Embed Size (px)
Citation preview
ARTICLE IN PRESS
Energy Policy 36 (2008) 2620– 2627
Contents lists available at ScienceDirect
Energy Policy
0301-42
doi:10.1
� Corr
E-m
margare1 Te
journal homepage: www.elsevier.com/locate/enpol
A theoretical analysis of procurement auctions for tertiarycontrol in Germany
Gernot Muller a,1, Margarethe Rammerstorfer a,b,�
a WIK Wissenschaftliches Institut fur Infrastruktur und Kommunikationsdienste GmbH, Rhondorfer Strasse 68, 53604 Bad Honnef, Germanyb Vienna University of Economics and Business Administration, Corporate Finance Institute, Heiligenstadter Strasse 46-48, 1190 Vienna, Austria
a r t i c l e i n f o
Article history:
Received 25 February 2008
Accepted 14 March 2008Available online 6 May 2008
Keywords:
Auction theory
Procurement auctions
Tertiary control
15/$ - see front matter & 2008 Elsevier Ltd. A
016/j.enpol.2008.03.017
esponding author. Tel.: +43131336 5995; fax
ail addresses: [email protected] (G. Muller),
[email protected] (M. Ramm
l.: +49 2224 9225 84; fax: +49 2224 9225 63.
a b s t r a c t
As far as energy policy is concerned, the design of the regulatory framework for energy transmission and
distribution is a key issue. Consequently, also the embodiment of balancing power markets drives
mainly the effectiveness of political implications for the energy sector. Initially, tertiary control in
Germany was solely offered by transmission system operators of the respective power control areas and
their associated power plant. The recast of the Energy Industry Act of 2005 led in last consequence to a
common procurement auction for the supply of tertiary control, which starts on December 1, 2006.
Admittedly, the reform has fallen short of expectations so far, first concerning the intensification of
market entry of tertiary control providers as well as the desired decline of the price level. Hence, this
article examines the effects of the changeover on observable demand charges. In order to identify
attributes of the common procurement auction for tertiary control hampering market entry of
providers, giving stimuli to collusion and strategic behavior, reducing intensity of competition
and encouraging an upswing of prices, we analyze the design under an auction theoretical approach
and deduce empirically whether structural components of the auction design have to be touched
up again.
& 2008 Elsevier Ltd. All rights reserved.
1. Introduction
Energy policy is part of the sectoral policy—which can bedivided into economical–structural and environmental policy—inorder to control the energy sector. Therefore, energy policyencompasses all stages of energy supply, starting with theproduction of energy followed by storage, transportation, trans-mission and distribution and ending up at the adhesive wastedisposal. The scope of energy policy changed extremely over thelast few decades. However, one common accepted aim is thesustainable usage of energy and production of energy under lowrisk as well as the cost-effective allocation and distribution ofenergy supplied. Energy policy in a narrower sense can be seen asthe regimentation of a technical, social and legal framework inorder to reach the previously mentioned aims. For the Germanelectricity sector the design of an efficient and appropriateregulatory framework is a key component for adequate energypolicy. Herein, the embodiment of the legal framework concerningenergy transmission and distribution as well as the adherent
ll rights reserved.
: +43131336 905995.
erstorfer).
markets at which the good—electricity or energy—is traded, haveto be mentioned. In this article the German balancing powermarket and its effective operation build the center of our analysis.Herein, we analyze the existing procurement auctions for tertiarycontrol from a theoretical as well as empirical point of view inorder to shed light on problems and potential improvements forfuture energy policies.
Since electrical energy cannot be stored, the maintenance ofbalance between supply and demand in the grid is one major taskof the transmission system operator. His duty is to estimatebeforehand how much electricity will be injected in or withdrawnfrom the network. Due to the increasing share of energy sourceswith variable output and the adjacent support schemes forrenewable energy sources, the real time balance becomes anincreasingly difficult task. An efficient and appropriate designedmarket is a necessary component in order to assist the balancingmanagement. This holds especially for the German power market,which constitutes the largest consumption and also the mostimportant transit market in the center of Europe. Initially,transmission system operators of the respective power controlareas and their associated power plants were the only supplier oftertiary control in Germany. Before the tendering procedure fortertiary control was established in 2001 and 2002, the systembalance had been provided by bilateral contracts betweentransmission operators and affiliated power plants.
ARTICLE IN PRESS
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–2627 2621
In 2001 and 2002, the Federal Cartel Office prescribed theimplementation of separate procurement auctions for the distinctpower control areas. According to the Energy Industry Act ofJuly 7, 2005, and the corresponding orders, the framework foracquisition and use of tertiary control has changed. The adjacentconsensual concept passed on August 29, 2006. Herein, the basisfor a common tendering procedure was found. Since December 1,2006, the transmission network operators offer a joint tenderingprocedure for tertiary control, which is supported by a web-based(electronic) platform (www.regelleistung.net). Among this, theregulator set uniform provisions, specified publication obligationsand limited the supply of tertiary control transmission systemoperators are allowed to purchase within their own power controlareas. Moreover, the timing of auctions, submitting of bids and theselection of providers in merit orders were defined. However, thereform, which targets the encouragement of the market entry oftertiary control providers and the decline of the current pricelevel, has not led to the desired outcome, yet. Certain character-istics of the new auction design are held responsible for thesedefects.
In order to identify attributes of the common procurementauction for tertiary control hampering market entry of providers,giving stimuli to collusion and strategic behavior, reducingintensity of competition and encouraging an upswing of prices,this paper analyzes the design under an auction theoreticalapproach. We examine features that have remained unchangedsince 2001/2002 and elements that were subject to modificationin 2006 separately. Supplemented with some empirical evidencewe want to deduce whether structural components of the auctiondesign have to be touched up again.
To achieve this aim, the remainder of the paper is structured asfollows. Section 2 reviews the pertinent literature. Section 3explains the structural characteristics of balancing power andminutes reserve markets as well as the design of procurementauctions for tertiary control in Germany. Section 4 presents thetheoretical analysis addressing the consequences of the auctiondesign for market structure and tendering behavior (barriers tomarket entry and transaction costs, collusion and strategicbehavior) and the implications for the price level and pricevolatility. The resulting outcomes are tested due to their evidencefor the German minutes reserve market. Section 5 concludes.
2 For further analysis see McAfee and McMillan (1987).
2. Literature review
The literature dealing with balancing power or balancingenergy focuses on three different strands. The first strand capturesthe evolution and efficiency of balancing power markets. This fieldof research is primary made up of country surveys in whichspecific markets for balancing power are analyzed. For example,Borenstein et al. (1999) measure the market power in theCalifornian electricity markets which cover in addition themarkets for load balancing and spinning reserve. Another articleis provided by Lloyd et al. (2003). They evaluate the success andfailures of electricity market reforms in UK, Norway, Canada(Alberta) and the US (California). One of their covered markets isagain the balancing power market. For the German market, suchan analysis is provided by Swider (2006). He investigates the legalframework, competitiveness and possible strategic behavior in theGerman balancing market until 2005.
The second field of research concerns the design of theauctions founding the basis for the balancing power market. Anauction is a process of buying and selling goods that are offered bya bidder at the bid price. As soon as all bids are collected, the goodis sold to the winning bidder at a certain price depending on theauction design. Therefore, an auction can be seen as a non-
cooperative multi-stage game.2 The literature in this area ofresearch analyzes the most appropriate auction design forbalancing power with respect to a certain legal framework andcountry-specific conditions. For California, this topic is discussedby Kahn et al. (2001), whereas Federico and Rahman (2003)examine the auction design for the UK electricity market.Theoretical evidence about the performance of modeling elec-tricity auctions is given in Fabra et al. (2002) and Klemperer(2002).
The third strand combines the two previously mentionedapproaches. Herein, the different types of auction awards areconsidered. The assigned award can be bound on the ‘‘pay-as-bidprice’’ or the ‘‘uniform price’’. Oppositional perceptions existconcerning ‘‘pay-as-bid’’ and ‘‘uniform prices’’ and their impact oncollusive behavior. In this context, the articles of Goswami et al.(1996), Klemperer (2002), Swider and Ellersdorfer (2005), Wawer(2005) and Swider (2007) have to be mentioned. In a monopolymarket on the one hand, the predominant position argues thatunder ‘‘uniform prices’’ the incentives for collusive behaviorincrease with available capacity. ‘‘Pay-as-bid prices’’ on the otherhand provide incentives for collusive behavior as the individualbid price and not the highest (possible) price is paid. In anoligopolistic market with pay-as-bid auction, the bidder can earnadditional profits if he prices minutes reserve close to theexpected marginal price of other competitors (as long as hisown costs are covered). This leads to strategic behavior becauseprices are set with reference to the expected behavior of othermarket participants and is not derived from the original costfunction. The opposite is true for uniform prices in an oligopolisticmarket.
This paper builds on the previous approaches and refines them.German procurement auctions for tertiary control are scrutinizedfrom the perspective of auction theory, particularly with regard tothe modification of auction design in December 2006 and itsimplications.
3. Structural characteristics of balancing power and theGerman auction design for minutes reserve
As it is not possible to store large amounts of energy, theoverall power generation has to match total energy demand ineach point in time. However, as production and demand are notconstant over time but are rather fluctuating (caused, for example,by station blackouts, stochastic feed-in (wind) or seasonalfluctuations in demand) this balance cannot be guaranteed.Consequently, a market mechanism had to be established in orderto assure a certain level of security of supply. This is the so-calledbalancing power system.
Differences of feed-in and withdrawal from a certain transmis-sion network can be of positive or negative sign. A positive signindicates that more energy is consumed than produced, leading toa reduction of frequency in the transmission network and in thewhole power grid, consequently. In contrast, a negative signimplies that network frequency is too high because more energy isproduced than consumed. The market for balancing energyprovides a mechanism to avoid this imbalance between genera-tion and load and presents an essential part for the stability of theoverall transmission and distribution system, therefore.
Balancing deviations between forecasted and actual feed-inand withdrawal can be substituted by the load-balancing group orby the transmission system operator itself. The traded energy forbalancing this system is called balancing power. Balancing power
ARTICLE IN PRESS
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–26272622
can be split up into three different groups of energy with differentcharacteristics and qualities, namely primary, secondary andtertiary control. Primary control starts within seconds as a jointaction. It must be balanced within 30 s, whereas secondary controlcan be seen as a long-lasting demand of primary control. Thebalance with secondary control has to be reached within 5 min.The system operator automatically calls up these two balancingpowers from the related power plants. The tendering for bothtakes place every six months. Tertiary control replaces secondarycontrol and implies a rescheduling in generation by the respon-sible suppliers. It is used as a scheduled delivery (positive ornegative delivery) based on quarters of an hour and needs to beactivated within 15 min.
The subject of a procurement auction in the German balancingmarket is the hold-out and ex-ante known amount of minutesreserve. This is called by the four main transmission systemoperators in Germany (EON, Vattenfall, EnBW and RWE). Thetransmission system operators are allowed to keep a maximum of50% of this needed amount for secondary and tertiary control ontheir own. The auction itself is divided into a number of separateauctions for positive and negative tertiary control as well as for sixtime slots per diem.3 Moreover, the auction takes place eachweekday for the following day (day-ahead auction). For weekendsand business holidays, the auction is scheduled at the lastweekday before. In-day timing for the auction is 8–10 a.m., rightbefore the spot market auctions take place.
For all transmission system operators, the auctioned amountfor tertiary control has to be stated on a common website at whichthe information has to be stored for 3 years. The informationprovided consists of the following aspects:
�
Dan
total amount of tertiary control demanded for the next day,4
�
the common share for all control areas, � the individual shares for each control area and � lists of all bids and the used minutes reserve of the previousday.5
For submitting bids, bidders have individual and private accessto information concerning the bidding process. A bid consists of
two different prices. The first one is the so-called demand (orcapacity) charge, which can be seen as a price for the option to useminutes reserve. Herein, the right for calling energy is founded,which can be determined by the opportunity costs the providershave to bear instead of selling energy at the secondary market(spot market). The second price is the commodity charge. Thisprice is primarily determined by the fuel costs for the adherentpower plant. Moreover, a bid has to specify the quantity of tertiarycontrol earmarked for the next day.Winning bids are published at 11 a.m. daily. The only criterionthat is taken into account is the demand charge. The list of allwinning bids is drawn up by following the concept of meritorders.6 Herein, bid prices are sorted in ascending order: the lowerthe rank, the more likely the acceptance of the bid. Afterwards,each transmission system operator who wants to use the minutesreserve builds a second merit order with respect to commoditycharges. Moreover, it is possible to use an amount with a highercommodity charge. This requires a justification; technical aspectscan be acceptable reasons.7
3 For deeper insight see BNetzA (2006).4 Concerning the calculation of balancing power see Bruckl et al. (2006) and
y et al. (2003).5 See BNetzA (2006).6 See BNetzA (2006).7 See Online trading place (2007).
4. Theoretical analysis of the German auction design
Subsequent to the description of the structural characteristicsof the German system, the procurement auction for tertiarycontrol is analyzed with respect to theoretical auction design. Ingeneral, auctions can be classified by the following character-istics8:
�
repeatability: single auction vs. repeated auction; � number of goods within an auction: single-unit auction vs.multi-unit auction; it is also distinguished whether a multi-unit auction is a simultaneous or a sequential auction;
� number of bidding rounds: single- vs. multi-stage auction; � open-bid vs. sealed-bid auction; � type of merit order; � pricing rules: first-price rule or second-price rule (onesuccessful bidder); pay-as-bid auction (or price-as-bid auction,discriminatory price auction) or uniform-price auction respec-tive market clearing price auction (several successful bidders);
� provision of information for bidders: private value, commonvalue or correlated value9;
� assumption of risk aversion: risk-neutral, risk-averse or risk-loving market participants.
Following the previously mentioned characteristics, the Germanauction design for minutes reserve can be characterized by thefollowing items10:
�
demand auction or procurement auction: sellers underbid eachother in pricing the supply of tertiary control; � repeated auction; � day-ahead auction; � multi-unit auction (positive and negative minutes reserve andtime slots are differentiated);
� electronic auction; � one-sided auction: bids are only provided by the seller; � single-stage auction; � sealed-bid auction; � multi-part auction: the bid price is divided into a demandcharge and a commodity charge;
� pay-as-bid auction: the payable amount is the individuallybidden demand charge.
4.1. Consequences for market structure and tendering behavior
4.1.1. Barriers to market entry and transaction costs
The intensity of competition and the behavior of marketparticipants, and correspondingly equilibrium prices and quan-tities, can be derived from the amount and relative importance ofbidders. In the context of competition, a main criterion is thepossibility of participating in the procurement auction deter-mined by the existence of legal, administrative and economicmarket entry barriers characterizing the German balancing powermarket.
The prequalification procedure constitutes the main barrier tomarket entry. Potential bidders proclaim that administrative costsfor entering the market are tremendous and cause extremeexpenditures besides auction costs. Furthermore, insufficientcoordination between different transmission system operators iscriticized. Technical requirements are labeled as too sophisticated
8 See for example Keuter et al. (1996).9 See Cheah (1994).10 See Swider (2006).
ARTICLE IN PRESS
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–2627 2623
and the necessity of complete availability of power plants isconsidered excessive.
In the first phase after implementing the procurementsystem, a lack of information about the tendering process, asdemanded amount, bid prices, winning bid and called minutesreserve as well as results of the procurement auctions, wascriticized because information asymmetries between incumbentsand newcomers were aggravated. This problem was overcomewith the new tendering procedure set in place in December 2006.The new bidding provisions have ensured an increase intransparency and a decrease in uncertainty such that possiblefirst mover advantages of incumbent companies have beenreduced.11
As previously mentioned, the German system allows transmis-sion system operators to hold a certain share of needed tertiarycontrol amounts a priori. This fact antagonizes free market entryfor participants from different balancing power regions. Areduction or dwindling amounts of these shares may lead tomore competition and lower costs. One step in this direction wasthe introduction of a common auction because it simplifies thepossibility of participating in the market and diminishes transac-tion costs. Another entry barrier is given by the minimumquantity bid. The new tendering procedure implemented inDecember 2006 has reduced this hurdle, and the now allowedpooling of different kinds of controlling energy allows smallercompanies to join the auction.12
Long-lasting time slots and the adjacent duration of theobligation to reserve capacities for minutes reserve imply anexclusion of certain capacities that are not usable for otherpurposes for the whole timestamp. Long-lasting time slotsforeclose small power plants from participating in the marketbecause these do not have the possibility to manage or hedge theupcoming risk. The decrease in the duration of time slots has ledto the possibility that excess capacity of power plants, which isrestricted in time, can be allocated in the market. Moreover, dailyauctions that have been in use since 2001 avoid the tendency tocollusive behavior and push the implementation of a diversifiedstructure and a high liquidity for the balancing power market.13
In the context of entry barriers, also problems concerningtransaction costs have to be mentioned. In general, these have tobe paid by the bidders themselves and cannot be socialized. Costsmay indeed rise as soon as market entry barriers are downsized,i.e. lower minimum quantity bids, shortened durations of timeslots and daily auctions induce higher transaction costs ascomplexity of running a procurement auction increases. Likewise,using single pricing instead of multivariate or bivariate priceswithin the tendering procedure can cause a decrease in transac-tion costs. The change in the market design which tends tobroaden the market, the common procurement of the certainshare of the needed tertiary control and the placing of bids whichfollows a common merit order have certainly reduced transactioncosts as soon as searching costs and information costs aredeclining with the establishment of a common procurementauction.14
4.1.2. Collusion and incentives for strategic behavior
Collusion and strategic behavior are highly correlated with themarket structure and, hence, with the existence of market entrybarriers. Collusion is given by coordinated behavior in settingprices and the signaling and influencing of other bidders in order
11 See Nailis and Ritzau (2006).12 See Turkucar (2006).13 See BNetzA (2006) and Turkucar (2006).14 See Wawer (2005).
to affect bids.15 An oligopolistic market with just a few actors onthe bidding side includes the danger of practicing market powerresulting in highly coordinated and strategic behavior. Thistendency grows as soon as bidders are linked by institutionalizedand informal relations obviously given for vertically integratedcompanies. Collusive behavior intensifies further if an auction isrepeated in regular intervals and concerns the same marketparticipants.
Unwritten agreements between the bidders for minutes reserveinvolve a lower number of independent market participants andenhance possible profits for incumbents. This holds becauserevenues are positively correlated with demanded compensation;collusive behavior reduces the bidding amount whereas thedemand charge remains high. Theoretically, this kind of strategicbehavior could lead that far that the bidder with the lowest costscan be determined a priori such that excess rents can be allocatedbetween market participants without taking any risk. Because ofthe mechanism socializing costs between transmission and systemoperators, there have to be doubts that transmission operatorsthemselves have an incentive to avoid collusive behavior.
In the context of auctioning minutes reserve, the mostcommon problem with collusive behavior is given by bidderswith market power to be inclined to clear the market incoordination with other oligopolies. This behavior signals smallercompanies or new entrants that they will not make profits whenjoining the market. Consequently, a single company or anoligopoly can restrict market access and create scarcity; fortransmission system operators, scarcity implies acceptable (low)costs. In general, there may be an economic incentive to offerextremely low demand charges and very high commodity charges.The opportunity costs of keeping capacity are compensated by thedemand charge, whereas actual provision of minutes reservebecomes unlikely.16
Whether collusion is stable depends on the incentive forparticipating agents to deviate from coordinated behavior. Inorder to avoid collusion, an adaptation of the auction design canbe necessary. For example, single-unit sealed-bid auctions do notprovide high incentives for strategic behavior because thestimulus to sheer off (choosing a lower demand charge) is ratherhigh. Concededly, in the case of repeated auctions a cheater isexposed to be identified more easily.
Moreover, collusive behavior may be more probable when theacceptance of a bid takes place by means of uniform prices asopposed to pay-as-bid prices. Incentives for collusive behavior arehigher under uniform-price auctions because a bidder withmarket power can force prices to increase by running energyshort. The other companies with minor market shares do not havean incentive to underbid these prices because their own revenueswould decrease. Consequently, inframarginal bids save the ownsupply whereas marginal bids raise the market price. Instead, pay-as-bid pricing reduces strategic incentives because the final priceis only determined by the own proclaimed price and notdetermined by the highest marginal price.
Another view claims that oligopolistic market structures andpay-as-bid pricing provide an incentive to price as close aspossible to the expected marginal price. The upcoming profit goesup the higher the bid price is above short-run marginal costs.Considering the lack of perfect market transparency and informa-tion, price setting becomes a guess for smaller companies nothaving the same possibilities to forecast the expected marginalprice.17 As a consequence, they do not enter the market and
15 See Keuter et al. (1996).16 See Nailis and Ritzau (2006, p. 44).17 Swider (2007) proclaims that this forecast is highly correlated with prices of
the previous day.
ARTICLE IN PRESS
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–26272624
contestability is reduced. However, uniform pricing implies thatbid prices equal short-run marginal costs but also that themarginal price is above marginal costs. Smaller companies are notdiscriminated and will furthermore benefit from the abuse ofmarket power by more influential bidders. Consequently, thisbrings more competition and the incentive for collusive behaviordecreases.
The modification of the tendering procedure in December 2006has certainly cut incitements for strategic behavior because themarket for minutes reserve follows a common structure. Ideally,supply forms a single supply curve and allocative efficiency isprovided by the market itself. Thus, abuse of market power isnearly prevented. The common procedure for auctioning and thecommon supply for all balancing zones have led to lessinformation asymmetry and have reduced strategic behavior.
At least, the duty to disclose results, amounts and bids canimprove transparency and can avoid collusion. A single biddergets information about demand and commodity charges andtendering amounts of all market participants. Estimating themarginal bid is facilitated and the probability of accepting the bidis enhanced. Further, production costs can be controlled moreeasily and the use of capacity can follow a better planning leadingagain to more competition.18
Furthermore, the phenomenon of the winner’s curse has to bequoted. In the context of minutes reserve, bidders have to takeprivate and common costs into account. Private costs aredetermined by individual factors for each bidder as for examplelong-term agreements and individual spot market activity. Theexistence of a common factor is given by the fact that all biddersare confronted with similar factor prices and with identicalprobabilities that the minutes reserve is called afterwards.Considering high common costs, uniform-price auctions providea better market system than pay-as-bid pricing.19 The reasontherefore is the so-called winner’s curse.20 If individual costs forall market participants are very similar, pay-as-bid pricing implieslower market action in bidding and provides higher costs fortransmission system operators because bidders add a premium totheir real costs in order to compensate for the possibility of thewinner’s curse. This tendency is more relevant to new entrants asthey are confronted with higher uncertainty of the real value of acertain good. Yet, demand auctions are affected by the winner’scurse to a lesser extent than supply auctions because the realvalue of a good is more likely known by the bidders.
4.2. Implications for price level and price volatility
The following section provides evidence of the development inprices and price volatility after the establishment of the newtendering procedure within the German balancing power market,especially for tertiary control. Therefore, we distinguish theore-tical consequences that are determined by the previouslymentioned literature and empirical evidence of the Germanmarket for minutes reserve.
4.2.1. Auction theoretical implications
Considering the previously mentioned aspects, two hypothesesdealing with the development of prices and price volatility can be
18 See Klemperer and Meyer (1989).19 See Wawer (2005).20 The winner’s curse takes place in auctions. Herein, the auctioned item
provides nearly for all bidders an equal value. However, the bidders do not know
the true item’s market value when they bid. The winner of the auction is the bidder
making the highest estimate. If it is assumed that the average bid is accurate then
the highest bidder overestimates the item’s value. Thus, the auction’s winner is
likely to overpay, which is called the winner’s curse.
formulated. First, lower market entry barriers, the reduction ofcollusive stimuli and avoidance of the winner’s curse imply lowerprice levels. Decreasing market barriers raises the number ofmarket participants and lessens individual costs. Second, theprovided market information ensures that the bidding strategydirectly follows the necessity of the market, leading to moreintensive competition and a reduction in demand charges.
Furthermore, the design of the German balancing powermarket can straightly have an impact on the price level. Long-lasting procurement times do not imply lower prices automati-cally. In the context of long-term aspects, opportunity costs canplay a major role.21 These originate from lost profit that otherwisewould have been realized at the spot market. The higheruncertainty of long-term procurement leads to a risk premium.On the other hand, short-term, e.g. daily, demand auctions areassumed to increase price volatility compared to long-termones.22
In addition, the timing of auctions is decisive. The auction forminutes reserve takes place daily and is settled before the day-ahead auction for the spot market. Therefore, the spot prices for acertain day (based on the spot market auction of the previous day)are known, whereas the spot prices for the next day determiningopportunity costs are unknown. Hence, it can be concluded thatprice volatility may be lower if the balancing power auctionterminates right before the spot market auction. However, thisschedule could also give incentives to surging bidding prices(demand charges) because bids that do not win the auction can besupplied at the spot market at the same day.23
Discussing uniform-price and pay-as-bid auctions in thecontext of collusive incitements, the conclusions may be drawnthat pay-as-bid pricing strengthens price competition and reducesthe adjacent price volatility. Newer and empirical analyses cannotdetermine a fixed effect of the pricing system on the level of spotprices. Following Ausubel and Cramton (1998, 2002) uniform-price auctions increase the price level if a private value auction isconsidered. As possible reasons, strategic behavior and lowerefficiency are mentioned. Staropoli et al. (2000) show that bothprocedures provide price premiums if there are certain bidderswith market power. Rassenti et al. (2000, 2003) prove that pricevolatility decreases if—under significant market power—uniformpricing is replaced by pay-as-bid pricing. They demonstrate thatincreasing competition in a market with pay-as-bid pricing letsprices converge to a level associated with uniform pricing undermarket power. Consequently, they show that the price level islower under uniform pricing.
If there is no asymmetric information, it can be shown that theprice level and volatility are similar for both methods. Referring tothis, Federico and Rahman (2003) provide evidence that underconsideration of symmetric information concerning costs andasymmetric information relevant to demand a switch fromuniform pricing to pay-as-bid pricing supports a reduction ofoutput, a welfare loss and an increase in consumer surplus. Undermarket concentration this changeover implies a profit and outputdecrease but also leads to a growth of consumer surplus.Concerning welfare, the finding is dependent on marginal costsand the degree of demand uncertainty.
4.2.2. Empirical evidence
As mentioned in Section 1, well-designed markets and awell-defined regulatory framework build the basis for effectiveenergy policy. A well-functioning balancing market will therefore
21 See Swider (2007).22 See BNetzA (2006).23 See Swider (2006) and Swider and Weber (2003).
ARTICLE IN PRESS
0
50
100
150
200
250
300
350
400
450
Pre_Mean
Price Level and Price Volatility
Positive Minutes ReservesNegative Minutes Reserves
Post_Mean Pre_Std. Post_Std.
Fig. 1. Price level and price volatility.
Correlation positive minutes reserves
0
0.70.60.50.40.30.20.1
RWE EnBW E.ON VE-T
Pre_correlations Post_correlations
Fig. 2. Positive minutes reserve correlation with spot prices.
Pre_Korrelation
0
0.4
0.3
0.2
0.1
Post_Korrelation
RWE EnBW E.ON VE-T
Correlation negative minutes reserves
Fig. 3. Negative minutes reserve correlation with spot prices.
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–2627 2625
contribute this. A market can be identified as being wellfunctioning and efficient if the integration of several markets isgiven or has increased due to a change in the market design.Therefore, we analyze the statistical features of the time series forpositive and negative minutes reserves prices as well as thecorrelation between these and the adherent spot market.
In this section, we analyze the impact of the balancing powermarket design on prices and price volatility of minutes reserve. Ina first step, we therefore compare price levels and price volatilitybefore and after the change in the tendering procedure atDecember 1, 2006. In a second step, we analyze the impact ofthe common tendering procedure on the interconnection betweenindividual demand charges, provided by the main market players(RWE, EON, Vattenfall and V-ET), and the spot prices of the EEX,given by the Phelix-baseload price.24 An increase in dependenceof prices would imply on the one hand that prices of theindividual power control areas converge and on the other handthat the termination of the auction leads (especially in the longrun) to higher integration and increasing dependence of spotand minutes reserve markets. It further means that high spotprices force minutes reserve prices to rise, too. Contrariwise, adecrease in the dependence would signalize that individual pricesdo not converge and, consequently, that the newly establishedtendering procedure does not have an impact on the balancingmarket at all.
The data used for the analysis are daily data for tertiary control.The considered period lasts from January 1, 2006, to June 11, 2007.We constructed a combined data set based on the data providedby the balancing group leaders (transmission system operators).Fig. 1 shows mean and standard deviation for balancing powerprices distinguished in positive and negative tertiary control.There is evidence that the mean for both price time seriesincreased after December 2006. For the standard deviation theopposite is true. Herein, the common tendering procedure led toabatement in volatility for positive and negative minutes reserve.
Concluding, within this short period after establishing thecommon trading platform the expected fall in the price level aswell as the increase in price volatility did not arise. With this shortanalysis it is not possible to exclude that several other parameters,besides the change at December 1, 2006, have an impact on theseresults. In order to attenuate the political implications of theseresults in a second step we provide insights in the dependencebetween the spot and reserve market.
Figs. 2 and 3 show the correlation before and after December 1,2006, between positive minutes reserve and Phelix-baseloadprices and, respectively, for negative minutes reserve andPhelix-baseload prices. It is obvious that the correlation increasedfor positive minutes reserve whereas for negative minutes reservethe opposite holds.
This result is not rather astonishing as positive minutes reserveacts as a substitutable good to energy provided at spot markets.On the other hand, negative minutes reserve builds some kind ofopposite product or mirror-inverted product. Bearing this in mind,it becomes clear that positive minutes reserve prices and spotprices converge and affect each other. Consequently, high spotprices lead to high prices for minutes reserve and vice versa. Theincreased price level, shown in Fig. 1, is therefore not only causedby the change in the tendering procedure but peculiarly induced
24 The data for the Phelix spot prices are provided by Bloombergs. Both prices
are average daily spot prices that result from aggregating the hourly spot prices for
power at the European Energy Exchange (EEX). The Phelix_Base spot price is
calculated via the following formula: Phelix_Base ¼P24
i¼1hourly spot prices=24.
This price is given as arithmetic mean of the hourly auction prices summed up over
a day without considering any congestion. Days with clock-change imply that the
number of considered hours reduce to 23 or increase to 25.
by the observable increase in spot prices. Furthermore, the effectthat all individual prices show higher correlation with spot pricesprovides evidence that price setting converges and certaincompanies are no longer able to set prices on their own. Theyare moreover price takers in the new arranged balancing powermarket.
Concluding, positive minutes reserve prices are driven by spotmarket prices whereas negative minutes reserves are not. Forbalancing power prices, the establishment of a common tradingplatform results in observable higher prices and lower pricevolatility. Consequently, bearing Fig. 2 in mind, the integrationbetween spot market and the market for positive minutesreserves increases due to the change in the tendering procedure.This leads to the conclusion that the efficiency of the balancing
ARTICLE IN PRESS
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–26272626
power market is improved. From an energy political point of view,this induces furthermore an improvement for the overallregulatory system.
5. Conclusions
The German tendering procedure for minutes reserve changedits auction design with December 1, 2006. This article deals withthe theoretical implications and consequences of the establishedauction design before and after this system change. Herein, wedistinguish between the aspects that were not revised inDecember 2006 and the aspects constituting the basis for themodification. The request for prequalifying, the repeated and dailyauction for the day ahead, the single-stage sealed-bid auction andthe two-part tariffs in combination with the merit order and pay-as-bid pricing are characteristics that were not subject to changein December 2006:
�
The used procedure for prequalifying restricts market entryand indirectly encourages collusive behavior through coordi-nated bidding. Furthermore, this procedure implies a high levelof transaction costs that can be seen as a premium on demandcharges. � A daily auction simplifies market entry and assists a lowerprice level but increases price volatility. The fact that theprocurement auction is a repeated auction increases thedanger of collusion.
� A single-stage sealed-bid auction restricts possible strategicbehavior and implies lower demand charges.
� Pay-as-bid pricing allows cost-oriented pricing and implies arestriction of price volatility. The phenomenon of winner’scurse is rather implausible.
Consequently, the previously mentioned characteristics ofthe German auction design for balancing power provide thepossibility for low price levels. The main criticism for the systemis still given by the prequalifying process that can prevent marketentry and can therefore induce a small number of marketparticipants.
In a second step, we analyze the results of the new auctiondesign dating December 1, 2006. In this context, we examine theconsequences of the conflation of the online trading places, thecommon auction for minutes reserve (demand independent fromeach balancing zone), the harmonization of time stamps, thestandardization of time slots and minimum quantity bids, and theobligation to release essential information.
�
The common auction design (time stamps and slots, meritorder, online trading places, etc.) and the broad duty to publishinformation about pricing and tendering quantities increasetransparency and reduce uncertainty and information costs.Further, this implies lower barriers to market entry andminimizes the incentive for strategic behavior. The level ofdemand charges can be expected to decline. � The termination of the auction for minutes reserve before theday-ahead spot market should reduce price volatility. Pricescan be anticipated to rise.
� The reduction of minimum quantity bids, the abridgement oftime slots and the duty to disclose information ease marketentry and diminish the risk of collusion. As transaction costswill pick up, the effect on demand charges is uncertain.
Summing up the theoretical arguments it remains unclearwhether prices should go up or down due to the reform ofDecember 2006. Up-front it is consequently not clear whether the
energy political aims are reachable due to theoretical evidence. Iftransaction costs can be neglected and if the reservation of sharesof minutes reserve to be purchased in the own control area will beabolished, demand charges will be lower owing to the revision ofthe auction design, at least in the long run. Additionally, otherthoroughgoing reforms of the auction design are not deemednecessary. The adherent empirical analysis provides evidence ofthe increase in price level and the reduction of price volatility fornegative as well as for positive minutes reserve. Furthermore, thedependence of demand charges and spot prices show—althoughother factors than the underlying tendering procedure can mainlydrive prices and antagonize theoretical implied scopes—that therecently political steps lead to a more efficient balancing powermarket and therefore to an improvement of the regulatoryframework. Our analysis showed that concerning the Germanbalancing power market no definite adjustment of energy policy isneeded.
References
Ausubel, L.M., Cramton, P., 1998. The optimality of being efficient. Department ofEconomics, University of Maryland (18 June 1999).
Ausubel, L.M., Cramton, P., 2002. Demand reduction and inefficiency in multi-unitauctions. Mimeo, University of Maryland.
Borenstein, S., Bushnell, J.B., 1999. An Empirical Analysis of the Potential for MarketPower in California’s Electricity Industry. Journal of Industrial Economics 47,285–323.
Bruckl, O., Neubarth, J., Wagner, U., 2006. Regel- und Reserveleistungsbedarf einesUbertragungsnetzbetreibers. Energiewirtschaftliche Tagesfragen 56, 50–55.
Bundesnetzagentur (BNetzA), 2006. Beschluss BK6-06-012 der Beschlusskammer 6vom 29.08.2006 in dem Verwaltungsverfahren wegen der Festlegung zuVerfahren zur Ausschreibung von Regelenergie in Gestalt der Minutenreserve,Bonn.
Cheah, C.W., 1994. Auction Design Strategies for Price-based Allocation ofSpectrum Licences. Centre for International Research on Communication andInformation Technologies.
Dany, G., Haubrich, H.-J., Schlecht, D., Joswig, R., 2003. Bemessung desRegelleistungsbedarfs im liberalisierten Strommarkt. Internationaler ETG-Kongress 2003 ‘‘Energietechnik fur die Zukunft,’’ Hamburg. ETG-Fachbericht94, Berlin–Offenbach, pp. 281–288.
Fabra, N., von der Fehr, N.H., Harbord, D., 2002. Modeling electricity auctions. TheElectricity Journal 15, 72–81.
Federico, G., Rahman, D., 2003. Bidding in an electricity pay-as-bid auction. Journalof Regulatory Economics 24, 175–211.
Goswami, G., Noe, T.H., Rebello, M.J., 1996. Collusion in uniform-price auctions:experimental evidence and implications for treasury auctions. The Review ofFinancial Studies 9, 757–785.
Kahn, A., Cramton, P., Porter, R., Tabors, R., 2001. Uniform pricing or pay-as-bidpricing: a dilemma for California and beyond. The Electricity Journal 14,70–79.
Keuter, A., Nett, L., Stumpf, U., 1996. Regeln fur das Verfahren zur Versteigerungvon ERMES-Lizenzen/Frequenzen sowie regionaler ERMES-Frequenzen. WIK-Diskussionsbeitrage 165. Bad Honnef, September 1996.
Klemperer, P., 2002. What really matters in auction design. Journal of EconomicPerspectives 16, 169–190.
Klemperer, P., Meyer, M., 1989. Supply function equilibria in oligopoly underuncertainty. Econometrica 57, 1243–1277.
Lloyd, D., Tishler, A., Woo, C.K., 2003. Electricity market reform failures: UK,Norway, Alberta and California. Energy Policy 31, 1103–1115.
McAfee, P., McMillan, J., 1987. Auctions and bidding. Journal of Economic Literature25, 699–739.
Nailis, D., Ritzau, M., 2006. Studie zur Marktgestaltung der Regel- undAusgleichsenergie vor dem Hintergrund des neuen EnWG, RAE-Studie. Aachen,16 January 2006.
Online trading place, 2007 /https://www.regelleistung.net/regelleistungWeb/statisch/auschreibung_mrl.jspS.
Rassenti, St.J., Smith, V.L., Wilson, B.J., 2000. Controlling market power and pricespikes in electricity networks: demand-side bidding. Working paper, Uni-versity of Arizona.
Rassenti, St.J., Smith, V.L., Wilson, B.J., 2003. Discriminatory price auctions inelectricity markets: low volatility at the expense of high price levels. Journal ofRegulatory Economics 23, 109–123.
Staropoli, C., Finnon, D., Glachant, J.-M., Jullien, C., Quatrain, R., Robin, S., Ruffieux,B., 2000. Modifying industry structure or market institution? An experimentalanalysis of the reform of the english electricity pool. Working paper, UniversiteParis I.
Swider, D.J., 2006. Handel an Regelenergie- und Spotmarkten. Methoden zurEntscheidungsunterstutzung fur Netz- und Kraftwerksbetreiber. ReiheWirtschaftswissenschaft, Wiesbaden.
ARTICLE IN PRESS
G. Muller, M. Rammerstorfer / Energy Policy 36 (2008) 2620–2627 2627
Swider, D.J., 2007. Wettbewerb am deutschen Regelenergiemarkt? Energie-wirtschaftliche Tagesfragen 57 (9), 32–37.
Swider, D.J., Ellersdorfer, I., 2005. Kosteneffizienz am deutschen Regelenergie-markt. Energiewirtschaftliche Tagesfragen 55, 802–806.
Swider, D.J., Weber, C., 2003. Ausgestaltung des deutschen Regelenergiemarktes.Energiewirtschaftliche Tagesfragen 53, 448–453.
Turkucar, T., 2006. Der deutsche Regelenergiemarkt und die Rolle der Ubertra-gungsnetzbetreiber. Intensiv-Seminar: Ausgleichs- und Regelenergie fur Stromund Gas, Berlin, 9 February 2006.
Wawer, T., 2005. Effiziente Ausgestaltung von Regelenergieauktionen zur Verrin-gerung der Netznutzungsentgelte. Proceedings IEWT 2005—Energiesystemeder Zukunft: Herausforderungen und Losungspfade, Wien.
