Behavioral Operations Management (BOM)esa2017.wu.ac.at/keynote_elena_katok.pdf · Behavioral...
Transcript of Behavioral Operations Management (BOM)esa2017.wu.ac.at/keynote_elena_katok.pdf · Behavioral...
Behavioral Operations Management (BOM):
Application of Experimental Economics
Elena KatokUniversity of Texas at Dallas
What is Operations Management?
¡ Operations Management (OM) is an area of business that deals with:¡ Designing and controlling manufacturing and
service processes¡ Two broad areas:
¡ Strategic: Supply Chain Management¡ Tactical: Logistics
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 2
o RelatedtoOperationsResearch(OR)o Moreaboutproblemsandlessabout
methods
OM in the 60’s and 70’s used to be Logistics, and mostly about central planning
¡ Supply Chain Design¡ Job-shop scheduling¡ Lot Sizing¡ Queuing Systems¡ Inventory Management¡ …
Copyright © 2014 Elena Katok, University of Texas at Dallas. All rights reserved. 3
The scope of OM
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 4
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
Inventory
The scope of OM
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 5
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
Inventory
Operations Management Evolution
¡ 1980s-1990s: decentralized models (Game Theory)¡ Incentives¡ Contract Negotiation¡ Social Norms¡ Trust
¡ Requires a model of individual (firm) decision-making.
Copyright © 2014 Elena Katok, University of Texas at Dallas. All rights reserved. 6
Decentralized Model
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 7
RawmaterialsSuppliers Manufacturing
Distribution
WarehousingCustomers
Retailer
Behavioral Operations Management (BOM)
8
¡ Understand what decision-makers are doing
¡ Develops behaviorally-robust models¡ The Goal: to explain, predict and improve
operations.
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
Decentralized Model
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 9
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
The case of the newsvendor problem
10
Customers Retailer
The profit:
The expected profit to be maximized:
The optimal order quantity 𝑞∗must satisfy:
Critical Ratio (CR)
𝜋 𝑞, 𝐷 = 𝑝min 𝑞, 𝐷 − 𝑐𝑞
𝔼 𝜋 𝑞, 𝐷 = 1 − 𝐹 𝑞 𝑝 − 𝑐 𝑞 +2𝑓 𝑥 𝜋 𝑞, 𝑥 𝑑𝑥6
7
𝐹 𝑞∗ =𝑝 − 𝑐𝑝
Arrow (1951)Copyright © 2017 Elena Katok, University of
Texas at Dallas. All rights reserved.
Decision Bias in the Newsvendor Problemwith a Known Demand Distribution:
Experimental EvidenceMaurice E. Schweitzer • Gerard P. Cachon
Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania 19104Fuqua School of Business, Duke University, Durham, North Carolina 27708
[email protected] • [email protected] • www.duke.edu/˜gpc
In the newsvendor problem a decision maker orders inventory before a one period sellingseasonwith stochastic demand. If toomuch is ordered, stock is left over at the end of the period,
whereas if too little is ordered, sales are lost. The expected profit-maximizing order quantity iswell known, but little is known about how managers actually make these decisions. We describetwo experiments that investigate newsvendor decisions across different profit conditions. Resultsfrom these studies demonstrate that choices systematically deviate from those that maximizeexpected profit. Subjects order too few of high-profit products and too many of low-profitproducts. These results are not consistent with risk-aversion, risk-seeking preferences, ProspectTheory preferences, waste aversion, stockout aversion, or the consequences of underestimatingopportunity costs. Two explanations are consistent with the data. One, subjects behave as if theirutility function incorporates a preference to reduce ex-post inventory error, the absolute differencebetween the chosen quantity and realized demand. Two, subjects suffer from the anchoring andinsufficient adjustment bias. Feedback and training did not mitigate inventory order errors. Wesuggest techniques to improve decision making.(Behavioral Operations; Newsvendor Inventory Decisions; Decision Bias; Anchoring; MinimizingEx-Post Inventory Error)
1. IntroductionIn the newsvendor problem a manager sells a productduring a short selling season with stochastic demand.The manager has one opportunity to order inventorybefore the selling season, and no further replenish-ments are possible. If the order quantity is greater thanrealized demand, the manager must dispose of theremaining stock at a loss. If the order quantity is lowerthan realized demand, the manager forgoes someprofit. Therefore, in choosing an order quantity themanager must balance the costs of ordering too littleagainst the costs of ordering too much.The newsvendor problem applies in a broad array of
settings. For example, fashion apparel retailers often
must submit orders well in advance of a selling season,without opportunity for replenishment during the sea-son. A manufacturer might need to choose its capacity(i.e., its order quantity) before the launch of a newproduct, knowing that the new product will becomeobsolete quickly (e.g., computers or cellular phones).Special promotions usually present a similar problem:order too little and the retailer faces irate customers, butorder too much and the retailer incurs additional inven-tory holding costs as it slowly sells the excess inventory.The newsvendor model also applies to individual choiceproblems, such as health care financing and insurancepurchasing (Rosenfield 1986, Eeckhoudt et al. 1991).There is plenty of anecdotal evidence to suggest that
Management Science © 2000 INFORMSVol. 46, No. 3, March 2000 pp. 404–420
0025-1909/00/4603/0404$05.001526-5501 electronic ISSN
The case of the newsvendor problem
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 11
Findings
① The “Pull to Center“ Bias ② Asymmetry: low CR is further from optimal (closer to mean
demand)③ Little learning
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 12
Low profit: CR = 0.25 High profit: CR = 0.75
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 13
These results cannot be explained exclusively by risk aversion, risk-seeking preferences, loss aversion, waste aversion, stockout aversion, or underestimation of opportunity costs.
The “Pull to Center“ Bias Appears to be Generally Persistent…
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 14
High profit: CR = 0.75 Low profit: CR = 0.25
0
10
20
30
40
50
60
70
80
90
100
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Ord
er Q
uant
ity
Period
50
60
70
80
90
100
110
120
130
140
150
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 O
rder
Qua
ntity
Period
--- AverageDemand-n- AverageOrder__OptimalOrder
Source: Bolton and Katok (2008)
The “Pull to Center“ Bias Appears to be Generally Persistent…
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 15
Bostian, Holt, and Smith: Newsvendor “Pull-to-Center” Effect592 Manufacturing & Service Operations Management 10(4), pp. 590–608, © 2008 INFORMS
to center at the low order frequency as well as at thehigh order frequency. The pull-to-center effect is alsonot diminished in sessions involving doubled payoffs.
2. Baseline ExperimentConsider a version of the newsvendor game in whichthe retail demand for the product is uniform over!0"X# and unsold units have no salvage value. Unitsare ordered at an exogenous per-unit cost c and aresold at an exogenous per-unit price p. When calcu-lating the expected sales quantity for an order ofx units, it is convenient to separate the two proba-bilistic events of being understocked (sales below x)and stocked-out (potential sales greater than x). Theuniform-distribution assumption admits an intuitiveformula (1) for expected sales. The probability of astockout is 1 − x/X, in which case x units are sold;the product of these elements yields the first twoterms on the right-hand side of (1). The third termaddresses the understock condition: the probability ofbeing understocked is x/X, in which case x/2 unitswill be sold on average.
expected sales= x− x2
X+ x2
2X(1)
The expected profit function is the price times theexpected sales quantity in (1), minus the order cost:
E$%&x"d'(= p
!
x− x2
2X
"
− cx" (2)
where the expectation is taken with respect to the ran-dom demand d. Expected profits are maximized atan order quantity of x∗ = X&p − c'/p. The quadraticpayoff structure results from the linear nature of thepayoffs and the uniform demand distribution, a sce-nario that is easy to describe to subjects in an exper-iment. However, a secondary consequence is that theprofit function can be relatively flat in the neighbor-hood of x∗ for some values of p and c.Schweitzer and Cachon (2000) report laboratory
experiments that use this payoff structure. In theirbaseline design with X = 300 and p = 12 “francs,”they construct two treatments by setting the whole-sale cost at c= 3 and c= 9 francs. (The conversion ratebetween experimental francs and real dollars is 300to 1, which corresponds to a price of $0.04 and costs of$0.01 and $0.03.) The implied optimal order quantities
Figure 1 Experiment 1: Orders and Feedback Every Round
100
50
75
25
00 5 10 20 3015 25
Round
Ave
rage
ord
er q
uant
ityc = 1c = 2c = 3
Note. Average order quantities = dots. Optimal predictions = dashes.
are x∗ = 225 for the low-cost treatment and x∗ = 75 forthe high-cost treatment, which translate into x∗ = 75and x∗ = 25 on the X = 100 scale. Subjects repeatedthe newsvendor game for 15 rounds. The actual aver-age order quantities over all subjects and rounds were#x= 176 and #x= 134, respectively, which correspond to#x= 59 and #x= 45 on the X = 100 scale.Figure 1 summarizes the results of our laboratory
sessions that used the X = 100 scale over 30 rounds.4
The resale price is p = 4¢ and the three alternativecosts are c = 1¢, c = 2¢, and c= 3¢, which yield opti-mal order quantities of x∗ = 75, x∗ = 50, and x∗ = 25,respectively. Subjects were informed that each inte-ger demand quantity in the range 1–100 would beequally likely and that the demand realization in around would be independent of those from previousrounds.5 Thus, the treatments involving c= 1¢ andc = 3¢ have direct analogs to Schweitzer and Cachon
4 Participants were recruited from undergraduate economics classesat the University of Virginia, with the promise that they wouldeach be paid $6 on arrival and that any subsequent earningswould be paid in cash. The sessions were run using the VeconlabWeb-based experimental platform, which generated instructionson the fly to conform to the selected design. The interestedreader can visit the administration site to view the setup details(http://veconlab.econ.virginia.edu/admin.php, under the “Mar-kets” category) and the participation site to log into a session thathas been set up (http://veconlab.econ.virginia.edu/login.php).5 Subjects in our experiments were asked to “imagine a hard spinof a roulette wheel with 100 stops marked from 1 to 100, with eachstop being equally likely.”
Source: Bostian, Holt and Smith (2008)
CR = 0.75
CR = 0.25
CR = 0.50
The asymmetry result is less robust: this dataset does not show much of an asymmetry
Source: Zhao, Geng, Chao and Zhang (2011)
order quantities, we make t-test with statistic results shown in Table 2. The results indicate that
the order quantities of subjects in sessions 1, 2 and 3 are significantly higher than their respective
standard optimal order quantities, whereas the order quantities of subjects in sessions 4, 5, 6 and
7 are significantly lower than their respective standard optimal order quantities. Basically, for
this part of experiment results, the ordering decisions of subjects exhibited similar behaviors as
those reported in the literature: too-high for low-profit setting and too-low for high-profit setting,
as well as the “pull-to-center” effect (see, e.g., Schweitzer and Cachon (2000) and Bostian, et al.
(2008)). A little surprise is session 4: it is a mid-price setting with critical fractile 0.5, but the
order quantities of subjects are significantly lower than the standard optimal order quantity.
100 110 120 180 400 500 900
050
100
150
200
250
300
Selling price
Ord
er q
uant
ity
Figure 3: Box-and-whisker plot and standard optimal order quantities
Table 2: Comparison between subjects’ order quantities and standard optimal order quantities
Session t-value, p-value remark
1 t = 29.06, p < 0.0001 significantly higher
2 t = 26.64, p < 0.0001 significantly higher
3 t = 15.00, p < 0.0001 significantly higher
4 t = −6.35, p < 0.0001 significantly lower
5 t = −30.07, p < 0.0001 significantly lower
6 t = −26.03, p < 0.0001 significantly lower
7 t = −32.01, p < 0.0001 significantly lower
Although subjects exhibit the “pull-to-center” effect, they should be heterogeneous. In any
session, we can find that there are either subjects with too-high order quantities or subjects with
10
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 16
Source: Stangl and Katok and Thonemann (2017)
Robustness to framing and somewhat robust to gender
Framing Gender
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
OederQ
uantity
CRx100
OptimalorderquantityNeutralframeOptimalorderquantity
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
OederQ
uantity
CRx100
MaleFemaleOptimalorderquantity
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
…while this one shows low CR is better (closer to optimal than high CR)
Source: Ockenfels and Selten (2013)
0"
10"
20"
30"
40"
50"
60"
70"
80"
90"
100"
0" 10" 20" 30" 40" 50" 60" 70" 80" 90" 100"
Observed(Av
erage(Order(Qua
n/ty(
Op/mal(Order(=(100(x(CR(
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
0
75 030
60
90 075
0153045607590
Optimal orderquantity=100xCR
Observedaverageorde
rqua
ntity
Buyback
19Source: Becker-Peth, Katok and Thonemann (2013)
Asymmetry persists with the Buy-back contract
20
Hypothesis Tree Branches are hypotheses addressing questions at the base. Cut branches indicate hypothesesfound “unsatisfactory.”
What explain newsvendor behavior?
Risk Aversion/Risk Seeking
Loss Avoidance
Underestimating opportunity costs
Waste Aversion
Learning
Cognitive Dissonance
Prospect Theory with dynamic reference point
Prospect Theory with static reference point
Random errors
Reference dependence
Overconfidence
Mental accounting
Subject pool
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
Decentralized Model
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 21
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
Supply Chain Contracting
Supplier(Seller)
Manufacturer(Buyer)
Demand𝑫~𝑭()Marketprice𝑷
µ
This has the structure of the the much-studies Stackelberg Game:• Seller is the first mover: proposes a contract
terms (eg. 𝑤)• Buyer id the “Newsvendor” chooses order
quantity (𝑞) Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
Wholesale Price Contract
Supplier(Seller)
Manufacturer(Buyer)
Demand𝑫~𝑭()Marketprice𝑷
µ
𝜋?@AABCDE = (𝑤 − 𝑐)𝑞
𝜋FDGHCBDE = 𝑝min 𝐷, 𝑞 − 𝑤𝑞
𝜋IJHKKDB = 𝑝min 𝐷, 𝑞 − 𝑐𝑞
The Buyer pays the Seller 𝒘 for each unit ordered:
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
Double Marginalization
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 24
¡ The Buyer orders less than the efficient amount under the wholesale price contract.
¡ Contracts can be designed to “coordinate” the channel.
Buyback Contract
Supplier(Seller)
Manufacturer(Buyer)
Demand𝑫~𝑭()Marketprice𝑷
µ
𝜋?@AABCDE = 𝑤 − 𝑐 𝑞 − 𝑏max(0, 𝑞 − min 𝐷, 𝑞 )
𝜋FDGHCBDE = 𝑝min 𝐷, 𝑞 − 𝑤𝑞 + 𝑏max(0, 𝑞 − min 𝐷, 𝑞 )
𝜋IJHKKDB = 𝑝min 𝐷, 𝑞 − 𝑐𝑞
The Buyer pays the Seller 𝒘 for each unit ordered and the Seller return to the buyer 𝒃 for each un-sold unit:
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
Automated Buyer
¡ Coordinating Contracts don’t work well ¡ Buyback: Katok and Wu 2009, ¡ Service Level Agreements: Davis 2014, ¡ Advance Purchase discount: Davis et al.
(2014)Sellers do not take sufficient risk to coordinate the channel.
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 26
Human Buyer and Seller
ContractProcessContractStructure Ultimatum Offer Negotiating Price Negotiating Price
& OrderQuantityWholesalePrice(Davis&Hyndman2017)
63% 63% 77%
Buyback(Coordinating)Becker-Peth et al.(2017)
69% 76% 82%
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 27
Efficiency
Negotiation Contract Improves Efficiency(Becker-Peth, Katok and Thonemann working paper)
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 28
UB N HNRejected 163.12 57.29 156.69Manufacturer 725.17 932.05 1,079.16Supplier 1,242.94 1,211.22 1,252.03
0
500
1,000
1,500
2,000
2,500
3,000 69% 76% 82%
(6%) (6%)(2%)
100% efficiency
Wholesale Prices and Buybacks: Observed and Efficient
Wholesale Prices (𝑤) Rebate (𝑏)
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 29
UB RN HNObserved 63.61 60.12 58.58Efficient 71.92 66.75 65.62
0
10
20
30
40
50
60
70
UB RN HNObserved 18.06 23.95 25.04Efficient 53.65 48.64 46.52
0
10
20
30
40
50
60
70
… but sellers still don’t accept enough risk to coordinate the channels.
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved.
BOM Studies of Contracting in the Bilateral Monopoly Setting
30
0"
0.25"
0.5"
0.75"
1"
Ho"and"Zhang"(2008)"
Lim"and"Ho"(2007)"
Katok"and"Pavlov"(2013)"
Efficien
cy(
Study(
Wholesale"Price"
CoordinaCng"
IntervenCon"
0"
0.25"
0.5"
0.75"
1"
Ho"and"Zhang"(2008)"Two5Part"Tariff"
Ho"and"Zhang"(2008)"Quan=ty"Discount"
Lim"and"Ho"(2007)"Two5Block"Tariff"
Lim"and"Ho"(2007)"Three5Block"Tariff"
Katok"and"Pavlov"(2013)"
MOQ"
Katok"and"Pavlov"(2013)"MOQ5full"info"
Rejec=ons"
Efficiency"
Coordinating contracts do not improve efficiency.
The Handbook of Behavioral Operations Management
Edited by Donohue, Katok and LeiderTo be published by Wiley (soon)
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 31
ExamplesofotherBOMtopics
Supply Disruptions
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 32
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Gurnani,H.,Ramachandran,K.,Ray,S.,andY,Xia(2013).Orderingbehaviorundersupplyrisk:anexperimentalinvestigation,Manufacturing&ServiceOperationsManagement,16(1),61-75.
• Chaturvedi,A.,Beil,D.,andE.Katok,SplitAwardAuctions:InsightsfromTheoryandExperiments,ManagementScience,forthcoming.
Capacity Planning
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 33
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Hyndman,Kraiselburd &Watson(2013),Aligningcapacitydecisionsinsupplychainswhendemandforecastsareprivateinformation:Theoryandexperiment,Manufacturing&ServiceOperationsManagement15(1),102-117.
• Chen,Su, &Zhao(2012),ModelingBoundedRationalityinCapacityAllocationGameswiththeQuantalResponseEquilibrium,ManagementScience58(10),1952-1962.
Distribution Networks
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 34
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Rapoport,A.,Gisches,E.J.,&Mak,V.(2014).Distributeddecisionsinnetworks:Laboratorystudyofroutingsplittable flow.ProductionandOperationsManagement,23,314-331.
• Mak,V.,Gisches,E.J.,&Rapoport,A.(2015).Routevs.segment:Anexperimentonreal-timetravelinformationincongestiblenetworks.ProductionandOperationsManagement,24,947-960.
Retailing Issues
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 35
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• DeHoratius,Gürerk,Honhon,Hyndman,“UnderstandingtheBehavioralDriversofExecutionFailuresinRetailSupplyChains:AnExperimentalStudyUsingVirtualReality”WorkingPaper.
• Vorotyntseva,Hanhon,Katok,“CanManagersPlanAssortments?AnExperimentalStudy”On-goingproject.
Queueing
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 36
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Kremer,M.,L.Debo.2012.Herdinginaqueue:Alaboratoryexperiment.Workingpaper,UniversityofChicagoBoothSchoolofBusiness.
• Kremer,M.,L.Debo.2016.Inferringqualityfromwaittimes.ManagementScience62(10)3023–3038.
Strategic Consumers
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 37
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Ovchinnikov andMilnerRevenuemanagementwithend-of-perioddiscountsinthepresenceofcustomerlearning,Productionandoperationsmanagement21(1),69-84.
• Kremer,Mantin andOvchinnikovDynamic PricinginthePresenceofMyopicandStrategicConsumers:TheoryandExperiment,ProductionandOperationsManagement26(1),116-133
Forecasting
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 38
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Kremer,M.,Moritz,B.,&Siemsen,E.(2011).Demandforecastingbehavior:Systemneglectandchangedetection.ManagementScience,57(10),1827-1843.
• Kremer,M.,Siemsen,E.,&Thomas,D.J.(2015).TheSumandItsParts:JudgmentalHierarchicalForecasting.ManagementScience,forthcoming.
• Moritz,B.,Siemsen,E.,&Kremer,M.(2014).Judgmentalforecasting:Cognitivereflectionanddecisionspeed.ProductionandOperationsManagement,23(7),1146-1160.
Information Sharing (Trust)
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 39
RawmaterialsSuppliers
Manufacturing
Distribution
WarehousingCustomers
Retailer
• Ozer,ZhengandChen(2011)TrustinForecastInformationSharing,ManagementScience57(6)1111-1137.
Manufacturing
Trusting the Forecast
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 40
RawmaterialsSuppliers
Distribution
WarehousingCustomers
Retailer
• GaryBolton&Katok,CryWolforEquivocate?Credibleforecastguidanceinacost-lossgame.ForthcomingatManagementScience2016.
• Stangl,Bolton&Katok,TrustingtheForecast:TheRoleofNumeracy,workingpaper
Conclusion
¡ Experimental Economics had a great deal of impact.
¡ These methods are successfully used in many disciplines.
Copyright © 2017 Elena Katok, University of Texas at Dallas. All rights reserved. 41