Newsvendor Model

Chapter 9

1
utdallas.edu/~metin

Learning Goals

Determine the optimal level of product availability

Demand forecasting
Profit maximization

Other measures such as a fill rate

utdallas.edu/~metin

ONeills Hammer 3/2 wetsuit

utdallas.edu/~metin

Hammer 3/2 timeline and economics

Economics:

Generate forecast
of demand and
submit an order
to TEC

Spring selling season

Nov Dec Jan

Feb Mar Apr May Jun

Receive order
from TEC at the
end of the
month

Each suit sells for p = $180

TEC charges c = $110/suit
Discounted suits sell for v = $90

Jul Aug
Left over
units are
discounted

The too much/too little problem:

Order too much and inventory is left over at the end of the season
Order too little and sales are lost.

Marketings forecast for sales is 3200 units.

utdallas.edu/~metin

Newsvendor model implementation steps

Gather economic inputs:

selling price,
production/procurement cost,
salvage value of inventory

Generate a demand model to represent demand

Use empirical demand distribution
Choose a standard distribution function
the normal distribution,
the Poisson distribution.

Choose an objective:
maximize expected profit
satisfy a fill rate constraint.

utdallas.edu/~metin

Choose a quantity to order.

The Newsvendor Model:

Develop a Forecast

6
utdallas.edu/~metin

Historical forecast performance at ONeill

.

7000
6000

Actual demand

5000
4000
3000
2000
1000
0
0

1000

2000

3000

4000

5000

6000

7000

Forecast

Forecasts and actual demand for surf wet-suits from the previous season

utdallas.edu/~metin

How do we know the actual

when the actual demand > forecast demand

If we underestimate the demand, we stock less than

necessary.
The stock is less than the demand, the stockout occurs.
Are the number of stockout units (= unmet demand)
observable, i.e., known to the store manager?
Yes, if the store manager issues rain checks to customers.
No, if the stockout demand disappears silently.

No implies demand filtering. That is, demand is known

exactly only when it is below the stock.
Shall we order more than optimal to learn about demand
when the demand is filtered?

utdallas.edu/~metin

Empirical distribution of forecast accuracy

Order by A/F ratio

utdallas.edu/~metin

Normal distribution tutorial

All normal distributions are characterized by two parameters, mean =

and standard deviation =
All normal distributions are related to the standard normal that has mean
= 0 and standard deviation = 1.
For example:
Let Q be the order quantity, and (, ) the parameters of the normal demand
forecast.
Prob{demand is Q or lower} = Prob{the outcome of a standard normal is z or
lower}, where

Q
z
or Q z
the same equation, the first allows you to
(The above are two ways to write
calculate z from Q and the second lets you calculate Q from z.)
Look up Prob{the outcome of a standard normal is z or lower} in the
Standard Normal Distribution Function Table.
utdallas.edu/~metin

10

Using historical A/F ratios to choose a

Normal distribution for the demand forecast

Start with an initial forecast generated from hunches,

guesses, etc.
ONeills initial forecast for the Hammer 3/2 = 3200 units.

Evaluate the A/F ratios of the historical data:

A/F ratio

Actual demand
Forecast

Set the mean of the normal distribution to

Expected actual demand Expected A/F ratio Forecast

Set the standard deviation of the normal distribution to

Standard deviation of actual demand
Standard deviation of A/F ratios Forecast

utdallas.edu/~metin

11

ONeills Hammer 3/2 normal distribution forecast

Product description
JR ZEN FL 3/2
EPIC 5/3 W/HD
JR ZEN 3/2
WMS ZEN-ZIP 4/3

ZEN 3/2
ZEN-ZIP 4/3
WMS HAMMER 3/2 FULL
Average
Standard deviation

Forecast Actual demand

90
140
120
83
140
143
170
156

3190
3810
6490

Error
-50
37
-3
14

A/F Ratio
1.5556
0.6917
1.0214
0.9176

1195
3289
3673

1995
521
2817

0.3746
0.8633
0.5659
0.9975
0.3690

Expected actual demand 0.9975 3200 3192

Standard deviation of actual demand 0.369 3200 1181

ONeill should choose a normal distribution with mean 3192 and standard
deviation 1181 to represent demand for the Hammer 3/2 during the Spring season.
Why not a mean of 3200?

utdallas.edu/~metin

12

Empirical vs normal demand distribution

1.00
0.90

Probability.

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0

1000

2000

3000

4000

5000

6000

Quantity

Empirical distribution function (diamonds) and normal distribution function with

13
mean 3192 and standard deviation 1181 (solid line)
utdallas.edu/~metin

The Newsvendor Model:

The order quantity that maximizes
expected profit

14
utdallas.edu/~metin

Too much and too little costs

Co = overage cost
The cost of ordering one more unit than what you would have ordered had you
known demand.
In other words, suppose you had left over inventory (i.e., you over ordered).
Co is the increase in profit you would have enjoyed had you ordered one fewer
unit.
For the Hammer 3/2 Co = Cost Salvage value = c v = 110 90 = 20

Cu = underage cost
The cost of ordering one fewer unit than what you would have ordered had
you known demand.
In other words, suppose you had lost sales (i.e., you under ordered). Cu is the
increase in profit you would have enjoyed had you ordered one more unit.
For the Hammer 3/2 Cu = Price Cost = p c = 180 110 = 70

utdallas.edu/~metin

15

Balancing the risk and benefit of ordering a unit

Ordering one more unit increases the chance of overage

Expected loss on the Qth unit = Co x F(Q), where F(Q) = Prob{Demand <= Q)

The benefit of ordering one more unit is the reduction in the chance of
underage:
Expected benefit on the Qth unit = Cu x (1-F(Q))

Expectedgainorloss.

80
70
Expectedmarginalbenefit
ofunderstocking

60
50
40
30

Expectedmarginalloss
ofoverstocking

20
10

As more units are ordered,

the expected benefit from
ordering one unit
decreases
while the expected loss of
ordering one more unit
increases.

0
0

800

utdallas.edu/~metin

1600

2400

3200

4000

4800

5600

6400

16

Expected profit maximizing order quantity

To minimize the expected total cost of underage and overage, order

Q units so that the expected marginal cost with the Qth unit equals
the expected marginal benefit with the Qth unit:
Co F (Q) Cu 1 F Q

Cu
C o Cu

Rearrange terms in the above equation -> F (Q)

The ratio Cu / (Co + Cu) is called the critical ratio.

Hence, to minimize the expected total cost of underage and

overage, choose Q such that we dont have lost sales (i.e., demand
is Q or lower) with a probability that equals the critical ratio

utdallas.edu/~metin

17

Expected cost minimizing order quantity with the

empirical distribution function

Inputs: Empirical distribution function table; p = 180; c = 110; v =

90; Cu = 180-110 = 70; Co = 110-90 =20
Evaluate the critical ratio:
Cu
70
Co Cu

20 70

0.7778

Look up 0.7778 in the empirical distribution function graph

Or, look up 0.7778 among the ratios:
If the critical ratio falls between two values in the table, choose the one that
leads to the greater order quantity

Convert A/F ratio into the order quantity

utdallas.edu/~metin

Q Forecast * A / F 3200 *1.3 4160.

24
25
26

Percentile
72.7%
75.8%
78.8%

1.25
1.27
1.30

212
635
1696

170
500
1300

HEATWAVE 3/2
HEAT 3/2
HAMMER 3/2

Forecast Actual demand A/F Ratio Rank

Product description

18

Hammer 3/2s expected cost minimizing order

quantity using the normal distribution

Inputs: p = 180; c = 110; v = 90; Cu = 180-110 = 70; Co = 110-90 =20; critical ratio = 0.7778; mean =
= 3192; standard deviation = = 1181
Look up critical ratio in the Standard Normal Distribution Function Table:
z

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.5 0.6915 0.6950 0.6985 0.7019 0.7054 0.7088 0.7123 0.7157 0.7190 0.7224
0.6 0.7257 0.7291 0.7324 0.7357 0.7389 0.7422 0.7454 0.7486 0.7517 0.7549
0.7 0.7580 0.7611 0.7642 0.7673 0.7704 0.7734 0.7764 0.7794 0.7823 0.7852
0.8 0.7881 0.7910 0.7939 0.7967 0.7995 0.8023 0.8051 0.8078 0.8106 0.8133
If the
falls0.8212
between0.8238
two values
in the
table,0.8315
choose the
greater
z-statistic
0.9 critical
0.8159ratio
0.8186
0.8264
0.8289
0.8340
0.8365
0.8389
Choose z = 0.77

Convert the z-statistic into an order quantity:

Equivalently, Q = norminv(0.778,3192,1181) = 4096.003

utdallas.edu/~metin

Q z
3192 0.77 1181 4101

19

Another Example: Apparel Industry

How much to order? Parkas at L.L. Bean
Demand
D_i
4
5
6
7
8
9
10
11
12
13
14
15
16
17

Probabability
p_i
.01
.02
.04
.08
.09
.11
.16
.20
.11
.10
.04
.02
.01
.01

Cumulative Probability of demand Probability of demand

being this size or less, F()
greater than this size, 1-F()
.01
.99
.03
.97
.07
.93
.15
.85
.24
.76
.35
.65
.51
.49
.71
.29
.82
.18
.92
.08
.96
.04
.98
.02
.99
.01
1.00
.00

Expected demand is 1,026 parkas.

utdallas.edu/~metin

20

Parkas 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 per parka = v = 50-10=$40
Profit from selling parka = p-c = 100-45 = $55
Cost of understocking = $55/unit
Cost of overstocking = c-v = 45-40 = $5/unit

utdallas.edu/~metin

21

Optimal level of product availability

p = sale price; v = outlet or salvage price; c = purchase price
CSL = Probability that demand will be at or below order quantity
CSL later called in-stock probability
Raising the order size if the order size is already optimal
Expected Marginal Benefit of increasing Q= Expected Marginal Cost of Underage
=P(Demand is above stock)*(Profit from sales)=(1-CSL)(p - c)
Expected Marginal Cost of increasing Q = Expected marginal cost of overage
=P(Demand is below stock)*(Loss from discounting)=CSL(c - v)

Define Co= c-v; Cu=p-c

(1-CSL)Cu = CSL Co
CSL= Cu / (Cu + Co)
utdallas.edu/~metin

22

Order Quantity for a Single Order

Co = Cost of overstocking = $5
Cu = Cost of understocking = $55
Q* = Optimal order size

Cu
55
CSL P( Demand Q )

0.917
C u C o 55 5
*

utdallas.edu/~metin

23

Optimal Order Quantity

0.917

Optimal Order Quantity = 13(00)

utdallas.edu/~metin

24

Parkas at L.L. Bean

Expected demand = 10 (00) parkas

Expected profit from ordering 10 (00) parkas = $499

Approximate Expected profit from ordering 1(00) extra

parkas if 10(00) are already ordered

= 100.55.P(D>=1100) - 100.5.P(D<1100)

utdallas.edu/~metin

25

Parkas at L.L. Bean

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

utdallas.edu/~metin

26

Revisit L.L. Bean as a Newsvendor Problem

Total cost by ordering Q units:

C(Q) = overstocking cost

+ understocking cost

C (Q ) C o (Q x) f ( x)dx C u ( x Q ) f ( x)dx
0

dC (Q)
Co F (Q) Cu (1 F (Q)) 0
dQ

Marginal cost of overage at Q* - Marginal cost of underage at Q* = 0

Cu
F (Q ) P( D Q )
C o Cu
*

utdallas.edu/~metin

27

Safety Stock
Inventory held in addition to the expected
demand is called the safety stock
The expected demand is 1026 parkas but
we order 1300 parkas.
So the safety stock is 1300-1026=274 parka.

utdallas.edu/~metin

28

The Newsvendor Model:

Performance measures

29
utdallas.edu/~metin

Newsvendor model performance measures

For any order quantity we would like to evaluate the following

performance measures:
Expected lost sales
The average number of units demand exceeds the order quantity

Expected sales
The average number of units sold.

Expected left over inventory

The average number of units left over at the end of the season.

Expected profit
Expected fill rate
The fraction of demand that is satisfied immediately

In-stock probability
Probability all demand is satisfied

Stockout probability
Probability some demand is lost
utdallas.edu/~metin

30

Expected (lost sales=shortage)

ESC is the expected shortage in a season

ESC is not a percentage, it is the number of units, also see next page
Demand - Q if
Shortage
0
if

Demand Q
Demand Q

ESC E(max{Demand in a season Q,0})

ESC

(x - Q)f(x)dx

x Q

utdallas.edu/~metin

31

Inventory and Demand during a season

Inventory

0
Season
utdallas.edu/~metin

Demand
During a
Season

Upside
down

Inventory=Q-D

D, Demand
During
A Season

32

Upside
down

Shortage
=D-Q

Shortage
Season
Demand
utdallas.edu/~metin

D: Demand During s Season

Shortage and Demand during a season

33

Expected shortage during a season

First let us study shortage during the lead time

Expected shortage E (max{D Q,0})

( D Q) f

( D)dD where f D is pdf of D.

D Q

Ex:

d1 9 with prob p1 1/4

Q 10, D d 2 10 with prob p2 2/4 , Expected Shortage?

d 11 with prob p 1/4
3
3

Expected shortage max{0, (d i Q)} pi

i 1

11

(d Q)}P( D d )

d 10

1
2
1 1
max{0, (9 - 10)} max{0, (10 - 10)} max{0, (11 - 10)}
4
4
4 4
utdallas.edu/~metin

34

Expected shortage during a season

Ex:

Q 10, D Uniform(6,12), Expected Shortage?

D 12

1 10 2

1
1 D2
1 12 2
Expected shortage ( D 10) dD
10 D

10(12)
10(10)
6
6 2
6 2
6 2

D 10
D 10
2/6
12

utdallas.edu/~metin

35

Expected lost sales of Hammer 3/2s with Q = 3500

Normal demand with mean 3192, standard deviation=1181
Step 1: normalize the order quantity to find its z-statistic.
z

Q 3500 3192

0.26

1181

Step 2: Look up in the Standard Normal Loss Function Table the expected lost
sales for a standard normal distribution with that z-statistic: L(0.26)=0.2824 see
Appendix B table on p.380 of the textbook
or, in Excel L(z)=normdist(z,0,1,0)-z*(1-normdist(z,0,1,1)) see Appendix D on p.389

Step 3: Evaluate lost sales for the actual normal distribution:

Expected lost sales L ( z ) 1181 0.2824 334
Keep 334 units in mind, we shall repeatedly use it

utdallas.edu/~metin

36

Measures that follow expected lost sales

Demand=Sales+Lost Sales
D=min(D,Q)+max{D-Q,0} or min(D,Q)=D- max{D-Q,0}
Expected sales = - Expected lost sales
= 3192 334 = 2858

Inventory=Sales+Left Over Inventory

Q=min(D,Q)+max{Q-D,0} or max{Q-D,0}=Q-min(D,Q)
Expected Left Over Inventory = Q - Expected Sales
= 3500 2858 = 642

utdallas.edu/~metin

37

Measures that follow expected lost sales

Expected total underage and overage cost with (Q=3500)
=70*334 + 20*642
Expected profit Price-Cost Expected sales
Cost-Salvage value Expected left over inventory
$70 2858 $20 642 $187, 221

What is the relevant objective? Minimize the cost or maximize the profit?
Hint: What is profit + cost? It is 70*3192=Cu*, which is a constant.

utdallas.edu/~metin

38

Type I service measure: Instock probability = CSL

Instock probability: percentage of seasons without a stock out
For example consider 10 seasons :
11 0 111 0 1 0 1
Instock Probability
10
Write 0 if a season has stockout, 1 otherwise
Instock Probability 0.7
Instock Probability 0.7 Probability that a single season has sufficient inventory
[Sufficient inventory] [Demand during a season Q]

InstockProbability P(Demand Q)
utdallas.edu/~metin

39

Instock Probability with Normal Demand

N(,) denotes a normal demand with mean and standard deviation
P N ( , ) Q
Normdist(Q, , ,1)

P N ( , ) Q Taking out the mean

N ( , ) Q

Dividing by the StDev

Q
P N (0,1)
Obtaining standard normal distribution

Normdist
,0,1,1

utdallas.edu/~metin

40

Example: Finding Instock probability for given Q

= 2,500 /week; = 500; Q = 3,000;
Instock probability if demand is Normal?

Instock probability = Normdist((3,000-2,500)/500,0,1,1)

utdallas.edu/~metin

41

Example: Finding Q for given Instock probability

= 2,500/week; = 500; To achieve Instock
Probability=0.95, what should be Q?

Q = Norminv(0.95, 2500, 500)

utdallas.edu/~metin

42

Type II Service measure: Fill rate

Recall:
Expected sales = - Expected lost sales = 3192 334 = 2858

Expected sales
Expected sales

Expected demand

Expected lost sales 2858

1

3192
89.6%

Expected fill rate

Is this fill rate too low?

Well, lost sales of 334 is with Q=3500, which is less than optimal.

utdallas.edu/~metin

43

Service measures of performance

100%
90%
80%

Expected fill
rate

70%
60%
50%

In-stock probability
CSL

40%
30%
20%
10%
0%
0

1000

2000

3000

4000

5000

6000

7000

Order quantity
utdallas.edu/~metin

44

Service measures: CSL and fill rate are different

inventory

CSL is 0%, fill rate is almost 100%

time
inventory

utdallas.edu/~metin

CSL is 0%, fill rate is almost 0%

time

45

Summary

Determine the optimal level of product availability

Demand forecasting
Profit maximization / Cost minimization

Other measures

Expected shortages = lost sales

Expected left over inventory
Expected sales
Expected cost
Expected profit
Type I service measure: Instock probability = CSL
Type II service measure: Fill rate

utdallas.edu/~metin

46

Homework Question:
A newsvendor can price a call option

Q1: Suppose that you own a simple call option for a

stock with a strike price of Q. Suppose that the price of
the underlying stock is D at the expiration time of the
option. If D <Q, the call option has no value.
Otherwise its value is D-Q. What is the expected value
of a call option, whose strike price is $50, written for a
stock whose price at the expiration is normally
distributed with mean $51 and standard deviation $10?

utdallas.edu/~metin

47