This item is only available as the following downloads:
1 ESSAYS ON THE NEGOTIATED SE TTLEMENT PRACTICE IN PUBLIC UTI LITIES REGULATION By SHOURJO CHAKRAVORTY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Shourjo Chakravorty
3 To my parents, Bhaswati and Swapan
4 ACKNOWLEDGMENTS I would like to thank my dissertation committee, David Sappington, Sanford Berg, Chunrong Ai and Robert Thomas for their help and guidance. I am grateful to Steven Slutsky, Ted Kury, Mark Jamison and seminar participants at the University of Florida for th eir helpful suggestions. I would also like to thank Stephen Littlechild, Dale Mailhot, Ann Cole and J. R. Kelly for providing me with the data used in this dissertation. I am grateful to the Economics Department at the University of Florida and the Robert F. Lanzilotti Public Policy Research Center for providing financial assistance during the course of my research.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 ABSTRACT ................................ ................................ ................................ ..................... 8 CHAPTER 1 THE NEGOTIATED SETTLEMENT PRACTICE IN PUBLIC UTILITIES REGULATION ................................ ................................ ................................ ........ 10 1.1 Introduction ................................ ................................ ................................ ....... 10 1.2 Why do Parties Settle? ................................ ................................ ..................... 14 2 NEGO TIATING WITH A CONSUMER ADVOCATE IN PUBLIC UTILITIES REGULATION ................................ ................................ ................................ ........ 31 2.1 The Model ................................ ................................ ................................ ......... 31 2.2 Results ................................ ................................ ................................ .............. 37 2.2.1 The Earnings review Case ................................ ................................ .... 37 2. 2.1.1 The full information case with a perfect consumer advocate ... 44 188.8.131.52 Bargaining results in the Settlement Stage .............................. 45 184.108.40.206 The imperfect information case ................................ ............... 49 2.2.2 The Company request Case ................................ ................................ 55 2. 3 Policy Implications ................................ ................................ ............................ 64 2.4 Conclusion ................................ ................................ ................................ ........ 68 3 DO CONSUMERS ALWAYS GAIN FROM NEGOTIATED SETTLEMENTS IN PUBLIC UTILITIES REGULATION? EVIDENCE FROM FLORIDA ........................ 70 3.1 Introduction ................................ ................................ ................................ ....... 70 3.2 Theory ................................ ................................ ................................ ............... 75 3.3 Empirical Model ................................ ................................ ................................ 80 3.3.1 Equation 3 ....................... 82 3.3.2 Equation 3 d Profit Function ................................ 85 3.3.3 Instrumental Variables for ................................ ............................. 87 3.4 Data ................................ ................................ ................................ .................. 87 3.5 Empirical results ................................ ................................ ................................ 89 3.5.1 OLS Regression of on Instruments ................................ ............... 89 3.5.2 Bivariate Probit Model with Partial Observability of the Negotiated settlement Pract ice at the FPSC ................................ .......................... 89 3.6 Conclusion ................................ ................................ ................................ ........ 91
6 APPENDIX A PROOF OF THE CANDIDATE SOLUTIONS TO EQUILIBRIUM 2 17 ................... 97 B PROOF THAT THE SPE PARTITIONS OF AND ARE EXPRESSIONS 2 18a, 2 18b AND 2 18c ................................ ................................ ............................ 98 LIST OF REFERENCES ................................ ................................ ............................. 103 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 107
7 LIST OF TABLES Table page 3 1 Distribution of negotiated settlements industry wise ................................ ........... 94 3 2 Summary statistics of independent variables ................................ ...................... 94 3 3 OLS regression of rate on the instrumental variables ................................ ......... 95 3 4 Bivariate probit model with partial observability of the negotiated settlement practice at the FPSC ................................ ................................ .......................... 96
8 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ESSAYS ON THE NEGOTIATED SETTLEMENT PRACTICE IN PUBLIC UTILITIES REGULATION By Shourjo Chakravorty August 2012 Chair: David E. M. Sappington Cochair: Sanford V. Berg Major: Economics The US and Canadian regulatory commissions use rate of return regulation (ROR) as a formal means to set consumer rates of public utilities firms. Ove r the past 25 years, negotiated settlements have increasingly come to replace ROR regulation In negotiated settlements, after a rate case has been initiated and before the formal hearing is scheduled, the utility, a consumer representative and other inter ested parties negotiate a settlement concerning consumer rates. Although this is common, economists have neglected this subject. The consensus in recent research is that the motivations for parties to settle in a rate case are different from those in litig ation. Consequently, a new theory is required to analyze such agreements. This dissertation examines the effect of such settlements on consumers when represented by a consumer advocate. The first chapter studies the background and motivations for negotia ted settlements in regulation.
9 The second chapter formulates a theoretical model of the negotiated settlement process. In the full information case in which the parties know the forthcoming regulation outcome and with a consumer advocate who accurately rep resents the because any decision by the regulator can be replicated in the settlement. In the imperfect information case, the consumer always benefits from a negotiated sett lement firm, sometimes the consumer is no better off with a consumer advocate with an identical preference than another whose preference is not. This result has impli cations for the delegated bargaining literature. The chapter ends with some policy recommendations. The third chapter uses data from the Florida Public Service Commission to ob servable bivariate probit model. The estimation suggests that the advocate puts more weight on present rate reductions for the consumer than future consumer welfare in his decision to settle. This leads to the possibility of settlements detrimental to cons umer interests. The results also show, unlike what parties claim, time savings caused by settlement are not a main reason for their occurrence.
10 CHAPTER 1 THE NEGOTIATED SETTLEMENT PRACTICE IN PUBLIC UTILITIES REGULATION 1.1 Introduction R ate of return (ROR) regulation has long been a popular means of regulating the earnings of public utilities in the US and Canada. Under ROR regulation, the regulator sets the prices charged by the utility to ensure a revenue that allows the utility to The rate base is What constitutes the rate base, the return the firm is allowed to earn on it s rate base and exactly what constitutes a fair return are all determined by the regulatory public commission after a process of formal hearings. Increasingly, this long established procedure of setting base rates is being replaced by a new regulatory technique. 1 This technique is negotiated settlement or stipulation Negotiated settlement entails the firm, the consumers and other interested parties (or their representatives) agreeing upon reasonable prices and other issues and submitting the agreement to the re gulator for approval. Traditionally, t he regulator determines base rates after the firm, consumer advocate (the consumer representative) and other interested have filed their testimonies and all witnesses have been cross examined at the hearing In negoti ated settlements, it is the firm, the consumer advocate and oth er parties who decide the rates. The regulator needs only to approv e the agreement for the rates to be implemented. The 1 Littlechild (2009a) shows the g rowth of the negotiated settlement practice at the Florida Public Service Commission (FPSC). Wang (2004) and Doucet and Littlechild (2009) show the increase of the same practice at the Federal Energy Regulatory Commission (FERC) and the Canadian National E nergy Board (NEB) respectively.
11 critics of negotiated settlement argue that negotiated settlements transfer decision making power from the regulator to the firms and large consumers. They worry that all consumers may not be protected in these settlements. The settleme nt process by its ve ry nature means that the compromises that helped to secure agreement are not made fully public. Even if they are made public, consumers may lack the sophistication to grasp all the implications of such an agreement. Consumer interests are more at risk when they are represented by an agent whose preferences may not be identical to theirs and may have their own motives for reaching settlement. In spite of these potential drawbacks to negotiated settlements regulatory commissions encourage this type of regula tion by t heir actions and some have even given explicit support 2 In spite of the growing use of t his regulatory instrument, little formal analysis of negotiated settlements has been undertaken 3 The consensus of existing research is that the main advantag e the settlement process provides is that more than one issue can be decided simultaneously as a total package This allows for potential trade offs between issues resulting in beneficial outcomes for all parties These trade offs are not possible in regul ation where the regulator decides each issue separately. The studies also agree that the main reasons why a plaintiff and a defendant come to a settlement before trial in a legal dispute (saving the time and cost of a trial, and reducing the uncertainty from a trial verdict) are not important when the parties of a rate case are 2 such settlement is entered into fairly and in good faith by competent parties, and is not procured by fraud or overrea Compromise, Accord and Release 9. 3 For a survey of the existing legal and economic literature on negotiated settlements see Littlechild and Doucet (2006).
12 deciding to settle. For this reason, models which analyze settlements in litigation literature 4 are not ideally suited for an investigation of negotiated settlements in regulation. A new model is required to analyze the latter based on the predominant reason why public utilities firms, their consumers and other interested parties settle a rate case. The next chapter uses the observations of past research on negotiated settlements in regulation to construct a model that investigates whether consumers can benefit from such agreements even when they are represented by a consumer advocate whose pref erences differ from theirs. The chapter formulates a two stage model. In the first stage, the consumer advocate and the firm bargain simultaneously over the shares of two amounts according to a Rubinstein infinite horizon alternating shortfall relative to the app roved revenue set in the last rate case. This amount establishes the rate consumers are to pay. The second amount is a result of other items sharing plans, and future behaviour of the parties frequently found in these types of agreements The se items serve as a mechanism for the firm to retain the future earnings it would otherwise have to give up to the consumer in ROR regulation The study models these retained future earnings of the firm as a reduction of the perceived future cost (the second amount) of operating under a ROR regulation regime. If negotiations fail then the regulator determines the shares of the overearnings or shortfall going to both parties. 4 For example, Bebchuk (1984) and Reinganum and Wilde (1986).
13 The central finding of the chap ter is that knowledge and the consumer advocate rep accurately then negotiated settlements always lead to Pareto improving outcomes for the consumer. In fact, in such a situation there would be no regulation stage at all. The simple intuition is that the negotiated settlement process can provide Pareto improving outcomes because it allows trade offs between issues unlike regulation. If the regulatory decision is known beforehand then any such outcome can be replicated in the settlement stage itself if no other better solution is feasible for both parties. But if the agreements may harm the consumer fo r certain values of the overearnings (or shortfall) and future cost For other values, however, the advocate will act in the best interest of preferences. The results are then used to expl ain the observations of previous studies and several policy steps for the regulator are recommended in order to safeguard consumer welfare. The model is a delegated bargaining model. It is shown that in some situations the consumer (the principal) is no better off during bargaining by employing an advocate (an agent) who internalizes her payoff function rather than another who does not. Since Schelling (1956) there has been a literat ure on how a principal may benefit by selecting an agent with different preferences to bargain on his behalf Examples include Rogoff (1985), Jones (1989), Persson and Tabellini (1994), Segendorff (1998) Brueckner (2000), Klumpp (2009) and Harstad (2010). such an agent in all these models is the same. By employing an agent who is a tougher bargainer than himself, the principal credibly commits to a stronger bargaining position.
14 If negotiations do not break down, the prin cipal is able to gain a bigger reward than by self representation. Extending this logic, it seems obvious that a principal will prefer to selfish agent (who only car es about his own utility). Indeed some models (for example, Sliwka 2007) 5 assume that the principal is always better employing an unselfish agent. Lammers (2010) shows that the principal may want to hire a selfish agent if the job requires bargaining with a third party. Lammers (2010) contends an unselfish agent who also has a tendency to internalize some of his better choice In th e ensuing analysis a perfect consumer advocate (who has the same utility function as the consumer) is always a tougher bargainer than a selfish one The perfect representative is for the same share of overearnings (or shortfall) The analysis shows that when bargaining simultaneously over two pies the consumer (the principal) sometimes can do no better by hiring a perfect advocate (the agent) rather than a selfish one. Unlike Lammers (2010), no behavioural assumptions are required to prove this result. 1.2 Why do Parties S ettle? A natural tendency is to compare the settlement process in regulation to that in litigation. In the economic literature of litigation, parties prefer to settle because trials are costly, their verdicts uncertain and they take time (Cooter and Rubinfeld 1989). These 5 Although in Sliwka (2007) the work the agent is employed to do is not restricted to bargaining.
15 are sometimes the reas on s given by the parties themselves in a rate case 6 But studies on negotiated settlements in regulation (Wang 2004; Doucet and Littlechild 2006; Litt lechild 200 9a, b ) argue that these are not the primary reasons why a utility and other parties in volved in a rate case settle. Littlechild (2009b) in his more detailed look at the stipulations determining base rates of ele ctric companies at the FPSC estimates that the costs saved by settling is at most slightly over one fourth of the total cost of th e full regulation process. I t is unlikely that cost savings are the main reason for utilities to settle when these settlements determine the outcome of revenues four orders of magnitude greater than the regulatory costs saved (Littlechild 2009b). He say s that this is also true of the Office of Public Counsel (OPC) in Florida which is responsible for representing consumers in regulatory formal hearings. The OPC would not be concerned about saving the cost of a formal hearing 7 when the revenues that are at stake are worth hundreds of millions of dollars to consumers 8 Time also do es not seem to be a major motivating reason for settlements. Littlechi ld (2009a, b) state that discussions of settlement between parties begin after a rate case is initiated at the FPSC and after testimonies are filed and counter filed. A settlement is reached frequently just before the hearing is meant to take place In the case of electric company rate cases, the expected duration of a hearing woul d be a week or two, and the FPSC would give its ruling shortly afterwards 6 See footnote 11 of Littlechild (2009b). Wang (2004) also states that the FERC also often claims that rate certainty and the legal costs saved are the advantages of the settlement process. 7 According to Littlechild (2009b), it costs OPC around $100,000 to bring a case in front of the r egulator. 8 Wang (2004) also states that the legal costs saved from avoiding a rate hearing at the FERC are too small a factor for a gas pipeline company and its customers to settle.
16 (Littlechild 2009b). This implies according to Littlechild (2009b), that the time saving from a settlement would be small and the uncertainty would also be resolved in a matter of w eeks. This is partly corroborated by the second chapter of this dissertation in which the payoff functions of the consumer advocate and the firm during rate cases at the FPSC are estimated using the same dataset as Littlechild (2009a, b). The study suggest s that saving the time of a form al hearing is important in nor I n his study of rate cases of the major interstate gas pipeline companies at th e FERC Wang (2004) agrees that uncertainty about the regula tory decision is not a likely rea son that parties settle rate cases. Wang (2004) suggests that the regulatory decision is more predictable than a trial outcome in a court of law He further observes that in the settled rate cases of gas pipeline companies at the FERC, the regulatory decision had the case gone for hearing -. 9 Because the most important motivating factors that cause parties to settle during litigation are not the same as in rate cases, this paper attempts to develop a model which is better suited to analyze such agreements than models which analyze settlements i n litigation literature The model is based on the predominant reason why public utilities firms, their consumer s and other interested parties settle a rate case. Wang (2004) Doucet and Littlechild (2006, 2009) and Littlechi ld (2009a, b) all agree 9 e whether the contesting parties had a valid reason for contesting, the FERC had to consider the outcome of the rate case if they had not been settled and report it. Because the alternative regulation outcome is available for these contested settlements, W
17 that the chief reason why parties in a rate case prefer settlement is that settlements achieve more desirable outcomes for all involved than the fo rmal hearing process Why are the parties able to achieve more desirable outcomes in settlement than by regulation? The se outcomes are shown to be possible in Wang (2004) because the settlement process is more flexible than the rigid regulation process. If there is more than one issue to be determined by the regulator, a formal hearing means the regulator decides these issues separately by issue merits not legally bound to adjudicate in this manner. By the Supreme Court ruling in Federal Power Commission v. Hope Natural Gas Company the FERC can make a valid regulatory decision even after making a material error in deciding one of the issues as long as FERC try not to make any material error in any of their rulings. Wang (2004) says there ar e two reasons for this The first reason, as n oted by Krieger (1995), is that many courts believe the final decision is necessarily invalid if there is a material error in any issue. Wang (2004) gives the example of the Public Service Commission of Wisconsin which tried to give a ruling without making findings of fact on rate base or rate of return. decision was found to be arbitrary and unlawful by the Wisconsin Supreme Court. 10 The second reason as stated by Noll (1971) is that sometimes the only threat of the 10 Commonwealth Telephone Company v. Public Service Commission 252 Wis 481 (1948).
18 The regulator is more suscept ible to appeals if it makes a material error in determining any issue. The settlement process on the other hand allows a trade off between issues and allows the parties to come to a better outcome for all by allowing the parties to focus on Because the settlement agreement is formulated by the main contesting parties themselves, the chance of appeal against a regulatory order approving it is minimized. The technique is further enhanced by the fact that the parties obviously know their preferences better than the regulator. The parties can agr ee to terms in whose favour the regulator may not rule bec ause it is unsure whether the parties will prefer such an outcome. For example, Littlechild (2009b) reports that some FPSC staff members wou ld recommend a settlement even when they estimated that the regulated rate reduction would be larger because consumers valued immediate gains. 11 view seems to be supported by the experience of Littlechild and Doucet (2009) at the NEB Littlech ild and Doucet (2009) study negotiated settlements of toll cases of major pipeline companies at the NEB T he Board would analyze each just as it would do in a formal he aring. It would take out any item of settlement which it only three fully negotia ted settlements. The NEB changed its approach in 1994 and announced that it would either accept or reject a settlement in its entirety. The Board 11 This is from the settlement signed by the OPC and the Florida Power and Light Company (FP L) in 1999.
19 now determined the reasonableness of a settlement by t he reasonableness of the process rather than by t he reasonableness of the outcome After this change of policy, there was a significant rise in the number of negotiated settlements at the NEB. In fact settlement from 1995 to 2010 This again shows that the main difference between the settlement process and regulation as perceived by the pipeline co mpanies and their customers was that settlement allowed parties to decide issues (tolls and base rates being among them) as a total package Wang (2004) constructs a two stage model to show how the settlement process yields a better outcome for a firm and its consumer by allowing a trade off between issues. In the first stage, the firm and its consumer bargain simultaneously over the shares of two pies of fixed size 1 If they come to an agreement the game ends with settlement At any period of this bargaining, the respondent of an offer may opt for litigation in which case the game moves on to the second stage. In this stage, the regulator decides the shares of both the fixed pies independently according to the Nash bargaining solution of the firm and consumer. Wang (2004) shows the settlement out come Pareto dominates the litigation outcome. But for a deeper understanding of t he settlement process and to determine whether consumers may be hurt from it a more detailed look at what issues are decided and an understanding of how interests of parties are represented during negotiations is needed Also Wang (2004 not explain why negotiations sometimes break down and the case goes to regulation 12 12 he calls it) though this is not stated explicitly by the author. Because the firm and consumer know what
20 The only danger to parties in settlements is that their interests may be misrepresented at the bargaining table. Therefore it is important to identify the parties and their ag ents, and to investigate whether there is a possibility that an agent may mis The main purpose of the next chapter is to see whether the final consumer may be harmed by settlement and if so, in w hat circumstances this will happen. So the chapter focuses on the role of the consumer advocate as the representative of the final consumer in the negotiations with and regulatory proceedings against a public utility firm 13 14 To examine if there is a possibility that the consum er advocate may misrepresent how the particular parties involved (the consumer, the firm, the consumer advocate and the regulator) benefit from settlement over regulation. Littlechild (2009a, b) identifies the benefi ts the parties receive in negotiated settlements regarding base rates at the FPSC. These settlements are signed by the public utilities firm whose base rates are in question, the OPC and other interested parties. they will get in the second stag e of litigation, they can come to the same outcome in the first settlement stage itself if there is no Pareto improving alternative. 13 The consumer advocate and the firm are not the only parties involved during negotiations and formal hearings. Other interested parties include large consumers, firm competitors and a wide variety of other interests. The paper, however, focuses on the role played by the consumer advocate and the firm during negotiations and regulation. Littlechild (2009b) reports that according to transcripts of the hearings the y 14 It must also be noted that there is another possibility of a principal agent analysis with the firm management representing the interests of the firm shareholders. This is briefly indicated in Littlechild (2009a) footnote 25.
21 The main benefit that consumers get from these stipulations at the FPSC is the lower rates they have to pay. Littlechild (2009b) examines how these rate reductions differ from the rate change that would have occurred if the case was not settled. Littlechil owned electric companies the rate reductions, refunds and revenue sharing from settlements w ere greater or earlier than those which would have been achieved by regulation. He estimates that t hree quarters of the benefits would not have been achieved without settlement. In some rate cases, even when the FPSC staff estimated that the rate reduction from the regulatory ruling may be higher than the stipulated rate reduction the settlement was sti ll approved by the FPSC on the basis that it would provide 15 The staff said that this would be more preferable to the consumer. Littlechild (2009b) also argues -unlike Wang (2004) -that the certainty of the settled rate reduction was another benefit to consumers. Settlements with rate reductions were concluded even at those times when the FPSC had a known policy of applying o ver earnings of electric co mpanies to write off assets rather than for consumer r ate reductions. After stating what benefits consumers get from negotiated settl ement, Littlechild (2009b) examines the reason s why firms agree to settlements that result in lower rates and increased refunds relative to those required by FPSC regulation. H e makes a detailed study of the negotiated settlements signed by the OPC and four of the five 15 This is from Lit
22 investor owned electric companies in Florida 16 Littlechild (2009b) finds that there are two main ways that stipulations create benefits for utilities. One, by settlement the OPC can commit to act in certain ways that fall outside the FPSC Littlechild (2009b) calls these commitments on conduct. Two, the OPC is w illing to make concessions that the FPSC is unwilling to make Two such concessions identified in these settlements are giving more discretion to the utility in its accountin g policy and revenue sharing The revenue sharing arrangements found in the settlements indicate a move from the ROR regulation towards incentive regulation. The que stion of how these benefits specifically translate into gains for the firms is important because it may be possible that these gains come at the expense of the final consumer. To explore this Littlechild (2009b) are examined in greater detail b elow. In the commitments to conduct category, Littlechild (2009b) talks about stipulated agreements where the OPC promises an action in return for a rate reduction or refund. This action may be the withdrawal of opposi tion in another forum such as oppositi on to a firm merger. More important, the OPC and the firm both agree not to initiate a rate case for a substantial p eriod of time. Wang (2004) refers to this as rate moratorium Wang (2004) admits that it is hard to estimate what the firm specifically ga ins here He says that the benefits for the firm depend on demand and cost uncertainty inflation, interest rates and the effect the settlement has on firm incentives and subsequent behaviour. What he means is that in the period that neither party can file a rate case, 16 The four companies were Florida Power and Light Company (FPL), Florida Power Corporation (FPC) which is now Progress Energy Florida (PEF), Tampa Electric Company (TECO) and Gulf Power Company (GPC). The fifth investor owned electric company, Florida Public Utilities Company (FPUC), did not participate in any settlement during the period of study.
23 the benefit to the firm lies in a portion of operating revenue that it retains This is the portion that may have been considered overe arnings by the regulator and it s amount is depend ent on the uncertain factors he lists 17 There is also another way the firm can profit which neither Littlechild (2009b) nor Wang (2004) explore s if the OPC agrees not to initiate a case for a given period Apart from operating revenue, the firm has income from other sources such as renting of its poles, sale of appliances and the renting of its properties. During regulation, the regulator sometimes has the discretion to subtract this other income component from the revenue requirement of a firm when calcula ting rates. By agreeing not to initiate a case the OPC allows the firm to keep all of its revenue from all its other operations for the durati on of the settlement period. T he agreement also motivates the firm t o increase these operations which it may not h ave done otherwise. The other benefits to firms that Littlechild (2009b) discusses are inter related. These consist of allowing the firm more flexibility in its accounting policy and revenue sharing Both these concessions by the OPC to the firms a re depa rtures from FPSC policy and the FPSC sta ff expressed their concerns about them Littlechild (2009b) uses the FPSC staff to examine how these settlement outcomes might have differ ed from the rulings. One of the stipulated items against which the staff protested was the greater discretion given to the firm in its depreciation and amortization polic y than the FPSC would have allowed. The advantage to the firm is that depreciation is a non cash expenditure and can be manipulated. Th e staff longer reflect the matching principle [matching capital recovery with consumption over 17 Wang (2004) says this is effectively price cap regulation.
24 (Littlechild 2009b) (Littlechild 2009b ) There is another use of a flexible depreciation policy for the firm It must be looked at in conjunction with the other new innovation found in settlements incentive regulation. From 1986 to 1999, earnings share arrangements were made in several negotiated settlements This incentive regulation would supersede the usual ROR regulation for the duration of the settlement period. For example, in 1996 FPSC approved a settlement between the OPC the Florida Industrial Power Users Group ( FIPUG ) and TECO (Order No. PSC 96 0670 S EI) In the settlement, TECO agreed to an earnings sharing plan whereby 60% of the net revenue that contributed to a ROE over a specified level would be deferred to the 1996 and 1997 plus any net revenue that contributed to a net ROE over 12.75% would be refun 18 This kind of arrangement along with the more flexible accounting policy enables the firm to increase depreciation (con sequently decreasing its ROE) in future years when its revenue is high Thus it can recover that portion whic h it would have had to give up to its consumers had conservative accounting policy been followed. The OPC might have caught on to this because Littlechild (2009b) reports that it wanted subject to manipulation 19 A 18 This plan was extended to 1999 by Order No. PSC 96 1300 S EI. 19 adjusting the level of equity maintained at the utility level, that the Company could affect the amount of 98 0802 FOF EI).
25 Revenue Sharing Incentive Plan was specified in four settlements with FPL and PEF made from 1999 to 2005 This plan consists of base revenue thresholds instead of ROE levels within which one third of the revenue would go to the firm and two third s to the consumers. Any revenue in excess of the revenue cap accrues entirely to the consumer. The firms however did have an authorized ROE range f or all regulatory (Littlechild 2009b) but it was not to be used for the revenue sharing mechanism. The staff was concerned that the ROE was now meaningless for surveillance purposes and estimated that the firms would earn above the upper limit of the ROE range. They also estimated that these earnings wou ld continue to grow and noted there was no earnings cap (Littlechild 2009b) This also suggests that with revenue sharing in settlements, the firm is allowed to retain a part of its future revenue which it could not have done if the case went for regulatio n. Another advantage that revenue sharing gives the firm is that prospective investments which otherwise would have needed FPSC approval are included in the rate base. Before a new investment by the utility, the FPSC must find that the 2009b) before including it in removed this function of the Commission. In one of them 20 an investment was included in the rate ba se with out a formal hearing of the issue. The final capital cost was included in the rate base and the entire operating expense of the new investment in the net operating income of the firm In the incentive regulation embodied in the settlements, 20 The negotiated settlement TECO signed with the OPC in 1996. The investment was the addition of a tion.
26 the regu regulation regime is only valid over the period specified in the settlement. If in the future a rate case went for regulation then the revenue requirements for the utility w ould be determined with the new investments now included in the base rate. Littlechild (2009b) mentions a few other firm benefits in settlements. As part of their more flexible accounting policy, firms were allowed to petition for storm cost recovery in c ase their storm damage reserve proved insufficient in the future instead of increasing the annual accrual for the reserve. Littlechild (2009b) also briefly mentions that some of the settlements allowed new costs of the firm to pass on to the consumers through fuel adjustment and other such clauses instead of base rates. The regulator may not have required consumers to pay these costs Fu el costs account ed for 13 to 38 percent of the total operating revenue of FPL, PEF and TECO in 2009 and around 46 21 A closer examination is required to see how much of the operating cost of firms is passed on by t hese clauses 22 to the consumer through settlements. The next chapter models the stipulated benefits as future cost savings of the firm. The firm expects that some of its revenue will be refunded to its consumers by the regulator in the future. The fi rm see s this as a future cost of its operations in a ROR regulatory regime If the OPC agrees not to file a rate case in a stipulation or agrees to a revenue sharing plan (whereby the firm can keep revenue over its authorized ROE) 21 Source is company annual reports. 22 Perhaps the reason why Littlechild (2009b) did not examine such clauses in more detail is because the impact of fuel adjustment and other clauses was not included in the FPSC database (Footnote 8 Littlechild, 200 9a).
27 then the firm can retain this expected future overearning for the duration of the settlement period. new investment in the rate base. This is another expected future cost of the firm which is saved by settlement when the OPC agrees to its inclusion in the rate base. 23 New costs of the firm which are passed on to the consumer automatically via fuel adjustment and other such clauses in settlements are also f uture cost savings It is the job of the consumer advocate to represent the consumer during negotiations and formal hearings. In Florida, it is the statutory duty of the OPC to represent the citizens of Florida. Therefore, the consumer advocate should only sign agreements which are beneficial to the consumer. Littlechild (2009a) states that advocate not only wants lower rates for consumers but he also wants to be seen achieving these lower rates. As seen from the above analysis, it is difficult to identify the exact gain the firm makes from the settlement. The identification problem will be greater for an average consumer who may lack the sophistication to fully interpret the concessions made to the firm. The most easily understood item of the settlement is the present rate reduction or incr ease. Littlechild (2009a) says a rate reduction represents 24 and is an observable signal that the advocate is doing a good job. The advoc ate may be tempted to give up more future income than 23 The stipulated firm benefits can equivalently be modeled as receiving future revenue that would have accrued to the consumers in a ROR regulatory regime. 24 Littlechild (2009a) also talks about how a stipulated rate reduction is also good news for t shareholders. A stipulated rate reduction as opposed to a regulated rate reduction signals to the shareholders that the rate reduction is manageable and the relationship with the regulator is good. The following analysis does not study this princ ipal agent relationship.
28 the consumer would otherwise want in exchange for a present rate reduction if the future loss is un observable to the consumer at the time of the settlement 25 The present show of good job performance lea ds to be tter chances of re appointment and longer term as advocate This can happen especially if the firm collects the cos t of the present rate reduction at a future date beyond office or if his tenure is uncertain 26 Here lies the divergence between the preferences and their agent represented at the negotiation table. The second chapter of this dissertation finds that in the rate cases a t the FPSC, the consumer advocate seemed to give more weight to the size of the present rate reduction than on future consumer welfare in his decision to settle. This suggests that there is the possible existence of negotiated settlements that the consumer would oppose. The FPSC must make its regulatory decision by balancing the needs of the utility and its shareholders with the needs of the consumers. But the regulator would also want to minimize the criticism of its regulatory decision from all parties in volved and insulate itself from further legal appeal against its decision 27 As mentioned previously, the settlement process can p rovide a means of achieving this objective. That is why, says Littlechild (2009a) ry settlement 25 Littlechild (2009a) mentions there may be a difference in interests of the parties with respect to calculating of the value of a settlement. The consumer advocate may want to maximize and make public the undiscounted nominal doll ar value of the rate reduction whereas the firm will be more interested in the real present discounted value. 26 The Public Counsel of Florida is appointed and re appointed every year. Littlechild (2009b) states that negotiated settlements in the Florida electricity sector typically determine rates for three to four years ahead. Some implications of the clauses of these settlements last even longer. 27 See Joskow (1974).
29 given to it. From the point of view, a negotiated settlement limits criticism from the parties involved and the chance s of further legal appeal The next chapter models two types of rate cases. The first is an earnings review case. A n earnings review case is initiated by the regulator, sometimes at the request of the consumer advocate, in the belief that the firm has earned above its revenue requirement and a rate reduction might be in order. The second type is a company request case. T his is a case initiated by the firm when it believe s a rate increase is necessary. In the earnings review case, it is assumed that the firm has earned more than its approved revenue and a rate case has already been initiated The model has two stage s The first stage is the Settlement Stage. Here the consumer advocate and the firm bargain simultaneously over the share of the overearnings and the share of the constant f uture cost that the consumer and the firm are to bear This simplified model assumes that the consumer advocate, the firm and the regulator all know the amount of overearnings and the total future cost of the firm 28 This total future cost e cost of ROR regulation discussed above. These are the t wo p ies whose shares are being negotiated The negotiations follow a standard infinite horizon alternating offers bargaining procedure (Osborne and Rubinstein 1990) If an offer during this stage is accepted then the game ends in a negotiated settl ement. B ut if the ad vocate and the firm disagree and cannot come to an agreement then the game eventually moves to the second stage of regulation. In the Regulation Stage which represents the formal hearing, the regulator decides the shares of the overearnings and the game ends. The company request case is modelled in the 28 As will be seen it is not required that the regulator know the total future cost of the firm.
30 revenue and it initiates a case to recover the shortfall. In the Settlement Stage, the cons umer advocate and firm bargain over how much of the applied for shortfall the firm s future cost 29 T he regulator determines how much of the shortfall the firm is allowed to recover in the Regulation S tage 29 Again in this simplified model it is assumed that the shortfall that the firm applies to the regulator to get back and the total future cost of the firm is known to the consumer advocat e, the firm and the regulator. In reality, in a company request case the firm does request a rate increase amount when filing a case so this is known to the parties.
31 CHAPTER 2 NEGOTIATING WITH A C ONSUMER ADVOCATE IN PUBLIC UTILITIES REGULATION The analysis of this chapte r proceeds as follows. Section 2.1 descri bes the formal model. Section 2.2 reports the results. Sect ion 2.3 develops the policy implications of the formal analysis. Section 2.4 concludes with suggestions for future research. 2.1 The M ode l A rate case is initiated when the firm earns more than its approved revenue (an earnings review case) or less than its approved revenue (a company request case). The prof it function of the firm before the game begins is: ( 2 1) In Equation 2 1 R is the revenue of the firm, is the operating and maintenance cost of the firm, is the future cost of the firm and costs. As discussed in the last chapter may be the amount of expected future revenue the firm will have to give up to the consumer expected future cost of a new investment or a new future cost of the firm which it expects to incur under ROR regulation. The revenue of the firm can be split into two components depending on whether the firm has earned over or under its approved revenue. If the firm has earned more than its approved revenue, then is: (2 2 ) where is the approved revenue as determined in the last rate case and is the amount of revenue that the firm earns in excess of If the revenue is below then is: (2 3)
32 where is the amount of revenue by which falls short of It is assumed that the regulator sets such that: (2 A1) Assumption 2 A1 reflects the fact that the regulator will not allow a firm revenue to be below its operating and maintenance cost. Such a revenue level would cause the firm financial distress 1 The regulator does not set costs, The reason for this i s the nature of considered here. It is either a future cost that the regulator cannot presently observe 2 or a future cost that the regulator determines the firm should solely bear 3 The regulator initiates an earnings review case if the firm earns over In such a is given by Equation 2 2 Alternatively, if the firm earns below the firm files a company request case with the regulator. Here, is expressed by Equation 2 3 T he first stage of the model which is the S ettlement Stage begins after the rate case is initiated 4 In the Settlement Stage of the earnings review case, the consumer advocate and the firm simultaneously bargain over the share of each party is to receive and the 1 If in the last rate case was set by a negotiated settlement instead, then it is ass umed that the firm would not agree to a revenue below 2 It is probable that is only observable to the firm. But during the Settlement Stage, the firm will make known to the consumer advocate as one of the amounts to be bargained in the Settlement Stage. Consequently, will be known to the consumer advocate at the beginning of the Settlement Stage. 3 The firm can always initiate a rate case when in future is inc urred. If causes the firm financial distress then becomes part of at the time of the future rate case. 4 Littlechild (2009a) reports that negotiations between parties begin after the rate case has been initiated and testimonies filed. Settlemen ts are usually concluded just days before the formal hearing. Wang
33 share of each party is to pay. is the part of that goes to the consumer and is the share of that the consum er has to pay. This implies that and are the corresponding shares of the firm. Therefore, and in an earnings review case. The only difference in the company request case is that the parties bargain over the shortfall instead of Consequently, and are the shares of that go to the consumer and the advocate respectively and in a co mpany request case. review case and the company request case. These are shown by equations 2 4a and 2 4b respectively: ( 2 4a) ( 2 4b) In Equations 2 4a and 2 4 b is the discount factor for each period in the Settlement Stage should not be confused with The firm uses to discount its profit in each successive period of bargaining within the Settlement Stage. In contrast, Settlement Stage. Therefore, it must be that tility in both the ea rnings review case and the company request is: ( 2 5) In Equation 2 5 5 The relation between and is assumed to be: 5 and are not the same for the same reason and are not related. Also
34 ( 2 A2) Assumption 2 A2 assumes that firms are at least as patient as the consumer 6 It is assumed that the consumer advocate is only concerned about the share of in an earnings review case or in a company request case that he can obtain for the consumer. This is because the value of s job performance. Th e consumer will perceive a higher value of as a better job done by the advocate and this will increase his chanc es for re appointment. 7 Therefore the utility function of the consumer advocate is: ( 2 6) During negotiations with the firm and the regulatory formal hearing, it is assumed that the advocate maximizes a weighted functi on of his own utility function and the s: 8 ( 2 7) where is the weight the advocate assigns to his own utility function. Equation 2 7 can be simplified to Equation 2 8 : ( 2 8) 6 A natural choice of would be where is the interest rate. 7 employment chances affecting his bargaining behaviour is not new. Cai (2000) develops a two phase bargaining model in which there is an election after the first phase. This model e xplains the unnecessary extension of strikes by labour union leaders as a result of their re election concerns. Cai (2000) shows that when the principal does not know the total size of the surplus over which the bargaining occurs, a tough agent may want to intentionally delay the bargaining to signal 8 This weighted payoff function is like the one used by Charness and Rabin (2002).
35 function. over time, for example 9 The consumer, however, has no knowledge of the consumer The assumption is that the typical ratepayer of a utility is not aware of the details of a rate case. The consumer can only observe and not after the game ends. is a cost that the consumer must pay in future and therefore, he cannot observe it at present. T he bargaining follows an infinite horizon standard alternating offers procedure. The consumer advocate proposes a division of the pots in even periods .The firm makes corresponding proposals in odd periods 10 The player has t wo options when responding to an offer: (1) accept the offer in which case the game ends in a negotiated settlement or (2) reject the offer and make a counteroffer in the next period of bargai ning Option (1) implies that the regulator approves any settlement that it is presented. 11 The disagreement vectors of the bargaining game in the first stage are the expected payoffs the parties will receive in the Regulation Stage. Because the rate case has already been initiated, if there is perpetual disagreement between the 9 he concluded settlements and was involved in rate cases repeatedly with the same major Florida public utility firms. 10 In the limit it does not matter to the outcome which player makes the first offer. 11 the years 1976 to 2002, only one settlement was overturned. This settlement was approved in 1989 after one item that the FPSC opposed was removed from the agreement.
36 advocate and firm then the Settlement Stage would eventually end and the Regulation Stage would begin The result would be the regulation payoffs for all parties 12 In the Regulation Stage, the regulator must decide the partition of between the parties in the earnings review case and the partition of in the company request case. The regulator must make this division without putting the firm in financial distress. This means the regulator must choose regulation must not be below The regulator does not choose According to the two cat egories of stipulated firm benefits of Littlechild (2009b), either the regulator cannot set or it has already set at 0 by its policy in a previous rate case. In both instances remains equal to 0 as it did before the rate case was filed. The regula tor chooses by utility function 13 regulation revenue is not less than In the earnings imization problem is: ( 2 P1) And in the company request case, it is: ( 2 P2) 12 This is consistent with Coot er and Rubinfeld (1989) who say that the disagreement vector or threat point (as they call it) in the settlement stage will be the expected trial verdict. 13
37 In the constrained maximization of Problems ( 2 P1) and ( 2 P2), is the weight that the regulator attaches to the consumer utility function. The regulator sets ts shareholders. Uncertainty of the regulatory decision is introduced in the model by assuming that the consumer advocate and firm do not know the value of They do, however, have beliefs about y density functions and for the advocate and firm respectively. The corresponding probability distribution functions are and 2.2 Results This section is divided into two parts. In subsection 2.2.1 t he model is solv ed for the e arnings review case and then it is solved for the co mp any request case in subsection 2.2.2 2.2.1 The Earnings review C ase In the Regulation Stage, the regulator has to choose to solv e the constrained maximization Problem 2 P1 Because in an earnings review case and Assumption 2 A1 implies that the constraint in Problem 2 P1 will hold trivially. ( 2 P1 a ) Differentiating with respect to provides: ( 2 9) Equation 2 9 reveals that if and if Therefore, the solution of Problem 2 P1a is: if ( 2 10a)
38 if ( 2 10b) Equations 2 10a 14 and 2 10b show that the regulator will award the entire to the party it weights most during regulation. T he refore, the regulation profit is: if ( 2 11a) if ( 2 11b) In the Regulation Stages of both the earnings review case and company request case, Because during regulation, the advocate strives to acquire as much of as he can. This is also best for the consumer when regulation weighted utility function, regulation util ity, ) is: if ( 2 12a) if ( 2 12b) From Equations 2 11a, 2 11b, 2 12a and 2 12b we can arrive at the negotiating The firm expected after regulation profit, and the payoff it expects the advocate will get from regulation, are given by Equations 2 13 and 2 14 respectively: ( 2 13) ( 2 14) 14 It is assumed that the regulator allows the status quo before the rate case to stand (i.e. the firm retaining all of its overearnings) when (or equivalently, when ).
39 regulation weighted utility, and the after regulation profit the consumer believes the firm will get, are given by Equations 2 15 and 2 16 respectively: ( 2 15) ( 2 16) The se expected payoffs serve as the disagreement vectors in the Settlement Stage. In the Settlement Stage, let represent the equilibrium offer of the consumer advocate and let be the equilibrium offer of the firm. The necessary condition for a subgame perfect equilibrium (SPE) is that each party is indifferent between accepting and not acc ondition 2 17 is : ( 2 17) We solve for in Equilibrium Condition 2 17 without considering the expected regulation payoffs (the expected disagreement vectors) The result is then compared with the expected regulation payoffs to see whether any one of the parties prefers regulation instead. higher than its settlement payoff and both parties believe each has a better chance of the reg ulatory decision being in its favour (i.e. ) then there is disagreement. This leads to the Regulation Stage where the regulator decides the rate case.
40 In equilibrium, it must be the case that if then If this were not the case, then the advocate could increase and reduce This would make the advocate better off w ithout reducing the firm profit. Similarly, if then In short, all possible tradeoffs between the two parties must have occurred in equilibrium. The result is that the possible eq uilibrium offers fall into four categories : 15 (i) and (ii) and (iii) and (iv) and The category of the equilibrium offer will depend on the pair It can be shown that : 16 (a) the equilibrium offers fall into category Candidate Solution iv when ; (b) the y fall in category Candidate Solution iii when ; and (c) they fall i nto category Candidate Solution i when If there is a SPE in the Settlement Stage, it occurs in th e first period of bargaining itself. This means in period Because the if ( 2 18a) 15 The proof of this is provided in Appendix A. 16 The proof is provided in Appendix B.
41 if ( 2 18b ) if ( 2 18c) In the limit as Expressions 2 18a, 2 18b and 2 18c become: 17 if ( 2 19a) if ( 2 19b) if ( 2 19c) The resulting payoffs of the advocate, firm and consumer from the equilibrium offers in Expressions 2 19a, 2 19b and 2 19c are: if ( 2 20a) if ( 2 20b) if ( 2 20c) 17 Because and This implies
42 Because utility function, term in the ranges of Expressions 2 19 and 2 20 is the present value discounted future cost of th e firm as perceived by the advocate The value of this expression is the weighted utility level if he receives all of but the consumer must pay all of in exchange. It serves as a relative measure of the sizes of the two pots We can make some observations about the SPE outcomes in the Settlement Stage. The trade off during bargaining is driven by the different valuation of by the advocate and the firm. As previously discussed in all possible SPEs either all of or all of (or both) is transferred to the consumer. For low values of relative to will be small. In this case, all of but not all of is transferred to the consumer The firm would like to transfer the remaining units of t o the consumer. But because there is relatively less of the firm has run out of units of to use as payment for the transfer of the se remaining units of This is why in the range where is lowest ( ) the solution is in category Candidate Solution iv and We can call the range the low e range Similarly, when is high relative to the SPE involves the transfer of the entire amount but not all of Consequently, the solution falls in category Candidate Solution i and in the range where is highest ( ) We can name this range the high e range
43 Between the ranges where takes its lowest and highest values, there is an intermediate range: (we can call this range the medium e range ) The SPE solution in this range involves the total transfer of and to the consumer. It is possible in the medium e range therefore to have multiple equilibria where is the same but the value of is different (or equivalently, multiple equlibria with the same but different values). The existence of this range is another outcome of the fact that the advocate and the firm value differently. During bargaining, exactly units of are required to exhaust all the units of 18 If another unit of is added then to transfer this unit to the consumer, the firm needs to transfer an additional units as well to the consumer. 19 But w hat if with the incursion of another unit of the firm has less than units of available for exchange? When this is the case, it is still in the best interest of the advocate and the firm (without considering the Regulatio n Stage) to transfer all of and to the consumer. 20 These are the SPE outcomes represented 18 This can be seen from the fact that to find the relevant range where the solution falls into category Candidate Solution iv and ( and ), the answer is We can then re arrange the upper limit of this range to get 19 The range where solutions fall into category Candidate Solution i and is Re arranging this range we get Therefore, if the solution falls into category Candidate Solution iv a nd all the units of and are exactly transferred to the consumer and then an additional unit of is incurred, then the firm needs ( ) units of to transfer this unit of as well. 20 This will be true as long as for the consumer advocate and for the firm.
4 4 in the medium e range It should be noted that if the parties valued the same (i.e. ) then no such range would exist. The only SPE solutions of the Settlemen t Stage would fall into categories Candidate Solutions i or iv This is because the one unit of would be transferred to the consumer for one unit of 220.127.116.11 The full information case with a perfect consumer advocate A special case is examined before considering more general results. In this full before the game begins Also, the consumer advocate in this case is a perfect repr esentative of the consumer, s o When there is full information the expect ed disagreement vectors in the Equilibrium Condition 2 17 become the regulation payoffs both parties know they are getting if the game moves to the Regulation Stage. In such a setting, the game would nev er g o to regulation. If there were no mutually beneficial outcome in the Settlement Stage, the parties would agree to the regulation outcomes in the Settlement Stage itself. The game would always end in a negotiated settlement. Because no agreement wou ld be to the detriment of the consumer. This leads to our first proposition Proposition 1: In the full accurately represented in the Settlement Stage (i.e. ), negotiated settlements will always result in Pareto improving payoffs for all parties relative to ROR regulation in rate cases. Proposition 1 is the outcome of the fact that any outcome in the Regulation Stage can be replicated in the Settlement Stage w ith the added possibility of more mutually desirable outcomes for all parties.
45 18.104.22.168 B argaining results in the Settlement Stage Two propositions concerning the bargaining procedure in the Settlement Stage are now presented. The implications of these pro positions hold beyond the context of negotiated settlements in public utilities regulation. The propositions are applicable for any situation of delegated bargaining where parties simultaneously bargain over two p ies Proposition 2: In the Settlement Stag e of the bargaining process the consumer may not be harmed if the preferences of the consumer advocate differ from its own (i.e. if The outcome reflected in Proposition 2 arises in the present model when are such that falls in the medium e range. W hen in Equation 2 7 preferences coincide with those of W hen the advocate bargains seeking to maximize only his own utility function. As increases, the advocate puts more weight on his utility function and less on increases we and preferences diverge Given certain relative sizes of and the magnitudes of the other parameters in the model 21 may remain entirely in the medium e range as increases from 0 to 1 This will be the case when In this situation even as the advocate places greater weight on his own utility (i.e. increases) the equilibrium partitions of the p ie s remain the same ( and ). The consumer is un harmed 21 The other parameters being and
46 because he gets the same and pays the same during bargaining regardless the value of The medium e range represents the value s of for which it is optimal for the advocate to receive all of in exchange for the payment of all of and it is optimal for the firm to propose this offer. This situation can also occur if the parameters are such that as increases from 0 to 1, moves from the low e range to the medium e range The condition for this is In such a case, initially as increases, the consumer is worse off because the advocate agrees to higher shares of that the consumer would have to pay. But after rises to equal and moves into the medium e range any further increase in will not hurt the consumer. Similarly, if initia lly an increasing will not harm the consumer as will remain in the medium e range. But as soon as becomes greater than (i.e. when falls into the high e range) an increase in will be to th e detriment of the consumer becau se the advocate will start to accept a higher share of for the same level. Proposition 2 is valid if the medium e range is not vacuous. This range characterizes equilibrium settlement solutions where there may be multiple values of for the same or vice versa. It is easy to think about settlements when are such that falls in the low e or high e ranges It i s natural to expect that the firm only agrees to settle if for a higher share of the consumer advocate is willing to pay a higher share of Or alternatively, the consumer advocate agrees to bear a larger burden of in exchange for a larger share of The question is are there
47 negotiated settlements that are characterized by the model when falls in the medium e range Such stipulations have been concluded at the FPSC. In May 1996, the FPSC approved a settlement (Order No. PSC 96 0670 S EI) signed by TECO, the OPC and the FIPUG in a TECO earnings review case. The settlement 31, 1998. In return, the OPC and FIPUG agreed not to seek a rate reduct ion until December 31, 1998, an earnings sharing plan for three years (1996, 1997 and 1998) 96 0670 S EI) regarding the regulatory treatment of the Polk Power Station a nd Port got from the earnings sharing plan was $11, 226, 598 (Order No. PSC 99 2007 PAA EI). In October 1996, the FPSC approved another settlement between TECO, the OPC and t he FIPUG (Order No. PSC 96 1300 S EI). This settlement provided an additional December 31, 1999. In return, the OPC and the FIPUG agreed to an earnings share plan for 1999 (t Order No. PSC 01 2515 FOP EI), allowed the actual final capital cost 22 of the Polk Unit to be included in the rate base for all regulatory purposes, and also allowed TECO t o include the fu ll operating expense of the Polk Unit in the calculation of its net operating income for all regulatory purposes 23 24 Comparing the two settlements that TECO 22 This came to an amount equal to one percent above the capital estimate of $506,165,000 plus related estimated working capital of $13,029,000. 23 This was estimated to be $20, 582, 000 net of Department of Energy (DOE) funding for the first year.
48 signed in 1996 25 it seems that the OPC agreed to burden the consumer with more of costs in the second settlement for the same share of overearnings ($25 million). If we include the refunds that the consumer received from the earnings sharing plan then it seems that the Public Counsel obligated costs for l settlement equilibrium when fall in the medium e range Expressions 2 19a, 2 19b and 2 19c imply with a higher ) will end up with a share of that is equal to or even less than a less (i.e. an advocate with a lower ). In the low e range the advocate will get the same share of (which is the entire amount ) for every value of Once in the high e range will actually decrease as increases ( as is evident from Expression 2 19c ) This is the case because the advocate must incur units of for every unit of to remain at the same utility level. If increases then the advocate will accept more units of for every unit of As a result the firm can make the advocate accept an offer with a lower share of for the same level of when increases. Ironically, the advocate can secure a hig her share of in the SPE of the Settlement Stage when he is less interested in doing so. This additional illustration stated in Proposition 3. 24 Additionally, OPC and the FIPUG did not seek a earni ngs review case during the duration of the settlement. This was probably one of the terms of the settlement like the May 1996 one. I could not check it as the settlement document was not available online. 25 Both settlements were agreed during the tenure of the same Public Counsel and in the same year. We can therefore assume that remained unchanged for both settlements.
49 Proposition 3: In the Settlement Stage, the share of the consumer advocate secures for the consumer in creases when he weights higher than his own (i.e. when the value of is lower than ). 22.214.171.124 The imperfect information case Whether t he parties receive the payoffs in Expre ssions 2 20a, 2 20b and 2 20c cases are settled and whether there is a possibility that some cases may be settled against consumer interests. The advocate and the firm wil l agree to a settlement only if it is mutually beneficial. An additional assumption mad settlement payoff, then the party prefers to settle. This means the advocate will want settlement if and similarly, the advocate will want it when We first look at what happens in the high e range. The results are equally applicable to the low e and medium e ranges The advocate wants to settle when and the firm wants to settle when In the range t hese two inequalities imply that the advocate and the firm want settlement when their beliefs and are such that: ( 2 21) ( 2 22)
50 Inequalities 2 21 and 2 22 imply that for a settlement equilibrium with the payoffs shown in Expressions 2 20a, 2 20b and 2 20c to take place, it must be that 26 The condition means that each party would want to settle when they believe that the regulatory decision will be more favourable to the other 27 It is important to note that rates may be set by negotiated settle ment even if one of the Inequalities 2 21 and 2 22 do not hold. However, t he partitions of and wi ll not be those shown in Expressions 2 19a, 2 19b and 2 19c For instance, suppose that Inequality 2 21 is violated but Inequality 2 22 hold s In the first period of the Settlement Stage, the advocate will make an offer which amounts to a payoff of for the firm. T he firm will accep t this offer if The firm expects from regulation and because the advocate offers more, it will accept. The case will end in settlement. The outcome would be a negotiated settlement where the payoffs to the advocate and firm would be and respectively. However, if the n the case will end in regulation 28 When the advocate make s an offer in the first period, the firm will decline expecting more from regu lation. There will be disagreement until the Regulation Stage begins The same is true if only condition Inequality 2 21 is violated but When Inequality 2 22 is violated but Inequality 2 21 holds then the advocate will want 26 Note that it must be because 27 The analogous result in Cooter and Rubinfeld (1989) is that t he plaintiff and the defendant would want 28 Again this corresponds to the result of Cooter and Rubinfeld (1989) where parties go to trial if they are
51 settlement. He will make the offer in Expression 2 19c in the f irst period of bargaining. The firm will reject this and offer in the second period. The advocate will then deduce from this offer and understand that the firm expects more from the regulatory decision than its payoff shown in Expression 2 20c The advocate will accept The result will be another settle ment where the payoffs of the advocate and firm will be and respectively. Only if both Inequalities 2 21 and 2 22 are satisfied, the payoffs shown in Expressions 2 20a, 2 20b and 2 20c are the stipulated payoffs I f both Inequalities 2 21 and 2 22 are violated then it must be that and the case ends in regulation. Littlechil d (2009a) reports that a formal hearing is scheduled 5 months after a rate a case is initiated at the FPSC. After testimonies have been filed and counter filed, discussions among parties begin usually after the first month. A settlement is often concluded a short time before the formal hearing is scheduled (sometimes even on the night befo re). If no agreement is reached then the formal hearing begins. Littlechild (2009a) observes that there is much less incentive to settle after the hearing begins. T his description of the settlement process suggests that the parties can deduce what the othe r expects from regulation during the discussions before the formal hearing. Like the model, if the parties cannot agree the case eventually goes to regulation. Both Littlechild and Doucet (2006) and Wang (2004) claim that the regulatory decision is more p redictable than any trial verdict. This implies that in our model and should nearly be equal. In the high e range is it po ssible to have the settlement shares as shown by Expression 2 19c when ? The answer is
52 yes. In such a ca se it must be that This case corresponds to the case when the advocate and firm have the same expectation about the regulatory decision (perhaps because it is predict able) but both prefer the outcome that can only be reached by settlement. As mentioned before the results for the high e range are applicable to the low e and medium e ranges. In the low e range, the values of and for which the advocate and the firm would want to settle with the payoffs shown in Expression 2 20a are: ( 2 23) (2 24) In Inequality 2 24 we know and Because the range under consideration is is also non negative. This implies the second term on the right hand side of Inequality 2 24 is non negative. In the medium e range, the advocate and the firm would want the st ipulated outcomes of Expression 2 20b when both Inequalities 2 26 and 2 27 hold: ( 2 26) ( 2 27)
53 The second term on the right hand side of Inequality 2 27 is non negative because The next question examined is settle knowing it will be beneficial to the consumer. The advocate always acts in the wh enever Because this condition becomes This means that in the high e range, the advocate always acts in the best interests of the consumer given his beliefs if he settles when : ( 2 28 ) Because and is bounded from ab ove by 1, comparing Inequalities 2 21 and 2 28 there are at least as many values of for which the consumer advocate would settle as th ere is if he was only agreeing to settlements beneficial to the consumer. More specifically, for values of where the advocate will enter into agreements with the firm whi ch he believes will harm the consumer This is also true in the low e and medium e ranges. If the advocate was only concerned about consumer welfare then he should only settle for the outcomes of Expression 2 20a in the low e range when: ( 2 29 ) And similarly, the consumer advocate should only conclude agreements with outcomes of Expression 2 20b if he was only concerned about consumers in this range when: ( 2 30)
54 Inequalities 2 28, 2 29 29 and 2 30 30 demonstrate that the consumer advocate can agree to settlemen ts which the consumer would not want i f they shared the same belief about the regulatory decision. This belief is represented by Proposition 4 summarizes these observations Proposition 4: Suppose represents the belief about the regulatory decision of both the consumer and the consumer advocate. Also suppose the consumer Then there exists a range of in which the consumer advocate will settle contrary to the wishes of the consumer. It should be noted that, by construction, when the range of in which the advocate wants to settle coincides with the range in which the consumer would want him to settle. That is, when the consumer advocate always acts in the way the consumer wants. Note also that an increase in relaxes constraints Inequal ities 2 22 and 2 2 4 31 This happens because as rises the advocate is willing to take a smaller share of for the entire or a larger share of for the same in the SPE settlement equilibria Expressions 2 19a and 2 19c This means a larger pr ofit for the firm in Expressions 2 20a and 2 20c Consequently, there are more values of for which the firm will want the stipulated out comes of Expressions 2 20a and 2 20c 29 Note that because 30 Again, since 31 It has no effect on constraint Inequality 2 27
55 Proposition 5: The range of for which the firm wants settlement will be non decreasing when the consumer advocate weights his preference more than the cons when is more than ). In Proposition 3, we saw that the advocate may actually receive a smaller share of as increases. advocate may not actually end up with a larger share of with a lower There are more values of for which the firm will agree to th e negotiated outcomes of Expressions 2 20a and 2 20c when is larger Hence, a larger concern for the consumer (embodied in a smaller ) does not always result in a larger share of if the firm believes it can gain mo re in regulation A bigger may have made the firm agree to the stipulated outcome and subsequently, the advocate may have ended up with a higher share of than the regulatory decision would allow 32 2.2.2 The Company request C ase The analysis of a company request case is analogou s to the earnings review case. T o avoid repetition, the details of the analysis and the results are only explained when they differ from the earnings review case. 33 In the Regulation Stage, the regulator ( 2 P1b) 32 An increase in also loosens constraints Inequalities 2 21 and 2 26 It loosens Inequality 2 23 only if Nevertheless, the result that when increases it makes the firm more amenable to the stipulated outcomes of Expressions 2 20a and 2 20c still stands. 33 All of the propositions stated in the earnings review case hold for the company request case as well.
56 There can be two subcases o f the constrained maximization Problem 2 P1b The first case is when In this situation, the constraint in Problem 2 P1b is non binding. The analysis of this subcase is identical to the earnings review case. The only difference is that in the equilibrium condition and results, and should substitute and respectively. The more interesting subcase is when The constraint in Problem 2 P1b is binding when This is the subcase that one should find more prevalent in reality 34 It is usually w hen the revenues of a firm are not enough to cover operating and maintenance expenditures that the firm files a company request case Therefore, it is this subcase that we examine i n more detail 35 When the solution to Problem 2 P1b is: if ( 2 30a) if ( 2 30b) The regulator maximizes by making the consumer pay the entire shortfall to the (i.e., when ) 36 ) 37 i t 34 This is corroborated by the data of Littlechild (2009a). The utilities were granted an average of 52.2% of the rate increase they had requested in the 76 company request cases which went to regulation between the years 1976 and 2002. 35 As already stated, the analysis and results for the case are identical to the earnings review case. The only difference is that and must be replaced with and for the corresponding analysis and results in the subcase. 36 In the company ion is the same as the 37 Again it is assumed that the regulator allows the status quo to stand when as seen from Equation 2 30b. In the company request case, the status quo is not to make the consumer any amount to the firm. But because the firm needs to at least avoid financial distress, the consumer is made to pay the minimum possible when This minimum possible amount is
57 maximizes by determining that the consumer should make no payme nts to the firm. But the regulator is restricted by the financial distress constraint of Problem ( 2 P1b) whereby the firm needs at least to stay in business. Therefore, when the regulator makes the consumer pay the minimum amount that it m ust which is This means the consumer retains of the shortfall that the firm had applied to get. The resulting after regulation profit for the firm is: if ( 2 31a) if ( 2 31b) regulation weighted utility function which is equal to the regulation utility function is: if ( 2 32a) if ( 2 32b) The above after regulation profit ( 2 33) ( 2 34) regulation profit to be and his own expected weighted utility function: ( 2 35) ( 2 36) Like the earnings review case, are the disagreement vectors of the bargaining game of the Settlement Stage
58 The necessary condition for a SPE in the Settlement Stage is identical to the earnings review case ( Equilibrium Condition 2 17). The points of difference are that and need to be replaced with and and also, are defined by Equations 2 36 and 2 33 The SPE equilibrium offers are found exactly the same way as the earnings review case. In the limit the SPE division of the pots (without considering the disagreement vectors ) are: if ( 2 37 a) if ( 2 37 b) if ( 2 37 c) This implies the resulting payoffs from the offer are: if ( 2 38 a) if ( 2 38 b) if ( 2 38 c) Like the earnings review case, we can call the ranges
59 and the low e range, the medium e range and the high e range respectively. 38 The main difference between the company request case and the earnings review case is that some of the stipulated payoffs shown in Expressions 2 38a, 2 38b and 2 38c will never be agreed between the parties in their relevant ranges. This is because the cons umer advocate will always expect a higher after regulation utility. To see this, we will examine each of the three ranges of Expressions 2 38a, 2 38b and 2 38c in turn. As in the earnings review case, the advocate will want the stipulated payoffs in Expres sions 2 38a, 2 38b and 2 38c when and the firm when Therefore, in the high e range the advocate and firm set tle with the SPE partitions of Expression 2 37c when: ( 2 39) ( 2 40) From Inequalities 2 39 and 2 40 we find it is possible for the stipulated payoffs of Expression 2 38 c to be negotiated as long as both inequalities hold. In this range as well, the advocate always if he agrees to the payoffs of Expression 2 38c when: 39 ( 2 41) In the medium e rang e the advocate will want to agree to the stipulated payoffs of Expression 2 38b when: 38 Analogous to the earnings review case, can be seen as a measure of the relative sizes of and 39 Again we see from Inequalities 2 39 and 2 41 that if then the inequalities become identical. That is the advocate always acts in the best interest of the consumer.
60 ( 2 42) Because in this range of it means This implies for Inequality 2 42 to hold it must be that This is not possible as is a probability distribution and cannot be greater than 1 at any point. C onsequently, we find that the advocate will never want to settle for the payoffs of Expression 2 38b The firm woul d want to settle for the payoffs in Expression 2 38b when: ( 2 43) is non negative because and by Assumption 2 A1 Also, because and is n egative. This means that has to be smaller than some negative number for Inequality 2 43 to hold. Because is a probability distribution and it can never be less than 0 at any point, Inequality 2 43 will never hold. In other words, the firm wi ll always fancy its chances with the regulator than settle for its payoff in Expression 2 38b 40 The consumer wants the stipulated payoffs of Expression 2 38a in the low e range when: ( 2 44) Now the numerator of the second term on the right hand side of Inequality 2 44 is negative in the range 40 The advocate would be always acting in the best interest of the consumer if he settled for the payoffs in Expression 2 38b when But in this range Therefore, would have to be more than 1 for the consum er to benefit from settlement. Because this is not possible, the consumer would never want the stipulated payoffs of Expression 2 38b either.
61 The denominator of this term is non negative bec ause Consequently, for Inequality 2 44 to hold it must be that In this range, the advocate may want the stipulated payoffs in Expression 2 38a only if (as cannot take higher values). Even when the advoc ate is indifferent between the stipulated payoff of Expression 2 38a and his expected after regulation ut ility The firm will want its stipulated profit in Expression 2 38a when: ( 2 45) In the range being considered, it is possible for Inequality 2 45 to hold. Therefore, the f irm may want to settle with the Expression 2 38a outcomes in this range 41 It is clear that the consumer advocate will not want the stipulated outco mes in Expressions 2 38a 42 and 2 38b This means the advocate never wants an outcome that can only be reached in the Settlement Stage when falls in the range This range can be re ar ranged to: ( 2 46) The left hand side of Inequality 2 46 is the maximum amount of that the advocate can retain. This is because the firm knows that if the case goes to regulation then it will definitely acquire of from the regulatory ruling. on the right hand 41 The advocate, if only acting in the best interest of the consumer, should settle for the payoffs of Expression 2 38a in this range when This implies that like the case of the advocate, the consumer wants the advocate to sign the agreement only when Even then, the consumer is ind ifferent between its stipulated payoff and its expected regulation payoff like the advocate. 42 As already discussed, it is possible that the payoffs of Expression 2 38a may be a valid settled outcome if in the model. But for the sake of exposition this is ignored for the moment. Nothing valuable is lost in discussion with this oversight.
62 side of Inequality 2 46 is the discounted loss the advocate makes from making the consumer pay the entire amount. It can be seen from Inequality 2 46 that the maximum gain the advocate can make in retaining as much as as possible is less than the loss (as perceived by the advocate) of the agreement to pay all of Consequently, the advocate would never agree to any stipulated outcome in this range where the SPE of the Settlement Stage entails the payment of t he entire 43 44 Proposition 6 : In a company request case, there is no negotiated settlement and the case goes to regulation if and is sufficiently small relative to In the dataset of Littlechild (2009a), 26 out of 91 earnings review cases ended in a negotiated settlement between 1976 and 2002 in Florida In contrast, only 6 out of 82 company request cases were settled. According to the data, company request cases rarely end in a stipulated agreement between t he consumer advocate and firm. Let us examine whether the present model predicts this empirical finding. Recall that the stipulated outcome Expression 2 38c is possible if the Inequalities 2 39 and 2 40 hold. That means a negotiated settlement is possible when the pair satisfies This is the high e range where is relatively larger in magnitude to Therefore experience suggests it is rare that is large relative to in company 43 It is possible that even in the range the advocate is willing to take in exchange for a lower share of But in the range where th is candidate solution is a SPE (i.e. which is negative. The advocate would never agree to this. 44 For a stipulat ed outcome to be possible, has to be large enough to offset the loss from the payment of This can be seen in the fact that Expression 2 38c is a possible stipulated outcome. It is valid in the range (the high e range).
63 request cases. Because is the already incurred shortfall and the open ended and uncertain nature of as described by Littlechild (2009b) being smaller than seems plausible 45 In spite of Proposition 6 negotiated settlements can still be concluded in the low e and medium e ranges albeit with different payoffs than shown by Expressions 2 38a and 2 38b The cases will be settled if i.e. both parties think the other has a better chance of winning if the case goes for regulation 46 The case goes to the re gulator if or in other words, both parties believe that they have a better c hance of claiming the entire pie during regulation. The Florida experience suggests that usually in company rate cases 47 One reason for this could be that in the low e and medium e ranges both parties know there is no mutually beneficial outcome which can only be reached by settlement. Consequently, th e firm will only file a company request case when it believes it has a high chance of winning (i.e. high ) the entire shortfall in a regulatory formal hearing. Another reason could be that because both know no mutually beneficial solution exists that can only be reached in the Settlement Stage, they do not even bother to sit down at a negot iating table. Because there are no negotiations, the belief of each party (i.e. and ) cannot be revealed to the other during the 45 high (at least 50%) for the stipulated payoffs of Expression 2 38c to occur. This can be seen from Inequality 2 39 where has to be equal to or higher than (because ) for a negotiated settlement. 46 This is also true for the high e range. 47 From Inequalities 2 21, 2 22, 2 24, 2 25, 2 27 and 2 28 we see that the stipulated outcomes of Expressions 2 20a, 2 20b and 2 20c are possible in an earnings review case even when
64 bargaining process So the possibility of a negotiated settlement when is lost. 2.3 Policy I m plications A negotiated settlement cannot be put into effect without the approval of the the needs of all involved in the rate case One of the main advantages of settlement is that the formulators know thei r preferences better than the regulator. Consequently, any outcome agreed among all relevant parties in a negotiated settlement is superior to any regulatory decision. This is true as long as all parties are represented reliably by their agents during neg otiations Hence, in the negotiated settlement process I t was assumed if an agreement was reached by the advocate and the firm in the Settlement Stage then the rate case ended with the agreement as settlement. It was therefore assumed that the regulator automatically approved all the settlements put to it 48 But I nequalities 2 23, 2 26, 2 29 and 2 41 show that settlements can be concluded between the advocate and firm which are disadvantageous to the consumer. For this reason, it is the job of the regulator to preclude such settlements In the past, regulators such as the NEB in Doucet and Littlechil d (2009) and the FPSC staff in Littlechild (2009b) have approached this problem by trying to estimate the gains and losses the customer would make by the settlement outcome. On the basis of 48 practice, Florida PSC has almost invariably adopted negotiated settlements put to i (2009b) reports that the FPSC staff stopped objecting to accounting flexibility in settlements when it was clear that the FPSC, in spite of staff reservations, would adopt such settlements as long as these agreements procured sign ificant rate reductions and freezes for consumers.
65 this estimation the regulators decide whether to approve the sett lement. In the context of the model, it means they have tried to estimate A simplifying assumption of the model was that the future cost of the firm was known to the advocate and the firm. In reality, it is very diffic ult to estimate is a cost incurred in the future (most of the time incurred as a stream of future costs) and is dependent on a number of uncertain variables. Even the firm, the party with the best information regarding may not know the exact value of Therefore, a long with the estimation of regulators have tried to estimate Apart from the difficulty of these particular estimations, the disadvantage of this approach is that it undermines one of the main benefits that the regulator can gain from the negotiated settlement technique. This benefit is that the regulator does not need to actively arbitrate the case because a superior regulatory outcome can be reached because via settlement the preferences of the parties are better represented than any ruli ng by the regulator (provided that consumer preferences are accurately represented) Arbitration requires the regulator to identify the gains and losses to all parties of a rate change This involves trying to estimate their preferences When the regulator analyzes a settlement by trying to simulate the prefer ences of a consumer, it is acting just as it would have done in a regulatory formal hearing. T he regulator may just as well have waived the negotiation process and moved directly to regulation. Even if the regulator could accurately estimate and this method of trying to protect consumer interests in the negotiated settlement proces s may not work in regulation In the eyes of the parties this regulatory strategy amounts to by Stage. The consumer advocate and the firm will perceive that the regulator is taking a
66 decision according to its preference if it disallowed an agreement. This would be true e ven if the regulator takes the right decision to protect the consumer after correctly estimating and Therefore, s uch a regulatory policy incentive to settle. Why should the parties expend their efforts to settle if their agreement is rejected on the grounds that it does not produce a decision t hat is in line ? In such a case, the partie s might as well wait for the One implication of this is that for the negotiated settlement technique to function smoothly, approval of the settlement by the regulator must be automatic or, at least, very likely This is one of the reasons why the FPSC in spite of staff objections appro ved nearly all the set tlements put to it between the years 1976 and 2002. Littlechi ld (2009a, b) claim that the reason for the almost universal approval of settl ements is that the FPSC wanted to avoid conflict with the consumer advocate and minimize critic ism from all other involved parties. In my opinion as long as the FPSC wanted regulation to go on by negotiated settlement it had to approve all settlements presented to it. This is true even if the Commission was only concerned by consumer and shareholde r welfare and not criticism from the involved parties. It is up to the regulato r to decide if it prefers regulation by negotiated set tlement or deems it more important that it should examine the just and reasonableness of every stipulated term before givin g approval The latter choice comes at the risk of jeopardizing the whole settlement process itself. The most effective we can ensure that no negotiated settlement harms consumer s is to limit the divergence between consumer and advocate preferences. In the above model this means to minimize When the advocate signs only those
67 settlements which he thinks are beneficial to the consumer Policies that will make as close to 0 as poss ible must then be implemented One such policy is to make long tenures for consumer advocates once appointed 49 A short term in office with uncertain re appointment possibilities will tend to make an advocate more interested in procuring the most visible re sults in a short period of time. If the costs that the advocate agrees to period of tenure, the advocate has more incentive to agree to them. A longer term period should de crease this tendency. The general policy that the regulator should follow in trying to reduce is to increase the transparency of settlements. The terms of a settlement between parties in a legal dispute should be kept private. But there is no reason th at should be the true with the negotiated settlements of rate cases. The terms of a settlement should be known to the parties involved and one such party are the consumers of the public utility firm. The implications of the settlements should be made clear to these consumers. The regulator should facilitate in making these settlements more transparent so that the consumers themselves can judge whether their agent is doing a good job or whether a replacement is necessary. One way this can be done is for the regulator to publish regular reports on the job performance of the advocate 50 This report should list every settlement signed by the advocate in his term in office and their cumulative effect on the 49 The Public Counsel of the OPC in Florida is appointed and re appointed every year by the Joint Committee of Public Counsel Oversight, a state Congress committee. 50 The FPSC does publish a news release after a settlement is made between parties. This news release is published at the time the settlement is concluded and it consists of a cursory list of the terms. In my opinion, this is not enough to make the average consumer fully grasp the consequences o f the agreement.
68 rates that the consumer are paying. Such a policy will s ignificantly reduce close to 0. Once this has been achieved, the regulator can be sure that any settlement presented to it is advantageous to all involved and approve it automatically. 2.4 Conclusion This chapter investigated whe ther negotiated settlements might harm consumers We found that such harm can arise if the preference s of consumers and their advocate diverge. It seems that the regulator is aware of this but the present methods used to prevent such agreements by estimating future consum er costs are not conducive to the smooth functioning of this regulatory technique. The best way to facilitate settlements that benefit the consumer is for the regulator to implement steps so that the advocate his can be done by extending the (though in Florida, only the state legislature can do this and not the regulator) as well as active monitoring of his performance. preference the regulator can be certain that the beliefs about the regulatory decision. decision then no rate case will go to regu lation. This is because any order by the regulator can be replicated by the parties themselves during the Settlement Stage. Because regulatory rulings are said to be predictable it may be the case that the advocate and the firm will know how the regulator will rule before the rate case in many instances This implies that there is full information or near full information in many rate cases. Littlechild and Doucet (2009) discuss the Generic Cost of Capital decision at the NEB in 1994. This decision allows for an automatic mechanism by which the return on
69 common equity for each pipeline is determined on an annual basis. It is viewed as a n important measure for facil itating settlements at the NEB because it reduces the uncertainty of the regulatory decision a nd aligns the regulatory expectations of all parties. It would be interesting to see whether any other such mechanism can reduce the uncertainty and enhance the settlement process. The Public Counsel in Florida is appointed by a Florida Congress committee But this institutional set up is not common to all states. In some states, the advocate is appointed by the governor In 19 US states, ther e is no consumer advocate The Public Service Commission staff is supposed to represent the interest of the consumer in these states An interesting future direction of research could be to investigate which of these institutional frameworks yield the best outcome for the consumer. There has been much debate whether the growth of negotiated settlements has reduced the r egulator to the passive role of a facilitator from the proactive decision maker that it is in conventional regulation. This study shows there certainly are advantages of negotiated settlements over traditional ROR regulation. But it calls on the regulator not to be a passive facilitator but to be an active monitor of such agreements so that all involved can benefit.
70 CHAPTER 3 DO CONSUMERS ALWAYS GAIN FROM NEGOTIATED SETTLEMEN TS IN PUBLIC UTILITIES REGULATION ? E VIDENCE FROM FLORIDA 3.1 Introduct ion Until recently, the prices that consumers in the US and Canada pay for essential services like electricity have been determined by rate of return (ROR) regulation. Under ROR regulation, the public utility commission considers formal testimony by intere sted parties, including the regulated utility and a consumer representative. The commission then sets prices to generate revenue for the utility sufficient to cover its prudently incurred costs and a fair rate of return on its rate base. 1 Negotiated settl ement (or stipulation) has been replacing ROR regulation over the past 25 years. Under the negotiated settlement process, the firm, the consumer representative, and other involved parties negotiate a settlement on consumer rates and other items before any formal regulatory hearing takes place. The other items these sharing plans, and future behavio u r of the parties. The parties then submit the settlement to the regulator f or approval. If the settlement is approved, no formal hearing takes place and all the stipulations of the agreement are upheld. Traditionally, negotiated settlements were seen as a less costly and less time consuming way to reach the same outcomes that RO R regulation provides. Recent research argues that this is not the main reason for settlements. 2 These studies contend the settlement process offers outcomes beneficial to all parties that cannot be 1 2 See Wang (2004), Doucet and Littlechild (2009), and Littlechild (2009a,b).
71 reached by ROR regulation. Critics argue that the negoti ated settlement process transfer the decision making power from the regulator to the firm and large consumers. This chapter analyzes the data from rate cases of the Florida Public Service Commission (FPSC) to see whether the negotiated settlement practice has been beneficial to the consumer. It also tries to resolve the debate about whether time saving is a major motive behind such agreements. In spite of its prevalence in practice, negotiated settlement has not received much careful study. 3 Several studies recount the negotiated settlement experience in various regulatory commissions. Wang (2004) studies settled rate cases of major interstate gas pipeline at the Federal Energy Regulatory Commission (FERC). Littlechild (2009a,b) examine negot iated settlements of rate cases of telecommunications, electric and gas companies at the FPSC. Doucet and Littlechild (2009) document the development of the negotiated settlement practice in toll cases of big pipeline companies at the Canadian National Ene rgy Board (NEB). These studies agree that reduction of costs and uncertainty, and saving time the alleged main reasons for choosing settlements over regulation are not the driving motivation behind these types of agreements. The consensus of these papers i because the regulator cannot legally prescribe such solutions or is unwilling to do so because it wants to protect itself from legal appeal of its decision. 3 See Doucet and Littlechild (2006) for a survey of economic and legal literature on negotiated settlements.
72 Because the reasons for settlement are different from those of a legal case (reduction of uncertainty and costs, and saving time), a theory of negotiated settlements in regulation is needed. The second chapter of this dissertation formulates a theoretical model where a firm and a consumer advocate 4 try to negotiate an agreement. If the parties fail to reach an agreement then the regulator decides consumer rates. The analysis finds that negotiated settlem ents are good for consumers as long as the interests of the consumer advocate and the consumers can diverge if the advocate values a present rate reduction or freeze more highl y than consumers. Rate reductions in settlements are an observable signal of good job performance by the consumer advocate. The advocate may negotiate present rate decreases from the utility in exchange for the transfer of future cost burdens from the firm to the consumer that may not benefit consumers. 5 4 A consumer advocate appointed by a state Congress committee and is charged with representing consumer interests in rate cases in Florida. This institutional set up is not common in all states. What is common, however, is that there always i s a representative of the consumer in rate cases. This may be a consumer advocate or staff of the regulatory commission entrusted with representing consumer interests. 5 Job performance may not be the only reason why advocates may agree to such settlements. The consumer advocate may actually believe that the consumer values present rate decreases more. This is a quote from State of Florida Public Counsel Activity Report p. 11 (June 30, 2003) (the Public Counsel is the consumer advocate in Florida) when talking about intervening in rate cases of water and wastewater and i nvolved in water and wastewater cases than either electric or telephone cases. While the aggregate dollar effect of a water and wastewater case is far smaller than the much larger electric or telephone cases, the per customer effect is often of the same ma gnitude. In addition, the greater tangibility of the product and the local management that often characterizes these small utilities tends to generate a repre had better information about the future.
73 Several stipulated items constitute these future cost transfers. 6 According to Littlechild (2009b), these future cost transfers are not set during regulation because, either the regulator cannot legally pr One common item is that settlement parties agree not to initiate a rate case for a specific time period. This means that the firm can retain any earnings more than its allowed revenue during that period The regulator cannot bar parties from filing a rate case. Another such stipulation at the FPSC enables the firm to have a more flexible depreciation and amortization policy than the commission would otherwise allow. The firm is then able to increase its reported operating expense in years in which it earns more than its revenue requirement to keep all of its annual earnings. Another feature of FPSC settlements is revenue sharing arrangements. The FPSC staff has stated that such arrangements allow the firm to earn revenue more than its authorized return on e quity (ROE) set in regulation. Both the depreciation and amortization policies and the revenue policy. These future cost transfe rs typically are not observable to consumers at the time of the agreement. If the costs are observable then the consumer may lack the sophistication required to fully grasp the implications of the stipulated items. In such a case, the consumer advocate may fashion agreements detrimental to consumer interests. The theoretical law and economics literature has discussed how a lawyer may not act in the best interest of the client during settlement negotiations owing to a divergence 6 For a detailed discussion of these items and others, see previous chapters.
74 of preferences (for example, Mnookin 1993). But there has been no empirical investigation on this principal agent relationship. In general, compared to the theoretical research on negotiated settlements in this literature (for a review, see Cooter and Rubinfeld 1989), there is much le ss empirical work. 7 According to Danzon and Lillard (1983), there are two challenges that confront standard econometric techniques in such studies. One is that some of the variables such as expected trial verdict that determine the observed settlement and trial outcome are unobservable. The second is that settled independent variables whose effects are being measured. Consequently, observed settlement and trial outcomes cannot be gene ralized into one population. The contribution of this chapter is two folds. Using data from the FPSC, it estimates the payoff functions of the consumer advocate and firm during a rate case. By this estimation, we see how the advocate values present rate ch anges compared to future cost transfers in settlements that negatively affect future consumer welfare. If the consumer advocate values rate reductions more than the future consumer welfare then negotiated settlements harmful to the consumer are possible. T his would be a serious problem with negotiated settlements as it is presently practiced. The second question answered by the estimation is whether saving time is a main reason of settling rate cases as claimed by parties (and opposed by researchers). 7 For empirical studies on negotiated set tlements in other contexts, see Danzon and Lillard (1983), Viscusi (1986), White (1989), Hughes and Snyder (1990), Farber and White (1991), Hughes and Snyder (1995), Kessler (1995), Chang and Sigman (2000), and Browne et al (2004). For experimental studie s, see Ashenfelter et al (1992), Coughlan and Plott (1997) and Babcock and Landeo (2004).
75 The a nalysis proceeds as follows. Section 3.2 describes the theoretical model that underlies the e mpirical estimation, Section 3.3 formulates the empirical model, Section 3.4 discusses the data, Section 3.5 presents the empirical results and Section 3.6 concludes. 3.2 Theory This section reviews the theoretical two stage model of Chapter 2 and presents the key results that are relevant to the empirical analysis. The theoretical model considers two types of rate cases separately: an earnings review case a nd a company request case. An earnings review case is initiated when the firm earns more than its allowed revenue. A company request case is initiated when the firm claims to have earned less than its allowed return. For simplicity, we focus on the earning s review case. The profit function of the firm before the game begins is: ( 3 1) In Equation 3 1 maintenance cost, re cost and factor for future costs. As discussed, may be the amount of earned future revenue more than the allowed revenue that the firm expects to give up to the consumer under ROR regulation. In an earnings review case, can be divided into two parts: 8 ( 3 2) 8 It is assumed That is the regulator allows the firm to cover its present operating and maintenance costs. Because this constraint is nonbinding in the ear nings review case and not important in the ensuing empirical analysis, we do not discuss further.
76 where is th e amount of revenue that the firm earns in excess of The game begins after the rate case has been initiated and the formal hearing date set. The first stage is the Settlement Stage. In this stage, the consumer advocate and the firm simultaneously ba rgain over the share of each party is to receive and the share of each party is to pay. is the stipulated portion of that goes to the consumer and is the corresponding stipulated share of that the consumer has to pay if there is agree ment between parties in this stage. This implies that and are the corresponding shares of the firm. Therefore, and esented by its profit function, Equation 3 3, if there are periods of bargaining within the Settlement Stage: ( 3 3) In Equation 3 3 is the discount factor for each period in the Settlement Stage. should not be confused with The fi rm uses to discount its profit in each successive period of bargaining within the Settlement Stage. In contrast, discount factor of a future cost that is incurred after the Settlement Stage. The ( 3 4) In Equation 3 4 9 We assume: ( 2 A2) reflecting the presumption that firms are at least as patient as the consumer. 10 9 and are not the same for the same reason and are not related.
77 It is assumed that the consumer advocate is only concerned about the share of in an earnings review case that he can obtain for the consumer. This is because the value of consumer will perceive a higher value of as a better job done by the advocate and this will increase his chances for re appointment. Therefore the utility function of the consumer advocate is: ( 3 6) During negotiations with the firm and the regulatory formal hearing, it is assumed that the advocate maximizes a weighted function of his own utility function and the 11 ( 3 7) where is the weight the advocate assigns to hi s own utility function. over time, for example. 12 The consumer, however, has no knowledge of th e consumer and not after the game ends. is a cost that the consumer must pay in future and therefore, he cannot observe it at present. 10 A natural choice of would be where is the interest rate. 11 Charness and Rabin (2002), for example, employ a similar payoff function. 12 years during which he concluded settlements and was involved in rate cases repeat edly with the same major Florida public utility firms.
78 The bargaining follows an infinite hori zon standard alternating offers procedure (Osborne and Rubinstein 1990). The consumer advocate proposes a division of and in even periods The firm makes corresponding proposals in odd periods. 13 The player has two options when responding t o an offer: (1) accept the offer in which case the game ends in a negotiated settlement or (2) reject the offer and make a counteroffer in the next period of bargaining. Option (1) implies that the regulator approves any settlement presented to her. 14 The d isagreement vectors of the bargaining game in the Settlement Stage are the expected payoffs the parties will receive in the Regulation Stage. Because the rate case has already been initiated, if there is disagreement between the advocate and firm then the Settlement Stage would eventually end and the Regulation Stage would begin. The result would be the regulation payoffs for all parties. 15 In the Regulation Stage, the regulator must decide the partition of between the parties. The regulator does not ch oose Because of the nature of the future costs considered here, either the regulator has no jurisdiction to set or it is against the above 0. In both instances remains equal to 0 as it did 13 In the limit it does not matter to the outcome which player makes the first offer. 14 the years 1976 to 2002, only one settlement was overturned. This settlement was approved in 1989 after one item that the FPSC opposed was removed from the agreement. 15 This is consistent with Cooter and Rubinfeld (1989) who say that the disagreement vector or threat point (as they call it) in the settlement stage will be the expected trial verdict.
79 before the rate case was fi led. The regulator chooses by maximizing a weighted 16 ( 2 P1) In the maximization of Problem ( 2 P1), is the weight that the regulator attaches according to her duty of balancing the needs of the firm with those of its customers. Uncertainty of the regulatory decision is introduce d in the model by assuming that the consumer advocate and firm do not know the value of They do, however, have beliefs about the value of These beliefs are represented by the probability density functions and for the advocate and firm respectively. The corresponding probability distribution functions are and Proposition 4 of Chapter 2 shows that if the advocate and the consumer share the ), then This happens when the advocate values an additional unit of more than the consumer (i.e. when in Equation 3 7 nd the consumer with more units of for an extra unit of than the consumer would want resulting in an agreement undesirable for the consumer. 16 regulator could as well as be maximizing a weighted function o
80 This chapter estimat es how much the advocate values rate reductions relative to future consumer welfare and thereby, sees whether the possibility of settlements harmful to consumers exists. 3.3 Empirical M odel functions in this model. These are shown by Equations 3 8 and 3 9 respectively: 17 ( 3 8) ( 3 9) In the above two equations, expected future profit, either by settlement or the regulator, is a measure of future consumer welfare, is the expected duration of the rate case, is the future total operating revenue of the firm, is a dummy variable that equals 1 when the rate case is of a telephone company and is anothe r dummy variable that equals 1 when the case is and capture the effect of unobserved variables in Equations 3 8 and 3 9 respectively. In the following subsections, we discuss the derivations of Equations 3 8 and 3 9 and their independent variables. is the utility of the consumer advocate at the time of settlement, or, if no settlement is reached, expected utility after regulation by the regulatory commission. is the future profit that the firm expects to 17 The model was also estimated with squared terms of the important independent variables in Equation 3 8 The bivariate probit model with partial observability either did not converge or the results remained unchanged when these squared terms were added.
81 earn d uring the period the settled or regulated rates are in place at the time of settlement or post regulation. In the dataset we use to estimate Equations 3 8 and 3 9 each observation is a docket which represents a rate case of a public utility. and can not be observed. We can, however, observe whether the rate case ended in a negotiated settlement. A settlement occurs only when both parties expect to gain more from the settlement outcome than from regulation. If we construct the variable which equals 1 if a rate case is settled and 0 if the case is regulated, we have: if and ot herwise. ( 3 10) The model specified by Equations 3 8, 3 9 and 3 10 can be estimated as a bivariate probit model with partial observability as described by Poirier (1980). Because is the actual total rate change that occurs in a rate case, it may be endogenous in Equations 3 8 and 3 9 This is because the rate change may be determined by the parties simultaneously with the decision to settle. If this is the case, then will be correlated to and To control for this endogeneity we follow the procedure formulated by Rivers and Vuong (1988) for endogenous independent variables in simultaneous probit models. This involves running an OLS regression of on all the independent variables and instrumental variables. We then save the residuals from this regression and estimate the bivariate probit model with the saved re siduals in both equations. The Equations 3 8 and 3 9 now become Equations 3 8a and 3 9a respectively: ( 3 8a)
82 ( 3 9a) where in Equations 3 8a and 3 9a are the residuals from the OLS regression of on the instrumental variables. Assuming have a bivariate normal distribution, the model specified by Equations 3 8a, 3 9a and 3 10 can be estimated using the likelihood function: ( 3 11) In Equation 3 11 we have: ( 3 12) ( 3 13) and Identification of this model requires that at least one independent variable present in one of the payoff functions be absent in the other. This is satisfied as can be seen from Equations 3 8a and 3 9a Before moving on to the next section, the following subsect ions discuss the derivation of Equations 3 8 and 3 9 from the theoretical model and the construction of the independent variables. We also discuss the instrumental variables used in the OLS regression at the end of the section. 3.3.1 Equation 3 8 : t unction ( 3 8) Equation 3 7 in the theoretical model and Equation 3 8 we see that corresponds to and corresponds to Because the advocate is assumed to only care about the rate change (i.e. in the limit ), the rate change represents is the
83 operating rev enue of the firm divided by the number of customers at least 11 months after the rate change has been implemented. It is the future average price paid by the Equation 3 7 can also be written as Eq uation 3 14 in the limit : ( 3 14) represents in Equation 3 14 and includes any share of future costs transferred to the consumer through the settlement as well as any gain ma de with the rate change We add the expected duration variable to test whether saving time and the expected signs for some is discussed below. : This variable is the total rate change (i.e. the total change in the allowed revenue granted to the company) that occurs in a rate case. It is measured in 1/100000 th of a dollar. 18 Rate reductions are negative and rate increases are positive. If there is no rate change, then is equal to 0. In some cases, a one time rate refund was ordered by the regulator. Because the average time for which rates were effective in the sample was 2.27 years, these one time refunds were divided by 2.27 and aggregate d with the permanent rate changes. 19 Since a rate reduction is good both for the consumer advocate and the consumer, the expected sign of is negative. 18 All money amounts were measured in 2010 dollars. 19 A similar approach is employed by Littlechild (2009a).
84 revenue divided by t he number of customers. 20 21 It is also measured in 1/100000 th of a dollar. The operating revenue and number of customers 22 used in calculating this variable are taken at least 11 months after the determined rates have become effective. There are two reasons for measuring this variable after 11 months. The first is that the revenue. After 11 months, most of the ordered annual rate change would have been implemented through customer bills. 23 The second reason is the future costs of the firm that may be transferred to the consumer via settlement may not immediately take effect after the rates have been implemented. 24 measures the average amount the consumer pays after some time has elapsed, allowing the possibility for some of the 20 The number used to divide the operating revenue wa s the number of customers at the time the operating revenue of the company was reported. In some years, this was unavailable for electric and gas companies. In its place the monthly average number of customers was used for the year the operating revenue wa 21 to 1984, th e FPSC Annual Reports reported both the number of customers and the number of access lines (or alternatively, number of main stations and trunks) for each regulated telephone company in Florida. The ratio of number of customers to number of access lines wa s 1.00093 and the standard deviation was 0.0652 for these years. 22 Number of customers includes industrial and commercial customers as well as residential. 23 Annual operating revenues and number of customers are reported usually at the end of the year (bet operating revenue after a year from the date they become effective. 11 months was chosen instead of 1 year because many of the rate changes were implemented on t he first day of the year. For example, if the operating revenue and number of customers was reported on 31 December and the rates of the final order became effective on 1 January as in many cases, we would calculate with the operating reve nue and number of customers for that year. If the rates became effective in March then we would calculate with the operating revenue and number of customers of the following year. 24 The previous chapters discuss what these future cost may be and their nature.
85 stipulated future cost transfers. Consumer welfare declines as increases. Therefore, one would expect to have a negative sign. : This variable was used a s an indicator for the duration in days that the parties expected the rate case to last if it went for regulation. The dockets were divided by type of cases following Littlechild (2009a). The first two types were earnings review cases and company request c ases. All other types such as periodic reviews on ROE, Modified Minimum Filing Requirements (MMFRs) and tax savings were grouped together in a third type. This last type involves small rate changes (nearly all the time a reduction). Within each type, the d uration of the cases that went for regulation was calculated. The duration is the number of days between a beginning date on which the docket was requested open and the date of the regulatory order that implemented the earliest 25 final 26 rates. The opening d ate was the earliest date in the docket index listing. The average duration for a regulated case was then calculated for each type and this was used as the variable. Because the parties prefer shorter rate cases, should have a negative sign. 3.3.2 Equation 3 9 unction ( 3 9) Equation 3 3 in the limit can be re written as: ( 3 21) 25 Sometimes in a docket there was more than one regulatory order implementing final rates. Much of the time, the orders that came after were just upholding or extending the items of the first one. 26 Regulatory orders which implemented interim rates were not considered.
86 in Equation 3 21 is the future net operating income of the firm. The total rate change 1 year after the rates become effect after at least 11 months 27 is used to represent in Equation 3 9 Like in the construction of revenue after 11 months to allow enough time for some of the cost transfers in the form of to pass on to the consumer. The total operating revenue is used instead of the reported net operating income because there is a chance that the firm may inflate costs to recover more revenue. 28 is included in Equation 3 9 to test the claim that saving time is one of the primary reasons for settlement. We also add the dummy variables and that equal 1 when the rate case is of a telephone company and electric company respectively. The construction and expected sign for some of the independent variables are discussed below. : The firm makes a decision about whether to settle by comparing what it expects its future profit will be under the regime of rates set by negotiated settlement to expected future operating revenue. is the operating revenu e of the firm at least 11 months after 29 the ordered rates of the case have been implemented. It serves as a 27 The reasons for choosing 11 months instead of 1 year are the same as in the construction of (see footnote 21). 28 actually corresponds to in Equation 3 21 29 It is reported in the same way as the operating revenue and number of customers were obtained to calculate (see footnote 21).
87 measured in 1/100000 th of a dollar like and profit should increase with an increase in its expected operating revenue, should be positive. : We expect to have a negative sign because the firm prefers a shorter proceeding to a longer one. : This dummy variable equals 1 if the rate case is of a telephone company. : This dummy variable equals 1 if the rate case is of an electric company. 3.3.3 Instrumental V ariables for The instrumental variables used in the OLS regression of are and : This variable is the amount requested by the firm when seeking a rate increase in a company request case. It is zero for earnings review cases and the third category of cases which includes all other types of cases other than earnings review or company request. : An earning review case is initiated because the firm is thought to have earned more than its allowed revenue. The third type of cases which include ROE reviews, MMFR s and tax savings cases also almost always result in a rate reduction. This rate reduction in both types of cases depends on the past earnings of the firm. is the total operating revenue that the firm earns the year before the rate case wa s filed. It is zero for company request cases. 3.4 Data We use the same dataset as Littlechild (2009a). This dataset was compiled by the FPSC. The database contains details of decisions in over 300 dockets that have
88 implications for the revenue of investor owned telephone, electric and gas companies between the years 1960 to 2008. The decisions cover the largest investor owned utilities regulated by the FPSC from 1968 onwards. These utilities are 13 telephone companies, 8 natural gas companies and 5 electri c companies. Specifically, the database is thought to cover (at least after 1976) (1) all decisions on base rate increases and decreases, (2) all decisions on changes in authorized return on equity (ROE), (3) results of all decisions relating to earning an d over earning cases, and (4) only those commission related decisions on amortization, reserves, incentive plans and other issues that had an implication on company revenue (Littlechild 2009a). Littlechild (2009a) uses one docket as one observation. This is the same approach taken here. We study the dockets in this database that were opened between the years 1978 2008. This period was chosen because some of the dates used to calculate were missing before 1978. There were 239 such dockets. 186 of these are analysed in the chapter 30 Out of these 186 dockets, 37 ended in a negotiated settlement. Table 3 1 shows industry wise, how many dockets ended with a settlement. From this database, we obtain whether the docket had a negotiated settlement outcom e, the date of the earliest regulatory order implementing final rates, the rate increase requested in a company request case, the total rate change, the date when the 30 37 dockets that were not included in the analysis were of the telephone companies: Florala Telephone Company, Fr ontier Communications of the South, Quincy Telephone Company and St. Joseph Telephone Company. The operating statistics of these companies included those from other states as well as from Florida. Their statistics from their operations in Florida alone cou ld not be separated from this data. Hence, rate cases of these companies were left out of the analysis. The data for the remaining 16 dockets was unavailable.
89 new rates ordered in the docket became effective and the type of case. The opening dates needed in the calculation of These index listings are on the FPSC website for the dockets opened in the year 1989 and onwards. For dockets opened in earlier years, the docket index listings are stored on microfi lm at the FPSC in Tallahassee. operating revenue and number of customers (number of access lines or number of main stations and trunks in the case of telephone companies) were reported from the FPSC Annual Repo rts. For some years and some companies, when these operating statistics were not available, we recorded this data from the company annual report filed with the FPSC or FERC (in the case of electric companies). 31 Table 3 2 presents the summary statistics of the independent variables. 3.5 Empirical results In this section, the results from OLS regression of are presented before moving on to the results of the main analysis. 3.5.1 OLS R egression of on I nstruments Table 3 3 presents the result s of the OLS regression of on the instrumental variables and the independent variables. Both and are statistically significant at the 1% level. 3.5.2 Bivariate Probit Model with Partial Observability of the Negotiate d settlement P ractice at the FPSC The results of the estimation of the partially observable bivariate probit model are presented in Table 3 4. is statistically significant at the 5% level in Equation 3 8a 31 Operating statistics and number of customers was recorded from company annual reports filed wit h the FPSC for 37 observations.
90 This means that is endogenous in Equation 3 8a is statistically significant at the 5% level. As hypothesized, is negative. in Equation 3 8a is statistically insignificant even though it is negative. This suggests that the rate change is more in Equation 3 8a is not statistically significant but it does have a negative coefficient. This indicates that the consumer advocate is not concerned about the time saved from avoiding the formal regulation process. None of the coefficients of the independent variables Equation 3 9a are statistically significant. In particular, the coefficient of in Equation 3 9a is statistically i nsignificant suggesting that, like the advocate, saving time is not one of the main reasons for the firm to settle. This result seems consistent wit h the analyses of Doucet and Littlechild (20 09), Littlechild (2009a, b) and Wang (2004) that time cost savings is not a significant driving motivation behind the negotiated settlement process even though it is claimed as such by parties. The marginal effect of the statistically significant variable on the probability of a settlement was calculated. The marginal effect of on the probability of settlement was 0.00043 when all the other variables were held at their mean and was 0. This means that an offer of a $10 million rate reduction (the average rate reduction was $44,839,210) will increase settlement probability by 4.3%. The average rate reduction increases settlement probability by around 19%.
91 3.6 Conclusion T he analysis of this chapter suggests that a larger rate reduction is the main reason why the consumer advocate agrees to a settlement. The high present rate reduction serves as an observable signal for good job performance. The longest serving Florida Publ ic Counsel, Jack Shreve, went on to be appointed Special Counsel for Consumer Affairs to the Florida Attorney General in 2003. In the press release from the highlights of his t ime as Public Counsel were reported as his negotiated settlements with Progress Energy Florida, Florida Power and Light, and BellSouth that secured large rate reductions for consumers. 32 These preferences of the advocate are of concern if these rate reduct ions are acquired in return for higher costs that the consumer must pay in future. Consequently, the consumer can be harmed by such negotiated settlement practice. The possibility of such harmful settlements requires some policy action. The NEB in Doucet and Littlechild (2009) and FPSC have approached this problem in the past by judging the merit of each individual item of an agreement before approval. But this approach is no different from the form al hearing in ROR in which the regulator examines each iss ue separately after and arbitrates accordingly. If parties knew that this was the policy being taken by the regulators, then they would have no incentive to negotiate and would just wait for the formal hearing itself. This is what the experience at the NEB suggests. Doucet and Littlechild (2009) report that between 1985 and 1995, when the 32 See http://myfloridalegal.com/newsrel.nsf/newsreleases/0FD32BAD8D00C2CB85256DAB0056A5B4?OpenDocument
92 NEB followed this policy, there were only three fully negotiated settlements. In 1995, the NEB changed its policy to accepting a settlement in its entirety. Since then, ap art from 4 cases, all major gas and oil pipeline company tolls have been set by settlement. Accordingly, the choice the regulator faces is between a no negotiated settlement regime or a negotiated settlement regime with near automatic approval. The questio n then becomes is that if the regulator chooses the latter 33 then what policy options are available to prevent harmful agreements. ssence, this means trying to make in Equation 3 7 equal to 0. With such policies in place, the regulator will be assured that any settlement brought to it for approval will be in the interest of consumers. One such legislative policy is to extend the te nure of the consumer advocate. 34 The shorter and secure present observable gains in exchange for future losses unobservable currently. Another policy is to increase the tr ansparency of the negotiated settlement process and the consumer advocate agency. 35 The regulator can play a role in this by publishing regular reports about the agency and recording the cumulative effects that past 33 As mentioned earlier, the FPSC has accepted almost all settlements put to it in spite of staff objections to some of them. Therefore, it seems that the FPSC has chosen a negotiated settlement regime. Littlechild (2009a, b) say the reason for this is that t he regulator wants to minimize conflict and criticism from parties. This may be an additional reason for automatic approval at the FPSC. 34 In Florida, the Public Counsel is appointed and re appointed every year. The appointment is made by a state Congress committee Joint Committee on Public Counsel Oversight. 35 In its existence since 1974, the Florida Office of Public Counsel (OPC) has published one activity report in 2003.
93 negotiated settlements signed by the pres ent advocate have had on present consumer rates. The results also show that time saving as measured by is not of much significance to either party in their decision to settle. This strengthens the claims made by Doucet and Littlechild (20 09), Li ttlechild (2009a, b) and Wang (2004). It must be noted that these results are applicable to the FPSC. Rate cases at other state and federal regulatory commissions may have different outcomes. It would be interesting to see how the results compare. This is especially so if the institutional framework of the commission is different. In some states, the consumer advocate is appointed by the governor. In 15 states, there is no consumer advocate. The public service commission staff represents the interests of th e consumer in rate cases in such states. The differences these cause in outcomes would be an interesting subject for future research and policy design.
94 Table 3 1. Distribution of negotiated settlements industry wise Industry Negotiated Settlement Regulated Cases Total Telephone 23 46 69 Electric 11 54 65 Gas 3 49 52 Total 37 149 186 Table 3 2. Summary statistics of independent variables Mean Standard Deviation Minimum Maximum rate 3.684493 1399.14 12502.63 5780.787 avgprice 0.0208135 0.0170582 0.000159 0.196655 exdur 259.0195 79.68544 168 347.66 oprev 14329.42 25660.83 5.088523 129655.6
95 Table 3 3 OLS regression of rate on the instrumental variables Coefficient (standard error) constant 241.0872 (387.2055) avgprice 3089.97 (5161.314) oprev 0.0033051 (0.0055336) exdur 0.7287268 (1.185068) tele 129.4369 (211.909) elec 191.2793 (226.9781) comreq 0.4265083*** (0.074171) pastearn 0.0285764*** (0.0067749) Notes: N=186. *=significant at the 10% level, **=significant at the 5% level, ***=significant at the 1% level.
96 Table 3 4 Bivariate probit model with partial observability of the negotiated settlement practice at the FPSC Coefficient (standard error) Equation 3 8a constant 0.4567369 (0.5901401) rate 0.0011958** (0.000555) avgprice 22.52024 (14.42497) exdur 0.001763 (0.0021899) resid 0.0008762** (0.0003953) Equation 3 9a constant 23.75832 (74.15168) oprev 0.0019223 (0.0060286) exdur 0.0667963 (0.1929536) tele 20.95455 (32.01735) elec 0.2289646 (17.71966) resid 0.0216145 (0.0582136) 0.84 (2.285803) Log likelihood 59.188619 Notes: N=186. *=significant at the 10% level, **=significant at the 5% level, ***=significant at the 1% level.
97 APPENDIX A PROOF OF THE CANDIDATE SOLUTIONS TO EQUILIBRIUM 2 17 Let the profit function of the firm equal a constant : (1A) Totally differentiation Equation 1A with respect to we get: (2A) : (3A) Totally differentiation Equation 3A with respect to we get: (4A) Equ ation 2A shows that the advocate must offer to increase by units in order to increase so that the firm earns the same amount of profit. For the consumer advocate to remain on the same utility level we see from Equation 4A that for a gain in must increase by units. Because it is always optimal for the advocate to offer or and (depending on the sizes of and ). A similar argument can be made to show that it is always optimal for the firm to offer or and This implies that the possible equilibrium offers fall into the four categories of Candidate Solutions i, ii, iii and iv
98 APPENDI X B PROOF THAT THE SPE PARTITI ONS OF AND ARE EXPRESSIONS 2 18A, 2 18B AND 2 18C To prove Expressions 2 18a, 2 18b and 2 18c, we examine which of the four Candidate Solutions i, ii, iii and iv solve Equilibrium Condition 2 17 Candidate S olution i : and Candidate Solution i implies that the Equilibrium Condition 2 17 becomes: (1B) (2B) Sol ving Equations 1B and 2B for and we get: (3B) (4B) For Equations 3B and 4B to be a valid solution, it must be that and Because of Assumption 2 A1 and, and being discount factors (i.e. both these parameters are between 0 and 1), it must be that and For it must be that and, for it must be that Therefore, the SPE partition in the range is This proves Expression 2 18c Candidate S olution ii : and Candidate Solution ii implies that the E quilib rium Condition 2 17 becomes: (5B)
99 (6B) Solving Equations 5B and 6B for and we get: (7B) (8B) For Equation 7B to be a valid solution it must be implies: (9B) For Equation 8B to be a valid solution it must be implies: (10B) From Equations 9B and 10B, we see that for Candidate Solution ii to be a valid solution it must be that: (11B) In the limit Inequality 11B becomes: (12B) Because of Assumption 2 A2 and the I nequalit y 12B cannot exist. This means that Candidate Solution ii cannot be a possible solution to Equilibrium Condition 2 17 Candidate S olution iii : and Ca ndidate Solution iii implies that the Equilibrium Condition 2 17 becomes: (13B)
100 (14B) Solving Equations 13B and 14B for and we get: (15B) (16B) For Equation 15B to be a valid solution it must be that implies: (17B) Because of Assumption 2 A1 Assumption 2 A2 and and being discount factors (i.e. these parameters are between 0 and 1), all the three terms on the left hand side of Equation 17B are nonnegative. Hence, I neq uality 17B is satisfied. implies: (18B) For Equation 16B to be a valid solution it must be that implies: (19B) Assumption 2 A1 Assumption 2 A2 and, and being discount factors imply that the first two terms on the left hand side of Equation 19B are positive and the third term is negative. In the limit Equation 19B becomes: (20B) The left hand side of Inequality 20B is nonnegative. Therefore, in the limit Inequality 20B exists. implies:
101 (21B) From Inequalities 18B and 21B, we see that for Candidate Solution iii to be a valid solu tion: (22B) In the limit Inequality 22B becomes: (23B) Inequality 23B implies that Candidate Solution iii is valid in the limit because of Assumption 2 A2 and Consequently, the SPE partition in the range is This proves Expression 2 18b Candidate S olution iv : and Candidate Solution iv implies that the Equilibrium Condition 2 17 becomes: (24B) (25B) Solving Equations 24B and 25B for and we get: (26B) (27B) For Equation 26B to be a valid solution it must be that implies: (28B)
102 implies: (29B) Similarly, for Equation 27B to be a valid solution it must be that implies: (30B) implies: (31B) For Candidate Solution iv to be a valid solution, Inequalities 28B, 29B, 30B and 31B must hold. Because Inequality 28B implies that Inequality 30B will hold and Inequality 31B implies Inequality 29B will hold. Multiplying both sides of Inequality 31B with we get: (32B) Because we can write: (33B) We see from Inequalities 28B and 33B that Candidate S oluti on iv is a valid solution in the range This means the SPE partition in the range is This proves Expression 2 18a
103 LIST OF REFERENCES Ashenfelter O., Currie, J., Farber, H. S., & Spiegal, M. ( 1992 ). An Experimental Comparison of Dispute Rates in A lternative Arbitration Systems. Econometrica 60 1407 1433. Babcock, L., & Landeo, C. M. ( 2004 ). Settlement escrows: an experimental study of a bilateral bargaining game. Journal of Economic Behavior and Organization 53 401 417. Bebchuk, L. A. ( 1984 ). Litigation and settleme nt under imperfect information. RAND Journal of Economics 15 404 415. Binmore, K., Rubinstein, A., & Wolinsky A. ( 1986 ). The Nash Bargaining Solution in Economic Modelling. RAND Jou rnal of Economics 17 176 188. Browne, M. J., Pryor, E. S., & Puelz, R. ( 2004 ). The Effect of Bad Faith Laws on First Pa rty Insurance Claims Decisions. The Journal of Legal Studies 33 355 390. Brueckner, M. ( 2000 ). Unanimity verses Consensus Bargaining Games: Strategic De legation and Social Optimality. Mimeo. University of Bonn. Cai, H. ( 2000 ). Bargaini ng on Behalf of a Constituency. Journal of Economic Theory 92 234 273. Chang, H. F., & Sigman, H. ( 2000 ). Incentives to Settle under Joint and Several L iability: An Empirical An alysis of Superfund Litigation. The Journal of Legal Studies 29 205 236. Charness, G., & Rabin, M. ( 2002 ). Understanding Social Preferences with Simple Tests. The Quarterly Journal of Economics 117 817 869. Cooter, R. D., & Rubinfeld, D. L. ( 1989 ). Economic Analysis of Legal Disputes and Their Resolution. Journal of Economic Literature 27 1067 1097. Coughlan, P., & Plott, C. ( 1997 ). An experimental analysis of the structure of legal fees: Ame rican rule versus English rule. California Institute of Technology, Division of Humanities and Social Sciences Social Science Working Paper No. 1025 Danzon, P. M., & Lillard, L. A. ( 1983 ). Settlment out of Court: the Disposition of M edical Malpractice Claims. The Journal of Legal Studies 12 345 377. Doucet, J., & Littlechild, S. C. ( 2006 ). Negotiated settlements: The development of legal and economic thinking. Utilities Policy 14 266 277. Doucet, J., & Littlechild, S. C. ( 2009 ). Negotiated Settlements and the N ational Energy B oard in Canada. Energy Policy 37 4633 4644.
104 Farber, H. S., & White, M. J. ( 1991 ). Medical malpractice: an empirical examina tion of the litigation process. RAND Journal of Economics 22 199 217. Fershtman, C. ( 2000 ). A Note on Multi issue Two sided Ba rgai ning: Bilateral Procedures. Games and Economic Behavior 30 216 227. Greene, W. ( 1998 ) New York: Econometric Soft ware, Inc. Harstad, B. ( 2010 ). Strategic delegation and voting rules. Journal of Public Economics 94 102 113. Holburn, G. L. F., & Spiller P. T. ( 2002 ). Interest group representations in administrative: The impact of consumer advocates and elected commissioners on regulato ry policy in the United States. University of California at Berkeley POWER Workin g Paper 002 Hughes, J. W., & Snyder, E. A. ( 1990 ). The English Rule for Allocating Legal Co sts: Evidence confronts Theory. Journal of Law, Economics and Organization 6 345 380. Hughes, J. W., & Snyder, E. A. ( 1995 ) Litigation and Settlement under the E nglish and Ameri can Rules: Theory and Evidence. Journal of Law and Economics 38 225 250. Inderst, R. ( 2002 ). Multi issue bar gaining with Endogenous Agenda. Games and Economic Behavior 30 64 82. Jones, S. R. G. ( 1989 ). Have your lawyer call my lawyer: B ilateral Deleg ation in Bargaining Situations. Journal of Economic Behavior and Organization 11 159 174. Joskow, P. L. ( 1972 ). The Determination of the Allowed Rate of Return in a formal Regulatory Hearing. The Bell Journal of Economics and Management Science 3 632 644. Joskow, P. L. ( 1974 ). Inflation and Environmental Concern: Structural Change in the Process of P ublic Utility Price Regulation. Journal of Law and Economics 17 291 327. Katz, M. L. (1991). Game playing agents: unobservab le contrac ts as precommitments. RAND Journal of Economics 22 307 328. Kessler, D. ( 1995 ). Fault, settlement and negligence law. Rand Journal of Economics 26 296 313.
10 5 Klumpp, T. ( 2009 ). Strategic Voting and Conservatism in L egislative Elections. Mimeo. Emory Uni versity. Krieger, S. H. ( 1995 ). Problems for Captive Ratepayers in Nonunanimous Settlemen t of Public Utility Rate Cases. Yale Journal on Regulation 12 257 343. Lammers, F. ( 2010 ). Fa irness in Delegated Bargaining. Journal of Economics and Management Str ategy 19 169 183. Littlechild, S. C. (2009a). Stipulated settlements, the consumer advocate and utility regulation in Florida. Journal of Regulatory Economics 35 96 109. Littlechild, S. C. ( 2009b ). The bird in hand: Stipulated settlements in the Florida electricity sector. Utilities Policy 17 276 287. Maddala, G. S. ( 1999 ) Limited dependent and qualitative variables in econometrics Cambridge: Cambridge University Press. Mnookin, R. H. ( 1993 ). Why Negotiations Fail: An Exploration of Barr ier s to Resolution of Conflict. The Ohio State Journal on Dispute Resolution 8 235 249. Noll, R. G. ( 1971 ) Reforming Regulation Washington, DC: Brookings Institution. Osborne, M. J., & Rubinstein, A. ( 1990 ) Bargaining and Marke ts Electronic version. Ac ademic Press Inc. Persson, T., & Tabellini, G. ( 1994 ). Representative democracy and capital taxation. Journal of Public Economics 55 53 70. Poirier, D. J. ( 1980 ). Partial Observabil ity in Bivariate Probit Models. Journal of Econometrics 12 210 217. Political Analysis 10: 101 112. Reinganum, J. F., & Wilde, L. L. ( 1986 ). Settlement, litigation and the allocation of litigation costs. RAND Journal of Economics 17 557 566. Rivers, D., & Vuong, Q. H. ( 1988 ). Limited Information Estimators and Exogeneity Tests for Simultaneous Probit Models. Journal of Econometrics 39 347 366. Rogoff, K. ( 1985 ). The Optimal Degree of Commitment to an Intermediate Monet ary Target. The Q uarterly Journal of Economics 100 1169 1189. Rubinstein, A. ( 1982 ). Perfect Equ ilibrium in a Bargaining Model. Econometrica 50 97 108.
106 Schelling, T. C. (1956). An Essay on Bargaining. The American Economic Review 46 281 306. Segendorff, B. ( 1998 ). Deleg ation and Threat in Bargaining. Games and Economic Behavior 23 266 283. Spulber, D. F. ( 1989 ) Regulation and Markets. Cambridge: MIT Press. Sliwka, D. ( 2007 ). Trust as a Signal of a Social Norm and the Hidden Costs of Incentive Schemes. The American Economic Review 97 999 1012. Viscusi, W. K. ( 1986 ). The Determinants of the Disposition of Product Liability Claims and Compensation for Bodily Injury. The Journal of Legal Studies 15 321 346. Wang, Z. ( 2004 ). Settling Utility Rate Cases: An Al ternative Ratemaking Procedure. Journal of Regulatory Economics 26 141 163. White, M. J. ( 1989 ). An empirical test of the comparative and contributory negl igence rules in accident rules. RAND Journal of Economics 20 308 330.
107 BIOGRAPHICAL SKETCH Shourjo Chakravorty was born in 1983 in Kolkata, India. He received a Bachelor of Arts in economics in Augus t 2005 and a Master of Arts in e conomics in August 2007 from Jadavpur University, Kolkata. He took a First in both his BA and MA examinations. He st arted graduate school at the University of Florida in F all 2007 and received a PhD in e conomics in Summer 2012. His research interests include industrial organization, applied econometrics, game t heory and b argaining.