Air Transport Industry is one of the best prototypes of the future world where all human activities will be integrated including administrations, companies and contractors. In that respect Air Transport Industry gives an idea of the world for which PageBox has been designed.

This document does not explain how to travel at a bargain price. However it also gives background information that can help you making clever choices.

To provide its service Air Transport Industry must combine the effort of:

• people who sell the service, for instance Travel Agents;

• people who operate aircrafts, airlines;

• people who operate airports;

• traffic controllers;

• customs;

• police;

• ...

Because this service is provided across countries, it also requires the cooperation of governments. Aircrafts are one of the most blatant applications of military research to civil needs. A company that knows how to make civil aircrafts also knows how to make military aircrafts. A civil aircraft can be used to carry troops. A pilot can pilot a military aircraft. For these reasons Air Transport activities have been subsidized, protected and heavily regulated.

Aircrafts are one century old. In the first half century progress has been extremely fast with frequent breakthroughs. In the second half century progress has been much smaller. Aircraft speed did not increase; the most significant changes, wide body and turbo fans took place at the same time as the legacy organization described below. On the other hand this is only then that Air Transport changed people life.

Like computers and Internet now, aircrafts used to incarnate modernity and to trigger the same kind of passion. Air Transport did not have to be profitable. Air Transport early became a global industry requiring the cooperation of independent organizations, private companies of different size and civil servants. These organizations communicated in a B2B model well before the 2000’. Another aspect is the weight of the legacy. With so many involved countries with so different resources, interests and timetables the big bang is never an option. The Air transport system must remain compatible with 20-years old equipments. As a consequence the Air Transport system is increasingly complex and maybe the main contribution of the computers has been to enable this growing complexity.

An optimal Air Transport system must combine conflicting interests like safety, low fares and noise acceptable for Airport neighbours. The involved organizations are different lifecycles. You can create a Travel Agency in a couple of days and an Airline in a couple of months but you must plan a new Airport twenty years in advance. Such a system needs regulations and the problem is to identify which regulations are needed and which regulations actually protect special interests.

We chose to start with the Air Industry organization put in place in the 70’s and the 80’s.

This organization has roots in the Convention relating to the Regulation of Aerial Navigation signed in Paris, October 13, 1929 and in the Convention for the Unification of Certain Rules Relating to International Transportation by Air signed at Warsaw, October 12, 1929. For an update of the Warsaw convention, just read a passenger ticket.

A regulation authority was put in place, the IATA that stood for International Air Traffic Association between 1919 and WWII and stands for International Air Transport Association since 1945. See the IATA history pages for the whole story. IATA has two functions:

1. trade Association (technical, legal, financial, traffic services and most agency matters);

2. tariff Coordination (passenger fares and cargo rates, agents' commissions).

There are three things to keep in mind:

1. The scope of IATA encompasses all Air Transport activities.

2. IATA only controls International flights. However because they must comply with the IATA processes the involved parties tend to apply the same processes to domestic flights.

3. http://www.iata.org is definitely worth a visit. However do note expect to find anything for free. Nowadays much of the IATA's revenues come from selling its products and services to Member airlines, to other airlines and to other companies involved in the travel, transport and tourism industry.

Some companies were created. The most important in an historical perspective are SITA, OAG and ATPCO.

Societe Internationale des communications aeronautiques (SITA) handles shared communication means, including network and devices. For instance when you check-in or board the PC and the device used to print and process the boarding pass does not belong to the Airline. For the next departure from the same gate another airline will reuse those PC and device. Such shared equipments are often installed and maintained by SITA.

Computer Reservation systems, later called Global Distribution Systems (GDS), were created by Airlines to allow Travel Agents to make online bookings. Then four GDSs emerged from a consolidation phase, which account for 85% of CRS terminals, Sabre created by American, Worldspan created among others by Delta and Northwest, Galileo and Amadeus created by European airlines.

These super GDSs are huge transactional systems that process around 5000 requests per second from 40000 to 50000 terminals. They operate mainframe clusters running Operating Systems from IBM (TPF) and Unisys (OS/2200). They initially operated only hierarchical networks with protocols like X25 and character datastreams. Now they also support TCP and Internet access. GDSs allow making booking on airlines that accept to pay a booking fee. GDSs are more and more independent of their founding companies.

OAG stands for Official Airline Guide. This book is published by the homonymous company and contains the airline schedule information, so for every flight essentially:

• the departure and destination airports,

• the departure and arrival times,

• the flight number.

Today the OAG maintains an airline schedules database, which holds flight details for 1000 airlines and more than 3000 airports and is updated around ten times a second.

Airline Tariff Publishing Company (ATPCO) is a fare distributing company. This paper presents its work.

On this diagram we see that airlines publish their fares as well as application rules. A rule can specify that a fare only applies if the traveler stays on the destination on Saturday night. Rules are written in a format that can be processed by computers.

For an entertaining presentation of the Air Transport regulations and Authorities play the Journey Flash file on UK Civil Aviation Authority site.

The Paris convention has been superseded by the Chicago Convention signed on December 7th 1944. You can find a printable excerpt that contains the most interesting articles here. Here is the most important:

Governments bargain with each other for (3), (4) and (5) to protect the market shares of their national airlines. There are three other freedoms that were not defined by Chicago convention.

The 6th freedom is the freedom to carry traffic between two foreign states via the state in which the airline is registered. Depending on its location a country has or does not have good opportunities to carry 6th freedom traffic. For instance European countries have good opportunities and Australia virtually none.

The 7th freedom is the freedom to operate a service between two foreign states. The 8th freedom is the freedom to operate domestic flights in foreign countries. It is also called cabotage and is almost never granted. Europe is a special case. EU airlines were granted the 5th and 7th freedom on international routes throughout the Union in 1973 and 8th freedom in 1997. The European Commission would like to negotiate bilateral agreements between EU and foreign countries.

We recommend reading How an airline can find efficiency with Unix of Paul Murphy. This paper describes the IT business of airlines with its combination of simple issues like booking and complex problems like resource scheduling.

Resource scheduling consists in defining when the aircrafts must take off to go where with the following constraints:

1. The aircraft must flight.

2. The aircraft must take off at time convenient for customers.

3. The aircraft and crew must be available at the right place. A crew in Bangkok cannot help if a flight is needed from London. This problem is especially hard to solve. To the opposite of machines people need to sleep and recover.

Things never go as planed because machines never stop breaking down, Air transport is still very sensitive to weather conditions, etc... The system involves third parties such as Aviation Authorities. For instance aircrafts queue before taking off thank to ground-hold strategies because the aircrafts would not be allowed to land when they would arrive if they were allowed to take off.

Paul Murphy’s paper describes a company with a business model close to a Low Cost Carrier (see below.) Traditional airlines are Network air carriers and generally operate a hub-and-spoke network. In this model an important objective of resource scheduling is to maximize the number of useful connections. A connection is useful only if arrivals precede departures in a sufficient time to permit the transfer of baggage from inbound to outbound flights. However if the connection time is too long the schedule no longer attracts the customer. So there is a balance to find.

Network air carriers moved to a hub-and-spoke network during the deregulation at the end of 70’s.

Hub and spokes network was not entirely new because of international transport constraints and because of technical constrains. A long-range aircraft like a Boeing 747 carries more passengers than a short-range aircraft like a Boeing 737. To fill the large aircrafts that were flying between international airports airlines needed to fly aircrafts from smaller airports to those international airports. In the same way to fill the short-range aircrafts in the smaller airports commuters were using even smaller aircrafts.

Air lines are just a special form of network. Because aircrafts differ in range and in capacity this network is partially hierarchical. Because customers want to go from an origin to a destination it is also partially a peer network. A well-known method to reduce the number of connections between nodes is to set up a concentrator:

In the first case we need 10 connections whereas in the second case we only need 4. Obviously it does not come for free: to go from A1 to A4 we need to first go in A5.

A computer network carries many messages per second almost instantly. Aircrafts are much slower and the network throughput is much slower. Furthermore aircrafts cannot board the same gate at the same time. Passengers must go from the arrival (terminal, gate) to the departure (terminal, gate.) Therefore a hub and spokes network works more like blood circulation.

Between T1 and T2 aircrafts fly from the spokes to the hub to feed it.

Between T2 and T3 employees clean the aircrafts, move the luggage, load the food and drinks and maybe fill the tanks. Passengers walk from their arrival gate to their departure gate.

Between T3 and T4 aircrafts fly from the hub to the spokes.

Then the process starts again with other passengers.

You can see potential drawbacks of hubs:

• All aircrafts have to take off and land at the same time.

• The drawback is augmented because usually the first take off happens at 7am and the last landing at 11pm. Furthermore to maximize the number of useful flights (with passengers) aircrafts should park at the spokes, which is bad for the crew that often has to go to the hotel and which complicates the aircraft maintenance.

On the other hand there are many airlines and a hub airport is usually the hub of only one airline and often a spoke of other airlines.

Hub airlines have a strong marketing power on the routes to and from their hub because of the flight frequencies. For business travelers living close to a hub airport the airline that owns the hub is the most convenient. Therefore these business travelers join the owner’s Frequent Flyer Program and become a sort of captive market for the owner that can increase its fares. The only network carriers to make money on a route between two hubs are the hub owners. It is not to say that hubs are bad. Hubs have not reduced the number of direct flights. They have dramatically increased the number of online connections (connections without airline change) and reduced the number of interline connections (connections with an airline change). They allowed increasing flight frequencies.

There are two kinds of passengers:

1. Leisure travelers: Passengers who want to travel and pay their tickets. They agree to plan their travel well in advance to get a better price.

2. Business travelers: Passengers who do not want to travel and do not pay their tickets. They often must travel under a short notice.

Airlines therefore define different fares for the same seat. Normally less restrictive are the rules that apply to a fare higher is the fare. The problem for the airlines is:

1. to fill the aircraft;

2. to sell the seats at the highest price;

They sell sets of tickets to consolidators well in advance. Then they sell discount tickets but they keep seats for business travelers that typically book a couple of days before the flight.

Depending on the fare rules a ticket can be transferable and even refundable. For the business traveler the only inconvenience if he does not show on time is that he is moved on the waiting list of the next flight. Therefore no-shows are frequent. Overbooking consists in selling more tickets than seats because experience shows that there will be a number of no-shows.

Revenue management (a.k.a. Yield management) was developed with a considerable success to automate the overbooking and conditional selling process. Here is roughly how it works:

The revenue management reads the number of seats sold and available from the Inventory and uses these numbers, historic data and a prediction model for instance to compute a bid price. If the customer or the Travel asks for a ticket at a price lower than the bid price a comparator returns “Sorry, we do not have seats at this price on the flight” even if there are still suitable seats. This page explains how Revenue management is working.

Today revenue management systems are operated by Airlines. Therefore it works the same for the travel agent and for the end consumer using a booking site on Internet. Revenue management used to be an advanced area of the Operational Research and Airlines tent to hide their secrets. Things seem to change. Revenue management improvements have a diminishing return on investment. Revenue management could become a commodity. For instance Navitaire partnered with Manugistics to offer a Revenue management in an Application Service Provider model. In that case Revenue management is an adaptation to Air Transport industry of an Operation Research infrastructure also used in supply chain management and planning. Revenue Management is becoming affordable not only to smallest Airlines but also for Hotels, resorts and Car rental.

When it starts being congested an airport introduces departure and arrival slots: an aircraft can take off only if its airline owns a departure slot and land only if its airline owns an arrival slot.

Local scheduling committees exist to coordinate airline requirements at slot-controlled airports. A document of the International Chamber of Commerce (ICC) presents this IATA based system: "This system of slot allocation acknowledges an incumbent airline's grandfather right to a particular slot time at an airport where that slot was used in the previous equivalent season. These grandfather rights continue until an airline ceases to utilize a slot or surrenders it back to the Scheduling Committee coordinator. Slots, which are not grandfathered are allocated by the Scheduling Committee coordinator in accordance with priorities set out in the IATA guidelines in a non-discriminatory manner. That is, the nationality of the airline plays no part in the decision of the coordinator allocating the slot. Slots allocated in this way become grandfathered once they have been used."

The system has worked rather well to date and both airlines and their governments have accepted it as fair and reasonable. However now in some regions (USA, EC, Japan) large airports are more and more congested for the following reasons:

• environmental concerns;

• safety regulations - aircrafts cannot be to close;

• increasing traffic.

In some cases the airport is full, which means that its slot number cannot grow anymore. Look at the Slot Pool page of Airport Coordination Limited. It allows comparing the London airports, Heathrow (congested), Gatwick and Stansted (plenty of slots available).

Local scheduling committees usually give priority to international flights over domestic flights for new slots or slots surrounded by incumbent airlines. The drawback of this system is that it favors the incumbent, non-growing airlines. “Once an airport is operated at maximum capacity, flexibility is lost, new entrants are denied access and incumbents are denied growth.” A new entrant (for instance a Low Cost Carrier) has often no choice but to use non-congested airports or to buy an incumbent airline.

An airline can exchange slots with another airline. Because some slots have more values than others there is a gray market where airlines can pay a fee in case of slot swap or even acquire slots. Proposals were made for the creation of a secondary market in airport slots. You can read the document of the UK Civil Aviation Authority (CAA). A secondary market would at least introduce some transparency.

The comments of the US Department of Justice for the American Airline and British Airways alliances recommend a market-based allocation system that would allow entrants to purchase divested slots. Authors note however that under a market-based allocation system many valuable slots have to be made available.

The US Federal Aviation Administration put a system in place to reallocate slots at the last time using a Collaborative Decision Making (CDM) paradigm. In this system airlines and airspace operators (FAA) share information and collaborate in determining the allocation of the arrival slots. The idea is that if each participating airline knows the needs of other airlines and agrees to trade its slots in real time then the average delay will be smaller. In some cases the system can avoid running a Ground Delay Program (Ground hold aircrafts.) You can find a description of the system here.

Frequent Flyer Programs (FFPs) are a kind of Loyalty Scheme. Loyalty schemes are used a lot in retailing (supermarket chains, bookshops.) By granting a discount after a given amount of purchases a loyalty scheme aims to attract customers to a service or good that they would not buy at its tag price. It is what marketing experts call reducing price elasticity.

FFPs are a special kind of Loyalty Scheme because:

• It is not necessarily easy to spend the granted miles. The number of unused miles represents now a huge amount of money.

• Usually the traveler who is granted miles is a business traveler who did not pay the ticket.

An article of the May 2nd, 2002 issue of the Economist called "One of the world's main currencies is heading for a fall" explains that Frequent Flyers miles is the second-biggest currency, now worth $500 billion. Furthermore airlines sold roughly $10 billion worth of miles to partners, such as credit-card firms.

As a consequence almost 50% of miles are earned without even leaving the ground. In America you can even get miles on your income-tax payments, if you pay by credit card.

"The total number of miles awarded each year by airlines worldwide has doubled over the past five years, but miles redeemed have increased by only one-third. In 2001, more than four times as many miles were earned as were redeemed, and miles do not expire so long as members have earned or used them in the past three years." Economist editors conclude, is that airlines have been printing too much of their currency. Eventually the value will fall. And the clear lesson: "Use your miles while you can."

The article is unfortunately not available for free. If you want to buy it go to http://www.economist.com/.

You can find an interesting comment of this article on Internet in http://www.hotelnewsresource.com/studies/et050302.htm.

Except in USA, airlines tend to award miles only for travel at premium fare (business or full economy) that leisure travel buy only in exceptional circumstances. Therefore FFPs programs primarily target business travelers. Business travelers do not pay their ticket but usually choose the airline. In such conditions they are influenced by their FFP membership in choosing the flights of a particular airline and they choose their FFP in order to fulfill two conditions:

1. frequent and convenient flights to their business flights to earn miles;

2. flights to their leisure destination to use their miles.

Therefore FFPs can be considered as anticompetitive and even as “a barrier to entry for new entrants with a small network.” Governments could hardly prohibit FFPs because other loyalty schemes were allowed before FFPs. They considered taxing the rewards as non-pecuniary income but this tax raises implementation difficulties. It is fair only to tax used miles and it is not easy to track miles granted for business trips. We also believe that governments missed a point in this case like in stock option case: executives are not the only ones to travel. Aircrafts are full of consultants who spend their life far from home and companies are desperately looking for cheap incentives to keep them on the road.

As we have seen airlines with few routes or routes within one particular geographic area cannot attract Frequent Flyers with their program. However they can address the issue by linking their FFPs with those of other carriers: miles earned on a flight of a carrier can be used on a flight of another carrier. FFPs played an important role in the creation of global alliances.

There is at least a site dedicated to frequent flyers problems, http://frequentflier.com. You can find on this site a 3-part series on global airline alliances titled “Global Airline Alliances Dominate Travel Landscape”, “Global Airline Alliances Offer Real Benefits for International Travelers” and “Global Alliances, Mergers & Acquisitions, and Miles.”

Airline alliances can cover many areas:

• Code sharing. Code sharing is an agreement between two airlines under which an airline that operates a flight allows another airline to offer that flight under its own flight code.

• Block spacing. Block spacing consists for an airline in allocating a number of seats to another airline on some of its flights.

• Links between FFPs.

• Joint sales and marketing.

• Joint ventures in catering, ground handling and aircraft maintenance.

• Joint passenger and cargo flights.

• Joint purchasing and insurance.

On one hand you find technical alliances, for instance ventures in catering, ground handling and aircraft maintenance. This sort of tactical alliance is created to reduce costs. For instance in one area a couple of airlines can agree to specialize their maintenance teams. One airline maintains 747 aircrafts and another A330 and A340 aircrafts.

The most famous alliances are strategic alliances created to handle code sharing, FFPs and some marketing programs. They do not only consist in a commercial agreement. Participants agree to adapt their schedules to increase the number of useful connections inside the alliance. These alliances aim to increase the marketing power of their members in order to successfully meet the challenges of globalization. In that respect it is interesting to compare founding members of GDSs with members of alliances created ten or twenty years later. GDS were created to interface travel agent market with airlines. Therefore GDSs were regional. For instance Amadeus was created by Air France, Iberia, Lufthansa and at the beginning SAS. Alliances aim to provide a global offer to international travelers. The trick in alliances is to have a member in each market. Air France is in Skyteam, Iberia in Oneworld and Lufthansa in Star Alliance. It has also to be said that even if airlines wanted to create regional alliances in order to dominate a market governments would not allow it. It is interesting to read the comments of the US Department of Justice for approval and antitrust immunity of OneWorld and Star alliances. This fascinating paper discusses many issues described on this page.

Bermuda II bilateral agreement refers to an agreement following the rules of the Chicago convention. Open Skies is a regime without regulatory constraints on pricing and entry. The US Department of Transport would like to get Open Sky as compensation to allow alliances. Interestingly enough the governments have to give something to airlines because they want to abrogate a bilateral agreement that was also a kind of privilege. However this bilateral agreement does not matter so much for incumbent airlines. First alliances allow airlines to circumvent the bilateral agreements. Second on congested airports grandfathered slots are a safe way to forbid new entrants. Third not only is airside capacity scarce, it is also difficult to obtain aircraft parking space and terminal facilities in some airports. Fourth new entrants are possible only if the economy enables their creation. As we noted above it is very difficult to compete on a route from or to a hub except if you own the hub on the other side. A great section of the DOJ comments starts with “In essence the parties (BA/AA) would have DOT (Department of Transport) believe that the complicated pricing structures and sophisticated yield management systems that the carriers have constructed (at great cost) to allow them to segment demand and discriminate between business and leisure passengers are ineffective.” A bit further authors note “Several characteristics of the airline industry increase the ability of carriers to engage coordinated interaction. More importantly carriers have almost instantaneous knowledge of competitors’ fare changes and the ability to quickly respond to any change.”

Airlines may have even better systems in the future. A patent application assigned to Sabre, illustrates what can be the future of airfares. You can find this application on http://www.uspto.gov. This application describes an improvement of power pricing or Low Fare Search (LFS). When you use an online booking system you basically enter the itinerary origin and destination and the departure and return dates. The LFS returns a list of fare availabilities sorted by growing price or by convenience (time between departure and arrival.) By the way the list illustrates a questionable feature of code sharing: it allows screen padding. You often need to scroll or go to a next page because the same flight is displayed more than once with different codes.

To return the list the LFS queries a large number of airlines for their fares and availabilities, the number of queries being somewhat reduced by sophisticated caches. The airlines’ Revenue managements return fares whose bucket is not empty. Because airlines’ Revenue managements do not talk to each other, LFS acts a bit like a reverse auction system. The customer expresses her intent to buy a flight seat and the airline whose fare is the lowest wins the bid. The Sabre’s invention aims to give a second chance to airlines that are member of the system. Before returning the availability list to the customer the Sabre LFS queries member systems. The member system that may have to be hosted on the LFS system for performance reasons receives the wining bid price. If its airline returned unavailable or an overpriced fare the system determines whether to make seats available at a price deemed to be competitive. Conversely the system can raise the fare if the airline fare was much lower than any of its competitors.

The Sabre’s invention has a couple of weaknesses. First this is us who add this concept of member airline. In the examples there is only one member, American Airline and one Online Booking system, Travelocity. However other airlines would not easily accept that sort of unfair competition except if they are also members. Second the member system must have access to the airline inventory to make sensible decisions, which is a problem if the member system has to be hosted in the LFS environment. However this invention might be important in two respects:

1. Dynamic pricing.

2. In fine re-pricing could become a sort of Revenue Management and impact conventional Revenue management.

If you want to know more about auction systems you can read the documents at http://www.agorics.com/Library/auctions.html. The conventional LFS looks like First price, sealed bid auctions. The Sabre’s invention unseals the participants’ bids.

There is another approach of interest, the Priceline "name your own price" apparatus. The traveler enters the origin and destination airports, the departure and arrival date, the number of passengers, her credit card data and the price she is ready to pay for the flight. Then two things can happen:

1. An airline is interested in selling a seat at this price. Then she is booked and charged.

2. No airline is interested in selling a seat at this price. Then she is not booked and not charged.

Through her credit card the traveler expresses a commitment to buy at the named price. However "name your own price" is not a reverse auction system. In a reverse auction system (widely used in procurement) buyer inform that they need a given good or service. Sellers make bids and the sale is awarded to the seller who makes the lowest bid. Priceline secured "name your own price" with a dozen of patents, including 6,510,418 that aims to prevent buyers - or intermediates - to use the apparatus to determine the market price.

Before closing the discussion about the DOJ comments we must mention two points of interest.

It is true that for a while carriers have almost instantaneous knowledge of competitors’ fare changes and the ability to quickly respond to any change. This fact had a considerable influence on the 80’s deregulation. Deregulation promoters believed that the theory of contestable markets could apply to the Airline industry. According to this theory, firms in oligopolistic industries will price at the same level as they would in more competitive industries, provided that a threat of competition exists. A threat is credible if:

1. There is no barrier to the entry of new firms to the market. For instance an entrant does not have to pay something that incumbents do not have to pay.

2. A firm must be able to go on the market, make profit for a short period of time and go out of the market without losing a lot of money.

3. The time it takes to incumbents to change their price is longer than the time the new firm needs to make its entry profitable.

Actually before the deregulation there were few fares that were not changing often just because there was no need. But because of computer improvements incumbents were able to change their fares extremely quickly when firms like People Express entered the market.

The DOJ comments also observe that today global “alliances do not entail comprehensive (or any) revenue pooling (...) The operating carrier gets almost all the revenue from a flight with the partner getting a commission for any ticket sold under their code.” In the same way alliance members do not pool or exchange slots.

The three main global alliances are:

One alliance, Qualiflyer founded by Swiss Air died on December 31, 2002.

The first alliance, Wings, founded by Northwest and KLM no longer exists though its core companies, Northwest, Continental and KLM still cooperate. See the Worldperks page for more information.

Global alliances will continue to change both in composition (it is easy for an airline to change of alliance) and responsibilities (for instance Star Alliance has an IT hub.)

With the advent of Internet three types of online booking sites were created:

1. Airline booking engines

2. Low cost carrier booking engines

3. Multiple carriers booking engines

Online booking is used to shortcut a part of the traditional distribution system:

Because an Airline online booking site is just a HTTP interface to the existing Airline system it is easy to develop. For the Airline the benefit is that it does not need anymore to pay a booking fee to the GDS and a commission to the Travel Agent.

Airline sites were not very successful:

• Airlines cannot however offer better prices than Travel Agents because it would be an unfair competition.

• For the same reason airlines may not want to aggressively promote their sites.

• Customers do not perceive airlines as able to provide the best fares.

• Customers like comparing prices.

Low Cost Carriers (LCCs) use a different business model:

• They target end consumers, not Travel agents.

• They do not use the GDS distribution channel and they refuse to pay a booking fee.

• They often do not even use tickets. Passengers making a booking are given a confirmation number, which is enough for baggage handling and flight check-in. At check-in passengers are given a reusable boarding pass.

• Their booking system is similar to hotel and car hiring systems and simpler than systems presented above.

Because they spare on wages, catering but also on distribution costs low cost carriers can offer lower fares than traditional airlines. Thank to their marketing effort customers know their URLs. Therefore their sites are quite successful.

Because LCCs sell direct flights:

• They have no concern about connecting flights, which simplifies their resource scheduling

• They can avoid large airport (typically the hub of a large airline) because they are congested and expensive. A shuttle to the city is often cheaper than the airport taxes. However in Europe at least this is often inconvenient for the traveler: secondary airports being farer from the cities and larger airports having better public transportation systems.

LCCs operate almost only domestic flights (partially because of the government agreements on International flights), which also means that they are less affected by political events.

LCCs also operate less airports and city pairs than traditional airlines. In 2003:

For instance EasyJet sold flights on 100 city pairs when a GDS can propose 500,000 city pairs.

LCCs do not try to replace traditional airlines; however because they concentrate on high traffic links LCCs hurt the interest of even large Airlines. They carry more passenger than many Network air carriers. United faces competition from low-fare carriers on about 70 percent of its routes, including Southwest Airlines Inc. in many locations and Frontier Airlines at its Denver hub.

Therefore large U.S. network air carriers consider starting low-fare units to compete with low-cost carriers, with two ideas:

1. Target price-sensitive leisure travelers.

2. Integrate their low-cost unit into their hub-and-spoke network.

Large carriers might be wrong and their low-cost units can actually cut their revenues on the most profitable customer segment, the business travelers:

• Business travelers look for convenience and are more and more price sensitive.

• In some cases LCC schedules and even airports are more convenient.

• Low cost carriers do not set complex rules and penalties on one-way flights.

As a consequence a significant number of LCC passengers are actually business travelers.

Some European airlines made the experience:

1. British Airways created Go in 1998 and sold it in 2002 to Easyjet notably because Go has competed with its parent on several European routes.

2. KLM founded Buzz in 1999 and sold it in 2003 to Ryan Air. The press release said that "KLM's decision to sell Buzz follows a strategic review of this business in light of the increasing competition in the European low cost arena over the last few months."

As a matter of fact traditional airlines often create low cost subsidiaries with old aircrafts and employees they do not need anymore. Then LCCs happily buy these subsidiaries once the Airline gives up for the slots. They keep only few employees and do not even want the aircrafts.

Large carriers have higher structural costs. For instance LCCs see IT as a commodity and often do not even develop or operate their reservation system. Many LCCs use the Navitaire’s Open Skies Reservation system. Some others use AirKiosk from Sutra. There is a good presentation of AirKiosk here. Because LCC operate fewer routes with almost no rules and fewer fares, a simpler Information system satisfies their needs. However their system still includes a sort of revenue management. LCCs sell the first few seats at a special promotional rate, often very low, but thereafter prices can increase substantially.

In this study we consider the following carriers:

The number of city pairs operated does not represent the size of a company. The number of flights and the type of aircraft are better metrics. Note that Sky Europe increased the number of city pairs with a bus shuttle, which makes sense in its area – Vienna is 40km distant of Bratislava. However we believe that this short study is useful because it is based only on publicly available and checkable information.

We consider the (origin, destination, carrier) triplets. These LCCs operate 1177 triplets and operate flights from 166 airports. The most popular airports are:

London, Stockholm and Paris have more than one airport with LCC flights. For historical reasons Germany and Italy are organized in regions with somewhat smaller cities. In these regions you may also choose between different airports.

The aggregate offer of Low Cost Carriers is quite significant. However to the opposite of Network Carriers, LCCs do not create alliances. We wrote an example of PageBox application, EuroLCC to study the connection opportunities offered by Low Cost Carriers.

A first generation of multiple carriers booking sites acts as big Travel Agents (Travelocity claims to be the 6th biggest TA and Expedia to be the 7th Travel Agent.)

Expedia used to be a wholly owned subsidiary of Microsoft up to its IPO in 1999. Then Expedia was acquired by USA Network (now USA Interactive) in 2001.

Travelocity is wholly owned by Sabre, the biggest GDS in the World. Travelocity competes with other travel agents that use Sabre products.

Orbitz was founded by five major airlines, American, Continental, Delta, Northwest and United. Orbitz has drawn wide-ranging criticism from Expedia, Travelocity as well as outfits like the American Association of Travel Agents with the claim that the airline-backed ticketing operation is antithetical to competition in the industry and hurts consumers. Though a Federal audit has cleared it, Orbitz chose to thoroughly explain its position. We recommend reading documents such as “Economics of Travel Distribution in Internet-Driven Environment” and “Revenge of the Disintermediated”. “Sabre/Travelocity, owned by the world's largest CRS, and Microsoft/Expedia have already secured nearly 70 percent of all on-line bookings by third-party travel sites through a variety of exclusive deals and mergers and acquisitions of smaller Web sites. The duopoly also has nailed down long-term deals with web portals, including AOL, Microsoft Network and Yahoo giving them exclusive distribution rights on portals used by 90 percent of Internet consumers” said G. Katz, Orbitz' chairman, president and chief executive.

Opodo is a European site using the same approach as Orbitz. Opodo has been created by Aer Lingus, Air France, Alitalia, Austrian Airlines, British Airways, Finnair, Iberia, KLM and Lufthansa. There were however two differences: rather than using a third party (ITA) engine, Opodo selected a GDS (Amadeus) engine and, in summer 2004, Amadeus acquired a 55% controlling interest of Opodo.

Katz made a point. An online booking site is successful only if end consumers know that it exists. There are only three ways to achieve this result: deals with big sites like Yahoo, acquisition of successful booking sites and advertising. The technical part of an online booking system is the cheapest part.

Furthermore the main part of a booking site is a low fare search display. The user enters the trip origin and destination, the departure and return date. The system must return 200 and more recommendations for flights with available seats at a given fare. The display should be sort-able per price and convenience (direct flights first.) The trend is to use the user address rather than the airport location and to consider alternate airports. Low fare search is resource consuming. It has to consider the different routes and to exercise the Revenue management of the considered airlines. Minimizing the processing cost requires advanced caching mechanisms. Therefore few companies (and among them GDSs) have the expertise to develop a low fare search.

We have seen above that this evolution endangers two species, GDSs and Travel Agents.

We need to first come back on the Orbitz comment. An interesting aspect is that Orbitz founders used to also be GDSs (Sabre and Worldspan) founders. They rightfully chose to create GDSs outside their core organization because this was not the same business:

• Airlines operate aircrafts and GDSs operate computers

• Airline industry is characterized by a relatively low fixed-to-variable cost ratio whereas GDSs have almost only fixed costs: booking fee = fixed cost + profit / number of bookings

• A GDS implies long-term investments and commitments

GDSs act both as Application Service Providers (ASPs) and as software companies that develop software for Travel Agents, airlines, alliances and third-party Web sites. This made sense when GDS were created. In a mainframe environment developing and operating programs in the same organization spares packaging and facilitate troubleshooting. The organization can be more reactive. Because development and test platforms are expensive to set up and operate the organization spares money when teams that operate the production environment also operate the development and test environments. Now GDSs also develop Windows applications, browser-based applications, Unix applications and Web applications. These applications require different technical, marketing and management skills and can hardly coexist in the same organization. Therefore GDSs tend to put Operations in Facility Management and to set up subsidiaries for new activities.

Another document from AirKiosk describes nothing less than the death of GDSs by booking engines. The author explains that online booking increases dramatically the system resource use and communication costs of GDSs and Airlines. As a matter of fact online booking increases the number of availability and power pricing/LFS requests per booking. The reason is simple: a travel agent is paid to book; to be productive she must quickly find a fare good enough for the customer. The customer of an online booking engine pays to book. She wants to be sure to get the cheapest and most convenient flight. However GDSs can address this issue in two ways:

GDSs have strong points:

• Arguably the best know-how of the Air Transport industry outside Airlines. They can operate the Airline Information systems more efficiently than Airlines because this is closer to their core business. They can also develop the sort of big software that Airlines, Alliances and Online Booking sites need.

• GDSs are profitable. They can acquire competences and market shares that they need to achieve their strategic objectives.

• GDS are behind multiple-carrier booking engines: Expedia and Orbitz use Wordspan, Travelocity Sabre and Opodo Amadeus.

Though very few is public today about the functions they will provide we can presume that GNSs will offer:

• Web site engines to power online travel sites.

• A power pricing / Low Fare Search (ITA already has a good one).

• An API, presumably implemented as a Web service. G2 SwitchWorks has already published its XML schemas.

• Somewhat simpler tools for travel agents.

GNEs plan to charge much lower booking fees than traditional GDSs.

Travel Agents

At the end of the 90’ it was commonly agreed that Travel Agents were declining and had to merge or die.

One after one US Airlines stopped paying commissions. Furthermore Travel agents had to face the competition of Online Booking sites. Travel Agents were also criticized for their incompetence.

However Travel Agents can survive and for a large part thank to Internet. On one hand a rational and experienced traveler will not use a Travel agent to buy a flight to a common destination. On the other hand a leisure travel may also want to buy a Cruise or a Tour and a business traveler typically needs other services like hotel booking and car rental. Commissions on hotels, Cruise lines, Tour Operators are still substantial. Travel Agents often lack training and experience but Travel Agent portals or Webtops can improve the situation. Such portals are easy to develop, use and support and they provides means:

• To list and compare the fares of low cost carriers, consolidators and airlines

• To book hotels, Cruises and Tours

With portals the Travel agent can:

• Provide a low airfare, which is important because the customer can easily compare this fare with fares available on Internet

• Provide a better service. GDSs are an effective intermediate to companies with sophisticated IT systems like the Airlines and Car rental companies but not for smaller companies like Tour Operators and hotels.

To summarize Travel Agents use Internet distribution channels as well as GDSs to serve their customers.

The revenues generated by package sales are substantial enough to attract multiple-carriers booking sites. These sites created dynamic packaging systems to address leisure traveler needs and corporate traveler products to address business traveler needs.

Dynamic packaging

Expedia was arguably the first company to implement a dynamic packaging. Originally (when it was owned by Microsoft) Expedia did not try to protect dynamic packaging with patents, which means that core functions of dynamic packaging are relatively safe from an Intellectual Property point of view. You can see the dynamic packaging of Expedia on the Vacation packages tab of their site. On this page end consumers may select a package including flight, hotel and car or flight and hotel or hotel and car. Packages may also include tours, cruises, rail travels and add-ons such as insurance.

GDSs support hotels and car rental companies for a while. They also support rail companies. For car rental and rail companies they use the same kind of connections and protocols as with airlines. With hotels things are more complicated. Hotels usually have their inventory hosted on a Hotel Reservation System (HRS) or Hotel Central Reservation System (CRS.) HRSs / CRSs interface with Property Management Systems (PMSs) dealing with various aspects of hotel management like accounting and reception, online reservation systems, and frequently with loyalty program and revenue management systems. HRSs / CRSs may also interface with GDS though, most of the times, they connect to switches like Pegasus and WizCom, themselves connected to all GDSs.

Expedia innovated in another way with the merchant model. They wrote: "We have a diversified business model: the agency model and the merchant model. Under the agency model, we act as an agent in the transaction, passing a customer’s reservation to the travel supplier (airline, hotel, car rental company or destination service provider), and receive a commission from the travel supplier for our services as an agent. Under the merchant model, we receive inventory (hotel rooms, airline seats, car rentals, destination services) from suppliers and then process the transactions as the merchant of record in the transaction. Acting as a merchant enables us to achieve a higher level of gross profit per transaction than in the agency model and provides better prices to customers than in agency transactions."

The key in the merchant model is to negotiate satisfactory agreements with providers. It seems that Expedia, which is the biggest multiple-carriers booking site, is one of the few companies to make money with the merchant model (half its revenue in 2002.) The reason is that big providers, which have sophisticated IT systems able to optimize prices and commissions, prefer the travel agent model, the merchant model being attractive primarily for smaller companies without complex systems like revenue management that want to sell online.

Corporate travel

In an article entitled "Accelerating e-Travel" published in December 2001 Demis Barlas wrote "U.S. companies spend approximately $185 billion a year on business travel. That is enough to make travel spending the second-largest controllable business expense (after salaries)."

Corporate Travel products were invented to control this expense at the end of eighties. The first generation of Corporate Travel products was running on mainframes and using expert systems to minimize expenses. The core components of Corporate Travel products were already present:

1. a traveler profile that contained employee data;

2. a travel policy which determined which flights and hotels an employee could select.

This first generation of products did not consider indirect travel expense like the employee recovery the day after an inconvenient travel and forced "a frequent traveler to sit through a two hour connection or stay in a less than mid-line hotel off the beaten path" driving "the typical frequent traveler to come up with creative ways to circumvent the travel policy in order to meet their own personal requirements." The second generation of corporate travel products was using a client / server model. Instead of displaying only cheapest flights, they sorted flights, policy-compliant flights first, flights requiring an approval next. They included an approval workflow.

The third generation of corporate travel products was Web-based. An example of such products is GetThere DirectCorporate. Because such products were Web-based:

1. they did not require to be installed and configured in the customer shop and could address the needs of smaller companies;

2. multiple-carriers booking sites such as Expedia could enter the corporate travel market.

Corporations frequently negotiate agreements with airlines. The most sophisticated corporate travel products are able to bias the display to fulfill the negotiated contracts and even to calculate optimum contract terms and conditions.

Conclusion

Air Transport industry is quite entertaining. Maybe someone could develop a video game where we could choose the government, Airport Authority, Airline, GDS, Travel Agent or Low cost carrier role...

Annexes

Air Transport industry has its jargon, which is presented in this document.

Consolidators

Consolidators are airline wholesalers. You can find presentations of Air Consolidators here and here.

Farebase manages a Consolidators’ airfare database that also includes charter flights and late holiday information. Farebase has been bought by TelMe.com and now provides online booking through its Farebase Travel Portal. Patheo has a similar offer.

For a list of consolidators you can check the Association of Special Fares Agents (ASFA) site. Another solution is to make a query on a search engine with the keywords “consolidator fare distribution.” Note that many portals share the same engines.

Discount Airfares is a good source of information. You can use it for instance to find airport phone numbers

Airport codes

This paper explains the Airport Identifier codes and here is one of the biggest Airport lists available on the Web. A reader suggested us another site, www.world-airport-codes.com, which provides for almost every airport in the world:

• The airport code

• The airport name

• Which country the airport is located in

• The ISO 3166 abbreviation code for each country

• World area code

• The GMT offset

• The runway length

• The runway elevation

• Latitude and Longitude data