8/10/2019 On Sc Newsvendor
1/44
utdallas.edu /~metin
1
Optimal Level of Product Availability
Chapter 12 of Chopra
8/10/2019 On Sc Newsvendor
2/44
8/10/2019 On Sc Newsvendor
3/44
utdallas.edu /~metin
3
Motivating News Article:Mattel, Inc. & Toys R s
Mattel !w"o intro#$ce# %ar ie in 1'(' an# r$n a stoc) o$t *orseveral years t"en on+ was "$rt last year y inventory c$t ac)s atToys R s, an# o**icials are also eager to avoi# a re eat o* t"e1''- T"an)sgiving wee)en#. Mattel "a# e ecte# to s"i a lot o*/erc"an#ise a*ter t"e wee)en#, $t retailers, wary o* e cessinventory , sto e# or#ering *ro/ Mattel. T"at le# t"e co/ any tore ort a 0( /illion sales s"ort*all in t"e last wee)s o* t"e year ...or t"e cr$cial "oli#ay selling season t"is year, Mattel sai# it willre $ire retailers to lace t"eir *$ll or#ers e*ore T"an)sgiving . An#,*or t"e *irst ti/e, t"e co/ any will no longer ta)e reor#ers in4ece/ er, Ms. %ara# sai#. T"is will ena le Mattel to tailor
ro#$ction /ore closely to #e/an# an# avoi# $il#ing inventory*or or#ers t"at #on5t co/e.
6 7all 8treet 9o$rnal, e . 1-, 1'''
8/10/2019 On Sc Newsvendor
4/44
utdallas.edu /~metin
;ey =ow /$c" s"o$l# Mattel or#er>7ill Mattel?s action "el or "$rt ro*ita ility>7"at actions can i/ rove s$ ly c"ain ro*ita ility>
8/10/2019 On Sc Newsvendor
5/44
utdallas.edu /~metin
(
Another !ample" Apparel #ndustry$o% much to order& Parkas at L'L' (ean
@ ecte# #e/an# is 1, 2 ar)as.
8/10/2019 On Sc Newsvendor
6/44
utdallas.edu /~metin
Parkas at L'L' (ean
Bost er ar)a C c C 0 (8ale rice er ar)a C C 014isco$nt rice er ar)a C 0(=ol#ing an# trans ortation cost C 018alvage val$e er ar)a C s C ( 61 C0
Dro*it *ro/ selling ar)a C 6c C 1 6 ( C 0((Bost o* overstoc)ing C c6s C (6 C 0(
8/10/2019 On Sc Newsvendor
7/44
8/10/2019 On Sc Newsvendor
8/44utdallas.edu /~metin-
Order uantity for a Single Order
C o = Cost of overstocking ) 34
C u = Cost of understocking ) 344
Q* ) Optimal order si,e
'1E.(((
((GF H =
+=
+=
ou
u
C C
C Q Demand P CSL
8/10/2019 On Sc Newsvendor
9/44utdallas.edu /~metin'
Optimal Order uantity
5
5'2
5'6
5'7
5'8
1
1'2
6 4 7 9 8 : 15 11 12 1; 16 14 17 89
CumulativeProbability
Optimal Order Quantity = 13(00)
.'1E
8/10/2019 On Sc Newsvendor
10/44utdallas.edu /~metin1
Parkas at L'L' (ean
@ ecte# #e/an# C 1 FK G ar)as@ ecte# ro*it *ro/ or#ering 1 FK G ar)as C 0 ''
A ro i/ate @ ecte# ro*it *ro/ or#ering 1FK G e tra ar)as i* 1 F? G are alrea#y or#ere#
C 1 '44 'P-D
8/10/2019 On Sc Newsvendor
11/44utdallas.edu /~metin11
Parkas at L'L' (ean
A##itional1 s
@ ecte#Marginal %ene*it
@ ecte#Marginal Bost
@ ecte# MarginalBontri $tion
11 t" (( . ' C 2 '( ( .(1 C 2(( 2 '(62(( C 2
12 t" (( .2' C 1('( ( .E1 C 3(( 1('(63(( C 12
13 t" (( .1- C '' ( .-2 C 1 '' 6 1 C (-
1 t" (( . - C ( .'2 C 6 C 62
1( t" (( . C 22 ( .' C - 22 6 - C 62
1 t" (( . 2 C 11 ( .'- C ' 11 6 ' C 63-
1Et" (( . 1 C (( ( .'' C '( ((6 '( C 6
8/10/2019 On Sc Newsvendor
12/44utdallas.edu /~metin12
Revisit . . %ean as a Newsven#or Dro le/Total cost y or#ering < $nits: BF
8/10/2019 On Sc Newsvendor
13/44utdallas.edu /~metin13
Ordering >omen?s Designer (oots@nder Capacity Constraints
A$t$/n Ieaves R$**leRetail rice 01( 02 02(
D$rc"ase rice 0E( 0' 0118alvage rice 0 0( 0'
Mean 4e/an# 1 ( 2(8tan#. #eviation o* #e/an# 2( 1E( 12(
Available Store Capacity ) 1 455'
8/10/2019 On Sc Newsvendor
14/44utdallas.edu /~metin1
Assuming Bo Capacity Constraints
Autumn Leaves +uffle
pi ! i 1( 6E(C0E( 2 6' C011 2( 611 C01:! i " i E(6: C03( ' 6 ( C 0: 11 6' C 02
Britical ractile E(M11 C .A- 11 M1( C .E3 1: M1A C .-E(Ni .:E .A1 1.1(< i 111- A E 3':
8torage ca acity is not s$**icient to )ee all /o#elsO
8/10/2019 On Sc Newsvendor
15/44utdallas.edu /~metin1(
Algorithm for Ordering@nder Capacity Constraints
#$nitiali%ation&orAll ro#$cts, Q i ") 5 . +emaining capacity") otal capacity .
#$terati'e "tep&7"ile +emaining capacity P #o
orAll ro#$cts,Compute t"e marginal contribution o* increasing Q i y 1I* all marginal contributions QC , 8T D
#Order "i%e" are already "uffi!iently lar e for all produ!t"&else Eind t"e ro#$ct wit" t"e largest marginal contribution , call it F#Priority i'en to t e mo"t profita*le produ!t&Q j ") Q j 1 an# +emaining capacity)+emaining capacity.1 #Order more of t e mo"t profita*le produ!t&
8/10/2019 On Sc Newsvendor
16/44utdallas.edu /~metin1
Garginal Contribution)-p.c/P-D< /.-c.s/P-D= / Order uantity Garginal Contribution
+emaining Capacity Autumn Leaves +uffle Autumn Leaves +uffle1455 5 5 5 96'::9 15:'79: 1;7';7516:5 5 5 15 96'::9 15:'79: 1;4'711
1;75 5 5 165 96'::9 15:'79: 15:'7:11;45 5 5 145 96'::9 15:'79: 157'15;1;65 5 15 145 96'::9 15:'719 157'15;1;;5 5 25 145 96'::9 15:'46; 157'15;1;25 5 ;5 145 96'::9 15:'649 157'15;1;15 5 65 145 96'::9 15:';49 157'15;
8:5 5 ;85 2;5 96'::9 9;'5;; 95'195
885 15 ;85 2;5 96'::7 9;'5;; 95'195895 25 ;85 2;5 96'::4 9;'5;; 95'195
2:5 485 655 2;5 7:'889 79'622 95'195285 485 655 265 7:'889 79'622 74'151
1 988 667 274 4;'1:7 4;'197 42';4:5 98: 667 274 4;'59; 4;'197 42';4:
8/10/2019 On Sc Newsvendor
17/44utdallas.edu /~metin1E
Optimal Safety #nventory and Order Levels"-+OP / ordering model
ea# Ti/es
ti/e
inventory
8"ortage
An inventory cycle
R D
8/10/2019 On Sc Newsvendor
30/44
utdallas.edu /~metin3
Dost one/ent is a c"ea er way o* rovi#ing ro#$ct variety
4ell #elivers c$sto/i e# DB in a *ew #ays@lectronic ro#$cts are c$sto/i e# accor#ing to t"eir #istri $tion c"annelsToyota is ro/ising to $il# cars to c$sto/er s eci*ications an# #elivert"e/ in a *ew #ays
Increase# ro#$ct variety /a)es *orecasts *or in#ivi#$al ro#$cts inacc$rate ee an# %illington F1'' G re orts [ *orecast errors *or "ig" tec"nology ro#$cts
4e/an# s$ ly /is/atc" is a ro le/U =$ge en#6o*6t"e season inventory write6o**s. 9o"nson an# An#erson F2 G esti/ates
t"e cost o* inventory "ol#ing in DB $siness ( [ er year.
Not rovi#ing ro#$ct *le i ility lea#s to /ar)et loss. An A/erican tool /an$*act$rer *aile# to rovi#e ro#$ct variety an# lost
/ar)et s"are to a 9a anese co/ etitor. 4etails in McB$tc"eon et. al. F1'' G.Dost one/ent: 4elaying t"e co//it/ent o* t"e wor)6in6 rocess inventoryto a artic$lar ro#$ct , a.).a. en# o* line con*ig$ration, late oint#i**erentiation, #elaye# ro#$ct #i**erentiation.
8/10/2019 On Sc Newsvendor
31/44
utdallas.edu /~metin31
Dost one/ent
Dost one/ent is #elaying c$sto/i ation ste as /$c" as ossi le Nee#:
In#isting$is"a le ro#$cts e*ore c$sto/i ation B$sto/i ation ste is "ig" val$e a##e# n re#icta le #e/an# Negatively correlate# ro#$ct #e/an#s le i le 8B to allow *or any c"oice o* c$sto/i ation ste
8/10/2019 On Sc Newsvendor
32/44
utdallas.edu /~metin32
or/s o* Dost one/ent y \inn an# %owerso F1'--G
a eling ost one/ent: 8tan#ar# ro#$ct is la ele##i**erently ase# on t"e reali e# #e/an#. =D rinter #ivision laces la els in a ro riate lang$age on to rinters a*ter t"e
#e/an# is o serve#.
Dac)aging ost one/ent: Dac)aging er*or/e# at t"e#istri $tion center. In electronics /an$*act$ring, se/i6*inis"e# goo#s are trans orte# *ro/ 8@ Asia to
Nort" A/erica an# @$ro e w"ere t"ey are locali e# accor#ing to local lang$age an#
ower s$ ly
Asse/ ly an# /an$*act$ring ost one/ent: Asse/ lyor /an$*act$ring is #one a*ter o serving t"e #e/an#. Mc4onal#s asse/ les /eal /en$s a*ter c$sto/er or#er.
8/10/2019 On Sc Newsvendor
33/44
utdallas.edu /~metin33
@ a/ les o* Dost one/ent=D 4es)9et Drinters Drinters locali e# wit" ower s$ ly /o#$le, ower cor# ter/inators, /an$als
Asse/ ly o* I%M R8 8erver ( 6E( en# ro#$cts #i**erentiate# y 1 *eat$res or co/ onents. Asse/ ly $se# to start *ro/
scratc" a*ter c$sto/er or#er. Ta)es too long. Instea# I%M stoc)s se/i *inis"e# R8 calle# vanilla o es. Wanilla o es are c$sto/i e#
accor#ing to c$sto/er s eci*ication.Vilin Integrate# Birc$its 8e/i6*inis"e# ro#$cts, calle# #ies , are "el# in t"e inventory. or easily *ast c$sto/i a le
ro#$cts, c$sto/i ation starts *ro/ #ies an# no *inis"e# goo#s inventory is "el#. or /oreco/ licate# ro#$cts *inis"e# goo#s inventory is "el# an# is s$ lie# *ro/ t"e #ies inventory.
New rogra//a le logic #evices w"ic" can e c$sto/i e# y t"e c$sto/er $sing a s eci*ic
so*tware.Motorola cell "ones 4istri $tion centers "ave t"e cell "ones, "one service rovi#er logos an# service rovi#er
literat$re. T"e ro#$ct is c$sto/i e# *or #i**erent service rovi#ers a*ter #e/an# is/ateriali e#.
8/10/2019 On Sc Newsvendor
34/44
utdallas.edu /~metin3
Dost one/ent
8aves Inventory "ol#ing cost y re#$cing sa*ety stoc) Inventory ooling Resol$tion o* $ncertainty
8aves solescence costIncreases 8ales8tretc"es t"e 8$ ly B"ain
8$ liers Dro#$ction *acilities, re#esigns *or co/ onent co//onality 7are"o$ses
8/10/2019 On Sc Newsvendor
35/44
utdallas.edu /~metin3(
Wal$e o* Dost one/ent: %enetton case
Eor each color 25 %eeks in advance forecasts Gean demand) 1 555* Standard Deviation) 455
Eor each garment
Sale price ) 345 Salvage value ) 315 Production cost using option 1 -long lead time/ ) 325
U Dye the thread and then knit the garment
Production cost using option 2 -short lead time/ ) 322U Mnit the garment and then dye the garment
7"at is t"e val$e o* ost one/ent> C( J sC1 J cC2 or cC22
8/10/2019 On Sc Newsvendor
36/44
utdallas.edu /~metin3
Wal$e o* Dost one/ent: %enetton case
B8 CF 6cG F 6cLc6sGC3 C .E(
8/10/2019 On Sc Newsvendor
37/44
utdallas.edu /~metin3E
A ly o tion 2 to all ro#$cts: %enetton case
B8 CF 6cG F 6cLc6sGC2- C .E4e/an# is nor/al wit" /ean 1 an# st.#ev s rtF G (
8/10/2019 On Sc Newsvendor
38/44
utdallas.edu /~metin3-
Dost one/ent 4ownsi#e
%y ost oning all t"ree gar/ent ty es, ro#$ction costo* eac" ro#$ct goes $7"en t"is increase is s$ stantial or a single ro#$ct?s#e/an# #o/inates all ot"er?s Fca$sing li/ite#$ncertainty re#$ction via aggregation G, a artial
ost one/ent sc"e/e is re*era le to *$ll ost one/ent.
8/10/2019 On Sc Newsvendor
39/44
utdallas.edu /~metin3'
Dartial Dost one/ent: 4o/inating 4e/an#Color %ith dominant demand" Gean ) ; 155 SD ) 855Other three colors" Gean ) ;55 SD ) 255
@ ecte# ro*it wit"o$t ost one/ent C 01 2,2 (@ ecte# ro*it wit" ost one/ent C 0'',-E2
Are t"ese cases co/ ara le> Total e ecte# #e/an# is t"e sa/eC
Total variance originally C H2( , C1, , Total variance nowC- H- L3F2 H2 GC , L12 , CE ,
4o/inating #e/an# yiel#s less ro*it even wit" less total variance.Dost one/ent can not e any etter wit" /ore variance.
8/10/2019 On Sc Newsvendor
40/44
utdallas.edu /~metin
Dartial Dost one/ent: %enetton case
or eac" ro#$ct a art o* t"e #e/an# is aggregate#, t"e rest is notDro#$ce Q1 $nits *or eac" color $sing tion 1 an# Q $nitsFaggregateG $sing tion 2, res$lts *ro/ si/$lation:
< 1 *or eac" < A Dro*it
133E 0' ,(E
(2 0'-, '2
11 (( 0'',1-
1 -( 01 ,312
- 1(( 01 , 3
8/10/2019 On Sc Newsvendor
41/44
utdallas.edu /~metin1
Tailore# F4$alG 8o$rcing
Tailore# so$rcing #oes not /ean $ying *ro/ two ar itrary so$rces.T"ese two so$rces /$st e co/ le/entary: Dri/ary so$rce: ow cost, long lea# ti/e s$ lier
U Bost C 02 (, ea# ti/e C ' wee)s
Bo/ le/entary so$rce: =ig" cost, s"ort lea# ti/e s$ lier U Bost C 02( , ea# ti/e C 1 wee)
An e a/ le B7D FBra*te# 7it" Dri#eG o* a arel in#$stry ringingo$t co/ etitive a#vantages o* $ying *ro/ #o/estic s$ liers vsinternational s$ liers.Anot"er e a/ le is %enetton?s ractice o* $sing internationals$ liers as ri/ary an# #o/estic FItalianG s$ liers asco/ le/entary so$rces.
8/10/2019 On Sc Newsvendor
42/44
utdallas.edu /~metin2
Tailore# 8o$rcing: M$lti le 8o$rcing 8ites
8/10/2019 On Sc Newsvendor
43/44
utdallas.edu /~metin3
4$al 8o$rcing 8trategies *ro/ t"e8e/icon#$ctor In#$stry
8/10/2019 On Sc Newsvendor
44/44
Learning ObFectives
Optimal order Nuantities are obtained bytrading off cost of lost sales and cost of e!cessstock
Levers for improving profitability #ncrease salvage value and decrease cost of stockout #mproved forecasting uick response %ith multiple orders
Postponement ailored sourcing