LINKS Tutorial #4:

Forecasting

Katrina A. Zalatan & Randall G. Chapman

In LINKS, your firm is evaluated by financial, operating, and customer-related results.

Planning is the process you go through to make decisions to achieve these results.
Forecasting is a critical part of the planning process.

This tutorial is divided into two parts. Each part ends with a "hands-on" exercise.
Here is an outline of tutorial and the questions that comprise each exercise:

Starting On Page:

PART 1: What is Forecasting? 3

Exercise Topic: Page:

Question:
1 Inputs to forecast sales 3
2 Input sources 3
3 Inputs to forecast parts 3
4 Forecast accuracy implications 3
5 Gross margin forecasting 3

PART 2: How Does Sales Forecasting Relate to Other

Supply Chain Management Decisions? 7

Exercise Topic: Page:

Question:
1 Costs of underestimation 8
2 Underlying demand patterns 8
3 Forecasting assumptions 9
4 Safety stock 9
5 Costs of overestimation 9

Zalatan & Chapman: LINKS Tutorial #4 Page 1 Revised January/2009

 1. What is Forecasting?

In LINKS, forecasting refers to your ability to predict sales volumes. Every simulation
round, you’ll prepare sales volume forecasts for each of your products. You' ll create the
1
following forecasts :

Forecasting Decisions Firm Month

Short-Term (i.e., Next Round)

Sales Volume Forecast, Channel #1 Region 1 Region 2 Region 3

Product 1

Product 2

…sales forecasts for next

round ("the short-term") by Short-Term (i.e., Next Round)
Sales Volume Forecast, Channel #2 Region 1 Region 2 Region 3
channel, by region, by
Product 1
product and…
Product 2

…the next two rounds Long-Term (i.e., 2- and 3-Rounds

Hence) Sales Volume Forecasts 2-Rounds Hence 3-Rounds Hence
after that ("the long-term")
by product. Total Sales Volume, Product 1

Total Sales Volume, Product 2

In all cases, great forecasts are very close estimates of what actually happens in the
future. For example, you forecast that you' d sell 100 units of product 1-1 in channel 1 in
region 1 in round 4, and when round 4 came around, you sell exactly that. You were
100% accurate. That' s the ideal. But forecasts could also be too high or too low in
varying degrees, with consequent accuracy of less than 100%.

1
Some LINKS simulation variants require only short-term sales volume forecasts; other LINKS
simulation variants require both short-term and long-term sales volume forecasts.

Zalatan & Chapman: LINKS Tutorial #4 Page 2 Revised January/2009

Your LINKS reports summarize your forecasting accuracy in two places:
1. An overall "forecasting accuracy" score that'
s one of the operational metrics
on your Performance Evaluation Report (page 1 of your financial reports).
2. A more detailed Forecasting Accuracy Report that reports the accuracy of
your sales volume forecasts individually and in aggregate for your firm as a
whole.
This tutorial focuses on the process of sales forecasting and the implications that
forecasting accuracy has for your firm's supply chain performance.

EXERCISE #1: Your Own Forecasting Process

Work through each of the following questions, then check out the "answers" on the next
page.

1. Inputs to Forecast Sales: What "inputs" should you consider when making your
sales volume forecasts?

2. Input Sources: Where can you find this information (from answer to question 1)?

3. Inputs to Forecast Parts: What inputs should you consider when forecasting
replacement part requirements?

4. Forecasting Accuracy Implications: Assume you have an 80% "Forecasting

Accuracy" score on your Performance Evaluation Report during a simulation round.
What are the effects of this 80% “Forecasting Accuracy” performance level?

Zalatan & Chapman: LINKS Tutorial #4 Page 3 Revised January/2009

 EXERCISE #1: ANSWERS

1. Several things. You should probably start by looking at your sales volume history.
What were your results? What was the size and variability (degree of change) of
demand in each region for each product and channel? Do you see any trends? Do
they appear to be long-term?

Eventually, you should seek to predict cause-and-effect relationships between

demand and the variables that impact it, variables like:
Price and
Customers'perceptions of your product quality, service quality, and
availability.
… variables that relate to your firm'
s marketing and operational strategies. Of
course, competitors' actions should also be discussed and anticipated. These
generate demand variables and the factors that influence them are detailed in your
LINKS manual.

2. Some information just needs to be deduced from experience and your reports. For
example, what happened when you increased price during the last simulation round?
Would you get the same results next time, or do you expect conditions to change?
What other variables influenced your sales?

Spreadsheets that track decisions, results, and other information from round to
round may also help you see trends. Other information could be obtained directly
(for a price) through research studies detailed in your LINKS participant’s manual.

3. Replacement parts requirements are predominantly a function of estimated failure

rates times historical sales through the warranty period. Failure rates vary by
supplier and by sub-assembly component.

4. A direct result is an increase in "Administrative Overhead (O/H)"—in this case by

20% (a 1% O/H increase for every 1% inaccuracy). Other supply chain management
areas are also adversely affected.

Zalatan & Chapman: LINKS Tutorial #4 Page 4 Revised January/2009

 2. How does sales forecasting relate to other
supply chain management decisions?

Each simulation round, you and your team will discuss and develop strategies (broad
courses of action) and decisions relative to each supply chain management area in
LINKS, including:

Product Development
Procurement
Manufacturing
Distribution
Transportation
Service
Generate Demand
Information Technology.

Ideally, customers should drive your generate demand strategies and the design of
your entire supply chain. Generate demand strategies should also drive your sales
forecasts:

" # $

$ %& #
'& () !

*
+
) + #

Zalatan & Chapman: LINKS Tutorial #4 Page 5 Revised January/2009

These sales forecasts have implications for many other supply chain management
decisions upstream (shaded diamond boxes, below):

& & #

)!

' ,
(

'/# * /(

-
!

Accurate sales forecasts could enable all supply chain elements to work together as an
integrated system, something critical to the success of your company. So what exactly
is the impact of sales forecasting? Explore this by working through EXERCISE 2 - A
LINKS Case.

Zalatan & Chapman: LINKS Tutorial #4 Page 6 Revised January/2009

 EXERCISE #2: A LINKS Case

Please read the scenario about Firm Z's Region #3, and answer the questions that
follow. Check your responses by referring to the answers starting on page 10.

SCENARIO: Firm Z just completed round 5. They currently sell both Products Z-1
and Z-2 in Region 3, both in channel 1 only. To date, Firm Z has made only three
changes to its initial starting position decisions (as of round #3) in Region 3. They:
Added a DC3 that they' ve owned and operated since round 4 (there was no
DC3 to start). They added it to try postponement and "see what happens."
Reduced FG and parts inventories to absolute minimums to cut
"unnecessary" costs.

All of the following have remained constant since round #1:

Products configurations.
Channels.
Prices.
Marketing budgets.

Recent Set-Top Box Industry Bulletins and some round #4 research has revealed
that some competitors have:
Opened channel 2 for both product lines.
Reconfigured products.
Changed prices and marketing budgets.

Here is an excerpt from one of Firm Z'

s spreadsheets that shows their sales results
since round #1):

Round #1 Round #2 Round #3 Round #4 Round #5

Forecast Actual Forecast Actual Forecast Actual Forecast Actual Forecast Actual
Z-1 15,000 15,000 15,000 10,500 15,000 15,120 15,000 11,300 12,000 6,000
Z-2 6,600 6,600 6,600 4,620 6,600 6,930 6,600 5,300 5,400 3,300

Continued…

Zalatan & Chapman: LINKS Tutorial #4 Page 7 Revised January/2009

 EXERCISE #2: continued…

Firm Z' s forecasts were just slightly above round #4 actuals because they explained,
"We' re not making any generate demand changes in this region, so sales probably
won' t change much." They forecasted increases of 6% and 2% for Products Z-1 and
Z-2, respectively. Other data from Firm Z' s rounds #4 and #5 follow:

$ 1
234 5 67 0& ) ! 1
234 5 67
# ;44
8 "944 8 )! 944
0<44 # :"44
=* "044
4"44

1. Costs of Underestimation: Assume that Firm Z ordered exactly the number of Z-1
and Z-2 products they forecasted for round #3 (15,000 and 6,600 units respectively).
Assume that there were no finished goods inventories for either product at the start
of round #3. What was the likely cost implication(s) of this underestimation?

2. Underlying Demand Patterns: Finish calculating the percentage changes in actual

sales for both products in rounds #4 and #5 in the table below. What do you notice
(if anything) about demand for each product?

Round Round Round Round Round

% % % %
#1 #2 #3 #4 #5
Change Change Change Change
Actuals Actuals Actuals Actuals Actuals

Z-1 15,000 10,500 -30% 15,120 +44% 11,300 6,000

Z-2 6,600 4,620 -30% 6,930 +50% 5,300 3,300

Zalatan & Chapman: LINKS Tutorial #4 Page 8 Revised January/2009

 EXERCISE #2: continued…

3. Forecasting Assumptions: What appear to be the primary reasons for Firm Z'
s lousy
round #5 forecast accuracy? Explain.

4. Safety Stock:
a. How much Product Z-0 safety stock did Firm Z have in round #5?

b. Was this a good decision? Explain.

5. Cost of Overestimating: Firm Z forecasts of unit sales were obviously too high in
round #5. What was the cost impact of this poor forecasting from a DC3/Product Z-0
inventory standpoint? Assume a beginning inventory of 700 units and that 16,200
units arrive on-time (about a 93% on-time arrival rate) of the original DC3 Z-0
shipment order of 17,400 units. (Show your calculations!)

Zalatan & Chapman: LINKS Tutorial #4 Page 9 Revised January/2009

 EXERCISE #2: ANSWERS

1. They had to pay a premium for emergency production and the emergency shipment
of this production. If there wasn'
t enough procurement inventory on hand to
accomplish this emergency production, they had to pay a premium for emergency
procurement, too.

2. Round #3 to Round #4:

Z-1: (11,300 - 15,120)/15,120 = -0.25 (actual sales were down 25%)
Z-2: (5,300 - 6,930)/6,930 = -0.24 (actual sales were down 24%)
Round #4 to Round #5:
Z-1: (6,000 - 11,300)/11,300 = -0.47 (actual sales were down 47%)
Z-2: (3,300 - 5,300)/5,300 = -0.38 (actual sales were down 38%)

It appears that the demand for products Z-1 and Z-2 tends to vary together, and
swings tend to be fairly large. Five rounds is a rather short time period, but if Firm Z
continues to track this information, they may be able to discern predictable demand
patterns. While no demand pattern is guaranteed to last forever, and no two
channels can be guaranteed to be the same from region to region or simulation
industry to simulation industry, the information may prove useful.

3. One reason appears to be their naïve assumption that their sales will only be
influenced by their own generate demand strategies, i.e., "We' re not making any
generate demand changes in this region, so sales probably won' t change much."
What about competitors? They also didn’t appear to consider underlying variability in
the demand patterns for this channel in this region. (That' s what you calculated in
question #1, above … the "variability" or round-to-round changes in "pure" demand
in the first three rounds of the simulation when all firms made the same decisions,
i.e., no generate demand strategy variation between firms).

4. a. None. They forecasted that they' d sell 17,400 total units in Region 3, and the
DC3 FG Inventory: Product Z-0, Round #5 Report shows they had only 16,900
Product Z-0's on hand, that's 500 fewer units than they needed to meet their round
#5 sales forecasts. This also indicates that Firm Z forgot that surface shipments do
not necessarily have 100% of the order available for use in production that round,
meaning those 500 units they needed were probably part of a "delayed shipment"
they received after production at the end of the round.

b. In hindsight, yes, because they sold so much fewer than they expected. Their
ending inventories would have been even higher. Given the apparent variability of
demand in this channel in this region, however, they could probably benefit by
considering higher safety stock levels in the future when their understanding of
generate demand strategies, competitors, and forecasting improves.

Zalatan & Chapman: LINKS Tutorial #4 Page 10 Revised January/2009

 EXERCISE #2: ANSWERS, continued…

5. (1) First, calculate Region 3'

s ending inventory of Product Z-0 for round #5:
Beginning Inventory 700
+ Plant Shipments, Surface 16,200
+ Plant Shipments, Air 0
+ Plant Shipments, Emergency 0
Postponed Production -9,300
= Available Inventory 7,600
- Sales, Region 3: 0
+ Plant Shipments, Delayed 1,200 (17,400 ordered - 16,200)
= Ending Inventory 8,800

(2) Calculate the inventory value = 8,800 x $102/unit = $897,600

Plus a 3% carrying cost = $897,600 x 1.03 = $924,528

That's a carrying cost of $26,928 ($924,528 - $897,600) that was caused by Firm
Z's poor forecasts. And that $26,928 and the inventory stored is cash that Firm Z
could not use for other value-producing uses in round #5.

To put this waste into perspective, consider that the just the carrying cost of
$26,928 was:
$5,486 (roughly 25%) more than they paid for service in Region 3 in round #5.
Twice their total Region #3 replacement parts cost in round #5.
25% of their Region #3 marketing budget.

Put that way, their round #5 carrying cost was one big chunk of change, and one
good reason to improve their forecasting!

Zalatan & Chapman: LINKS Tutorial #4 Page 11 Revised January/2009