NBER WORKING PAPER SERIES ON THE PERSISTENT FINANCIAL LOSSES OF U.S. AIRLINES: A PRELIMINARY EXPLORATION. Severin Borenstein

Transcription

1 NBER WORKING PAPER SERIES ON THE PERSISTENT FINANCIAL LOSSES OF U.S. AIRLINES: A PRELIMINARY EXPLORATION Svrin Bornstin Working Papr NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachustts Avnu Cambrig, MA January 2011 In 2010, I was a mmbr of th USDOT's Futur of Aviation Avisory Committ. I also avis th U.S. Dpartmnt of Justic in its analysis of th propos mrgr btwn Continntal an Unit Airlins. For hlpful commnts an iscussions, I am gratful to Silk Forbs, Will Gillspi, Kn Hyr, Paul Joskow, Nancy Ros an Cathrin Wolfram. A short vrsion of this papr is forthcoming in th May 2011 Amrican Economic Rviw Paprs an Procings unr th titl "Why Can't U.S. Airlins Mak Mony?''. This papr is icat to th mmory of Alfr E. Kahn who pass away on Dcmbr 27, I was lucky nough to work for Fr at th Civil Aronautics Boar in 1978 an to spak with him occasionally sinc thn about th airlin inustry an govrnmnt rgulation. His approach to inustrial organization an rgulation, an th application of rsarch to non-partisan policy making, st a stanar to which all IO conomists shoul aspir. His insights continu to influnc th bst rsarch on conomic rgulation. H was th vry mol of a morn (an thoughtful) IO conomist. Th viws xprss hrin ar thos of th author an o not ncssarily rflct th viws of th National Burau of Economic Rsarch. NBER working paprs ar circulat for iscussion an commnt purposs. Thy hav not bn prrviw or bn subjct to th rviw by th NBER Boar of Dirctors that accompanis official NBER publications by Svrin Bornstin. All rights rsrv. Short sctions of txt, not to xc two paragraphs, may b quot without xplicit prmission provi that full crit, incluing notic, is givn to th sourc.

2 On th Prsistnt Financial Losss of U.S. Airlins: A Prliminary Exploration Svrin Bornstin NBER Working Papr No January 2011 JEL No. L1,L93 ABSTRACT U.S. airlins hav lost narly $60 billion (2009 ollars) in omstic markts sinc rgulation, most of it in th last ca. Mor than 30 yars aftr omstic airlin markts wr rgulat, th ismal financial rcor is a puzzl that challngs th conomics of rgulation. I xamin som of th most common xplanations among inustry participants, analysts, an rsarchrs -- incluing high taxs an ful costs, wak man, an comptition from lowr-cost airlins. Dscriptiv statistics suggst that high taxs hav bn at most a minor factor an ful costs shocks play a rol only in th last fw yars. Major rivrs sm to b th svr man ownturn aftr 9/11 -- man rmain much wakr in 2009 than it was in an th larg cost iffrntial btwn lgacy airlins an th low-cost carrirs, which has prsist vn as thir pric iffrntials hav gratly clin. Svrin Bornstin Haas School of Businss Univrsity of California, Brkly Brkly, CA an NBER bornst@haas.brkly.u

3 In 2008 an 2009, U.S. passngr airlins rport aggrgat nt losss, bfor xtraorinary incom an chargs, of $14 billion on rvnus of $270 billion. 1 About 76% of th losss wr on omstic U.S. oprations, which hav bn rgulat sinc th fall of Most intrnational routs rmain mor havily rgulat, an gnrally mor lucrativ for thos carrirs that ar prmitt to srv thm. Th vry poor financial rsults in again spark iscussions of why th airlin inustry has far so baly sinc rgulation. From 1979 through 2009, U.S. airlins lost $59 billion (in 2009 ollars) on omstic oprations. 2 Figur 1 shows nt incom on omstic oprations for th inustry sinc 1979, scal by th siz of oprations (availabl sat-mils). 3 It illustrats that th losss hav bn ramatically wors in th last tn yars than in th prvious two cas of rgulation. In fact, in 2009 ollars, omstic passngr airlin oprations lost $10 billion from 1979 to 1989, ma profits of $5 billion in th 1990s an lost $54 billion from 2000 to To put ths numbrs in contxt, at th n of 2009, th ntir book valu of U.S. passngr carrirs assts was about $163 billion an th book valu of sharholr quity was $10 billion. Evn at th n of th 2000, aftr six conscutiv profitabl yars, thir assts wr $159 billion an sharholr quity was $40 billion (all in 2009 ollars). 4 Thr cas aftr rgulation th inustry s financial track rcor is ismal. This isn t what conomists, analysts or inustry participants prict in It is a puzzl to inustrial organization conomists an a challng to th viws of rgulation avocats. Th puzzl is compoun by th fact that th inustry saw robust invstmnt until 2001 an has sn only most isinvstmnt in th financially isastrous 2000s. From 1979 to 2001, th U.S. airlin passngr flt grw in vry yar, by an avrag of 4.9% pr yar 1 Th arnings figurs I rport throughout this papr xclu asst writowns, pnsion sttlmnts, rorganization costs an frsh start accounting ajustmnts, which ar oftn associat with bankruptcis an mrgrs. Incluing ths ajustmnts os not chang th basic pictur, but causs larg swings in yar-to-yar rport arnings that ar not attributabl to markt activitis in th spcific yar. Capital gains an losss from ful hging ar gnrally inclu in oprating xpnss (vint in avrag ful purchas prics that iffr substantially from th markt pric), not xtraorinary incom an chargs. Ths ata inclu only U.S. carrirs that rciv at last $1 million pr quartr from passngr rvnus, so thy xclu cargo carrirs such as UPS an F Ex. S th Appnix for a list of th carrirs inclu. Th nt incom bfor xtraorinary chargs os inclu bt paymnts an taxs. All rfrncs hr to arnings, nt incom, profits an losss us this masur. In trms of DOT incom statmnt accounting, this is nt incom minus othr nt incom. 2 Th losss ar slightly largr, $67 billion, whn intrnational oprations ar inclu. 3 S th appnix for tails of all calculations for figurs. 4 Carrirs assts inclu aircraft an othr facilitis on long-trm las. 2

4 Figur 1: Nt Incom (/c2009) pr Availabl Sat-Mil: U.S. Domstic Markts masur by aircraft an 3.6% pr yar masur by aircraft-sats. 5 Th flt siz pak in From th n of 2001 to th n of 2008 (latst availabl at), aircraft an aircraft-sats clin by 1.7% an 1.4% pr yar rspctivly. Bornstin & Ros (2008, hncforth BR ) arss th volatility of airlin profits, showing that fluctuations in man an ful prics along with fix capital costs an sticky labor costs can xplain th inustry s arnings volatility. But that analysis i not arss th lvl of profits, th fact that th omstic airlin inustry has rport ngativ nt incom in 23 of 31 yars sinc rgulation an a strongly ngativ aggrgat nt prsnt valu of arnings. Thr is no convntional long-run quilibrium xplanation for an inustry that prptually loss mony, but thr ar a numbr of isquilibrium thoris that hav bn suggst by inustry participants, financial analysts, an rsarchrs. In this short papr I iscuss ths thoris an attmpt to narrow own th rang of plausibl xplanations. 5 Ths aircraft ata covr omstic an intrnational oprations of narly all U.S. passngr airlins oprating 19-sat an largr aircraft. As suggst by this iffrnc, th avrag siz of commrcial aircraft in th U.S. clin uring this prio, u in part to th growth of rgional jts. 3

5 l W l l l l l l l l W l > W Figur 2: Avrag Tickt Tax I. Exognous cost rivrs: taxs an ful Inustry lars argu that th tax an f burn on airlin tickts is xcssiv, toay incluing a 7.5% tickt tax an fs of $6.20 pr sgmnt flown. In aition, many airports impos passngr facilitis chargs (PFCs) of up to $4.50 on ach passngr boaring a flight at th airport. On can argu about whthr ths taxs ar xcssiv givn th govrnmnt costs of supporting th inustry, but it is ifficult to s how ths woul la to losss for mor than a short transitional prio. Figur 2 shows that th avrag tax (incluing fral tickt taxs an PFCs) as a prcntag of th bas tickt pric has climb staily, an is toay about twic as high as whn it was 8% through most of th 1980s. 6 But th avrag ollar tax pr tickt (in 2009 ollars) is toay at about th sam lvl it was in th profitabl lat 1990s. Ovr th last 30 yars, th primary form of taxation has transition from strictly a prcntag xcis tax to a mix of prcntag an pr-sgmnt taxs. In th 1980s, th ntir tickt tax was a prcntag of th tickt valu. Th passngr facility chargs wr a in th arly 1990s, th sgmnt tax in 1997 an th Sptmbr 11 scurity f in 6 Th calculation of avrag tax as a prcntag of pric inclus baggag an tickt-chang fs in bas pric. 4

6 l l l l l l l Figur 3: Ral Avrag Tickt Pric Rlativ to 1979 (ajust for avrag trip istanc) (inclusiv of govrnmnt taxs/fs an avrag airlin baggag an tickt-chang fs) arly 2002, all bas on th numbr of flights th passngr boars, rgarlss of th far pai. As a rsult, as ral fars hav clin, ropping significantly aftr th Sptmbr 11 attacks, th tax burn incras as a prcntag of th bas far. 7 Th problm sms to b not that taxs hav risn, but that th bas fars hav falln an stay so low. Evn th post-9/11 tax incras has mostly rvrt in ral trms. Figur 3 shows th avrag tickt pric rlativ to 1979, ajust for inflation an trip istanc (which has incras substantially ovr tim). 8 Whil taxs an fs hav chang incrmntally, th inustry scal has chang massivly. In th stanar long-run ajustmnt ynamics, it sms that th inustry shoul hav bn abl to achiv th scal chang ncssary to incorporat an pass through ths taxs. My own rsarch in progrss suggsts that changs in passngr facilitis chargs ar narly ntirly pass through to customrs within two quartrs. 7 Th substantial f incras in arly 2002 rais rvnu for significantly xpaning scurity srvics aftr 9/11. 8 S th appnix for tails of this calculation. 5

7 Figur 4: Jt Ful Pric ($2009) Ful costs incrass hav crtainly bn a significant componnt of losss in som yars, most obviously Ovr th rgulation ra, howvr, oil costs wr highst in th first 7 yars an th most rcnt 5 yars, ovr $40 pr barrl in 2009 ollars, an much lowr uring th 19 intrvning yars. Figur 4 shows that from 1986 to 2004 th avrag jt ful pric was blow $1.40 pr gallon rlativly stabl an much lowr than in th arly prio of rgulation. Yt, th inustry still lost mony in 13 of thos 19 yars an on nt lost $31 billion in 2009 ollars. Whil thr is no qustion that th airlins arnings ar affct in th short run by xtrm oil pric fluctuations such as occurr in th last fw yars, thr osn t appar to b a barrir to capacity ajustmnt ovr 3 to 6 months in rspons to oil pric changs. Th rapi ructions in schuls in th scon half of 2008 mak that clar. Still, rucing flight schuls osn t liminat costs if thos costs ar fix or sticky. In tims of growing man, carrirs can ajust fairly smoothly to unanticipat cost incrass by growing mor slowly, without having to groun aircraft or ruc workforc siz. Whn man is stagnant or clining, howvr, rscaling oprations in rspons to upwar cost shocks is mor ifficult an costly. 6

8 l l l l l l l l Z'W Figur 5: Impli Domstic Airlin Dman an Ral GDP, Rlativ to 1979 II. Exognous man shocks Th rol of man shocks in airlin losss is most notabl in an in Prior to 9/11, howvr, it appars that omstic man grw fairly staily. Infrring man shifts from avrag yil an rvnu passngr-mils, man changs ar prsnt in figur 5. 9 Dman incras by 110% from 1979 to 2000, growing in 16 of thos 21 yars. Yt, th inustry ma mony in only 8 of thos yars an ovrall lost $3 billion ($2009) ovr this prio. Th conomic ownturns uring this prio crtainly affct airlin inustry profits, but w wouln t xpct invstors to bliv that man growth woul b compltly constant an stay. It is har to s how unanticipat man shocks uring this tim coul b a cribl xplanation for th ovrall poor prformanc. Dman shocks ar a mor plausibl xplanation for th losss of th 2000s. Th post- 9/11 man rop, which was about 20% from 2000 to 2002, was unprcnt. By 2008, man was still about 3% lowr than it ha bn in 2000, an thn it ropp about 11% in Bcaus of th fix capital costs an sticky labor costs, th ca of prss man was accompani by a ca of prss prics. In ral trms prics wr 20% 9 I follow th sam basic approach as in BR, but us a pric masur that ajusts for avrag trip istanc. S th appnix for furthr xplanation. 7

9 lowr in 2009 than in 2000 (ajust for trip istanc) spit th fact that jt ful prics wr about $0.59 pr gallon (52%) highr, which, bas on 2009 rvnu passngr-mils pr gallon of ful, rais ovrall costs by about 9%. Th othr notabl chang in omstic srvic ovr this prio was th incras in avrag passngr loa factor from 71% in 2000 to 81% in This was a continuation of th vry stay incras in th 1990s from about 60% in 1990 (an most of th 1980s). Th incrass ar mostly inpnnt of man shocks, rising or holing constant in vry yar xcpt a 2% ownwar tick aftr Sptmbr 11, which was compltly rvrs by Incras ful costs woul mak highr loa factors mor conomic, but thr is no vinc that th loa factor incrass hav bn gratr uring prios of rising than prios of falling ful costs. Mor likly, loa factor incrass hav bn a rsult of improving yil managmnt tchnologis. III. Entry an xpansion of low-cost carrirs Many inustry obsrvrs an participants point to low-cost (an low-far) carrirs (LCCs) as part of th rason for low inustry profits, but thr is wi isagrmnt on what th connction is. If LCCs ar simply offring a lowr-quality prouct, thn thir iffrntiat prouct shoul fin its nich in th markt if thr is sufficint man for that quality lvl, yiling an quilibrium with both typs arning normal rturns. Among inustry an labor lars, a common viw is that nw low-cost ntrants an LCC incumbnts hav ma xcssiv capacity invstmnts uring growth prios, an somtims vn uring ownturns, that hav prss prics for all. In orr to iscourag xcssiv invstmnt, th largst airlin pilots union has call for incrasing capital rquirmnts as part of FAA licnsing of nw airlins. 10 But th vinc osn t appar to support th ia that nw ntrants or olr LCCs ar mor pron to ovr-invstmnt than th lgacy airlins. Figur 6 prsnts th aircraftsat flt siz of LCCs an non-lccs (incluing lgacy carrirs an rgional carrirs who gnrally oprat as coshar partnrs to th lgacy carrirs). Two things ar clar from this figur. First, LCCs in aggrgat hav xprinc no mor rratic flt siz ajustmnts spit bing lss wll-stablish on avrag. 11 In fact, thy continu to 10 If LCC s wr mor inclin to ovrinvstmnt, w woul xpct thir financial prformanc to b wors on avrag an to b mor volatil than lgacy carrirs. That osn t sm to b th cas as iscuss blow. 11 Th clins in 1987 an 1988 ar caus by th purchass of PSA by US Airways an Air California 8

10 >Z > Figur 6: Aircraft-Sats in Flt grow graually vn aftr 9/11 whil rmaining much lss unprofitabl than th lgacy carrirs, as shown latr. If anything, it appars to b th lgacy carrirs who ar mor pron to ovr-invstmnt rlativ to th growth of thir traffic. Scon, th changs in flt siz of th LCCs is warf by th variation of th non-lcc flt, suggsting that LCC invstmnt cisions hav not bn th primary rivr in inustry capacity changs. 12 An altrnat viw of LCCs is that thy hav bn graually chipping away at th ntrnch positions of lgacy carrirs that hav much highr costs. Th chang has bn graual, bcaus th lgacy carrirs ar also protct by ntwork markting programs an othr activitis that rais barrirs to ntry by mor fficint firms. Potntially xclusionary activitis of lgacy carrirs inclu frqunt-flyr an corporat iscount programs that xchang iscounts for customr loyalty on a portfolio of unrlat routs, 13 as wll as by Amrican Airlins, in both cass transfrring LCC flts to th control of lgacy carrirs. Similarly, th clin in 2007 is a rsult of th US Airways-Amrica Wst mrgr. 12 Ths figurs inclu ntir carrir flts, som of which ar us on intrnational routs, but th conclusion is not chang if th analysis is limit to narrow-boy aircraft, which ar us primarily for flights within North Amrica. 13 Svrin Bornstin (1996) iscusss th potntial anti-comptitiv ffcts of such rpat-buyr programs in mor tail. Mara Lrman (2007, 2008) prsnts vinc on th impact of FFPs. 9

11 l l l l l l l > ^ Figur 7: Domstic Markt Shar of Low-Cost Carrirs (by rvnu passngr-mils) rlationships with airports that allow larg incumbnts to rstrict th availability of gats, laning slots an othr rsourcs to potntial ntrants. LCCs hav bn growing staily sinc th arly 1990s. Figur 7 shows thir omstic markt shar, by rvnu passngr-mils sinc LCCs now compt (fin as at last 10% passngr shar) on ovr 60% of all airport pairs, an ovr 80% of all city-pairs if on assums that th iffrnt airports in Dallas, Houston, Chicago, San Francisco, Los Angls, Nw York, an Washington DC ar in th sam markts. An LCCs hav much lowr costs than th lgacy carrirs. Figur 8 shows th oprating cost pr availabl sat-mil ajust for avrag flight istanc (or haul ) lngth. 15 Ajust for th avrag flight istanc, lgacy carrir costs hav rmain 30%-60% highr than th LCCs for narly all of th rgulation ra, avraging about 40% highr in th last ca. Th caus of this prsistnt cost iffrnc is an ara in n of furthr rsarch. Ful 14 Th ownticks in 1986, 1987 an 2008 ar from lgacy carrirs absorbing th oprations of LCCs. 15 Dtails of th ajustmnt ar prsnt in th appnix. 10

12 >Z > Figur 8: Avrag Oprating Costs (/c2009) pr Availabl Sat-Mil (ajust for avrag flight lngth) costs ar approximatly th sam for all airlins. Wags for flight attnants an groun crw at Southwst airlins ar about comparabl to thos for similar staff at lgacy carrirs, an pilots of comparabl aircraft ar pai about th sam. That may not b tru for som of th othr LCCs, howvr. I am not awar of an analysis that incorporats full bnfits packags incluing pnsions. Work ruls ar clarly mor flxibl at Southwst an th othr LCCs, an mploys in gnral ar assign to a wir array of activitis, probably laing to highr labor utilization. Aircraft utilization is highr at Southwst than at lgacy carrirs, though it woul b valuabl to stuy how that has chang as Southwst has xpan to airports with mor congstion an wathr isruptions. Quality factors also iffr on-tim rats, passngr complaint rats, lgroom, sat assignmnt policis, among othr factors. Whn all of ths factors ar consir, howvr, it is not clar which carrirs hav th high-quality prouct in omstic markts. Whil th cost iffrntial btwn LCCs an non-lccs has rmain larg, th avrag pric iffrntial has bn shrinking, as shown in figur 9. Figur 9 is ajust for th avrag trip istanc of passngrs flying on ach typ of carrirs. 16 LCC fars hav 16 Dtails of th ajustmnt ar prsnt in th appnix. 11

13 l l l l l l l l l l l Figur 9: Pric Prmium of Non-LCCs Ovr LCCs (ajust for trip istanc) clin much lss than thos of lgacy carrirs in th 2000s, rflcting in part thir lowr burn of xcss aircraft capacity. This is no oubt a larg part of th rason that LCCs hav suffr much milr losss in th 2000s, as shown in figur IV. A sris of unfortunat vnts? Dman an cost shocks hav crtainly play a significant rol in th airlin inustry s poor financial rsults, but thr is littl rason to think thos isruptions will b lss frqunt in th futur. Furthrmor, aftr mor than 30 yars, it sms unlikly that airlin losss ar u ntirly to a sris of unfortunat xognous vnts rlativ to what managmnt an invstors shoul hav xpct. Throughout rgulation, th lgacy carrirs hav maintain much highr costs than LCCs, but th pric prmia thy hav bn abl to charg hav graually clin ovr th last 20 yars, shrinking by mor than 60% ovr that tim. As a rsult, whil th 17 Ths ata inclu th roughly $2 billion Southwst airlins nt profits uring from hging ful prics. Rmoving thos savings maks th iffrnc somwhat smallr, but osn t chang th qualitativ conclusion. 12

14 >Z > Figur 10: Domstic Nt Incom (/c2009) pr Availabl Sat-Mil for Non-LCCs an LCCs xognous man an cost shocks hav affct all carrirs, th lgacy airlins hav far much wors financially, an LCCs hav grown staily. Th rspons of lgacy carrirs has bn to xpan thir ntworks through mrgrs an alliancs. Thr is littl vinc that such movs narrow th cost gap with LCCs, but ntwork xpansion may hlp iffrntiat thir proucts an improv srvic. It also may incras thir ability to us ntwork markting vics to ampn LCC comptition. 18 Th airlin financial prformanc has improv substantially in 2010 an th inustry sms likly to b clos to brak-vn on omstic oprations for th yar. Still, th xprinc of th last ca suggsts that until lgacy carrirs can ithr clos th cost gap with LCCs or incras th pric prmium thy maintain, thy will likly hav ifficulty arning consistnt profits through th typical cycls in th airlin businss nvironmnt. This short papr obviously osn t sttl th issus surrouning airlin profitability. I bliv that th topic woul bnfit from much mor invstigation by inustrial organization conomists. 18 Thr is a lngthy litratur on th impact of airlin alliancs that xpan ntwork ffcts. S Armantir & Richar (2008) an citations thrin. 13

15 Appnix For th analysis in this papr, I inclu th following carrirs: Lgacy: (Amrican, Alaska, Braniff (pr-1990), Continntal, Dlta, Eastrn (pr-1992), Frontir (pr-1987), Frontir (post-1994), Northwst, Ozark (pr-1987), Pan Am, Pimont (pr-1998), Rpublic (pr- 1998), TWA, Unit, US Airways, Wstrn, Trump, National (pr-1983), Hawaiian, Aloha, Rgional: Air Miwst, Air Wisconsin, Amrican Eagl, Atlantic Coast, Atlantic Southast, Businss Exprss, CCair, Chautaqua, Colgan, Comair, Commutair, Continntal Exprss, Exprss Airlins, Grat Laks Aviation, Gulfstram Int, Msa, Msaba, PSA Airlins, Skywst, Trans Stats Airlins, Miwst Exprss, Horizon, Pinnacl, Businss Exprss, Wstair, Rpublic, Shuttl Amrica, GoJt, Compass, an Low-Cost Carrir: Pacific Southwst (pr-1989), Air California (pr-1988), Air Floria (pr-1985), Airtran, Amrica Wst, JtBlu, Miway, Morris, Nw York Air, Popl Exprss, Southwst, Spirit, Sun Country, Valujt, Rno Air, Jt Amrica, Virgin Amrica, Allgiant, USA3000, Amrican Trans Air, Kiwi, National (post-1998), Wstrn Pacific, Worl). All airlins ar inclu in aircraft capacity ata. Som of th vry small airlins ar not inclu in th financial an traffic ata. All pric lvl ajustmnts ar to 2009q4 using th all-urban Consumr Pric Inx. Figur 1: USDOT Burau of Transportation Form P-12 for incom statmnts an Form T-1 for ASMs. S Figur 2: Author s calculations from USDOT BTS Origin & Dstination Survy (DB1A an DB1B) an ffctiv ats of tickt tax changs. PFC changs ar availabl at rports/mia/airports.xls. S figur 3 scription for ajustmnt ma to account for baggag an chang/cancllation fs. Figur 3: Bas on th Markt Data Datast, which is riv from DB1A/DB1B an is scrib at Aftr ajusting all prics to 2009q4, for vry quartr, I stimat th OLS rgrssion AvgP ric rcnt = α 0t + α 1t NSDist r + α 2t NSDist 2 + ɛ, r whr an obsrvation is all of th passngrs flying rout r on carrir st c with n coupons in th trip uring quartr t. Each trip is a on-way journy btwn two airports that may b on-coupon (no chang-of-plan) or two-coupon (on chang-of-plan at an intrmiat airport. Both irctions on th rout ar collaps into th sam obsrvation. Roun-trips ar brokn into two on-way obsrvations. Th carrir st is a singl carrir obsrvation for on-coupon trips. It is th pair of carrirs for two-coupon trips (which ar oftn th sam carrir on both coupons). NSDist r is non-stop istanc btwn th airports. Th rgrssion is wight by passngrs in ach obsrvation. On avrag, ach rgrssion inclus 80,000 obsrvations covring about 7 million passngrs, with mor obsrvation an passngrs in latr yars I thn calculat th fitt valus for th tickts sol in th sam quartr on yar latr, ÂvgPric rcnt =ˆα 0t 4 +ˆα 1t 4NSDist r +ˆα 2t 4NSDist 2, r 14

16 an calculat th ratio of aggrgat rvnus actually pai to th aggrgat rvnus customrs woul hav pai if thy ha pai th fitt valu pric bas on th yar-arlir rgrssion paramtrs. Ths pric o not inclu baggag fs or tickt chang/cancllation T ickt Rvnus+Baggag Fs+Chang/Cancllation F s fs. So, for ach quartr, I calculat T ickt Rvnus for omstic oprations of all passngr carrirs inclu in th analysis (from Form P-12). Th chang in this ratio from on quartr to th sam quartr in th following yar is us to rscal th rvnu chang for ach quartr. I aggrgat th rvnu iffrnc ovr ach yar to gt th annual chang in prics. Figur 4: USDOE Enrgy Information Aministration, Figur 5: Dman is assum to b Q = A t P ɛ for all yars, t with ɛ = 1. Ral P is takn from th calculations for figur 3. Q is aggrgat omstic rvnu passngr-mils (from USDOT BTS Form T-1). Figur 5 prsnts A t by yar. Figur 6: USDOT BTS Form B-43 Figur 7: USDOT BTS Form T-1 Figur 8: USDOT BTS Form P-12 for incom statmnts an Form T-1 for ASMs. Th ajustmnt is on by first stimating th rgrssion ln(opxp/asm) ct = α 0 + α 1 ln(avghaul) ct + α 2 (ln(avghaul) ct ) 2 + c δ c + t γ t + ɛ on annual carrir-yar ata (for all commrcial passngr airlins with at last 20 parturs pr ay in th yar) for whr δ ar fix carrir ffcts an γ ar fix yar ffcts. Th opxp/asm for LCCs is thn normaliz to th avrag haul lngth of lgacy ˆα0+ˆα1ln(AvgHaul lg,t )+ ˆα 2 (ln(avghaul lg,t )) 2 ˆα 0 +ˆα 1 ln(avghaul LCC,t )+ ˆα 2 (ln(avghaul LCC,t )) 2 carrirs in th sam yar by multiplying by for ach yar, whr ˆα 0 =5.16, ˆα 1 = 1.92 an ˆα 2 = Th ajustmnt for haul lngth maks a significant iffrnc up to th mi-1980s whn LCCs flw much shortr istancs than lgacy/rgional carrirs, but maks almost no iffrnc in th last ca. Figur 9: Author s calculations bas on Markt Data atast (s figur 3 scription). This calculation corrcts for rout istanc in much th sam way as for figur 3, but th rgrssion is run only on lgacy plus rgional carrir obsrvations an th out-of-sampl priction an comparison is for LCC carrirs. For vry quartr, I stimat an OLS rgrssion of avrag far on rout istanc an istanc squar using only ata from lgacy an rgional carrirs. I thn calculat th fitt valus for th LCC obsrvations an th aggrgat rvnu iffrnc if LCC customrs ha pai th fitt valu pric insta of th actual LCC pric thy i pay. Th corrction for baggag an chang/cancllation fs is similar to figur 3. Figur 10: USDOT BTS Form P-12 for incom statmnts an Form T-1 for ASMs 15

17 Rfrncs Armantir, Olivir an Richar, Olivr. Domstic Airlin Alliancs an Consumr Wlfar, RAND Journal of Economics, Autumn 2008, Vol. 39(3), Bornstin, Svrin. Rpat-Buyr Programs in Ntwork Inustris, in Wrnr Sichl., Ntworks, Infrastructur, an Th Nw Task for Rgulation, , Ann Arbor, MI: Univrsity of Michigan Prss, Bornstin, Svrin an Ros, Nancy L. How Airlin Markts Work...Or Do Thy? Rgulatory Rform in th Airlin Inustry, National Burau of Economic Rsarch (Cambrig, MA) working papr #13452, Lrman, Mara. Ar Frqunt Flyr Programs a Caus of th Hub Prmium? Journal of Economics an Managmnt Stratgy, Spring 2008, Vol. 17(1), Lrman, Mara. Do Enhancmnts to Loyalty Programs Affct Dman? Th Impact of Intrnational Frqunt Flyr Partnrships on Domstic Airlin Dman, RAND Journal of Economics, Wintr 2007, Vol. 38(4),