Introduction to the Theory and Practice of Yield Management

This article was downloaded by: [85.247.199.130] On: 29 August 2015, At: 10:33Publisher: Institute for Operations Research and the Management Sciences (INFORMS)INFORMS is located in Maryland, USA

INFORMS Transactions on Education

Publication details, including instructions for authors and subscription information:http://pubsonline.informs.org

Introduction to the Theory and Practice of YieldManagementSerguei Netessine, Robert Shumsky,

To cite this article:Serguei Netessine, Robert Shumsky, (2002) Introduction to the Theory and Practice of Yield Management. INFORMSTransactions on Education 3(1):34-44. http://dx.doi.org/10.1287/ited.3.1.34

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial useor systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisherapproval, unless otherwise noted. For more information, contact [email protected].

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitnessfor a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, orinclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, orsupport of claims made of that product, publication, or service.

Copyright © 2002, INFORMS

Please scroll down for article—it is on subsequent pages

INFORMS is the largest professional society in the world for professionals in the fields of operations research, managementscience, and analytics.For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org

http://pubsonline.informs.org

http://dx.doi.org/10.1287/ited.3.1.34

http://pubsonline.informs.org/page/terms-and-conditions

http://www.informs.org

NETESSINE AND SHUMSKYIntroduction to the Theory and Practice of Yield Management

INFORMS Transcations on Education 3:1 (34-44) INFORMS34

Introduction to the Theory and Practice ofYield Management1

Serguei NetessineThe Wharton School

University of Pennsylvania

[email protected]

Robert ShumskyW. E. Simon Graduate School of

Business AdministrationUniversity of Rochester

[email protected]

1. Introduction

A variety of concepts and analytical tools fall underthe label yield management. The term is used in manyservice industries to describe techniques to allocatelimited resources, such as airplane seats or hotelrooms, among a variety of customers, such as busi-ness or leisure travelers. By adjusting this alloca-tion a firm can optimize the total revenue or "yield"on the investment in capacity. Since these techniquesare used by firms with extremely perishable goods,or by firms with services that cannot be stored at all,these concepts and tools are often called perishableasset revenue management.

The techniques of yield management are relativelynew - the first research to deal directly with theseissues appeared less than 20 years ago. These days,yield management has been an enormously impor-tant innovation in the service industries. AmericanAirlines credits yield management techniques for arevenue increase of $500 million/year and DeltaAirlines uses similar systems to generate additionalrevenues of $300 million per year2 While the air-lines are the oldest and most sophisticated users ofyield management, these practices also appear in

other service industries. For example, Marriott Ho-tels credits its yield management system for addi-tional revenues of $100 million per year, with rela-tively small increases in capacity and costs. Broad-casting companies use yield management to deter-mine how much inventory (advertising slots) to sellnow to the "upfront market" and how much to re-serve and perhaps sell later at a higher price to the"scatter market." Even manufacturers are using yieldmanagement techniques to increase profits. After all,manufacturing capacity is as perishable as an air-line seat or an advertising slot - if it is not used whenit is available, that opportunity to use the capacity isgone forever.

The purpose of this note is to introduce the funda-mental concepts and trade-offs of yield managementand to describe the parallels between yield manage-ment and the newsvendor framework that is an im-portant model for inventory management. In par-ticular, this note focuses on how a manager mightallocate perishable inventory among a variety ofcustomer segments. To acquire some intuition aboutthe problem we will spend most of our time with anapplication that involves two types of customers.Consider an establishment we will call the EastmanTowers Hotel of Rochester, NY. The hotel has 210King/Queen rooms used by both business and lei-sure travelers. The hotel must decide whether to sellrooms well in advance at a relatively low price (i.e.,to leisure travelers), or to 'hold out' and wait for asale at a higher price to late-booking business trav-elers.

Our solution to this problem will be simple - wewill find a single 'cap' or 'booking limit' for the num-ber of rooms to sell in advance to leisure travelers.There are also variations of the problem and the so-lution that are more complicated, and these varia-tions can improve the performance of the yield man-agement system. For example, we might change thebooking limit up or down as time passes, or we mightassume that some rooms may be used only for somecustomers while other rooms are more flexible (think

1We would like to thank anonymous referees, Krishnan Anand, Gerard Cachon, Yu-Sheng Zheng, MBA students inthe OPIM632 class at the Wharton School and the OMG402 class at the Simon School for comments and sugges-tions.2Andrew Boyd, “Airline Alliance Revenue Management”'. OR/MS Today, Vol. 25, October 1998.

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



of 'economy' and 'deluxe' rooms). We will discussthese extensions in the last section of this note.

In general, the yield management/booking limit de-cision is just one aspect of a more general businessquestion: How should a firm market and distributegoods to multiple customer segments? To answerthis question, a firm must use tools for pricing andforecasting as well as for inventory management.All of these tools are often grouped under the termrevenue management, although the boundaries be-tween yield management and revenue managementare often ambiguous. In this note we will specifi-cally focus on just a single aspect of revenue man-agement - yield management.

We next identify the environments in which yieldmanagement techniques are most successful. Sec-tions 3-5 examine a specific solution to the basicEastman Towers problem, while Section 6 will dis-cuss a similar important problem, 'overbooking.' InSection 7 we describe complications and extensionsof the basic problem. Appendix A is a list of sug-gested additional readings on yield management,Appendix B contains exercises that test our under-standing of the yield management problem3 .

2. Where and Why Firms Practice Yield Man-agement

Business environments with the following five char-acteristics are appropriate for the practice of yieldmanagement (in parentheses we apply each charac-teristic to the Eastman Towers Hotel):

1.It is expensive or impossible to store excess re-source (we cannot store tonight's room for useby tomorrow night's customer).

2.Commitments need to be made when future de-mand is uncertain (we must set aside rooms forbusiness customers - "protect" them from low-priced leisure travelers - before we know howmany business customers will arrive).

3.The firm can differentiate among customer seg-ments, and each segment has a different demandcurve (purchase restrictions and refundability re-

quirements help to segment the market betweenleisure and business customers. The latter aremore indifferent to the price.).

4.The same unit of capacity can be used to de-liver many different products or services (roomsare essentially the same, whether used by busi-ness or leisure travelers).

5.Producers are profit-oriented and have broadfreedom of action (in the hotel industry, with-holding rooms from current customers for fu-ture profit is not illegal or morally irresponsible.On the other hand, such practices are contro-versial in emergency wards or when allocatingorgans for transplantation).

Given these characteristics, how does yield man-agement work? Suppose that our hotel has estab-lished two fare classes or buckets: full price and dis-count price. The hotel has 210 rooms available forMarch 29 (let us assume that this March 29 is aMonday night). It is now the end of February, andthe hotel is beginning to take reservations for thatnight. The hotel could sell out all 210 rooms to lei-sure travelers at the discount price, but it also knowsthat an increasing number of business customers willrequest rooms as March 29 approaches and that thesebusiness customers are willing to pay full price. Tosimplify our problem, let us assume that leisure de-mand occurs first and then business demand occurs.Hence we must decide how many rooms we are will-ing to sell at the leisure fare or in other words, howmany rooms shall we protect (i.e., reserve) for thefull price payers. If too many rooms are protected,then there may be empty rooms when March 29 ar-rives. If too few are protected, then the hotel for-goes the extra revenue it may have received frombusiness customers.

Notice we assume that the hotel can charge two dif-ferent prices for the same product (a room). To sepa-rate the two customer segments with two differentdemand curves, the hotel must practice market seg-mentation. In the hotel example, we assume that thefirm can charge different prices to business and lei-

3Solutions are available for interested faculty upon request

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



sure customers4 . In order to differentiate betweenthese two groups, a firm often introduces bookingrules that create barriers or "fences" between mar-ket segments. For example, a Saturday-night staymay be required to receive a discounted room onMonday, because most business travelers prefer togo home on the weekend while leisure customer aremore likely to accept, or may even prefer, the week-end stay. Again, leisure customers are more price-sensitive so the hotel wishes to sell as many roomsto business customers at a higher price as possiblewhile keeping room utilization high.

Booking Limits and Protection Levels

Before we look at the mathematics that will help usto make this decision, some vocabulary would behelpful. We define a booking limit to be the maxi-mum number of rooms that may be sold at the dis-count price. As we noted previously, we assume thatleisure customers arrive before business customersso the booking limit constraints the number of roomsthat these customers get: once the booking limit isreached, all future customers will be offered the fullprice. The protection level is the number of roomswe will not sell to leisure customers because of thepossibility that business customers might book laterin time. Since there are 210 rooms available in thehotel, and just two fare classes, in our example,

booking limit = 210 - the protection level

Therefore, the hotel's task is to determine either abooking limit or a protection level, since knowingone allows you to calculate the other. We shall evalu-ate the protection level. Suppose the hotel considersprotection level 'Q' instead of current protection levelQ+1 (Q might be anything from 0 to 209). Further,suppose that 210-Q-1 rooms have already been sold(see Figure 1). Now a prospective customer callsand wishes to reserve the first 'protected' room atthe discount price.

Should the hotel lower the protection level from Q+1to Q and therefore allow the booking of the (Q+1)throom at the discount price? Or should it refuse thebooking to gamble that it will be able to sell thevery same room to a full price customer in the fu-ture? The answer, of course, depends on (i) the rela-tive sizes of the full and discount prices and (ii) theanticipated demand for full price rooms. The deci-sion is illustrated as a decision tree in Figure 2.

Figure 1: Protection Level and Booking Limit in the Hotel

4As we mentioned in the introduction, finding the appropriate price for each customer segment is an important part ofrevenue management. However, the pricing problem is not within the scope of this note. Here we assume that pricesare established in advance.

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



In the next section we associate numbers withthisdecision and find the optimal protection level,Q*.

Solving the Problem

To determine the value of each branch of the deci-sion tree in Figure 2, we need to know the probabil-ity for each 'chance' branch and the values at theend of the branches. Suppose that the discount priceis $105 per night while the full price is $159 pernight. To find the probability on each branch, de-fine random variable D to represent the anticipateddemand for rooms at the full price. The hotel mayestimate the distribution of D from historical de-mand, as well as from forecasts based on the day ofthe week, whether there is a holiday, and other pre-dictable events. Here we will assume that the distri-bution is derived directly from 123 days of histori-cal demand, as shown in Table 1 below.

In the table, the 'Cumulative Probability' is the frac-tion of days with demand at or below the number ofrooms in the first column (Q).

Table 1. Historical demand for rooms at the full fare.

Figure 2 : Deciding the protection level

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



Now consider the decision displayed in Figure 2. Ifwe decide to protect the (Q+1)th room from sale, thenthat room may, or may not, be sold later. It will besold only if demand D at full fare is greater than orequal to Q+1, and this event has probability

1-F(Q). Likewise, the protected room will not besold if demand is less than or equal to Q, with prob-ability F(Q). Figure 3 shows our decision with thesealues included.

Now, F(Q) is the third column in Table 1. We sim-ply scan from the top of the table towards the bot-tom until we find the smallest Q with a cumulativevalue greater than or equal to 0.339. The answer hereis that the optimal protection level is Q*=79 with acumulative value of 0.341. We can now evaluate ourbooking limit: 210 - 79 = 131. If we choose a largerQ*, then we would be protecting too many roomsthereby leaving too many rooms unsold on average.

If we set Q* at a smaller value, we are likely to selltoo many rooms at a discount thereby turning awaytoo many business customers on average.

3. A General Formula

The solution described above is an example of a stan-dard technique that was developed for the airlineindustry. The technique was named "Expected Mar-ginal Seat Revenue" (EMSR) analysis by PeterBelobaba at MIT5. In our example we had two fareclasses with prices rL and rH (rH is the higher price,$159, while rL is the lower price, $105). We alsohad a random variable, D, to represent the distribu-tion of demand at the high fare. Since there are justtwo fare classes, the optimal booking limit for lowfare class is equal to the total capacity minus Q*.

The EMSR analysis can actually be described as anewsvendor problem. We make a fixed decision, Q,the protection level, then a random event occurs, thenumber of people requesting a room at the full fare.If we protect too many rooms, i.e., D<Q, then we

Given Figure 3, we can calculate the value of lower-ing the protection level from Q+1 to Q. Loweringthe protection level results in selling the (Q+1)throom at a discount which guarantees revenue of$105. Protecting Q+1 rooms has an expected valueequal to:

(1 - F(Q))($159) +F(Q)($0) = (1 - F(Q)) ($159)

Therefore, we should lower the protection level toQ as long as:

(1 - F(Q))($159) <= $105

or

F(Q) ($159 - $105) / $159 = 0.339.

Figure 3 : Booking Limit Decision with Data

5Peter P. Belobaba “Application of a Probabilistic Decision Model to Airline Seat Inventory Control”. OperationsResearch Vol. 37, No. 2, 1989.

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



end up earning nothing on Q-D rooms. So the over-age penalty C is rL per unsold room. If we protecttoo few rooms, i.e., D<Q, then we forgo rH-rL inpotential revenue on each of the D-Q rooms thatcould have been sold at the higher fare so the under-age penalty B is rH-rL. The critical ratio is

From the newsvendor analysis, the optimal protec-tion level is the smallest value Q* such that

4. Overbooking

Another important component in the yield manage-ment toolbox is the use of overbooking when thereis a chance that a customer may not appear. For ex-ample, it is possible for a customer to book a ticketon an airline flight and not show up for the depar-ture. If that is the case, the airline may end up flyingan empty seat resulting in lost revenue for the com-pany. In order to account for such no-shows, air-lines routinely overbook their flights: based on thehistorical rate of no-shows the firm books more cus-tomers than available seats. If, by chance, an unusu-ally large proportion of the customers show up, thenthe firm will be forced to 'bump' some customers toanother flight. Hotels, rental car agencies, some res-taurants, and even certain non-emergency health-careproviders also overbook. When determining the op-timal level of overbooking, the calculation is simi-lar to the calculation used for yield management.The optimal overbooking level balances (i) lost rev-enue due to empty seats and (ii) penalties (financialcompensation to bumped customers) and loss of

customer goodwill when the firm is faced with moredemand than available capacity.

Let X be the number of no-shows and suppose weforecast the distribution of no-shows with distribu-tion function F(x). Let Y be the number of seats wewill overbook, i.e., if the airplane has S seats thenwe will sell up to S+Y tickets6. As in the newsvendormodel, define the underage penalty by B and theoverage penalty by C. In this case C represents ill-will and the net penalties that are associated withrefusing a seat to a passenger holding a confirmedreservation (the 'net' refers to the fact that the airlinemay collect and hold onto some revenue from thispassenger as well). Here, B represents the opportu-nity cost of flying an empty seat. To explain further,if X>Y then we could have sold X-Y more seats andthose passengers would have seats on the plane. SoB equals the price of a ticket. If X<Y then we needto bump Y-X customers and each has a net cost of C.Thus, the formula for an optimal number overbookedseats takes following familiar form: is smallest valueY* such that

As an example, suppose the Eastman Towers Hoteldescribed above estimates that the number of cus-tomers who book a room but fail to show up on thenight in question is Normally distributed with mean20 and standard deviation 10. Moreover, EastmanTowers estimates that it costs $300 to "bump" a cus-tomer (the hotel receives no revenue from this cus-tomer, and the $300 is the cost of alternative ac-commodation plus a possible gift certificate for aone-night stay at Eastman in the future). On the otherhand, if a room is not sold then the hotel loses rev-enue equal to one night sold at a discount since at

'6Of course, the number of no-shows depends upon the total number of tickets that we sell. In the followingcalculations we assume that the distribution of X is the distribution of no-shows given that we sell exactly S tickets,and that the Y additional customers are guaranteed to show up. A more accurate model would account for no-shows among the extra Y customers, but for most applications this more complicated model would produce a

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



the very least this room could have been sold to lei-sure customers7 . How many bookings shouldEastman allow? The critical ratio is

From the Normal distribution table we get ϕ(-0.65)= 0.2578 and ϕ(-0.64) = 0.2611, so the optimal z* isapproximately -0.645 and the optimal number ofrooms to overbook is Y*=20-0.645*10=13.5 (onecan also use Excel's Norminv function for this cal-culation: "=Norminv(0.2592,20,10)" gives us ananswer of 13.5). If we round up to 14, this meansthat up to 210+14=224 bookings should be allowed.

5. Complications and Extensions

There are a wide variety of complications we facewhen implementing a yield management system.Here we discuss a few of the more significant chal-lenges.

Demand Forecasting

In the examples above, we used historical demandto predict future demand. In an actual application,we may use more elaborate models to generate de-mand forecasts that take into account a variety ofpredictable events, such as the day of the week, sea-sonality, and special events such as holidays. In someindustries greater weight is given to the most recentdemand patterns since customer preferences changerapidly. Another natural problem that arises duringdemand forecasting is censored data, i.e., companyoften does not record demand from customers whowere denied a reservation.

In our example in Sections 3-5 we used a discrete,empirical distribution to determine the protectionlevel. A statistical forecasting model would gener-ate a continuous distribution, such as a Normal or tdistribution. Given a theoretical distribution and itsparameters, such as the mean and variance, we wouldagain place the protection level where the distribu-

tion has a cumulative probability equal to the criti-cal fractile.

Dynamic Booking Limits

By observing the pattern of customer arrivals, firmscan update their demand forecasts, and this may leadto changes in the optimal booking limit. In fact, manyairline yield management systems change bookinglimits over time in response to the latest demandinformation. For example, it is possible that an air-line may raise a booking limit as it becomes clearthat demand for business-class seats will be lowerthan was originally expected. Therefore, during oneweek a leisure customer may be told that economy-class seats are sold out, but that same customer maycall back the next week to find that economy seatsare available.

Variation and Mobility of Capacity

Up to now, we have assumed that all units of capac-ity are the same; in our Eastman example we as-sumed that all 210 hotel rooms were identical. How-ever, rooms often vary in size and amenities. Air-lines usually offer coach and first-classes. Car rentalfirms offer subcompact, compact, and luxury cars.In addition, car rental firms have the opportunity tomove capacity among locations to accommodatesurges in demand, particularly when a central officemanages the regional allocation of cars. The EMSRframework described above can sometimes beadapted for these cases, but the calculations are muchmore complex. Solving such problems is an area ofactive research in the operations community.

Mobility of Customer Segments

In the example of Sections 3-5 we assumed that aleisure customer who is not able to book a room atthe discount price (because of the booking limit) doesnot book any room from that hotel. In fact, someproportion of leisure customers who are shut outfrom discount rooms may then attempt book a roomat the full price. The possibility that a customer may'buy-up' complicates the model but, most modernyield management systems take such customermovements into account.

7Technically, this empty room could have been sold at either low or high price so it is possible that lost revenue isactually higher than we suggest. Also, often hotels charge a penalty for late cancellations.

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



Nonlinear Costs for Overbooking

In some cases the unit cost of overbooking increasesas the number of 'extra' customers increases. This isparticularly true if capacity is slightly flexible. Forexample, a rental car agency faced with an unex-pected surge of customers may be able to obtain, fora nominal price, a few extra cars from a nearby sis-ter facility. However, at some point the agency's ex-tra supply will run out - and then customers will belost, resulting in a much greater cost.

Customers in a Fare Class Are Nor All Alike

While two leisure travelers may be willing to paythe same price for a particular night's stay, one maybe staying for just one day while the other may oc-cupy the room for a week. A business traveler on anairplane flight may book a ticket on just one leg ormay be continuing on multiple legs. Not selling aticket to the latter passenger means that revenue fromall flight legs will be lost.

In each of these cases, the total revenue generatedby the customer should be incorporated into the yieldmanagement calculation, not just the revenue gen-erated by a single night's stay or a single flight leg.

There are additional complications when code-shar-ing partners (distinct airlines that offer connectingflights among one another) operate these flight legs.If code-sharing occurs, then each of the partners musthave an incentive to take into consideration the otherpartners' revenue streams.

A similar complication occurs when there are groupbookings. How should the hotel consider a groupbooking request for 210 rooms at a discount rate?Clearly, this is different from booking 210 individualrooms since denying a reservation to one out of 210group customers may mean that all 210 will be lostto the hotel. Frequently, companies restrict groupreservations during the peak seasons but up till nowthere are no general rules for handling group reser-vations.

Summary

Table 2 summarizes the impact of these issues onthree industries: airlines, hotels and car rental firms8

. Sophisticated yield management tools have beendeveloped in all three industries, and these tools takeindustry-specific factors into account. However, allof these tools are based on the basic EMSR modeldescribed above.

Table 1: Comparison of yield management applications

8Adopted from Carrol W. J. and R. C. Grimes. “Evolutionary change in product management: experiences the car rentalindustry.”' Interfaces, 1995, vol.25, no.5, 84-104.

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



Appendix A. Suggested further readings andcases.

Handbook of Airline Economics. 1995. AviationWeek Group, a division of McGraw-Hill Compa-nies, Chapters 47 - 50.

Handbook of Airline Marketing. 1998. AviationWeek Group, a division of McGraw-Hill Compa-nies, Chapters 23 - 30.

R. G. Cross. 1996. Revenue Management: Hard-Core Tactics for Market Domination. BroadwayBooks.

Dhebar, A. and A. Brandenburger. 1993. AmericanAirlines, Inc.: Revenue Management. Harvard Busi-ness School Case.

M. K. Geraghty and E. Johnson. 1997. “RevenueManagement Saves National Car Rental”. Interfaces.Vol.27, pp. 107-127.

J. I. McGill and G. J. Van Ryzin. 1999. “RevenueManagement: Research Overview and Prospects.”Transportation Science. Vol.33, pp. 233-256.

R. D. Metters. Yield Management at Pinko Air.Southern Methodist University Case.

B. C. Smith, J. F. Leimkuhler, and R. M. Darrow.1992. “Yield Management at American Airlines”.Interfaces. Vol.22, pp. 8-31.

K. T. Talluri and G. J. van Ryzin. 2002. The Theoryand Practice of Revenue Management. Kluwer Aca-demic Publishers, Dordrecht, The Netherlands. Tobe published.

G. J. van Ryzin. 1998. Transportation NationalGroup. Columbia University Case.

Appendix B. Suggested exercises.

Problem 1

Recall the yield management problem faced by theEastman Towers Hotel, as described in the note "In-troduction to the Theory and Practice of Yield Man-agement." A competing hotel has 150 rooms withstandard Queen-size beds and two rates: a full priceof $200 and a discount price of $120. To receive thediscount price, a customer must purchase the roomat least two weeks in advance (this helps to distin-guish between leisure travelers, who tend to bookearly, and business travelers who value the flexibil-ity of booking late). You may assume that if a lei-sure traveler is not able to get the discount rate, shewill choose to book at another hotel.

For a particular Tuesday night, the hotel estimatesthat the average demand by business travelers has amean of 70 rooms and a standard deviation of 29rooms. Assume that demand follows a Normal dis-tribution around the forecast.

a)Find the optimal protection level for full pricerooms (the number of rooms to be protected fromsale at a discount price).

b)Find the booking limit for discount rooms.

c)Suppose that for a short time, the hotel's fore-cast of business customer demand is biased up-ward: the forecast of 70 rooms is too high andfewer business customers appear, on average.Qualitatively describe the economic conse-quences of using the protection level and book-ing limit derived in (a) and (b).

d)Suppose that for a short time, the hotel's fore-cast of business customer demand is biaseddownward: the forecast of 70 rooms is too lowand more business customers appear, on aver-age. Qualitatively describe the economic con-sequences of using the protection level andbooking limit derived in (a) and (b).

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



Problem 2

An airline offers two fare classes for coach seats ona particular flight: full-fare class at $440/ticket andeconomy class at $218/ticket. There are 230 coachseats on the aircraft. Demand for full-fare seats hasa mean of 43, a standard deviation of 8, and the fol-lowing empirical distribution:

Economy-class customers must buy their ticketsthree weeks in advance, and these tickets are ex-pected to sell out.

a)Find the (i) protection level and (ii) booking limitfor low-fare seats.

b)Suppose that unsold seats may sometimes besold at the last minute at a very reduced rate(similar to USAirways' "esavers" for last-minutetravel). What effect will this have on the protec-tion level calculated in (a)?

The protection level (Q*) will be…

Circle one: Higher Lower The same

Justification:

Problem 3.

Sunshine Airlines flies a direct flight from Baltimoreto Orlando. The airline offers two fares: a 14-dayadvance purchase economy fare of $96 and a fullfare of $146 with no advance purchase requirement.In its yield management system, Sunshine has twofare classes and, based on historical demand pat-terns, has established a booking limit of 90 seats forthe economy fare class.

Because KidsWorld is located near Orlando, Sun-shine strikes a deal with KidsWorld Inc. The dealallows Sunshine to offer a 5-day KidsWorld pass toanyone who purchases an economy-class Sunshineticket (the customer pays $96 for the Sunshine ticketand an additional $180 for the KidsWorld pass, adiscount from the regular price of $200 for a 5-daypass purchased directly from KidsWorld). Sunshinepasses along the $180 pass revenue to KidsWorld,but as part of the agreement, Sunshine receives a$50 incentive or 'kick-back' from KidsWorld forevery pass sold through Sunshine.

Given that Sunshine has accepted the deal withKidsWorld, should Sunshine change the bookinglimits and protection levels for the Orlando flight?Should it make other changes to its fare class struc-ture? If not, why not? If yes, how? Your answershould be qualitative, but you should be as specificas possible.

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.



Problem 4.

WZMU is a television station that has 25 thirty-sec-ond advertising slots during each evening. The sta-tion is now selling advertising for the first few daysin November. They could sell all the slots now for$4,000 each, but because November 7 will be anelection day, the station may be able to sell slots topolitical candidates at the last minute for a price of$10,000 each. For now, assume that a slot not soldin advance and not sold at the last minute is worth-less to WZMU.

To help make this decision, the sales force has cre-ated the following probability distribution for last-minute sales:

a)How many slots should WZMU sell in advance?

b)Now suppose that if a slot is not sold in advanceand is not sold at the last minute, it may be usedfor a promotional message worth $2500. Nowhow many slots should WZMU sell in advance?

Problem 5

An aircraft has 100 seats, and there are two types offares: full ($499) and discount ($99).

a)While there is unlimited demand for discountfares, demand for full fares is estimated to bePoisson with mean l=20 (the table below givesthe distribution function). How many seatsshould be protected for full-fare passengers?

b)An airline has found that the number of peoplewho purchased tickets and did not show up fora flight is normally distributed with mean of 20and standard deviation of 10. The airline esti-mates that the ill will and penalty costs associ-ated with not being able to board a passengerholding confirmed reservation are estimated tobe $600. Assume that opportunity cost of fly-ing an empty seat is $99 (price that discountpassenger would pay). How much should air-line overbook the flight?

Dow

nloa

ded

from

info

rms.

org

by [

85.2

47.1

99.1

30]

on 2

9 A

ugus

t 201

5, a

t 10:

33 .

For

pers

onal

use

onl

y, a

ll ri

ghts

res

erve

d.

Introduction to the Theory and Practice of Yield Management

Documents

Transcript of Introduction to the Theory and Practice of Yield Management