What to order?
What are the issues?
A Sample Problem
Commit 10,000 units before show
Commit 10,000 units after show
Minimum of 600 units
A First Approach
Ignore differences in
Profit margins
Salvage values
Ignore minimum lot sizes
Consider only first order cycle
Sample Problem
Style
Mean Forecast Std Deviation in Demand
Gail
1,017
388
Isis
1,042
646
Entice
1,358
496
Assault
2,525
680
Teri
1,100
762
Electra
2,150
807
Stephanie
1,113
1,048
Seduced
4,017
1,113
Anita
3,296
2,094
Daphne
2,383
1,394
Normal Distribution
0.45
0.4
0.35
0.3
0.25
0.2
0.15
Std Dev.s
0.1
0.05
0
-5
-4
-3
-2
-1
0
1
2
3
4
5
Idea 1
Make all products equally likely to sell out
Choose a single std dev. To set production quotas for all products
What should the Std. Dev. Be?
Style
Gail
Isis
Entice
Assault
Teri
Electra
Stephanie
Seduced
Anita
Daphne
Mean
Forecast
1,017
1,042
1,358
2,525
1,100
2,150
1,113
4,017
3,296
2,383
Std Deviation in
Demand
388
646
496
680
762
807
1,048
1,113
2,094
1,394
Total Production
Order
Amount
1,017
1,042
1,358
2,525
1,100
2,150
1,113
4,017
3,296
2,383
20,001
Std. Devs
0
0
0
0
0
0
0
0
0
0
0
Probability of Sell out
Probability of Sell Out
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
Number of Standard Deviations
50%
Normal Distribution
0.45
Set order Qty to this many std. devs
0.4
Probability we discount last item =
Probability we stock out =
Probability
demand exceeds over qty =
0.86
0.35
0.3
0.25
Probability demand is smaller than order quantity =
0.2
0.15
Std Dev.s
0.1
0.05
0.14
0
-5
-4
-3
-2
-1
0
1
2
3
4
5
What’s Wrong with This?
What else should we be looking at?
Still just worried about
Order up to 10,000
One order cycle
No minimum order qty.
A Second Idea
Look at 1 Product
How to trade off risks of overstock
(discounting) vs risks of understock (lost sales)? If we order Q
The last item faces what risk of being discounted?
Probability Demand < Q = F(Q)
The last item faces what risk of selling out
Probability Demand > Q = 1 - F(Q)
We want to be indifferent
We want two to be equal
Expected loss from Overstock =
CO*F(Q)
Expected loss from Lost Sale = CL*(1F(Q))
A little Algebra:
F(Q) = CL/(CO+CL)
Example
Oversimplification
Lost Sale: CL = Selling Price - Cost
Discount: CO = Cost - Salvage Value
Electra:
Selling Price $173
Cost $ 50
Salvage
$ 0
Lost Sale: CL = $123
Discount: CO = 50
Want Probability of Discount = F(Q) = 123/173 = 0.71
Find Q with this cumulative probability: ~2,599
Balancing Risks
Style
Gail
Isis
Entice
Assault
Teri
Electra
Stephanie
Seduced
Anita
Daphne
Mean
Forecast
1,017
1,042
1,358
2,525
1,100
2,150
1,113
4,017
3,296
2,383
Probability of
Style
Sell Out
Gail
0.86
Isis
0.86
Entice
0.86
Assault
0.86
Teri
0.86
Electra
0.86
Stephanie
0.86
Seduced
0.86
Anita
0.86
Daphne
0.86
Std Deviation in
Demand
388
646
496
680
762
807
1,048
1,113
2,094
1,394
Total Production
Expect Cost of
Lost Sale
$
51.33
$
41.92
$
25.67
$
34.22
$
62.45
$
105.23
$
71.01
$
19.68
$
36.79
$
83.84
Order
Amount
Std. Devs
606
(1.06)
357
(1.06)
832
(1.06)
1,804
(1.06)
292
(1.06)
1,294
(1.06)
2
(1.06)
2,837
(1.06)
1,075
(1.06)
905
(1.06)
10,003
-1.0605
Probability of Sell out
Probability
Last Item is Discounted
0.14
0.14
0.14
0.14
0.14
0.14
0.14
0.14
0.14
0.14
Expect Cost of
Discount
$
7.22
$
7.22
$
7.22
$
7.22
$
7.22
$
7.22
$
7.22
$
7.22
$
7.22
$
7.22
Salvage
Price
Cost
Value
$
110 $
50 $
$
99 $
50 $
$
80 $
50 $
$
90 $
50 $
$
123 $
50 $
$
173 $
50 $
$
133 $
50 $
$
73 $
50 $
$
93 $
50 $
$
148 $
50 $