.
The evaluation of analyst output and performance was a very
subjective task because of varying degress of project difficulty and analyst
experience. Even the EDP audit specialists in firm C did not pretend
to quantitatively evaluate human outputs. Rather, they relied upon
direct observation of the analysts at work.
The GAO summary of resource accounting systems was quite vague
and inconclusive. The following definition was used:
A resource accounting system tracks, accumulates, aggregates,
and reports the use of data processing resources in terms of
physical consumption measures, such as staff time in terms
of hours or years, computer accounting units, etc.
Using this definition, the GAO study found that there were 12 out of 22
commercial installatins visited which used resource accounting systems
as a tool for managing and controlling data processing activities.*
Other conclusions included:
a. Output from these systems is also being used as a basis for

applying costs to systems development projects and on-going
jobs.

b. Output from resource accounting systems was considered an
excellent tool for internal management and control at the
data processing installations visited.

c. However, personnel at these installations still consider
"dollar costs" as the most common, understandable expression
for managing and controlling data processing resources.

 

*This number (12) seems to be inconsistent with other GAO data.
For example, it was found that 20 out of 22 firms made users aware of
EDP costs incurred for their benefit, either through cost assignment or
memo billing. It would seem that an elementary resource accounting
system would be the basis for even the most simple billing algorithm.
Therefore it is difficult to understand why only "12 out of 22" firms
used resource accounting systems. Perhpas the people interviewed were
unfamiliar with GAO terminology.

128

Long-range Planning

 

Perhaps no other facet of control was subject to such wide
variations between firms as was that of long-range planning for the EDP
area. Because of this wide divergence of views and practices, the six
case study firms will be discussed individually.

Firm A had perhaps the most comprehensive and unified long-range
plans. First, the overall goals and strategies of the corporation
were laid out for a five year period and then translated into the
outline of a corporate five-year plan. Then operating and staff
groups, including the corporate EDP systems group, analyzed the plans in
terms of the support required from their group. The EDP area makes
hardware and manpower projections and converts these to dollar terms for
the five year period. Under consideration was the lengthening of the
time frame to ten years. Approval of a five year plan does not commit
funds to a particular group. For example, approval of a particular
hardware increase or improvement in the five-year plan does not
necessarily mean that it will be approved in future operating or capital
expenditures budgets. In this particular firm, personnel increases in
the EDP area were becoming more and more difficult to get approved (both
in the five-year plan and in operating budgets).

All hardware acquisitions and planned system projects had to be
justified on a cost/benefit basis excluding intangible benefits. Personnel
there claimed a 302 ROI (return on investment) on hardware expenditures,
but this was on an.§ REESE} basis. As in the other case study firms there
was little post-implementation examination or audit of hardware or
personnel expenditures to see if the claimed benefits (real or intangible)

had ever been realized.

129

The EDP area in firm B was the only one which had a specific
group in charge of planning--entitled Computer Systems Planning and
Administration. This group was responsible for budgeting, planning,
sepcial studies, and "Office Administration" for the EDP activities in
the firm. A long-range corporate plan had last been done in 1973 with
the EDP area basing its expectations and plans on corporate goals and
strategies. For some reason, (possibly the drastic and continuing changes
in the corporate environment during this period which probably invalidated
most of the long-range work) further corporate long-range plans were not
prepared. The EDP area was hit with cost reduction pressures and
"insufficient manpower." Besides, it was explained that it would have
particularly difficult (add: and probably useless) to attempt a
long-range EDP plan without an overall corporate plan. Thus the
"planning office" is left with only short-run budgeting and planning
functions despite its desire to be more future oriented.

Firm C was the organization which had initially attempted to
centralize its EDP facilities, only to reverse it course of action.

(See pages 98-99.) Before the split-up of the centralized data center
there had been a unified five year plan for the EDP area in terms of
manpower, hardware, and dollar costs. For some reason, this practice

was discontinued after the data center was divided and the smaller groups
put under line authority.

Long-range planning was done in two ways: (1) a Project Budget

*
Request (PBR) and (2) Program Approval. The PBR was something like a

 

*Note: Do not confuse a PBR with a typical user request for
systems work. One recent PBR in this firm was the result of six months
of collaboration between the corporate staff systems group and a divisional
data center. It requested an optical scanning system with an annual cost
of $500,000 and expected annual benefits of $1,500,000.

130

capital expenditure request but encompassed much more and included
projected manpower requirements and costs. Very high level approval was
required. For PBR's in the EDP area, the data processing personnel would
prepare most of the required estimates, but a corporate profit analysis
group drafted the final written project request. Projects had to be justified
on the basis of tangible benefits, but intangible factors were included
in the project request and supposedly considered. ROI was used as a
major criterion for acceptance although the cutoff return was not given.
There was no way of accurately measuring the resultant benefits after
a project was implemented and no attempt was made to do so. Follow-up
procedures overall appeared quite weak. Once a project made the
approval/disapproval hurdle, it was usually free sailing.

The second long-run planning device, the "Program Approval,"
was not an authorization to expend funds but rather approval of a course
of action. It encompassed a longer time span than a PBR and was consider-
ably broader in scope. It required the highest level of approval--the
administrative committee. In such a document, manpower and hardware
projections were made for as far as six years into the future. Updates
were made as often as quarterly.

A major problem with this means of long-range planning is that
it seemed to start with what the EDP personnel want or think they need
1J1 the future, and then is examined to find if this fits in with overall
<morporate goals and strategies. It would seem that corporate goals should
Come first, and then EDP personnel should determine how they can help
reach those goals. Despite the use of these two means of long-range
Plianning, the company's own internal EDP auditor declared that "the

abeence of solid long-range plans in the data processing area is one of

131

this firms's greatest failures." (The comment is probably too harsh
on firm C, particularly in light of the quality fo the long-range
planning done in firm D.)

The directors of the overall EDP area for firm D at first said
that no long-range planning was done, but then backed down from that
position a bit. What plans do exist seem to be very informal (even to
the point of existing only in the back of the manager's mindl). Since
all equipment was leased, the EDP area felt no need for separate budgets
for operations and capital expenditures. Evidently additional hardware
rental charges were approved only through the annual budgeting procedure.
There seemed to be little effort made to analyze users' long-range
requirements and coordinate these in any formal long-run EDP planning.

In firm E the corporate five-year plan consisted of a series of
pro forma income statements and supporting schedules. It was soon to be
expanded to include pro forma balance sheets. The main types of project-
ions made by the EDP area were manpower and hardware estimates. These
were translated into dollar figures and "other costs" were added to the
hardware costs so that total EDP costs in the five-year plan were broken
down into two groups--personnel and other. The plan was updated annually.
Users' long-run requirements were not specifically estimated or coordi—
'nated and provided little input into the EDP projections.

Firm F had a corporate five—year plan in which the EDP area
participated. In addition to the two usual types of projections--hardware
and personnel-—a third area was considered--space or occupancy needs.
TPhese three were translated into dollar mounts for planning purposes.

Ihlans were updated annually. There was some effort made to coordinate

132

users' requirements in these plans, especially if major new systems were
to be implemented during the period.

Firm F purchased tape drives and disc packs and leased CPU's,
but the two groups were handled differently. Since all hardware purchases
had to be projected and properly approved via a capital expenditures
budget, the tape drives and disc pack units were included there. Long-
run lease commitments, and thus the CPU's, were included in the operating
budget, but were supposedly looked at and reviewed in the same manner as
capital expenditures.

The GAO study reported very little in the area of long-range

planning.

 

Tracking Users' EDP-related Costs

The case study firms varied widely in their ideas about and

approaches to tracking users' EDP-related costs. Two firms (A and E)

did not attempt to even summarize total data processing costs incurred

in the firm, much less determine users' EDP-related costs. In firm E

the stated reason for the unconcern in this area was that all hardware
acquisitions within the firm had to be approved by the MIS manager. Thus
users' could not acquire hardware (likewise presumeably incur personnel
and other costs) without his knowledge.

In firm A part of the reason for unconcern in this area was
probably the difficulty of collecting these types of costs because of the
widespread use of data processing equipment throughout the firm. However,
firm A needed to know user-related EDP costs as much or more than the
other case study firms. An example of such a need had occurred within
the last several years. The corporate data center, in an effort to

discourage card input, had gradually reduced the number of keypunch

133

operators at the center from 30 to 3. Users were forced to do their own

keypunching or find some other means of data entry. However, there was

no way to track the users' EDP-related costs, and thus it is not known

how much these increased after the cutbacks in keypunch operators.

The remaining four case study firms at least attempted to prepare

an annual report on total corporate data processing costs, and included

varying degrees of users' EDP-related costs. The most common types of

costs included were "dedicated" hardware costs and related personnel

costs within user areas. Next came the costs of liason or coordinator-

tYPe personnel who, although not located within the data centers or

SYStems groups, nevertheless spent most of their time on data processing

CYPe activities. Firm D was probably the most careful in this respect.

It went to great lengths to estimated the percentage of an individual's
time spent on data processing related tasks, and charged that percentage

Of his salary as a data processing expense to be included in the annual

summary of such. Personnel in the firm considered this annual summary

‘10 be a very accurate report, and it probably was by comparison with the

0timer firms' efforts. However, one factor that made the going easier in

firm D was its high degree of centralization. Only 20% of its EDP costs

were incurred outside of the centralized corporate data center. By
Comparison, firm P which had a centralized corporate data center incurred
52% of its total EDP costs outside of this center. And this did not
include the salaries of individuals whose time was split between the

EDP and some other area.

The GAO study reported little information in this area.

134

Charging System for Users
Although all case study firms had some type of charging

nne:cihanism.for computer operations, there was considerable variation in the

1:1:<)cedures and methods used. Firms also differed widely in their

£11>1>roach to systems and programming costs.

I?imrm A

Firm A operated 18 decentralized data centers throughout the

czcatintry. Prior to 1975 each center had had considerable leeway in

(leatermining its own charging policies and algorithms. However, in 1975

titre corporate systems group began issuing a series of releases giving

Czllarging policy guidelines to data center managers. The guidelines were

1>£1sed on policies used at the corporate data center and were a fairly

<:<)mplete documentation of how rates were set at that installation. The

illithor was fortunate enough to obtain a copy of the first and most

:Lmnportant release of over 500 pages which had just been printed and

<:<Dllated on the day of his visit to this firm. This was by far the best

clocumentation of charging procedures, goals, and algorithms that any

Case study firm had or at least made available to the author. Because

C>f the large amount of material covered in this document only the

Elighlights will be touched upon in this chapter. The reader is referred

t:o Appendix A for further interesting and important extracts from this

document .

One of the most important statements made in the document was

t:hat regarding billing algorithm goals.

The billing algorithm of an accounting system should meet
several goals:

a. Accuracy - Usage data gathered should indeed represent the

true amount of a resource consumed by a user task.

135

b. Repeatability - The charges for a given user task run at
different times in different mix environments should be
identical :_a small acceptable tolerance.

c. Flexibility - A number of resources should be monitored
for usage and charged to the user task. A given
installation charges for those items it has determined
are significant. However, since the computer operations
environment varies from site to site, a flexible billing
formula is necessary. The items of charge may change
dynamically.

d. Simplicity - Although flexibility is important in order
to insure a user is charged for those resources used,
simplicity must temper the tendency to create a complex
billing charge function containing too many terms for the user
to be able to comprehend or to respond.

e. Linearity - A linear billing charge function meets the goal
of simplicity while allowing recomputation of charge by
providing for the capability of accumulating units used and
than later extending these by a constant rate.

f. Reflection of true cost - It can be shown that when charges
to users reflect the true cost of resources used, the
system configuration will tend to be modified (excluding
non-economic political considerations) to achieve an economic
optimality. In other words, the system will reach maximum
cost efficiency because users will (given a choice) buy
resources which cost the least to the user. Since the
resources chosen really are the cheapest, the system is
optimized - each subsystem unit is justified or removed
as not cost effective.

g. Modification of User Behavior - This goal is a corollary
to (f). Demand should be stimulated for resources which are
most cost effective. This stimulation is carried out by a
favorable pricing structure to those resources which have
low value either by virtue of low cost to the data center
or by the resource's oversupply (due to a relatively
permanently imposed improper configuration).

Most of these goals can be found in the EDP literature. Several of these
directly parallel the list of "attributes of a charging technique" made
by one firm in Sollenberger's NAA study. [142, pp. 133-134] However,

a goal such as "flexibility" is tailored to this firm's particular needs--

the variations in environment and users from data center to data center.

136

Some goals are obviously conflicting--such as "accuracy" and "simplicity"--
so we are forewarned that there will probably be tradeoffs in each area.

The list of billable items and charging algorithm are presented
in Figure 4-5 on page 137. The charging scheme appears quite complex
at first glance, yet several EDP personnel in the corporate systems group
were adamant that most users were quite sophisticated in interpreting and
responding to the charging scheme. The following example of a user
modifying his behavior in response to computer charges was cited
(although it is doubtful whether it proves he was a "sophisticated" user).

The user requested a production planning report which was run
monthly and used tremendous amounts of I/O. It was costing him around
$1,500 a month in charges for this report alone or $18,000 annually.

So the user went to the corporate systems area and in effect said: I
need the report, but it has to be done cheaper than this. He asked for
something in the range of $100 a month. The systems area revised some
of the routines and made the peripherals controlling the disc packs more
efficient. The result: the user did get "virtually" the same report
for a monthly cost of around $100 or $1,200 annually.

Despite the detailed charging algorithm for computer operations
and the detailed documentation in this area, there is no charge for the
services of the corporate systems group. Thus users are quite willing
to use its "free" services as cited in the above example. Systems
development costs are considered part of corporate overhead and are
included in the proration of that total figure.

Further documentation of firm A's activities is presented in
Appendix A and includes:

1. Overview of related subsystems and their functions.

137

Billable Items:

Task related - 1. Processor time

2. I/O time

3. Cards read

4. Cards punched
5. Lines Printed
6. Code Core

7. Data Core

8. Connect time
9. Queue premium
10. Shift premium
11. Tape mounts
12. Tape Open time

Miscellaneous - l. Inhouse keypunch
2. Vendor keypunch
3. Tape library (can be used for pack library)

4. Offline

5. Clerical

6. Xerox

7. Miscellaneous items

The billing formula for tasks is:

n
task
charge - 1:: 1 (units used of 1) (rate for i) x
m
1 + 2 premium j
j = 1
where: n = number of billable items

m = number of premiums (discounts)
rate for i = some constant (perhaps 0)
premium j a a function surcharge (+) or discount (-)

Figure 4-5. Billable items and charging algorithm-—firm A.

138

2. Output reports generated.
3. Explanation and rationale of billable items.

4. Overview of policies and procedures.

Firm B

Firm B had a long history of data processing experience and
initiated a charging scheme in 1964 for its IBM 360. At first charges
were based on wall clock time, but with the advent of multi—programming
there was an attempt to have a more detilaed algorithm including
CPU time and I/O. However, because of wide fluctuations of costs
from the latter algorithm on different runs of the same application,
users were billed for the lower of this method or wall lock time.

There were other problems. Computer operations changed rates
quarterly in an attempt to make chargeout "revenues" equal department
costs. Thus none of the users understood how they were being charged,
much less how to react in an economic and rational manner to the
fluctuating rates.

With the installation of two IBM 370/158's in 1973 the firm
began to use the SMF package for resource measurement and inputs into
the charging algorithm. Rates were set annually instead of quarterly.
Charges were much more "repeatable" in varying mix environments. The
schedule on page 139 shows the billable items and comparative rates for
1974 and 1975. The reduction of rates in 1975 on a number of items was
attributed to "reduction in both the amount and cost of the equipment
along with a comparable reduction in the number of personnel required
to operate the equipment." Rates were set on the basis of usable
(practical) capacity of the equipment, not on the basis of expected usage.

Thus all computer operation costs were not charged out and the overall

1974-1975 COMPUTER RESOURCE CHARGING COMPARISON

Resource Center

 

158-158 CPU
Core Storage
Magnetic Tapes

Disk Storage -
Dedicated

Disk Storage -
Nondedicated

Printers

Card I/O

Data Entry
Remote Job Entry
360/30 Emulation

7074 Emulation

Microfilm - Original

Microfilm — Duplicate

139

 

 

 

1974 Rates 1975 Rates
Unit of Measure gper Unit per Unit
CPU Second .04557 .03902
KeByte Hours .07448 .01552*
Tape Drive Hours 11.15 6.28**
Disk Spindle per 1055.00 1135.00
Month
Disk Track - Hours .00279 .00183***
Lines Printed .0006 .00052
Cards Read or .00062 .00048
Punched
Elapsed Hour 7.45 7.45
Lines Printed .00075 .00052
Elapsed Hours for 31.50 Same as
Old Rates Only Other Jobs
Cost per Run 1973 Avg. 1973 Avg.
Cost Cost
Original Fiche 1.50 1.50
Produced
Duplicates Produced .08 .08
The following items will be charged directly to the applicable

using departments on the basis of costs incurred.

--Optical Scanner
--Time-Sharing
--Telecommunications
-[code name] Data Link
--Custom Forms
--Dedicated Keypunches

* A KrByte hour represents occupancy of 1000 bytes of core storage for
one hours of elapsed time.

** A tape drive hour represents the occupancy of one tape drive for one
hour of elapsed time.

*** A disk track hour represents the occupancy of one disk track for one
hour of elapsed time.

Figure 4-6.

Comparative charging rates--firm B.

140

D.P. area absorbed the remainder which became, in effect, part of
corporate overhead. At least users were not charged for capacity costs.
Users were required to use the published rates for the upcoming year to
determine their dollar budget for EDP costs. Thus a fairly high degree
of user sophistication was expected and assumed.

A short study had been made of the desireability of incentive
charging rates, i.e., a priority pricing scheme. The study is reproduced
in Appendix B. It is interesting to follow the reasoning in the study,
particularly in the main reason given for rejecting the priority pricing
scheme-that user rates might be forced to fluctuate and operating costs
would probably not be reduced. Both ideas have some validity. The users'
bad experience with fluctuating rates prior to 1973 had no doubt left
them dubious of any system where rates might change more often than once
a year. And no doubt operating costs would probably not be significantly
reduced; however, an effective priority pricing system might smooth user
demand and postpone possible increases in hardware and personnel. It
should be noted that not all EDP personnel were in favor of scrapping
the priority pricing plan. The Director of Computer Systems Planning
and Administration believed that such an idea was feasible and would
eventually be implemented.

In a way, it was a sort of back-handed compliment to the EDP
users in this firm to have a priority pricing system rejected because of
fear of being swamped by nighttime, lowbpriority jobs to be run at the
cheaper rates. The implication was clear that EDP personnel believed
most users to be quite sophisticated and responsive to changes in the
charging algorithm. Whether this was in fact true was debatable. One

reason it could be questioned was the fact that interdepartmental EDP

141

charges apparently did not have a separate account in a user's chart of
accounts. They were lumped with other interdepartmental charges or even
included under "Miscelleneous--Business Expenses." EDP personnel felt
strongly that a separate account should be used for their charges in
users' cost reports, and this change was probably to be made in the
next budget period.

Firm B used a memo billing procedure for program maintenance
and systems design work done for users. The Director of Computer Systems
Planning and Administration felt that these items would eventually be

formally charged out and included in users' monthly cost reports.

Firm C

Firm C had a large number of computer systems located throughout
its various divisions. A corporate staff group collected various
utilization measures on hardware used throughout the firm. However,
there was no company-wide policy requiring the divisional data centers
to charge users although they were encouraged to do so. Charging
practices varied from no charging scheme to full chargeout of all computer
operations and systems costs. The policies of two data centers at
opposite ends of this spectrum will be examined.

The Finance/Personnel Data Center is an example of a large
installation in firm C that did not charge users. Two IBM 370-145's
were in use in batch mode. Approximately 100 people were employed at
the center, with almost 50% of these in data entry (CRT to disc). Three
staff groups are the sole users of this data center--the personnel office,
the controller's office, and the treasurer's office. One of the main

reasons that users were not charged was that the jobs run at this center

142

were not discretionary (e.g., empoyees must be paid) and the user
groups were not particularly interested in knowing the cost of the jobs.
In fact, the users could not really control the length of jobs such as
payroll which was a function of how many employees the company currently
had. There were no charges for systems design and programming efforts,
because again much of this was nondiscretionary and of a maintenance
programming nature.

The data center that has "the most sophisticated charging
system" is located in the Sales and Marketing Division. Two IBM 370-158's
were in use. The billable items and related rates are given in Figure
4-7 on page 143. However, the schedule is misleading regarding the way
the billing operation really works. The rates are used to estimated the
total charges for a year for a given user based on his prior year's usage
of these resources. This total amount is then split into twelve "monthly
assessments" which are charged as lump sum costs to the user. These
assessments remain constant for the year unless there is a "material
change in the number and size of applications run for that user". Thus
there is no direct relationship between the resources consumed in a
monthly period and the charge for that period. Users are not charged on
a resources used basis but rather on an average anticipated resources
usage basis. If "material changes" in a user's ongoing applications and
systems work occur during a year then the given rates are used to calculate
the gross charge billings for the excessive unanticipated usage.

No prime time charge is used. There had been some experimentation
with the idea several years back, but it had been deemed too costly to

administer. It is difficult to imagine how a priority pricing scheme

143

1975 CROSS CHARGE & ESTIMATING RATES

 

 

Systems 3 20.00 Hr.
Programming 18.00 Hr.
Data Entry 8.00 Hr.
E. A. M. 15.00 Hr.
370: Core (CPU hours x average K—Bytes) x s 1.50 Unit
Tape (Tape EXCP's/15,000) x 8.50 Unit
Disk (Disk EXCP's/15,000) x 8.00 Unit
Peripheral (Periph EXCP's/15,000) x 10.00 Unit
1975 PERMANENT DISK STORAGE CHARGES
Per Byte Per Track Per Cyl. Per Pack
g1} (13,020 (247,570) (100,018,280)
Month (1) .00000769c .09902c $1.881 $760.14
Day (30) .00000025¢ .00330c .062 25.34
Hour (720) .OOOOOOOl¢ .00013¢ .002 1.06
Figure 4-7. Billable items and charging rates-~firm C.

144

would fit into a charging system using lump sum charges. Thus users

" for prime time work by experiencing a longer turnaround during

"pay
this time.

The rates quoted for systems and programming services were
somewhat of an anomaly. Most of this type of work was done on program
maintenance and was not directly charged to users. Rather these
systems charges were included in the fixed "monthly assessment." However,
new systems development work was billed to users at the quoted rates if
it was "material." Specific criteria for "materiality" were not given.

Not everyone in the firm was pleased with the charging system.

The head of the internal EDP audit team made the statement that the
company "really doesn't have a chrging system." He, along with others,

believed that users should be charged on a resources used basis for both

computer operations and systems work.

Firm D

Firm D had a policy that the EDP area must be a "disappearing
cost center" and chargeout 100% of its costs. Rates were developed
annually for various billable items, and these estimated rates were
used during the month. Appendix C presents portions of a recent study
done within firm D listing the billable items and documenting how the
"estimated rates" are developed. A schematic of the charging or alloca-
tion process is presented.

At the end of each month there was a "spreadback" of the difference
between costs charged out and actual costs. This "spreadback" was based on

the percentage of EDP costs intially charged to a users. Thus, using a

very simplistic example, if there were only two users and user A and B

had estimated charges of $150 and $50 respectively, and actual EDP costs

145

were $240, the "spreadback" procedure would allocate an additional $30
to A and $10 to B.

This procedure was in effect a very precise form of allocating
actual costs. It was not a transfer pricing system be cause of the
uncertainty of the actual amounts charged to users, even if their exact
resource usage was known in advance. Although the "estimated rates" were
set for annual periods, the "spreadback" procedure created actual rates
that varied from month to month.

Other drawbacks to this sytem included the following:

1. Budgeting on the part of users was made more difficult
because of variations in the effective rates.

2. A user's charge was affected by what other users did or
did not do.

3. Users were charged capacity costs which they were not
responsible for.

In defense of the EDP personnel at the firm, it should be noted
that it was not their idea that the EDP area should be a disappearing
cost center. This was a requirement passed down by top management, and
they were forced to live within this constraint. The detailed measurement
of selected resources (as described in Appendix C) is certainly better
than the two earlier chargeout methods that preceded the method currently
in use.

Prior to 1971 Firm D based computer charges on CPU hours. In
that year the basis was changed to core hours in an effort to be more
accurate and equitable. No attempt was made to charge for peripherals
on a use basis. The reasons given at that time were:

1. It would be too expensive to accumulate the data and use

it for charging purposes.

146

2. Users should feel that the peripheral equipment is for
everyone's use and not just for one job.

Evidently the costs to accumulate such data dropped over the
next several years and the latter, somewhat nonsensical, argument
disappeared. The current charging system as outlined in Appendix C
does charge for certain peripherals on a use basis.

Evidently there was a strong user reaction when the change was
made to a detailed charging algorithm. Some on-line users' charges
jumped by approximately 500%. This was reduced somewhat after "adjusting"
the algorithm, but these types of charges were still much higher than
before. Interstingly enough, despite the protests, no one cut back in
their demand for services.

The Director of Data Processing felt that if a lump-sum, fixed
charge were made to all users or no hcarge at all, that there would be
a substantial increase in requests for services. On the other hand,
he felt that if the rates were raised to include a profit figure, 95%
of the jobs would remain because users feel they are so essential.

Firm D was the only firm that could state even a rough dollar
figure as the cost of their charging system--$20,000. It was felt that
the benefits derived from the system very much outweighed the costs.

The GAO study found nine commercial firms that could give a rough range
for the annual costs of their charging system. The following categories
were used and results were obtained: "minimal annual costs"--three firms;
$50,000—100,000~-one firm; $lO0,000—$200,000--five firms. Thus firm D
was probably in the low end of the "minimal" category.

Firm D made excellent use of its charging system as an indirect

cost conrol tool. EDP costs in users' budgets were closely scrutinized

147

and controlled. This forced users to have only their most important
applications run and lessened or controlled demand for EDP services.
Thus, without a continuous flood Of new applications and constant demands

for new systems work, cost within the EDP area were more easily controlled.

Firm E

Firm E was particularly interesting because it was in the process
Of developing and implementing a charging system base on detailed
resource usage. Most Of the system was to be Operational by and in use
during 1976.

Prior to 1971 there was no apparent attempt to allocate data
processing costs directly to users. Such costs were lumped in with other
corporate overhead costs and allocated on the easiest basis--sales. In
1971 the firm began to make assessments of computer Operations costs at
the divisonal (not the departmental) level based solely on wall clock
time. Charges were made for systems development projects and the EDP
area attempted to be a disappearing cost center.

A new Resource Accounting System (RAS) was being deveIOped and
implemented over 1975-1976. The following points were given as reasons
to undertake such a project and are taken from a presentation made to
users:

Why have RAS?

1. M18 resources are expensive.

2. Without RAS, we had no way to determine what systems
development costs were or what on-going systems operating
costs were.

3. Charging provides an economic basis for resource allocation
if resources are limited.

4. Cost benefit analysis can be done by the user; otherwise,
only MIS can evaluated the economic feasiblity of proposed
systems.

5. User has a "stake" in the efficiency Of systems and the
services provided by MIS.

148
6. Users will be "self-governing" in their utilization of
M18 resources if the resources are not free.

Statement six was often stated or implied by personnel in other
firms, but was not documented elsewhere in such a straightforward manner.
The direct implication of the statement is that i£_MIS resources are
free, users cannot be self-governing (presumeably because Of their
high demand for such resources), and therefore some priority system
must be used. This could take the form of a user group or committee,

a high-level executive, a top management committee, or an EDP manager or
committee. However, the fact that charges are made for MIS resources
does not mean that users are completely "self-governing," particularly
in the systems design area. Projects must still be approved at some
level. In most firms the requisite level Of approval increased as the
estimated costs Of the project increased.

Prior to the development of RAS in firm E, project approval had
to be given by the manager of systems development, the director of MIS,
and the manager Of the user division. It was not clear if this policy
was to be continued after the introduction Of RAS, but if so, it would
definitely put some limitations on the "self-governing" aspects of MIS
resource utilization.

The requirements for which RAS was designed to fulfill were given
as:

1. Need to know what resources are being consumed and in what
quantity.

Need to know who consumer are and the quantity of consumption.
Need to identify what the consumption is for.
Need to cross—charge users based upon consumption.

Need to be able to determine what the return on data
processing expenditures is.

VII-\UJN

The ability of RAS to meet the last requirement is quite dubious.

There was no provision made for measuring benefits resulting from the

149

various ongoing applications. How the firm was going to attempt to
measure "return on data processing expenditures" using only cost data was
not made clear. As discussed in Chapter Three (see page 80) there are
several reasons why a ROI in the EDP area is difficult if not impossible
to calculate.

Although the specific charging algorith for ongoing computer
applications was not yet avilable, it was based on the following factors:
Central processor utilization factors.

Peripheral device utilizaton factors.
Memory utilization factors.

Operator intervention factors.
Supplies factors.

. System degradation factors.
. Geographic factors.

VGU§WNH

The term "geographic factors" was not explained, but apparently
had something to do with RJE and time-sharing jobs. Personnel costs were
loaded onto the various resource centers for rate determination, but
occupancy costs were not included. The systems personnel in charge Of
RAS had foresight enough to sense potential conflicts with users, and
one of the current objectives was to "fine tune" the RAS sub-systems to
provide equitable charges. Users were to be sent invoices for a several
month period (for information only) before the forma charging system

affecting budgetary accounts was to begin.

£322.:

Firm F made the initial move toward centralized corporate computer
facilities in 1965. At that time D.P. costs were "pro—rated" out to
users; there was no charge for individual jobs. In 1968 the firm began
charging on the basis of CPU time and continued this method until 1973
when, using the SMF package, they changed to a resources used basis--CPU,

disc, tape, card reader and printer, and RJE.

150

Initially they also charged for core storage. However, these
charges for on—line users took a large jump and fluctuated wildly from
run to run. This was attributed to the variable partitioning of core,
and when it could not be solved in a reasonable manner, "core usage"
was dropped from the charging algorithm.

The following were given as reasons for their confidence in and
use of a charging scheme:

1. It reduces the role the systems personnel must play in

priority setting. [Notice, however, that this role is
‘ngt eliminated.]

2. It encourages cost—benefit analysis at the user level.

3. It reduces the requests for services.

4. It gives a basis for the acquisition Of new resources.

The stated philosophy at firm F is that the charging system
should be based on an efficient shop basis rather than on an expected
level of Operations basis. In other words, users should not be charged
for capacity costs.

However, the actual practices used did not follow the stated
philosophy. Around 1971 the firm charged back about 852 of computer
operations budget to users. By 1975 this percentage had gradually risen
to 902 and the goals was 1002 for 1976. The reason that costs were not
fully charged out between 1971 and 1975 was not that rates were based
on practical capacity; but that rates for the coming year were based on
the previous year's actual cost and usage figures. Because costs were
rising, actual costs were never fully charged out. For 1976, however,
the rates were being based on projected cost estimates for the year
(not actual 1975 figures) and on estimated usage for the year (not actual
usage in 1975). The stated goal was to arrive at equitable rates which

would charge on 100% Of computer operations costs for the year. This is

not charging on the basis of practical capacity.

151

The company was planning on giving premium/discount rates a try.
A number of users have RJE's and many jobs are remote batch jobs that are
run during prime time--the day shift-~when on-line, timesharing users
are the heaviest. In an effort to shift some jobs from prime time to
evening, runs during the second shift will be charged only 80% of "normal"
rates. There are a few users who have jobs so important that they have
been given a priority coding (which they can use at their Option) to
be "next-jOb-in." There will be a lS-ZOZ premium above normal rates on
all jobs using this Option. The premium/discount rates are not based
on any cost factors; they are merly an attempt to spread demand more
evenly.

The general feeling in the EDP area is that that the average

user's level of EDP sophistication is quite low when it comes to under-

standing the "hows, whys,’ and "whats" of the amounts he is charged.
The current method was chosen because it was "more equitable" than its
predecessor; but the users cannot appreciate what that means, nor do
they generally have the ability to respond to changes in the relative
pricing of resources--e.g., a decrease in the cost of disc storage
relative to tape.

Marketing (which was the single largest user) had annual charges
for computer Operations alone of around $300,000 annually, yet the
division seemed to care very little about how it was charged and seldom
had any questions. Accounting and finance, on the other hand, incurred
annual charges for computer operations of around $20,000, but they went
through the charges with a "fine tooth comb" and were continually asking

questions.

It is interesting to compare the firm's willingness to use a

152

premium/discount rate structure vis-a-vis the sophistication of its users.
Firm F used a premium/discount plan hOping that it would persuade é;££El
users to switchto second shift runs. Other case study firms, such as
firm B, (which viewed their users as more sophisticated) shied away from
such a plan, fearing that large numbers of users would request second
shift runs at the discounted rates.

"work centers" in the EDP area besides computer

Three other
applications were charged to users--data entry, data coordinators, and
systems design. Each work center had included in its budget all direct
costs and a pro-rated share of (l) managerial costs (costs of EDP
management), (2) occupancy costs, and (3) corporate overhead.

The data entry "work center" charged users only for jobs
requiring eight hours or more per month to prepare. Small jobs were
not charged. This approach recovered about 752 Of the center's costs.

Data coordinators were a functional group whose main responsibi-
lities lay in file maintenance and liason work. If a member of this
group dealt almost exclusively with one user, then his "cost" was
charged directly to that user. If an employee's time was split, then
his time was identified with individual users when possible. About
75% of the costs in this work center were "recovered" through user charges.

Systems design personnel charge out their time whenever it can be
specifically identified with a given user. A rate was determined individ-

ually as follows:

Total annual costs per man (including prorated costs)
252 x 8 x 702

 

= hourly rate

The 252 is the expected days of work during the year, and eight is the

hours in a work shift. Seventy percent was the estimate of time actually

153

available for user oriented work after educational programs, staff
meetings, coffee breaks, etc. were deducted. This procedure charges
about 65-702 Of systems design costs to users.

One of the problems mentioned at this and other case study firms
was the problem caused by the charging system when a very large integrated
system was under consideration. It became difficult to measure or even
estimate the total benefits accruing to each segment of the firm. Usually
the accounting department was called in to set up a percentage of costs
to be allocated to the resepctive user departments. Often this allocation
was quite arbitrary. As such, it was hard to get all benificiaries to
"sign" for a share of the total cost. In some cases, if all the users
were not willing to accept their share of costs, justification and approval
of the new system was difficult or impossible.

All firms had this problem in some form when the new system
became Operable and charges for computer usage were made to the various
users. However, firms that did not charge users for systems development

costs, at least partially avoided this problem at the design stage.

Support from GAO and CASB Findings

All case study firms visited charged at least some users for
computer Operations, and all but one of the algorithms were based on a
resources used basis (one firm used the information but mainly as an
input into the next year's charges). Both GAO and CASB studies indicate
that the majority of business firms charge users for computer services.
The two studies were structured along different lines, and hence their

findings were summarized differently.

154

Table 4—1. GAO and CASB findings--firms charging users.

GAO data:

Number Z

Cost Assignment--A1ways 18 81
Cost Assignment--Sometimes l 5
Memo Billing 1 5
Users Not Made Aware of Costs ._2 9

22_ 100
CASB data:
Users Charged a Composite Rate 31 8
Users Charged Specific Rates 243 61
Other 2 1
Users Not Charged 117 30

393 100

The GAO study found a considerably higher percentage Of firms
that charged users than did the CASB survey. The explanation for this
difference--and for further differences in the results of the studies--
lies chiefly in the manner in which firms were selected for the two
studies. Of those in the GAO study, most were chosen for their prowess
(or reputation of such) in the EDP area. Neither study attempted to
define a specific population,randomly sample it, and generalize the
results to the population. However, the CASB study was a more
representative selection of firms, chosen from a wide spectrum of
industries and services.

Both studies attempted to classify charging and noncharging
firms by the size of their D.P. budget as shown in Table 4—2 on page 155.
The GAO data are somewhat inconclusive. However, there is a distinct
pattern in the CASB statistics. About three fourths of all firms in
the survey having EDP service centers costing $2 million or more annually
use specific rates (detailed resource usage) to charge users. In the
categories below $2 million, the percentage of firms using specific

rates decreases markedly. However, even in the lower two categories

155

Table 4—2. GAO and CASB findings--D.P. budget vs. charging practices

 

 

 

 

 

 

 

 

 

GAO data:
Data Processing Budget
$l-9 mil 4§10-24 mil $25-50 mil
Assigns Costs NO. _Z_ No. 2 No. Z
Always 10 91 3 50 5 100
Never 1 9 2 33
Sometimes -- -- 1 17
11 100 6 100 5 199
CASB data:
Cost Related to EDP Service Center
Less Than $.l-.5 $.5-2 $2-5 $5-10 Over
$7.1 mil mil mil mil mil $10 mil
Method NO. X No. 2 No. 2 No. 2 NO. 2 No. 2
Users Not
Charged 6 46 26 50 41 36 22 21 12 22 8 19
Composite
Rate 4 31 8 15 8 7 6 5 2 4 3 7
Specific
Rates 3 23 18 35 66 57 78 74 4O 74 32 74

 

 

 

 

13 100 52 100 115 100 106 100 55 100 43 100

approximately half the firms used some type of charging scheme. Although
Turney's ("few companies") and Heitger's ("most firms") comments mentioned
on pages 97-98 regarding the use Of transfer pricing in the EDP area
were made in 1972, Heitger's conclusion rings most true today. One reason
for the prevalent use of charging systems may be found in Gibson and
Nolan's stage hypothesis. [58] Over the last several years many firms
have no doubt passed from stages one and two to stages three and four—-
where control mechanisms such as charging systems begin to be used.

Both GAO and CASB studies listed various consideratons in the
approach used in charging users. The GAO data consists of two parts,
the first being oriented more to the objectives of charging users and

the second to the rationale or criteria used in designing a specific algo-

rithm,

156

Table 4-3. GAO findings--Objectives of and criteria used in designing
charging systems

GAO data: Objectives in charging users

Frequency
Make users more cost conscious 18
Improve D.P. management control over cost 16
Reduce user demand for services 8
Assist D.P. management in controlling
service demands 5

(firms could indicate more than one objective)

GAO data: Criteria for charging scheme
Most equitable 18
Consistent ll
Understandable 10
Simple and easy to administer 9
Acceptable 5
Relevant 4
Legal 1
Other 3

Tflue difference between the last two factors in the first set of data is
INJC clearly distinguishable, yet only one firm in the study marked both.
Likewise, some of the criteria--e.g., "acceptable" and "relevant"--in the
secorui set of data seem too vague to have any Operating interpretability.
Although the CASB entitled their summarized results "Considerations
in the Approach Used in Charging Users," their data are more a statement
Of what is being done, not why. There are some apparent discrepancies
in Clue.Casb data presented in Table 4-4 on page 157. 325 firms
8nPPO‘rted the third statement, yet only 274 firms in the study supposedly
charged users. (See page 154.)
The GAO summary of the charging methods being used is presented
in Table 4-5 on page 157. The difference between standard full cost
and standard operating cost was not identified. Possibly the former
c“mains prorated costs, such as occupancy costs, whereas the latter does

not. 'There is some reason to doubt the stated results when they are

157

Table 4-4. CASB findings--considerations in charging users

CASB data:

Frequency
Charges are made on the basis of data that are
easily gathered 147
Charges are made on the basis of methOd(s) and data
that are easily understood by users 130
Charges are made on the basis of specific hardware
components employed in processing each job 325

Charges are made in a consistent manner such that a

user can recognize that the charge to him/her

reflects his/her own use of data processing

resources and does not reflect a fluctuation

in the rate(s), volume of activity within the

EDP service center, job mix, and the like 211
Charges are made, where components are substitut-

able, on the basis of a schedule that would

influence users' choice Of components (which,

in turn, would promote balance in resource

utilization in the EDP service center) 65
Charges are made, where resource utilization

fluctuates between excess and inadequate

capacity within a short time period, on the

basis of a schedule that would influence

users' timing of job submission (which, in

turn, would promote evenness in resource

utilization in the EDP service center) 60
Other 21

Table 4—5. GAO findings--charging methods used

 

GAO data: Methods used
EL- Z
Standard Full Cost 9 42
Standard Marginal Cost 5 26
Standard Operating Cost 3 16
Actual Full Cost _j;_ __11
a .129

 

Table 4-6. GAO findings-~types of cost accounting systems used

 

GAO data:
Type of cost accounting system No. Z
Job Order 10 46
Standard Cost 4 18
Process 2 9
Installations Not Using Cost

Accounting System 6 27

22 100

158

compared with other GAO data. The GAO study asked the firms to identify
the type of cost accounting system used. The results obtained are
presented in Table 4-6 on page 157. Seventeen firms in Table 4-5
reported using standard cost charging methods,yet only four firms in
Table 4-6 reported having "standard cost" cost accounting systems. The
types of cost accounting systems used in the GAO questionnaires and
finds were defined in either place. "Standard"_g£ "actual" costs may
be used in conjunction with either job order or process cost accounting
systems--the two groups 353 Egg mutually exclusive. It is difficult to
imagine how a true process cost accounting system could be used in the
EDP area except possibly in connection with a project reporting system
for the systems design area. Finally, the stated result that six firms
or twenty-seven percent of the total were 595 using any type of cost
accounting system seems most dubious. How could 19 of the 22 firms
regularly charge users for services without having some type of cost
accounting system?

The finding in Table 4-5 regarding the use of standard marginal
cost seems questionable, particularly inlight of CASB findings, if it
is really a true marginal cost being used. There is some reason to
believe that firms using "standard marginal cost" charging methods
excluded only hardware/software costs, but included relatively fixed
personnel costs. If both were excluded, in a true marginal cost situation,
the remaining variable costs would be so small as to be hardly worth the
time and effort to charge users. It is doubtful that the objectives
given on page 156 would be reached. Even Turney [149] did not suggest

that only marginal costs be charged to users. The CASB study did find

159

that some firms did not include either hardware or software costs in

their charging rates, but these firms were very few.

Table 4-7. CASB findings--inclusion of hardware/software costs in
charging rates

 

NO. Z
Rate includes equipment rental or depreciation 4O 10
Rates includes equipment rental or depreciation
22d systems software rental or amortization 223 57
Rate includes neither 13 3
Firms not charging users .111 ._39
323 100

One one of the case study firms did not use standard cost rates.
This was firm D which allocated the undistributed costs to users at the
end of each month on the basis of their previous charges for the month.
The other five firms left the undistributed amount in EDP overhead.

The usuable responses from the GAO and CASB findings regarding
the disposition of variances are presented below. In a way the CASB
analysis is somewhat confusing and vague because the first three

alternatives are virtually the same. However, if the third alternative

Table 4-8. GAO 8 CASB findings--disposition of variances

GAO data:
31.0.; A
Left undistributed 7 41
Used to adjust rates 0 0
Allocated to users 6 35
Some combination of the three __43 __§4
.11 .192
CASB data:
Transferred to EDP service center overhead 117 46
Transferred to general corporate overhead 28 11
Transferred to income statement 67 26
Carried forward as adjustment to rates of
subsequent period _44

160

means transferred or allocated to the income statements (or cost
reports) of users, then the CASB data begins to look comparable to the
GAO data.

Only one case study firm used a premium/discount plan in
conjunction with its charging system, although plans were under way to
try this in another firm. The GAO reported eight out Of twenty-two
firms using priority pricing plans. The following circumstances gave
rise to as premium or discount in the rates.

Table 4-9. GAO & CASB findings--circumstances resulting in premium]
discount rates

GAO data:
so;
Job submitted at a specific time of day 3
Job submitted to be processed before a
specific deadline 2
Job has a priority coding 4
Other 3
CASB data:
Job submitted at a specific time Of day 48
Job submitted to be processed before a
specific deadline 13
Job has a priority coding 70
Other 40

The GAO responses1h>not total eight because some firms responded to
more than one reason. The data indicates that considerably more firms
appear to use a priority pricing plan than the one out of six ratio

found in the case study firms.

Chapter Summary

 

This chapter has outlined the findings from the case studies in
the seven areas of the model described in the previous chapter. Where

possible the findings have been supplemented with data from GAO and

161

CASB studies. Although it is difficult to compare the results from a
sample of six case study firms with that from a sample of nearly 400 as
in the CASB survey, the case study firms appear to have better than
average cost control practices. For example,_all case study firms used
some type of charging scheme.
The one area where a significant difference was noted was in
the use of premium/discount rates. However, as noted earlier in this
chapter, several case study firms di not use premium/discount rates
because their users were, in effect, "too sophisticated"--too ready
and willing to adjust their timing to changing rates.
Overall, the EDP cost accounting systems in the case study
firms were neither amazingly perfect nor terribly bad, but rather quite
adequate and in some areas very good. The very poor cost accounting
systems that Turney suggested existsed, simply were not found. The
case study firms' efforts lend credence to the idea that cost accounting
systems are of considerable use in controlling EDP costs and operations.
In the next chapter the findings of this chapter will be
summarized and the model components of chapter three will be examined
in light of these findings. Fruitful areas for further research will be

delineated.

CHAPTER V

CONCLUSIONS AND NEW DIRECTIONS

Introduction

 

The purpose of this chapter is to summarize the case study
findings. These will be examined in light 0f the CV0 PI‘OPOSitiODS
suggested in Chapter One:

(1) The presence or nonpresence of the model EDP cost accounting
system as a whole system in existing organization practices.

(2) The presence and need for each of the individual model
components in a reasonably complete cost control system.

Suggested improvements in EDP cost control practices will be made. New

areas for further research will be proposed.

Case Study Conclusions

 

It is difficult to completely quantify case study data. Any
attempt to do so will no doubt omit much information about the firm(s)
which is needed to understand more fully the practices used and the
reaons behind them. However, in an attempt to summarize the previous
chapter and provide a basis for discussing various conclusions, the
figure on page 163 is presented. Various aspects of each case study
firm's control system are ranked on a simple scale. The rankings are
based on the expectations of the observer after surveying the relevant

EDP and cost accounting literature and are in terms of the proposed cost

162

163

ouoouanaoo
Huauoouooao

oosuuamaoo
couscous

oonuaaasoo
unannou-

macaquesoo
ow uoa

moonwaaaoo
Ououoooa

monouaaaou
as non

uncouuano
unwouv
unnuahm

ouaoaanaou
ouuuoooa

ouuoqansou
anauuuuaona

oonuaunlou
ousuovoa

sausage-co
ouuuovos

ousuaaafioo
Huauaduonau

oonoaanaoo
aoauouuoouu

unouuuunu
unowuuuono
ususnaoo

uuooaansoo
«unusual

ooouaamaou
ad uoa

soundneaou
Hosanna-nan

massage-co
monsoon-

ooouaaaaou
Ououuvoa

ooauaaqaou
ad and

 

uuaoo
wouuaouumafi
.uuou: wuaxuouh

moonwaeaou
«unwoooa

moonwaoaou
ououovos

oooodagloo
ow uon

moonwagsoo
cannons!

«unusaaaoo
as you

ououwaoloo
Hoauouuooao

umadfldam

Huuuoonau
undounuooq

coauaaoaou
ououovoa

oosoaueaou
uuuuovoa

ouooaaoaoo
Huwusuuuoau

oussaHAIou
Huaunuuuosu

ooouwHoaou
ouuuovoa

moooaaglou
Huauouuuoao

ufigflH—dflsa

huw>wuuu
no usouoo

.mmcfiooam xenon owmo mo humaaom

ouaowdoaoo
Ouuuovoa

mucosansou
«aqueous

sausage-cu
ouuuuvoa

«unquaosoo
ououovoa

ouomwaQBOO
ououovoa

ooou«HQEOO
ououuooa

nonnaaosoo
Hoauouuooau

ounouansoo
Huauouuuoso

ooouqnnaou
unannouoaao

oonueaalou
noduaouoosu

ousuaaaaou
Hoaunuuuoso

moonwaeaou
Huauououoau

 

 

souowo

mduuuoaou vow

waauomvan

uuouoou uooo
use uuuuooo
huwawowoooouum

moonwaaaou
couscous

moonwamaoo
nauseous

oooouualou
uuuuuooa

ooaounasoo
cassava-

ououuanaou
ouuwovoa

oouuuunaou
cassava-

nOAuouuuuuo vac
soauooaaou unoo
no HO>UA

.Hum magmas

fihfim

N0.H

no

RN.

«0.

NH

Nn.a

noun.

«0 N u an
ounce man

164

control model. At the end of the discussion of each model component in
Chapter Three the factors used to evaluate the case study firms in that
area were presented. These factors are summarized in Figure 5-2 on pages
165-166.

There is no single ideal measurement of the effectiveness of an
EDP cost control system. The EDP costs as a percentage of sales statistics
in Figure 5-1 are given as a very rough "ball-park" measurement. These
would be expected to vary considerably between industries. For example,
the firms with the two highest percentages are both in the pharmaceutical/
chemical industry. Both incur substantial data processing costs in their
research efforts, and each has a separate data center dedicated to research
efforts. In firm P, if research D.P. costs are omitted, the percentage
Of EDP costs to sales drops to .8Z. The two firms with the loweset
percentages were both in the manufacturing and materials conversion
classification. Such interindustry idfferences are to be expected and
somewhat lessen the comparability of the EDP cost to sales ratios.

Firm E--which had the highest percentage—-experienced several
problems which partially explain the high percentage figure.

(1) D.P. efforts in the research division.

(2) A "niggardly" attitude toward the EDP area on the part of
Op management prior to 1971. The firm is still playing "catch-up ball"
in the EDP area. For example, it was the last of the case study firms to
develop a detailed charging algorithm on a resources used basis.

(3) A major change of hardware in midstream. About 1970, the
firm changed form all Univa equipment to IBM. All programs were left in
Autocoder and were run using an emulator. These programs are gradually
being converted to COBOL.an expensive task expected to take several more

years.

165

1. Level of Cost Collection and Aggregation

l.

2.

The degree of classification and detail by descriptive
e1ement--particu1arly in the areas of personnel and
equipment costs.

The degree of classification and detail by specific
program or project--the existence of a project reporting
system and detailed charges by individual programs.

11. Responsibility Centers and Cost Centers

Agreement Of responsibility centers with formal
organization chart.

Homogeneous activites within cost centers.

Responsibility centers used--not profit or investment
centers.

Attempt made to separate controllable from noncontrollable
costs.

111. Budgeting and Reporting System

IV. Output

1.

The existence of an annual budgeting procedure.

The use of participative budgeting.

Monthly cost/budgeting reports with monthly and
year-to—date cost/budget comparisons.

The degree to which managers must "live within" their
budgeted amounts.

The gneration of data to allow continuous cost/benefit
analysis for ongoing and projected systems.

or Activity Measurement

The existence Of system software or other method for
measuring individual resource utilization in the hardware
area.

The use of the utilization data to manage and control
computer operations, not just to provide input into a
charging scheme.

The existence of work measurement and reporting systems
for the data entry area.

A reporting system for human output in the systems design
and programming area.

V. Long-range Financial Planning

Figure 5-2.

The existence of reasonably complete long-range goals and
plans at the corporate level.

The existence of long-range goals and plans for future
projects in the EDP area.

The existence of long-range plans for personnel and hardware,
and projected dollar costs.

Factors used to evaluate case study firms in model component
areas. (Continued on page 166)

166

Hardware acquisition decisions made on a basis consistent
with other capital budgeting decisions in the firm.

Use of a life cycle, cost/benefit approach to system
project decisions.

VI. Tracking Users' EDP-related Costs

Annual summary of data processing costs for the firm

is prepared.

EDP personnel approve or are Otherwise aware Of hardware
acquistions by other functional areas in the firm.

An effort is made to collect and aggregate EDP—related
costs in user areas and report these on a systematic basis.

VII. Charging System for Users

1.

3.

Figure 5-2.

A charge to users is made for EDP resources used in the

computer operations area.

a. The charging system is based on standard full costs
and practical capacity--thus no attempt is made to
charge out all of computer Operations costs.

b. Charges are based on a detailed measurement of EDP
resources consumed.

c. Charging rates are not changed more Often than once a
year except in unusual circumstances.

d. Consideration is given to priority pricing schemes.

A charge to users is made for systems design and programming

services.

a. A lump—sum amount is negotiated for larger jobs.

b. Not attempt made to charge out all systems design
costs--e.g., no charges made for preliminary and
feasibility studies.

c. Charges are made for user-initiated maintenance
programming work.

d. Satisfactory resolving of problems in charging
multiple sponsors of large-scale, integrated systems.

A mechanism exists for resolving user complaints in both

charging areas.

Factors used to evaluate case study firms in model component
areas. (Concluded)

167

The EDP cost/sales percentages from the GAO data are given below.

Table 5-1. GAO findings--EDP costs as a percentage of sales

GAO data:
as; .1.
Less than 1% 6 31
l-lOZ 10 53
ll-24Z 2 11
Over 25Z ‘__l ___5
_1_9_ 199.

The case study firms compare favorably with the GAO firms--which were
also chosen partly for their prowess in the EDP control area. It would
have been more helpful if the GAO summary had used a smaller classification
group than 1-10Z which provides a 10 to 1 leeway for the EDP cost percen-
tages of firms in this group.

In terms of Gibson and Nolan's four stage model of EDP growth
(see pages 22—24) all but one of the case study firms appeared to be in
the latter part Of the third stage ("formalization") or the early part
of the fourth stage ("maturity"). Sophisticated applications were common,
and the control techniques mentioned in Gibson and Nolan's work were well
developed.

Firm E, however, was just entering or partially into the third
stage. The Resource Accounting System (RAS) was under development
including a more sophisticated charging algorithm, and various controls
were beginning to proliferate.

Thus, it must be realized that the conclusions drawn from the
case study firms may not apply to firms in Gibson and Nolan's stage one
and two. As mentioned previously, however, part of the reason for
proposing the model was to allow and encourage firms in stages one and
two to control EDP costs in those stages and thereby promote a more

orderly growth pattern.

168

Presence of the Entire Model

 

The proposed EDP cost control model as a whole was found to
exist in a state of "moderate compliance" or better in Firm F. (See
page 163.) Although most of the model components were found to be
only in "moderate compliance", rather than in "substantial compliance",
all were present in this firm.

Most of the other firms were very close to having the complete
model present in their existing practices. Firms B and D lacked only
a more adequate long-range financial planning system. Firm C did not
charge users for systems development costs. Firm E did not attempt
to track user's EDP-related costs, but this unconcern was prompted by and
partially compensated by the fact that all hardware aquisitions within
the firm had to be approved by the MIS manager.

Thus the model, which was based chiefly on traditional cost
accounting literature, was found in a reasonably complete state in one

firm, and nearly so in the other firms.

Individual Model Components

The first three model components are very closely related. The
first component, the level of cost collection and aggregation, was found
to be in a "moderate compliance" state in all case study finms. Adequate
detail was generally present except for the high percentage of costs
carried in two accounts-~hardware costs and personnel costs. MOst firms
used essentially the same chart of accounts in the EDP area as they did
in the other areas of the firm.

Weaknesses in the initial classification of accounts were at

least partially compensated by the relative strength of the responsibility

169

accounting systems observed. All firms went beyond the expected minimal
subdivision of "computer operations" and "systems design." Since the
responsibility accounting systems directly paralleled the formal
organization charts, what we are really saying is that the firms had
well-organized EDP areas. The firms were somewhat weaker in the use and
understanding of the idea of "controllable costs." Some firms attempted
to subdivide "controllable" and "prorated" costs, yet they did not

adhere to the concept of committed costs being "uncontrollable," at least
in the short run at a given responsibility level. This particularly true
in the computer operations area.

All firms had standard budgeting and reporting practices which
were part of an overall corporate procedure. Monthly cost reports were
used in all firms, and were very similar in the types of costs and
budgeted amounts reported. Flexible budgeting was found to be in use in
one firm, but was a highly inappropriate technique of which the EDP
managers had taken advantage. Although monthly budget/cost reports were
used in both the systems design and computer operations areas, the
reports had greater usage as a control device in the computer operations
area.

All firms had a project reporting system which was the chief
control device in the systems design area. A potential problem of the
project reporting system was the practice of allowing system analysts to
estimate the percentage Of completion on a given project. Projects
tended to quickly "reach" a high level of completion where they remained
unfinished for some time. But on the whole, the case study firms deemed

the project reporting system a very useful and necessary control device.

170

Because of the widespread use of the IBM SMF package among case
study firms, output or activity measurement in computer operations areas
were nearly identical. The usage made of these measurement varied
considerably however. Some firms viewed the measurements as merely
inputs into the charging algorithm. But other firms also used the soft-
ware package to monitor the utilization of various decentralized data
centers throughout the firm or to increase the efficiency of a specific
configuration. The four firms given a "substantial compliance" rating
in this area were attempting to use SMF and similar adata to manage and
control the computer Operations area, not just as input for a charging
scheme.

All case study firms had fairly sophisticated work measurement
and reporting systems for the data entry area. But the productivity
of analysts and programmers was more difficult to measure. Typically
work measurement in this area was limited to information contained in a
project reporting system--the number of hours an individual spent on
various assignments or projects. Productivity in this area is difficult
if not impossible to quantify. Although the project reporting system was
of some help in evaluating an individual's efforts, this was usually done
more on the basis of a personal knowledge and observation of one's work.

The next two areas-~long-range financial planning and tracking
users' EDP-related costs-~were the model components that had the weakest
support from the case studies. This result was somewhat expected regarding
the latter area, but came as a surprise in the first area. Especially
surprising were the results from the two case study firms given "not in

compliance" ratings in this first area. Neither firm claimed to have

171

much in the way of formal plans beyond one year, and in firm D hardware
expenditures--even long-term leases-—were approved via the Operations
budget.

Firm A bad the most complete long-range planning system, but
there was little in the way of concrete, realized benefits that could
be pointed out. Evidently, five years later, Sollenberger's conclusion
regarding long-range planning in the systems area still holds true and
probably can be extended to include computer operations.

Planning beyond one year has yet to contribute large dividends

to systems development. Activity was still too much crisis

oriented rather than plan oriented. The idea of a master plan
was either never developed or was seriously downgraded as
impractical or not pertinent in the firms studied. Five-year
plans were common, but their values were limited primarily to
manpower budgeting and broad systems concept development.

[142, pp. 10-11]

Note that Sollenberger is not saying that long-range planning is useless,
but that it has not yet paid "large dividends." This could be because:

(1) Long-range planning is inappropriate in the EDP area because
of rapid technological changes and other reasons; or

(2) Long-range planning has not been properly employed or
executed in the EDP area.

One further conclusion can be noted. Long-range planning in the
EDP area is impossible or useless if corporate goals, objectives, and
strategies are not outlined first. The EDP area is a service center
which contributes to corporate goals. It should not attempt to chart
its own course unrestrained. Over five years ago Clarke predicted:

Top management realizes the importance of long range planning

in determining the future position of the company. However,

this understanding has not included the EDP function. Many top
managers today look upon EDP as a necessary evil and fail to
consider the impact of the large investment in data processing

men and equipment. . . . Management is going to ask the EDP
organization for the long range plans and how they will use

172

the computer. . . . The days of plodding along aimlessly

are almost over. Today's top management has had some

exposure to systems and it is going to expect constructive,

definitive plans to meet the informational needs of the

company. To effectively benefit from a resource one must

plan for its use. [26, p. 8]

This expected emphasis on long-range planning was not found in the
case study firms. Rather Clarke's comment made in 1971 seemed quite
apropos: "Few companies have been or are involved now in future
systems planning." [26, p. 8]

The idea of tracking users' EDP-related costs is a novel one
suggested by certain incidents reported in the EDP literature. (See
pages 81—82.) Firm D included a very complete analysis of such costs
in its annual summary of EDP costs. Two firms, A and E, made no attempt
to summarize on an annual basis the EDP costs for the firm. The remaining
firms prepared annual EDP cost summaries with some minor efforts to include
users' EDP-related costs. In general, EDP personnel were more responsive
to the concept of tracking users' EDP-related costs than to long-range
financial planning, and considered the former to be a useful idea.

The firm which received the "substantial compliance" rating in
this area had perhaps the best overall record of cost control. Not only
did it have the lowest EDP costs to sales ratio, but EDP costs had
remained virtually constant for a five year period. Most of this cost
control had been achieved by a significant reduction in personnel. In
1971, 200 people were employed in the EDP area; nearly five years later
this figure had been reduced to 144, with much of the reductions coming
in the systems and programming areas. The firm appeared reasonably
satisfied with the status of its ongoing applications--most users desiring

fast turnaround were already online—-and much of the programming efforts

were in the area of maintenance programming and small jobs.

173

Five case study firms charged users on a resource usage basis for
computer operations costs. Three firms were given "substantial compliance"
ratings because their charging system approximated standard full costs
based on practical capacity. Costs not charged out were absorbed as
part of EDP overhead. The other three firms were considered in "moderate
compliance" because (1) capacity was based on actual usage (firm D) or
estimated actual usage (firm F) or (2) the prior year's actual usage
determined the flat rate charge in the present year (firm C).

Three case study firms charged users for systems development costs.
Of these, two firms attempted to charge out all systems development costs.
The third--rated in "substantial compliance"--charged out approximately
7OZ of such costs. Another firm used a memo billing procedure for
systems work and the remaining two did not charge at all.

Some reasons previously advanced by various firms for not charging
users for systems work have included:

(1) There are different levels of competence in systems analysts.
A flat rate hourly charge would cause "unfair" charging because some jobs
would get the "good analysts" and some would get the "bad". The alterna-
tive to this is placing programmers and analysts into different productivity
classes and establishing a separate rate for each class--a difficult and
costly procedure.

(2) Analysts may have trouble accurately keeping track of their
time if they are currently active on several jobs.

(3) It is hard to stimulate change if users are charged for
systems development.

(4) The EDP area (including systems development) can most

effectively control its own costs from within.

174

(5) The cost of creating a charging system and a group in charge
of this area may outweigh the benefits received.

(6) Users are reluctant to support large scale integrated systems
where all benefits do not accrue to their area.

The first argument can be overcome by establishing a lump-sum,
contractual charge for a project regardless of actual time spent (unless
delays are caused by the user). The second argument is no longer valid
because firms having a project reporting system already collect this kind
of data. The third argument is difficult to follow. The systems area
should not stimulate change for "change's sake" anyway. The fourth
argument is debatable. It can be argued that a coordinated approach--
charging users 39d stressing internal cost control--is more desirable than
concentrating on either internal or external control to the exclusion of
the other.

The final two arguments bear more weight. The fifth argument is
definitely a possibility, particularly in light of the difficulties of
measuring the cost of a charging system, much less the benefits resulting
from such. Only one case study firm could attempt to set a cost figure
on its entire charging system. None could determine the incremental
costs of charging users for systems work.

Finally, if top management desired to implement large-scale,
integrated systems, it could ask the systems area not to charge for
develOpment of such systems where multiple users benefit. This might
be preferable to relying on arbitrary cost allocations for sponsoring
users. If users were charged for other (nonintegrated) systems, they
would be motivated to request integrated systems with other users--a

"free" good--thus fulfilling the desires or objectives of top management.

175

In summary, the model found considerable support on a component
by component basis. Charging for systems design efforts was not as
strongly supported as charging for computer operations, but the general
trend was toward, rather than away from charges in this area. For
example, the firm using memo billing planned to formalize those charges
in the budgetary accounts.

Long-range financial planning and tracking users' EDP—related
costs received the least support, but firms were generally more favorable
to the latter idea. Thus long-range financial planning is the most dubious
of the model components. Though it found moderate support, firms were
unsure of its eventual benefits. In addition, it was relatively costly
to perform, not necessarily in an incremental dollar sense, but in taking
analysts and managers away from more pressing (and presumably immediately
productive) duties.

In the final analysis, it must be remembered that an adequate
EDP cost system does not control costs. People control costs. Their job
is made easier by the presence of adequate information and means of control.
The best of cost accounting systems will do little good unless management--
both top level and within the EDP area--is actively involved in the cost
control process. An adequate EDP cost accounting system cannot replace

proven managerial skill and involvement, but it may encourage both.

 

Spggested Improvements in Observed EDP Cost Accounting_Practices
Various improvements in EDP cost accounting practices have been
implied in the last two chatpers. Some of the more important ones will
be summarized at this point.
(1) Examine carefully the chart of accounts used to control the

EDP area. Are large percentages of costs grouped in one or two accounts?

176

Is the set of accounts exactly the same as used in other areas of the
firm? Does cost data for charging algorithms and other purposes have to
come from supplementary records and schedules? Perhaps a restructuring

of the chart of accounts for the EDP area will help meet this area's needs
better and improve control.

(2) Examine the use made of SMF or similar data if available.

If the sole usage of such data is merely as an input into the charging
algorithm and a means of determining rates for the coming year, then much
of the potential benefits from the SMF system are not being realized.
Investigatetfluauses of this data in the areas of workload analysis and
classification, standards establishment and enforcement, and identification
of resource wastage and other factors which may degrade system performance.

(3) Consider the trade-Offs made in cutting EDP costs and
increasing users' EDP-related costs. This may entail tracking the
latter and preparing at least annually an EDP cost summary which includes
such.

(4) Consider the behavioral effects of the current charging
practices in both the systems and computer operations area. If systems
services are a "free good", are users motivated to control their use of
this resource? Would contractual, fixed-amount charges encourage analysts
and programmers to meet time and cost estimates for projects under
development? In what ways do users react to the charging system for
computer operations and changes made therein? Do they understand clearly
the basis on which they are charged? Are they informed enough to budget
their own EDP resource usage and costs for the coming year?

In most firms, EDP personnel had a relatively free hand in

designing the charging algorithms used. Although nothing is wrong with

177

practice--they are the experts in this area--tOp management should get
involved to the extent of setting goals and criteria and determining if
these are met by the practices in use. This was the thrust of Heitger's
dissertation, and his general conclusion--use standard cost rates based
on practical capacity--holds true today.

Finally in situations where users are charged, ensure that there
is a coordinated approach with equal emphasis on cost control in both
users' and EDP budgets.

(5) Have top level management examine closely the need for
long-range planning at both the corporate and EDP levels. The need for
formal plans in this area may vary widely between industries and firms.
However, the area should not be ignored. Top level management and EDP
managers must be satisfied that orderly, long-term growth and development
in this area is not impaired by current actions and plans. For example,
the sudden change of hardware in firm E might not have been necessary if

long-term considerations had been previously scrutinized.

New Directions in Future Research

 

A number of questions were raised by this study which seem
fruitful ground for further research. Such topics include:

(1) The usefulness of long-range planning in the EDP area.
Perhaps one way to approach this problem is to use a case study approach
over an extended period of time.

(2) The effect of industry practices on a firm's EDP cost control
techniques and chargeout practices. CASB data indicate considerable
variation in charging practices between industries. Only ZOZ of the
public utilities did Egg charge internal users for EDP services (sample

size a 25), while 48Z of financial institutions did not charge users

178

(sample size = 44). Possibly industry differences extend to other
cost control techniques, although this was not detected in the case
study firms because of their limited number.

(3) User reaction to complex chargeout algorithms. Is
modification of users' behavior made easier with simpler, possibly
transaction-based billing practices? Also, are EDP activities which
are charged out subject to "greater control" than those that are not
charged to users?

(4) The need for and most effective structure of a priority
setting device. It is doubtful whether a charging system will solve
all priority problems in the systems development or computer operations
areas, especially if these costs are not controlled in users' budgets.
Often the setting of priorities is left largely in the hands of EDP
managers who cannot hope to please everyone. This was essentially the
problem that led to the decentralization of the corporate data center
in firm C several years ago. For those firms having centralized data
centers the issue is particularly relevant. Various forms and levels of
committees and other priority setting devices are in use, and all are
probably not equally effective.

(5) EDP capacity measurement. Heitger has previously cited this
as an area for further research. [71, pp. 127—128] Practical capacity
was the basis suggested in Chapter Three for determination of charging
rates, and some firms were attempting to use this basis. However, as
Heitger puts it, "a great deal of subjective evaluation was necessary in
arriving at practical capacity." [71, p. 122] In one of the case study
firms a lengthy two-page worksheet was used to convert maximum available

capacity to practical usable capacity. Subtractions for unstaffed

179

emergency reserve hours, hardware downtime, and software downtime were

only the beginning of a long list of necessary "subjective evaluations."

Chapter Summapy

 

This chapter summarized the findings (presented in Chapter Four)
in terms of support or nonsupport of the model EDP cost accounting system
as a whole and by its various components. The model was found to be
reasonably complete in one firm and nearly so in the others. Model
components received considerable support on an individual basis. Long-
ragne financial planning and tracking users' EDP-related costs received
the least support, although the latter was gnerally received with more
favor than the former.

Suggestions were made for improving EDP cost control practices
based on the literature review, model construction, and case study

findings. Finally, new directions for further research were presented.

B I BL IOGRAPHY

10.

11.

12.

13.

BIBLIOGRAPHY

Abdel-khalik, A. Rashad. "On Gordon's Model of Transfer-Pricing

System." Accounting Review, XLVI (October, 1971)

, 783-787.

Abdel-khalik, A. Rashad, and Lusk, Edward J. "Transfer Pricing--
A Synthesis." Accounting Review, XLIX (January, 1974),

8-230

Albrecht, Leon K. Organization and Management of Information
Processing Systems. New York: Macmillan Company, 1973.

 

Ameiss, Albert P.; Ulett, George A.; Thompson, Warren A.;

and Wood,

Harry E. "Pay for Service Hospital Plan." Journal of

y§ystems Management, XXIV (May, 1973), 8-13.

 

American Accounting Association, Committee on Internal Measurement
and Reporting. "Report of the Committee on Internal
Measurement and Reporting." Accounting Review, Supplement

to Vol. XLVIII (1973), 209-241.

American Management Association. Cost Control and the Supervisor.

New York: American Management Association, 1966.

Anderson, John J. "Direct Chargeout of Information Systems Services

Costs?" Management Adviser, XI (March, 1974), 27-33.

 

Arpan, Jeffrey S. International Intracorporate Pricing.
Praeger Publishers, 1972.

New York:

Arrow, Kenneth J. "Control in Large Organizations." Managgment

Science, X (April, 1964), 397-408.

"Asking the Computer for Separate Checks." Management Review, LXII

(November, 1973), 32+.

Axelrod, Clive Warren. "The Allocation of Computing Resources in
Organizations with Semi-automonous Users." Unpublished

Ph.D. dissertation, Cornell University, 1971.

Benston, George J. "Multiple Regression Analysis of Cost Behavior."

Accounting Review, XLI (October, 1966), 657-672.

 

Bierman, Harold, Jr. "Pricing Intracompany Transfers."
Review, XXXIV (July, 1959), 429-432.

180

Accounting

 

/

/

V/f

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

181

Black, Donald L. "Controlling A Computer System.‘ Datamation,

XX (April, 1974), 98—99+.

 

Blevins, Herbert H.; King, Alfred M.; and Miller, Fred W. "Managing
MIS: Controlling Costs." Financial Executive, XLIII
(June, 1974), 26-36+.

Bocchino, William A. Management Information Systems: Tools and
Techniques. Englewood Cliffs, N.J.: Prentice-Hall, Inc.,
1972.

 

Borovits, Israel. "The Pricing of Computer Services." Data p/“
Processing. XVI (May, 1974), 160-162.

 

Brabb, George., and Grosso, Donald. "Making EDP Effective and
Efficient." Journal of Systems Management, XXXIII
(November, 1972), 40-43.

Brink, Victor 2. Computers and Management. Englewood Cliffs, N.J.:
Prentice-Hall, Inc., 1971.

 

Burlingame, John F. "Information Technology and Decentralization."
Harvard Business Review, XXXIX (November, 1961), 121-126.

 

Cardenas, Alfonso F. "Data Entry: A Cost Giant." Journal of
Systems Management, XXIV (May, 1973), 35-42.

 

"Charge-Back, Cost Center EDP Budgeting Debated at New York Meeting."
Management Adviser, XI (January, 1974), 11-12.

 

Charman, John. "Effective Project Costing and Killing Techniques."
Information Systems Administration. By F. Warren McFarlan,
Richard L. Nolan, and David P. Norton. New York: Holt,
Rinehart, and Winston, Inc., 1973.

 

Churchill, Neil C.; Kempster, John H.; and Uretsky, Myron. Computer-
based Information Systems for Management: A Survgy. New
York: National Association of Accountants, 1969.

Clarke, Lawrence J. "Profitability through an EDP Subsidiary."
Journal of Systems Management, XXI (September, 1970), 34-36.

 

. "Why Plan for Systems Development?" Journal of Systems
Management, XXII (June, 1971), 8—11.

 

Coe, Ted L. "Allocating the Corporate Information Processing Resource."
Journal of Systems Managament, XXV (August, 1974), 18-22.

 

Coleman, Raymond J., and Riley, M. J. MIS: Management Dimensions.
San Francisco: Holden-Day, Inc., 1973.

Cook, Paul W., Jr. "Decentralization and the Transfer Pricing
Problem." Journal of Business, XXVIII (April, 1955), 87-94.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

182

. "New Techniques for Intracompany Pricing." Harvard Business

 

Review, XXXV (July, 1957), 74-80.

Copley, Lewis L., Jr. "Still Need Business Knowledge in EDP." Data

Management, XIII (July, 1975), 5.

 

Cost Accounting Standards Board. Prgggess Report to the Coagress 1974.

 

Washington, D.C., 1974.

. Progress Report to the Coagress 1975. Washington, D.C.,

 

1975.

Crompton, Walter H. "Transfer Pricing: A Proposal." Management

 

Accounting, LIII (April, 1972), 46-48.

 

Currah, Brian and Morino, Mario. "Uses for OS SMF Data."

Applied Technology, Inc., Arlington, Va., 1972. (Mimeographed.)

PACE

Dascher, Paul E. "Transfer Pricing--Some Behavioral Observations."

Managerial Planning, XX (November, 1972), 17-21.

 

Dean, Joel. "Decentralization and Intracompany Pricing."
Business Review, XXXIII (July, 1955), 65-74.

 

Harvard

Dearden, John. "Computers and Profit Centers." The Impact of

 

Computers on Management. Edited by Charles A. Myers.

Cambridge, Mass.: The M.I.T. Press, 1967.

. "Interdivisional Pricing." Harvard Business Review,

 

XXXVIII (January, 1960), 117-125.

, and Nolan, Richard L. "How to Control the Computer Resource."

Harvard Business Review, LI (November, 1973), 68-78.

De Brander, B.; Deschoolmeester, D.; Leyder, R.; and Vanlommel, E.
"The Effect of Task, Volume and Complexity upon Computer
Use." Journal of Business, XLV (June, 1972), 56-84.

DiGuglielmo, Anthony E. "Computer Job Costing." Management Controls,

 

XVIII (March, 1971), 51-56.

"Two Approaches to Calculating Cost of Simultaneous

Computer Jobs." Banking, XXXII (April, 1971), 12+.

Ditri, Arnold; Shaw, John C.; and Atkins, William. Managing the EDP

 

Function. New York: McGrawbHill Book Co., 1971.

Dittman, David A. "Transfer Pricing and Decentralization."

Management Accounting, LIV (November, 1972), 47-50.

Dolkas, James B. "Better Cost-Effectiveness at Computing Facilities

Through Better User Communications and Services."
Management, XI (November, 1973), 26-29.

Me

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

183

Dopuch, Nicholas, and Drake, David F. "Accounting Implications of a
Mathematical Programming Approach to the Transfer Price
Problem." Journal of Accounting Research, II (Spring, 1964),

 

10-24.

Emery, James C. "Problems and Promises of Regional Computing."
Datamation, XIX (August, 1973), 55-58.

 

Estes, Neil. "Step-by-Step Costing of Information Systems."
Journal of Systems Management, XX (August, 1969), 31-35;

 

and XX (October, 1969), 20—28.

Finney, Frederick D.

"Pricing Interdivisional Transfers." Management

 

Accounting, XLVII (November, 1966), 10-18.

 

Finney, John E. "Costing in a Data Processing Department."
Management Accounting, LVI (October, 1974), 29-35.

 

Foy, Nancy. Computer Management. Philadelphia: Auerbach Publishers

 

Inc., 1972.

Fredericks, Ward A.

"A Case for Centralized EDP." Business

Automation, XIX (January, 1972), 20-24.

 

Fried, Louis. "Estimating the Cost of System Implementation." Data
Processinngagazine, XI (March, 1969), 32-35; and XI (April,

 

1969), 24-28.

Gayle, John B. "Multiple Regression Techniques for Estimating Computer
Programming Costs." Journal of Systems Management, XXII
(February, 1971), 13-16.

Gellman, Harvey S.

 

"Evaluation of the Effectiveness of EDP Facilities."

Journal of the Society of Industrial Accountants of Canada,
XLIX (November-December, 1975), 16-19.

Gemar, W. J. "Making the Most of Your Computer." The Accountant
(England), May 8, 1975. Quoted in the Journal of Accountanqy,
CXL (September, 1975), 103.

Gibson, Cyrus F. and Nolan, Richard L. "Managing the Four Stages
of EDP Growth." Harvard Business Review, LII (January, 1974),

76-88.

 

Guidice, John J., and McElroy, John J. "Allocating Job Costs for
Multi-programming Systems." Data Processing Magazine, XIV

(Spring, 197

2), 16-190

Goetz, Billy E. "The Effect of a Cost-Plus Contract on Transfer
Prices." AccountingyReview, XLIV (April, 1969), 398-400.

 

. "Transfer
Congruence."

Prices: An Exercise in Relevancy and Goal
Accounting Review, XLII (July, 1967), 435-440.

 

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

184

Gordon, Jyron J. "A Method of Pricing for a Socialist Economy."
Accounting Review, XLV (July, 1970), 427-443.

 

. "A Method of Pricing for a Socialist Economy, A Reply."
Accounting Review, XLVI (October, 1971), 788-790.

 

. "The Use of Administered Price Systems to Control Large
Organizations." Management Controls: New Directions in
Basic Research. Edited by Bonini, Jaedicke, and Wagner.
New York: McGrawbHill Book Co., 1964.

 

 

Gould, J. R. "Internal Pricing in Firms When There Are Costs of Using
An Outside Market." Journal of Business, XXXVII (January,
1964), 61-67.

 

Greewood, Frank. Managing the Systems Analysis Function. American
Management Association, Inc., 1968.

 

Greer, Howard C. "Divisional Profit Ca1culation--Notes on the
'Transfer Price' Problem." NAA Bulletin, XLIII (July, 1962),
5-12.

Grillos, John M. "Pricing EDP Resources." Computer Decisions, VI ,zv,

 

Hammer, Glenn B. "Cutting Time-Sharing Costs." Datamation, XXI
(July, 1975), 36-39.

 

Hartman, W.; Matthes, H.; and Proeme, A. Management Information
Systems Handbook. New York: McGrawbHill Book Company, 1972.

 

 

I - ('3' .

Heitger, Lester Edward. "Criteria for Transfer Pricing of Data
Processing Services within Business Firms." Unpublished
Ph.D. dissertation, Michigan State University, 1972.

Helmkamp, John G. "Technical Information Center Management: An
Accounting Deficiency." Accounting Review, XLIV (July, 1969),
605-610.

Heyel, Carl. John Diebold on Management. Englewood Cliffs, N.J.:
Prentice-Hall, Inc., 1972.

 

Hirsch, Rudolph E. "Data Processing Can Be Cost-Controlled."
Design and Management of Information Systems. Edited by
David H. Li. Chicago: Scicnce Research Associates, 1972.

"The Value of Information." Journal of Accountancy,
CXXV (June, 1968), 41-45.

 

Hirschleifer, Jack. "Economics of a Divisionalized Firm." Journal
of Business, XXX (April, 1957), 96-108.

 

. "Internal Pricing and Decentralized Decisions." Contemporany

 

Cost Accounting and Control. Edited by George J. Benston.
Belmont, California: Dickenson Publishing Company, Inc., 1970.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

185

. "On the Economics of Transfer Pricing." Journal of
Business, XXIX (July, 1956), 172-184.

 

Homes, Robert W. "Twelve Areas to Investigate for Better MIS."
MIS: Management Dimensions. By Raymond J. Coleman and
M. J. Riley. San Francisco: Holden-Day, Inc., 1973.

 

Holstrum, Gary L. and Sauls, Eugene H. "The Opportunity Cost
Transfer Pricefl' Management Accounting, LIV (May, 1973),
29-33.

 

Horgren, Charles T. Cost Accounting:, A Managerial Emphasis.
3rd ed. Englewood Cliffs, N.J.: Prentice-Hall, Inc.,
1972.

Humfrey, Robert D. "The Organizational Side of Centralized EDP
Services." Management Controls, XXI (January, 1974), 11-13.

 

Hurtado, Corydon D. "A System to Measure EDP." Journal of
Systems Management, XXIII (January, 1972), 32-35.

 

 

Jacobs, Donald. "Equitable Machine Cost Accounting in A Multi-

Programming Environment." Data Management, XI (November,
1973), 21-25.

Jones, G. Hunter. "Managing the Data Processing Manager."
Journal of Accountancy, CXXXVII (May, 1974), 72-75.

 

Jones, J. Bush. "Centralizing Computer Facilities." Journal
of Systems Management, XXII (June, 1971), 28—31.

 

Joslin, Edward O. "Costing the System Design Alternatives."
Data Management, IX (April, 1971), 23-27.

 

Kanter, Herschel; Moore, Arnold; and Singer, Neil. "The Allocation
of Computer Time by University Computer Centers." Journal
of Business, XLI (July, 1968), 375-384.

 

Kaplan, Robert 8., and Thompson, Gerald L. "Overhead Allocation via
Mathematical Programming Models." Accounting Review,
XLVI (April, 1971), 352—364.

 

Kelly, Joesph F. Computerized Management Information Systems.
New York: The Macmillan Company, 1970.

Kenny, John J. "Objectives for HOP Organizations." Data Management,
X (January, 1972), 33-35.

 

Khtaian, George A. "Costing of Data Processing Services." Journal
of Systems Management, XXVI (November, 1975), 32-36.

 

Knutsen, K. Eric, and Nolan, Richard L. "Assessing Computer Costs
and Benefits." Journal of Systems Management, XXV (February,
1974), 28-34.

 

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

109.

186

Krasney, Melvin. "Accounting Controls for Corporate EDP Costs."
Management Accounting, LII (March, 1971), 17-l8+.

 

Larson, Harry T. "EDP: A 20 Year Ripoff!" Infosystems, XXI
(November, 1974).

Larson, Raymond L. "Decentralization in Real Life." Management
Accounting, LV (March, 1974), 28-32.

 

 

Li, David H. Accounting; Computers; Management Information Systems.
New York: McGraw-Hill Book Company, 1968.

. Design and Management of Information Systems. Chicago:
Science Research Associates, 1972.

. "Interdivisional Transfer Planning." NAA Bulletin, XLVI
(June, 1965), 51-54.

Logue, James A. "Cost Control in Troubled Times--l7 Ways Companies
Reduce EDP Costs." Management Review, LX (February, 1971),
8-17 0

 

Martin, Donald D. "Data Processing Managers Need to Know Accounting."
Data Management, XIII (November, 1975), 26-28.

 

Martin, George Robert. "Accounting Data and Their Decision Use: An
Empirical Study of Transfer Prices and Their Use in Management
Decisions." Unpublished Ph.D. dissertation, University of
California, Berkeley, 1967.

Massey, L. Daniel. Managing the Human Element in EDP. Braintree,
Mass.: D. H. Mark Publishing Co., 1969.

McCosh, Andrew M. "Computerized Cost Classification System."
Management Accounting, LIII (April, 1972), 42-45+.

 

McCullers, Levis D., and McDill William R. "'Quasi-Pricing' for
Intracompany Transactions." Journal of Accountancy, CXXXIII
December, 1970), 80-82.

 

McFarlan, F. Warren. "Management Audit of the EDP Department."
Harvard Business Review, LI (May, 1973), 131-142.

 

; Nolan, Richard L.; and Norton, David P. Information
System Administration. New York: Holt, Rinehart, and
Winston, Inc., 1973.

 

 

Menkus, Belden. "What Data Cost to Process." Administrative
Management, XXXVII (June, 1975), 22-24.

 

Moriarity, Shane. "Another Approach to Allocating Joint Costs."
Accounting Review, L (October, 1975), 791-795.

 

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

121.

122.

123.

124.

187

Morino, Mario. "SMF Data Base Design Considerations." PACE Applied
Technology, Inc., Arlington, Va., 1972. (Mimeographed.)

Most, Kenneth S. "Gordon's Transfer Price Model for a Socialist
Economy: A Comment." Accounting Review, XLVI (October, 1971),
779-782.

 

Myers, Charles A., ed. The Impact of Computers on Management.
Cambridge, Mass.: The M.I.T. Press, 1967.

 

National Academy of Sciences. Information Systems Management and
Evaluation. A Report by the Computer Science and Engineering
Board. Washington, D.C., 1972.

 

 

Nielson, N. R. "Flexible Pricing: An Approach to the Allocation of
Computer Resources." Proc. AFIPS 1968 Fall Joint Computer
Conference, XXXIII, 521-531. Wayne, Pa.: MDI Publications,
1968.

 

 

. "The Allocation Of Computer Resources--Is Pricing the
Answer?" Communications of the ACM, XIII (August, 1970),
467-474.

 

Nolan, Richard L. "Plight of the EDP Manager." Harvard Business
Review, LI (May, 1973), 143-152.

 

Nunamaker, J. F., and Whinston, A. "Computer System Management:
A Macro Planning Cost Allocation Procedure." Purdue
University, 1973. (Mimeographed.)

Onsi, Mohamed. "A Transfer Pricing System Based on Opportunity
Cost." Accounting Review, XLV (July, 1970), 535-543.

Palmer, Carl R. "Computer-Based Information Systems and Related
Activities--Accounting Entity Definitions" (Draft Copy).
Paper presented at the GAO Symposium on Cost Accounting
and Cost Control, New Orleans, La., May, 1974. (Mimeographed.)

Phillippakis, Andreas S., and Thompson, Howard E. "Reward Functions,
Transfer Prices, and Decentralization." Quarterly Review
of Economics and Business, X (Spring, 1970), 57-66.

 

Rau, Paul. "Evaluating the EDP Function." Datamation, XVIII
(September, 1972), 72-73.

 

Reichardt, Karl E. "Capitalizing Costs of Information Systems."
Management Accounting, LV (April, 1974), 39-43.

Reichenbach, Robert R., and Tasso, Charles A. Organizingyfor Data
Processing. American Management Association, Inc., 1968.

 

Riley, Robert T., and Williams, Thomas A. "Predicting Information
Costs." Journal of Systems Management, XXIV (June, 1973),
26-290

 

125.

126.

127.

128.

129.

130.

131.

132.

133.

134.

135.

136.

137.

138.

139.

188

Ronen, Joshua, and Livingstone, J. L. "An Expectancy Theory Approach
to the Motivational Impacts of Budgets." Accounting Review,
L (October, 1975), 671-685.

 

, and McKinney, George, III. "Transfer Pricing for Divisional
Autonomy." Journal of Accounting Research, VII (Spring, 1970),
99-112.

 

Ruskin, Vernon W. "Comparing Computer Operations." Journal of

Systems Management, XXIV (December, 1973), 34-38.

 

 

Sass, C. Joseph. "Benefits of Equitable Charges for a Batch Computer."
Data Management, IX (March, 1971), 23-25.

 

Schaller, Carol. "Survey of Computer Cost Allocation Techniques."
Journal of Accountancy, CXXXVII (June, 1974), 41-46.

 

Sharav, Itzhak. "Transfer Pricing--Diversity of Goals and Practices."
Journal of Accountancy, CXXXVII (July, 1974), 56-62.

 

Sharpe, William F. The Economics of Computers. New York: Columbia
University Press, 1969.

 

Shaw, John C., and Atkins, William. Managing Computer System
Projects. New York: McGraw-Hill, Inc., 1970.

 

Shillinglaw, Gordon. Cost Accounting: Analysis and Control. 3rd ed.
Homewood, Illinois: Richard D. Irwin, Inc., 1972.

 

Shubik, Martin. "Incentives, Decentralized Control, the Assignment
of Joint Costs and Internal Pricing." Management Controls:
New Directions in Basic Research. Edited by Bonini, Jaedicke,
and Wagner. New York: McGraw-Hill, Inc., 1964.

 

 

Smidt, Seymour. "Flexible Pricing of Computer Services." Management
Science, XIV (June, 1968), 581-600.

 

. "The Use of Hard and Soft Money Budgets, and Prices to
Limit Demand for Centralized Computer Facility." Proc. AFIPS
1968 Fall Joint Computer Conference, XXXIII, 499-509. Wayne,
Pa.: MDI Publications, 1968.

 

 

Smith, Hendrick S. "Cost Control for Computers." Business Horizons,
XVI (February, 1973), 73-82.

 

Snydermand M., and Kline, R. A. "Job Costing a Multiprogramming
Computer." Journal of Data Management, VII (January, 1969),
19—20+o

 

Sobczak, J. J. "Pricing Computer Usage." Datamation, XX (February,
1974), 61-640

 

140.

141.

142.

143.

144.

145.

146.

147.

148.

149.

150.

151.

152.

153.

189

Sollenberger, Harold M. "A Cost Accounting Framework for EDP
Management." Paper presented at the GAO Symposium on Cost v//’

Accounting and Cost Control, New Orleans, La., May, 1974.
(Mimeographed.)

. "Factors Which Influence the Use of Cost Accounting Data
in EDP Management." Paper presented at the GAO Symposium on
Cost Accounting and Cost Control, New Orleans, La., May,
1974. (Mimeographed.)

. Management Control of Information Systems Development. V”
New York: National Association of Accountants, 1971.

. "Management Information Systems: A Charge to Users and
Cost Control." Management Accounting, LII (November, 1970),
25-28+.

I

, and Arens, Alvin A. "Assessing Information Systems Projects."

Management Accounting, XXV (September, 1973), 37-42.

Solomons, David. Divisional Performance: Measurement and Control.

New York: Financial Executives Research Foundation, 1965.

Stone, Williard E. "Intracompany Pricing." Accounting Review,

 

XXXI (October, 1956), 625-627.

Tomlin, Roger. Managing the Introduction of Computer Systems.

London: McGraw-Hill Book Co., 1970.

Tosi, Jr., Henry L. "The Human Effects of Budgeting Systems on

Management." MSU Business Topics, XXII (August, 1974),
53-63.

Turney, Peter Benjamin Bolton. "An Accounting Study of Cost Behavior

and Transfer Pricing of Management Information Systems."
Unpublished Ph.D. dissertation, University of Minnesota, 1972.

U.S. General Accounting Office. "Interview/Question Guide for Current

Vendig,

Wagner,

Practices Survey of Cost Accounting and Cost Control of
Computer-Based Information Systems —- Final Version."
Washington, D.C., November 6, 1974. (Mimeographed.)

. Report to the Congress. Opportunity for Greater Efficiency
and Savings through the Use of Evaluation Techniques in the
Federal Government's Computer Operations. Washington, D.C.,
1972.

Richard E. "A Three-Part Transfer Price." Management
Accounting, LV (September, 1973), 33-36.

 

L. C. "Computers, Decentralization, and Corporate Control."
California Management Review, IX (Winter, 1966), 25-32.

(J y.

a)

. .,.
/(- fr ,. ‘
,0 \
I
f
'.

154.

155.

156.

157.

158.

159.

160.

161.

162.

163.

164.

190

Wallace, John B. "When to Bill for Computer Services." Data
Management, XIII (July, 1975), 12-17.

 

Warner, C. Dudley. "Monitoring: A Key to Cost Efficiency."
Datamation, XVIII (January, 1971), 41-43+.

 

Watson, David J. R., and Baumler, John V. "Transfer Pricing: A
Behavioral Context." Accounting Review, L (July, 1975),
466-474.

 

Wiorkowski, Gabrielle K., and Wiorkowski, John J. "A Cost Allocation ’
Model." Datamation, XIX (August, 1973), 60-65.

 

Witham, Robert B. "Controlling Computer Results." Management
Accounting, XXV (October, 1973), 48-51.

 

 

Withington, Frederic G. "Crystal Balling: Trends in EDP Management."
Infosystems, XX (January, 1973), 20-21.

 

. The Organization of the Data Processing Function. New

 

 

York: John Wiley and Sons, Inc., 1972.

. The Use of Computers in Business Organizations. Reading,

 

 

Mass.: Addison-Wesley Publishing Co., 1966.

Wood, Donald R., and Ditri, Arnold E. "Managing EDP Costs."
Financial Executive, XXXIX (October, 1971), 72-74+.

 

Wormley, James T. "Ensuring the Profit Contribution of a Corporate
Data Processing Department." Management Accounting, XLVIII
(January, 1967), 3-12.

 

Zeyhe, Lewis R. Cost Reduction in the Plant. Englewood Cliffs, N.J.:
Prentice-Hall, Inc., 1965.

 

APPENDIX A

l.|u .Iul.‘IlI‘

APPENDIX A

SELECTED EXTRACTS FROM
USER SPECIFICATIONS OF DATA CENTER
ACCOUNTING AND RESOURCE UTILIZATION SYSTEM

(Firm A)

Subsystems and Their Function

The Accounting and Resource Utilization is divided into

four subsystems:

(1) Logout and Log Preprocessing

(2)

(3)

(4)

Function: To capture and preprocess system log files and

prepare a job oriented file for Accounting and a merged
raw log for the Resource Utilization Subsystem.

Accounting

Function: To redistribute data center costs to its users

in a equitable and auditable manner through an invoice which
displays resource units consumed.

Resource Utilization

Function: To gather statistics on the usage of the various

system resources (hardware units, software and personnel) and
produce management reports which will enable the data center
to reconfigure and reorganize to optimize throughput.

System Software Modifications

Function: TO validate the right of a potential user to use

computer resources (charge code validation) and to inform
a user at the end of each job as to the cost of that job
in resource units used and dollars.

191

192

Description of Output Reports

 

Error Diagnostic

 

Each system program will have an error diagnostic printer file
to report system failures due to [system software]* errors as

well as any errors detected that cause abnormal termination of
the run.

Job Structure By Begin Time

 

Jobs are listed by begin time with their corresponding tasks
also ordered by begin time. All known information about each
job is shown in this report. Certain data may be eliminated by
parameter inputs set by the user.

Billing Report By Account

 

Jobs processed are ordered by account number then by begin
time. Information contained in this report is similar to
the Job Structure By Begin Time Report.

Audit Of Billing Clerk Activity

 

A sequential listing of all external debits, transfer and
reversals to accounts performed by the billing clerk are
reported. Also included in this report are all other
transactions concerning modification of [certain] files.
This report serves as a hard copy record of all actions
performed by the clerk.

Audit Of Operator Comments

 

A sequential listing by computer system of all Operator log
comments and all Operator log to job messages. The report
serves as a summary of all messages entered bythe system
operators.

If the message is input in a specific format, the computer
operator can generate a transaction from the message to allow
for automatic reversal of the job if it has an error termination
because of a Data Center error. The billing clerk would have to
verify the reasonability or accuracy of the reversal.

List of DS-ed Jobs

 

This report calls the billing clerks attention to jobs that
terminated abnormally. These DS-ed jobs are good candidates
for charge reversal.

 

*
Items in brackets indicate that certain specific terms have been
changed so that the firm is not identifiable.

193

Daily Revenue Statistics

 

Summarized information by system is displayed to reflect

where the data center is receiving revenues. Breakouts are

given by Job Queue, Run Type, Shift, Compiler Usage and

Revenue Sources. An additional Revenue Summary breakout is
produced indicating reasons for lost billing. The information
produced on this report will be saved in a database bile to allow
printing of the monthly report.

Hardcopy Of Billing Clerk Activity

All information for each session of the online inquiry and

update program (excluding inquiry) is recorded on theline printer
file. This file serves as a permanent record of billing clerk
transactions performed during the period logged.

Monthly Account Invoice

This report is designed to be a tear apart report to be
distributed to each account (user). A minimum of two pages per
user will be produced. The first page will show a summary of all
units billed while the second page will provide detail information
concerning jobs charged to the account as well as billing clerk
credit and debit entries.

Monthly Report By Project

 

This is a specialized report to meet the needs of those data
centers required to produce a monthly statement by Corporate
Unit, Project Number, and Cost Center sequence.

Monthly Revenue Statistics

This report will be produced from the data base file generated by
the DAILY UPDATER program. As with the daily statistics

report, summarized information is reported by Job Queue, Run
Type, Compiler Usage, Revenue Source and a Revenue Summary
indicating reasons for lost billing.

General Accounting Journal Entries

This report will show data processing charges to all cost
centers in standard journal entry form for either the
Manufacturing or Marketing Accounting Systems. This report
can be used by the accounting department in the preparation
of journal entries for data processing charges.

Listing of Accounts

 

A sorted listing is produced containing all accounts within the
system. All information pertaining to each account is printed
including account attributes, expiration data, charge number and
password for each account.

194

Explanation and Rationale of Billable Items
(a) Processor Time

Definition:

 

Processor time is the total time during which a central processing
unit (CPU) is executing instructions for the user task.

Rationale:
Processor time is a possible (and suggested) billable item.

1. The processor is a major component of the computer system
in terms of cost.

2. Processing data is a major function of the computer system.
3. It is an easily measurable resource.
4. Times are fairly mix independent Q: 10%).

(b) I/O Time

Definition:

 

I/O time is the total time an I/O channel is in use controlling
an input or output operation for a user task.

Rationale:
I/O time is a possible (and suggested) billable item because:

1. The I/O multiplexor is a major system component relative to
total system cost.

2. I/O is a major data processing activity.

3. All I/O operations to all peripheral devices accumulate
I/O time except those to a remote file.

4. I/O time is easy to measure.
5. I/O time is mix independent Q: lOZ).
(c) Code Core Integral

Definition:

 

The code core integral of a user task is the average size of
its code area times the sum of its processor and I/O times.

195

Rationale:

Code core is a possible item because:

Gore is expensive.
Charging for code core will encourage efficient program coding.
The code core integral is calculated by the [system software]

and is easily Obtained from the EOT (end of task) record in
the system log.

(d) Data Core Integral

Definition:

 

The data core memory integral is the accumulation, for all data
areas, of the size of the memory area allocated times the sum of
the processor and I/O times accumulated during the period in
which the data area was present.

Rationale:

Data core integral is a possible billable item because:

1.

2.

NOTE:

Core is expensive.

If data core was a free resource users would substitute large
arrays for disk space which to accesss requires I/O time
which is not free (and implies increased job run time).

The data core integral is calculated by the [system software]
and is easily obtained from the EOT record.

If both data and code core integrals are billed, data core
should be more expensive. Data must be written to disk
overlay before the core area may be reused. Code may be
simply forgotten since a copy of all code (all of which is
not modified) is always present on disk.

(e) Cards Read

Rationale:

Cards read is a possible billable item because:

1.

The cost of reading cards in from the reader to disk
(spooling) is not covered by I/O time which is accumulated
when reading from disk to core.

It will represent a nuisance charge for card handling.

It will indirectly charge for disk space used to store the
card images.

196

4. It is a readily available statistic in the EOT log record.
5. It is mix independent.
(f) Cards Punched

Rationale:

Cards punched is a possible billable item because:

1. The cost of the program which punches spooled punch backup
disk files, is not billable to any user. Therefore, a
charge for cards punched may recover that cost.

2. It can be used to recover the cost of cards and card handling.

3. It can be used to recover the cost of the card punch.

4. It is a readily available statistic in the EOT log record.

5. It is mix independent.

(g) Lines Printed

Rationale:

Lines printed is a possible billable item because:

1. The cost Of the program which prints spooled printer backup
disk, is not billable to any user. Therefore, a charge for
lines printed may recover that cost.

2. It can be used to recover the cost of paper and paper handling.

3. It can be used to recover the cost of the printer.

4. It is a readily available statistic in the EOT log record.

5. It is mix independent.

(h) Connect Time

Definition:

 

Connect time is the elapsed time between LOGON and LOGOFF for
[online] sessions.

Rationale:
Connect time is a possible billable item for the following reasons:
1. The cost of the datacomm subsystem is a significant cost not

otherwise billed for directly by any other unit. Note l/O
time is not accumulated for remote files.

3.

4.

197

The following items can be billed for using connect time:

a. Datacomm Subsystem including line adapters and adapter
clusters.

b. Modems

c. Terminals

d. Supply costs at the remote terminals

e. Line costs

A user will be encourage to maximize the effectiveness of his
use of terminals. He will plan a session more carefully.

This statistic is easy to gether for it is present in the
system log LOGOFF record.

(1) Tape Mounts

Rationale:

Tape mounts is a possible billable item because:

1.

2.

3.

4.

It is a major time consuming task for an operator to mount,
dismount, and watch for tape status.

Users should be discouraged from tape usage when an alternative
(disc pack or elimination of the tape altogether) is available.

This item complements tape storage charges in that it reflects
a greater charge to tapes which are used more often and

charges the user, not the owner.

This statistic is mix independent.

(j) Tape Open Time

Definition:

 

Tape open time is the sum of all the open times for all files
Opened by the user task. An individual file's open time is the
elapsed time from open to close of that file.

Rationale:

Tape open time is a possible billable item because:

1.

The cost of a tape drive can be recouped via this charge item.
Note that a user is tying up the whole unit for the entire
time a file is open.

Users are discouraged by this charge from using tape files
if an alternative exists.

198

(k) Shift Premium

Definition:

 

Shift premium is a surcharge (or discount) which is added to the
normal cost of a task based on the time of day the task is initiated.

Rationale:

Shift premium is a possible billable item because:

This item will give economic incentives to run in non-prime shifts
and so decrease prime shift demand thus load leveling the computer
system.

(1) Queue Premium

Definition:

 

Queue premium is a surcharge or discount on the normal cost of a
task based on the queue from which the task is active. Queue
parameters indicating level of service include limits on:

1. Priority
2. Process Time
3. I/O time
4. Cards punched
5. Lines printed

The queue used by a job's tasks is noted in the EOT log record.
(m) Keypunch (in-house)

Rationale:

In-house keypunching is a significant and easily isolatable

expense. In fact, by Management Systems' policy the Data Entry

department must be a separate cost center. The General Accounting

system interface produces two journal entries: one for keypunch

services, and one for all other computer services.

(n) Tape Library

Explanation:

 

The billing unit used is reel-months (one reel of tape storage

for one month). It is possible to modify the tape library interface
program to charge some multiple number of times for packs entered
into the tape library. (For example let 1 pack-month 8 10 reel-
months).

199

Rationale:

Tape library storage is a significant expense to the data center
and it is easy to redistribute the cost back to its source.

(0) Disk and Pack Usage

Explanation:

 

A program will be produced (on release 2+) to sample disk and packs
to determine usage and build debit entries. For now a manual
entry may be made. Unit of measure is mega-segment-months.

Rationale:

Disk and pack space represent a major cost. There is, however,
trouble in identifying the owner of a file.

(p) Offline

Explanation:

 

A manual DR entry may be made for offline owrk. The unit of
measure is hours.

Rationale:
To account for offline machine and operator time.
(q) Clerical

Explanation:

 

A manual DR entry may be made for clerical work. The unit of
measure is hours.

Rationale:
To account for miscellaneous clerical time.
(r) Xerox

Explanation:

 

A manual DR entry may be made for Xerox backup file printing.
The unit of measure is pages. A DR may be filled out from a
summary sheet produced by the Xerox machine at the end of each job.

Rationale:

Xerox printing costs are a substantial data center expense and the
user of this service is easy to distinguish.

200

(5) Dollar Billing

Explanation:

 

A DR entry for a miscellaneous charge in dollars may be entered.
Rationale:

Some items are charged for so rarely that they do not warrant a
category of their own. A convenient unit for this catch-all
category is dollars.

(t) 5 Miscellaneous Units

Explanation:

 

Five spaces for user defined accumulators are provided. Units of
measure and meaning is [sic] user specifiable.

Rationale:

To be rationalized by the user who defines the accumulator.

Overview Of Policies and Procedures

 

The general policies which apply to and influence the Operating philosophy
and practices of the data processing facility are shown in draft form in
this section. They define the approach to be used in administering a

Data Center Accounting System.

The procedure drafts contained herein indicate the manner by which the
policies may be put into practice and administered on an on-going basis.
These procedures should be placed in the Standard Practice Manual of the
individual unit; they have titled here for the "Data Center Procedure
Manual".

This section is organized in the following manner: the policy draft,
which in final form would appear in one of the functional manuals
(Controller's, Management Systems, Organization & Policy, Data Centers,
etc.) is followed by the applicable procedures necessary to implement
this policy and provide for its continuance on an on-going basis. The
overall functioning of the system was viewed in terms of seven policies,
each of which occupies a separate subsection of this section.

[Author's note: Because of the length of the procedure drafts only the
policy statements are included in this Appendix.]

The first policy is entitled Accounting for the Use of Data Center Resources,
and explains why accounting is an appropriate action by a Data Center.

The procedures accompanying this policy indicate the actions that need be
taken and the responsibility assignments that must be made to implement

the system, and the fact that activity reporting is to be performed on a
periodic basis.

201

A key to accounting is prOper identification of the consumer of resources.
The Change Identifier policy indicates the objectives here, and
accompanying procedures establish how this is to be controlled.

 

Reporting of use is in large measure determined by the manner in which
resources are allotted to individual organizations within the data
processing center. Data Center Organization for Cost Control and its
accompany procedures indicate the manner in which this allocation
process is to function.

Cost Reflexive Pricing established the philosophy to be adapted by a
Data Center in charging for services. The four procedures accompanying
this policy describe the manner by which unit prices for services are
established and validated.

 

The fifth policy draft established how Data Center Services Billings are
to be handled, and indicates the relationship that is to exist among data
centers relative to the interchange of work and the resultant billings.
The four procedures here indicate the manner by which services are to be
requested, types of services to be offered, billings rendered, and price
disputes resolved.

Adjustment to bills are a normal function of data center accounting, and
the Customer Credits policy states the principles governing such adjust-
ments. Associated procedures include those for establishing minimum
amounts required for credit, the means by which credits are requested,
and the manner in which credit transactions are handled.

 

The seventh and last policy draft is Retention of Accounting Information.
Supporting procedures indicate how retention schedules are to be estab-
lished, how log pooling as a technique can be employed, how report
retentions are entered on the data base, where microforms may be used, and
how records are transferred from the data processing center to a records
storage area.

1. Accounting for the use of Data Center Resources

Rationale

The commitment of the aggregate of resources represented within
each data processing facility is indicative of a conscious decision to
forego other economic alternatives in favor of this particular use.
The prudent management of this aggregate involves the measurement
and control of the manner by which the Data Center provides services.
Policy

Each data processing center shall instituteaistandardized mechanism

of internal accounting which has as its basic Objective the identification

202

and measurement of activity for those organizations utilizing its
services. These services represent a consumption of limited resources
having a real economic cost, the use of which must be neither
indiscriminate nor capricious.

This cost is to be assessed users of these services in a manner
which is comprehensible and which encourages them on a continuing
basis to measure the benefits they realize against the costs incurred,
and to influence or modify their demands accordingly.

Applicability

 

All data processing centers for which the prime budgetary
responsibility rests with the Management Systems function.

Responsibilities

 

Each installation manager shall be responsible for the institution
of procedures which permit the adequate identification of all users
of data processing services, the accumulation of resource activity to
their accounts, and the rendering of periodic statements of services
provided to each. The management of each activity utilizing the data
processing facility shall be responsible for the development of a
forecast of the nature and volume of demand required from the data
center. These forecasts shall be used as a resource planning tool
within the data processing function.

Relation to Operational Analysis

 

The measure of activity and the variance from forecast, with
associated cost and service details, shall be analyzed on a continuing
basis withe the objective of improving the effectiveness and efficiency
of the data center function. The responsibility for this analysis lies
with the management of the data center, with the active cooperation

and assistance from its users.

203

Charge Identifiers
Rationale

Proper control over the validity of data processing service request
is essential as the honoring of these requests represents a consumption
of data processing center resources, the cost of which is to be
subsequently distributed to the requestors of service. The mechanism
through which this control can be ahieved, in part, is a properly
administered program utilizing charge identifiers.
Policy

The fulfilling of service requests by a data processing facility
shall be dependent on the existence of a valid charge identifier
used to specify the account to which charges for this work shall be
accumulated.

Definitions

 

A charge identifier is a non-significant identifier used to
designate an account to which data processing work done is to be
charged. An account is a record to which charges for transactions
are entered and subsequently distributed to the account holder. A
project is an identifier utilized to specify an approved undertaking
for which the expenditure of company resources is authorized over
a period of time.

Establishment and Maintenance Of Account Numbers

The issuance and control of account numbers is the responsibility
of the ranking financial manager (or his delegate) at each Corporate
unit. Requests for addition, deletions and revisions to accounts
are to be handled in accordance with procedure and are properly directed

to the financial activity of the unit for review and approval.

204

Establishment and Maintenance of Project Numbers

 

The issuance and control of project numbers is the responsibility
of the manager (or his delegate) who must monitor and periodically
measure the progress on the identified project.

Establishment and Maintenance of Charge Identifiers

 

The issuance and control of charge identifiers is the responsibility
of the manager of the data processing facility (or his delegate) at
which the work is to be performed. The determination of the validity
of the account number and project number (if used) must be made
at the time the charge identifier is established.

Use of Charge Identifiers

 

It is the responsibility of the data processing function to assure
at the time of submission of requests for services that the requestor
has the authority to use the charge identifier. This may result from
the identification to a line item on an approved processing schedule,
use of an authorized usercode and password, authorized signature matching
that on specimen signature cards, or other means deemed appropriate. On
a periodic basis an audit of the use of charge identifiers should be

made to insure the adequacy of the supporting procedural controls.

Data Center Organization for Cost Control
Rationale

As interrelated yet distinct functions exist within a data
processing facility, the organizational structure for cost control
should reflect the distinction of function.
Policy

Data processing facilities shall distinguish, minimally, among

data entry, computer maintenance, and computer operations and support

205

functions. This distinction shall be reflected in the accounting
treatment for the expenses incurred within each function and the
expense Offset made as a result.

Responsibilities

 

The manager of each data processing facility is responsible for
initiating the request for and implementing of cost control numbers
for each function, and for assuring that all associated expenses are
properly placed to each function. The assignment of cost control
numbers is the responsibility of the Controller's activity in each
corporate unit.

Relief of Incurred Expense

 

The full cost of operating the function shall be redistributed to
all customers in accordance with their use of the resource of the
function. This relief shall be made through an expense offset.
Independence of offset rates among functions is essential if the true
resource costs are to be evidenced. The expense of each function and
the offsets made shall be accumulated on an ongoing basis, and period-
ically examined for adequacy. The reestablishment of offset rates is
a drastic action having significant budgetary impact and should be
considered only in those instances where existing rates have ceased

to have a meaningful relation to existing conditions.

Cost Reflexive Pricing

Rationale

The total cost of providing sufficient quality computing services
for a given demand level can be calculated from a number of independent
cost pools that can be established. A breakeven pricing formula,

reflecting rates used to relieve each independently measured cost pool

206

and also reflecting the ratio among these cost pools must be developed
and then administered over a period of time.
Policy

The pricing formula adopted must be tailored to the environment of
the individual facility and must be adjusted when the environment
changes. It must include elements that can be measured independently
and which are explainable. Foremost, it must permit the redistribution
of the true costs of services to those demanding the services without
encouraging arbitrary or capricious allocations of expenses to cost
pools.

Expense Budgeting and Annual Demand Forecasts

 

On an annual basis it is the responsibility of data processing
management at the facility level to solicit demand forecasts from all
customers and to develop and secure approval of an expense budget
that will permit the servicing of the forecasted demand. Several
iterative cycles may be required before the expense level and the
corresponding demand levels are in balance.

Whereas in the development of the expense budget costs are grouped
into expense accounts, in the pricing calculation they are grouped
into cost pools. A cost pool is an accumulator wherein similar
costs are grouped and against which expenses are relieved. All costs
incurred in the operation of a data processing facility are to be
assigned to a cost pool, and the gross expense to meet a given demand
level is thus the sum of the cost pools.

While the establishment of individual cost pools is at the option
of each data processing center, as a general guideline when identifiable
costs to a pool reach five percent of gross expense, a distinct cost

pool should be created for purposes of expense relief.

207

Cost Pool Ratios

 

The Relationship between the cost pools is then shown by the
relative proportion they bear to the gross expense. As changes in
environment influence the size of each cost pool, and adjustment either
in the ratio to other pools or to the unit price should be made.
Resource Unit Pricing

In determining the price of a resource unit, the usable capacity,
as distinguished from the theoretical capacity, should be viewed as
the upper boundary. Demand forecasts for each unit should then be
applied to the total cost to be relieved to determine the price to be
charged.

Cost of Applications

 

The total cost of an application is the sum of the resources
used multiplied by the respective rate of each, plus a portion of the

unallocated costs of operating the data processing facility.

Data Center Services Billings
Rationale

The orderly handling of data center services billings on an
interdepartmental, inter-unit, and interdivisional basis makes
mandatory the statement of governing principles and criteria for
the handling of these transactions.
Policy

In terms of the goods of the Corporation, Data Center Services
are goods of a miscellaneous nature for internal use and are handled
through negotiation between the buying and selling units. Negotiations

shall be substantiated in writing and will be binding. The transactions

208

will handled in accordance with the accounting procedure of the
Corporation.
Interdepartmental Billinge for Data Center Services

Interdepartmental charges for data center services will be
handled monthly and will be computed on the basis of pre-established
rates of charge for that period. The charge rate used will be
determined by the level of expense offset necessary to cover the
costs of handling and approved demand forecast level. The transfer
of charges will be handled directly by journal entries from the Data
Center Accounting System to the Corporate General Accounting System.
Interdivisional & Inter-unit Computer Services Billings

Requests for data center services between Corporate units will be
negotiated by the respective data processing organizations or their
functional equivalent. Any special conditions or terms are to be
specified at the time of negotiation, and in the event none are specified
the types Of services provided by the selling facility to its internal
customers will be assumed.

When the purchasing and selling facilities each Offer data center
services, the charge assessed will be at the lower of the two billing
rates used. In no event are they to exceed the cost Of comparable
processing if done at the purchasing facility. Billings shall be
redistributed directly to the departments involved.

Where the purchasing facility neither has nor intends to acquire
an internal computing capability, it shall negotiate the puchase of
services with potential sellers. A negotiated demand forecast for an
extended period shall be binding on both parties and charges will be
assessed at the rate calculated when the external demand is incorporated

with internal demand to determine the base.

209

Disputes and Settlements

 

Where price disputes arise, the established mechanism for the
resolution of interdivisonal price disputes is to be utilized. All
reasonable efforts to resolve the dispute at the unit level are to be

exhausted before appeal is made.

Customer Credits for Erroneously Billed Data Center Services
Rationale

The statement of governing principles and criteria promotes the
understanding between a data center and its customers as to those
circumstances wherein credits for erroneous charges will be made to
customer accounts. This statement encourages the impartial admini-
stration of data center billings.
Policy

Where it is reasonably determinable that a data center customer
has, without fault, been assessed charges for services incorrectly
performed or billed, corresponding credit will be issued and
notification supplied the customer of the credit. The original
charge transaction will remain unaltered, but will be offset by a
credit to the appropriate customer account.

Responsibilities

 

The cost of handling customer credits should be considered in
determining that level of charge below which no credits will be given.
It is the responsibility of the data center manager to make and secure
approval of this minimum credit amount. This information is to be
available to all customers of the data processing center. It is further

the responsibility of the data center management to insure that adequate

210

procedures exist to document the credits made to customer accounts.
Where customer billings have been made and credits are requested, it
istfluaresponsibility of the requestor to present adequate documentation
to identify the erreneous transaction and determine the appropriate
credit.
Auditability of Credit Transactions

A record of credits issued, with supporting documentation, is to
be separately maintained for purposes of auditing the customer billings
made. Where a request for credit is made but subsequently denied,
documentation of this fact must also be available. These records will
be maintained in the data processing center for the appropriate period

and then discarded.

Rentention of Data Center Accounting Information
Rationale

The use of data center services for which charges are assessed
dictates that the data processing facility maintain for an adequate
period records of activity which are sufficient to substantiate
these charges.
Policy

The rentention schedules established for documents and computer-
readable records will be a nature to provide ease in auditing and
general checks of the integrity of these records.

Responsibilities

 

The Data Center Management in each facility shall be responsible
for establishing, adding, or modifying retention periods for its
records and shall obtain approval from the Division or Plant Management,

as appropriate.

211

The Data Center manager shall be responsible for confirming that
computer readable file retentions are adequate to meet the retention
requirements, and that all appropriate procedures to insure records
integrity are established and observed.

Where appropriate the use of pooling techniques or microforms
will be adopted to facilitate the retention of documents or records.
The use of these approaches will be determined at the installation
level. [A number of attachments showing record retention schedules

and control forms were presented.]

L... .

APPENDIX B

APPENDIX B

COMPUTER COST ALLOCATION PHASE II STUDY AND UPDATE
INCENTIVE CHARGING RATES

(Firm B)

A - OBJECTIVE

 

The primary objective of adopting an incentive rate structure is
the optimization of computer resource utilization through the load
leveling effect of this type rate. A secondary objective may be one of
discouraging large volumes of high-priority, fast turnaround jobs during
peak processing periods through the use of a penalty rate structure.
These higher rates may be justified as the result of the additional
scheduling work required when providing this type service.

B - GENERAL

 

The use of incentive rates only becomes practical under circum-
stances where an undesirable amount of unused computer center resource
capacity is available. Under this situation, incentive rates might be
considered as a means Of improving resource utilization. However, if
incentive rates are to be considered, it should be noted that unused
capacity seldom exists in proportionate quantities among the various
resource centers of the Operation. Therefore, for an incentive rate
structure to work to its fullest advantage, it should be directed
toward each individual available resource.

A rate structure that provideszapenalty for priority processing
(either a graduated scale of priorities and rates or a penalty rate for
fast processing during peak periods) does not appear practical at the
present time. There can be no assurance under such a scheme that any one
job will be intiated before any other jobs of that same priority. Past
experience with the use of priorities has indicated that, unless the
assignment of such priorities is constantly policed, each using department
eventually assigns the highest possible priority to all of its work.
Therefore, unless the penalties for high priority processing were estab-
lished of [sic] ridiculously high levels, a majority of the jobs would
be submitted under this classification. Therefore, this approach would
be self-defeating. It should also be noted that a high priority on job
initiation does not necessarily assure the user of specific or fixed
turnaround time for his work. Such other factors as job mix under identical
priorities, system work load, bursting and decallating schedules and mail
service all contribute to turnaround time. Therefore, only the approach
to load leveling will be discussed in the rest of this paper.

212

213

C - INITIATING INCENTIVE RATES

 

Identifying Unused Capacity

The first step in establishing an incentive rate structure is
to specifically identify unused capacity. Assuming that the quantity
of each resource available in the processing center is no greater than
that required to satisfactorily service peak-load periods, unused capacity
can be identified as that available during non-peak periods. Therefore,
such unused capacity take the form of portions of specific resources
during a particular period(s) during the day, week, month or year.

Establishing Goals

 

The second step to be performed is that of establishing the goals
to be achieved as the result of implementing an incentive charging scheme.
Such goals as increasing overall revenue, decreasing overall costs and/or
providing better service during peak periods were considered. The first,
increasing overall revenue, would dictate that the new charging method
should be one that attracts new jobs and has a minimal effect in encour-
aging existing work to be shifted so as to process under the lower rate.
The second approach would be one that encourages some portion of the
' existing peak-load jobs to be shifted to off-peak periods. However,
if this were to be the goal it would be necessary to determine that
resource costs would be reduced in a meaningful measure prior to
implementing the new rate. The third approach is similar to the second
except that, rather than reduce operating costs, better processing service
would be provided for more jobs during peak periods.

Determining Rates

As the result Of establishing incentive rates for off-peak
processing, in all probability the overall recovered revenue for the
department will be decreased. This would be the result of a certain
number of jobs that will be shifted from on-peak to off-peak periods to
take advantage of the lower rates. The actual impact on revenue of
this type of redistribution of work load can only be speculative at
this time. However, it will in all probability not be in a direct
ratio to the difference between the two rates.

Determining Applicability of Incentive Rates

To serve the load leveling objective, incentive rates should
only be available for jobs that can be started and completed during
off-peak periods. A work load profile of the central processing units
indicates that presently this Off-peak period extends from approximately
3:00 AM to 8:00 AM during normally staffed priods of Operation. There-
fore, if incentive rates are to be adopted, it would be the recommendation
that such be appliable to any job normally requiring four or less hours
elapsed processing time that can be started any time during the 12:00
Midnight to 6:00 AM period.

214

Monitoringnand Controlling:the System

 

As mentioned earlier, an incentive charging scheme requires
continual monitoring to assure that it is achieving the desired goals.
This monitoring activity may require as little time as a few hours per
month under situations where everything is performing as anticipated, or
it may require a substantial amount of effort under circumstances where
new rates need to be prepared and the using departments informed of the
change. Over and above the cost of the continuing monitoring activity
there will be a rather substantial programming effort required to prepare
the necessary monitoring reports at the time the jobs are being processed.

Impact on Charging System

 

If an incentive rate plan is to be implemented, it will require
a certain number of modifications ot the charging system. These changes
would involve not only the charging computations, but in all probability
require additional information be carried on the reports for use by the
using departments.

D - CONCLUSIONS AND RECOMMENDATIONS

 

It is concluded from this Study that, if even a simple incentive
rate structure were to be implemented, a rather substantial initial
investment would be incurred. That, based on the experience of one company
that tried such a scheme (reference EDP ANALYZER, Page 4, November 1973),
the results are highly unpredictable. That to be effective such a plan
must control the distribution of work between peak-load and off peak-load
periods. This control can only be accomplished through proper rate
administration. Proper rate administration can be expected to result in
fluctuating rates which are normally undesirable from the users' viewpoint.
That, even if successful, the overall benefits to the Company would be
minimal unless a meaningful reduction in Operating costs could be
achieved. Therefore, it is not recommended that an incentive rate plan
be initiated at the present time.

If further interest is generated in this area, it is recommended
that some trial rates be established and administered and evaluated on
a manual basis before incorporating this type of rate into the normal scheme
of charging.

APPENDIX C

APPENDIX C

ALLOCATION OF DATA PROCESSING EXPENSE

(Firm D)

onjectives:

 

a. Provides the data for the monthly control of Data Processing
Expense to user (or requesting) departments.

b. Provides the reasonable prorates to be applied to the actual
Data Processing Expense allocation for accounting purposes.

c. Provides the historical percentage data of prorate by charged
department, for the budgeting purpose.

Approach:
Three basic concepts are used to develop prorates:

Job Code: All operations must belong to a particular job code.
A job code consists of an average of over ten programs.

Requsting Department: Called a user department, which has a
direct contact with Data Processing Department and
has the responsibility for the expense control.

Charged Department: A inal burden center to which the expense
is charged.

The expense is accumulated by job code, then allocated by requesting

 

 

department and ny_job code, and finally reallocated by charged department.

This step is taken by using the predetermined rates of facotrs involved.

Total Data Processing Expense in 8890 [the centralized data center] is

divided into two types in terms of the approach method:

215

216

Computers Expense: The expense concerning routine program running,
-Central Processing Unit Expense (CPU)
-Line Printer Expense
-Tape Drive Expense
-Online Expense
-Disc Expense
-Atcom Expense

Others: The other expenses,
-Systems Development Expense
-Programming Expense
-Data Preparation Expense
-Microfilm Expense
-Flat Charges
Deve10pment of "Computers Expense" Allocation
The basic method is to estimate the burden, to keep it unchanged, to
accumulate the actual (or the other suitable units), then to develop

rates, and finally to break down the burden by job code.

Estimation of Burden

 

Computers Expense is divided into the seven basic burden centers based
on the function of each. The monthly burder amount in each basic
burden center is accumulated based on the estimation of the expense
which is considered to be charged to the particular basic burden center.
The basic burden centers and the structure of the burden are as follows:
(1) Basic CPU Burden

CPU Rental (from Purchase Order)
Miscellaneous Hardware Rental (P.O.)
Program Rental (P.O.)
Special "Slop" Miscellaneous Expense
(Prorates and so on)
X Number of Directors
X Supervisors
X Softwares
X Operators
X Stenographers
Total

217

(2) Basic Printer Burden
Hardware Rental (P.O.)
Expense Material
X Supervisors
X Operators

Total

(3) Basic Tape Burden
Drives at $ (P.O.)
Drives at $ (P.O.)
Miscellaneous Drives (P.O.)
Tape Controllers Rental (P.O.)
X Supervisors
X Operators

Total

(4) Basic Online Burden
Program Rental (P.O.)
Controllers Rental (P.O.)
X Software
X Operators
Total

(5) Basic Disk Burden
3330's (P.O.)

II” lFllll JVllllll lfllll

(6) Fixed Batch Burden
Hardware Rental (P.O.)
X Supervisors
X Operators
X Librarians
Total

(7) Basic Atcom Burden
Hardware Rental (P.O.)
X Software
Total

[ll ||”||||

 

Accumulation of "Actual" CPU Hours and the Other Operation Factors

 

In order to develop the rates to applied to the above basic burdens,
the following factors are counted and accumulated by built-in-counters.

Accumulation is made py job code and then totaled.

 

(1) CPU Hours: Batch CPU Hours (All other than below)
Printer CPU Hours
Online CPU Hours
Atcom CPU Hours

(2) Line Count

218

(3) Tape Drive Hours
(4) Disk Tracks: WOrk Space Disk Tracks
Online Disk Tracks
Permanent Disk Tracks
(5) Atcom Blocks
Develupment of Burden Rates
The objective of developing Burden Rates is to allocate the total basic
burden into each job. For instance, dividing the Basic CPU Burden by
Total CPU Hours provides CPU Rate per hour. CPU Rate times CPU Hours of
a particular job code becomes CPU Burden Amount of that job. However,
there exist some transfers between basic burden centers, because some

function must bear the burden of the other functions. Printing uses CPU,

for example.

The following calculation steps are taken:

Basic CPU Burden :
Basic Printer Burden:
Basic Tape Burden

Basic Online Burden :
Basic Disk Burden :
Fixed Batch Burden :
Basic Atcom Burden .

WHICDD-OO'N

Total Burden in "Computers Expense"
[ a lb b <: l (I ‘ e " f

Transfer between basic
burden centers

 

0°
H—n-u—I-II— .

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

a c d e f g
I Iz.am10madnku, :c”_‘ 7
CPU CPU .CPU 9» '.
' ¢)uJ l ' J ' fi)
5 wk "it"
Burden 1 1 [ we
L

 

 

 

 

Figure A-l. Illustration of Calculation of Burden Amounts

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

219

Calculation of Total CPU Hours

Total Batch CPU Hours
+ Total Printer CPU Hours
+ Total Online CPU Hours
+ Total Atcom CPU Hours

Total CPU Hours

 

Basic CPU Burden (a)

 

Total CPU Hours = a Rate

a Rate x Total Batch CPU Hours 8 Batch CPU Charge

a Rate x Total Printer CPU Hours = Printer CPU Charge
a Rate x Total Online CPU Hours = Online CPU Charge

a Rate x Total Atcom CPU Hours a Atcom CPU Charge
Determination of CPU Rate (Batch CPU Rate)

Batch CPU Charge + Fixed Batch Charge + Disk Work Space
Total Batch CPU Hours

= CPU Rate per CPU Hour
Determination of Line Rate

Basic Printer Burden + Printer CPU Charge
Total Line Count

 

= Line Rate per Line
Determination of Tape Drive Rate

Basic Tape Burden
Total Tape Drive Hours

 

= Tape Drive Rate per Hour

Determination of Online Rate

Basic Online Burden + Online CPU Charge + Online Disk
Total Online CPU Hours

 

= Online Rate per CPU Hour
Determination of Disk Rate

Basic Disk Burden - (Work Space Disk + Online Disk)
Total Disk Tracks

= Disk Rate per Track
Determination of Atcom Rate

Basic Atcom Burden + Atcom CPU Charge
Total Atcom Blocks

 

= Atcom Rate per Block

220

Development of Burden Amounts by Function by Job Code

(1) CPU Amount = CPU Rate x CPU Hours
Consists of: Basic CPU Burden
+ Fixed Batch Burden
+ Work Space Disk Charge
- Printer CPU Charge
- Online CPU Charge
- Atcom CPU Charge

(2) Line Amount = Line Rate x Line Count
Consists of: Basic Printer Burden
+ Printer CPU Charge

(3) Tape Amount 8 Tape Drive Rate x Tape Drive Hours
Consists of: Basic Tape Burden

(4) Online Amount - Online Rate x Online CPU Hours
Consists of: Basic Online Burden
+ Online CPU Charge
+ Online Disk Charge

(5) Disk Amount - Disk Rate x Disk Tracks
Consists of: Basic Disk Burden
- WOrk Space Disk Charge
- Online Disk Charge
(6) Atcom Amount - Atcom Rate x Atcom Blocks
Consists of: Basic Atcom Burden
+ Atcom CPU Charge
Each job code has the above amount by function, as well as the total.
The total amount by job code and its grand total provide "All Data

Processing Expense" data with "computed expense".

Development of "Others" Allocation

 

The other type of approach in Data Processing Expense Allocation is
based on the development of the predetermined rates and the accumulation
of actual manhours.

Development of Rates per Manhour

 

(1) Systems Developent Rate

(2) Programming Rate
(1) and (2) are calculated together.

221

Salaries

41 People - 5 systems analysts, 36 programmers = $ A
41 x 8 hours x 21 days x 12 months = 82,656 hours per year
$ A é 82,656 hours 8 $ a per hour basic rate

 

Employee Benefits

41 People x s (given) each= $_
S B - $_ A - x $_ a -

B
$_ b per hour benefits

Overtime Premiums & Cost of Living Allowance

 

Cost Of Living - 41 People x $ (Given) each - $_ C
$_ C 3 $_ A= _Z x $_ a -$_ c per hour C. O. L. A.
Overtime Premium- (given2Z x $_ a = $ d per hour O.T.

Buildings, Services, Utilities, Etc.

 

 

 

$ (Prorate) x 41 People = 3 E
Total Data Processing People -———

$ E f 82,656 hours a S e per hour Buildings, etc.

Supplies and Miscellaneous Office Equipment

 

$ (Prorate), x 41 People
Total Data Processing People

 

 

=$_F__
S F 4 82,656 hours 8 S f per hour Supplies, etc.

Supervision,nClerical, and Administrative Expense

 

7 People - l steno., l analyst, l director, 4 supervisors
calculation is the same as above

3 g per hour Supervision, etc.

Total a + b + c + d + e + f + g - Systems Development Rate

(3)

(4)

(5)

8 Programming Rate
Data Preparation Rate
Same Method, includes rental of specialized equipment
Microfilm Rate
Same
Flat Rate

Fixed charge of machine rental located outside General
Operation Computer Room, but charged to 8890.

222

Accumulation of "Actual" Manhours

Actual manhours are measured by job code and by category. After totaling
of manhours by category, the above rates are adjust for the actual manhours.
The adjustment is made because the above rates are calculated on the basis
of 8 hour working day but the actual hours for allocation are measured
based on the actual working hours. That is, if the rate is $13 per hour
and the actual average working day is 6 hours, the rate to be applied
becomes $13 x‘% = $17.33.

Requesting Department Splits

Data Processing Expense is finally charged to burden centers. However,

as a step, requesting departments are designated to each job code.
Approximately 20 jobs out of over 300 have more than one requesting
department. The determination of the allocation between these departments
is rather arbitrary like 33Z for Manufacturing, 33Z for Sales, and 34%

for Administration. The Cost Department develops splits annually.
Development of All Data Processing Expense by Job Code by Requestinngepartment
Adjusted Rate x Actual Manhours by Job Code = Expense by Job Code

Computers Expense is consolidated with "others" at this step. As a
results, the aggregate Data Processing Expense is prorated into each
requesting department on the predetermined basis. A requesting department,
which has the responsibility for cost control, can check not only this
total expense of each job but the expense breakdown by function (Such as
Programming, Printer, Online, etc.) of that. The data can contribute to
cost reduction in the direct way such as the change from online to line

Printer, for example.

223

Allocation of Data Processing Expense to Charged Departments

 

Data Processing Expense must be charged to each burden center as a
cost for the departmental cost control purpose. Furthermore, this

cost must reflect the actual cost incurred in data processing centers.

Charged Department Splits

 

The accumulated expense by job code in each requesting department

is allocated to burden centers. Cost Department develops prorate
percentages for this purpose annually, through Budget Prorate Study.
The basic idea of percentage development is that the specific job
expense should be allocated to burden centers following the extent that
these centers receive benefits through the job. As a practical method,
volume oriented allocation and expense oriented allocation are most

common 0

Allocation Of Expense

 

The allocation is made by using the charged department splits, by job

code on the requesting department basis.

Allocation of Expense by Data ProcessingnExpense Sub Account

The expense is rearranged by charged department in order to meet the
need of accounting, where the breakdown is made by sub account. Charge
amount is totaled by charged department and percentage by charged
department is gained, which is to be used as the historical data for

prorate percentage in budgeting.

"mmmES