Dual Ful Comtition in th British Enry Rtail Markts Richard Grn * Univrsity of Hull Businss School Univrsity of Hull Hull HU6 7RX Tl: 048 46570 Fax: 048 4666 Email: R.J.Grn@hull.ac.uk May 005 Rvisd: 6 May, 005 Prliminary Abstract This ar studis intraction btwn comanis sllin both as and lctricity to housholds in Grat Britain. British Gas is th formr monooly in th as markt, whil fiv comanis now own th formr rional lctricity monoolis. Switchin costs iv ths incumbnts a dr of markt owr. Th ar modls comtition btwn ths comanis, and shows that th incumbnts rtain lar markt shars dsit charin hihr rics than thir rivals. Kywords: intractions JEL: Elctricity rics, rtail comtition, as lctricity L94, L95 * Suort from th Lvrhulm Trust, throuh th award of a Phili Lvrhulm Priz, is ratfully acknowldd. I would lik to thank th MIT Cntr for Enry and Environmntal Policy for its hositality, and sminar audincs at th Univrsity of Surry and th Univrsity of California Enry Institut for hlful commnts.
. Introduction For th ast fiv yars, all British housholds hav bn abl to choos which comany thy buy thir as or lctricity from. Full rtail comtition was art of th lan for th lctricity rstructurin imlmntd in 990, with an oriinal tart dat of Aril, 998. Dlays in roducin th IT systms rquird, and th dsir to has in th markt onin ovr tim, mant that th first housholds did not t a choic until Stmbr 998, and th last in May 999. Full rtail comtition in as was not adotd as a olicy until 993, but th ilot schm ban in Aril 996, and th whol country was on to comtition by Aril 998. In 994, all domstic customrs bouht thir as from British Gas, and thir lctricity from thir local Public Elctricity Sulir. Thr wr fourtn of ths rional monoolis: th twlv formr Ara Elctricity Boards in Enland and Wals, and two vrtically intratd comanis in Scotland. At th tim of writin, ths comanis hav an avra markt shar of rouhly thr-fifths in thir hom aras (with som rional variation), and aroximatly half of housholds hav switchd sulir at last onc som hav switchd away, and thn back to thir local incumbnt. In th rulator s viw, this dr of switchin comars wll with a numbr of othr markts whr customrs normally buy from a sinl comany at a tim, such as bankin, tlcoms and insuranc. Elctricity and as ar homonous roducts, and chanin sulir dos not man that you chan th comany that dlivrs th roduct that is still th task of a rulatd ntwork comany. Whil som comanis offr scial faturs such as air mils or rn lctricity, it sms natural that comtition would mostly b basd uon ric. Th Brtrand modl dos not sm to aly, howvr th formr monoolis (th incumbnts) ar sinificantly mor xnsiv than thir rivals in narly vry markt smnt. Ths ric diffrntials iv a motivation for switchin, althouh th numbr of consumrs who mana to switch to a mor xnsiv sulir shows that (actual) diffrntials cannot b th only motivation. Anothr ossibl motivation is th convninc from buyin both as and lctricity from th sam comany, and rcivin a sinl bill. Ofm (004,. 60) Th subjct of this ar is th nry markt in Grat Britain (Enland, Scotland and Wals), as Northrn Irland has no rtail comtition in lctricity, and vry fw as customrs. Waddams Pric (004) shows that almost 40% of low-incom customrs who had switchd sulir in 000 had movd to a mor xnsiv tariff than thy wr rcivin from thir rvious sulir.
rorts that 80 r cnt of consumrs who switch mov on to a dual ful dal. British Gas is now th larst sulir in th houshold lctricity markt, whil th incumbnts in lctricity ar th only othr comanis with lar markt shars in as. Whil Ofm dos not ublish th rlvant rional data, intuition surly imlis that thy will hav thir larst shars of th as markt in th aras whr thy ar th lctricity incumbnt. Tyically, a comany will strss th low ric it can offr for its non-traditional ful, whil avoidin th subjct of th hih ric it is still charin as an incumbnt. Th low ric in th non-traditional markt may not lav much of a rofit marin (which is not to say that it is actually rdatory), but has th rat advanta of hlin to rtain customrs in th traditional markt, whr marins rmain much hihr. Dsit th imortanc of th intractions btwn as and lctricity, and an incrasin litratur on th British rtail markts, I am not awar of any acadmic work which studis dual ful comtition. This ar is an attmt to fill that a. Th nxt sction ivs mor information on th markt, and on th acadmic litratur to dat. Sction 3 outlins a siml duooly modl, to illustrat th ky trad-offs in a tractabl modl with analytical solutions. Sction 4 movs to a mor comlicatd modl without analytical solutions, which nvrthlss shows th intraction of incumbnts and ntrants across svral markts. Sction 5 comars th rdictions of th modls with data from th markts. Sction 6 concluds.. Ovrviw of th markts Thr ar 4 million houshold lctricity customrs in Grat Britain, and million as customrs many housholds in rural aras, and som in towns, ar not connctd to th as mains. Th as and lctricity industris startd in th nintnth cntury with a mix of rivat and municial ownrshi, and wr nationalisd in 949 and 948 rsctivly. In both cass, distribution and sals wr oranisd on a local basis, with fourtn Ara Elctricity Boards and twlv Ara Gas Boards. Th British Gas Cororation was formd in 97 to ovrs th task of convrtin th country from coal as to natural as, and th Ara Gas Boards lost thir sarat idntitis. British Gas was rivatisd in 986, with rlativly liht rulation, and (notional) comtition to sll as to lar customrs. This comtition rovd inffctiv, sinc British Gas could st both rics to customrs and th transmission char any rival would hav to ay it in ordr to comt with it. Prssur from th 3
Offic of Fair Tradin and an inquiry by th Monoolis and Mrrs Commission mad comtition in th industrial markt mor ffctiv, but th nd to st a ric control for as transmission and distribution ld to a scond MMC inquiry in 99-93. This inquiry rcommndd that British Gas should b slit into a ilin comany and a rtailin comany, in ordr to nsur a lvl layin fild for othr rtailrs, and that onc th slit had takn ffct, comtition should b introducd in rtailin to housholds. Th ovrnmnt dcidd to introduc houshold comtition on a fastr timtabl than that rcommndd by th MMC, and not to slit u British Gas, in cas this dlayd comtition. In ractic, British Gas chos to slit itslf in two shortly aftrwards. BG lc ownd th ilins 3 and most of th xloration and roduction assts, tradin abroad as British Gas, whil Cntrica lc took ovr th suly businss, tradin in th UK as British Gas. From this oint onward, I will us British Gas to rfr to Cntrica. Suly comtition startd with a ilot schm in th southwst of Enland in Aril 996. It was radually xtndd to th rst of th country durin 997 and 998, so that vry as customr was abl to choos thir sulir by May 998. From th arlist days, th most ffctiv comtitor to British Gas aard to b th local lctricity comany in ach rion, which was a caus of som tnsion, ivn that th lctricity markt was not yt on to comtition. Th lctricity industry was rivatisd aftr th as industry. Th ara boards, now known as Rional Elctricity Comanis (RECs) wr rivatisd in 990, whil two lar nrators, National Powr and PowrGn, and two Scottish comanis wr sold in 99. Partly in rsons to th roblms that had followd as rivatisation, th lctricity schm had slit transmission from th rst of th industry, introducd xlicit rulation for distribution ntwork chars, and st a timtabl for introducin full comtition in suly. Th first has of markt onin, for lar industrial customrs, was rlativly smooth, but th scond has, for mdium-sizd customrs, was a fiasco. Customrs wr allowd to chan sulir bfor thy had th ncssary mtrin and communications quimnt in lac and ristrd, and many thrfor rcivd incorrct bills. It was clar that th industry had not had an aroriat forum to dsin systms that would work, ivn th numbr of comanis that xctd to los from th introduction of comtition. 3 Ths wr subsquntly also slit off, and ar now ownd, tothr with th lctricity transmission rid, by National Grid Transco. 4
To rvnt a similar fiasco whn th markts wr ond to domstic customrs, th rulator took control of a lar-scal roram that was to cost about ¾ billion. To nsur that th IT systms could co, ach comany would on its markt in stas ovr six months, startin with an ara containin tn r cnt of its customrs (basd on ost cods (zi cods), so that th boundaris would b clar to all). Non of th comanis IT systms wr rady in tim (thr wr rumours that som comanis had xctd th Labour ovrnmnt lctd in 997 to sto th roram, and had dlayd sndin mony accordinly). Th first markts wr ond in Stmbr 998, and th final ons in May 999. By that tim, th industry s cororat structur was startin to chan. Scottish Powr had bouht Manwb, on of th RECs, in 995. Hydro-Elctric mrd with Southrn Elctric in 998, formin Scottish and Southrn Elctricity. National Powr and PowrGn had trid to mr with RECs in 995, and wr blockd by th ovrnmnt, but wr allowd to do so in 999. Eastrn Elctricity, latr known as TXU aftr its takovr by a US firm, had intratd into nration and bouht th suly businss of Norwb. Scottish and Southrn bouht Swalc, th suly businss srvin th south of Wals. Elctricité d Franc bouht London Elctricity, thn th SWEB suly businss from th south-wst of Enland, thn Sboard. National Powr slit itslf into two, with th rdominantly UK-basd businss takin th nam Innoy and tradin as nowr. Innoy latr took ovr two mor REC suly businsss, Northrn and Yorkshir. Innoy was itslf takn ovr by RWE of Grmany, whil PowrGn was takn ovr by E.On. Finally, PowrGn took ovr th suly businsss and owr stations ownd by TXU, whn th US arnt withdrw from most of its Euroan orations. Tabl summariss ths amalamations, whil fiur shows th aras concrnd. 5
Tabl : Nativ customr numbrs, by ara and comany Nam now Ara(s) Nativ Customrs (million) 994 999 00 003 British Gas All GB EdF Enry Exortr to GB 0.8 3.0 4.8 London.8 SWEB.. Sboard.8.8.8 PowrGn Gnrator 0.. 7.0 E Midlands. Eastrn.9.9 4.9 Norwb.0.0 nowr Gnrator 0.0 3.9 5.3 Midlands.0 Yorkshir.9.9 Northrn.4.4.4 Scottish and Southrn.4 3.0 3.9 3.9 Southrn Scottish Hydro 0.6 Enry Swalc 0.9 0.9 Scottish S Scotland.6.8.8.8 Powr Manwb. Sourc: Ofm (004). Not that th fiurs rflct mrrs, but not rowth in total customr numbrs ovr tim. Onc thir markts ond, th incumbnts in both industris startd to los customrs. Thir rics wr initially rulatd, with ric cas st in rlation to thir avra costs. In both industris, th comanis had ntrd into lon-trm wholsal urchas contracts at rics abov thos subsquntly rulin, and so it was asy for comtitors to undrcut thm on th basis of currnt wholsal rics. In th lctricity industry, th most imortant comtitor was British Gas, whil in th as industry, th most imortant comtitor in ach rion was th incumbnt lctricity comany. Rional markt shars ar not ublishd, howvr, and so tabl can only 6
iv markt shars for a tyical lctricity rion on th basis of fiurs for th incumbnts shars in thir own rions, and British Gas shar of th markt as a whol. Tabl : Markt shars in a rion, basd on GB avras Industry Ty of comany Markt shar 994 S 999 S 00 Dc 003 Gas British Gas 00 75 67 6 All ntrants 0 5 33 39 Elctricity Incumbnt 00 90 70 59 British Gas 0 7 4 0 All othr ntrants 0 3 7 Sourc: Ofm (004) On this basis, th markt still aars to b vry concntratd, whil th national markt shars in tabl 3 and 4 imly a much mor framntd markt structur. 7
Tabl 3: National markt shars in houshold as suly Nam now Formr comany Markt shar (r cnt) 994 S 999 S 00 Dc 003 British Gas British Gas 00 75 67 6 EdF Enry EdF 0 n/a 5 London 0 SWEB 0 n/a Sboard 0 n/a PowrGn PowrGn 0 n/a 4 E Midlands 0 Eastrn 0 3 6 Norwb 0 nowr National Powr 0 n/a 6 9 Midlands 0 Yorkshir 0 n/a Northrn 0 n/a Scottish and Southrn 0 3 5 7 Southrn Scottish Hydro 0 Enry Swalc 0 Scottish Scottish Powr 0 n/a 5 6 Powr Manwb 0 All othrs 0 n/a a too small to b saratly rortd by Ofm, includd in all othrs Sourc: Ofm (004) for British Gas; various Ofm mrr consultations 8
Tabl 4: National markt shars in houshold lctricity suly Nam now Formr comany Markt shar (r cnt) 994 S 00 Dc 003 British Gas British Gas 0 7 4 EdF Enry EdF 0 0 4 London 8 SWEB 5 Sboard 8 6 PowrGn PowrGn 0 8 E Midlands 9 Eastrn 5 Norwb 9 nowr National Powr 0 4 5 Midlands 8 Yorkshir 8 Northrn 6 5 Scottish and Southrn 0 4 4 Southrn Scottish Hydro Enry Swalc 4 Scottish Scottish Powr 7 0 Powr Manwb 5 All othrs 0 Sourc: Ofm (004) and author s stimats 9
Tabl 5: National markt shars in houshold nry suly, Dcmbr 003 Comany Gas Elctricity As th As an As th As an Total Incumbnt Entrant Incumbnt Entrant British Gas 6 0 0 4 4 EdF Enry 0 5 4 Nowr 0 9 3 5 PowrGn 0 6 5 Scottish & Southrn 0 7 3 4 Scottish Powr 0 6 7 4 All othrs 0 0 All comanis 6 40 58 4 00 Sourc: Author s stimats from Ofm data Tabl 5 braks down ths national markt shars into th customrs which a comany sulis as an incumbnt, and thos sulid as an ntrant: th formr lctricity incumbnts hav bn far lss succssful than British Gas in attractin lctricity consumrs from outsid thir traditional aras, individually, and vn as a rou. It is not ossibl to brak down th scond column of th tabl to divid th lctricity comanis as customrs into thos livin whr th comany is th lctricity incumbnt, and thos whr th comany is an ntrant for both fuls. It would b surrisin, howvr, if th lctricity comanis did not find most of thir as customrs in th aras in which thy ar incumbnts. Thr has bn a substantial amount of acadmic work on th UK rsidntial nry markts, althouh almost all of it looks ithr at as or at lctricity, rathr than takin a dual ful rsctiv. Furthrmor, thr has bn littl work on markt shars, and most authors concntrat on th attrn of rics and of discounts, rlativ to incumbnts, and th switchin costs that may xlain ths. Otro and Waddams Pric (00) xlord th attrn of rics across aymnt tys shortly aftr th lctricity markt was ond to comtition. Entrants nrally undrcut incumbnts dirct dbit tariffs, but wr (on avra) mor xnsiv for raymnt customrs. In many cass, of cours, an ntrant in on rion is an incumbnt lswhr, and ths comanis raymnt tariffs wr 0
closr to thir dirct dbit tariffs whr thy wr incumbnts than whr thy wr ntrants. Otro and Waddams Pric sust that this is du to ric discrimination, and in articular to olitical rssur to k raymnt tariffs down, rlativ to th costs involvd in srvin raymnt customrs. Entrants wr unwillin to comt hard for raymnt customrs at thos ric lvls, and whil thy wr offrin bttr rics for dirct dbit customrs, customr inrtia allowd th incumbnts to rtain hih markt shars dsit hihr rics. Salis and Waddams Pric (004) show that incumbnts continud to char hihr rics for dirct dbit and standard crdit customrs in 00, whil thir chars for raymnt customrs wr not sinificantly diffrnt from ntrants chars. Rional diffrncs in transmission and distribution costs wr translatd almost on-for-on into th tariffs for dirct dbit and standard crdit customrs, whil th rlationshi btwn ths costs and raymnt chars was much wakr. Salis and Waddams Pric intrrt this as rflctin wakr comtition in th raymnt sctor. Giulitti t al (004) show that lctricity rics rmaind disrsd, and customrs wr abl to ain from switchin to a nw sulir, svral yars aftr th markts wr ond to comtition. Thy sust that this shows that consumrs still hav sinificant switchin costs. Giulitti t al (003) assss th xtnt of switchin costs in th as industry, and th factors affctin thm. Thy show that British Gas could incras its rics sinificantly rlativ to ntrants rics (and its costs) whil still rtainin many of its customrs, and that to do so would b rofit maximisin. Thy calculat what this imlis for social wlfar, ivn th ains to consumrs who switch, th losss to thos who do not, and th costs that comanis incur in rsuadin consumrs to switch. Ovrall, svral of thir scnarios imly that as comtition has rducd wlfar. Thir stimats of th rofit-maximisin mark-us ar similar to thos of Grn (000), who uss a calibratd thortical modl to calculat lctricity comanis bst rsonss to utility-maximisin customrs with switchin costs. Sturluson (003) diffrntiats btwn sarch and switchin costs, and stimats both mirically for Swdish data. H also shows how ths conctually diffrnt costs can hav diffrnt imacts uon comanis. Whil all of ths ars ar basd on utility-maximisin, rational, consumrs, Waddams Pric (004) shows that, rmarkably, many amon a saml of low-incom
customrs from 000 who had switchd had actually mad thmslvs wors off, ayin mor aftr thy switchd to a nw sulir than thy would hav don, had thy stayd with thir formr sulir. Whil som consumrs may hav switchd to a sulir that was char at th tim that th dcision was mad, but had subsquntly raisd its rics, twnty four out of thirty nin as customrs who had just arrand to switch sulirs wr makin thmslvs wors off, as wr twlv out of twnty nin lctricity customrs. At th vry last, ths rsults imly that w should xct to find that non-ric factors ar imortant in consumrs dcisions about nry sulirs. Non of ths ars considrs th rol of dual ful comtition, in which customrs buy as and lctricity from th sam sulir. Such comtition has stadily rown in imortanc, howvr, and Ofm now stimat that 80 r cnt of switchin consumrs do so with a dual ful dal ithr buyin a scond ful from a comany that alrady sulis thm, or buyin both fuls from th sam (nw) comany. British Gas succss in lctricity suly has bn drivn by th comany s ability to sll owr to its xistin customrs, whil tabl 5 rvals how (rlativly) unsuccssful th lctricity comanis hav bn in sllin lctricity to customrs in othr aras. Whil th data ar not ublishd, it sms obvious that thir much ratr succss in th as markt must com from th sam kind of cross-sllin. Th ffcts of such cross-sllin on consumr choics and comany bhaviour ar th subjct of this ar. 3. A siml modl W will start to xlor th imact of as-lctricity intractions with a dlibratly siml modl, which ivs analytical solutions. W will assum that thr ar only two activ comanis in any ivn rion British Gas and th lctricity incumbnt. Normalisin th siz of th oulation to on, a roortion of consumrs, α, buy just on unit of lctricity, whil (-α) buy both γ units of as and on unit of lctricity. (Thr ar no as-only customrs.) Th lctricity comany s ric for lctricity is. British Gas ric for as is. Th lctricity comany slls as at a ric of, whil British Gas slls lctricity at a ric of. Th subscrit rfrs to th roduct, th surscrit (whr
rsnt) to th main roduct of th sllin comany. Th comanis marinal costs of as and lctricity ar ivn (in th sam ordr) by c, c, c and c. With a liquid wholsal markt and common carria on th ntworks, w would xct th cost of actually srvin th customr with a ivn roduct to b vry narly th sam for both comanis, but th cost of acquirin a customr is likly to mak th ntrant s ovrall cost hihr than th incumbnt s in ach industry. W trat th customr, rathr than th kwh, as th marinal unit, and so it is aroriat to trat ths acquisition costs as marinal. Whil as and lctricity ar homonous roducts, w mov away from ur ric comtition by assumin that ach customr has a rlativ rfrnc btwn British Gas and th lctricity comany this may rflct thir ast xrinc with th two comanis, or th ffctivnss of th comanis marktin and advrtisin activitis. W will modl this as a variabl β, which rflcts th rlativ rfrnc for British Gas, uniformly distributd on [, ]. A consumr who switchs on of thir ful sulis to British Gas will add β to thir utility, whil a consumr who switchs to th lctricity comany will subtract β from thir utility. (If thy rfr th lctricity comany, of cours, thir valu of β will b nativ, and so this will rais thir utility.) A customr who buys both fuls and switchs on of thm incurs a switchin cost of s. A customr who buys only lctricity and switchs to British Gas incurs a switchin cost of t s. Thir switchin cost may b larr bcaus thy will b movin to a comany that thy may not hav dalt with bfor; β is now a masur of th customr s attitud to th lctricity comany, and to advrtisin by British Gas, rathr than to th comany s ast rformanc in srvin this articular customr. A customr who buys both as and lctricity will hav th followin utility: v γ if thy buy both roducts from th incumbnt sulirs v γ β s if thy buy both roducts from British Gas () v γ β s if thy buy both roducts from th lctricity comany Siml utility maximisation imlis that th lctricity comany will sll both as and lctricity to customrs with valus of β γ ( ) s both fuls to customrs with valus of β s ( ), that British Gas will sll, and that customrs with 3
intrmdiat valus will buy as from British Gas and lctricity from th lctricity comany. A customr who buys only lctricity will hav th followin utility: v v β t if thy buy from th lctricity comany if thy buy from British Gas () In this cas, British Gas will attract customrs with valus of t ( ) β. Aratin across both tys of customr, w t th followin sals fiurs: Th lctricity comanyslls : q ( α )[ s ( )] α[ t ( )] ( α )[ s γ ( )] γ q British Gas slls : q q γ ( α )[ s ( )] α[ t ( )] ( α )[ s γ ( )] units of units of as units of lctricity and units of as lctricity and (3) This st-u ivs us convnint dmand curvs, which ar linar in th two comanis rics for th ful. Thr ar no cross-ful ffcts, howvr. Th modl in th nxt sction allows for th ossibility that British Gas (for xaml) miht cut th ric of lctricity in ordr to rtain mor as customrs. Addin a sub-rou of consumrs who hav no switchin costs would chan th dmand curvs so that th quantity of ach ful dndd uon all four rics. In quilibrium, howvr, this chans th lvl of rics, but not th attrn of ric diffrntials, which only dnd uon th aramtrs scific to ach markt. Whil w could add othr comlicatin factors (such as a non-uniform distribution of β) to t a richr attrn of ric diffrntials, it is bttr to stick with th simlr modl for th rsnt sction. c c British Gas rofits ar ivn by ( ) ( ) π q q, whil th lctricity comany s rofits ar ivn by ( ) ( ) π c q c q. If w diffrntiat with rsct to ach ric in turn, w t raction functions: 4
( ) ( ) (4) British Gas for and ) ( and th lctricity comany, for and ) ( γ γ α α γ γ α α c s t s c c s t s c W can combin th raction functions and solv to t: (5) for as 3 3 and 3 3 and for lctricity, 3 ) ( 3 and 3 ) ( 3 γ γ γ γ α α α α s c c s c c t s c c t s c c W thus find that th ric diffrntials dnd uon th diffrncs in cost btwn th ntrant and th incumbnt, and th avra switchin cost in ach markt: ( ) for as 3 3 and for lctricity, 3 ) ( 3 c c s c c s t s γ α (6) Not that th amount of as bouht rlativ to lctricity ntrs into quation (6), but is basically a scalin factor th diffrnc in as bills (as oosd to rics) will qual on third of th diffrnc in costs, lus two-thirds of th switchin cost for that markt, just as in lctricity. As arud abov, th ntrants ar likly to hav hihr marinal costs than th incumbnts, onc w includ customr acquisition costs, and so th quations ar writtn to mak th final fractions ositiv. From quation (6), w can driv quantitis for ach firm: 5
Th lctricity comanyslls : q ( c c ) ( α ) s αt 3 3 ( α ) γ ( c c ) γ q British Gas slls : q q γ ( c c ) ( α ) s αt 3 3 ( α ) γ ( c c ) 3 3 s 3 s 3 units of units of units of lctricity and units of lctricity and as as (7) Th rsults of th modl can b summd u in Proosition : An incras in th lvl of switchin costs for consumrs takin both fuls will rais th incumbnts rics and markt shars, and rduc th ntrants rics and markt shars. An incras in th lvl of switchin costs for customrs takin only lctricity will rais th lctricity comany s ric and markt shar in lctricity (and rduc thos of British Gas). An incras in th roortion of customrs takin only lctricity will rais th lctricity comany s ric and markt shar in lctricity. An incras in th incumbnt s costs will rais both comanis rics and rduc th incumbnt s markt shar. An incras in th ntrant s costs will rais both comanis rics and incras th incumbnt s markt shar. An incras in th cost diffrntial btwn th two comanis in ithr markt (assumin th incumbnt to hav th lowr costs) will rduc th ric diffrntial btwn thm, and rais th incumbnt s markt shar. Proof: From insction, or diffrntiation, of quations (5) to (7). This modl was dlibratly kt siml, but shows th ky rol layd by consumrs switchin costs (ral or rcivd) and cost diffrncs btwn comanis. Th ratio of as to lctricity consumrs varis across Grat Britain, and this will allow us to tst roosition in th nultimat sction of th ar. First, howvr, w will mov to a lss simlifid modl of th markts, but on that dos not allow for analytical solutions. 6
4. A numrical modl Th modl of sction 3 was kt siml, to nsur analytical solutions, but only includd two comanis. Whil British Gas and th incumbnt lctricity comany tyically srv around thr-quartrs of th housholds in ach rion, and th othr comanis individual markt shars ar much smallr, thir collctiv markt shar is sufficint to justify buildin a modl that includs thm. In this sction of th ar, w us a loit modl of consumr choic to xlor how rics and markt shars vary with switchin costs and othr markt charactristics. W follow Bsanko t al s (990) drivation of th loit modl, but assum that thr is no outsid otion all consumrs will buy nry from th sulirs that w modl. W assum that thr ar m comanis, and that numbr is th lctricity incumbnt. Th consumrs utilitis ar similar to thos ivn by quations d d di () and (), in that consumr i rcivs a utility of v k s η if sh buys lctricity and as from comany k, whr v is a common symmtric valu aind k k whn buyin nry from any of th comanis, d k is comany k s dual-ful ric for as and lctricity combind, d s k is th cost (to any consumr) of switchin to comany k, and di η k is a consumr- and comany-scific valuation of buyin from comany k. If th consumr buys both fuls saratly from th incumbnts, howvr, di hr utility will b, whr is th incumbnt lctricity comany s v η0 di ric for lctricity alon, is British Gas ric for as alon, and is th customr-scific valuation of buyin from th two incumbnts. In this cas, of cours, no switchin costs ar incurrd. For customrs who only urchas lctricity (which is an xonous dcision, as all customrs connctd to th as mains ar assumd to buy as), th utility from i i buyin from th incumbnt is ivn by whr is th utility from buyin lctricity (alon) from th incumbnt. A customr who buys lctricity from i an ntrant, k, rcivs a utility of v k s η, whr is that comany s ric for lctricity, i η k s k v η is th cost of switchin to that comany for lctricity alon, and is th idiosyncratic valuation of buyin lctricity from that comany. k k k η η 0 7
W assum that th idiosyncratic valuations ar idntically and indndntly distributd accordin to a ty- xtrm valu distribution. This has location aramtr zro and scal aramtr θ. Th distribution and dnsity functions ar ivn by F( x) x f ( x) θ x / θ { } x / θ F( x) Bsanko t al intrrt θ as a masur of divrsity, or th xtnt to which consumrs idiosyncratic valuations affct thir urchas dcisions. With a low valu of θ, consumrs lac rlativly littl wiht on comany-scific factors, and mor wiht on ric and switchin cost. A hih valu of θ corrsonds to a cas in which comany-scific factors ar imortant. In this alication, it maks mor sns to think of θ as a masur of th ffctivnss of marktin and sals camains in rducin consumrs switchin costs. Whn w calibrat th modl, th bst match to th data for 003 coms from sttin rlativly hih lvls of switchin costs so hih that no consumr would switch on ric savins alon. If w intrrt θ as a masur of th comanis ability to rduc ths switchin costs (tyically by contactin customrs with dtails of th savins on offr), thn w will find that in aras with hih valus of θ, customrs ar mor rsonsiv to such contacts, and mor switchin taks lac. Rturnin to th dtails of th modl, ach consumr will choos to buy from th comany that offrs hr th ratst lvl of utility, takin ric, switchin cost, and idiosyncratic comonnt tothr. Followin Bsanko t al, markt shars of dual-ful customrs ar x x d k d 0 (, m) x (, m) x d d {( v k sk )/ θ} {( v )/ θ} whr (9) D m k x / D / D d d {( v s )/ θ} x{ ( v )/ θ} k k (8) for customrs who switch for customrs who stay with both incumbnts Th markt shars of lctricity-only customrs ar ivn by 8
x (, m) x k x (, m) x {( v k sk )/ θ} {( v )/ θ} / E whr (0) E m k x / E {( v s )/ θ} x{ ( v )/ θ} k k for customrs who switch for customrs whostay with thincumbnt Givn th siz of ach markt, and th variabl costs of suly, it is straihtforward to writ down th rofit functions for ach firm: π d d d [ ( c c )] x, m) ( α ) [ c ]( x (, m) ( α ) x (, ) α ) ( 0 m () Th incumbnt lctricity comany arns th diffrnc btwn its dual-ful ric and th cost of sulyin two fuls, on its shar of th dual-ful customrs; tothr with th diffrnc btwn its lctricity-only ric and cost, on th customrs who tak both fuls and do not switch, and its shar of th lctricity-only customrs. π d d d [ ( c c )] x (, m) ( α ) [ c ] x0 (, m) ( α ) [ c ] x (, m) α () Th incumbnt as comany arns th diffrnc btwn its dual-ful ric and th cost of sulyin two fuls, on its shar of th dual-ful customrs; th diffrnc btwn its as-only ric and cost, on th customrs who tak both fuls and do not switch; and th diffrnc btwn its lctricity-only ric and cost, on its shar of th lctricity-only customrs. d d [ ( c c )] x (, m) ( α ) [ c ] x (, m α π ) (3) k k k k k Th othr comanis arn th diffrnc btwn thir dual-ful ric and th cost of sulyin two fuls, on thir shar of th dual-ful customrs; tothr with th diffrnc btwn thir lctricity-only ric and cost, on thir shar of th lctricity-only customrs. Bsanko t al dscrib two solution concts monoolistic comtition, in which firms tak E as ivn, and th Nash quilibrium, in which firms tak account of 9
th imact of thir rics on D and E. On fatur of thir monoolistic quilibrium is that all firms hav th sam mark-u, which is dfinitly not a fatur of th British nry markt. W thrfor calculat th Nash quilibrium, which has to b don numrically. To calculat such an quilibrium, w nd to calibrat th modl. Th consumr rotction body nrywatch ublishs data on th rics chard in ach of th fourtn rional markts by ach comany, for lctricity, as, and a dual-ful acka, at ach of thr consumtion lvls 4 and for thr aymnt tys. Th aymnt tys ar standard crdit, for consumrs who rciv a bill vry thr months and thn ay it; dirct dbit, for consumrs who ay with an automatic bank transfr ach month (tyically a fixd amount throuhout th yar, basd on an stimat of thir consumtion); and raymnt, for consumrs who hav to char thir mtr, usually with an lctronic card, bfor takin any ful. Dirct dbit is th chast aymnt mthod (it involvs th lowst costs for th comanis) and raymnt th most xnsiv. 5 Usin rics for Aril 003, w calculat th ric in ach rion for ach aymnt ty as th avra of th total bills at th thr consumtion lvls. W assum that thr ar only svn ossibl outcoms for dualful customrs: buyin both fuls from th oriinal incumbnts, buyin both from British Gas, or buyin both from on of th fiv rmainin lar lctricity comanis. This involvs two simlifyin assumtions: firstly, thr frin sulirs (with a collctiv markt shar of undr on rcnt) ar inord, and scond, w cut down th numbr of rmutations by assumin that all switchrs mov to a dual ful dal. Th Dartmnt of Trad and Industry ublishs th incumbnt s quartrly markt shar in ach of th fourtn lctricity rions, and in ach of British Gas twlv oratin rions, for ach aymnt ty. Usin data for as consumtion in ach local ovrnmnt ara, and assumin that th roortion of customrs switchin within a British Gas rion is uniform across local ovrnmnt aras, w can stimat th numbr of as customrs, and th roortion who switch, in ach lctricity rion. 4 nrywatch dfins low consumtion as,650 kwh of lctricity and 0,000 kwh of as r yar; mdium consumtion as 3,300 kwh of lctricity and 0,500 kwh of as r yar, and hih consumtion as 4,950 kwh of lctricity and 8,000 kwh of as r yar 5 In ractic, dual-ful dals ar not availabl for customrs with r-aymnt mtrs, and th comanis rics for sarat sulis of as and lctricity wr usd instad. 0
W assum that th cost of switchin varis btwn rions and btwn th tys of comanis, but that th ratio of costs btwn comany tys within a rion dos not nrally chan. By comany ty, w man that th cost may dnd uon whthr th comany alrady has a rlationshi with th customr, but will not dnd uon th idntity of that comany. W allow th cost of switchin lctricity alon to b lss than th cost of dual-ful switchin (and assum that British Gas has no switchin cost advanta for ths customrs, with whom it has no xistin rlationshi). In ractic, th constraint that th lctricity-only cost was lss than or qual to th cost of dual-ful switchin turnd out to b bindin, whil a constraint that dual-ful switchin to ithr incumbnt was lss costly than switchin to a comltly nw comany was not bindin. Kin th ratio of costs within a rion th sam across rions mant that th cost of switchin to a non-incumbnt comany was always thr tims that of switchin to th lctricity incumbnt, whil th cost of switchin to British Gas was nrally two-thirds that of switchin to th lctricity incumbnt. (Ths ratios wr chosn by Microsoft Excl s solvr, tothr with th switchin cost and valus of θ.) Th xction was th cost of switchin to British Gas in th Scottish Hydro rion, whr th lowr roortion of as customrs was likly to lac th comany at a disadvanta comard to th lctricity incumbnt. W assumd that th ffctivnss of th comanis fforts to ovrcom switchin cost was th sam in ach rion, but varid with th customr s aymnt ty. In ractic, th valus of θ for standard crdit and r-aymnt customrs wr similar (4 and 0), whil th valu for dirct dbit customrs was 50% hihr (7). Th switchin costs and ffctivnss aramtrs wr chosn to minimis th sum of squard dviations btwn th incumbnts actual markt shars and thir rdictd shars for ach aymnt ty and ach ful. In most rions, th cost of dual-ful switchin to British Gas was around 80, to an lctricity incumbnt was around 0, and to a nw comany was rouhly 360. Th last fiur also alis to lctricity-only switchin. Ths fiurs should b contrastd to annual bills for two fuls of btwn 480 and 690, and for lctricity of btwn 00 and 330. Clarly, switchin costs of this manitud will nsur that customrs stay with th incumbnts unlss sulirs can find ways of ovrcomin th costs by ffctiv marktin.
Tabl 6 shows th stimatd switchin costs and th roortion of customrs still with th incumbnt comanis, avrad across thr aymnt tys and two fuls, xct for th Hydro-Elctric rion, whr it is for lctricity alon. 6 Th rions ar dividd into thr rous. Thr ar thr rions with articularly hih lvls of switchin, all in th north of Enland, and ths had rlativly low switchin costs. It is ossibl that th comanis snd mor on marktin in this art of th country, sinc th cost of transortin as is rlativly low, but th imlicit rvnus from sllin as ar vry clos to thos in othr rions, imlyin hihr rofit marins. 7 Th north of Scotland has th lowst lvl of switchin. Th stimatd switchin cost for lctricity comanis was hihr than lswhr, and th cost of switchin to British Gas, which was allowd to vary from its normal ratio, was much hihr. 6 Gas switchin fiurs ar only ublishd for Scotland as a whol, althouh switchin bhaviour in lctricity diffrs dramatically btwn th north and th south. Th calibration thrfor matchd th avra of th fiurs calculatd for th two rions to th national lvl of switchin, for ach as aymnt ty. Th rsultin fiurs wr a clos match for lctricity, but would hav aard a oor match for as in th north of Scotland. 7 By imlicit rvnus from sllin as w man th diffrnc btwn a comany s dual ful tariff and its tariff for lctricity alon. This oint aros in discussion with Danil Sroi.
Tabl 6: Estimatd switchin costs and incumbnt shars Switchin costs Incumbnt shars Elc. Inc. BG Othr actual fittd Low switchin Scottish Hydro 70.0 49.7 507.9 74 76 Avra switchin Southrn 37.6 87.3 4. 67 67 London 37.0 86.9 409. 66 66 SWEB.4 77. 36.7 64 65 Scottish Powr 6. 80.0 376.7 64 64 Eastrn.7 77. 363.5 64 64 Swalc 4.4 79.0 37.6 64 64 Sboard 0.4 76.4 359.6 63 63 Midlands.4 77.7 365.7 63 63 East Midlands.8 70.9 334.0 6 6 Manwb 6.0 73.6 346.5 6 60 Hih switchin Norwb 04. 66. 3. 59 59 Yorkshir 06.9 67.8 39.4 58 58 Northrn 97.9 6. 9.5 56 56 Havin stimatd th ky aramtrs, w wr thn abl to solv th modl numrically. W modlld thr rions on with low ffctivnss and switchin, on with intrmdiat, and on with hih ffctivnss and switchin. W assum ach rion has th sam numbr of dual-ful customrs. Th rlativ numbr of lctricity-only customrs also varid btwn rions, bin hihst in th lowswitchin rion and lowst in th hih-switchin rion. Comanis wr allowd to choos a dual-ful and an lctricity ric for ach rion, but British Gas had a sinl nationwid ric for as-only customrs, in lin with its xistin ractic. Th marinal cost of an lctricity customr was st to 00, and of a as customr to 3
300. 8 Usin th built-in solvr in Microsoft Excl, ach comany in turn maximisd its rofits, holdin th othr comanis rics constant, until th rics convrd on a Nash Equilibrium. Th rics and markt shars for th thr tys of rion ar shown in Tabl 7. 8 In ractic, th cost of sulyin customrs varis from rion to rion, for both lctricity and as, ivn rional diffrntials in transmission and distribution chars and takin such (stimatd) variations into account is a ossibl xtnsion to this modl. 4
Tabl 7: Simulatd rics and markt shars, for standard crdit customrs Lowswitchin Normal Hihswitchin Cost from incumbnts 037 934 903 Pric of as from BG 50 50 50 Dual-ful ric British Gas 697 709 697 Elctricity incumbnt 748 694 684 Elctricity ntrant 6 63 65 Elctricity-only ric British Gas 3 34 36 Elctricity incumbnt 536 433 40 Elctricity ntrant 3 34 36 Elctricity markt shar British Gas 8.9 30.6 9.8 Elctricity incumbnt 58.7 4.8 38.4 Elctricity ntrant (ach) 5.6 6.6 7.9 Gas markt shar British Gas 4. 45.4 4.5 Elctricity incumbnt 36. 8.5 6.9 Elctricity ntrant (ach) 5.6 6.5 7.9 Siz Dual-ful Elctric-only 0.7 0. 0. θ 5 5 5 Switchin cost Elc Incumbnt 70 0 00 British Gas 50 80 67 Othr comany 50 360 300 All of ths rics ar wll abov marinal cost, althouh it is worth notin that th comanis also hav som non-marinal costs that thy will nd to rcovr. It is clar that th incumbnt comanis can afford to k thir rics wll abov thos of 5
thir rivals Marins ar ratr for lctricity-only customrs than for dual ful customrs, in art bcaus British Gas is (by assumtion) lss ffctiv as a comtitor. (Addin an xtra aramtr, for lctricity-only customrs cost of switchin to British Gas, is an obvious xtnsion hr.) Givn that th calibration av British Gas a switchin cost advanta in attractin dual-ful customrs in most markts, British Gas has th hihst ric, whil th lctricity incumbnts rics ar much closr to British Gas than to th non-incumbnts. Th xction is th lowswitchin markt, basd on th north of Scotland, in which British Gas is at a disadvanta to Hydro-Elctric, and chars a lowr ric. For th lctricity incumbnt, rics ar hihst in th markts with hih switchin costs, but ths ar also th markts in which thy can sustain th hihst markt shars. Th ntrants rics hardly vary across markts, althouh thy ris slihtly as switchin costs fall, in lin with th simlr modl. British Gas has a hihr markt shar, and dual ful ric, in th markt with mdium switchin costs than in th markt with low costs. Th incumbnts markt shars ar much lowr in th simulations than thy ar at rsnt. Thir rics ar currntly blow th lvls sn hr, imlyin that it would b rofit-maximisin to rais rics and los markt shar. 5. Emirical analysis In this art of th ar, w comar th rdictions from th modls with markt data. Pric data for Aril 003, May 004, and May 005 wr takn from th nrywatch wb sit. W hav rics for th six lar sulirs (British Gas, EdF Enry, nowr, PowrGn, Scottish and Southrn Enry, and Scottish Powr) for thr aymnt mthods (standard crdit, dirct dbit, and raymnt) and two roducts (lctricity only and dual ful) in ach of th 4 rions. W also hav th ric of as from British Gas, which is ndd to calculat th cost of continuin to buy from both incumbnt comanis, rathr than takin a dual ful suly. As bfor, w tak th unwihtd avra of ach comany s ric for low, mdium and hih consumtion lvls. All rics ar normalisd for diffrncs in cost btwn rions by dividin thm by th avra of th rics chard by th four non-incumbnt lctricity comanis in that rion. Thos avra rics will xcd marinal costs, from th discussion in th rvious sctions, but may not b too far from comtitiv 6
lvls, ivn th rlativly small markt shars of non-incumbnt lctricity comanis. Tabl 8: Avra rics by yar Dual Ful Elctricity only Incumbnt in: Both BG Elc. BG Elc. 003.44.086.048.008.074 Standard 004.3.076.04.03.083 Crdit 005.7.067.07.006.070 003.3.055.030.0.08 Dirct Dbit 004.0.049.039.0.097 005.09.036.07.00.083 003.07.008.00 0.963.008 Praymnt 004.05.03.09 0.99.040 005.069.059.09.038.063 Th attrns for crdit and dirct dbit customrs ar similar, and rathr diffrnt from raymnt customrs. It is around tn r cnt mor xnsiv to buy from both incumbnts than to tak a dual ful suly from a non-incumbnt lctricity comany, althouh th diffrnc has bn fallin ovr tim. Both British Gas and th incumbnt lctricity comany offr savins to consumrs who tak a dual ful suly from thm, but thy ar lss comtitiv than th non-incumbnts, and British Gas is lss comtitiv, on avra, than th incumbnt lctricity comanis. Onc aain, ths ric diffrncs sm to hav bn fallin. For lctricity-only customrs, incumbnts ar around svn to nin r cnt mor xnsiv than nonincumbnt lctricity comanis, with no clar trnd ovr tim. British Gas is slihtly mor xnsiv than th othr ntrants to this smnt, but within on standard dviation. For raymnt customrs, ntrants st thir rics much closr to th incumbnts rics, on avra, at th start of th riod. Th lss intns comtition to srv raymnt customrs was bcaus thy wr oftn blivd to b lss rofitabl than thos ayin by othr mthods, dsit thir hihr tariffs. Durin th riod, howvr, four of th fiv lctricity comanis incrasd thir 7
raymnt tariffs in aras whr thy wr ntrants by lss than thy incrasd thir othr tariffs (raymnt as incumbnt, and othr aymnt mthods). This mant that th incumbnts rics bcam rlativly hihr, as shown at th bottom of tabl 8. Th nxt tabls rort th rsults of rrssions on th fiv lctricity comanis attrns of (rlativ) rics. In ach rion, th man of th four non-incumbnt lctricity comanis rics is qual to on. Thr ar dummy variabls for bin th local incumbnt in ach yar, and dummy variabls for 004 and 005. Th yar dummis aly to all obsrvations in that yar, so that th incumbncy variabls masur th additional imact of bin th incumbnt in an ara. Tabl 9 ivs th rsults for crdit customrs. Exct for PowrGn, incumbncy is sinificant at th on r cnt lvl in vry yar, and involvs raisin dual ful rics by btwn two and four r cnt in 005. Tabl 9: Rrssions on (rlativ) dual ful rics, crdit customrs EdF nowr PowrGn Scottish and Southrn Scottish Powr Inc 003 0.0458 *** (0.00897) 0.0370 *** (0.00670) 0.084 *** (0.00985) 0.046 *** (0.0065) 0.094 *** (0.00894) Inc 004 0.0507 *** (0.00897) 0.040 *** (0.00670) 0.08 (0.00985) 0.0338 *** (0.0065) 0.0646 *** (0.00894) Inc 005 0.078 *** (0.00897) 0.04 *** (0.00670) 0.036 ** (0.00985) 0.068 *** (0.0065) 0.090 *** (0.00894) 004 0.0508 *** (0.00587) -0.00096 (0.00438) 0.0096 (0.00645) -0.090 *** (0.00403) -0.0370 *** (0.00477) 005 0.048 *** (0.00587) -0.057 *** (0.00438) -0.057 ** (0.00645) -0.05 *** (0.00403) 0.0 ** (0.00477) Constant 0.973 *** (0.0045).07 *** (0.0030) 0.9807 *** (0.00456).036 *** (0.00338).036 *** (0.00338) R 0.884 0.84 0.7439 0.7888 0.830 Standard Errors in arnthss * Sinificant at 0%; ** Sinificant at 5%; *** Sinificant at % 8
Tabl 0 ivs th rsults for dirct dbit customrs. Aain, incumbncy rsults in a statistically sinificant ric incras in twlv of our fiftn comany-yar combinations. Th imact, of btwn two and four rcnt in most cass, is similar to that for crdit customrs. Tabl 0: Rrssions on (rlativ) dual ful rics, dirct dbit customrs EdF nowr PowrGn Scottish and Southrn Scottish Powr Inc 003 0.0406 *** (0.00834) 0.033 *** (0.00749) 0.007 (0.00539) 0.0433 *** (0.00648) 0.088 *** (0.079) Inc 004 0.0503 *** (0.00834) 0.0437 *** (0.00749) 0.0038 (0.00539) 0.039 *** (0.00648) 0.0503 *** (0.049) Inc 005 0.068 *** (0.00834) 0.0443 *** (0.00749) 0.063 *** (0.00539) 0.030 *** (0.00648) 0.03 (0.049) 004 0.058 *** (0.00546) -0.00096 (0.00490) 0.079 *** (0.00353) -0.067 *** (0.0044) -0.0 *** (0.00668) 005 0.00 *** (0.00546) -0.057 *** (0.00490) -0.0075 ** (0.00353) -0.047 *** (0.0044) 0.045 *** (0.00668) Constant 0.9953 *** (0.00386) 0.9965 *** (0.00347) 0.9974 *** (0.0050).036 *** (0.00300).036 *** (0.00338) R 0.7005 0.7598 0.6787 0.7856 0.6499 Standard Errors in arnthss * Sinificant at 0%; ** Sinificant at 5%; *** Sinificant at % Tabl ivs th rsults for raymnt customrs, which ar far mor mixd. Aftr th first yar, all th cofficints on incumbnts rics ar ositiv, but only thr cofficints ar sinificant at th 5% lvl (or bttr). In th most rcnt data, howvr, th bst stimat of th imact of incumbncy on rics is similar to that for th othr aymnt tys. 9
Tabl : Rrssions on (rlativ) dual ful rics, raymnt customrs EdF nowr PowrGn Scottish and Southrn Scottish Powr Inc 003 0.005 (0.00843) -0.0043 (0.00663) 0.069 * (0.009) 0.0 (0.0090) -0.0056 (0.0664) Inc 004 0.008 ** (0.00843) 0.033 * (0.00663) 0.075 * (0.009) 0.004 (0.0090) 0.030 * (0.0664) Inc 005 0.0497 *** (0.00843) 0.039 *** (0.00663) 0.093 * (0.009) 0.038 (0.0090) 0.076 (0.0664) 004-0.0060 (0.0055) 0.08 *** (0.00434) 0.0097 (0.00596) -0.0309 *** (0.0059) -0.0009 (0.00890) 005-0.009 (0.0055) 0.0084 * (0.00434) 0.084 ** (0.00596) -0.0557 *** (0.0059) 0.057 *** (0.00890) Constant 0.988 *** (0.00390).06 *** (0.00307) 0.9783 *** (0.004).0054 *** (0.0048) 0.9697 *** (0.0069) R 0.5453 0.787 0.546 0.7655 0.3844 Standard Errors in arnthss * Sinificant at 0%; ** Sinificant at 5%; *** Sinificant at % It is not ossibl to run th sam rrssions for British Gas, sinc it is an incumbnt in vry rional markt. In tabl, howvr, w rrss British Gas dual ful rics aainst th roortion of lctricity customrs who also tak as in ach rion, yarly dummis, and th (rlativ) ric aid by customrs who hav not switchd from th two incumbnts in ach rion. British Gas chars th sam ric for as in ach rion (as is nrally th cas with othr as sulirs), and so th variation in customrs aymnts is du to th diffrnt rics chard by incumbnt lctricity comanis. This mans that w can trat th variabl as xonous to British Gas conduct in th dual ful markt. W find that British Gas rics for crdit and dirct dbit customrs hav bn in (rlativ) dclin, whil thir raymnt rics hav bn risin rlativ to thos chard by ntrant lctricity comanis. British Gas chars rlativly mor for crdit customrs and dirct dbit customrs in rions whr thr ar a hih roortion of as customrs, which could rflct a stronr brand in ths rions. 30
Th comany chars mor, rlativ to ntrant sulirs, in rions whr stayin with both incumbnts is articularly xnsiv. This ffct is stronst in th raymnt markt, but also statistically sinificant in th dirct dbit markt, and not contradictd in th crdit markt). This is consistnt with th raction functions drivd in quation (4) and with th numrical stimats in th scond two columns of tabl 7. Tabl : Rrssions on (rlativ) dual ful rics, British Gas Crdit Dirct Dbit Praymnt Rlativ char of 0.935 0.97 ** 0.5345 *** incumbnt sulirs (0.337) (0.306) (0.0604) Proortion of as customrs 0.0404 ** (0.0546) 0.0346 * (0.0765) 0.0053 (0.0459) 004-0.0078 * (0.00406) -0.005 (0.00434) 0.003 ** (0.00384) 005-0.043 *** (0.00507) -0.033 ** (0.0054) 0.089 *** (0.00437) Constant 0.8305 *** (0.4040) 0.700 *** (0.4378) 0.4540 *** (0.0677) R 0.4976 0.4596 0.8870 Standard Errors in arnthss * Sinificant at 0%; ** Sinificant at 5%; *** Sinificant at % Tabl 3 shows th corrlation matrics for th incumbnts rional markt shars in lctricity and as for ach aymnt mthod, at thr oints in th dvlomnt of comtition. By Stmbr 004, th roortions still buyin from th incumbnt ar vry similar in as and in lctricity for dirct dbit customrs (5%) and crdit customrs (63%), but mor raymnt customrs hav stayd with th incumbnt in as (73%) than in lctricity (63%). Th standard dviation of incumbnt shars is ratr in lctricity than in as, howvr. 9 9 Whil th standard dviation of incumbnt markt shars in lctricity is incrasd by th low switchin rats in th north of Scotland (which is not saratly idntifid in th as data), th diffrnc rsists whn lookin at Enland and Wals alon. 3
Tabl 3: Incumbnt markt shars, by aymnt ty Stmbr 00 Elctricity Gas Crdit DD Pray Crdit DD Pray Man incumbnt shar 68.8 6.7 65.5 67. 56.9 78. Standard dviation 7.0 7.4 8.4 3.7 4.4 7.3 Crdit Elctricity DD 0.586 Praymnt 0.786 0.670 Crdit 0.34-0.06 0.3 Gas DD -0.8 0.080-0.53 0.447 Praymnt 0.39 0.07 0. 0.68 0.356 Stmbr 003 Elctricity Gas Crdit DD Pray Crdit DD Pray Man incumbnt shar 64.4 56.4 64. 64.6 55.4 73.6 Standard dviation 8.7 7.3. 3. 4.6 7.6 Crdit Elctricity DD 0.694 Praymnt 0.686 0.807 Crdit 0.053-0.60-0.03 Gas DD 0.033-0.030-0.006 0.667 Praymnt 0.069-0.073-0.34 0.393 0.48 Stmbr 004 Elctricity Gas Crdit DD Pray Crdit DD Pray Man incumbnt shar 63. 5.9 6. 63.0 5.4 7.8 Standard dviation 9.5 9..7 4. 3.9 8.9 Crdit Elctricity DD 0.68 Praymnt 0.667 0.864 Crdit -0.7-0.396-0.379 Gas DD -0.57-0.345-0.33 0.53 Praymnt 0.086-0.99-0.4 0.480 0.490 Within a ful, incumbnt markt shars ar ositivly corrlatd across rions whr th incumbnt has bn ood at rtainin dirct dbit customrs, it tnds to hav bn ood at rtainin crdit and raymnt customrs. Th rlationshi is stronr for lctricity than as. At first, thr was a wak ositiv corrlation btwn markt shars in as and in lctricity for ach aymnt ty, imlyin that in an ara whr customrs wr switchin for on ful, thy wr mor likly to b switchin for th othr on. By Stmbr 004, howvr, this had switchd to a wak nativ corrlation. This is consistnt with dual ful comtition bcomin incrasinly imortant, so that whr an incumbnt lctricity comany has bn abl 3
to dfnd its markt shar, it has don so, in art, by sllin as to an incrasin numbr of customrs, rducin British Gas markt shar in th rion. 6. Conclusions Th first art of this ar outlind som stylisd facts about th British nry rtail markts. Rouhly sakin, about two-fifths of customrs ar still buyin thir as from British Gas and thir lctricity from th rional lctricity incumbnt. Slihtly ovr on fifth ar buyin both fuls from British Gas. About on fifth ar buyin both fuls from th rional lctricity incumbnt. Most of th rmainin customrs ar buyin both fuls from on of th othr lar lctricity comanis, althouh thr ar still som small indndnt sulirs, and som customrs hav switchd on ful, or switchd both but to diffrnt sulirs. Tabl 8 showd that buyin from any incumbnt was slihtly mor xnsiv than buyin as and lctricity from a nonincumbnt lctricity comany, and buyin from both incumbnts could cost around 0% mor. Sinc th avra nry bill is around 700 a yar, it is difficult to xlain th incumbnts hih markt shars without assumin that customrs fac hih (ral or imaind) switchin costs. Th followin sctions of th ar usd an analytical modl to show how th rics of as and lctricity miht dnd uon th lvl of switchin costs and th rlativ sizs of th two markts, and a calibratd numrical modl to calculat rofitmaximisin rics for ntrants and incumbnts. Th simulations imlid that th incumbnts could find it rofitabl to rais rics wll abov currnt lvls, vn thouh thir markt shars would fall furthr. Rturnin to th data, w found that th fiv main lctricity comanis chard mor for a dual ful suly in aras whr thy wr th lctricity incumbnt, and that British Gas chard mor in aras whr th lctricity comany s ric for lctricity was hih. W did not rsnt any conomtric analysis of markt shars, 0 but not that th incumbnts rional markt shars in as and in lctricity ar bcomin nativly corrlatd. This can b intrrtd as a 0 Th main roblm is idntifyin th rlationshi btwn markt shars and attrns of rics, vn if ast rics ar usd as instrumnts. In th short trm, a comany s markt shar will fall if its rics ar rlativly hih, but in th quilibria of th modls in this ar, comanis that ar rotctd by hihr switchin costs will char hihr rics and still hav hihr markt shars. 33
sin of th imortanc of dual ful comtition th bst way for an lctricity comany to dfnd its own markt is to attack that of British Gas. Dual-ful comtition is still volvin in th British rtail nry markts two yars ao, incumbnt markt shars in as and lctricity wr ositivly corrlatd but it is clarly an imortant fatur of thos markts. Analysin th markts for as and lctricity on thir own is likly to miss som imortant intractions. At th olicy lvl, sinc both Ofm and nrywatch ar rsonsibl for both as and lctricity, w can ho that ths intractions will b takn into account. Th ky olicy qustion is whthr comtition is rotctin consumrs bttr than continud rulation (or a rturn to rulation) would hav don. Any ar that dos not rsnt a rasond cas for what would hav hand undr rulation is only a artial answr to that qustion. Howvr, it is clar that about half of th consumr switchin that romtd th nd of rulation cam about bcaus consumrs wr switchin from th incumbnt in on ful in ordr to stay with th incumbnt in th othr. Whil this did iv thm savins, comard to stayin with both incumbnts, ths customrs wr still ayin mor (on avra) than if thy had switchd to a non-incumbnt sulir. It is clar that thr is comtition in ths markts, but it is far from clar that it is workin as ffctivly to rotct consumrs as w miht ho. 34
Bibliorahy Bsanko, D., M.K.Prry and R.H.Sady (990) Th Loit Modl of Monoolistic Comtition: Brand Divrsity Journal of Industrial Economics, vol. XXXVIII, no. 4, 397-45 Giulitti, M., J.Otro and M.Watrson (004) Suly comtition and ric bhaviour in th UK lctricity suly industry, mimo, Aston Univrsity Giulitti, M., C.M. Waddams Pric and M.Watrson (003) Consumr Choic and Industrial Policy: A Study of UK Enry Markts Univrsity of California Enry Institut, CSEM WP Grn, R.J. (00) Can Comtition Rlac Rulation for Small Utility Customrs? CEPR Discussion Par 406, Cntr for Economic Policy Rsarch Ofm (004) Domstic Comtitiv Markt Rviw 004, London: Offic of Gas and Elctricity Markts Otro, J and C.M. Waddams Pric (00) Pric Discrimination in a Rulatd Markt with Entry: th rsidntial UK lctricity markt, Bulltin of Economic Rsarch, vol. 53, no. 3, July, 6-75 Salis, E and Waddams Pric, C (forthcomin) Chars, Costs and Markt Powr in th Drulatd UK Elctricity Rtail Markt, Th Enry Journal, Jun 004 Sturluson, J.T. (003) Consumr Sarch and Switchin Costs in Elctricity Rtailin, Chatr in Toics in th Industrial Oranization of Elctricity Markts, PhD Thsis, Stockholm School of Economics, Swdn Waddams Pric, C.M. (004) Soilt for Choic? Th Costs and Bnfits of Onin UK Rsidntial Enry Markts, CSEM WP 3, Univrsity of California Enry Institut 35
Fiur : Th Elctricity Rions of Grat Britain Hydro- Elctric Scottish Powr Norwb Northrn Yorkshir E Midlands Manwb Midlands Swalc Eastrn London Sboard SWEB Southrn 36