You are on page 1of 10

I N F O R M S

Transactions on Education
Vol. 8, No. 2, January 2008, pp. 6574
issn1532-0545 08 0802 0065
informs

doi 10.1287/ited.1070.0003
2008 INFORMS
Teaching Problem Solving Techniques to
MBA Students Through Supply Chain Management
Michael J. Fry
Department of Quantitative Analysis and Operations Management, University of Cincinnati,
Cincinnati, Ohio 45221, mike.fry@uc.edu
T
his paper describes the use of a supply chain management course to teach problem solving techniques to
MBA students. An introduction to problem solving techniques and a step-by-step methodology are presented
that can be used in supply chain management courses. We discuss two specic supply chain management cases
and provide examples of how these cases can be used to illustrate steps in the problem solving methodology.
We also describe how problem solving skills can be reinforced and assessed through case study write-ups. We
conclude with some feedback from students who have taken this class.
Key words: problem solving; supply chain; case studies
History: Received: August 7, 2006; accepted: August 13, 2007. This paper was with the author 5 months for
2 revisions.
1. Introduction
MBA students, and business students in general, are
notorious for their dislike and fear of quantitative
classes. One researcher describes this as a math pho-
bia among business students, explaining that, Busi-
ness students, although able, are often math phobic.
Courses should strive to lessen math phobia [and]
enable students to be comfortable with mathematics
(Lamoureaux 2000). Although the focus of this paper
will be on MBA students, the importance of teach-
ing problem solving skills is not unique to busi-
ness schools. A recent report from the Association of
American Colleges and Universities identied intel-
lectual and practical skills such as teamwork and
problem-solving as an essential learning outcome for
university students in the liberal arts (Association of
American Colleges and Universities 2005).
Traditionally, management science courses have
played the role of (hopefully) making MBA stu-
dents more comfortable with basic analytic skills
and problem solving techniques. However, Powell
(2001) notes that, Management science courses have
been removed from the core in a number of highly
respected schools. As a result, some rms, particu-
larly management consulting companies, are increas-
ingly hiring students trained in the hard sciences
(e.g., math, physics, and engineering) who then must
be taught business concepts, rather than trying to
teach problem solving skills to business students
(Montell 1999, Leonhardt 2000). Therefore, it would
seem imperative that we nd other ways in which to
broaden MBA students exposure to analytic methods
and strengthen their problem solving abilities.
Few of those in the elds of operations research
and management science (OR/MS), question the ben-
et of enhancing students problem solving abilities.
Outside of OR/MS we are also seeing strong sup-
port for creating MBA graduates with strong analytic
abilities. BusinessWeek rankings of top MBA programs
lauded the ability of the top programs (namely North-
western University and the University of Chicago)
to instill analytical skills in their graduates (Merritt
and Chambers 2004). A recent Harvard Business Review
article describes the increased use of analytics as a
competitive advantage in a wide range of industries
and job functions (Davenport 2006). The January 23,
2006, cover story from BusinessWeek discusses the
ever-increasing importance of math skills in the busi-
ness world (Baker and Leak 2006). Thus, it is all the
more ironic that some schools are scaling back their
basic management science core courses. However,
the news is not all bad for OR/MS courses. Supply
chain management (SCM) courses have continued to
increase in popularity and number of offerings, even
as management science core courses decline. This cre-
ates a clear opportunity for an alternative format in
which to introduce problem solving techniques to a
wide audience.
The growth of SCM in business schools has been
dramatic. Johnson and Pyke (2000) point to the fact
65
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
66 INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS
that as late as 1995 only a handful of universities
taught a course with the title supply chain man-
agement. Today, most top business schools at least
offer elective courses in SCM. Many have signicantly
altered their core operations management courses to
include signicant SCM content (Johnson and Pyke
2000). Several schools (including Michigan State Uni-
versity and Wright State University) have replaced
their core operations management class with classes
explicitly labeled as supply chain management. Per-
haps even more indicative of the prevalence of SCM
is the number of schools offering specialized mas-
ters degrees in SCM and even a few universities
(notably Pennsylvania State University and Arizona
State University) that have renamed entire depart-
ments to reect a wide emphasis on SCM.
Because SCM encompasses so many different
opportunities for tackling difcult problems (many of
which are rmly grounded in traditional operations
research concepts) SCM classes provide an excellent
opportunity for exposing our MBA students to basic
problem solving techniques that they may not be get-
ting in other classes. In this discussion we highlight
several ways in which basic problem solving tech-
niques can be included in an MBA-level SCM course.
We present a problem solving methodology that is
used to walk students through challenging problems.
We specically mention two cases that can be used in
a SCM course and discuss how each case study can be
used to illustrate the problem solving methodology.
For each case we detail the factors in the case that
are particularly suited for illustrating certain steps in
the problem solving methodology. The goal is to bet-
ter acquaint MBA students with the formal process
of problem solving that they will need in facing real
business problems. SCM-type classes present an effec-
tive means for introducing these basic concepts. We
also discuss how the problem solving methodology
can be reinforced through case study assignments,
and we offer some observations from our experiences
in such a problem-solving-focused SCM class.
2. Basic Problem Solving
Skills for MBAs
As others have noted (most notably Powell 1995,
2001; Liberatore and Nydick 1999; Grossman 2001),
the most important analytical techniques for MBA
students to learn are not necessarily specic nonlin-
ear solution techniques, the best heuristics for solv-
ing traveling salesperson problems, or even specic
inventory models. Rather, it is learning the basic con-
cepts of how to conceptualize, formulate, and model
difcult problemswhat Powell (2001) calls the art
of reasoning logically with models. As Evans (1997b,
p. 106) notes, A problem well dened is half solved,
Figure 1(a) Connect All Nine Dots with Four Straight Lines Without
Lifting Pencil or Retracing a Line
Note. This is a simple brain-teaser example that illustrates the importance
of explicitly dening constraints.
Figure 1(b) Completed Solution
Notes. Students often assume a constraint that does not exist that lines must
stay within the array of dots. See Figures 1.4 through 1.6 in Ackoff (1978)
for an ingenious, alternative solution to this problem.
which exemplies the importance of the conceptual-
ization steps in any problem solving methodology.
Therefore, it is essential to provide MBAs with at least
some of the basic fundamentals of how to conceptu-
alize and attack difcult problems. SCM courses pro-
vide an excellent opportunity for teaching these tech-
niques.
The rst several hours of our SCM course are spent
on introducing problem solving techniques, includ-
ing a step-by-step methodology, and then applying
these techniques to some simple problems. We present
material from Rubensteins Patterns of Problem Solv-
ing (1975), Ackoffs The Art of Problem Solving (1978),
and Giordano and Weirs A First Course in Mathemat-
ical Modeling (1985) to the students in the rst class
meeting. Specically, we walk the students through
several brain-teaser type examples and then use
these as illustrations for good and bad problem solv-
ing techniques. For instance, both Rubenstein (1975)
and Ackoff (1978) include the example shown in
Figures 1(a) and 1(b), where the reader is asked to
connect all nine dots with four straight lines without
lifting the pencil from the paper and without retrac-
ing a line. Rubensteins text provides several other
brain-teasers that can be used in an introductory
class.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS 67
The example in Figure 1 immediately illustrates the
importance of verbalizing the constraints of a problem
and not being bound by assumed constraints that are
not actually constraints at all. In this particular exam-
ple, the typical student assumes that the drawn lines
are conned to the boundaries of the array formed by
the dots; thus, they may often not be able to visualize
a solution. Such an example can be used to illustrate
what many refer to as thinking outside the box.
However, a more appropriate way to consider this
may be simply dene the box and consider all pos-
sible solutions within that box. The simple act of
explicitly dening the constraints often leads students
to identify and discard assumed constraints that do
not actually exist. In the words of Russ Ackoff, this
can prevent a problem that one cannot solve because
of a self-imposed constraint (Ackoff 1978, p. 9).
Examples such as the one above are used to moti-
vate the need for a structured approach to prob-
lem solving. We present the following list of steps to
the students, which are adapted from Steve Pollocks
notes for his Modeling Design Studio at the Univer-
sity of Michigan (others, including Stephen Powell at
Dartmouth College and Mike Magazine at the Uni-
versity of Waterloo and the University of Cincinnati
have taught similar courses).
1. Identify (dene) problem
2. State assumptions
3. Dene model (consider alternative models)
(a) What is the objective/output?
(b) What is the measure?
(c) What are the decision variables?
(d) What are the constraints?
(e) What are the important data for the problem?
4. Solve (consider an SIEsimplest interesting
example)
5. Verify
6. Validate
7. Implement
8. Monitor and continuously improve
After requiring the students to solve small, gen-
eral types of problems using their step-by-step pro-
cess, we introduce a slightly more realistic example
to demonstrate the methodology. This example uti-
lizes Goldratts Product Mix problem depicted in
Hopp and Spearmans Factory Physics (2001) instruc-
tor materials (Chapter 1). The basic problem is shown
in Figure 2 and consists of determining the optimal
product mix to be produced by a prot-maximizing
rm. There are two products sold by the rm, prod-
ucts P and Q. Product P is composed of three compo-
nents: P
1
, P
2
, and P
3
. Components Q
1
and Q
2
comprise
product Q. The components are purchased for the
costs shown in Figure 2. Each product follows a dif-
ferent production sequence on the available machines
Figure 2 An Illustration of the Goldratt Production Problem Included
in Factory Physics
Product P Q
Price (selling) $90 $100
Max weekly sales 100 50
Machines A, B, C, D
Machines run 2,400 min/week
Fixed expenses of $5,000/week
D
C C
A B
C
B
B
A
D
$5 $20 $20 $20
$20
25
10
15
5
15
5
20
15
10
15
P
P
3
P
1
P
2
Q
1
Q
2
Process
time
Component
cost
Component
Q
Note. This illustration provides a useful introduction to using a problem solv-
ing methodology and examples of various modeling techniques.
(labeled A through D in Figure 2) and requires dif-
ferent process times at the machines. Each machine
is limited to 2,400 minutes per week of production
capacity. The maximum weekly demand for product P
is 100 units, and the maximum weekly demand for
product Q is 50 units. Weekly overhead results in a
cost of $5,000 per week. The objective is to determine
the optimal amount of products P and Q to produce
that meets the processing time and sales constraints.
The correct solution to this two-product produc-
tion problem can be found through a simple linear
program (see Equations (1)(8), where X
p
denotes the
number of P products produced and X
q
the number
of Q products produced). However, the main concern
at this point is for the students to go through the step-
by-step process outlined above: to dene the problem
(determine production strategy), state the objective
(maximize prots), list the constraints (time available
on the machines, market demand limitations, etc.),
identify the pertinent data (cost values, etc.), etc. No
explanation of linear programming as a solution tech-
nique is given, but the students are asked to attempt
a solution anyway. In the most recent teaching of this
class, just less than half of the students identied max-
imizing the production of product Q (because it is the
more protable product) and then lling remaining
capacity by producing product P as a viable solution.
This heuristic solution suggests producing 50 units
of product Q and 43.33 units of product P, which
results in a weekly prot loss of $50. However, the
optimal solution is to produce 73.85 units of prod-
uct P and 36.92 units of product Qresulting in a
weekly prot of more than $538. Only two students
(out of the 24 enrolled students) immediately identi-
ed the correct mathematical model for this problem
and were able to explain why the heuristic solution
was suboptimal. We should note that all students
have been introduced to basic optimization proce-
dures including linear programming (LP) in a previ-
ous class. Thus, we observe that teaching students a
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
68 INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS
tool (such as LP) is not the same as teaching students
how to solve problems. Even though all students in
this class had seen LPs in a previous course, many
do not see this problem as an obvious place to apply
such a formulation.
max {(90 45)X
p
+(100 40)X
q
5,000] (1)
subject to 15X
p
+10X
q
2,400 (2)
15X
p
+35X
q
2,400 (3)
15X
p
+5X
q
2,400 (4)
25X
p
+15X
q
2,400 (5)
X
p
100 (6)
X
q
50 (7)
X
p
, X
q
0. (8)
When we do nally reintroduce the tool of lin-
ear/nonlinear programming through Microsoft Excel
Solver, the emphasis is on Solver as a tool with the
problem solving techniques as the important skills.
This includes emphasizing steps to verify and val-
idate the solution as well as explorations into the
effects of the given assumptions (through sensitivity
analysis). The students are told that such problem
solving techniques will be a continual emphasis of the
SCM class and that they are expected to demonstrate
such abilities for each of their required case study
write-ups.
3. Specic SCM Cases to Illustrate
Problem Solving Techniques
Johnson and Pyke (2000) provide an excellent list-
ing of cases that can be used in an MBA-level SCM
class. Many of these case studies can be presented in
such a way to emphasize problem solving techniques.
The two case studies discussed below are particu-
larly suited to analytical analysis and to emphasiz-
ing a problem solving methodology. Whichever case
studies are chosen, it is important that the instructor
continue to stress the importance of following a prob-
lem solving procedure and that students case study
write-ups include evidence of following such a proce-
dure. Students are often quick to abandon such pre-
liminaries as explicitly stating objectives, constraints,
and assumptions when diving into a complicated case
study. The instructor must be sure to emphasize that
such a procedure should be followed in all situations
and that evidence of such a procedure will be part of
a students evaluation.
3.1. Applichem (A)
The Harvard Business School Case Applichem (A)
(Flaherty 1985) describes a company, Applichem, that
manufactures a product known as Release-ease used
in the production of molded parts. Applichem can
manufacture Release-ease at six manufacturing plants
located around the world. Customer demand for
Release-ease occurs in six different regions. Data are
provided for the capacity limitations at each manu-
facturing plant and the expected customer demand in
each region. Detailed data related to manufacturing
costs (broken down to a very detailed level including
operating costs, raw material costs, utility costs, etc.)
and transportation costs are also given in the case.
Students are asked to decide how Applichem should
structure its global supply chain (how much Release-
ease should be produced at each plant and how
it should be allocated across the different demand
regions). The Applichem case is particularly useful in
illustrating the difculties encountered in such prob-
lem solving steps as dening an objective, determin-
ing appropriate measures for the objective, stating
assumptions, tracking down relevant data, and imple-
mentation issues.
Following our list of problem solving steps, stu-
dents tend to quickly identify the objective of the
Applichem case study as fullling customer demand
at the minimum total cost where costs include man-
ufacturing and transportation costs. Transportation
costs between each manufacturing plant and each
customer demand region are given, and customer
demand is generally assumed to be known and con-
stant. However, signicant complications remain in
identifying appropriate measures for the objective. In
particular, students struggle to dene the best mea-
sure for manufacturing cost. A thorough reading of
the case reveals that the manufacturing plants differ
greatly due to when they were built (the technology
they employ), the size of each plant, the abilities
and cost of the labor force that is employed, packag-
ing requirements, etc. Therefore, students are forced
to spend considerable time thinking about how the
objective of minimizing cost can be best measured.
What is the fairest cost measure for comparing
the different manufacturing plants? Should costs be
adjusted for packaging requirements or differences in
labor costs? If so, how?
As with all case studies, there are many assump-
tions that must be made to complete the analysis.
One issue that often comes up in students analyses
is how to separate out xed versus variable costs.
As production is scaled back at a plant, how does
this affect the manufacturing cost? There are suf-
cient data provided in the case so that students can
attempt to separate out many xed costs from vari-
able costs (e.g., indirect versus direct labor). Further-
more, we encourage students to consider factors that
might not be explicitly reected in their objective
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS 69
function statement that could impact their recommen-
dations and analysis. If the solution that results in
the minimum cost is to close one or more plants, are
there any other considerations that should be made?
Would they expect there to be costs incurred to close a
plant? Would they be substantial? What effect would
this have on employee morale and performance at the
remaining plants?
One interesting aspect of the Applichem case has
to do with data acquisition (step 3e in our problem
solving methodology). Because manufacturing plants
are located in different countries around the world,
exchange rate uctuations play a big part in the case
analysis. The case presents all costs in terms of 1981
U.S. dollars, and it includes exchange rate and ina-
tion data for the years 19771982. As students work
on their case analyses, they often express concern that
these data are quite limited (only six data points) and
that they are quite old. We strongly encourage the
students to supplement these data by nding more
recent values. Students are often surprised that they
are required to nd data not explicitly provided in the
case. Similar to the problem of connecting the dots
in Figure 1, students tend to dene the scope of the
problem as including only the data that are presented
in the case. We use this as an opportunity to con-
vey that nding useful data is often the most chal-
lenging aspect of many practical problems and that
one should not be constrained by using only data
that are easily available. We generally do not give
much guidance to students about where to nd such
data, but the International Financial Statistics Yearbook
from the International Monetary Fund is probably the
best source for more recent data (Lowe et al. 2002
contains real exchange rates for the Applichem case
through 1993). However, even once they locate the
data, questions will remain on what range of data are
relevant (e.g., How should they handle years when
a currency was pegged/unpegged? If the decision is
being made today, how many years worth of data
should be considered?).
Although demand is generally assumed to be con-
stant, the uctuations in exchange rates also provide
excellent opportunities to explore the effects of uncer-
tainty in decision making. Students generally grasp
that they can use a linear program to determine the
optimal shipping strategy to minimize costs once they
account for uctuations in exchange rates and ina-
tion in any given year. However, they often fail to
consider the dangers of solving for a single realization
of exchange rates and ination.
The most common optimal solution using the 1981
data included in the Applichem case (depending on
the exact assumptions made by the students) is that
the manufacturing facility in Japan should be closed.
However, this ignores the benet of reactive capac-
ity where the Japanese facility can be used as an
operational hedge if exchange rates and/or ination
move in such a way that Japan becomes a preferred
manufacturing option. This issue is examined in detail
in such papers as Huchzermeier and Cohen (1996),
Cohen and Huchzermeier (1999), and Lowe et al.
(2002). We use this opportunity to describe the basic
ideas of stochastic programming, recourse actions,
and reactive capacity. In some cases the class discus-
sion will lead to a more thorough consideration of real
options and evaluating the value of excess capacity.
Generally, some students will suggest a simpler
approach to handling the inherent uncertainty in
exchange rates and ination such as considering aver-
age values over time or using regression to predict
future values. This is an excellent opportunity to dis-
cuss the dangers of considering only expected, or
average, values from a complicated process. We draw
upon many of the examples used by Sam Savage in
his Flaw of Averages materials. Savage explains the
Flaw of Averages as the fallacies that occur from
using a single number, typically the average, to rep-
resent an entire distribution of possible uncertain out-
comes (Savage 2002). The Applichem case is a good
illustration of this because making a decision to close
a plant based on average conditions or a limited
set of outcomes ignores the benets of keeping excess
capacity as a form of operational hedge. This also
helps the students understand the importance of look-
ing at a distribution of outcomes instead of simply
the average outcome and can motivate the use of
what-if analysis. Cohen and Huchzermeier (1999)
generate distributions of outcomes for the Applichem
case using Crystal Ball to show the benets of keep-
ing the Japanese plant open as a form of operational
hedge. Because most students have ignored this anal-
ysis in their case write-ups (seven of the nine groups
from our most recent class recommended closing the
Japanese plant), students observe rst-hand the dan-
gers of ignoring the effects of uncertainty.
Finally, we make sure to include a discussion on
how the student team recommendations would be
implemented and monitored in practice (step 7 in our
problem solving methodology). For those who recom-
mend closing a plant, what do you do with the laid-
off workers? Because the Japanese plant has a record
of being the most technologically advanced manu-
facturing site, one student group suggested forming
an internal consulting team from members of this
site as a best-practices implementation group. We dis-
cuss whether, and how, this might be feasible. We
also consider the fact that the shipping strategy that
Applichem is currently using according to the case
may have indeed been optimal at some previous time.
How should Applichem monitor its performance to
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
70 INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS
prevent its shipping strategy from becoming outdated
in the future? Changing production strategies can be
expensive. How often should Applichem reevaluate
its strategies and change production amounts? What
factors would impact this decision? Of course, these
considerations do not do justice to the actual difcul-
ties faced in implementing such global strategies, but
they do at least force the students to consider factors
they may not have included in their initial analysis.
1
3.2. Steel Works, Inc.
Steel Works, Inc., is a case prepared by David Kletter
and Steve Graves at the Massachusetts Institute of
Technology (Kletter and Graves 1996). It describes
a manufacturer of custom and specialty use steels
that is facing difculties in customer service and ex-
cessive holding costs. The case includes a fair amount
of data for many of Steel Works specialty steel lines
including historical demand data, past inventory lev-
els, nancial information, etc. Students are asked to
evaluate Steel Works current policies and to make
suggestions for improvement.
The Steel Works case provides useful illustrations of
such problem solving skills as dening a model, stat-
ing assumptions, validating models, and using actual
data (as well as the frustrations of limited data sets). It
can also be useful for motivating the use of stochastic
inventory models and for discussing the limitations
of such models.
For the assignment of the Steel Works case, we
try to provide as little guidance as possible on what
type of analysis is needed for the case. This forces
the students to dene the actual problem and scope
for the case. Based on statements made in the case
description about falling customer service levels, stu-
dents tend to focus on the companys current inven-
tory policies. However, students still must determine
the scope of the problem and determine which sys-
tem variables it is likely that they could adjust and
what things are outside the scope of the problem. Can
Steel Works reduce the number of different specialty
steel products they produce? How would this be per-
ceived by customers? Should they consider changing
their production methods? What would this mean in
terms of capital expense and implementation? Den-
ing the scope of the problem in this case can lead to
valuable discussions.
Because this case is generally assigned after an
introductory lecture on inventory models, students
will often try to use such models to develop a bet-
ter inventory policy. However, several challenges are
1
The PowerPoint presentation and Excel le that can be used
in a lecture covering class discussions on the Applichem case
(Applichem.ppt and Applichem_soln.xls) can be found at http://
ite.pubs.informs.org/.
Figure 3 Steel Works Example
0.00
500.00
1,000.00
1,500.00
2,000.00
2,500.00
3,000.00
1 2 3 4 5 6 7
Product
A
v
e
r
a
g
e

i
n
v
e
n
t
o
r
y
90%
95%
98%
99%
99.87%
Actual
Note. This chart shows how tradeoffs can be illustrated between increasing
inventory levels and higher customer service. it can be used to introduce
students to What-If types of analysis.
immediately apparent. Should the students use a peri-
odic or continuous review model? There are hints in
the case that a periodic review model is more appro-
priate for all items because of Steel Works use of a
xed production schedule, but it is not immediately
obvious. What are the relevant parameters? How
should the students handle parameter values that
are not clear from the case description? For instance,
desired customer service levels are not directly given
in the case. This can prompt discussions of what-if
analysis to see what effect different parameter values
have on the model outputs. A chart similar to that
shown in Figure 3 can be produced to show the effect
on average inventory levels for different customer ser-
vice levels across products. Figure 3 illustrates the
tradeoffs between providing better customer service
and accepting higher inventory costs.
2
The Steel Works case includes one years worth
of demand data for seven different specialty steel
product lines. This proves to be a very limited
data set and makes it difcult to t a demand dis-
tribution to the data. Most students will tend to
assume that the demand follows a normal distribu-
tion for convenience. Simple statistical tests (such as
the KolmogorovSmirnov test) quickly show that this
is a dubious assumption for some of the product lines.
Thus, the students are forced to consider the trade-
offs of a possible faulty assumption with the benets
of generating an analytically tractable model. Dur-
ing class discussions we consider the downsides of
assuming normality when the data suggest otherwise
and what other possible approaches could be taken.
We discuss simple heuristics that offer clear improve-
ments for Steel Works. Such heuristics include setting
the safety stock to the maximum monthly sales over
the past year for some very slow-moving (i.e., least
2
The Excel le used in these calculations (SteelWorks_InvLevels.xls)
can be found at http://ite.pubs.informs.org/.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS 71
likely to be normally distributed) but high-margin
items (Willems 2002). This facilitates a discussion of
ABC inventory analysis (see Nahmias 2005) and other
simple inventory policies. Such heuristics, although
not optimal, do not depend on the assumptions of
normality and serve as an illustration of the possi-
ble benets and drawbacks to suboptimal solution
methods.
Class discussion of the Steel Works case will also
generally lend itself to a consideration of the impor-
tance of model validation (step 6). Many students
become confused between whether they should have
used a continuous review or a periodic review inven-
tory policy for their analysis. In the most recent class
offering (which consisted of nine group case study
write-ups) three groups recommended the use of a
continuous review model for all products, four recom-
mended periodic review, and two recommended peri-
odic review for some products and continuous review
for others. We discuss how one would go about
validating the choice of model with reality. What
conditions would violate the assumptions of peri-
odic/continuous review models? How would one go
about identifying such violations? Once you choose
a model, how would you validate the model param-
eters you have chosen? For instance, a production
lead time must be dened for the inventory model.
There is very little direct information given in the
case about the timing of the production sequence.
A single sentence in the case description states that
one of Steel Works products is always produced
during the rst week of the month. Based on this
statement, many students reluctantly choose a review
interval of one week for all seven product lines. How-
ever, what students often fail to realize is that there
is an opportunity to validate this assumption for all
product lines. The case data include historical batch
sizes for each product line. Therefore, we can work
backwardscomparing the historical batch sizes to
historical demandto work out the production lead
time for the model (Willems 2002). Using this method,
it appears that our assumption of a one-week produc-
tion lead time is supported by the data for all but
one product line. These examples begin to solidify the
importance of model validation in students minds.
3
4. Reinforcing Problem Solving
Methods Through Case Study
Write-Ups
Our students are required to detail their problem
solving methodology, including a clear statement of
3
A PowerPoint presentation that can be used in the teaching of the
Steel Works (SteelWorks.ppt) case can be found at http://ite.pubs.
informs.org/.
problem scope, stating assumptions, clearly dening
model outputs (recommendations), validating their
models, and discussing implementation issues. It is
essential to connect the problem solving methodol-
ogy listed above with student deliverables and to pro-
vide clear feedback and reinforcement to the students.
Most of our students have never before considered a
structured, methodological approach to problem solv-
ing. Thus, repetition is needed to reinforce the power
of such methods.
Outside of the main body of their write-up, stu-
dents are asked to discuss the iterations they went
through in developing a suitable model to analyze the
case study. What attempts did they discard and why?
After class discussions, students often realize that
they were too hasty to abandon a particular approach
(such as more sophisticated ways to deal with the
uncertainty in exchange rates in the Applichem case).
Students are almost always too eager to begin the
actual modeling and number crunching phase of a
case study without spending enough time on the rst
steps of the problem solving methodology: dening
the problem, stating assumptions, considering alter-
native modeling possibilities, etc. Likewise, students
are often too willing to take whatever output they get
from their rst modeling attempt and translate that
into their recommendations. This leads students to
dismiss the later steps of verication, validation, and
considering the effect of their assumptions on their
model outputs.
4
Interestingly, we have found that students often fall
short of our write-up expectations in surprising ways.
Students seem reluctant to give specic recommenda-
tions about policy changes and instead prefer to offer
general platitudes. For instance, students will sug-
gest instituting better forecasting techniques, reduc-
ing inventory levels, etc., without giving any details
on what these better methods would be or how
to develop them. Most surprisingly, we notice that
even students who have done extensive analysis will
often not use this analysis or their model outputs
when giving recommendations. It appears that stu-
dents are more fearful of offering a suggestion that
could be proven wrong than offering no specic rec-
ommendations at all. We explain that such overly
general suggestions are unacceptable and that all rec-
ommendations must be specic, supported by ana-
lytical analysis, and grounded in a problem solving
methodology. Why should forecasting be improved?
What will be the effect? How should it be improved?
However, students must also be reassured that their
entire grade will not suffer if they suggest specic
4
Bohl et al. (1995) present an excellent discussion of the importance
of assumptions in problem solving as it relates to the classic Monty
Hall Problem.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
72 INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS
recommendations that are counter to those that we,
as instructors, believe to be correct. Students have
often heard that there is no one right answer,
but they also rmly believe that there are wrong
answers. Thus, students are reluctant to provide an
answer that could be shown to be wrong. To alle-
viate some of these concerns, our grading is heavily
inuenced by the problem solving methodology that
students have followednot just the correctness of
their solutions. Additionally, identifying inferior solu-
tions should be part of the problem solving method-
ology. In particular, students should be taught to
consider alternative modeling possibilities and how to
quickly evaluate the different approaches for expected
effectiveness. In the best cases, even wrong answers
provide excellent learning opportunities.
Less surprisingly, students often neglect to consider
validation techniques and implementation issues.
Granted, validating models through a case study is
not always easy. However, both the Applichem case
and the Steel Works case provide enough data that
simple validations can be carried out. In particular,
the Steel Works case presents the opportunity to val-
idate the choice of modeling methodology using the
production batch sizes presented in the case and com-
paring these values to the expected output from peri-
odic and continuous review inventory models.
The implementation of solutions is an exception-
ally important topic in the problem solving method-
ology, but it can be challenging to communicate in
a case-based class. Most of our MBA students have
several years of previous work experience, so they
possess rst-hand knowledge of how difcult it can
be to implement operational changes at a company.
However, students still tend to ignore these issues
in their case write-ups. Therefore, we dedicate sig-
nicant portions of the class discussion to examin-
ing such issues as how to get general buy-in from
employees, where would you expect resistance to
come from, how would you deal with such resistance,
etc. Such discussions do not completely replicate the
experiences of actual solution implementations, but at
least they make the students aware of some of the dif-
culties that can arise in the implementation step and
how these difculties might be addressed.
5. Conclusions
The need for producing MBA students with better
problem solving abilities is clear. Recruiters continu-
ally point to analytical thinking as a requirement in
job candidates. However, it is also clear that there
are fewer opportunities to teach such skills in core
management science classes. The continued increase
in popularity of supply chain classes, both as core
classes and as elective classes, presents an opportunity
to teach problem solving techniques through specic
supply chain concepts such as logistics, inventory, and
others.
This discussion presents suggestions for how a
basic problem solving methodology can be incor-
porated into an MBA-level SCM class. Two cases,
Applichem (A) and Steel Works, Inc., are presented
as examples of the use of this methodology. The
Applichem case is particularly useful for teaching stu-
dents how to dene an objective and appropriate
measures, how to deal with assumptions, and how
to supplement limited data. The Steel Works case is
useful in illustrating the problem solving steps of
dening a model (and considering alternative mod-
els), stating assumptions, validating a model, and
dealing with actual data. It is also an excellent way
to demonstrate periodic review versus continuous
review inventory models and to explain the benets
of what-if analysis.
There are many other cases used in SCM courses
that could be approached using the problem solv-
ing methodology presented here (see Johnson and
Pyke 2000 for a complete listing of SCM cases).
Even cases that do not contain the detailed numerical
data found in the Applichem and Steel Works cases
can be approached using this (or a similar) problem
solving methodology. The important point is to get
students comfortable with breaking down a very dif-
cult problem into manageable chunks and identifying
the important characteristics of the problem. Evans
(1997a, p. 79; 1997b, p. 106) nicely denes a problem
solving process that can be applied to the messi-
est and least data-intensive of business problems.
His process includes the following steps: (1) Mess
nding, (2) Fact nding, (3) Problem nding, (4) Idea
nding, (5) Solution nding, (6) Acceptance nding.
Each of our methodology steps can be mapped on
to those of Evanss process, but he also describes the
application of this methodology to a very broad range
of business problems and gives a set of questions
that can be asked to stakeholders at each step in the
methodology to provide relevant information.
We also discuss how to reinforce the development
of the problem solving methodology in our students
case study write-ups. We nd that it is extremely
effective to require the students to document their
use of the problem solving methodology and to keep
track of approaches that were ultimately not used in
the nal analysis. This provides an opportunity for
reection on the students part to examine how they
could have improved their problem solving approach.
We have taught this supply chain management
course more than half-a-dozen times, focusing on
problem solving methods and tools. We cover approx-
imately one case a week during a ten-week quarter;
all cases are geared to emphasize problem solving
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS 73
methods while also teaching important supply chain
management concepts.
5
Our experiences suggest that students leave the
course much more able to confront difcult business
problems. In a recent teaching of this course, students
were asked to rate their comfort level with solving
complex problems both before and after the course
and to provide feedback on their problem solving
skills.
6
Surprisingly, there was very little change in
the actual scores that students assigned to their prob-
lem solving abilities (the mean score on a rating of
15 remained virtually unchanged over the 20 stu-
dents). However, student feedback indicated that they
had gained a much greater understanding of problem
solving methods. Student comments included such
statements as
Course helped me to think in terms of breaking down
complex business problems into smaller chunks that
are easier to handle and solve.
|This course] allowed me to foresee outcomes of vari-
ous decisions before having to be fully implemented.
I learned to look at processes in a more in-depth man-
ner while still keeping sight of larger goals.
|This course] extremely improved my problem solving
skills by being able to break down large problems into
smaller, easier ones.
We attribute the dichotomy between the student
responses and the nonincreasing scores in problem
solving to the possibility that, although students did
truly learn to be better problem solvers, they were
also exposed to quite difcult problems that proved
to the students that they may have initially overesti-
mated their own problem solving skills. Students also
provided feedback such as
The problems (case studies) were very challenging and
extremely difcult . . . but it gave me the ability to be
challenged and learn from attempting hard problems.
Before this class I have never encountered problems as
difcult as these. I was very challenged throughout the
whole course. Helped to approach difcult problems.
It is essential to provide our students with such prob-
lem solving abilities. As specic management science
5
A sample course schedule (SampleCourseSchedule.pdf) can be
found at http://ite.pubs.informs.org/.
6
Several studies have shown correlations between students
self-condence in mathematical abilities and their performance;
however, there is some disagreement on the magnitude of these
correlations. Bandura (1997), Casey et al. (2001), and others argue
that increases in condence with mathematical ability do correlate
with better mathematical skills. However, Hawkins (1995) argues
that this may not be the case. Our data are not nearly sufcient to
make further claims on this subject.
courses continue to be removed from many MBA cur-
ricula, we need to nd new ways to expose our stu-
dents to such methods and tools. Students have been
very complimentary of our approach and teaching
methods in this course, even years after their grad-
uation. We have been contacted many times by past
students who let us know that they have success-
fully applied many of the general and specic top-
ics of this course to their jobs. We hope to expand
the teaching methods described here to undergradu-
ate courses in the future so that students will become
acquainted with basic problem solving techniques as
soon as possible.
References
Ackoff, R. L. 1978. The Art of Problem Solving. John Wiley & Sons,
New York.
Association of American Colleges and Universities. 2005. Lib-
eral education outcomes: A preliminary report on student
achievement in college. Association of American Colleges
and Universities, Washington, DC, http://www.aacu.org/
advocacy/pdfs/LEAP_Report_FINAL.pdf.
Baker, S., B. Leak. 2006. Math will rock your world. Business-
Week Online (January 23). http://www.businessweek.com/
magazine/content/06_04/b3968001.htm?chan=search.
Bandura, A. 1997. Self-Efcacy: The Exercise of Control. Freeman,
New York.
Bohl, A. H., M. J. Liberatore, R. L. Nydick. 1995. A tale of two
goats . . . and a car, or the importance of assumptions in prob-
lem solutions. J. Recreational Math. 27(1) 19.
Casey, M. B., R. L. Nuttall, E. Pezaris. 2001. Spatial-mechanical
reasoning skills versus mathematics self-condence as media-
tors of gender differences on mathematics subtests using cross-
national gender-based items. J. Res. Math. Ed. 32(1) 2857.
Cohen, M. A., A. Huchzermeier. 1999. Global supply chain man-
agement: A survey of research and applications. S. Tayur,
R. Ganeshan, M. Magazine, eds. Quantitative Models for Sup-
ply Chain Management. Kluwer Academic Publishers, Norwell,
MA, 669702.
Davenport, T. 2006. Competing on analytics. Harvard Bus. Rev.
84(January) 99107.
Evans, J. 1997a. Creativity in OR/MS: The creative problem-solving
process, Part 1. Interfaces 27(5) 7883.
Evans, J. 1997b. Creativity in OR/MS: The creative problem-solving
process, Part 2. Interfaces 27(6) 106111.
Flaherty, T. 1985. Applichem (A). HBS Case 9-685-051. Harvard
Business School, Boston.
Giordano, F., M. Weir. 1985. A First Course in Mathematical Modeling.
Wadsworth, Belmont, CA.
Grossman, T. A. 2001. Solving the MBA teaching riddle. OR/MS
Today 28(6) 14.
Hawkins, R. M. 1995. Self-efcacy: A cause of debate. J. Behav. Ther-
apy. Experiment. Psychiatry 26(3) 235240.
Hopp, W. J., M. L. Spearman. 2000. Factory Physics, 2nd ed.
McGrawHill/Irwin, New York.
Huchzermeier, A., M. A. Cohen. 1996. Valuing operational exibil-
ity under exchange rate risk. Oper. Res. 44(1) 100113.
Johnson, M. E., D. F. Pyke. 2000. A framework for teaching supply
chain management. Production Oper. Management 9(1) 218.
Kletter, D., S. C. Graves. 1996. Steel Works, Inc., MIT Case.
Massachusetts Institute of Technology, Cambridge, MA.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.
Fry: Teaching Problem Solving Techniques to MBA Students Through Supply Chain Management
74 INFORMS Transactions on Education 8(2), pp. 6574, 2008 INFORMS
Lamoureaux, C. 2000. Business and management. Presentation,
CRAFTY Curriculum Foundations Project (October 2829),
University of Arizona, Tucson.
Leonhardt, D. 2000. A matter of degree? Not for consultants. New
York Times (October 1, Sec. 3, 1). http://query.nytimes.com/gst/
fullpage.html?res=9C0DEFDD173DF932A35753C1A9669C8B63.
Liberatore, M. J., R. L. Nydick. 1999. Breaking the mold: A new
approach for teaching the rst MBA management science
course. Interfaces 29(4) 99116.
Lowe, T. J., R. E. Wendell, G. Hu. 2002. Screening location strategies
to reduce exchange rate risk. Eur. J. Oper. Res. 136(3) 573590.
Merritt, J., E. Chambers. 2004. The best b-schools. BusinessWeek
Online (October 18), http://www.businessweek.com.
Montell, G. 1999. Another career choice for Ph.D.s: Manage-
ment consulting. The Chronicle of Higher Ed. (November 12),
http://chronicle.com/jobs/99/11/99111203c.htm.
Nahmias, S. 2005. Production and Operations Analysis, 5th ed.
McGraw Hill Higher Education, New York.
Powell, S. G. 1995. Teaching the art of modeling to MBA students.
Interfaces 25(3) 8894.
Powell, S. G. 2001. Teaching modeling in management sci-
ence. INFORMS Trans. Ed. 1(2), http://ite.pubs.informs.org
/Vol1No2/Powell/Powell.php.
Rubenstein, M. 1975. Patterns of Problem Solving. Prentice Hall,
Englewood Cliffs, NJ.
Savage, S. 2002. The aw of averages. Harvard Bus. Rev. (Novem-
ber). doi: 10.1225/F0211B.
Willems, S. P. 2002. Personal correspondence regarding the teaching
of the Steel Works, Inc., case.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
s
u
p
p
l
e
m
e
n
t
a
l
m
a
t
e
r
i
a
l
a
n
d
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
i
t
e
.
p
u
b
s
.
i
n
f
o
r
m
s
.
o
r
g
.

You might also like