Calendars

A Critique of Wolak s Evaluation of the NZ Electricity Market:

Description
A Crtque of Wolak s Evaluaton of the NZ Electrcty Market: The Incentve to Exercse Market Power wth Vertcal Integraton and Transmsson Loss* Seamus Hogan 1 and Peter Jackson 2 1 Correspondng Author Department
Categories
Published
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
A Crtque of Wolak s Evaluaton of the NZ Electrcty Market: The Incentve to Exercse Market Power wth Vertcal Integraton and Transmsson Loss* Seamus Hogan 1 and Peter Jackson 2 1 Correspondng Author Department of Economcs Unversty of Canterbury (03) Department of Management Unversty of Canterbury Abstract: Ths paper provdes a crtque of the 2009 report by Frank Wolak nto the New Zealand electrcty market that concluded that there had been a cumulatve total of $4.3b (NZD) of overchargng over a perod of seven years. We fnd that hs concluson that the market structure has gven generators an ncentve to exercse substantal market power s hghly senstve to some (unrealstc) assumptons made about the elastcty of demand and the extent of transmsson loss, and that the observed correlatons between estmates of that ncentve and market prce are also consstent wth perfect competton. Key Words: Wolak Report; electrcty markets; market power. JEL Codes: L41, L13 * Ths paper s part of a symposum of papers by the authors together wth Graeme Guthre and Lew Evans of Vctora Unversty of Wellngton that together provde a crtque of the 2009 Wolak report nto the wholesale electrcty market n New Zealand. Ths paper, prepared for the NZAE 2010 conference focuses on the measurement of market power n the Wolak report A Crtque of Wolak s Evaluaton of the NZ Electrcty Market: The Incentve to Exercse Market Power wth Vertcal Integraton and Transmsson Loss 1. Introducton. In 2009, the Commerce Commsson of New Zealand released a commssoned report by Professor Frank Wolak: An Assessment of the Performance of the New Zealand Wholesale Electrcty Market (Wolak, 2009). Ths report was an analyss of the extent to whch electrcty generators n New Zealand have exercsed unlateral market power to acheve wholesale electrcty prces n excess of the levels that would result from perfect competton. The report concluded that there was evdence that generators have had an ncentve to exercse unlateral market power, and evdence that they have exercsed that market power. It also estmated a counterfactual benchmark of what prces would have been n the absence of market power and used ths counterfactual to conclude that there had been a cumulatve total of $4.3b (NZD) of overchargng over a perod of seven years. The Wolak report contans three broad strands of emprcal evdence that generators have exercsed unlateral market power n New Zealand. The frst strand calculates emprcally from the observed offers, the extent to whch frms have had the ablty and the ncentve to exercse unlateral market power usng standard olgopoly theory. The second strand uses regresson analyss to estmate the extent to whch those measures of market power can explan the observed prce movements n the New Zealand wholesale market. Fnally, the report uses drect estmates of frms short-run margnal cost curves to calculate the extent to whch market prces have exceeded margnal cost. The estmate of $4.3b of overchargng s derved from ths thrd strand of emprcal analyss, and not surprsngly, ths s the part of the report that has receved most attenton and crtcsm. 1 Fgure 5.1 n the Wolak report, reproduced below, shows the calculaton of the $4.3b. The bars n red are the total payments made to generators n excess of the estmated 1 See, for example, Electrcty Techncal Advsory Group (2009), NZIER (2009), and Unversty of Auckland (2009), 1 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market short-run margnal cost. The bulk of the $4.3b was obtaned n durng the three dry-year perods of hgh wholesale prces, n 2001, 2003, and The consensus n the crtques of the report seems to be that the method used has lkely led to an overstatement of the relevant margnal cost of producton and hence an overstatement of the amount of market-power rents. These crtques, however, do not provde an alternatve estmate of the rents, and so the other two strands of emprcal evdence n the report acqure greater sgnfcance and are worth of analyss. Ths paper and the parallel paper, Evans and Guthre (2010), seek to do ths. Evans and Guthre consder the regresson evdence n relatng market power to prces. In ths paper we dscuss the estmates of market power themselves. In the followng secton, we brefly outlne the bass of the varous crtcsms of the drect-estmaton of market rents n the Wolak report. In Secton 3, we descrbe the report s estmates of the ablty and ncentve for frms to exercse market power. In Secton 4 we suggest some reasons why these estmates are lkely to be overstated. 2 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market 2. The Drect Estmate of Market Rents. In order to estmate the extent to whch frms have been exercsng market power n dry years, t s necessary to establsh a counterfactual for how much prces would have rsen n those years due to the normal nteracton of supply and demand n a compettve market. The Wolak Report s method for estmatng ths perfectly compettve benchmark has been subject to four man crtcsms: that the short-run margnal cost s the wrong benchmark for dentfyng the perfect-competton counterfactual; that measures of excess capacty gnored the mportance of ancllary servces; that the report ncorrectly measured the opportunty cost of fuel n thermal statons and water n hydro statons; and that the report s assumpton of zero elastcty of demand s napproprate. We wll brefly address each one of these n turn. 2.1 Short-Run Margnal Cost. When the New Zealand Electrcty Market was formed n 1996, the phlosophy of the market desgn was to balance the sometmes conflctng goals of short run and long run effcency, whle provdng securty of supply resemblng a 1 n X dry year standard. After much delberaton t was decded to mplement an energy-only, or sngle-payment, market. In such a market nvestors must not only varable costs, but fxed costs also, from a sngle payment for energy. An energy-only payment equal to short-run margnal cost can never provde a market return on the margnal captal cost of plant. Ths s partcularly so n an ndustry lke electrcty wth seasonal fluctuatons n both supply and demand n whch the margnal unt of capacty s lkely only to be needed n hgh-demand perods n dry years. Hgh prces durng these perods wll be a necessty to pay for that margnal capacty and provde an ncentve for nvestment n new capacty. Ths logc mples that the perfectly compettve benchmark prces wll be at the pont at whch demand ntersects a vertcal supply curve at the pont of full capacty. The argument for usng short-run margnal cost n the Wolak report was that, at all tmes n the perod of the data beng consdered, total output from all power statons was consderably less than the mpled capacty derved by consderng the sum of the maxmum output of each ndvdual staton over the perod n queston. Ths concluson of permanent excess capacty gnores two addtonal factors. 3 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market 2.2 Ancllary Servces. The New Zealand wholesale market combnes a market for power, and a market for ancllary reserves, whch s a planned excess capacty kept avalable for generaton at short notce n case of a fault n the transmsson network. Ancllary reserves are not so mportant n countres wth two-dmensonal networks and a larger number of power statons, but they are are partcularly mportant n New Zealand because of the one-dmensonal hgh-voltage drect current (HVDC) lnk connectng the North and South Islands. In the event of a falure of the HVDC, t s mportant that there be enough ancllary reserves n the sland to whch power f flowng n order to prevent cascadng falure. Snce power sometmes flows north and, less frequently, south, the amount of capacty kept n reserve for ancllary servces wll fluctuate over tme between North Island and South Island statons. 2.3 Opportunty Cost of Fuel and Water. The Wolak Report s perfectly-compettve benchmark assumed that the opportunty cost of fuel used n thermal power statons s smply the replacement cost of that fuel, and that the opportunty cost of water n a hydro staton s the hghest value n a season of the thermalstaton, short-run margnal cost calculated usng the fuel replacement cost. Both assumptons can be questoned. New Zealand s not stuated close to ready sources of fuel. Because of take or pay contracts, thermal statons are not able to access sgnfcantly more or less fuel accordng to the partcular hydrologcal outcome of the year, and so need to husband ther fuel reservors wth an assocated opportunty cost of use that exceeds smple replacement cost. Smlarly, hydro statons need to take account of the opton value of a lmted reservor of water n a dry year, takng nto account uncertanty about hydrologcal nflows and demand patterns. The ex post realsed market prces throughout the wnter s not a relable measure of the true opportunty cost of water at any pont n tme gven the nformaton avalable at that tme. For a fuller dscusson of ths pont, see Evans and Guthre (2009). 2.4 Elastcty of Demand. Even f t were the case that short-run margnal cost could be estmated from the replacement cost of thermal staton fuel, and that there was permanent excess capacty n the New Zealand market, short-run margnal cost would stll not be the approprate benchmark f 4 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market demand were to exceed capacty n a dry year at that prce. The Wolak report assumed away ths possblty by assumng that demand s perfectly nelastc. Ths s a reasonable assumpton for the descrbng the performance of the market n the very short term. The market operates by sellers submttng supply offers every half-hour, buyers smply drawng as much power as they wsh, and the market prce beng realsed ex post to equate demand to aggregate supply. By defnton, ths demand s nsenstve to prce. Over a perod of several weeks or months n a dry year, however, demand can and does respond to ncreases n the wholesale prce. Ths arses through three mechansms: Frst, there s a small percentage of the fnal demand that buys drectly from the wholesale market and so has an ncentve to reduce demand when prces rse; second, retal companes offer fnancal ncentves to customers to reduce demand; and thrd, the threat of supply shortages assocated wth the hgh prces lead to government-sponsored conservaton campagns. The non-zero medum term elastcty of demand wth respect to wholesale prces s also an mportant factor n ndrect measures of market power consdered n the sectons below. 2.5 Summary. The four crtques above overwhelmngly pont n the drecton that the estmate of market power n the Wolak report s too hgh, but they don t provde an alternatve measure. There stll remans the other strands of emprcal evdence n the report that are also consstent wth the hgh prces n dry years beng the result of market power. Ths leads the authors of the NZIER dscusson document prepared for the Major Electrcty User s Group (NZIER, 2009), to comment: Wolak s analyss has drawn a lot of crtcsm, not all of t vald. Most crtcsms have been drected at the way n whch he constructs compettve benchmark prces and thereby estmates market power rents. Crtcsm of ths addtonal step n hs analyss does not detract from the earler steps n hs methodology of examnng the slopes of resdual demand curves to determne ablty and ncentve to nfluence market clearng prces. Ths basc methodology s well-establshed and wdely recognsed and appled worldwde to assess competton n short-term wholesale markets. In the remander of ths paper, we turn our attenton to these measures of the ablty and ncentve to exercse unlateral market power. 5 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market 3. Measures of the Ablty and Incentve to Exercse Unlateral Market Power. In ths secton, we present the theory for how one can nfer from market data on frms offer curves, each frm s ablty and ncentve to nflate wholesale prces by overstatng the prces at whch they would be wllng to supply gven quanttes. Consder frst, a smple Cournot model wth no transmsson loss and no vertcal ntegraton. Let Sj ( p) be the offer curve (supply) submtted by generator j, and let the demand curve be D(p). The resdual demand curve, R ( p ), facng frm gves the quantty that frm would need to supply nto the market to produce a market prce of p, takng the demand curve and the offer curves of the remanng generators as gven. That s R( p) D( p) S ( p). = j j Let c ( q ) be frm s cost of supplyng q unts of output. Frm s proft maxmsaton problem then s Max pr ( p) c ( R ( p)), p for whch the frst-order condton s ( p c ) R ( p) + R ( p) = 0. Ths equaton can be re-wrtten as p= c η, (1) where R ( p) η =. 100 R ( p) (2) The term, η, s what the Wolak report terms an nverse sem-elastcty. It measures the amount by whch a generator could ncrease prce by wthholdng 1% of ts supply from the market and so s a measure of the market-power of a sngle frm. In the case where frms have no fxed-prce forward contracts, as shown by Equaton (1), t s also a measure of the ncentve of a frm to wthhold supply n order to push prce hgher than margnal cost. If frms do have a commtment to sell a certan amount of power at a fxed prce determned pror to submttng supply offers nto the market, then, whle η stll reflects a frm s ablty to exercse market power, the ncentve to do so s mtgated by the extent to whch t needs to be a buyer n the wholesale market n order to meet ts fxed-prce 6 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market oblgatons. To show ths, let generator, have fxed-prce forward contracts to sell a quantty, Q of power at an average prce of p. 2 In ths case, the frm s optmsaton problem becomes, Max pr( p) c( R( p)) + ( p p) Q, (3) p and the assocated frst-order condtons and sem-elastcty mark-up rule become ( p c ) R ( p) + ( R( p) Q ) = 0, and p= c η δ, (4) where R( p) Q δ =. R( p) The mtgaton term, δ, gves the extent to whch fxed-prce forward oblgatons reduce the ncentve to exercse market power. In the specal case where the generator wll supply exactly the same quantty of power as ts forward oblgaton, the mtgaton s 100% and the optmal quantty wll be such that prce equals margnal cost. If ts forward oblgatons are greater than ts quantty suppled, the ncentve s to expand output to force prce below margnal cost. 4. Potental Modfcatons to the Measures. The theory lad out n the prevous secton replcates, wth slght notaton changes, that presented n the Wolak report. Of course, the theory descrbes the ncentves facng frms n a very smple model that abstracts away from many features of the actual New Zealand electrcty market. Gven the complcated nature of that market, smplfcaton s necessary to get a tractable model, but there are two aspects of realty n partcular that, f not consdered, wll lead to a basng up of the estmates of η and δ and consequently of the estmates of the ncentve to exercse market power. The frst concerns the mpact of the elastcty of demand for wholesale-market electrcty on the calculaton of η, and the second concerns the mpact of transmsson loss on the formula δ. 2 Snce p has no effect on the frm s offer strategy, t doesn t matter whether all ts fxed-prce oblgatons are at the same prce, or contract-specfc prces. 7 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market 4.1 The mpact of demand elastcty on market power. As noted n Secton 2, the Wolak report assumes that demand s perfectly nelastc. Ths s a perfectly sensble assumpton for the short-run elastcty, and so mght accurately represent the market power that could result from, say, a falure n the transmsson network. When prces are consstently hgh over a perod of weeks or months, however, the market does contan mechansms by whch load can respond to prce ncentves. That s, generators lookng to explot market power over longer perods, may well face perfectly nelastc demand curves n the very short-run of a sngle 30-mnute prcng perod, but they must also take nto account the mpact that persstent hgh prces wll have on that nelastc short-run demand curve over tme. Ths s partcularly pertnent n assessng measures of market power. Although a small amount of demand elastcty wll not have much mpact on the sem-elastcty most of the tme, t can have a very large mpact when the sem-elastcty s very hgh. To see ths, magne that we have an estmate of the market-power elastcty, ˆ η, made under the assumpton that demand elastcty s zero when the absolute elastcty s n fact postve. Note that R ( p) = S ( p) D ( p). j j We therefore have R ( p) η = 100( S ( p) D ( p)) j j (5) whle R ( p) ˆ η =. 100 S ( p) j j (6) Puttng Equaton (6) nto (5) gves 1 η =, 1 ε100θ + ˆ η p where ε = pd ( p)/ D( p) s the absolute elastcty of demand and θ = D( p) / R( p) s the nverse of frm s market share. 8 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market Now consder Fgure 4.5 from the Wolak report, reproduced below. The hghest average (across all generators) value for the estmated nverse sem-elastcty, ˆ η, s 10, whch occurred at the peak of the dry-year prce-spke of That corresponded to a market wholesale prce of 200. If we generously assume an nverse market share of θ = 4, then a demand elastcty as low as ε = 0.05would be enough to lower the true nverse sem-elastcty from 10 to The Impact of Transmsson Loss on the Incentve to Inflate. The model presented n Secton 2 descrbes a network wth no transmsson loss. In realty, frms supply power to njecton ponts on the natonal grd, and demanders n the wholesale market purchase power from ext ponts. The offers of generators are offers to supply a quantty of electrcty at the njecton pont at the specfed prces, whereas the prces pad by buyers refer to the quantty taken at the ext ponts. Wth transmsson loss along the grd, the total quantty suppled at njecton ponts wll exceed that demanded at ext ponts, wth a correspondng dfferental n the prces. To see how ths can affect the emprcal estmates of the ncentve to nflate, consder a slght modfcaton to the model of Secton 2, by consderng a two-node system n whch all 9 S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market generators supply at one node, and demanders purchase at the other. Further, assume that a fracton, λ, of the energy suppled s lost n transmsson. s Let S( p ) be the total energy suppled at the njecton node as a functon of the prce receved by sellers, and let d D( p ) be the total energy demanded at the ext node, as a functon of the prce pad by buyers. Equlbrum n ths market requres that the energy suppled s enough to meet demand takng nto account the transmsson loss, d s D( p ) = (1 λ) S( p ), (7) and that the prces are such that total payments equal total recepts, d d s s p D( p ) = p S( p ). (8) Equaton (7) nto (8)gves p d s p =, 1 λ and the modfcaton to Equaton (3) to account for transmsson loss s s s s s Max p R( p ) c( R( p )) + ( p p /(1 λ)) Q, s p for whch the frst-order condton s R( p) R( p) Q /(1 λ) p= c + R ( p) R( p) δ λ = c η. (1 λ) (9) To get a sense of how mportant transmsson loss mght be, consder Fgure 4.8 below, agan reproduced from the Wolak report. Agan, we focus on the dry-year prce spke n The mean net nverse sem-elastcty at the peak, takng nto account both vertcal ntegraton and generators fxed-prce forward contracts, s approxmately 1.5. Recall from the prevous graph that the mean nverse sem-elastcty was 10. That s, we have mean values of η = 10 and ηδ = 1.5, and so can nfer a mean value of δ of approxmately S. Hogan, P. Jackson Crtque of Wolak s Evaluaton of the NZ Electrcty Market Transmsson loss s estmated to average approxmately 6.5% to 7.5%. Although some of ths loss wll come n the lnes connectons between generators and the grd, and between the grd and fnal users, t s also the case that transmsson loss s greatest when load s hgh, partcularly when the flow on the HVDC s hgh, whch are also the tmes whch put the most demand pressure on prce. In other words, the exte
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks