Airline Auctions

Airline Auctions X-Terrestrials Anuj Jenveja Mariette Fourmeaux du Sartel Menaka Shroff Sarah Diegnan Haas School of Business UC Berkeley 5/15/2008 Airline Auctions MBA 211 Executive Summary I. Background Given the rising costs of fuel, current security concerns and competitive online ticketing options, the airline industry is struggling to increase revenue. American Airlines alone cancelled approximately 179425 flights last year. Our paper focuses on looking at new ways to help airlines increase revenue. Using the game theoretic principles of auctions we created a model to sell excess airline capacity (seats). We used this model to evaluate auction impact on American Airline’s (AA) revenue and its implications of such a model in the airline industry. II. Methodology and Analysis To understand the target audience for an auction, we segmented the air travelers in three categories (a) Early Birds or Leisure travelers, (b) Business travelers and (c) Deal Hunters. Additionally, for simplicity of analysis we only used one key airline route from Dallas, Texas to Newark, New Jersey. The auction revenues were calculated using an English Auction. However since the Revenue Equivalence Theorem holds we would advice airlines to use a first price sealed bid auction. Finally, we performed several sensitivities on (a) Load Factors (b) ticket willingness to pay and (c) number of bidders. III. Recommendation Based on analysis outlined in this report, we recommend that American Airlines to launch a pilot program holding online auctions for excess tickets on their website. This will not only give them the first mover advantage but also open up new ways to making revenues. Additionally we recommend American Airlines to open the auction for excess seats seven days prior to departure. The target audience would be Business Travelers and Deal Hunters. The expected range of tickets prices for each segment would be competitively lower than third party websites. These ticket sales would represent pure revenue for the airlines because they would not be paying commissions to the third party websites. X – Terrestrials Page 2 of 24 Airline Auctions MBA 211 IV. Airline Industry The airline industry has been faced with a number of challenges recently. These have had a strong negative impact on airline revenues. Challenges Facing Airline Industry With the increasing popularity of the internet, comparative shopping sites have proliferated while the airline industry has struggled to break even. As a large percentage of travelers search for airline tickets on these sites, most airlines cannot afford to be present in this market. Yet, these sites require airlines to pay large commission, cutting deeply into the airline revenues. The exact amount of commission paid by the airlines is not widely available, it is known that hotels pay a 25% commission1 and the profit margin of third party websites is approximately 11%. From this information it can be assumed that the airlines roughly pay a 10-25% commission to these ticket sites. Following this blow to revenues, there was the 2001 downturn in airline transportation. The combination of the tech bubble bursting with the events of September 11th led to the greatest dip in airline traffic the industry had ever seen. As a result, several airlines declared Chapter 11 bankruptcy protection. Just starting in late 2006, airlines begun recuperating from this crisis. Today, the pressure on the airline industry comes in the form of skyrocketing oil prices. In the past, fuel costs have ranged from 10% to 15% of U.S. passenger airline operating costs. Currently, these have increased to between 30% and 50%. From 2000 to the first half of 2007, the price of jet fuel more than doubled and the US Consumer Price Index rose 19.5%, whereas the price to fly a mile fell 10.2%.2 1 2 Mr. Chip Connelly, Chairman and Chief Executive Officer, joie de vivre ATA Office of Economics X – Terrestrials Page 3 of 24 Airline Auctions MBA 211 Chart 1: Administrative cost allocation Chart 2: Fuel Costs, Cents per Gallons Source: ATA Office of Economics Load Factors Airlines keep a close eye on their load factor, the industry term for their capacity, as this drives their profitability. The load factor is calculated as the number of seats sold over the total number of available seats, either for a plane, for a flight route, for an entire airline or for the whole industry. Today, the industry break-even load factor is close to 80%.3 This value has sharply increased in recent years due to the increased fuel prices and the lower ticket prices. Airlines regularly operate very close to their breakeven load factors, with an additional seat sold often the difference between making or losing money. 3 Air Transport Association X – Terrestrials Page 4 of 24 Airline Auctions MBA 211 Chart 3: Load Factor Source: ATA Office of Economics American Airlines For this project, we look specifically at American Airlines (AA), the largest airline in the world with 2007 revenues of $22.9 billion.4 American Airlines has also experienced challenging times recently and has taken actions to increase its profitability such as canceling flights. The airline finally began breaking . even in 2006. In 2007, AA cancelled 17,942 flights to increase its load factor.5 Yet, even with this change, it continues to operate at a near breakeven 2007 load factor of 81.5%.6 V. Ticket Sales The airline industry uses many computer reservations systems e.g Sabre System, across the world. systems, Through these computer systems, airfares can be published quickly and efficient to the airlines sales efficiently channels. Computers also allow airlines to accurately predict, number of tickets sold allowing airlines to 4 5 American Airlines 2007 Annual Report Same AJ source as in first paragraph of paper 6 American Airlines and American Eagle Monthly Traffic Reports, December 2007. X – Terrestrials Page 5 of 24 Airline Auctions MBA 211 estimate excess capacity for each leg. Airline tickets are sold through online and offline (travel agents) channels. Listed below are two broad online categories: a. Bread and Butter Ticketing Websites. Tickets are available through sale using many aggregator websites such as Expedia, Travelocity, Orbitz etc. Additionally, many airlines, such as Southwest, Jet Blue, and American Airlines, sell tickets through their own websites. These ticket sales provide the airlines with the greatest portion of their revenues and are typically purchased at least 14 days in advance of travel. b. Excess Ticketing Websites: As a flight’s load factor approaches profitability, an additional passenger seat has a marginal cost of zero; the additional fuel and food needed for that customer has no significant bearing on the total cost of the flight. Therefore, as the travel date approaches, the airlines have increased pressure to sell excess seats at a lower price. To accomplish this, airlines provide their excess tickets to third party online sites, such as Priceline.com and Hotwire, both using auction as a method to sell opaque airline tickets. The opaque fare service, where one has to guess what price an airline is selling its tickets for, was the airlines' sales channel of last resort - the way they sold off tickets they didn't feel they could get rid of by any other means. But capacity cuts and the generally dire condition of the airline industry (as mentioned above) have reduced the number of empty seats available to Priceline and its competitor Hotwire. Trying to fill empty seats and to face down competition from lowfare airlines such as Southwest, the major airlines have been undercutting Priceline and Hotwire with their own sales. Currently Priceline and Hotwire both offer both bread and butter and the opaque ticket options. X – Terrestrials Page 6 of 24 Airline Auctions MBA 211 VI. Customer Segmentation In 2007, approximately 156.8million person-trips were taken by airplane. Of those trips 60% were for leisure purposes and 40% were business related travel7. Airline customers can be further broken-down into three types of segments depending on when tickets are purchased. The early birds generally purchase tickets more than 21 days in advance. Business travelers are usually on a tighter schedule due to last minute meetings and purchase tickets up to seven days in advance. The last consumer segment is the deal hunters who purchase tickets up to fourteen days in advance. How far in advance tickets are purchased, is the first signal that the airlines get from their customers. There exists a second signal which is used by the airlines to differentiate the business travelers from the deal hunters. Generally, business travelers travel during the weekdays while deal hunters desire a Saturday night stay-over. Based on these two signals, the airlines can distinguish the different customer segments and price their tickets accordingly. All three costumer segments have different willingness to pay. The early birds will demand the lowest price while the business travelers are less price sensitive because they must get to their destination regardless of price. The deal-hunters are looking for last minutes deals so their willingness to pay is between the early birds and the business travelers. Searches were conducted on all the major travel websites and American Airlines website in order to find the different prices that were associated with the three customer segments. All the flights were direct, round-trip tickets departing from Dallas - Fort Worth airport and arriving at Newark airport. Two airlines fly direct between these two locations, American Airlines and Continental. The first search was simulating the early bird travelers, regardless of travel dates as long as the tickets were purchased more than 21 days in advance the ticket prices ranged from $300 - $500 on all the websites. The next search 7 Domestic Travel Report, 2007 Edition, Travel Industry Association X – Terrestrials Page 7 of 24 Airline Auctions MBA 211 conducted was a ten day advance ticket purchase for a flight leaving on Wednesday morning and returning on Friday night to simulate the business traveler. The results can be seen in the table below. Business Travelers: Depart Wednesday Morning, Return Friday Night American Airlines Flight Expedia Travelocity Orbitz Priceline Hotwire Continental Website American Airlines Website $1,141 $938 $1,141 $1,163 $1,142 $992 Continental Flight $1,141 $1,000 $1,141 $1,163 $1,142 $1,142 - Table 1: Business Traveler fight pricing The next search was again a ten day advance purchase for a flight leaving on Wednesday morning but returning on a Sunday night to simulate a deal-hunter. The results are summarized in the table below. Deal Hunter: Depart Wednesday Morning, Return Sunday Night American Airlines Flight Expedia Travelocity Orbitz Priceline Hotwire Continental Website American Airlines Website $679 $461 $632 $750 $461 Continental Flight $642 $429 $642 $642 $663 $429 - Table 2: Deal Hunter flight pricing From these searches it can be seen that airlines do effectively use the two signals of advance purchase date and travel days in order to differentiate the customers segments. Once the travel websites and airline website know this information they are able to charge different ticket prices because of the differences in willingness to pay between the customer segments. X – Terrestrials Page 8 of 24 Airline Auctions MBA 211 VII. Auction Subtle differences in how an auction design could have significant impact on profits. Therefore, the auction we developed for American Airlines would be similar to airline auctions held on Priceline.com with a few key differences. First, the auction will be held online at American Airlines’ website, www.aa.com as a first price sealed bid auction. Second, seats for a specific departing flight will be auctioned one week in advance, which naturally eliminates the early birds customers since this customer type generally purchases tickets more than 21 days in advance. Therefore, the two customer segments that will tend to participate in this auction are business travelers and deal hunters. Ideally the auction would be run daily with bids submitted throughout the day and the “winners” announced at the end of day. This model does not work for the business traveler segment because they will require a shorter lag time between submitting a bid and being notified of a successful bid. Therefore, to minimize each bidder’s waiting time between the auction and its results, the auction would be held each hour by releasing only a few seats at a time. The number of seats released would be based on the number of excess seats available. The airline would have a secret reservation price to prevent extremely low ticket prices and too much price volatility for the same flight. A reservation is also needed so that in the occasion when the number of bidders is less than the number of seats available the price will not drop to an artificially low value. The reservation price will depend on the number of remaining seats available for the flight and on the specific customer segment(s) that demand the flight. As discussed in the “Customer Segmentation” section, business travelers have a higher willingness to pay than do deal hunters. Therefore, a flight that is demanded by more business travelers would retain a higher reservation price. We will also demonstrate in our model that fewer seats will generate a higher ticket price and therefore should have a higher reservation price. Finally, the auction will disclose the number of available seats for that X – Terrestrials Page 9 of 24 Airline Auctions MBA 211 auction round. This is designed to give each bidder more information which generally helps to increase bids. VIII. Auction Model Assumptions We made the following assumptions when we modeled the potential revenues per seat to the airline: • Revenue Equivalence Theorem – Although the actual auction on the airline’s website would be structured as a first price sealed bid auction, we can use the revenue equivalence theorem and model it as an English auction when estimating the expected revenues from the auction. This will significantly simply our analysis and eliminates the need for complicated calculus. Applying this theorem assumes that the auction is efficient, the bidders are free to exit and bidders are alike. All three of these assumptions are applicable to this auction. • Specific AA flight – Our model analyzes two roundtrip American Airlines flights from Dallas to Newark departing on a Wednesday. The day of the return flight depends on the customer segment, as shown in the chart under “bid ranges”. We analyzed seat prices for two separate auctions, one for business travelers and one for deal hunters. The aircraft used for these flights is an M83 with 124 seats available for purchase.8 • Load factor – The load factor used for this analysis was 82%9, which is above the airline industry’s estimated breakeven load factor of 80%.10 With 124 seats on the aircraft, an 82% load factor yields approximately 22 excess seats available for purchase one week prior to the flight. (18% * 124 seats). • Number of bidders – For each auction, we assume that there are more bidders than there are available seats. This assumption is later verified in the “Online Web Traffic Model” section. 8 9 Type of aircraft and number of seats were found on AA.com AA’s specific load factor data for 2007 was found on AA.com and the Bureau of Transportation Statistics 10 Breakeven load factor data was found at the Air Transport Association’s website (www.airlines.org) X – Terrestrials Page 10 of 24 Airline Auctions MBA 211 Therefore, we assume that with 22 excess seats available in each flight, we expect at least 23 bidders for each auction. • Uniform distributions –We assumed uniform distributions among the range of bids for each auction. We felt this was a good approximation of how expected bids would be distributed. • Bid ranges – In an English auction, a bidder’s optimal strategy is to bid up to their willingness to pay. Therefore, after considering the market price data outlined in the charts under the “Customer Segmentation” section, we extracted the following bid ranges for each segment: Business Travelers Travel Dates Low Bid High Bid Wed - Fri $750 $938 Table 3: Expected bid price Deal Hunters Wed – Sun $100 $461 IX. Results - sensitivities We analyzed the results of our model by performing sensitivities on a number of variables. For each sensitivity analysis, we kept all other variables outlined in our assumptions constant. The first sensitivity that we conducted was varying the number of bidders and analyzing the expected range of seat prices. The results are graphed below for each customer segment: Figure 1: Business Traveler Bidders v/s Seat Price X – Terrestrials Page 11 of 24 Airline Auctions MBA 211 Figure 2: Deal Hunter Bidders v/s Seat Price For both customer segments, the increase in expected seat price as the number of bidders increases is nonlinear and indicates diminishing returns. The expected seat prices approach the high end of the bid ranges (“high bids”) as the number of bidders increase to just over 100. The prices meet the lower bounds of the bid ranges (“low bids”) where there are just a few more bidders than there are available seats for sale. The second sensitivity was varying the low bids and analyzing the expected range of seat prices, as shown below: Figure 3: Business Travelers WTP v/s Seat Price Figure 4: Deal Hunters WTP v/s Seat Price In this analysis, the high bids were kept constant as they depend on the prices offered in the market and the number of bidders was held constant at 50. This sensitivity indicates a linear relationship between expected seat prices and the low bids. Seat prices are noted to be relatively sensitive to these low bids, X – Terrestrials Page 12 of 24 Airline Auctions MBA 211 ranging from $750 - $900, or a difference of 20%, for business travelers and from $300 - $430, or a difference of 43%, for deal hunters for the given bid ranges. Finally, we conducted a 2-dimensional sensitivity by varying number of bidders & the load factor. The following charts summarize the results: Figure 5: Varying number of bidders and load price for Business Travelers v/s Seat Price Figure 6: Varying number of bidders and load price for Deal Hunters v/s Seat Price In this analysis, we examined a number of load factors for a range of bidders. The low bids increased as the load factor increased as fewer seats are available in the plane thus resulting in a higher willingness to pay. Assuming that the market prices remain the same for each of these load factors, we kept the high bids constant. The results show a diminishing seat price gap among the various load factors as the X – Terrestrials Page 13 of 24 Airline Auctions MBA 211 number of bidders increases. The key takeaway here is that regardless of the load factor or the low bids, a high number of bidders is the key driver to maximize seat prices and obtain the upper bound of the customer’s willingness to pay or the high bid.. X. Online Web Traffic Model A web traffic model was created in order to provide a realistic estimate of the hourly number of bidders online. For the sake of this model we made the following assumptions: 1. Based on comScore Inc. data, the number of unique website users (UUs) in March 2008 on Expedia was approximately 14.83 million. Since ticket buyers cross utilize many website to make an airline ticket purchase decision, we assumed the Expedia UUs as the total online airline ticket seekers. 2. We expect (and recommend) that American Airlines will have an aggressive go to market strategy to promote the Airline Auctions and therefore would attract 30% of Dallas Texas UUs to this site 3. We assumed 35% of UUs to be from the deal hunters customer segment and 65% to be business travelers. Because deal hunters are purchasing tickets seven to fourteen days in advance some of that customer segment has already purchased tickets seven days from departure and therefore a smaller percentage of the UU’s will be from that segment. Traffic Model to estimate bidders per day Total Unique Users 14,830,000 Total Unique Users 494,333 30% Conversation 148,300 Deal Hunters- 35% 51,905 Business Travelers- 65% 96,395 # AA Flights [AJ what was the source?] 1688 Estimated Deal Hunter Bidders 31 Estimated Business Bidders 57 Table 4: Online web traffic model / month /day /day /day /day /day /day X – Terrestrials Page 14 of 24 Airline Auctions MBA 211 Bid Reasonable Range (Figure 1 and 2) Expected based on traffic model Business Travelers 30 – 50 57 Deal Hunters 25- 40 31 Table 5: Expected Bids From the results of the auction model and the traffic study we believe the airlines can make money with an online auction. Based on the expected range of seats available seven days prior to departure and the expected number of bidders, tickets prices for business travelers will range between $750 and $925 with the listed prices on the third party websites being $950. And for deal-hunters we expect the ticket price to range from $225 to $425 with the third party price being approximately $450. This would be generating ticket prices very close to the third party websites. XI. Reality - Why airlines don’t do auctions on their site? There are several reasons why airlines might be apprehensive about conducting their own online auctions. Resistance: Airlines have stayed lean and have always outsourced ticketing to third party sellers. Traditionally they have been a risk averse industry. Undertaking an auction might require a considerable amount of learning curve for airlines hence hesitation to adopt a new model. • Channel Conflict: Third party websites might oppose AA’s direct sales strategy. This might backfire AA’s other ticket sales. • Brand Dilution: Currently cheaper tickets are associated with website such as Priceline and Hotwire. Should tickets sell for cheaper values, American Airlines might be worried about brand dilution which might prohibit them from charging higher prices in the future. • Consumer Behavior: Lastly but most importantly, it would take considerable effort and marketing initiatives for AA to draw estimated users online and actually bid on airline tickets. X – Terrestrials Page 15 of 24 • Airline Auctions MBA 211 XII. Conclusion Airline industry has faced several challenges over the last 10 years. With the significant increase in fuel cost several airline companies have recently filed for bankruptcy. The recommendation of this report is for airlines to hold auctions for excess tickets on their own website. Based on the analysis, revenue can be made by the airlines. As previously mentioned the currently the third party websites, such as Expedia, Orbitz, and Travelocity take a 11 – 25% commission on all airline tickets sold. Therefore, based on the results of this report, if the airlines held the auctions on their site they would be able to achieve similar prices that these websites are receiving but not pay a commission, thus resulting in higher margins for the company. In order for the auction to be successful several mechanisms need to be in place: 1. The auction will focus on excess tickets starting seven days before the departure of the flight. 2. There needs to be a secret reservation price in order to keep the ticket prices high. Also, the reservation price is needed to ensure that if the number of bidders is less than the number of available seats the ticket price will not be artificially low due to low bidders. 3. Tickets will be released on an hourly basis in order to minimize the time between a customer placing a bid and the airline accepting the bid. This is critical for attracting the business travelers. As previously mentioned there are four major reasons why airlines have not already held auctions on their website, resistance, channel conflict, brand dilution, and consumer behavior. In order to overcome the resistance problem, a pilot program should be conducted with a limited number of flights. This will help the airlines climb the learning curve faster and make quick adjustments as needed. Also, airlines currently sell tickets on their own website and run promotions. So the upgrade in technology to conduct an auction would not be a major investment. As for channel conflict currently the third party websites X – Terrestrials Page 16 of 24 Airline Auctions MBA 211 are making a substantial amount of money whereas the airline industry is struggling to breakeven. This auction only focuses on excess tickets and therefore would not affect the ticket sales to Expedia, Orbitz, Travelocity, etc. These websites will still always attract the early bird customer segment and portions of the deal-hunters and business travelers. Because of the secret reservation price, brand dilution is also not an issue. The reservation price will ensure that the tickets will not be sold at a price level that would hurt the airline’s brand. And lastly, consumer behavior can be overcome through targeted marketing. The deal-hunters and business travelers are naturally savvy with third party ticket sites and auction sites. These are the travelers that predominately search (or have their assistants search) all the major websites for the best possible deal. Therefore, once the marketing is in place these customers will not shy-away from the site because of an auction. The ebays of the world have helped to make online auctions a common place in our society. Finally, not only can the airline companies increase their profits by holding online auctions they also can increase their information gathering. By understanding which flights have higher excess capacity and which flights see higher last minute ticket purchases the airlines can better allocate their resources. For example they would be able to combine certain flights based on accurate “live” information instead of just canceling flights because they seem empty. Possibly that flight would have been full three days in advance due to the business travelers and the airlines would have known from past auctions. In conclusion, the current economic times and rising fuel costs represent an excellent time to rethink one’s business model. One of the major airline companies, still in business, should immediately launch a pilot program testing the validity of online auctions and gain the first-mover advantage. * * * X – Terrestrials Page 17 of 24 Airline Auctions MBA 211 Appendix X – Terrestrials Page 18 of 24 Airline Auctions MBA 211 X – Terrestrials Page 19 of 24 Airline Auctions MBA 211 X – Terrestrials Page 20 of 24 Airline Auctions MBA 211 X – Terrestrials Page 21 of 24 Airline Auctions MBA 211 X – Terrestrials Page 22 of 24 Airline Auctions MBA 211 X – Terrestrials Page 23 of 24 Airline Auctions MBA 211 X – Terrestrials Page 24 of 24