Part IV: Planning and Managing

Inventories in A Supply Chain

Chapter 12: Determining the Optimal Level of

Product Availability

© Chopra / OPNS 455 / Optimal Availability 1

 Mattel, Inc. & Toys “R” Us

Mattel was hurt last year by inventory cutbacks at Toys “R” Us, and
officials are also eager to avoid a repeat of the 1998 Thanksgiving
weekend. Mattel had expected to ship a lot of merchandise after the
weekend, but retailers, wary of excess inventory, stopped ordering
from Mattel. That led the company to report a $500 million sales
shortfall in the last weeks of the year ... For the crucial holiday
selling season this year, Mattel said it will require retailers to place
their full orders before Thanksgiving. And, for the first time, the
company will no longer take reorders in December, Ms. Barad
said. This will enable Mattel to tailor production more closely to
demand and avoid building inventory for orders that don't come.
- Wall Street Journal, Feb. 18, 1999

© Chopra / OPNS 455 / Optimal Availability 2

 Key Questions

 How much should Toys R Us order given demand

uncertainty?
 How much should Mattel order?
 Will Mattel’s action help or hurt profitability?
 What actions can improve supply chain profitability?

© Chopra / OPNS 455 / Optimal Availability 3

 Drivers of Supply Chain Performance
Efficiency Responsiveness

Supply chain structure

Facilities Transportation

Inventory

•Seasonal Inventory
•Cycle Inventory
•Safety Inventory
•Level of Product Availability
•Newsboy tradeoff for Seasonal items; continuously stocked
items; multiple products under capacity constraints
•Levers to improve supply chain profits and decrease seasonal inventory

© Chopra / OPNS 455 / Optimal Availability 4

 Estimating Optimal Level of Product Availability
Buyers’ Estimate of Demand Distribution at L.L. Bean

Demand Probabability Probability of demand Probability of demand being

[100s] being this much or less greater than this much
4 .01 .01 .99
5 .02 .03 .97
6 .04 .07 .93
7 .08 .15 .85
8 .09 .24 .76
9 .11 .35 .65
10 .16 .51 .49
11 .20 .71 .29
12 .11 .82 .18
13 .10 .92 .08
14 .04 .96 .04
15 .02 .98 .02
16 .01 .99 .01
17 .01 1.00 .00

Expected Demand = 1,026 Parkas

© Chopra / OPNS 455 / Optimal Availability 5
 Estimating Optimal Level of Product Availability
Cost of Over- and Understocking at L.L Bean

Cost per parka = c = $45

Sale price per parka = p = $100
Discount price per parka = $50
Holding and transportation cost = $10
Salvage value = s = $50-$10 = $40

 Profit from selling parka = Cu = p-c = $100-$45 = $55

 Cost of overstocking = Co = c-s = $45+$10-$50 = $5

© Chopra / OPNS 455 / Optimal Availability 6

 Estimating Optimal Level of Product Availability
Profit from Ordering the Expected Demand at L.L. Bean

Order Quantity = 1000 Parkas (Expected Demand = 1,026)

Probability Demand Sold Overstocked Understocked Profit
0.01 400 400 600 0 $ 19,000
0.02 500 500 500 0 $ 25,000
0.04 600 600 400 0 $ 31,000
0.08 700 700 300 0 $ 37,000
0.09 800 800 200 0 $ 43,000
0.11 900 900 100 0 $ 49,000
0.16 1,000 1,000 0 0 $ 55,000
0.20 1,100 1,000 0 100 $ 55,000
0.11 1,200 1,000 0 200 $ 55,000
0.10 1,300 1,000 0 300 $ 55,000
0.04 1,400 1,000 0 400 $ 55,000
0.02 1,500 1,000 0 500 $ 55,000
0.01 1,600 1,000 0 600 $ 55,000
0.01 1,700 1,000 0 700 $ 55,000
Expected: 1,026 915 85 111 $ 49,900

© Chopra / OPNS 455 / Optimal Availability 7

Estimating Optimal Level of Product Availability
Expected Marginal Contribution of Increasing Order Size by 100 units at
L.L. Bean

If we order 1,000, the CSL=probability(demand ≤ 1,000) = 0.51

Additional 100 units sell with Additional 100 units do not sell
probability 1-CSL = 0.49. with probability CSL = 0.51.
We earn margin Cu=p-c = $55 / unit. We lose Co= c-s = $5 per unit.

Expected marginal contribution of an additional 100 units =

0.49 x 100 x $55 - 0. 51 x 100 x $5 = $2,440

© Chopra / OPNS 455 / Optimal Availability 8

 Estimating Optimal Level of Product Availability
Expected Marginal Contributions as Availability is Increased

Additional Expected Expected Expected Marginal

100s Marginal Benefit Marginal Cost Contribution
11th 5500.49 = 2695 500.51 = 255 2695-255 = 2440
12th 5500.29 = 1595 500.71 = 355 1595-355 = 1240
13th 5500.18 = 990 500.82 = 410 990-410 = 580
14th 5500.08 = 440 500.92 = 460 440-460 = -20
15th 5500.04 = 220 500.96 = 480 220-480 = -260
16th 5500.02 = 110 500.98 = 490 110-490 = -380
17th 5500.01 = 55 500.99 = 495 55-495 = -440
Optimal Order Quantity = 1,300 Parkas
Expected Profit = $ 54,160
Service level = 92%
© Chopra / OPNS 455 / Optimal Availability 9
 Estimating Optimal Level of Product Availability
Seasonal Items with a Single Order in a Season: Summary
p = sale price
s = outlet or salvage price Cu = p-c
c = purchase price
O* = optimal order size Co = c-s
CSL* = optimal cycle service level = probability (demand ≤ O*)

At the optimal cycle service level CSL* and order size O*:
Expected marginal profit from raising the order size by one unit to O*+ 1 ≤ 0
Expected Marginal Revenue = probability the unit sells  Cu = (1-CSL*) Cu
Expected Marginal Cost = probability the unit does not sell  Co = CSL* Co

Therefore: (1-CSL*)  Cu ≤ CSL* Co

Optimal Cycle Service Level: CSL* ≥ Cu / (Cu + Co ) = (p-c) / (p-s)

Critical fractile

© Chopra / OPNS 455 / Optimal Availability 10

 Product Availability for Continuous Distributions:
Example

Motown studios is deciding on the number of copies of a CD to

have manufactured. The manufacturer currently charges $2 for
each CD. Motown sells each CD for $12 and currently places
only one order for the CD before its release. Unsold CDs must
be trashed. Demand for the CD has been forecast to be
normally distributed with a mean of 30,000 and a standard
deviation of 15,000.
How many CDs should Motown order?

© Chopra / OPNS 455 / Optimal Availability 11

 Evaluating Expected Profits, Overstock, and
Understock

Expected profits = (p-s) NORMDIST((O - )/, 0, 1, 1)

-  (p-s) NORMDIST((O- )/, 0, 1, 0) – O(c-s) NORMDIST(O, , , 1)

+ O (p-c) [1 - NORMDIST(O, , , 1)] (12.3)

Expected overstock = (O - )NORMDIST((O - )/, 0, 1, 1)

+  NORMDIST((O - )/, 0, 1, 0) (12.4)

Expected understock = ( - O)[1- NORMDIST((O - )/, 0, 1, 1)]

+  NORMDIST((O - )/, 0, 1, 0) (12.5)

© Chopra / OPNS 455 / Optimal Availability 12

 Product Availability for Continuous Distributions
Under Quantity Discounts
Motown studios is deciding on the number of copies of a CD to have
manufactured. The manufacturer currently charges $2 for each CD. Motown sells
each CD for $12 and currently places only one order for the CD before its release.
Demand for the CD has been forecast to be normally distributed with a mean of
30,000 and a standard deviation of 15,000. How many CDs should Motown order?
What is the expected profit?

What is the expected overstock?

What is the expected understock?

The manufacturer now offers a price of $1.95 for orders of at least 50,000 CDs and
a price of $1.90 for orders of at least 60,000 CDs.
How should Motown respond?

© Chopra / OPNS 455 / Optimal Availability 13

 Drivers of Supply Chain Performance
Efficiency Responsiveness

Supply chain structure

Facilities Transportation

Inventory

•Seasonal Inventory
•Cycle Inventory
•Safety Inventory
•Level of Product Availability
•Newsboy tradeoff for Seasonal items; continuously stocked
items

© Chopra / OPNS 455 / Optimal Availability 14

 Estimating Optimal Level of Product Availability
Continuously Stocked Items

Given
– CSL = probability of not stocking out in a cycle with current
level of safety stock = Cycle Service Level
– H = cost of holding one unit for one year
– D = Annual demand
– Q = Replenishment lot size
 Basic tradeoff
– Benefit from increasing safety inventory (additional sales if
demand is high) versus cost of increasing safety inventory
(holding cost of one unit)

© Chopra / OPNS 455 / Optimal Availability 15

 Estimating Optimal Level of Product Availability
Continuously Stocked Items

 Benefit from increasing safety inventory by one unit

= (1- CSL*) Cu
 Cost of increasing safety inventory by one unit
= HQ/D

Equating the two gives optimal level of product availability

CSL* = 1-HQ/(CuD)

© Chopra / OPNS 455 / Optimal Availability 16

 Estimating Cost of Understocking for
Continuously Stocked Items
Data
D = 100 gallons/week; D= 20;
H = $0.6/gal./year
L = 2 weeks; Q = 400; ROP = 300.

What is the implied cost of stocking out?

– Safety Inventory = ROP – D*L = 100
– Standard deviation of lead time demand: 20*sqrt(2)=28.3
– With given policy, CSL=NORMSDIST(100/28.3)=0.9998
– Implied cost of stocking out:
Cu= HQ / (1-CSL) / D = 0.6*400/ 0.0002 / 5,200 = $230.8

Source: Example 12.3 in C & M

© Chopra / OPNS 455 / Optimal Availability 17
 Drivers of Supply Chain Performance
Efficiency Responsiveness

Supply chain structure

Facilities Transportation

Inventory

•Seasonal Inventory
•Cycle Inventory
•Safety Inventory
•Level of Product Availability
•Newsboy tradeoff for Seasonal items; continuously stocked
items; multiple products under capacity constraints

© Chopra / OPNS 455 / Optimal Availability 18

 Optimal Availability for Multiple Products
Under Capacity Constraint

High End Mid Range

Retail price pi $150 $100
Purchase price ci $50 $40
Salvage price si $35 $25
Mean Demand 1000 2000
Standard deviation 300 400
of demand

Available Capacity = 3,000.

© Chopra / OPNS 455 / Optimal Availability 19
 Optimal Availability Assuming No Capacity
Constraint

High End Mid Range

Cu = pi - ci $150- $100-$40=$60
$50=$100
Co + Cu =pi - si $150- $100 –
$35=$115 $25 =$75
Critical 100/115 = 60/75= 0.80
Fractile 0.87
Optimal 1,337 2,337
*
Order iO

Total Order Quantity = 3,674 > 3,000

© Chopra / OPNS 455 / Optimal Availability 20
 Optimal Availability Under Capacity Constraint
Optimal Ordering Policy

High End Mid Range

Order qty, O 1089 1911
Exp. Mar. Profit $ 29.10 $ 29.10
Expected Profit $ 89,416 $ 105,736
Total Expected Profit $ 195,152

At optimality, expected marginal contribution

of each item ordered is equal

© Chopra / OPNS 455 / Optimal Availability 21

 Optimal Availability Under Capacity Constraint
Optimal Ordering Policy

 Step 1: Set order quantity Qi = 0 for all products i.

 Step 2: For all products i, compute/update the expected marginal
contribution at the current order quantity
 Step 3: For the product j with the largest positive expected
marginal contribution, increase order quantity Qj by the minimum
increment. This is equivalent to auctioning off capacity one
unit at a time and assigning it to the highest bidder (the one
with the largest expected marginal contribution)
 Step 4: If there is still capacity available and there is some
product with a positive expected marginal contribution, return to
Step 2, else stop.

© Chopra / OPNS 455 / Optimal Availability 22

 Drivers of Supply Chain Performance
Efficiency Responsiveness

Supply chain structure

Facilities Transportation

Inventory

•Seasonal Inventory
•Cycle Inventory
•Safety Inventory
•Level of Product Availability
•Newsboy tradeoff for Seasonal items; continuously stocked
items; multiple products under capacity constraints
•Levers to improve supply chain profits and decrease seasonal
inventory
© Chopra / OPNS 455 / Optimal Availability 23
 Levers to Improve SC profits and decrease Seasonal
Inventory

 Increase salvage value (over stock outlets)

 Decrease cost of under stocking (substitution)
 Improve forecasts
 Multiple orders in a season
 Postponement

What happens to profits, understock, and overstock in

each case?

© Chopra / OPNS 455 / Optimal Availability 24

Levers to Improve SC profits and decrease Seasonal
Inventory: Postponement of Product Differentiation

Supply Chain Flows Without Postponement

Supply Chain Flows With Component Commonality and Postponement

© Chopra / OPNS 455 / Optimal Availability 25

 Value of Postponement: Benetton
Data
 Demands (uncorrelated)
– Each color: Mean = 1,000; SD = 500
– Aggregate: Mean = 4,000, SD = 1000
 For each garment
– Sale price = $50
– Salvage value = $10
– Production cost using option 1 (long lead time) = $20
– Production cost using option 1 (greige thread) = $22
 What is the effect of postponement?
– Expected overstock and under stock reduced
 What is the value of postponement?
– Expected profit increases from $94,576 to $98,092

© Chopra / OPNS 455 / Optimal Availability 26

 Value of Postponement: Benetton

 How does the value of postponement change as the

demand uncertainty increases/decreases?
 Which components are better postponed – most or
least expensive? Short or long lead times?
 How does value change as number of colors
postponed increases/decreases?
 With a process that allows postponement do you want
to sell more or fewer colors?

© Chopra / OPNS 455 / Optimal Availability 27

 Value of Postponement with Dominant Product
 Demand
– Color with dominant demand: Mean = 3,100, SD = 800CV=0.26
– Other three colors: Mean = 300, SD = 200 CV=0.67
– Aggregate: Mean = 4,000, SD = 872 CV=0.22

 Expected profit without postponement = $102,205

 Expected profit with postponement = $99,872

Why is postponement not valuable with a dominant product?

How should we react?
© Chopra / OPNS 455 / Optimal Availability 28
 Tailored Postponement: Benetton
4 colors, for each: mean demand = 1,000, SD=500
Produce Q1 units for each color, and QA units undyed

Q1 Aver. Aver.
(colored) QA(neutral) Total Aver. Profit Overstock Understock
0 4,524 4,524 $ 98,092 715 190
1337 0 5,348 $ 94,576 1648 300
700 1,850 4,650 $ 102,730 308 168
800 1,550 4,750 $ 104,603 427 170
900 950 4,550 $ 101,326 607 266
900 1,050 4,650 $ 101,647 664 230
1000 850 4,850 $ 100,312 815 195
1000 950 4,950 $ 100,951 803 149
1100 550 4,950 $ 99,180 1026 211
1100 650 5,050 $ 100,510 1008 185

© Chopra / OPNS 455 / Optimal Availability 29

 Cautions in Implementing Postponement
 The cost of postponement
– Postponement often increases the manufacturing cost
– Design and production costs can only be justified over a family of products
 Value of postponement is larger the more uncertain and the less
correlated the individual product demands
 Cautions
– End products must look suitably different to the consumer
– Do a small set of products provide most of the sales?
– Do products have low uncertainty?
 Tailored postponement
– Higher manufacturing cost is justified only for uncertain portion of demand.
– Consider more efficient process for stable portion of demand.

© Chopra / OPNS 455 / Optimal Availability 30

 Managing Inventories and Uncertainty in the
Supply Chain: Summary of Lessons
Push Pull

Cycle Inventory Safety Inventory Seasonal Inventory

• Aggregation • Quick response •Increase salvage value and
• Volume based •Reduce uncertainty decrease lost margin
discounts over •Reduce lead time •Shorten lead time to reduce
rolling horizon •Reduce lead time variability uncertainty
• EDLP, promote •Increase reorder frequency •Multiple orders in season.
to limit forward • Accurate response by pooling •Tailored postponement
buy •Tailored pooling based on
•Demand correlation
•Coefficient of variation
•Value of product
•Level of service
•Holding cost
© Chopra / OPNS 455 / Optimal Availability 31