1、MULTIMARKET CONTACT AND SERVICE QUALITY:EVIDENCE FROM ON-TIME PERFORMANCE IN THEU.S. AIRLINE INDUSTRYJEFFREY T. PRINCEDANIEL H. SIMONCornell UniversityWe examine the relationship between multimarket contact and service quality. Weoffer two hypotheses: (1) multimarket contact negatively affects servi
2、ce quality and (2)multimarket contacts in less competitive markets more negatively affect service qual-ity. We test these hypotheses using U.S. airline on-time performance data to measureservice quality. We find that multimarket contact increases delays and that this effectis greater for contacts on
3、 more concentrated routes, although the effect diminishes onvery highly concentrated routes. These findings provide support for the mutual for-bearance hypothesis and suggest that multimarket contact facilitates tacit collusion onquality as well as price.One of the most widely studied questions inst
4、rategy and in industrial organization economicsis how market structure influences firms competi-tive behavior. For many years, research on thisquestion focused exclusively on firms competingon price in single markets. However, more recently,researchers have branched out in two directions,examining f
5、irms that compete against each other inmultiple markets, and examining the influence ofmarket structure on nonprice forms of competition.Over the past 20 years, a stream of papers (BaumEvans Fer-nandez Gimeno Jans however, the effect diminishes onvery highly concentrated routes.Additional analysis s
6、howed that nearly all ofthe additional delay was due to increased timespent on the ground, primarily before aircraft leftthe gate. These results suggest that mutual for-bearance likely results in reduced investments incheck-in, baggage, and/or maintenance staff andequipment. Moreover, we found that
7、the increasein arrival delays is not caused by spending moretime in the air, nor is it caused by shorter sched-uled flight times; scheduled flight times actuallyincrease with multimarket contact. We alsofound that airlines cancel more flights, flysmaller planes, and offer fewer flights as theirmulti
8、market contact with rivals increases. Theseresults provide additional evidence that multi-market contact actually influences airlineschoices, because flight schedules, as well asflight cancellations and the sizes of the planesflown, are more directly under the airlinescontrol.THEORETICAL BACKGROUNDM
9、arket Structure and QualityJust as market structure may influence a firmspricing behavior, it may also affect a firms provi-sion of product or service quality. However, al-though competition should unambiguously reduceprices, its effect on quality is less clear, becausechanges in quality affect cons
10、umers willingness topay and producers costs.Several authors have discussed and analyzedthe relationship between competition and servicequality, both theoretically and empirically. Thetheoretical literature has been largely inconclu-sive, with some studies suggesting that competi-tion increases quali
11、ty (Schmalensee, 1974), oth-ers suggesting that competition reduces quality(Gal-Or, 1983), and still others positing either norelationship between competition and quality(Swan, 1970), or that the relationship depends ona variety of assumptions (Schmalensee, 1979).Two issues seem to underlie these di
12、vergent the-oretical results: (1) difficulty in defining qualityin a way that is mathematically tractable and (2)sensitivity of results to assumptions of a partic-ular theoretical model (Schmalensee, 1979).We posit that for competition to lead to “exces-sive” service quality, it must be the case tha
13、t, at thecompetitive level of quality, the aggregate marginalrevenue (for all firms) from all firms jointly provid-ing lower quality is strictly greater than the aggre-gate marginal cost. In this situation, we would ex-pect a reduction in competition to result in areduction in product (service) qual
14、ity, as firmsmove toward a more cooperative (monopolistic)outcome. In keeping with this expectation, Mazzeo(2003) argued that when customers have morechoices, firms have an incentive to offer higher-quality goods, better service, and lower prices inorder to maintain their market shares. Moreover,thi
15、s incentive to improve quality is enhanced be-cause the cost of attracting new customers is higherin markets with more rivals (Estelami, 2000).Most empirical tests of the impact of competitionon quality have focused on various measures ofservice quality in the service sector. However, asDranove and
16、White (1994) noted, difficulties inmeasuring quality make inference difficult. None-theless, researchers have found a variety of ways tomeasure quality. Liao and Chuang (2004) examinedthe impact of local competition on customer satis-faction ratings in the fast food industry. They foundthat the numb
17、er of rivals in a given local market hasa positive effect on customer ratings of overall sat-isfaction, customer service, and loyalty (Liao however, itis worth noting that combining the results fromstudies of the effect of competition on pricing,which have generally shown that competition re-duces p
18、rices, with the results from the studies de-scribed above, showing that competition increasesservice quality, provides evidence for the strongcompetitive response that we propose. That is,these two research streams jointly suggest that (asfound in Domberger et al. 1995) firms respond toincreased com
19、petition by cutting price and increas-ing service quality.Multimarket Contact and Mutual ForbearanceAccording to the mutual forbearance hypothesis,firms that compete in multiple markets will bemore likely to cooperate, because a competitiveattack in any one market may draw responses in allother join
20、tly contested markets. Therefore, the pun-ishment for deviations from collusive behaviorwould increase with the number of markets inwhich rivals meet. Bernheim and Whinston (1990)pointed out that if multimarket contact simply in-creases the costs and benefits of deviating fromtacit collusion in equa
21、l proportions, then there isno reason to think that multimarket contact wouldaffect incentives to cooperate. However, they alsoshowed that under a variety of plausible condi-tions, multimarket contact may facilitate coopera-tive behavior. These include cost differencesamong firms, differences in the
22、 number of rivals indifferent markets, and differences in market de-mand growth ratesconditions that all potentiallyhold in the airline industry.Recently, several studies have found evidence ofa relationship between multimarket contact andhigher prices. In the airline industry, Evans andKessides (19
23、94) and Gimeno and Woo (1996, 1999)showed that airlines charge higher prices on routesin which they face rivals against whom they com-pete on many different routes. Gimeno and Woo(1999) showed that contacts on routes that share anend-point airport with a focal route have a greatereffect on price on
24、the focal route. Fernandez andMarin (1998) found that Spanish hotels chargehigher prices in markets that they share with rivalsagainst whom they compete in many other markets.In the U.S. cement industry, Jans and Rosenbaum(1996) found that the effect of multimarket contacton prices is positive when
25、weighted by a firmsmarket share or a Herfindahl index value in eachmarket where the contact occurs. Parker and Roller(1997) found that multimarket contact increases theprice of mobile phone service. Busse (2000) pro-vided some evidence for the mechanism throughwhich multimarket contact may affect pr
26、ices, asshe showed that cell phone providers are morelikely to offer identical price schedules across mar-kets when they compete against the same rivals inmultiple markets.Several studies have also examined the link be-tween multimarket contact and other types of firmbehavior. Baum and Korn (1996, 1
27、999) showed thatmultimarket contact among California commuterairlines influences market entry and exit rates, withthe effect being greatest on routes dominated by asingle airline. Shankar (1999) found that multi-market contact reduces marketing expenditures byfirms introducing new products and weake
28、ns re-sponses from incumbent competitors. Jayachand-ran, Gimeno, and Varadarajan (1999) provided fur-ther theoretical development of the link betweenmultimarket contact and competitive behavior,while extending the mutual forbearance hypothesisto product line rivalry and discussing how multi-market c
29、ontact may influence firms entry strate-gies. Vonortas (2000) explored the relationship be-tween multimarket and multiproject contact andfirm incentives to engage in R the hub-and-spoke systememployed by nearly all major airlines creates costdifferences among airlines in different markets; andthere
30、are differences in the number of carriers serv-ing different routes (Evans Gim-eno friends couldprovide information on their past experiences fly-ing a route; for example, a traveler from Chicago,planning a trip to New York, might be told byfriends that they experienced delays when flyingUnited Airl
31、ines from OHare to LaGuardia. Third,business travelers, who often must fly the sameroute many times, could use their own past expe-riences to inform their future expectations. Conse-quently, this group is especially capable of learningabout airlines on-time performance on particularroutes from perso
32、nal experience and from the ex-periences of colleagues.Providing additional support for Assumptions 1and 2 is research showing that demand for air tra-vel increases with on-time performance. Both Ja-nuszewski (2004) and Morrison and Winston(1983) showed that airlines increase prices in re-sponse to
33、improvements in own on-time perfor-mance, and Januszewski also showed that airlinesreduce prices in response to improvements in ri-vals on-time performance. Similarly, Suzuki(2000) found that airline market shares increasewith on-time performance. These results supportthe contention that customers c
34、are about on-timeperformance and provide evidence that customerscan observe route-level on-time performance (or atleast that airlines believe they can).Assumptions 3 and 4 ensure that airlines cancompete on on-time performance at the route leveland link this competition to the level of multimar-ket
35、contact. Although several studies have shownthat airlines raise prices on routes with highermultimarket contact, it might seem less plausiblethat airlines would adjust their investments inservice quality on a route-by-route basis and beable to link their adjustments to route-level mul-timarket conta
36、ct.We justify Assumption 3 by calling upon argu-ments and findings in the prior literature. AsMazzeo (2003) discussed, airlines can make differ-ent kinds of route-level investments that affect on-time performance. For example, they can hire ad-ditional employees to speed up processes such asloading
37、and unloading baggage, check-in, and soforth. Similarly, airlines can avoid delays resultingfrom mechanical failures by maintaining an addi-tional airplane(s) at an airport, or by having a readysupply of mechanics available. Additionally, theycan reduce time spent in the air by using largerairplanes
38、 with greater flying speed. They can alsoacquire more slots at airports.Although many of the investments describedabove are more airport- than route-specific, they arealso highly fungible and can be quickly reallocatedacross routes within the airport. Drawing on theexample used by Mazzeo (2003), if
39、the preflightmaintenance check reveals a problem with anAmerican Airlines plane scheduled to fly from Dal-las to Nashville, then American might replace itwith a plane that had been scheduled to fly toIndianapolis. Similarly, airlines may transfer flightcrews or baggage personnel to different routes
40、inresponse to route-specific conditions. Moreover,advances in technology have greatly facilitated thissort of resource reallocation by airlines in recentyears (Mazzeo, 2003).Mazzeo (2003) and Rupp et al. (2003), as well asMayer and Sinai (2003), have provided empiricalsupport for the idea that airli
41、nes adjust their invest-ments in service quality route-by-route. As notedabove, Mazzeo (2003) and Rupp et al. (2003) bothfound that airlines provide worse on-time perfor-mance on less competitive routes. Similarly, Mayerand Sinai (2003) found that airlines provide worseon-time performance on flights
42、 leaving their hubs,because they schedule many flights to arrive at thehubs at the same time, in order to facilitate conve-nient connections, and because they hold flights towait for connecting passengers. Finally, Borensteinand Netz (1999) showed that route-level competi-tion influences airlines fl
43、ight schedules.Typically, a local airport manager allocates air-port resources to different routes. Is it reasonable,then, to assume, as in Assumption 4, that this man-ager can link his or her decisions to contacts be-tween rivals in multiple routes? We argue that suchlinking is plausible for two re
44、asons. First, managersat an airlines central office are likely to be aware ofmultimarket contact at the route levelas evi-denced by pricing strategies. Although local man-agers may not be aware of multimarket contact atthe route level, the central managers can providedirection to them by emphasizing
45、 particular routeson which on-time performance is a priority. Sec-ond, mutual forbearance does not necessarily im-ply that airline managers adjust their operations2009 341Prince and Simonroute-by-route in response to changes in multima-rket contact on each route. They may also adjusttheir service qu
46、ality efforts on a rival-by-rival basis.That is, when an airlines multimarket contact witha rival changes, the airlines management maychoose to adjust its operations on all routes whereit competes with that rival. For example, as Unitedand American airlines compete against each otheron more routes,
47、each airline may choose to reduceits investments in service quality on all (or onsome) of the routes where they meet.Assumption 5 ensures that increased cooperationamong airlines will lead to worse on-time perfor-mance. Underlying this assumption is the likeli-hood that improving on-time performance
48、 is costly,with the aggregate costs of airlines jointly improv-ing on-time performance above the competitivelevel increasing faster than the benefits. This as-sumption is necessary in order to extend the mu-tual forbearance hypothesis to on-time perfor-mance. It provides the incentive for firms to r
49、elaxon-time performance efforts. Doing so enables firmsto reallocate resources, ticketing agents, baggagehandlers, mechanics, and airplanes to routes withlower levels of multimarket contact, where on-timeperformance competition is likely to be more in-tense, and where airlines face stronger incentives toimprove on-time performance.To justify this assumption, we again call on theresults of Mazzeo (2003) and Rupp et al. (2003).These works both provide evidence that lower lev-els of competition lead to worse on-time perfor-mance. Moreover, this evidence is consistent withfindi
