GERMAN ECONOMIC ASSOCIATION OF BUSINESS ADMINISTRATION GEABA X. SYMPOSIUM ZUR ÖKONOMISCHEN ANALYSE. Another look at commercial airport services

Transcription

1 X. SYMPOSIUM ZUR ÖKONOMISCHEN ANALYSE DER UNTERNEHMUNG Another look at commercial airport services Achim I. Czerny Session A1 GERMAN ECONOMIC ASSOCIATION OF BUSINESS ADMINISTRATION GEABA

2 Another look at commercial airport services Achim I. Czerny WHU Otto Beisheim School of Management Vallendar, July 11, 2009 Abstract I consider a model of a private, congested airport that provides aeronautical services to airlines and concessions to commercial service providers. Some concessionaires are in competition with shops outside the airport area (e.g., retailers). For other concessionaires this competition is limited (e.g., car rental). Moreover, the overall consumption of commercial services may be independent of traveling decisions. I show that, in this context, the effect of commercial services on aeronautical service charges is ambiguous and that, in welfare optimum, marginal congestion costs are fully internalized. These results are in contrast to the ones obtained earlier by other economists. JEL Classification: D42, D62, H42, L12, L51, L93, R41, R48. Keywords: Airports, privatization, congestion, aeronautical services, commercial services.

3 1 Introduction A growing number of airports are fully or partially privatized, and airport privatization is almost always accompanied by some form of price regulation because many airports are supposed to possess monopolistic market power in the area of aeronautical services, which includes the supply of runway, terminal, gangway, and parking capacity for aircraft. The benefits of regulation seem to be straightforward in this situation, but the value of the textbook result on monopolies is limited in the case of airports. One reason is that airports are frequently congested and that greater levels of aeronautical service charges are useful in limiting congestion. Another reason is that airport business areas include aeronautical and commercial services. Commercial services include retailing, advertising, car rental, car parking, and land rents. Today, the share of revenues from commercial services among airports worldwide has reached an average level of roughly 50% (ACI, 2008). 1 There are two relevant differences between commercial airport services on which I concentrate. Consider retail and car rental services as examples. First, the demand for retail services is often independent of travel decisions. For instance, the decision to consume food and beverages does not depend on whether individuals will fly or not. If individuals fly, they may consume food and beverages at the airport. However, if they do not fly, they will also consume food and beverages but outside the airport; consequently, the consumption of food and beverages is independent of flying. This is different in the case of car rental services because the decision to rent a car may only be relevant in the case of flying. Second, in the case of retail services, shops inside the airport area are in competition with shops outside the airport area. This is because individuals who fly will not buy, e.g., clothes or other 1 (Graham, 2009) discusses the importance of commercial revenues to today s airports. 2

4 products at the airport if prices are above the market level. In contrast, competition between car rental service providers in- and outside the airport area is limited. To cover the different market environments of commercial services, I consider a theoretical model of a congested airport that provides aeronautical services and concessions to retail and car rental service providers. 2 The purpose is to explore the behavior of a private, unregulated airport and to test whether airport regulation is useful in this specific context. Note that the regulators objectives can differ in reality. Some regulators may focus on total surplus and others on consumer surplus; I consider both objectives for this reason and find the following. To maximize consumer surplus, aeronautical airport services and airport concessions to car rental service providers should be provided for free. 3 This demonstrates that congestion pricing is against the interests of a consumer surplus-maximizing regulator and provides a straightforward explanation for the lack of serious congestion pricing in the air transport industry. To maximize total surplus, concessions to car rental service providers should also be provided for free, but aeronautical service charges should internalize marginal congestion costs. I also find that the supply of commercial airport services can reduce the benefits of regulation. This is because both car rental and retailing services can imply downward pressure on aeronautical service charges in the case of a private, unregulated airport. However, the effect of car rental services on aeronautical service charges is ambiguous in this situation because they can also have a positive effect on aeronautical service charges. Furthermore, car rental service charges themselves can be excessive in the 2 Retail and car rental services may also represent other service areas, e.g., car parking services or advertising. 3 This is based on the assumption that variable airport costs are zero. 3

5 case of private airports. Overall, commercial airport services appear to be no substitute for regulation. These results are partly consistent with earlier results obtained by other economists, but there are also differences. (Starkie, 2001), (Zhang and Zhang, 2003), (Oum et al., 2004), and (Starkie, 2008) already found that, in the case of profit maximization, commercial airport services can create downward pressure on aeronautical service charges; this is consistent with my results. In addition, (Zhang and Zhang, 2003) and (Oum et al., 2004) found that total surplus-maximizing aeronautical service charges should not fully internalize marginal congestion costs to increase the surplus generated in the commercial business areas; this is in contrast to my results. Furthermore, (Zhang and Zhang, 2003) found that aeronautical service charges of a private, unregulated airport are always greater than total surplus-maximizing aeronautical service charges; this is another result that is in conflict with my results. Note that (Zhang and Zhang, 2003) consider a single, monopolistic commercial airport service; hence, they made no distinction between the different commercial business areas and their specific competitive environments, and this is the reason for the discrepancy between their results and the results presented in my paper. In particular, they did not take into account that (i) the demand for commercial services may be independent of traveling, (ii) commercial service providers inside the airport area may be in competition with shops outside the airport area, and (iii) service charges in the commercial business areas may affect passenger demand. (Czerny, 2006) considers a highly stylized model of an uncongested airport that provides aeronautical and commercial services. He found that commercial services increase aeronautical service charges in the case of private, unregulated airports, as is consistent with the results presented in this paper. 4

6 However, he ignores the competitive environment of retailing markets and underestimates the downward pressure of retailing services on aeronautical service charges in the case of private, unregulated airports. In this paper, I extend the model of (Czerny, 2006) such that both congestion and car rental and retailing services are elements used to run numerical simulations. The contributions of this paper are as follows. It provides an airport model that is general with regard to demand and average congestion costs functions 4 and that captures relevant differences between commercial airport business areas and their specific competitive environments, which have been ignored by other economists. Hence, this airport model provides a more realistic picture of the relationship between aeronautical and commercial business areas. It considers a regulator focused on total surplus or on consumer surplus (all economists mentioned above consider a total surplus-oriented regulatoe only). It also provides a highly stylized model of a congested airport that provides car rental and retailing services, and this model can be used to run numerical simulations. This paper is organized as follows. Section 2 introduces the general airport model. Section 3 explores the benefits of airport regulation in a context where commercial services are relevant. Section 4 presents a highly stylized airport model and the results of numerical simulations. Section 5 provides the concluding remarks. 2 The general model There is a private, monopoly airport that provides aeronautical services to airlines and concessions and shopping areas to commercial service providers. 4 I consider general downward sloping demand functions and a general upward sloping average congestion costs function. 5

7 This airport captures all revenues from commercial services via take-it-orleave-it offers to concessionaires. Marginal costs of the airport and concessionaires are constant and normalized to zero. Commercial airport services are of two types including car rental and retail services, and there is a relevant difference with regard to market structure between these types of services. Car rental services are essential for certain passengers, especially for business passengers, to limit total travel time, and these passengers need to rent cars directly inside the airport area. For this reason, I take the extreme assumption that there is no competition between car-rental service providers in- and outside the airport. The airport charges car rental service providers with a price p 2 0 per customer, which also determines the price charged to customers by concessionaires because concessionaires are in competition. In contrast, passengers have no preference to shop inside the airport area, so that retail service providers inside the airport area are under circumstances of competition with shops outside the airport area. One consequence is that overall consumption of retail services does not depend on whether individuals fly or not because individuals who do not fly buy retail services outside the airport. Another consequence is that the airport earns a given profit margin p 0 per customer from concessions in the retail business area. Note that I choose extreme market conditions no competition between car rental service providers in- and outside the airport and a given profit margin in the case of retailing to clearly identify the role of market structure in commercial airport businesses. I focus on car rental and retail services, but recall that these two types of services may also represent other commercial airport business areas where the airport may possess market power or is in competition with firms outside the airport area (see Footnote 1). 6

8 Runway and terminal capacity is limited so that flying creates congestion and congestion costs. The number of passengers (i.e., the number of individuals who fly) is denoted by q 1 0. Average congestion costs, denoted by Γ 0, depend on passenger numbers, q 1, and they are convex, i.e., Γ/ q 1 > 0 and 2 Γ/ q Total congestion costs are q 1 Γ, and marginal congestion costs are (q 1 Γ) q 1 = Γ + q 1 Γ q 1. (1) Without loss of generality, we can say that congestion costs are borne by airlines. Airlines exist under circumstances of perfect competition and consider average congestion costs as constant and given. 5 Other marginal airline costs are constant and normalized to zero. Under these conditions, the airfare is equal to the sum of the aeronautical service charge, p 1, and average congestion costs, Γ. In this context, the internalization of marginal congestion costs in (1) to airlines and passengers depends on p 1. For instance, marginal congestion costs are partly internalized if p 1 = 0 and fully internalized if p 1 = q 1 Γ/ q 1. Denote passenger demand by D 1 (p 1 + Γ, p 2 ) > 0 with D 1 / p 1 < 0 and the demand for car rental services by D 2 (p 1 + Γ, p 2 ) > 0 with D 2 / p 2 < 0. Passenger demand and the demand for commercial airport services are heavily interdependent. The model employs the following four features to capture the demand interdependency. (i) There is a negative relationship between the demand for car rental services, D 2, and aeronautical service charges, p 1, that implies D 2 / p 1 < 0. This is because aeronautical service charges determine 5 There is debate about the adequacy of the concept of given average congestion costs in the context of airlines because airline markets are oligopolistic (see, e.g., (Brueckner and van Dender, 2008), (Daniel and Harback, 2008), and (Brueckner, 2009)). However, I adhere to given average congestion costs in this paper. This is to abstract away the airline market-structure issue so as to focus on the differences between commercial airport services with regard to market structure. 7

9 passenger demand and the demand for car rental services depends on passenger numbers. (ii) Only passengers buy commercial airport services (i.e., individuals do not travel to the airport only for shopping reasons); hence, D 1 = 0 D 2 = 0 holds true. (iii) There is a negative relationship between car rental service charges, p 2, and passenger demand, D 1, that implies D 1 / p 2 < 0. The following example provides an economic rationale for this relationship. Consider a business traveler who achieves profit π > 0 by flying and renting a car. Without car rental services, the business traveler uses public transport services, which increases total travel costs by amount κ > π (p 1 +Γ). In this situation, the business traveler flies if and only if car rental service charges, p 2, are low enough. As a consequence, there are altogether two reasons for the negative relationship between car rental service charges and the demand for car rental services. One is that an increase in car rental service charges reduces the demand for car rental services for a given number of passengers. The other is that an increase in car rental service charges reduces the passenger number due to D 1 / p 2. Both reasons are captured here. (iv) All passengers buy one unit of retailing services, which implies that, inside the airport, the demand for retailing services is equal to D 1. However, passenger demand, D 1, is independent of retailing services because individuals can buy these services outside the airport area. Airport profits are Π = (p 1 + p) D 1 + p 2 D 2 (2) and consumer surplus is CS = (, ) (p 1 +Γ,p 2 ) ( 2 ) D i (x 1, x 2 ) dx i. (3) i=1 8

10 Note that, in this context, the integrability condition is satisfied because income effects are zero (Crew and Kleindorfer, 1979). Therefore, the solution of the line integral in (3) is independent of the particular path along which integration is taken. One way to calculate consumer surplus is CS = D 2 (p 1, x 2 ) dx 2 + p 2 p 1 +Γ D 1 (x 1, ) dx 1. (4) Total surplus is CS + p 1 D 1 + p 2 D 2. (5) Observe that both consumer surplus in (4) and total surplus in (5) ignore retailing services. This is because shopping opportunities exist outside the airport and the individuals decision to buy retail services is independent of their travel decision. The regulator s objective function is Ψ = CS + β (p 1 D 1 + p 2 D 2 ) (6) with β {0, 1}. 6 If β = 1, the total surplus in (5) is relevant from the regulator s viewpoint (I call this the total-surplus case), and if β = 0, it is the consumer surplus in (4) only (I call this the consumer-surplus case). 6 The objective function in (6) is similar to the one considered by (Baron and Myerson, 1982), except that I do not include subsidies or taxes and that I consider a discrete variation of β {0, 1} (not β [0, 1]). 9

11 3 Optimal airport behavior from the regulator s viewpoint versus the behavior of a private, unregulated airport To identify whether airport regulation can be useful in a context where commercial services are relevant, I explore airport charges, p 1 and p 2, that are optimal from the regulator s viewpoint and the behavior of a private, unregulated airport. Then I compare the results. Optimal airport charges from the regulator s viewpoint are (p 1, p 2) = arg max p 1,p 2 Ψ s.t. p 1, p 2 0. (7) Differentiating the regulator s objective function in (6) with regard to p 1 and p 2 gives the first-order conditions for optimal airport charges in (7) given by Ψ p 1 = p 2 +β D 2 (p ( 1, x 2 ) dx 2 D 1 (p 1 + Γ, ) 1 + Γ ) p 1 p 1 ( D 1 + p D p 2 p 1 ) D 2 + λ 1 = 0 (8) p 1 and ( Ψ = D 2 + β p D D 2 + p 2 p 2 p 2 ) D 2 + λ 2 = 0 (9) p 2 where λ 1 and λ 2 are Lagrange multipliers attached to the non-negativity constraints for p 1 or, respectively, p 2. The first-order conditions in (8) and (9) imply the following. Proposition 1 For a congested, monopoly airport that offers aeronautical, car rental, and retailing services, the following holds true from the regula- 10

12 tor s viewpoint. In the consumer-surplus case (β = 0), aeronautical services charges, p 1, should be 0 (i.e., p 1 = 0). In the total-surplus case (β = 1), aeronautical service charges, p 1, should be chosen such that airfares, p 1 + Γ, fully internalize marginal congestion costs (i.e., p 1 = q 1 Γ/ q 1 ). Airport concessions for car rental services should be provided for free (i.e., p 2 = 0). Proof See Appendix A. The results described in Proposition 1 are in opposition to the findings of (Zhang and Zhang, 2003). They also consider a congested, monopoly airport that offers commercial services but find that total surplus-maximizing aeronautical service charges should not fully internalize marginal congestion costs. In their model, this is to increase the surplus generated in commercial business areas. Zhang and Zhang assume that the airport is a monopoly provider of commercial airport services but that commercial services do not affect flight decisions. Hence, they ignore that (i) the consumption of commercial services can be independent of traveling, (ii) shops in- and outside the airport area may compete, and (iii) a positive effect of commercial services (e.g., car rental services) on passenger demand exists. Aeronautical services and concessions for car rental services should be provided for free in the consumer-surplus case, which provides a straightforward explanation for the lack of congestion pricing in reality. Note that fixed cost recovery is possible in this situation because profits from retailing services, p D 1 (0 + Γ, 0), exist. In the total-surplus case, concessions for car rental service providers should also be provided for free, but aeronautical service charges, p 1, are positive. However, it is unclear whether the change in aeronautical service charges will increase airport profits. This is because the increase in aeronautical service charges reduces passenger numbers and 11

13 revenues in the retailing business area. What about the pricing behavior of a private, unregulated airport? The pricing behavior of a private, unregulated airport is (p M 1, p M 2 ) = arg max p 1,p 2 Π s.t. p 1, p 2 0, (10) and differentiating airport profits in (2) with regard to aeronautical service charges, p 1, and car rental service charges, p 2, gives the first-order conditions and Π p 1 = D 1 + (p M 1 + p) D 1 p 1 + p M 2 Π p 2 = (p M 1 + p) D 1 p 2 + D 2 + p M 2 D 2 p 1 + λ M 1 = 0 (11) D 2 p 2 + λ M 2 = 0. (12) Observe that aeronautical service charges of a private, unregulated airport depend on the profit margins in the retailing business area p, which is in contrast to optimal airport charges from the regulator s viewpoint in (7). The first-order condition in (11) and (12) imply: Proposition 2 For a congested, private and unregulated airport, the following holds true. Without commercial airport services, aeronautical service charges are excessive from the regulator s viewpoint (i.e., p M 1 > p 1 holds true in this case). With car rental and retailing services, aeronautical service charges can be optimal from the regulator s viewpoint (i.e., parameter constellations that lead to p M 1 = p 1 exist). The effect of car rental services on aeronautical service charges is ambiguous, but the effect of retailing services is clear-cut: they reduce aeronautical and car rental service charges. Proof I start with considering a congested airport that does not provide commercial services and then include commercial airport services. 12

14 Denote the price elasticity of demand for aeronautical and car rental services as with i {1, 2} and x {, M}. η x i = px i D i D i p i (13) If commercial airport services are not provided, a private, unregulated airport charges a price, p 1, for aeronautical services that implies η M 1 = 1. In the total-surplus case, multiplying optimal aeronautical service charges in (29) by D 1 / p 1 and rearranging yields η 1 = Γ p 1. (14) Recall that 1 + Γ/ p 1 > 0 holds true (see the proof of Proposition 1 in Appendix A), which is equivalent to Γ/ p 1 > 1; this shows that profitmaximizing aeronautical service charges are excessive in the total-surplus case and, consequently, in the consumer-surplus case as well, with p 1 = 0. The picture changes when commercial airport services are taken into account. Rearranging the first-order condition for aeronautical service charges in (11) leads to η M 1 ( p M = 1 2 D 2 + p D ) 1 D 1 p 1 D 1 p }{{ 1 } > 0 > 1, (15) which is not in contradiction to p M 1 = p 1; hence, in a context where commercial services are relevant, parameter constellations that imply p M 1 = p 1 may exist. The elasticity in (15) demonstrates that car rental services create downward pressure on aeronautical service charges (the price elasticity of demand is greater with car rental services). However, the total effect of car rental services on aeronautical service charges is ambiguous. This is be- 13

15 cause the supply of car rental services increases passenger demand, which follows from D 1 / p 2 < 0. Therefore, it is possible that car rental services increase aeronautical service charges, p M 1, compared to a situation in which car rental services are not provided. p M 1 p2 =p M 2 Hence, a constellation that implies > p M 1 p2 = and η M 1 > 1 may exist. In contrast, the supply of retailing services always reduces aeronautical service charges p M 1 (the price elasticity of demand is greater with retailing services) because they do not affect passenger demand (recall that D 1 / p = 0). Finally, rearranging the first-order condition in (12) leads to η2 M = 1 pm 1 + p D 1, (16) D 2 p }{{ 2 } > 0 which shows that retailing services reduce car rental service charges, p M 2. I can draw the following conclusions with regard to the benefits of airport regulation due to Proposition 2. Aeronautical service charges of a private, unregulated airport are excessive from the regulator s point of view; hence, regulation is useful in this situation. This has also been found by (Basso, 2008) for the linear case. However, the supply of commercial airport services can effectively reduce aeronautical service charges of a private, unregulated airport and, as a consequence, they may also reduce the need for regulation. This is especially true for retailing services, which always reduce aeronautical and car rental service charges of a private, unregulated airport. In contrast, the effect of car rental services on regulation is ambiguous. First, their effect on the aeronautical service charges of a private, unregulated airport is ambiguous. Second, commercial service charges themselves can be excessive and therefore can create an additional rationale for airport regulation. These re- 14

16 sults reproduce some results obtained earlier by other economists, but there are also important differences. (Starkie, 2001), (Zhang and Zhang, 2003), (Oum et al., 2004), and (Starkie, 2008) already found that, in the case of profit maximization, commercial airport services can create downward pressure on aeronautical service charges; this is consistent with my findings in this paper. However, they ignore the positive effect of commercial airport services on passenger demand and, in the case of a private, unregulated airport, the possible positive relationship between commercial services and aeronautical service charges. Another difference between my results and the results obtained by (Zhang and Zhang, 2003) is the following. Zhang and Zhang find that aeronautical service charges of a private, unregulated airport are always higher than are total surplus-maximizing ones, but this does not hold for my model. Recall that the supply of retailing services by airports does not contribute to consumer or total surplus here because the demand for retailing is independent of travel decisions. Therefore, optimal aeronautical service charges are independent of retailing services from the regulator s viewpoint but not from the airport s viewpoint, and this is what causes the difference. (Czerny, 2006) considers a stylized model of an uncongested airport that provides aeronautical and commercial services. He finds that commercial services increase aeronautical service charges in the case of a private, unregulated airport, as is consistent with the effects of car rental services on aeronautical service charges found in this paper. However, he ignores the competitive environment of retailing markets and underestimates the downward pressure of retailing services on aeronautical service charges. In the next section, I extend the context employed by (Czerny, 2006) so that both congestion and car rental and retailing services are included. Then, 15

17 I use the extended setting to run numerical simulations that illustrate the effects of commercial airport activities on airport charges. The simulations are also used to further clarify the benefits of airport regulation. 4 The highly stylized airport model and numerical simulations The numerical simulations are based on the following highly stylized model. There is a set of individuals with mass one. Every individual flies at most once and buys at most one unit of car rental and one unit of retailing services. The individual benefit of flying is V 1 [0, 1], and the individual benefit of car rental services is V 2 [0, 1]. Individuals are uniformly and independently distributed over the V 1 -V 2 -space, which is equal to the unit square. Figure 1 illustrates this case. For simplicity s sake, average congestion costs are determined according to passenger numbers such that Γ = q 1, which implies convex total congestion costs q 1 Γ = q1. 2 The consumer surplus of a passenger who flies and rents a car is V 1 + V 2 (p 1 + Γ + p 2 ). (17) Passengers rent a car if and only if V 2 p 2, and individuals fly if and only if V 1 + max{0, V 2 p 2 } (p 1 + Γ) 0. (18) Observe that retailing services are not relevant for the flight decision ( p does not enter the left hand side of the inequality in (18)) because shopping deci- 16

18 V A p 1 +Γ B C E D 0 V 2 0 p Figure 1: Passenger demand and demand for commercial services for airport charges p 1 = p 2 = 0.2 (Γ 0.56 in this instance). Passenger demand and the demand for retailing services is A + B + C. The demand for car rental services is B + C. sions are independent of traveling. I derive the demands for airport services based on the passenger s consumer surplus in (17) and the inequality in (18). Figure 1 illustrates the demand for aeronautical and car rental services for airport charges p 1 = p 2 = 0.2, which imply average congestion costs, Γ, approximately equal to Passenger demand is D 1 = A + B + C (19) for the following reasons. Individuals in areas A and B fly, because the benefit that they receive by flying, V 1, is greater than the airfare, p 1 +Γ. Observe that individuals in area C obtain great benefits from car rental services (V 2 > p 2 in this area) and that they fly because the benefits of aeronautical and car rental services together exceed the total airfare. Furthermore, note that individuals in area C would not fly without car rental services because V 1 < p 1. Consider 17

19 the example of the business traveler that I already introduced in Section 2. This business traveler achieves profit π > 0 by flying and renting a car. Not renting a car would cause extra costs κ. In this situation, V 1 = π κ and V 2 = κ. Hence, if V 1 < p 1 + Γ and V 1 + V 2 p 1 + p 2 + Γ, the business passenger flies, since π p 1 + p 2 + Γ. Finally, individuals in areas E and D do not fly, since V 1 < p 1 and V 1 + V 2 < p 1 + p 2 holds in these areas. To calculate passenger demand as a function of airport charges p 1 and p 2, I distinguish three cases: p 2 1 (which implies zero demand for car rental services), p 1 + p 2 + Γ > 1 and p 2 < 1, and p 1 + p 2 + Γ 1. Passenger demand in (20) is determined by areas A, B, and C, which implies 7 D 1 = max{0, 1 (p 1 + Γ)} for p 2 1 { max 0, 3 2 (p } 1 + p 2 + Γ) + p for p 1 + p 2 + Γ > 1, p 2 < 1 1 (p 1 + Γ) p 2 (p 1 + Γ) 2 2 for p 1 + p 2 + Γ 1. (20) Substituting q 1 for Γ and D 1 in (19) and solving for q 1 gives average congestion costs as a function of airport charges p 1 and p 2, which are Γ = { max 0, 1 p 1 2 } for p 2 1 { max 0, 3 2 (p } 1 + p 2 ) + p (p1 + p 2 ) + p 2 2 (1 + p 1 + p 2 ) for p 1 + p 2 + Γ 1. for p 1 + p 2 + Γ > 1, p 2 < 1 The demand for car rental services is (21) D 2 = B + C D 1. (22) 7 (Czerny, 2006) derives demand expressions for the case of a non-congested, monopoly airport. 18

20 Individuals in area A do fly, but they do not rent a car because V 2 < p 2. Individuals in area D do not rent a car because they do not fly although V 2 > p 2 holds true. This is because the demand for car rental services depend on travel decisions. Individuals in area E do not fly and do not rent a car. The demand for car rental services is determined by areas B and C, which implies D 2 = max { 0, (1 p 2)(3 2(p 1 + Γ) p 2 ) 2 1 p 2 (p 1 + Γ) } for p 2 1 for p 1 + p 2 + Γ > 1, p 2 < 1 for p 1 + p 2 + Γ 1. (23) Furthermore, the demand for retailing services is simply equal to passenger demand, D 1, in (20). Substituting passenger demand in (20) and the demand for car rental services in (23) for D 1 and D 2 in (4) gives the consumer surplus as a function of airport charges p 1 and p 2, which is CS = 1 (p 1 +Γ) min{p 2,1 (p 1 +Γ)} 1 y (p 1 + Γ) 2 2 dy + 1 max{p 2,1 (p 1 +Γ)} (1 y) (3 2 (p 1 + Γ) y) 2 dy + 1 p 1 1 (x + Γ) dx, (24) for p 1, p 2 1. With passenger demand in (20), average congestion costs in (21), car rental service demand in (23), and consumer surplus in (24) one can also calculate airport profit, Π, total surplus, and the regulator s objective function, Ψ, as functions of airport service charges p 1 and p 2. 19

21 0.6 p 1 * 0.4 M p p M p Figure 2: Aeronautical and car rental service charges in the totalsurplus case (β = 1), p 1 = 1/2 and p 2 = 0, and in the case of a private, unregulated airport, p M 1 and p M 2, depending on the profit margin in the retail business area, p [0, 1/2]. I know that p 1 = p 2 = 0 holds in the consumer-surplus case (β = 0) and that p 2 = 0 also holds in the total-surplus case (β = 1) due to Proposition 1. Furthermore, p 1 = 1/2 holds true for the total-surplus case in this instance. What about the charges of a private, unregulated airport in this highly stylized setting? Figure 2 depicts airport charges in the total-surplus case (β = 1), p 1 = 1/2 and p 2 = 0, and in the case of a private, unregulated airport, p M 1 and p M 2, depending on the profit margin in the retail business area, p [0, 1/2]. The figure demonstrates that there is a negative relationship between aeronautical service charges, p M 1, and profit margins in the retail business area, p. The same holds true for the relationship between car rental service charges, p M 2, and p. The figure also demonstrates that compared to optimal airport charges in the total-surplus case, aeronautical service charges of a private, unregulated airport can be excessive or too low, and that car rental service charges 20

22 of a private, unregulated airport are excessive for all p [0, 1/2]. Observe that, in the consumer-surplus case, both aeronautical and car rental service charges of a private, unregulated airport are excessive for all p [0, 1/2]. Recall that car rental services can raise aeronautical service charges in the case of a private, unregulated airport due to Proposition 2. Using the highly stylized model introduced in this section, I illustrate this result. Equations (20) and (21) imply passenger demand (1 p 1 )/2 for all p 2 1 (car rental services are not relevant in this case). Aeronautical service charges of a private, unregulated airport are equal to 1/2 in this situation (without car rental services). On the other hand, Figure 2 shows that aeronautical service charges of a private, unregulated airport can be greater than 1/2 for low values of p due to the supply of car rental services or because p M 2 < 1. Overall, the simulation results confirm that commercial airport services are not a substitute for regulation and that commercial airport services, such as car rental services, create an additional rationale for regulation. 5 Conclusions In this paper, I considered a theoretical model of a congested airport that provides aeronautical services and concessions to commercial service providers. There is a strong dependency between the demand for aeronautical and commercial airport services. Furthermore, the market environments of commercial airport services are different, which has been ignored by economists so far. For instance, the competition between car rental service providers inand outside the airport area is limited, but there is a significant amount of competition between retailers in- and outside the airport area. Furthermore, the demand for overall commercial services can be independent of travel- 21

23 ing decisions. The model I considered here is built to capture the demand interdependencies and the different market environments in detail. My core results were the following. In the case of consumer surplus maximization, aeronautical services and concessions for car rental services should be provided for free (recall that constant marginal airport costs were normalized to zero). In the case of total surplus maximization, concessions to car rental service providers should also be provided for free, but aeronautical service charges should fully internalize marginal congestion costs. Furthermore, I found that the supply of commercial airport services can reduce the benefits of regulation. This is because both car rental and retailing services can entail downward pressure on aeronautical service charges in the case of a private, unregulated airport, which can reduce the benefits of regulation. However, car rental services can also increase aeronautical service charges of a private, unregulated airport, and car rental service charges themselves can be excessive in this situation. Overall, it turns out that commercial airport services are not a substitute for regulation, and this was confirmed by the results of the numerical simulations. There are some aspects that are not adequately covered in this paper and that should be addressed to obtain a more complete and realistic picture of optimal airport charges from the regulator s perspective and of the relevance of commercial services to airport regulation. These aspects include airline market structures, regulation regimes, and capacity choice. I found that, in a context where commercial airport services exist and the regulator s objective is to maximize total surplus, aeronautical service charges should be used to fully internalize marginal congestion costs. However, it has been shown that the level of aeronautical service charges that is needed to internalize marginal congestion costs depends on the airline mar- 22

24 ket structure. 8 Note that airlines mostly exist in the context of oligopolistic rather than perfect competition (the latter being the condition that is assumed in this paper). Hence, to obtain a more realistic picture of total surplus-maximizing aeronautical service charges, oligopolistic airline market structures should be considered. However, I do not expect that variations in airline market structures will affect the core results of this paper with regard to the relationship between aeronautical and commercial airport services. There is a debate about whether airport regulation regimes should pursue a single-till or dual-till approach. 9 The key difference between single-till and dual-till regulation is that with single-till, commercial revenues are used to cover fixed airport costs, which includes runway costs (Czerny, 2006). In comparison, the dual-till approach attributes portions of fixed airport costs to aeronautical and commercial business areas so that cost recovery then has to be achieved in both areas separately. I found that aeronautical and car rental services should be provided for free in the case of consumer surplus maximization. In this case, implementation is possible under single-till regulation only (or with subsidies, which are usually not available for airports). Under a dual-till regulation regime, in contrast, aeronautical service charges would have to be positive to achieve cost recovery in this business area. It is possible to expand airport capacity in reality. To obtain a more complete picture of the benefits of airport regulation in a context where commercial services are relevant, it would therefore be essential to consider airport investments. (Zhang and Zhang, 2003) consider lumpy airport in- 8 For a detailed discussion of the relationship between airport congestion charges and airline market structures, see (Brueckner, 2002), (Daniel and Harback, 2008), (Morrison and Winston, 2007), (Brueckner and van Dender, 2008). 9 For a detailed discussion of single-till and dual-till regulation regimes, see (Crew and Kleindorfer, 2001), (Starkie, 2001), (Lu and Pagliari, 2004), (Oum et al., 2004), (Czerny, 2006), and (Starkie, 2008) 23

25 vestments in a growing airport market and find that a private, unregulated monopolistic airport slows down airport expansion in comparison with expansion under conditions of total surplus maximization. This result could be different in my setting, where the supply of retailing services by airports does not contribute to the consumer or total surplus but is likely to increase the pace of investments under profit maximization and with no regulation at all. Overall, the research provided in this paper is based on a more realistic model of commercial airport services, and the results therefore provide a clearer picture of the relationship between aeronautical and commercial airport services as well as the principles underlying the benefits of airport regulation in a context where commercial airport services are relevant. However, future research is required to complete the picture further. A Proof of Proposition 1 I begin with car rental service charges and then turn to aeronautical service charges. Rearranging the first-order condition for optimal car rental service charges from a regulator s viewpoint in (9) leads to p 2 = 0 directly follows. λ 2 = D 2 (1 β) β ( p D p 2 p 2 ) D 2 > 0. (25) p 2 24

26 Substituting p 2 by p 2 = 0 into the first-order condition for optimal aeronautical service charges from a regulator s viewpoint in (8) and rearranging leads to Note that 0 D 2 (p 1, x 2 ) (p 1 + Γ) + β ( dx 2 D 1 (p 1 + Γ, ) 1 + Γ ) p 1 ( D 1 + p 1 ) D 1 + λ 1 = 0. (26) p 1 0 D 2 (p 1, x 2 ) (p 1 + Γ) dx 2 = D 1 (p 1 + Γ, ) D 1 (p 1 + Γ, 0). (27) Then, the equations in (26) and (27) together lead to ( λ 1 = D 1 (p 1 + Γ, 0) 1 + Γ ) ( β D 1 + p 1 p 1 ) D 1. (28) p Γ/ p 1 > 0 holds true because D 1 / p 1 < 0, which implies p 1 = 0 for β = 0. Marginal congestion costs are not fully internalized in this case. On the other hand, rearranging the equation in (28) leads to p 1 = D 1 Γ D 1 (29) and an airfare p 1 + Γ that equals marginal congestion costs in (1) for β = 1; thus, marginal congestion costs are fully internalized in this case. References ACI (2008). ACI Airport Economics Survey

27 Baron, D. P. and Myerson, R. B. (1982). Regulating a monopolist with unknown costs. Econometrica, 50(4): Basso, L. J. (2008). Airport deregulation: Effects on pricing and capacity. International Journal of Industrial Organization, 26(4): Brueckner, J. K. (2002). Airport congestion when carriers have market power. American Economic Review, 92(5): Brueckner, J. K. (2009). Price vs. quantity based approaches to airport congestion management. Journal of Public Economics, 93(5-6): Brueckner, J. K. and van Dender, K. (2008). Atomistic congestion tolls at concentrated airports? seeking a unified view in the internalization debate. Journal of Urban Economics, 64(2): Crew, M. and Kleindorfer, P. (2001). Regulation for privatized airports: Single-till versus multi-till pricing methodologies for sydney airport. Crew, M. A. and Kleindorfer, P. R. (1979). Public Utility Economics. Czerny, A. I. (2006). Price-cap regulation of airports: single-till versus dual-till. Journal of Regulatory Economics, 30: Daniel, J. I. and Harback, K. T. (2008). (When) Do hub airlines internalize their self-imposed congestion delays? Journal of Urban Economics, 63(2): Graham, A. (2009). How important are commercial revenues to today s airports? Journal of Air Transport Management, 15(3): Lu, C.-C. and Pagliari, R. I. (2004). Evaluating the potential impact of alternative airport pricing approaches on social welfare. Transportation Research Part E, (40):

28 Morrison, S. A. and Winston, C. (2007). Another look at airport congestion pricing. American Economic Review, 97(5): Oum, T. H., Zhang, A., and Zhang, Y. (2004). Alternative forms of economic regulation at airports. Journal of Transport Economics and Policy, 38(2): Starkie, D. (2001). Reforming UK airport regulation. Journal of Transport Economics and Policy, (35): Starkie, D. (2008). Aviation Markets: Studies in Competition and Regulatory Reform. Ashgate. Zhang, A. and Zhang, Y. (2003). Airport charges and capacity expansion: Effects of concessions and privatization. Journal of Urban Economics, 53(1):