|
“Common
... and Global:” Choosing Standard Artifacts and Processes
for ERP Systems
Daniel
O’Leary
University
of Southern California
Marshall
School of Business
Leventhal
School of Accounting
Los
Angeles, CA 90089-1421
oleary@rcf.usc.edu
Abstract
Implementation
of enterprise resource planning (ERP) systems has led to
the motto, “Common ...
and
Global.” As a result, employing standard ERP systems
has led to firms adopting common
artifacts
(e.g., vendor numbering systems) and processes (e.g., sales
management) for global
implementation.
Although there are a number of important reasons for adoption
of standards in
an ERP
system, unfortunately, adoption of a standard may have differential
costs for divisions.
What
is optimal for one division or corporate is not optimal
for each of the divisions.
As a
result, there is concern about how standard artifacts and
processes are chosen for ERP
systems.
Is it possible to find standard artifacts and processes
that are optimal for the corporation
and
each of the divisions? What choice process for ERP standards
are rational? How can choice
processes
for selection of ERP standards be manipulated? This paper
addresses these issues
using
analytic methods and information gathered from a number
of ERP implementations.
Keywords:
Enterprise Resource Planning (ERP) Systems; Choosing Standard
Artifacts and
Processes
for ERP Systems;
1. Introduction
Enterprise
Resource Planning (ERP) software is one of the most rapidly
growing software
phenomena.
It has been estimated by the Gartner Group that the Fortune
1000 is implementing
or has
implemented an ERP system (Michandi and Digrius 1997). Further,
ERP are a significant
portion
of major consulting firm revenues. For example, roughly
50% of Price Waterhouse’s
consulting
has to do with ERP implementation (Public Accounting Reports
1997). Currently, the
ERP
market is dominated by Baan, J.D. Edwards, Oracle, PeopleSoft
and SAP, with estimates
indicating
that SAP has over 30% of the market (Busse, 1997). It is
because of this wide spread
impact
that ERP are beginning to be integrated into both curriculum
and research.
One
of the critical sets of decisions that needs to be made
in any ERP relates to the choice
of standards
used in the system, including standard artifacts (e.g.,
vendor numbering schemes and
lists)
and business processes (e.g., order management). The expression
“common ... and global”
has
received attention as a guideline to help implement those
standards in some companies (e.g.,
CIO
1996), forcing the notion that each division or plant will
not chose their own standards, but
instead
the company will use a similar set of standards across each
division. Unfortunately,
adoption
of a standard is not necessarily optimal for all divisions
or even the company as a
whole:
Particular standards may not benefit each division or plant
equally. As a result, it is
important
to investigate how companies chose standard artifacts and
processes, if it is possible to
find
standard artifacts and processes that are optimal for the
corporation and each of the
divisions,
determine which choice process for ERP standards are rational,
and how choice
processes
for selection of ERP standards be manipulated?
This
Paper
This
paper proceeds as follows. Section 2 provides some background
information on ERP
systems.
Section 3 investigates why ERP systems use standard artifacts
and business processes,
drawing
on evidence from ERP implementations, from the literature.
Section 4 analyzes how
firms
generate a portfolio of potential standards and choose between
them, drawing on case
discussions
of two companies. Section 5 investigates the potential for
and evidence of conflict
between
divisions and the company over choices of ERP standards.
Sections 6 and 7 contain the
analytic
results of the paper. Section 6 lays out notation and conditions
for “rationality.” Section
7 discusses
a number of alternative decision making approaches that
companies could use in
choosing
standards for ERP systems. Section 8 discusses some implications
of the choice rules
and
investigates issues such as agenda setting and strategic
choice making. Finally, section 9
provides
a brief summary of the paper.
2. Background:
ERP Systems
ERP
software allows integration of multiple divisions and plants
within an enterprise, typically
employing
a relational database system to ensure that all users have
access to the same
information,
in order to improve company responsiveness to customer needs.
For example,
Owens
- Corning (White et al. 1997) chose to implement the ERP,
SAP to facilitate visibility of
the
data and allow quick response to customers.
Up until
now, customers called an Owens - Corning shingle plant to
get a load of
shingles,
placed a separate call to order siding, and another call
to order the
company’s
well-known pink insulation.
(The
company’s new vision was that) Owens - Corning should
offer one
call
shopping for all the exterior siding, insulation, pipes
and roofing material that
builders
need. (SAP’s) R/3 will give Owens-Corning the ability
to make that
happen
by allowing sales people to see what is available at any
plant or warehouse
and
quickly assemble orders for customers.
Historically,
much corporate software has been customized and local, so
that individual
divisions
and plants could exploit unique characteristics. Artifacts,
such as vendor lists and
charts
of accounts, were chosen and developed with local processes
and decision making needs
in mind.
In addition, business processes, such as sales management
or purchasing, often were
chosen
to meet local needs and resource constraints. Unfortunately,
disparate software, artifacts
and
processes made it difficult to integrate across multiple
divisions. As a result, it became
increasingly
difficult to coordinate divisions to meet global customer
needs and competition.
Contrary
to customized local software, ERP package software is developed
by a software
company,
facilitating sale to many companies. Further, unlike customized
software, common
ERP
package software allows integration across all divisions
and plants in a company,
facilitating
a common corporate view. As noted by Brownlee (1996, p.
R18), “When Colgate
employees
log on to the network, the same menu options appear on their
screens, regardless of
whether
they’re in Cambridge or Burlington or New York.”
Since
ERP systems are developed for the enterprise, an ERP implementation
requires
many
enterprise decisions, including which standard artifacts
and processes will be used. For
example,
in the case of “Vandelay” (McAfee and Upton,
1996), a number of standards were
adopted,
including a common chart of accounts, common vendor numbers
and common part
numbers,
in order to facilitate communication and coordination between
different business units,
such
as factories and divisions. Unfortunately, some standards
benefits particular divisions more
than
other divisions. Some standards may be optimal for the company,
and not for all divisions.
As a
result, an important part of implementing an ERP system
is developing approaches to chose
standards.
3 Why
Standards for ERP? -- Some Evidence of Benefits
Why
do firms adopt a single set of standard processes or system
artifacts when they implement
an ERP
system? This section summarizes a number of reasons gathered
from a range of ERP
implementations
including, the software requires it, to get control of processes
that seem to have
gotten
out of control, because of the need for a common view of
the organization, and value
creation
and cost reduction.
Software
Requires It
Perhaps
the most immediate reason that standard ERP artifacts and
processes are adopted is
because
ERP software requires it. As noted by Chris Roon, a Vice
President of Red Pepper
Software
(Vaughn 1996, p. 72) “Whether its steel, rubber, or
electronics, with SAP it is a
common
set of parameters that drive individual plants.” Further,
the need for standard product
lists
and standard price lists also was seen in the SAP implementation
by Owens - Corning
(White
et al. 1997)
...
Owens-Corning traditionally had operated as a collection
of autonomous fiefs.
“Each
plant had its own product lines,” says Domenico Cecere,
president of the
roofing
and asphalt units. Each plant also had its own pricing schedules,
built up
over
the years of cutting unique deals with customers. ... (SAP’s)
R/3, however,
effectively
demanded that Mr. Cecere’s staff come up with a single
product list
and
a single price list. (bolding added)
Get
Control of “Out of Control” Processes
In some
cases different organizational locations evolve processes
that appear to get out of
control.
In order to get those processes back under control and to
facilitate organizational
adoption
of best practices, firms adopt ERP systems. As seen in Vandelay
(McAfee and Upton
1996,
p. 4-5)
Vandelay’s
sites’ operations practices were as varied as their
information systems.
There
was no uniformly recognized “best” way to invoice
customers, close the
accounts
at month end, reserve warehouse inventory for a customer
order or carry
out
an of the hundreds of other activities in the production
process that required
computer
usage or input. ...
To alleviate
... problems with systems and practices, Vandelay decided
to
purchase
and install a single ERP system, which would incorporate
the functions
of all
the previously fragmented software. The company would also
standardize
practices
across sites.
Common
View of the Data
In order
to generate a common view of the data a common set of organizing
artifacts, such as
chart
of accounts, vendor lists and customer lists are required.
Those artifacts provide the ability
to gather
information in disparate settings in order provide a common
view across each of those
settings.
For example, as noted by Vaughn (1996, p. 74),
...
Elf Atochem North America Inc., Philadelphia ... is moving
13 business units
over
to SAP software. ... Elf Atochem came to SAP because its
various companies
had
been reorganized to work as one. (As a result, the company)
... had inherited
“a
lot of different computer systems, a lot of different ways
of doing business, and
a lot
of hand-offs.” A common view of diverse data was important
...
Value
Creation and Cost Reduction
Implementing
standard artifacts and processes can create value and reduce
costs. For example,
as noted
by Pirelli’s director of information technology Arrigo
Andreoni, (Wahin, 1998, p. 48)
“The
more standardization there is, the easier it is to implement
new ideas and respond to new
opportunities.”
In addition, Andreoni notes that standardization can reduce
costs. As an
example,
before standardization, Pirelli had a full service back
office and customized software in
each
of five countries. ERP software was used to replace the
multiple back office staffs with a
single
back office staff in Switzerland, cutting costs by 25%.
Now offices in each country send
the
data to Pirelli’s central servers.
4. Generating
and Choosing Alternative Standards
Where
do the standard artifacts and processes come from? An interview
with the ERP project
manager
Nestle’s (US) found how Nestle’s generated different
business process options for its
SAP
implementation. First, Nestles’ decided that standardization
was appropriate, choosing to
implement
each of the same business artifacts and practices required
for SAP implementation in
all
three of its US divisions. Second, each of the three division’s
existing artifacts and practices
became
candidates. Third, both SAP and the consultant’s best
practices databases were used to
generate
candidates. Fourth, a multifunctional team used both sets
of inputs to decide on
company
standard artifacts and business processes.
Which
of the portfolio of generated artifacts and processes were
chosen? At Nestle’s, in
some
cases, existing division practices were adopted. In other
cases hybrid systems were
developed
and adopted. Finally, in some cases existing consultant
and SAP best practices were
adopted.
An interview
with the Deputy ERP project manager at Litton Data Systems
Group found
that
a similar approach was used to generate the portfolio of
standards for its Baan
implementation,
but that the source of standards was somewhat different
because of industry
considerations.
As with Nestles, a multiple function and multiple departmental
group was
founded
to generate choices. However, in addition to other sources,
Federal Government-based
standards
and artifacts were also a part of some of the portfolios
of standard artifacts and
processes.
Which
Standards will be Adopted?
Since
ERP committees generally include representatives from each
of the effected
divisions,
divisions can express their preferences as to which standard
is adopted through their
representatives
on these standard generating and choosing committees. Divisional
preferences
may
be driven by any of a wide range of motives, including some
of the following.
Maximizing
Corporate. Ideally, each division will put aside its individual
goals and work toward
developing
a set of standards that maximize the global needs.
Minimizing
Divisional Change Costs. Alternatively, a division may work
to keep its own
artifacts
or processes in order to minimize change costs such as training
personnel who would
use
the new standards and hiring different personnel to use
and implement the new standards.
However,
if a division is unable to promote its own standards, as
standards for the company, then
its
change and change costs are probably minimized by choosing
the standard that is closest to
their
own.
Responding
to Competition. Global competition has forced many companies
to outsource many
activities
that used to be done by corporate divisions. Any division
within a company faced with
this
potential threat may choose to work to respond to that competition
by doing extensive
reengineering.
Accordingly, in such settings we might see divisions work
to implement
substantial
change in the standard artifacts and processes (e.g., Hammer,
1989).
It is
not the purpose of this paper to choose between these or
other motives. Instead, the
purpose
is to provide evidence, in conjunction with the following
section that companies need to
think
about how to choose these common and global standards. In
addition, these varying
motives
suggest that there are incentives for divisions to try to
get the corporation to choose
standards
that benefit particular divisions.
5. Evidence
of a Conflict of Needs: Global vs. Local
What
is adopted for global usage is not always what is best or
preferred locally. For example,
Vaughn
(1996, p. 72) quotes Chris Roon, a Vice President of Red
Pepper Software, who admits
that
standard ERP artifacts are “... useful where financial
viewers want to consolidate
information
across diverse operating units, but ... the common view
may not be optimum for
individual
divisions.” Although standardization coming from implementation
of enterprise
software
by standardizing processes and artifacts has global benefits,
it comes from sacrificing
local
customized capabilities.
In addition,
because of differential benefits to divisions and corporate,
arising from
standard
artifact and process choices, the group decision making
can become a political process,
subject
to strategic decision making behavior designed to meet division
needs. Ultimately,
differences
in interests between global views and division views can
result in conflicts. Further,
choices
of standard artifacts and processes do not necessarily always
maximize corporation’s
utility,
but may benefit particular divisions, differentially.
This
section finds that there is both the potential for conflict
between divisions over
choices
between standards and evidence of conflict over the choices
that have been made.
Example:
A Common Standard Artifact -- Product Lists
As an
example of a common standard artifact consider the product
list. A product list gives a
unique
identifier to each product for purposes of entering and
reporting data about those
products.
If a common product list is adopted across all divisions,
then that means each division
uses
basically the product list. Typically, a product list must
provide sufficient flexibility to add
new
products and to capture each of the existing products. Unfortunately,
information needs are
not
necessarily the same in each division of a company, since
divisions have different products,
markets,
etc. Some products are likely to be used in only one division;
some divisions are likely
to need
more extensive and detailed product lists than others, etc.
As a
result, in each case, different business divisions and units
would each experience a
different
set of costs of implementing a standard product list, depending
on whose product list
was
chosen. If a particular division’s product list were
used as the common standard across the
company
then that division would face minimal training and implementation
constraints.
Further,
the product list may fit its interests better than other
product lists, facilitating addition of
new
products, etc. However, for other divisions it would be
an entirely new product list,
requiring
training and may not fit that division’s needs.
Further,
divisions may find that the common product list is not as
efficient as their
previous
one. For example, a product list with twice as many digits
can take twice as long to
enter
those digits for a particular product. Further, the existence
of more digits is likely to mean
a higher
probability of a data entry errors, suggesting that the
quality of the information could
decrease.
Conflicts
Occur Over Artifacts and Processes
ERP
systems demand common artifact and processes. However, as
noted above,
divisions
have differential returns associated with particular standards.
As a result, conflict
occurs
between divisions over choices for both artifacts and processes.
For
example, as seen in the SAP implementation by Owens - Corning
(White et al. 1997),
there
apparently was substantial conflict over an artifact R/3,
however, effectively demanded that
Mr.
Cecere’s staff come up with a single product list
and a single price list. The staff initially
fought
ceding control over pricing and marketing to a computer-wielding
central command. “My
team
would have killed if we’d let them,” he says.
Further,
as noted in a discussion about Owens - Corning (CIO, 1996)
obtaining common
and
global processes was not easy, and apparently resulted in
some conflict over how much
divisions
would be able to represent their individual needs in the
ERP.
“You’ll
always have variations in processes at the business unit
level with SAP,
particularly
around customers ... its my job to make sure the variations
are the
exception
rather than the rule. It’s a constant struggle.”
Since
there is evidence of both the opportunity for conflict and
the existence for conflict,
there
is reason to pursue analysis of rational choice of standards.
Rational group choice of ERP
standard
artifacts and processes is the subject of the next two sections.
6. Choice
Between Standards: Some Notation and Rationality Conditions
1
How
do organizations implementing an ERP system make the choice
between different standards
(artifacts
and processes)? Before this issue is examined, consider
some notation and some
rationality
conditions.
Let
x and y be different standards that could be adopted by
the ERP implementing
company.
For example, x and y could represent two different charts
of accounts or two different
vendor
numbering schemes or two different definitions of (common
and global choices) business
processes,
for say, order management. Let xRy indicate that x is preferred
to or indifferent to y.
R is
used to stand for the knowledge of relationships between
all x and y. For example, R can be
specified
through a utility function à where xRy if and only
if à (x) > à (y).
Let
S = (x, y, z, ...) be the set of alternatives available
for some artifact or business
process
being chosen for the ERP system, termed the “environment.”
S is the set of alternatives
from
which the choice will be made. Let |.| indicate cardinality,
and let |S| > 3.
Suppose
that there are n-1 divisions and a “corporate office”
of the company
implementing
the ERP system. Each will be treated as one of n divisions
in the set C, the
company.
Each division is assumed to have preferences between the
choices made for an ERP.
So that
each division is assumed to have a preference for artifacts
such as a chart of accounts and
business
processes. Let Rj represent the preference relationship
of company j and Ãj be the utility
function
of company j. Let |xR jy| be the number of divisions for
which xRjy, for j ? C.
Rj is
assumed to be both complete, reflective and transitive among
choice between
different
standards that could be adopted.
Property
1 (Completeness): Either xRjy or yRjx ? x and y ? S.
Each
division is assumed to have a preference between any particular
pair of
business
artifacts or process choices, e.g., purchasing, that divisions
could chose
between.
A lack of completeness can translate to an apparent lack
of
“decisiveness,”
since a lack of completeness makes it appear that the group
cannot
make
up its mind (e.g., Pattanaik 1978).
Property
2 (Reflectivity): xRjx ? x ? S.
In the
case of reflectivity, for each division, the comparison
between any
particular
business artifact or process is indifferent to itself.
Property
3 (Transitivity): xRjy, yRjz, implies xRjz ? x, y and z
? S.
Suppose
that there are at least three definitions of a business
artifact or processes
for
some activity requiring standard ERP definition across divisions.
Transitivity
assumes
that for each division there is a preference between each
pair, and that
one
is preferred among the three.
If Rj
satisfies properties 1-3 then it is called a weak preference
order relationship. Throughout
the
choice of ERP standards, each of the divisions j is assumed
to be rational, adhering to
conditions
1, 2, and 3.
7. Approaches
used to Choose Standards
As noted
above there are a number of sources of ERP standards. Further,
the cost to
individual
divisions will vary based on which standards are adopted.
How
do organizations choose which set of standards, e.g., chart
of accounts, to adopt?
Which
approaches are used in real world applications? As noted
by McAfee and Upton (1996)
companies
have some questions as to how to chose standards for ERP
systems. For example, in
McAfee
and Upton (1996) the project manager is quoted, as having
a strong bias toward “input
by many,
design by few,” but did not know how to put that statement
into practice. How do we
gather
input by many? What group choice rules can we use? Are those
methods “rational?”
This
section investigates four classic approaches, with evidence
of their use in ERP system
development,
including “majority votes,” “Borda rules,”
“Pareto optimality,” and “dictatorship.”
Majority
Votes
According
to Wakin (1998), Pirelli’s director of information
technology, uses what he
calls
“democratic governance” across divisions to
achieve standardization in core business areas
required
for ERP implementation. Although democratic governance can
take many forms, it is
suggestive
of a “majority votes” approach - whichever standard
the majority of divisions prefers
becomes
the standard.
Let
xRCy indicate that the company (“C”) as a whole
prefers x over y.
Choice
Rule 1 -- Majority Votes
xRCy
if and only if ?
j ?C|xRjy|
> ?
j ?C
|yRjx|, ? x,y ? S.
A choice
rule of majority votes is one where the company adopts standard
x over standard
y if
more divisions prefer standard x over standard y. Unfortunately,
the majority votes rule is
not
generally transitive. For example, suppose there are three
different divisions each with their
own
different business process for billing, b1, b2 and b3. The
lack of transitivity is seen in the
following
choices:
(b1)R1(b2)
,(b2)R1(b3) ; (b2)R2(b3) ,(b3)R2(b1) ; (b3)R3(b1) ,(b1)R3(b2)
Since
two of the three divisions prefer b1 to b2, (b1)RC(b2).
Similarly, (b2)RC(b3), and
(b3)RC(b1).
As result, RC is not transitive.
Borda
Rule
An alternative
way to gather input from many regarding standards artifacts
and processes
is to
have a panel establish a ranking of the options and then
and then give a certain number of
points
for first place, etc. For example, if there are five different
ways to execute the business
process
“reserve inventory for clients” then each panel
member would rank each of the
alternatives,
first place would get five points, second place four, etc.
This is known as the Borda
rule
(e.g., Fishburn 1971).
An interview
with the Deputy Project Director of Litton Data Systems
Division (LDSD),
PeopleSoft
implementation found that LDSD used a Borda-like approach.
First, LDSD defined
“Key
Aerospace and Defense Features and Thrusts” including
Defense
Business Processes
•
Government Forms
•
Progress Payments
•
Moving Average Actual Costs ...
Financial
Management
•
Division Budgeting and Resources Planning
•
Cost Schedule Control
•
Multiple Rates for Labor and Burden
Second,
given these and other categories, different ERP vendors
were evaluated and ranked
on their
ability to provide each feature, with points given for their
particular rankings. Third,
choice
of ERP system was made based on the relative ranking.
Choice
Rule 2 -- Borda Rule
Define
a rating function r(x|S, Rj ) = Ê(y ? S| yRjx), where
Ê(.) is conditional cardinality, and
the
number of alternatives that division j finds at least as
good as x. The Borda rule is xRCy if
and
only if ? (j ? C) r(x|S, R j) > ? (j ? C) r(x|S, R j)
? x, y ? S
In order
to illustrate the Borda rule, consider Fishburn’s
(1971) example
Division
Preference Order
1 b1;
b2; b3; b4; b5
2 b2;
b3; b5; b4; b1
3 b5;
b1; b2; b3; b4
4 b1;
b2; b4; b5; b3
5 b2;
b4; b3; b1; b5
The
Borda quantities are, respectively, 17, 21, 13, 12, 12,
for b1, b2, b3, b4, b5. As a result, the
choice
for the corporation as a whole would be b2 since it has
the highest total. Unfortunately,
none
of the options has a majority of the possible 65 points.
As a result, next suppose that the
least
preferred options b3, b4, and b5 are removed from consideration.
In this situation, the new
preference
order would be as follows.
Division
Preference Order
1 b1;
b2
2 b2;
b1
3 b1;
b2
4 b1;
b2
5 b2;
b1
In this
reduced set b1 would be chosen (8 compared to 7). The Borda
rule provides a
group
choice rule that meets conditions 1-3 and so has a weak
preference order. However the
Borda
rule violates the following property.
Property
4 (Independence of Irrelevance Alternatives).
Any
two set of the rankings must have identical corporate choices
in S or subsets of S. In
particular,
a choice rule has independence of irrelevant alternatives
if and only if xRCy (for x,y ?
S’,
? S’ ? S) implies xRCy (for x,y ? S).
Pareto
Optimality
As noted
above (Wakin 1998), Pirelli’s director of information
technology uses what he
calls
“democratic governance” to achieve standardization
in core business areas required for ERP
implementation.
Pareto optimality establishes ordinal preference between
pairs of ERP
standards,
determining which preferences all of the divisions agree
on. As a result, another way
to achieve
democratic governance in the choice of ERP standards, is
Pareto optimality, if it
exists.
Choice
Rule 3 -- Pareto Choice Rule
xRCy
if and only if xRjy for all j ? C and ? x, y ? S.
Property
5 (Pareto Optimality)
A choice
rule is Pareto optimal if and only if satisfies Pareto’s
choice rule.
Unfortunately,
Pareto’s choice rule is not complete. Using Fishburn’s
example, the only
pairs
that have defined Pareto relationships are (b2)RC(b3) and
(b2)RC(b4). Thus, in some cases,
such
as the current example, groups will not be able to establish
relationships between pairs of
standards
for an ERP.
Dictatorship
In other
settings it is apparent that rather than a majority rules
approach, the solution used in
choosing
standards for an ERP system is closer to a “dictatorship.”
For example, a discussion of
Owens-Corning
(CIO, 1996) notes that “it would be relatively easy
for the Toledo-based team to
huddle
and make SAP process choices for the company unilaterally.”
Further,
Colgate (Brownlee, 1996) has implemented a network of “supplier
managed”
inventories,
for both its customers and suppliers, using SAP. Colgate
provided a number of its
most
important suppliers with SAP so that they can directly access
Colgate’s SAP system. As
part
of the trade-off, Colgate does not have to pay for product
ingredients it until it actually uses
them.
In this setting Colgate specified many of the business processes,
and provided standards in
terms
of various pricing lists, etc. A division k is likely to
be a “dictator” if it is the corporate
office
division or a large division with substantial relative power.
Choice
Rule 4 (Dictatorship)
Suppose
the company chooses division k’s preference, regardless
of any other division’s
preferences,
so that k is the dictator, i.e., xRCy if and only if xRky,
? x,y ? S.
Arrow’s
Impossibility Theorem
A question
for firms implementing ERP systems is which of the choice
rules meet the properties
1 -
5. Of the four choice rules, majority rules, Borda, Pareto
optimality and dictatorship,
unfortunately
only dictatorship satisfies properties 1 - 5. This result
is known as Arrow’s
impossibility
theorem (Arrow 1963).
This
means that except for those settings where dictatorship
is used, ERP standard choice
will
not include choice methods that are “rational”
as defined above. Further, as seen in the next
section,
unfortunately, virtually all choice rules are subject to
manipulation.
Discussion
Earlier
in the paper two questions were asked that can now be answered.
First, is it possible to
find
standard artifacts and processes that are optimal for the
corporation and each of the
divisions?
Arrow’s impossibility theorem indicates that in general
the answer is no. Second,
what
choice process for ERP standards are rational? There is
only one choice process presented
here
that is rational, dictatorship (e.g., by the project leader).
Although dictatorships were a
concern
to Arrow (1963), with for profit corporations, equity is
probably not an issue. However,
dictatorship
does not insure a portfolio of artifacts and processes that
is optimal for the
corporation,
it simply ensures a “rational” solution.
8. Implications of Group Choice Rules
Not
only do the choice rules (other than dictatorship) not meet
conditions 1-5, but the choice
rules
are also subject to manipulation through agenda setting
and strategic voting behavior.
Agenda
Setting as an ERP Implementation Strategy
The
impossibility theorem has some implications for the types
of negotiations that can take place
between
agents in the choice of standards. As noted by McKelvey
(1976), so called "agenda
setting"
can influence the choice of alternatives that ultimately
are adopted if pairwise
comparison
and adoption is done. Suppose that xR(C)y, yR(C)z and zR(C)x,
that is on a pairwise
basis,
x is preferred to y, y is preferred to z, and z is preferred
to x. If the agenda is set (x versus
y) versus
z then pairwise choice leads to z . However, if the agenda
is set as (y versus z) versus x
then
the choice leads to x .
ERP
standard adoption processes that employ pairwise analysis
are subject to issues such
as "agenda
setting." For example, if there were three product
lists used in the divisions of a
company,
and the company was to chose one, we could see agenda setting
determine which was
chosen,
if they were evaluated on a pairwise basis. As a result,
ERP system implementers and
developers
need to be particularly sensitive to using pairwise comparisons
to choose standards.
In addition,
the choice of which rules to follow in ERP standard adoption
can result in
manipulation
of solutions. For example, in the Borda example, if division
2 or 5 were setting the
rules
they would prefer to have the decision set be the entire
set C, since the use of C generates a
solution
of b1. Whereas, divisions 1, 3 and 4 would prefer requiring
a majority of points, then
using
the reduced set, in order to generate their preferred solution
of b2. Control of the agenda
can
lead to control of the choice of the standard ERP artifact
or process.
Not
all group choice processes are subject to agenda setting.
For example, “majority
votes”
has the same outcome has the same result no matter what
order is used (Condorcet 1785).
Strategic
Behavior
In addition
to agenda setting, divisions may exhibit “strategic
behavior” by voting or ranking
standard
choices in order to facilitate adoption of the standards
that they prefer.
Strategic
behavior can invalidate group choice methods by eliminating
feasible solutions
or change
outcomes. Feasible solutions are easily invalidated when
Pareto Optimality is the
criteria
- it only takes one division to invalidate a Pareto Optimal
pair. In the above example
discussion
for Pareto Optimality, only division 3 is limiting a preference
relationship between b1
and
b3, b1RCb3. As a result, if Pareto optimality is used as
a criteria, we might find divisions
developing
their rankings strategically to limit certain choices. In
addition, if a division sees that
other
divisions are voting for a particular choice, then it can
alter the outcome with its own votes,
voting
for choices it does not prefer in order to see options that
it prefers least are not chosen,
e.g.,
in majority votes or Borda.
Interestingly,
dictatorship is the one choice rule that is not subject
to manipulation. Since
which
ever standard process or artifact the dictator chooses is
adopted by the firm, it is assumed
that
choice rule is not manipulatable.
Discussion
Earlier
the question was asked, how can choice processes for selection
of ERP standards be
manipulated?
In this section it is clear that both agenda setting and
strategic voting can be used
to manipulate
which artifacts and processes are chosen for the ERP system.
Unfortunately, those
same
issues again illustrate the importance of the project leader.
9. Summary
ERP
systems offer a number of benefits deriving from the adoption
of standard artifacts
and
processes, including control of out of control processes,
a common view of corporate wide
data,
value creation and cost reduction. As a result, firms that
adopt ERP systems generally
standardize
certain system artifacts (e.g., vendor numbering systems)
and processes (e.g., sales
management).
Typically
companies generate a portfolio of standards to choosing
from existing company
artifacts
and practices, best practices and hybrid approaches that
combine company and best
practices.
Unfortunately, it is not clear which of a portfolio of artifacts
and standards should be
adopted,
particularly since adoption of artifacts and best practices
can benefit certain divisions
more
than others. As a result, it is important to examine how
such ERP standards choices are
made
and how rational decisions about ERP standards can be made.
This
paper used analytical models to find that, besides dictatorship,
there are no methods
that
allow rational choice between a portfolio of ERP standards
that will be optimal for a
company
and a division implementing an ERP system. In addition,
virtually all such group
choice
approaches are subject to manipulation through agenda setting
and strategic behavior.
Although
“common ... and global” seem to guide ERP development,
it is important to remember
the
limitations of trying to adopt a common set of artifacts
and processes for an ERP system in a
multiple
division company. Companies can do “common ... and
global”, but there will be
differential
benefits to divisions. Arrow’s impossibility theorem
shows the limited possibilities
for
generating a solution using a rational approach or one that
is not subject to manipulation.
References
Arrow,
K., Social Choice and Individual Values, Cowles Foundation
Monograph 12, New York,
Wiley,
1963
Brownlee,
L., “Overhaul,” The Wall Street Journal, Wednesday,
November 18, 1996, p. R12
and
R. 17.
Busse,
T., “Known World-Wide SAP Contemplates its Future
Domain,” Information Week,
1997.
CIO,
“SAP Key Roles,” http://www.cio.com/forums/061596_sap_roles.html
Condorcet,
Marquis de, Essai sur l’application de l’analyse
a la probabilite des decisions
rendues
a la pluralite des vois, Paris, 1785.
Fishburn,
P., A Comparative Analysis of Group Decision Methods,”
Behavior Science, Volume
16,
Number 6, November 1971, pp. 538-544.
Hammer,
M., “ “Reengineering Work: Don’t Automate,
Obliterate,” Harvard Business Review,
July/August
1990, pp. 104-112
Kleindorfer,
P., Kunreuther, H., and Schoemaker, P., Decision Sciences:
An Integration
Perspective,
Cambridge, Cambridge University Press, 1993.
Michandi,
V. and Digrius, B., “The SAP Implementation Marketplace,”
Gartner Group, RSAP-
104,
1997.
McAfee,
A. and Upton, D., “Vandelay Industries,” Harvard
Business School, 9-697-037, April
16,
1997.
McKelvey,
R., “Intransitivities in Multidimensional Voting Models
and Some Implications for
Agenda
Control,” Journal of Economic Theory, Volume 12, 1976,
pp. 472-482.
Pattanaik,
P., Strategy and Group Choice, North Holland, Amsterdam,
1978.
Public
Accounting Reports, “How Do the System Integrators
Rate?,” Public Accounting
Reports,
March 15, 1997.
Vaughn,
J., “Enterprise Applications,” Software Magazine,
May 1996, pp. 67-73.
Wakin,
E., “Global Strategies Drive Pirelli,” Beyond
Computing, January/February 1998, pp.
46-48.
White,
J., Clark, D. and Ascarelli, S., “This German Software
is Complex, Expensive and
Wildly
Popular,” Wall Street Journal, March 17, 1997, p.
1.
|
