Revisiting the Relationship between Competition and Price Discrimination in the Airline Industry∗ Ambarish Chandraa,b

Mara Ledermana

May 4, 2015 a b

: University of Toronto, Rotman School of Management : University of Toronto at Scarborough, Department of Management Abstract We revisit the relationship between competition and price discrimination in the airline industry. Building on early theoretical work, we develop a simple model showing that, if consumers differ in terms of both their underlying willingness-to-pay to travel and their degree of brand loyalty, competition may increase fare differences between some types of travellers while decreasing them between others. Using a novel source of data on airfares that distinguishes tickets sold to different types of travellers, we show that competition increases cross-cabin fare differentials but decreases fare differentials between most Discount Coach passengers, thus reconciling conflicting findings from earlier research.

1

Introduction

Understanding the conditions under which firms can practise price discrimination has been a central question in Industrial Organization. It is now generally accepted that firms can price discriminate in both monopoly and oligopoly settings. Yet, the precise relationship between market structure and price discrimination is less straightforward. As Stole (2007) shows, while competition will lower the prices charged to all consumers, the impact of competition on the price differential across consumers is theoretically ambiguous. In particular, if price discrimination is based on differences in consumers’ underlying value of a good, competition will drive prices towards ∗

Corresponding author: [email protected]. This research was supported by the Social Sciences and Humanities Research Council. Zhe Yuan provided excellent research assistance. We thank Heski Bar-Isaac, Ken Corts, Joshua Gans, Eugene Orlov, Christian Ruzzier and seminar participants at the Kellogg School of Management, the University of Toronto, and the 2015 IIOC for helpful comments.

marginal cost and reduce the differential. On the other hand, if consumers also differ in their degree of brand loyalty, competition can increase price differentials. Empirically, a number of papers have studied the relationship between market structure and price discrimination, most in the context of the airline industry.1 Early work in this area by Borenstein and Rose (1994) measured the relationship between market structure and the overall amount of price dispersion on a route. Their analysis found greater price dispersion on more competitive routes, leading them to conclude that, in this setting, price discrimination was based on differences in the degree of travellers’ brand loyalty. More recently, Gerardi and Shapiro (2009) investigate the impact of competition on fare dispersion and find the opposite relationship. Using a longer sample and a more convincing empirical strategy, they find that competition reduces dispersion, specifically by decreasing fares at the top of the distribution by more than fares at the bottom. Gerardi and Shapiro do not explicitly discuss what this implies about customer heterogeneity; however, their findings are consistent with price discrimination based on differences in travellers’ underlying willingness to pay. In this paper, we revisit the relationship between market structure and price discrimination in the airline industry. Theoretically, we consider the possibility that travellers differ in terms of both their underlying value of a trip and their degree of airline loyalty and, furthermore, that travellers who are similar on one dimension may differ on the other. Using a simple extension of the set-up in Stole (2007), we show that this can give rise to more than two types of travellers, in contrast to what is often assumed in the literature. Moreover, we show that, since traveller types may differ in terms of their underlying value of travel, their degree of brand loyalty or both, competition may increase the price differential between some types while simultaneously decreasing it between others. Empirically, we investigate the relationship between market structure and price differentials using a new dataset and novel empirical setting—the Canadian airline industry. We measure the impact of increased competition on the prices of different classes of tickets as well as the prices of tickets at different parts of the fare distribution. We find that competition has little impact on very cheap or very expensive tickets but has a large impact on tickets in the middle of the fare distribution. This results in competition reducing the price differential between some travellers while increasing the differential between others. Thus, our findings encompass the results in both the Borenstein and Rose and the Gerardi and Shapiro studies and indicate that both effects can operate at the same time. Borenstein (1985) points out that, when products are differentiated, price discrimination can be based on differences in consumers underlying value of a product or differences in the strength of their brand preferences. Holmes (1989) develops a model of third-degree price discrimination under duopoly. He shows that a firm’s 1

The reason much of this work has been carried out in the context of the airline industry is likely due to its wealth of available data, frequent changes in market structure and heterogeneity in types of travellers.

2

price elasticity of demand in a market can be expressed as the sum of the industrydemand elasticity and the cross-price elasticity and that, with more than one firm, price discrimination can be based on differences in both industry and cross-price elasticities. In his review article, Stole (2007) highlights the important role played by cross-elasticities of demand in determining equilibrium prices and output under oligopoly price discrimination. He explicitly shows that the relationship between competition and the price differential between consumers is theoretically ambiguous and will depend on whether consumers have similar or different cross-elasticities of demand. We use the set-up in Holmes (1989) and Stole (2007) to generate predictions about how competition will affect fare differences among airline passengers. The literature on airline pricing has long recognized that travellers are heterogeneous in their willingness-to-pay and has typically distinguished between business and leisure travellers. The source(s) of heterogeneity will determine whether competition increases or decreases the price differential between traveller types. If business and leisure travellers have a different underlying value of travel but similar levels of brand loyalty, competition will reduce the price differential between them, as was found by Gerardi and Shapiro (2009). On the other hand, if business travellers have a higher value of travel and are also more brand loyal, competition may increase the price differential between them, as was found by Borenstein and Rose (1994). In addition to these scenarios, we consider the possibility that business travellers have a higher value of travel than leisure travellers but are themselves heterogeneous in their degree of airline loyalty—for example, due to differences in corporate travel policies. We introduce an intermediate type of traveller whom we refer to as a “brand indifferent business traveller”. Using a simple three-type model, we show that competition may have the largest impact on the fares of the intermediate type since these are the consumers who will be charged high prices by a monopolist but whose price will move towards marginal cost with competition. We further show that this implies that competition will reduce the fare differential between some travellers while increasing it between others. More generally, our model suggests that, if airlines segment travellers along both dimensions, competition will increase the price differential between travellers who have different levels of airline loyalty but reduce the price differential between travellers whose only source of heterogeneity is their underlying willingness-to-pay for a trip. Empirically, we test this prediction using data on the Canadian airline industry. The Canadian industry has several features that make it well-suited for an empirical analysis of the relationship between market structure and price discrimination. First, Canada has had only a single full-service price-discriminating airline—Air Canada— operating throughout our sample period. Second, market structure is straightforward to measure in Canada because there is little connecting service, no domestic codesharing, limited use of regional partners and only one multi-airport city. Third, the small number of carriers operating in the domestic Canadian market means that the 3

changes in market structure that we observe map more closely to the simple comparison between monopoly and duopoly which is the basis of much of the theoretical work in this area. Our analysis uses data from the Airport Data Intelligence (ADI) database produced by Sabre Airline Solutions.2 The ADI database provides monthly fare and booking information for most itineraries worldwide. An important advantage of the ADI database is that it disaggregates tickets by cabin class as well as fare class. This allows us to investigate how competition affects the fares paid for different types of tickets as well as tickets at different points of the fare distribution.3 Importantly, this allows us to distinguish probable business travelers based on information other than just their ticket’s location in the overall fare distribution, which is a drawback of studies that use publicly available data for the U.S. market. Our empirical analysis proceeds in several stages. We start by estimating the impact of route-level market structure on Air Canada’s fares for tickets in different cabin classes (Discount Coach, Coach, and Business). Given that about 70% of tickets are classified as Discount Coach and that there is considerable price variation across Discount Coach tickets, we then investigate how market structure impacts different classes of tickets within Discount Coach. Finally, we estimate the impact of competition on different percentiles of the Discount Coach fare distribution. All of our regressions include route, year and month fixed effects and therefore capture how Air Canada adjusts its fares for a given type of ticket, as the degree of competition on a route changes over time. A clear pattern of results emerges from our empirical analyses. We find that competition has little impact on Air Canada’s very cheap or very expensive tickets but a significant impact on tickets in between. In particular, when we compare the impact of competition across cabin classes, we find that having an additional competitor on a route has no impact on Air Canada’s Coach or Business fares but reduces its average Discount Coach fare by approximately 7%. When we explore the impact of competition on different types of Discount Coach tickets, we find that an additional competitor has a larger impact on more expensive Discount Coach tickets than less expensive tickets (about 8% as compared to 4%). Finally, when we estimate the impact of competition on the percentiles of the Discount Coach fare distribution, we uncover a U-shaped relationship between competition and fare reductions. Competition has the largest impact on fares between the 15th and 75th percentiles of the distribution and a smaller impact on fares below and above these percentiles. 2

The Canadian government does not collect and disseminate detailed data on airfares in the way that the U.S. government does through the Department of Transportations Databank 1B, which is a random 10% sample of domestic tickets. 3 Recently, other studies have also employed airline data with information on ticket characteristics, although the source and setting is different from ours; see Hernandez and Wiggins (2014) and Sengupta and Wiggins (2014).

4

The pattern of results we estimate suggests that competition acts to reduce the fare differential between some types of travellers but increase the differential between others. In particular, the fare differential between Business Class/Coach tickets and Discount Coach tickets rises while the differential between different Discount Coach tickets falls. While it is not possible for us to perfectly map different types of tickets to different types of consumers, our findings suggest that price discrimination across passengers who purchase differentially priced Discount Coach tickets is likely based on differences in their underlying value of travel while price discrimination across passengers purchasing tickets in different cabin classes is based on differences in both their underlying value of travel and their degree of brand loyalty. This work contributes to the literature on price discrimination and market structure. The airline industry has been the setting for much of the empirical work on competition and price discrimination; yet, this work has uncovered conflicting empirical relationships and offered alternative economic arguments. In addition to the papers by Borenstein and Rose (1994) and Gerardi and Shapiro (2009), there have been a number of follow-on studies which have also found evidence of different relationships. Stavins (2001) finds that price dispersion due to ticket restrictions increases with competition, consistent with the Borenstein and Rose finding. Gaggero and Piga (2011), using data from the Irish airline industry, find that competition reduces fare dispersion, similar to Gerardi and Shapiro. Hernandez and Wiggins (2014), using a cross-section of U.S. domestic routes and data on ticket characteristics, find that competition from Southwest compresses the overall menu of fares though competition from other low-cost carriers does not. Dai et al. (2014) uncover a non-monotonic relationship between competition and fare dispersion, with competition increasing dispersion in concentrated markets but decreasing it in competitive markets.4 Kim (2015) extends the analysis of Gerardi and Shapiro and finds that the negative relationship between competition and price dispersion does not persist in more recent data. Finally, Zhang et al. (2015) show that competition from high-speed rail increases airline price dispersion, which is also consistent with Borenstein and Rose. Our paper reveals the source of the conflict in this earlier literature. We show that competition can increase or decrease price differentials and moreover, that with more than two types of consumers, competition may increase differentials between some while reducing them between others. It is perhaps not surprising then that earlier work which has measured the impact of competition on overall price dispersion has uncovered conflicting findings. Our empirical analysis instead estimates the impact of competition on fare differentials directly and finds evidence consistent with both Borenstein and Rose (1994) and Gerardi and Shapiro (2009)—that is, we find that some fare differentials rise with competition and others fall. More generally, our paper emphasizes that estimating the relationship between market structure and 4

However, they support this with a theoretical model which generates predictions regarding the intensity of competition in a market, not the number of competitors in the market which is what they measure in their empirical analysis.

5

price differentials in an industry requires a nuanced understanding of the sources of consumer heterogeneity and segmentation in that setting. The remainder of this paper is organized as follows. The next section lays out the theoretical considerations. In Section 3, we describe our empirical setting and our data. Section 4 presents our empirical strategy and explains how we test the basic predictions of the model. The results of our empirical analysis are presented in Section 5. A final section concludes.

2

Theoretical Considerations

In this section, we present a simple and highly stylized model to illustrate how competition may increase price differences between some groups of customers while decreasing price differences between others. The intuition that drives our results is similar to Borenstein (1985), which is explored further in Holmes (1989). Specifically, the key insight from these papers that we build on is that the effect of competition on price differentials depends on whether price discrimination is based on differences in consumers’ tendency to drop out of the market or their tendency to switch suppliers. We first summarize the key result from Holmes (1989), with slight modifications to fit our extension. Assume that two differentiated firms, A and B, face a set of potential consumers of two types, 1 and 2. Holmes assumes that the firms can practise thirddegree price discrimination, meaning that the firms can set separate prices for the two different groups of consumers. Holmes shows that the demand for each firm’s output, by each type of consumer, has an elasticity that can be decomposed into an industry-elasticity component and a cross-price elasticity component. Specifically, for either firm, the elasticity of demand by consumers of type i is given by: I C eF i (p) = ei (p) + ei (p)

(1)

Here, eI , the industry elasticity, measures how responsive aggregate industry demand is to changes in prices while eC , the cross-price elasticity, measures the impact on one firm’s demand from changes in the other firm’s price. Holmes then shows how the familiar inverse elasticity pricing rule determines equilibrium prices for each group of consumers: 1 1 (p∗i − c) = F ∗ = I ∗ ∗ pi e (pi ) e (pi ) + eC (p∗i )

(2)

As Holmes points out, this expression shows that, in symmetric oligopoly, price discrimination can be based on differences in consumers’ industry-demand elasticity and/or differences in consumers’ cross-price elasticities. Stole (2007) uses Holmes’ set-up to illustrate why, with two types of consumers, the relationship between competition and price differentials is ambiguous. Stole explains that if the goods are close substitutes (i.e.: both types of consumers have high 6

cross-elasticities of demand), then competition will drive prices in both segments towards marginal cost and the price differential across segments will be negligible. On other hand, if consumers with a high industry elasticity consider the goods to be close substitutes while consumers with a low industry elasticity have strong brand loyalty, then competition will lower prices to the former while firms maintain high prices for the latter. In this case, competition will lead to greater price differentials across consumer segments, relative to the case of monopoly. It is clear from Stole that both the Borenstein and Rose (1994) and Gerardi and Shapiro (2009) findings are theoretically possible and that the relationship between competition and fare differentials depends on the underlying source(s) of heterogeneity between travellers. We extend the two-type model from Stole to consider the possibility that travellers differ in terms of both their underlying value of a trip and their strength of brand loyalty and, moreover, that travellers who are similar on one dimension may still differ on the other. This gives rise to more than two types of travellers and that competition may increase price differentials between some types while decreasing them between others. We illustrate the intuition using a simple three-type model. In particular, we assume that Type 1 consumers have a low willingness-to-pay for a trip and no brand loyalty. These travellers, whom we call, price-sensitive leisure travellers, will choose to fly with the cheapest possible airline and, if prices are too high, they will choose not to fly at all. We assume that Type 2 consumers are travellers with a high willingness-to-pay for a given trip but little brand loyalty. These travellers, whom we call brand-indifferent business travellers, will purchase a ticket even if fares are high but will choose to fly with the airline offering the cheapest fare. The third type of travellers are brand-loyal business travellers who have both a high willingness-to-pay to take their trip and a high degree of brand loyalty. We focus on these particular segmentations of travellers because we believe they are consistent with key institutional features of the airline industry. A fundamental source of heterogeneity between travellers is their basic willingness-to-pay for a trip. Business travel is conducted to support some type of commercial or income-generating activity and therefore the reservation price for a business-related trip will typically be higher than that of a leisure-related trip. Therefore, we model business and leisure travellers as differing in their underlying willingness-to-pay or industry elasticity.5 In addition, travellers are heterogeneous in their degree of brand loyalty. In the airline industry, brand loyalty can result from both actual differentiation between airlines’ offerings as well as perceived differentiation resulting from airlines’ use of frequent flyer programs. These programs, which reward travellers for cumulative travel on a given airline, lower the degree of substitutability between otherwise very similar flights. Because these programs generally have a non-linear reward structure, they will be more highly valued by business travellers than leisure travellers since the 5

Note that travellers must differ in terms of their underlying value of a trip for there to be price discrimination in monopoly markets.

7

former fly more frequently.6 For this reason, business travellers are often assumed to be more brand loyal than leisure travellers. However, we recognize that business travellers themselves may differ in terms of their degree of loyalty, due to differences in corporate travel policies (which may offer the traveller varying amounts of flexibility in his choice of carrier and ticket type), differences in their preferences for in-flight amenities or even differences in the frequency or destination of travel which will impact the value of collecting frequent flyer points. We therefore assume that leisure travellers have low brand loyalty and that business travellers differ in terms of their degree of airline loyalty. We capture these sources of heterogeneity in travellers’ willingness-to-pay and their willingness-to-switch by assuming that Types 1 and 2 have the same crosselasticity of demand and differ only in terms of their industry elasticity while Types 2 and 3 have the same industry elasticity and differ only in their cross-elasticity.7 Specifically: eI1 > (eI2 = eI3 ) (3) (eC1 = eC2 ) > eC3

(4)

Similar to Holmes (1989), we assume that airlines are able to set separate prices for each of these three types of travellers, if they so choose. That is, we assume airlines practice third-degree price discrimination. In reality, airlines price discriminate through a combination of third-degree and second-degree strategies, both taking advantage of known information about travellers’ willingness to pay and also offering menus of fare and ticket characteristic bundles for travellers to choose from. For simplicity, we abstract from the self-selection problem and assume the airline can observe enough about each traveller’s type—for example, from his frequent-flyer program profile and the search parameters he enters—to charge them a different price. We begin by considering a price discriminating monopoly airline facing these three traveller types. In the case of a monopolist, the cross-price elasticity, eCi , is zero for all consumer types, implying that the firm’s elasticity is the same as the industry elasticity. The monopolist will set each group’s price, which we denote pM , according to the standard inverse elasticity rule. Therefore, for each Type i: 1 (pM i − c) = I M pi ei

(5)

M M Given equation 3 this implies that pM 1 < (p2 = p3 ). We now consider the impact on prices when there is competition from a second airline. Each firm in this symmetric duopoly sets a price for each group of consumers, 6

See Borenstein (1989), Borenstein (1991), Lederman (2007) and Lederman (2008) for discussion and empirical evidence on how frequent flyer programs impact fares and market shares. 7 These equality assumptions may be unrealistic but are used to starkly illustrate how the different sources of heterogeneity affect the relationship between market structure and price differentials. Assuming weak monotonicity in the inequalities below will not change the result.

8

denoted pD , according to the inverse elasticity rule. Therefore, for each Type i: 1 (pD i − c) = I D pi (ei + eCi )

(6)

D D Given equations 3 and 4 this implies that pD 1 < p2 < p3 . Note that, with competition, the consumers cross-elasticities of demand become relevant. We can now compare how the change in market structure affects prices to each group and examine how price differentials between each pair of types changes with M competition. Note first that, for all i, pD i < pi , or that prices are lower in duopoly than monopoly for all consumers. For each Type i, Equations 5 and 6 imply that the eC ratio of the monopoly to duopoly markup is: 1 + eIi . i Equations 3 and 4 imply that

eC2 eC1 eC2 eC3 > , and > eI2 eI1 eI2 eI3 Thus, competition reduces Type 2’s fares by more than either of the other types. Note the intuition behind the result that the Type 2 fares fall more than the other two types. Type 2 travellers need to fly, like Type 3’s; however, they are willing to switch carriers, like Type 1’s. Their low industry elasticity but high cross elasticity means that the airline can charge them high prices when it is a monopolist but cannot once there is competition. In contrast, the Type 1’s high industry elasticity means the airline cannot charge them very high prices even under monopoly and so competition does not impact their fares as much. The Type 3’s low cross elasticity means that the airline can continue to charge them high prices even under competition and so competition also does not impact their fares as much. What does this imply for how competition affects price differentials between different groups of consumers? It is clear that whether competition increases or decreases price dispersion will depend on which groups’ fares are compared. Since fares for Type 2’s fall by more than the other two types, competition should decrease the differential between Type 2’s and Type 1’s and increase the differential between Type 3’s and Type 2’s. Without additional structure on the model, we cannot determine whether competition lowers Type 1 or Type 3 fares more. However, we know that competition should either increase the differential between Type 3’s and Type 1’s (which will occur if fares to leisure travellers fall by more than fares to brand loyal business travellers) or decrease the differential between them but by less than the change between Type 2’s and Type 1’s. More generally, the model suggests that, if airlines are able to segment travellers based on both their underlying value of a trip and their degree of brand loyalty, competition will increase the price differential between travellers who have different levels of brand loyalty but decrease the differential between travellers whose only source of heterogeneity is their underlying willingness-to-pay.8 8

For completeness, we could also consider the existence of a fourth type of traveller with a

9

This simple model illustrates two key points which impact an empirical analysis of the relationship between competition and price discrimination. First, we have shown that with more than two types of consumers, competition may increase the price differential between groups while decreasing it between others. This implies that empirical analyses that measure changes in overall price dispersion using a metric like the Gini coefficient may not be informative about the underlying changes in price differentials that have occurred. Second, we have shown that the largest impact of competition may be on neither the cheapest nor most expensive fares but rather on fares in the middle. Since it is typically not possible to know which tickets are sold to which types of travellers, previous work has often compared the impact of competition on the top and bottom of the fare distribution as a way to distinguish tickets sold to business and leisure travellers. Our simple model suggests that it may more informative to estimate the impact of competition on the overall distribution as focusing on the extremes may miss the largest effects.

3 3.1

Empirical Setting and Data Empirical Setting: The Canadian Airline Industry

Our empirical setting is the Canadian domestic airline industry. The Canadian market has several features that make it well suited for a study of market structure and price discrimination. First, Canada has had only one full-service price-discriminating airline—Air Canada—operating in recent years. Air Canada is, by far, the largest airline in the country, in terms of both the number of routes served and passengers carried. Air Canada provides service on virtually all of the top domestic routes in Canada and on 100% of the routes in our regression sample. We therefore focus our empirical analysis on Air Canada’s pricing behavior, investigating how its fares for different types of tickets change as it faces varying of levels of competition on a route. Second, market structure is straightforward to measure in the Canadian setting. There is little connecting service in Canada because Canadian airlines do not operate large hub-and-spoke networks.9 Rather, they mostly operate point-to-point flights, low willingness-to-pay to travel but high brand loyalty, whom we could call a brand-loyal leisure traveller. Assume that the brand loyal leisure traveller had the same industry elasticity as our leisure traveller above but the same cross-elasticity as the brand loyal business travellers. Using the same logic as above, we can show that competition has the smallest effect on these travellers. Intuitively, this is because their prices are already relatively low under monopoly and, due to their high brand loyalty, fall little with competition. In terms of differentials, competition would increase the differential between these travellers and the (brand-indifferent) leisure travellers and the brand indifferent business travellers but decrease the differential between these travellers and the brand loyal business traveller. These patterns are consistent with the more general implication of our model that the impact of competition on fare differentials between consumers will depend on whether those consumers differ in terms of their industry elasticities, cross elasticities or both. 9 Air Canada does have a hub in Toronto. However, Air Canada also offers non-stop service

10

focusing on the larger cities in the country. In contrast, in the U.S., there are typically multiple carriers offering connecting service between any two cities, leading to different measures of market structure depending on whether the researcher focuses on only direct service or on direct and connecting service. In addition, there is no domestic codesharing between Canadian carriers so there is no need to distinguish between operating and marketing carriers when measuring competition. With the exception of Air Canada, there is also no use of regional partners. Finally, there is only one multi-airport city in Canada (Toronto). The existence of multi-airport cities can make market structure measures sensitive to the researcher’s decision about market definition. Third, the Canadian market offers the opportunity to examine changes in fares and fare differentials as routes move between monopoly and duopoly. Because of the small number of carriers serving the domestic Canadian market, and Air Canada’s long-standing dominance, there are routes on which the airline is a genuine monopolist for at least part of our sample period. By contrast, even with recent consolidation, it is rare to find routes in the U.S. with only a single airline offering direct service, especially when restricting attention to travel between large airports, as we do in this paper. The comparison between monopoly and duopoly forms the basis of much of the theoretical work in this area, and therefore our setting provides a better and more direct test of the theoretical implications than in the U.S. market. Since there is virtually no previous empirical work on the Canadian industry, we provide some background information to illustrate how the Canadian industry compares with the U.S., which has been extensively researched.10 Table 1 presents the 10 largest Canadian airports based on total annual enplanements in 2011. To demonstrate how Canadian airports compare to U.S. airports in size, we also show, for each Canadian airport, a U.S. airport of comparable size and indicate the rank of that airport. As the table shows, Canadian airports are generally significantly smaller than U.S. airports, with the third largest airport in Canada roughly the same size as the 19th largest in the U.S. and the tenth largest roughly the same size as the 77th largest in the U.S.11

3.2

Data and Construction of Sample

The primary source of data for our empirical analysis is the Airport Data Intelligence (ADI) database, compiled by Sabre Holdings. Sabre is a travel technology company that owns a global distribution system (GDS) used by thousands of travel agents (including several of the large online agencies). Based on its GDS bookings, as well between all of Canada’s large cities and the vast majority of its passengers fly non-stop itineraries. 10 We refer the interested reader to Chandra and Lederman (2014a) for a detailed description of the Canadian industry. 11 These rankings are based on enplanements, not originations or trips. The low enplanement numbers at Canadian airports reflect both the smaller number of passengers in the market as well as the lack of connecting service since connecting itineraries generate multiple enplanements per trip.

11

Table 1: Top Canadian Airports, and Comparable US airports Canada Rank

Airport

1 2 3 4 5 6 7 8 9 10

Toronto Pearson Vancouver Montreal Trudeau Calgary Edmonton Ottawa Halifax Victoria Kelowna Quebec City

U.S. Comparable Enplanements 32,278,458 16,394,986 13,228,564 12,073,264 6,156,730 4,359,055 3,482,421 1,456,782 1,355,975 1,343,021

Airport

Rank

Chicago O’Hare Newark Boston New York LaGuardia St. Louis Sacramento Cincinnati El Paso Tulsa Manchester

2 14 19 20 31 40 51 72 76 77

Source: Statistics Canada’s “Air Carrier Traffic at Canadian Airports” (2011); Federal Aviation Administration’s “Passenger Boarding and All-Cargo Data” (2011). Both sources include domestic and international passengers.

as data it collects to capture bookings that do not go through its GDS, Sabre produces the ADI database, which contains fare and booking information for most passengers and flights worldwide, from January 2002 until the present. Our analysis uses data on travel within Canada from 2002 until 2011. The level of observation in the ADI data is the airline-itinerary-year-month-cabin class-fare class. This means that—for each month and for each pair of airports in Canada—we observe every airline that offered direct or connecting service between those airports, the number of passengers who travelled that itinerary on each airline in that month in a given cabin and fare class, and the average fare they paid.12 The data are further broken down by direction of travel so that passengers flying from Toronto to Vancouver, for example, can be distinguished from those flying from Vancouver to Toronto, and are also broken down by point of origin. We complement the ADI data with flight schedule data from the Official Airlines Guide (OAG). The OAG data provide a one-week snapshot of all scheduled flights between all Canadian airports over the same period. The OAG data provide measures of capacity and frequency by carrier and also provide a second source of data on entry and exit dates which is useful for constructing and checking our market structure measures. For our regression analysis, we limit our sample to routes between the top 15 cities in Canada.13 Travel between these 15 cities accounts for approximately 65% of all domestic travel in Canada. The average route in this sample has about 8,000 12

Cabin and fare classes are explained in more detail below. These 15 cities contain 17 airports, since there are three airports in the Toronto area. The top ten airports appear in Table 1. 13

12

monthly passengers in the ADI data and about 7,000 direct monthly passengers.14 The largest route in the sample (Toronto-Montreal) has, on average, over 100,000 monthly passengers in the ADI data. Averaging across routes in this sample, 59% of the passengers on a route travel on direct itineraries. However, in this sample as a whole, direct passengers account for over 87% of passengers, indicating that connecting passengers are concentrated on the smaller routes. We focus our empirical analysis on the impact of competition on Air Canada’s fares on direct itineraries on routes between the top 15 cities. Air Canada provided service on 158 routes between the top 15 cities, with 118 of these routes being served non-stop. These 118 routes form the basis of our regression sample. Across these routes, Air Canada’s average share of direct or one-stop passengers on the route is 47% and its average share of direct passengers on the route is 48%.

3.3

Cabin Class and Fare Class Data

A novel and important feature of the ADI data is that it includes information on the cabin and fare class of purchased tickets. This information helps us distinguish tickets of different types. In contrast, the commonly used Databank 1B provides no information as to why observationally similar tickets are associated with different fares. The cabin class refers to the broad categorization of a ticket by an airline. In the data, across all airlines, the cabin class variable takes on one of six possible values. However, most airlines only use a subset of the values. In our regression sample, over 98% of Air Canada’s itineraries take on one of three values which correspond to the following three ticket categories: Discount Coach, Coach or Business. Aggregating across all route-months in the regression sample, 70% of Air Canada’s passengers travel in Discount Coach, 26% in Coach class and 4% in Business class. Air Canada does not necessarily sell tickets in all three classes on every route. Business class tickets require a separate cabin on the aircraft and some of Air Canada’s smaller planes do not have separate business class cabins. Within our sample of Air Canada’s direct itineraries between the top 15 cities, we observe Discount Coach tickets on all of Air Canada’s routes in every month, Coach tickets on 98% of route-months and Business tickets on 70%. Note that while Business tickets provide passengers with a seat in a separate cabin and a higher level of pre- and in-flight service, Coach and Discount Coach passengers travel in the same cabin of the aircraft and experience the same service during the flight. Instead, these tickets are typically distinguished based on features such as refundability, fees for flight changes, and frequent flyer point accumulation. Within our data, tickets are also distinguished based on their fare class or booking 14

We exclude routes that have fewer than 400 passengers per month in our data. Routes are distinguished by direction of travel so Toronto to Vancouver is a distinct route from Vancouver to Toronto.

13

code.15 Fare classes are a finer level of categorization than cabin classes and multiple fare classes will be associated with a given cabin class. Fare class codes are used by airlines to distinguish tickets that are associated with different fares and, in some cases, different features. For example, during our sample, Air Canada offered two types of Business Class tickets which differed in terms of their refundability and change fees. These two types of Business Class tickets would be associated with different fare classes. Similarly, Air Canada offered several types of Discount Coach tickets, differing in terms of features such as change fees, eligibility for same-day standby and frequent flyer point accumulation. These would also be associated with different fare classes. In addition, fare classes are used to distinguish tickets which are identical from the customers’ point of view but which are given different prices by the airline. For a given flight, airlines will sell a given type of ticket at different fares as a form of price discrimination and/or due to differences in the shadow cost of a seat. For example, the price of a given type of ticket on a given flight often increases as the departure date approaches. This may reflect an increase in the shadow cost of that seat or it may reflect a price discrimination strategy in which airlines use advance purchase requirements to charge higher fares to consumers who book last minute as they are presumed to be business travellers with a higher willingness-to-pay. Table 2 shows how the tickets in our data map to cabin classes and fare classes. The table shows that cabin classes are associated with multiple fare classes. But, in most cases, a given fare class implies a unique cabin. For example, all the codes that fall alphabetically between N and X are used exclusively to denote Discount Coach tickets while the C, D and J codes are used exclusively for Business Class travel. The codes that do not imply a unique cabin are B, I, M, Y and Z. These exceptions appear to be due to Air Canada changing, over time, the set of codes associated with a given cabin class. Appendix B (which was produced by Air Canada) shows a mapping of Air Canada’s fare classes to its different ticket types in 2009. There are a number of things to note from this table. First, the table shows how Air Canada’s ticket types in 2009 map to the cabin classes in our data. The airline offered two types of Discount Coach tickets (called Tango and Tango Plus), one type of Coach ticket (called Latitude) and two types of Business tickets. Second, the table shows that, at a point in time, fare classes map uniquely to ticket types. Third, the table shows that different fare classes are associated with different types of tickets but a given ticket type will have multiple fare classes associated with it. Finally, the mapping of fare classes to ticket types appears to be largely consistent with the patterns from our data shown in Table 2. For example, classes C, D, J and Z are associated with Business class tickets in our data, just as they are in the Air Canada document. In both cases, classes Y and B are associated with Coach tickets. The mapping is not perfect, however, because over the course of our sample period Air Canada could have changed its use of specific 15

Fare classes are typically designated using a single letter of the alphabet.

14

Table 2: Passengers by Class and Cabin (000s) Class

A B C D E G H I J K L M N P Q R S T U V W X Y Z Total

Cabin

Total

Business

Coach

Discount Coach

0 0 885 98 0 0 0 52 1,075 0 0 0 0 0 0 0 0 0 0 0 0 0 0 71 2,181

0 343 0 0 0 0 0 0 0 0 0 35 0 0 0 0 0 0 0 0 0 0 13,652 0 14,030

3,177 1,344 0 0 906 1,110 1,761 312 0 325 4,441 1,497 962 469 3,441 439 1,634 1,067 978 3,251 961 14 8,652 1,027 37,767

3,177 1,687 885 98 906 1,110 1,761 364 1,075 325 4,441 1,532 962 469 3,441 439 1,634 1,067 978 3,251 961 14 22,304 1,098 53,979

Note: Data for 2002–2011 for top 15 domestic routes in Canada.

15

fare codes. We encounter one challenge in utilizing the cabin and fare class information in the data. For most of the sample, the Y fare class is classified as a Coach ticket. However, there are a number of route-months in our data (specifically, in some months in 2008, 2009, 2010 and 2011) in which the Y code appears to have been assigned to almost all passengers on the route, including all of those in Discount Coach. In those months, Y class tickets, on average, account for 62% of all passengers on the route while, in all other months, Y class tickets, on average, account for around 4% of tickets. Given that Air Canada’s official documents consistently indicate that Y class denotes a fullprice coach ticket, we believe that this classification is an error in the data. The fact that this occurred simultaneously on all routes in our sample, during these particular months, further suggests these codes were assigned in error. To ensure that these observations do not impact our analysis, we exclude any route-month observations in which the fraction of Y class passengers exceeds 17%, which is the maximum share of passengers accounted for by Y-fares in any route-month prior to 2008.16 This drops a total of 3755 route-months from our sample.17

4

Empirical Approach and Identification

The goal of our empirical analysis is to investigate how competition impacts differences in the fares charged to different types of passengers. While previous work in this area has largely focused on the impact of competition on the overall amount of fare dispersion on a route (captured by an index such as the Gini coefficient), we instead try to distinguish tickets of different types and estimate whether competition affects their fares differently. We also estimate how competition impacts different parts of the overall fare distribution.18

4.1

Regression Specification

Our main estimating equation is a simple reduced-form specification. Denoting routes by r and time-periods by t, we estimate the effect of competition on a specific fare using: log prt = β0 + β1 Competitionrt + λr + θt + rt (7) 16

The results are robust to using alternative cutoffs and even to treating the original data as accurate. 17 As we describe in more detail below, because much of variation in market structure arises in the earlier years of the sample, dropping some observations from these latter years has little impact on our findings. 18 Borenstein (1989) estimated the impact of hub dominance on different percentiles of the fare distribution. In their analysis, Gerardi and Shapiro (2009) estimate the impact of competition on both the Gini coefficient and various percentiles of the fare distribution.

16

where λ and θ denote route and time fixed-effects, respectively. An observation is a route-month combination.19 We cluster standard errors at the route-level. We express prices in logs to measure the proportional effect of competition on various fare measures. Doing so allows us to compare differences in the estimated coefficients in order to determine the effect of competition on the ratio of fares for different tickets. In particular, assume that for two distinct types of fares, i and j, the estimated coefficients on the competition variable are βˆ1i and βˆ1j . Since these estimated coefficients represent the proportional effect of competition on fares, price dispersion will rise or fall depending on the value of βˆ1i − βˆ1j .

4.2

Variables used in the Regressions

Fare Measures We explore the relationship between market structure and fare differentials in three ways. First, we compare the impact of competition on the average fare of tickets in each of the three different cabin classes: Business, Coach and Discount Coach. This allows us to examine, at a broad level, whether the prices of Air Canada’s tickets in different cabin classes are affected differently by competition from rival airlines. These average fares are weighted by the number of passengers in each fare code, subject to the cut-off that each fare code accounts for at least 5% of passengers in the route-month.20 Second, we measure the impact of competition on different types of fares within the Discount Coach class. Recall that approximately 70% of the passengers in our data travel on Discount Coach tickets and there are over 20 different fare classes within Discount Coach.21 While the ADI data allow us to distinguish tickets by fare class, we are unable to map the fare classes to specific types of tickets. Put differently, we know the different fares associated with different fare classes but not why those fares differ. Therefore, to explore the relationship between market structure and within-Discount Coach price differentials, we estimate how competition affects fares in the most and least expensive codes within Discount Coach. As above, we impose passenger number cut-offs to ensure that the fare codes constitute a non-negligible number of passengers. Third, we estimate how competition affects the full distribution of fares within the Discount Coach category. We construct the empirical distribution of Discount 19

Recall that the regression sample only includes observations on Air Canada’s fares. Such cut-offs are necessary for eliminating small outlier observations containing a few passengers, which can reflect either errors in the recording of fares and codes, or else unusual means of purchasing tickets such as frequent-flier rewards, employee discounts, upgrades granted after the time of booking etc. In the online appendix we show that our measures of price dispersion are not sensitive to these cut-offs. 21 By contrast, the full-fare Coach and Business cabins use just one or two booking codes each, and therefore there is little point in examining dispersion within these service classes. 20

17

Coach fares on each route by assuming that every customer within a given fare class paid the average fare associated with that fare class. This then allows us to calculate the percentiles of the Discount Coach fare distribution and estimate how different percentiles are affected by changes in market structure. It is worth pointing out that competition can lead to lower fares in two ways. Air Canada could reduce the fares it charges for tickets in a given fare class or it could increase the number of tickets it allocates to inexpensive fare classes or possibly both. If the fare reduction occurs through a reduction in the price of tickets in a given fare class, this should be captured in both the regressions which use the prices of different fare classes as the dependent variable and the percentile regressions. On the other hand, if the airline keeps the average prices of tickets in each class the same but reallocates capacity from expensive to less expensive classes, this would not appear in our regressions that examine the prices of various fare classes but would appear in the percentile regressions. Market Structure Measures We measure the competition faced by Air Canada on a given route-month in three ways: (i) the number of carriers, other than Air Canada, that provide direct service on the route in the corresponding month, (ii) indicators for whether the market is a duopoly or competitive (which we define as having three or more carriers), with AC’s monopoly routes being the omitted category, and (iii) the negative of the log of the Herfindahl Index in the route-month.22 When constructing the market structure measures, we restrict the sample to the main nationwide-carriers that existed during our sample period. Along with Air Canada, there were four such carriers, all of which were essentially low-cost carriers: Westjet, Porter, Canjet and Jetsgo.23 All of these four carriers offer only a single class of service on their aircraft. Together, these five airlines account for over 85% of domestic airline passengers in Canada, and over 99% of passengers within our sample of routes between the top 15 cities.24 22

These are the same measures used in Gerardi and Shapiro (2009) although Gerardi and Shapiro (2009) use the log of the number of carriers and we use the level since Air Canada is a monopolist on a number of routes. 23 Porter is not exactly a low cost carrier; it features amenities that are more commonly associated with a ‘Premium Economy’ class of service, such as leather seats and free snacks on board and in its airport lounges. However, Porter offers a single aircraft cabin and its fares are usually lower than Air Canada’s Discount Coach fares. See Chandra and Lederman (2014b) for a note on Porter Airlines and its effects on Air Canada’s fares. 24 Note, in particular, that we drop charter airlines, as well as small carriers such as Bearksin Airlines which operate small planes on some of the routes in our sample.

18

Table 3: Summary Statistics: Regression Sample

Business Fare Coach Fare Discount Coach Fare (mean) Discount Coach Fare (min.) Discount Coach Fare (max.) Num. Direct Rivals Duopoly Competitive HHI

Mean

SD

Min.

Max.

N

924.4 606.4 235.1 158.5 340.0 0.89 0.62 0.13 0.69

474.6 301.9 99.3 75.2 156.4 0.64 0.49 0.34 0.22

250 200 65 50 74 0 0 0 0

2749 2588 808 517 1528 3 1 1 1

4203 5688 7308 7308 7308 7308 7308 7308 7308

Note: An observation is a combination of origin-destination-year-month.

Summary Statistics Table 3 presents summary statistics on our fare and markets structure variables. The level of observation in the table is the route-month and we have a total of 7,308 observations in the regression sample. Air Canada serves all of these routes in all months by construction. As the table indicates, across route-months, the average Discount Coach fare is $235 while the average Coach and Business fares are $606 and $924, respectively.25 On average, Air Canada faces less than one direct competitor on its routes. About 62% of route-months have Air Canada facing one competitor in direct service while 13% of route-months have two or more such rivals. Based on the distribution of passengers across carriers, the average Herfindahl index on a route is a very high 69%. Table 4 provides a sense of how fares vary across various booking classes and percentiles within Discount Coach for a sample of large routes, as well as for the full sample. The top panel shows average fares for five booking classes and the lower panel does the same for a selection of percentiles of the fare distribution. Among the five booking classes summarized, A class fares are consistently the cheapest, followed by V, H, R and then M fares. Across all route-months, the 99th percentile fare within Air Canada’s Discount Coach class is almost four times as expensive as the very cheapest fare on the same route in the same month.

4.3

Identification

Our empirical analysis consists of a series of reduced-form regressions in which we relate various fare measures to route-level market structure. All of our regressions 25

As mentioned in Section 3, not all routes have Coach or Business class service which explains the lower number of observations for these fares.

19

Table 4: Variation in Discount Coach Fares: Selected Routes Fare Class

A H M R V

Toronto to/from: Ottawa

Montreal

Vancouver

168 205 272 276 201

187 220 290 284 209

302 360 566 556 334

Percentile

1 25 50 75 99

All Routes

Toronto to/from:

211 304 481 417 267 All Routes

Ottawa

Montreal

Vancouver

90 155 186 215 326

96 161 198 225 351

192 262 286 323 729

134 198 224 270 561

Note: Values represent mean fares in the corresponding fare class or percentile within the Discount Coach fare distribution.

include route, year and month fixed effects. Thus, our analysis exploits variation in market structure within routes over the 120 months of our sample and our estimates capture how Air Canada changes its fares for different types of tickets as market structure on a route changes. Even with the inclusion of route fixed-effects, one might be concerned about the endogeneity of the market structure measures. While the route fixed effects control for route-level unobservables that may be correlated with market structure and fares, changes in market structure over time—which result from the entry and exit decisions of competing airlines—could still be correlated with time-varying unobservables which could also affect Air Canada’s fares. In order to assess this issue, it is helpful to understand the sources of variation in market structure in our ten-year Canadian sample. Out of the 118 routes served by Air Canada in our sample, 77 experience at least one change in market structure during the period we study and many of these experience multiple changes in market structure. Variation in market structure arises from three sources. First, during the early years of our sample, three Canadian low-cost carriers (WestJet, CanJet and Jetsgo) were expanding their domestic networks. Between 2002 and 2004, WestJet entered 39 routes, CanJet entered 21 routes and JetsGo entered 32 routes. Second, between 2007 and 2010, Porter Airlines, which began operations in 2007 out of Toronto’s Billy Bishop Airport, entered 18 routes. Third, in 2005 and 2006 respectively, JetsGo and CanJet exited the industry. Their exit provides a change in market structure on 20

all of the routes they had been previously serving. 27 routes had service by JetsGo in the months preceding its exit and 17 routes had service by CanJet in the months preceding its exit. Concern about the possible endogeneity of the market structure measures results from the fact that airlines’ entry and exit decisions may be correlated with route-level time-varying factors that are unobservable to the researcher. For example, airlines might choose to enter routes that are experiencing demand growth or growth from particular customer segments which could also affect the prices charged for different types of tickets. While this is possible, we think this concern is largely alleviated in our setting. First, a significant portion of the variation in market structure that we exploit results from the full-scale exit of CanJet and JetsGo. Their exit decisions were likely based on industry- and firm-level considerations and not driven by particular unobservable changes on the routes they were serving right around the time of their exit. As a robustness check, we also estimate our models using only the variation in market structure that results from their exit. Second, the variation in market structure that results from airlines’ entry decisions is also less likely to be as influenced by route-specific time-varying unobservables than in other settings. This is because the airlines which expanded their operations during our sample were constrained—both due to geography and due to their aircraft technology—in terms of the routes they could serve. WestJet, CanJet and JetsGo were all low-cost carriers operating one (or, in JetsGo’s case, two) aircraft type(s) and flying mostly point-to-point operations. This means that they could only enter routes that could be served by the type of plane they had and that also had large enough populations to sustain sufficient point-to-point traffic. In addition, each began with a particular geographic focus and expanded outward from their initial focus. For example, WestJet began in western Canada, initially providing service between Vancouver, Calgary, Edmonton, Winnipeg and Kelowna. It then added Regina, Saskatoon and Victoria—additional western destinations—to its network. Much of the entry we observe by WestJet during our sample is routes in and out of Toronto, Montreal and Ottawa as it began its expansion eastward. Eventually, it added service to destinations in eastern Canada, such as Halifax and Moncton. While there is obviously still some endogenous selection by the airlines of which route to enter when, the set of routes these airlines could theoretically enter was largely determined by route distance, geography and exogenous demand conditions such as population, and the timing of their entry was influenced by a natural expansion outward from their initial geographic focus. Similarly, the expansion of Porter Airlines in the latter part of our sample is also partly driven by exogenous considerations. Porter Airlines began operations out of Toronto’s Billy Bishop Airport in 2007. This was a small airport in downtown Toronto that had not been used for commercial flights for many years. Porter operates only Bombardier Q400 planes and has been constrained in adopting any other type of aircraft because of the runway length at the airport and because of city regulations. 21

As a result, as Porter expanded, it only entered routes that are within the flying range of the Q400. In addition, Porter’s expansion has been largely focused out of Billy Bishop Airport and, to a lesser degree, Ottawa. It has explicitly not expanded westward. Thus, as in the case of the three low-cost carriers, the set of routes that Porter could theoretically enter was largely determined by distance, geography, and demand conditions.

5

Results

Our main results are presented in Tables 5 through 7. Table 5 investigates the impact of competition on cross-cabin price differentials while Tables 6 and 7 investigate how competition impacts within Discount Coach price differentials. We then present a number of extensions and robustness checks. We conclude the section with a discussion of how our results relate to the theoretical considerations laid out in Section 2. Table 5 presents estimates of the relationship between market structure and average fares in each of the three cabin classes—Business, Coach and Discount Coach. For each cabin class, we show the results of estimating equation 7 using each of the three market structure variables described above. Looking first at the specifications that include the number of non-stop rivals as the measure of competition (columns 1, 3 and 5), the coefficient estimates indicate that having an additional non-stop rival on a route lowers Air Canada’s Discount Coach fares by about 7%, but has no statistically significant effect on average fares in either of the two higher cabins. Similarly, when we measure market structure using the dummy variables for a duopoly or competitive market (columns 2, 4 and 6), we find that competition only impacts Air Canada’s Discount Coach fares. The estimates in column 2 suggest that moving from a monopoly to duopoly setting reduces Air Canada’s average Discount Coach fares by about 8%, and that the introduction of additional competition reduces fares by another 7 percentage points. The estimates using the Herfindahl index as the measure of competition show a similar pattern. Since competition reduces Discount Coach fares but has no impact on Coach or Business class fares, the findings in Table 5 indicate that competition increases cross-cabin fare differentials, relative to monopoly. These results are consistent with the finding in Borenstein and Rose (1994) which found that competitive routes were associated with greater fare dispersion, albeit in a different setting and using cross-sectional estimation strategy. We next investigate how market structure affects price differentials across different types of Discount Coach tickets. Table 6 examines how the number of direct rivals affects the most and least expensive fare codes in Discount Coach on each route. We also examine the impact on the two most and two least expensive fare codes.26 The coefficient estimates in the first two columns of the table indicate that competition 26

From this point on, we only report results using the number of rivals as our measure of market structure, since the results are very similar using the other competition measures, as was seen in Table 5. Interested readers can consult the online appendix to this paper to view the results using

22

23 0.884 5688

0.885 5688

0.884 5688

0.001 (0.026) 6.053a (0.022)

(6)

0.958 4203

6.341a (0.016)

-0.008 (0.007)

(7)

(9)

0.958 4203

0.958 4203

-0.027c (0.015) a 6.342 6.343a (0.016) (0.015)

-0.010 (0.013) -0.018 (0.016)

(8)

Business

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses.

c

0.910 7308

0.911 7308

R2 Obs 0.912 7308

6.062a (0.023)

Constant

Competitive

Duopoly

6.045a (0.021)

(5)

-0.108a (0.018) a a 4.932 4.940 4.925a (0.014) (0.015) (0.014)

0.017 (0.012)

(4)

-Ln(HHI)

(3)

-0.022 (0.018) 0.044c (0.025)

-0.069a (0.010)

(2)

Coach

-0.083a (0.014) -0.154a (0.019)

Num. Direct Rivals

(1)

Discount Coach

Table 5: Regression of Cabin Level Average Fares on Competition Measures

Table 6: Regression of Fare Codes within Discount Coach

Num. Direct Rivals Constant R2 Obs

(1) Lowest

(2) Highest

(3) Low-2

(4) High-2

-0.045a (0.012) 4.345a (0.012)

-0.078a (0.015) 5.429a (0.030)

-0.061a (0.011) 4.573a (0.011)

-0.082a (0.014) 5.278a (0.028)

0.864 7308

0.747 7308

0.918 7308

0.804 7308

c

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses.

has a larger impact on expensive Discount Coach tickets than on cheap ones. Air Canada’s fares for tickets in the most expensive booking class fall by about 7.8% with the entry of an additional non-stop rival while fares for tickets in the cheapest class fall by about 4.5%. This suggests that competition reduces intra-class fare differentials. Examining the two lowest and two highest classes tells a similar story although the impact of competition on the two lowest fare classes is larger than the impact on the single lowest fare class. This suggests that competition may have a larger impact on fares as one moves up the fare distribution, something we explore further in our next set of results. We note, however, that the findings in Table 6 are consistent with those in Gerardi and Shapiro (2009) which found that competition reduced overall fare dispersion, in particular by reducing high-priced tickets more than low-priced tickets.27 In Table 7 we estimate the effect of the number of rival carriers on selected percentiles of Air Canada’s Discount Coach fare distribution. For comparison, we also present the regression coefficients for Air Canada’s average fares in the two other cabin classes. The coefficient estimates suggest that competition has a different impact on different types of Discount Coach tickets. In particular, the greatest impact of competition on Air Canada’s fares lies somewhere in the middle of the Discount Coach fare distribution. Among the selected percentiles, the largest effect of an additional carrier is between the 25th and the 75th percentile—a 7% to 8% reduction in fares. In contrast, we estimate smaller effects of competition on the tails of the distribution—between approximately 2% and 3.5%. The coefficients in the middle of the Discount Coach distribution—specifically, those at the 25th, 50th and 75th the other measures. 27 Gerardi and Shapiro (2009) is based on a sample that only includes Coach tickets. Their data, however, does not allow them to distingiush Discount Coach from (full-price) Coach tickets.

24

Table 7: Regression of Discount Coach Percentiles (1) (2) (3) (4) (5) (6) (7) 1 25 50 75 99 Coach Biz Num. Direct Rivals -0.035a -0.082a -0.072a -0.080a -0.020b 0.017 -0.008 (0.007) (0.012) (0.011) (0.016) (0.009) (0.012) (0.007) Constant 4.325a 4.706a 4.860a 5.121a 5.577a 6.045a 6.341a (0.011) (0.015) (0.022) (0.037) (0.022) (0.021) (0.016) 2 R 0.834 0.920 0.891 0.844 0.861 0.884 0.958 Obs 7308 7308 7308 7308 7308 5688 4203 c

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses.

percentiles—are significantly different from those at the extremes.28 To get a better sense of how competition affects the entire range of fares, we present Figure 1, which plots the coefficient estimates on the number of rival carriers for every fifth percentile in the fare distribution. The figure has a clear U-shape, indicating that the greatest effect of competition occurs between the 15th and 75th percentiles of the Discount Coach fare distribution. By contrast competition has a much smaller effect at either end of this distribution as well as on fares in the higher service classes. This implies that, even within Discount Coach, competition reduces the differential between some tickets but increases the differential between others.

5.1

Extensions and Robustness Checks

We now present a number of extensions and robustness checks. We first examine the effect of competition from two specific rivals to Air Canada: Porter Airlines and Westjet Airlines. There are reasons to believe that these two carriers may have had distinct effects on Air Canada’s fares for certain types of tickets. Porter Airlines is a relatively new, regional airline focused on travel out of its hub in Toronto, as was discussed in Section 4. Porter uses the Billy Bishop airport in downtown Toronto, which is often much more convenient for travelers than Air Canada’s hub at Pearson Airport. Porter is believed to appeal especially to business executives who work downtown, for whom the airport is a short distance from their offices. Thus, if competition has the greatest impact on business travellers who are willing to switch between airlines, then competition from Porter on routes in or out of Toronto should have a particularly large impact. 28 An intuitive way to see that these differences are significant is to examine the ratios of various percentiles, since these are direct measures of price dispersion. Table 11 in the Appendix shows that competition has a negative effect on the ratio of the 50th or 75th percentiles to the 1st or 5th, and a positive effect on the ratio of Coach and Business class fares, to the average Discount Coach fare. These effects are statistically significant.

25

Figure 1: Effect of the Number of Rivals on Percentiles of AC’s fare distribution

Effect of Num Carrs

0.00

−0.03

−0.06

1

5

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 99 LatExec

Selected Percentiles Notes: Values represent coefficients from regressing each percentile in the fare distribution on the number of direct rivals faced by AC on a route. Other controls include route, month and year FEs.

26

Table 8: Regression of Discount Coach Percentiles: The Effect of Porter Airlines

Porter Toronto Porter non-Toronto Other direct carriers Constant R2 Obs

(1) (2) (3) (4) (5) (6) (7) 1 25 50 75 99 Coach Biz -0.023 -0.231a -0.276a -0.331a -0.130a 0.028 0.001 (0.015) (0.026) (0.023) (0.031) (0.021) (0.033) (0.013) 0.001 -0.018 -0.061b -0.183a -0.057c 0.129a 0.011 (0.025) (0.021) (0.029) (0.025) (0.033) (0.026) (0.047) -0.038a -0.074a -0.058a -0.057a -0.010 0.013 -0.009 (0.008) (0.010) (0.009) (0.013) (0.009) (0.013) (0.008) 4.327a 4.699a 4.849a 5.103a 5.570a 6.048a 6.342a (0.011) (0.013) (0.020) (0.035) (0.021) (0.021) (0.015) 0.835 0.922 0.895 0.851 0.861 0.885 0.958 7308 7308 7308 7308 7308 5688 4203

c

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses.

To explore this, we estimate our percentile regressions with separate variables to capture the impact of competition from Porter on Toronto routes, competition from Porter on non-Toronto routes and competition from other carriers. Table 8 presents the results of these regressions. The estimates indicate that the U-shaped pattern of fare reductions that we found earlier is most pronounced when Air Canada faces competition from Porter on its Toronto routes. The impact of competition from Porter on those routes is much larger than the impact of Porter on other routes or the impact of other carriers. This suggests that travelers who purchase tickets in the middle and upper portions of the Discount Coach distribution have a greater crosselasticity with respect to Porter in Toronto than they do to other carriers or to Porter in other markets.29 We now turn to effects of competition from Westjet Airlines, which is a low-cost carrier competing nationally with Air Canada on most major routes. We exploit the fact that, for the early part of our sample, Westjet’s service from the Toronto area was from Hamilton airport, which is located about 40 miles from downtown Toronto. Over time, Westjet gradually shifted operations from Hamilton to Toronto’s Pearson airport. This means that, for a sample of routes to and from Toronto, we observe periods when Westjet’s operations were from a considerably less desirable location than Air Canada’s flights from Toronto. This would have implied possibly weaker competition for Air Canada, especially for business travelers who would not have been expected to make the drive to Hamilton. To explore this, we re-estimate our percentile regressions allowing competition 29

The coefficient on Coach fares for non-Toronto routes has an unexpected positive sign, but all other coefficients accord with expectations.

27

Table 9: The Effect of Westjet’s Competition from Hamilton Airport (1) 1 Westjet at Pearson Westjet at Hamilton Other direct carriers Constant R2 Obs

(2) 25

(3) 50

(4) 75

(5) 99

(6) Coach

(7) Biz

-0.039a -0.064b -0.131a -0.141a -0.035 0.044c -0.027b (0.010) (0.023) (0.043) (0.043) (0.022) (0.022) (0.013) 0.029 -0.073c -0.056 -0.074 0.015 -0.061 -0.017 (0.018) (0.037) (0.069) (0.063) (0.034) (0.042) (0.021) -0.023c -0.070a -0.038c -0.044c -0.025c 0.020 0.022a (0.013) (0.020) (0.021) (0.024) (0.014) (0.013) (0.006) 4.417a 4.930a 5.074a 5.234a 5.677a 6.224a 6.293a (0.013) (0.022) (0.038) (0.050) (0.038) (0.045) (0.022) 0.836 2208

0.887 2208

0.848 2208

0.815 2208

0.815 2208

0.862 1898

0.962 1826

c

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses.

from Westjet at Hamilton to have a different impact than competition from Westjet at Toronto and controlling for the number of other carriers serving the route.30 Table 9 presents results of this analysis. The results show that competition by Westjet from Hamilton has a smaller impact on the fares in the upper portion of the Discount Coach distribution than Westjet’s competition from Pearson. This is consistent with these tickets being purchased by business travellers who may be willing to switch carriers but not commute to an alternative airport to do so. We run a series of additional analyses to confirm the robustness of our findings. In Table 10 we restrict the sample to routes that were served by Canjet or Jetsgo. By doing so, we exploit only changes in market structure that result from these airlines’ full-scale exit from the industry. As argued in Section 4, these types of changes in market structure might be considered more exogeneous that carriers’ selective entry or exit from particular routes. The results when we restrict the sample to these routes are similar to the results from the full sample. We also carry out a number of other robustness checks which we describe here. The results of these checks are available in the online appendix. First, we show that our findings are very similar when we split our sample by routes that involve Toronto and routes that do not. This suggests that the pattern we have uncovered occurs on a wide set of routes. Second, we break up the sample into the periods before and after 2007 and show that the U-shaped pattern emerges in both periods. Third, we add a control for the average size of the planes used by Air Canada on each route, using 30

For this analysis, we limit the sample to routes into or out of Toronto. If Westjet provided service on a given route-month from both Pearson and Hamilton, we code this as service from Pearson, since that is a more relevant and direct measure of competition for Air Canada.

28

Table 10: Regression of Discount Coach Percentiles: Routes served by Canjet or Jetsgo (1) 1 Num. Direct Rivals -0.038a (0.010) Constant 4.527a (0.017) 2 R 0.803 Obs 3314

(2) 25 -0.075a (0.017) 4.818a (0.025) 0.902 3314

(3) 50 -0.069a (0.014) 4.983a (0.035) 0.861 3314

(4) 75 -0.082a (0.017) 5.211a (0.042) 0.816 3314

(5) 99 -0.032a (0.008) 5.700a (0.021) 0.826 3314

(6) Coach 0.010 (0.013) 6.177a (0.033) 0.848 2809

(7) Biz 0.007 (0.006) 6.319a (0.018) 0.960 2363

c

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses. Sample includes routes that were ever served by either Canjet or Jetsgo.

data from OAG, since this variable was identified by Gerardi and Shapiro (2009) as an explanation for the discrepancy between their finding and that of Borenstein and Rose (1994). Our results are robust to including this control. Fourth, our findings are similar if we ignore the direction of travel between airport pairs and estimate our models at the city-pair level. Finally, are results are robust to alternative ways of identifying and excluding the problematic ‘Y’ fare observations.

5.2

Discussion

In Section 2, we developed a simple model of airline price discrimination in which travellers differed in terms of both their underlying value of a trip and their degree of brand loyalty. We distinguished between leisure travellers and two types of business travellers: “brand loyal business travellers” and “brand indifferent business travellers”. We showed that competition would have the largest impact on the fares charged to brand indifferent business travellers and, as a result, would reduce the fare differential between these travellers and leisure travellers but increase the fare differential between them and the brand loyal business travellers. Consistent with existing results in Borenstein (1985), Holmes (1989) and Stole (2007), our simple model illustrated that competition increases price differences between consumers when discrimination is based on differences in cross-price elasticities (or the strength of brand preferences) but decrease price differences between consumers when discrimination is based on differeces in industry-demand elasticites. The results from our empirical analysis appear to be consistent with the predictions of our simple model. Our results suggest that Air Canada offers (at least) three distinct types of tickets which are differentially affected by competition. The fares for Air Canada’s very cheap tickets are hardly impacted by competition, suggesting that these tickets are sold to travellers who are highly price sensitive and who are charged low prices even when Air Canda is a monopolist. The fares of Air Canada’s 29

very expensive tickets are also hardly impacted by competition suggesting that these tickets are sold to travellers with a low industry-demand elasticity and a low crossprice elasticity who can be charged high prices both when Air Canada is a monopolist and when it faces competition. However, the fares of the remainder of Air Canada’s tickets do fall with competition, suggesting the existence of a set of travellers who can be charged high fares under monopoly but not when there is a competing airline in the market. This set of travellers is consistent with the brand indifferent business travellers that we consider in our model. Furthermore, consistent with our model, our results indicate that price differentials between some groups of travellers fall while price differentials between other groups rise. In particular, we find that the differential between Business and Coach class tickets and Discount Coach tickets increases with competition. This suggets that the consumers who buy Business and Coach tickets have a higher level of brand loyalty than other travellers and that airlines are able to segment them along this dimension. It is perhaps not surprising that the fares of Business tickets do not fall with competition given that none of the competing airlines in this setting offer Business class service. But note that this is entirely consistent with the theoretical model. Our model (and the theoretical work on which it is based) predicts that fare differentials will rise with competition if consumers have different degrees of brand loyalty. Brand loyalty arises if consumers perceive competing products to be poor substitutes for each other. In this setting, the degree of substitutability between airlines depends on actual differentiation across airlines’ offerings and perceived differentiation due to frequent flyer programs. The fact that competitors to Air Canada do not offer business class cabins but do offer relatively similar coach class products means that the degree of substitutaiblity between airlines—i.e.: cross-elasticities of demand—will almost surely vary across customers and act as a potential basis for segmentation. Frequent flyer programs, which are likely to be most highly valued by the same types of travellers who prefer Business class tickets, will further contribute to this heterogeneity in brand loyalty. On the other hand, we find that the differential between moderate-to-expensive and inexpensive Discount Coach tickets falls with competition. This suggests the main source of heterogeneity among travellers who purchase more and less expensive Discount Coach tickets is their underlying willingness-to-pay for their trip and not differences in brand loyalty. For example, corporate travel policies that require employees to purchase the cheapest available ticket would create a set of business travellers with a high willingness to pay to travel but effectively no brand loyalty. Alternatively, business travellers who travel infrequently may be less heavily invested in Air Canada’s frequent flier program and may therefore perceive competing offerings as more substitutable. Finally, while our findings are consistent with our simple model of price discrimination, it is worth discussing whether cost-based explanations of price dispersion may play a role as well. Variation in the fares observed on a given route-month may reflect 30

not only price discrimination but also differences in the marginal costs (including the shadow costs) of a seat. While it is certainly the case that the marginal costs of Business and Coach tickets are somewhat higher than Discount Coach tickets—for example, due to the costs of providing greater in-flight service, more frequent flier points or refundability—it is unlikely that these cost differences are large enough to account for the fare differences between these types of tickets. In our data, the average Coach fare is more than twice the average Discount Coach fare and the average Business fare is more than four times the average Discount Coach fare. Given this, our finding that Coach and Business fares do not fall with competition is unlikely to be the result of these high fares simply reflecting higher marginal costs associated with these tickets. A more significant source of cost variation can arise from differences in the shadow cost of capacity. Because capacity is hard to adjust and demand is uncertain, the marginal cost of an airline seat includes a shadow cost—i.e.: the cost of not being able to sell that seat at a later time. Shadow costs will vary across flights in both predictible and unpredictable ways. For example, the shadow cost of capacity will be higher at peak times of the day, peak times of the season and when airline and/or airport resources are scarce. This will result in fare differences across flights on a given route, reflecting the higher expected shadow cost of capacity on certain flights. In addition, the shadow cost of a seat may change over time as demand for a flight is realized. Since airlines can adjust fares as demand is realized, fares will be adjusted to reflect the shadow cost of a seat at the time that the seat is sold. This will result in fare differences across seats on a given flight. It is likely that some of the within route-month fare variation observed in our data reflects differences in the shadow costs of seats. Since we do not have data on specific flights or data on expected or realized load factors, it is not possible for us to directly control for the factors that affect the shadow cost of a seat. Thus, the relevant question becomes whether the results we find could result if the fare variation were the result of differences in shadow costs rather than price discrimination. Specifically, one might worry that competition could lower Air Canada’s load factors, thereby lowering the shadow costs of capacity and reducing fares. We believe this is unlikely to explain the full pattern of results that we find, for a number of reasons. First, because we have data on cabin class, we know that the high fares associated with Coach and Business seats reflect the fact that these are different types of tickets, not just tickets with different shadow costs. Were these high fares partly the result of shadow costs, then we would expect these fares to fall with competition but they do not. Second, we find that competition also has little impact on the most expensive Discount Coach seats. If much of the within-Discount Coach fare variation reflected differences in shadow costs which fall with competition, we would expect the price of the most expensive Discount Coach tickets to fall but we find that these fares are hardly impacted. Third, our findings with respect to the differential effects of competition from Porter in Toronto and Westjet in Hamilton are additional evidence 31

that the U-shaped pattern of fare reductions reflects the impact of competition on travellers different cross-elasticities of demand. Finally, there is mixed evidence as to the general importance of scarcity based explanations for fare variation. Puller et al. (2012)—using a detailed dataset of fares, ticket characteristics and load factor measures—find that the majority of fare variation is explained by ticket characteristics and that scarcity-based explanations have virtually no effect on fare dispersion. On the other hand, Williams (2013)—using a dataset of fares, bookings and remaining capacity—finds that fares vary significantly with remaining capacity on a flight.

6

Conclusions

In this paper, we have revisited the relationship between market structure and price discrimination in the airline industry. Previous empirical work has examined how competition impacts fare dispersion and uncovered conflicting findings. Borenstein and Rose (1994) found that competition increased fare dispersion while Gerardi and Shapiro (2009) found the opposite. Although the Gerardi and Shapiro study utilizes a longer dataset and more credible identification strategy, the theoretical literature makes it clear that both relationships are possible. Indeed, the literature shows that the effect of competition on price differences across consumers depends on whether segmentation is based on differences in consumers’ industry-demand elasticities or cross-price elasticities. Buildling on this basic insight, we developed a simple model with three types of travellers. Our model allowed travellers to differ in terms of both their underlying value of a trip and their degree of brand loyalty and, further, allowed travellers with a high value of travel to differ in terms of their brand loyalty. Our model suggests that competition may have the largest impact on the fares charged to travellers who have a high underlying value of completing their trip but little airline loyalty, whom we call “brand indifferent business travellers”. Because these travellers’ fares fall by more than those of other types of travellers, competition reduces the fare differential between some types of tickets while increasing the differential between others. Our empirical analysis estimated how changes in market structure on routes served by Air Canada affected the airline’s fares for various different types of tickets. The results showed that competition has little impact on Air Canada’s very cheap fares or its very expensive fares, including its Coach and Business Class fares. On the other hand, competition leads to a 7-8% reduction in fares of tickets in the middle of the Discount Coach distribution. Overall, we find a U-shaped relationship between competition and fare reductions over the fare distribution. The relative fare effects imply that competition increases fare differentials between Air Canada’s Coach and Business tickets and Discount Coach tickets but reduces fare differentials within Disocunt Coach. The pattern of results we find is consistent with our model. Our theoretical model, as well as our empirical findings, are consistent with both the previous papers in this literature and show that the findings in both Borenstein 32

and Rose (1994) and Gerardi and Shapiro (2009) may operate simultaneously. In this way, this paper helps reconcile the conflicting findings and arguments in these papers. More generally, the paper highlights the fact that, in non-monopoly settings, the impact of competition on price discrimination will depend on whether price discrimination is based on differences in industry elasticities or cross elasticities or both and that measuring this relationship requires a nuanced understanding of the sources of consumer heterogeneity in an industry.

References Borenstein, S. (1985). Price discrimination in free-entry markets. The RAND Journal of Economics, 380–397. Borenstein, S. (1989). Hubs and high fares: dominance and market power in the US airline industry. The RAND Journal of Economics, 344–365. Borenstein, S. (1991). The dominant-firm advantage in multiproduct industries: evidence from the us airlines. The Quarterly Journal of Economics, 1237–1266. Borenstein, S. and N. L. Rose (1994). Competition and Price Dispersion in the US Airline Industry. Journal of Political Economy 102 (4), 653–683. Chandra, A. and M. Lederman (2014a). The Airline Industry in Canada. Working Paper . Chandra, A. and M. Lederman (2014b). The Effects of Porter Airlines’ Expansion. Working Paper . Dai, M., Q. Liu, and K. Serfes (2014). Is the Effect of Competition on Price Dispersion Nonmonotonic? Evidence from the US Airline Industry. Review of Economics and Statistics 96 (1), 161–170. Gaggero, A. A. and C. A. Piga (2011). Airline market power and intertemporal price dispersion. The Journal of Industrial Economics 59 (4), 552–577. Gerardi, K. S. and A. H. Shapiro (2009). Does competition reduce price dispersion? new evidence from the airline industry. Journal of Political Economy 117 (1), 1–37. Hernandez, M. A. and S. N. Wiggins (2014). Nonlinear Pricing Strategies and Competitive Conditions in the Airline Industry. Economic Inquiry 52 (2), 539–561. Holmes, T. J. (1989). The effects of third-degree price discrimination in oligopoly. The American Economic Review , 244–250. Kim, M. (2015). Market Structure, Competition, and Price Dispersion in the Airline Industry Revisited. Technical report. 33

Lederman, M. (2007). Do enhancements to loyalty programs affect demand? The impact of international frequent flyer partnerships on domestic airline demand. The RAND Journal of Economics 38 (4), 1134–1158. Lederman, M. (2008). Are Frequent-Flyer Programs a Cause of the “Hub Premium”? Journal of Economics & Management Strategy 17 (1), 35–66. Puller, S. L., A. Sengupta, and S. N. Wiggins (2012). Does scarcity drive intra-route price dispersion in airlines? NBER Working Paper 15555. Sengupta, A. and S. N. Wiggins (2014). Airline Pricing, Price Dispersion, and Ticket Characteristics On and Off the Internet. American Economic Journal: Economic Policy 6 (1), 272–307. Stavins, J. (2001). Price discrimination in the airline market: The effect of market concentration. Review of Economics and Statistics 83 (1), 200–202. Stole, L. A. (2007). Price discrimination and competition. Handbook of industrial organization 3, 2221–2299. Williams, K. R. (2013). Dynamic airline pricing and seat availability. Technical report, mimeo. Zhang, L., H. Li, Z. Du, and F. Wei (2015). Market Competition and Price Dispersion: Evidence from China’s Airline Market. Technical report.

34

A

Appendix A: Regressions of Fare Ratios Table 11: Regression of Fare Ratios (1) 50:1

Num. Direct Rivals Constant R2 Obs

(2) 50:5

(3) 75:1

(4) 75:5

-0.077a -0.062a -0.131a -0.109a (0.021) (0.016) (0.046) (0.041) 1.799a 1.751a 2.451a 2.381a (0.055) (0.050) (0.120) (0.116) 0.337 7308

0.295 7308

0.372 7308

0.360 7308

(5) Coach:Disc

(6) Biz:Disc

0.217a (0.052) 3.058a (0.084)

0.151a (0.036) 3.723a (0.081)

0.637 5688

0.728 4203

c

p < 0.1, b p < 0.05, a p < 0.01. All regressions include route, month and year FEs. Standard errors, clustered by route, in parentheses.

B

Appendix B: Booking Codes on Air Canada

The document below shows how various booking codes, and the class of service that they imply, are associated with different characteristics on Air Canada. The document was posted by an Air Canada representative on a popular online forum on air travel. Executive class, generally associated with the J and C booking codes, is equivalent to business class service on other airlines. Latitude class tickets do not entitle the traveler to a special cabin, but do provide many of the benefits associated with business travel, such as free changes to the itinerary, lounge access, and a free meal. Tango Plus tickets are distinguished from those in Tango in the following ways: full credit for miles flown; free seat selection; changes to the itinerary on the day of travel for a low fee; and the ability to upgrade to Executive class in exchange for miles.

35

North American Fare Structure TANGO K, N, G, P, T, E

TANGO PLUS M, U, H, Q, V, W, S, L

LATITUDE Y, B

EXECUTIVE CLASS® LOWEST D, Z

EXECUTIVE CLASS® FLEXIBLE J, C

Changes

$ 75 + difference in fare

Difference in fare may apply

$ 150 $ 75 on Rapidair routes n/a

Complimentary

$ 50 Canada, $ 75 Transborder + difference in fare $ 75

Difference in fare may apply

Same Day Change Upon Check-in

$ 50 Canada, $ 75 Transborder + difference in fare $ 75

Complimentary

Available only on Rapidair

Available

Available

Available

Same Day Airport Standby Refunds

Non-Refundable

Non-Refundable

Refundable

Non-Refundable

Refundable

Advance Seat Selection

$15, $17, $22¹ (Optional)

Complimentary

Complimentary

Complimentary

Complimentary

Maple Leaf TM Lounge Access

$ 45

$ 35

$ 30

Yes

Yes

Onboard Café

Prepay $7 for $9 value, at aircanada.com/agents.

Complimentary

Complimentary Executive Class meal

Complimentary Executive Class meal

Aeroplan® Accumulation

25 % Aeroplan Miles

100 % Air Canada Status Miles

100 % Air Canada Status Miles

150 % Air Canada Status Miles

150 % Air Canada Status Miles

Air Canada Top Tier Upgrade Certificates

n/a

As per the terms and condition on the certificates

As per the terms and condition on the certificates

n/a

n/a

Priority Service Check-in, Bags, Boarding

No

No

At airports in Canada, where available

Yes

Yes

On My Way™

$ 25: up to 1,000 miles $ 35: 1,000 + miles

$ 25: up to 1,000 miles $ 35: 1,000 + miles

$ 25: up to 1,000 miles $ 35: 1,000 + miles

$ 25: up to 1,000 miles $ 35: 1,000 + miles

$ 25: up to 1,000 miles $ 35: 1,000 + miles

¹ $15, 0- 350 miles, $17, 351 – 1000 miles, $22, 1001 + miles. This is a summary of the fare attributes for travel within North America when purchased on the Air Canada website. ® ® TM Aeroplan is a Registered Trademark of Aeroplan LP. Executive Class is a Registered Trademark of Air Canada. Maple Leaf is Trademark of Air Canada. ™On My Way is a Trademark of Air Canada. Information subject to change without prior notice. Sales Communication, Updated Nov. 24, 2009.