I. INTRODUCTION Under the long-established but unwritten "regulatory compact,"2 a regulated firm agrees that the prices it charges will be set by regulators, and regulators agree that the prices they set will allow the firm to recoup its operating costs plus a reasonable profit. For a regulated firm, "reasonable profit" is defined as the rate of return that is sufficient to attract the capital the firm needs to continue to meet its obligations. Regulators rely on the regulated firm's overall cost of capital to estimate such a rate of return.3

There are two main components to any firm's overall cost of capital: the cost of debt and the cost of equity. The cost of debt generally can be directly measured, but the cost of equity cannot. As a consequence, determining an appropriate return on equity and an overall fair allowed rate of return for a regulated firm is one of the oldest issues in rate regulation.

Beginning in the 1890s, state regulators relied on the "fair value" of a regulated firm's assets to determine the rate of return. This approach culminated in the U.S. Supreme Court decision in Smyth v. Ames,4 and came to be known as the "Fair Value" Doctrine. The Fair Value Doctrine did not last long; it collapsed under its inherent circularity - the value of a regulated firm was whatever regulators said it was.5 A decade later, in Consolidated Gas,6 the Court began to discuss the relationship between risk and return directly; it reasoned that a fair rate of return encompassed a return on invested capital and a return for risk.7

By 1923, in its Bluefleld decision,8 the Court had begun to zero in on the idea of comparable risk, stating as follows:

A public utility is entitled to such rates as will permit it to earn a return on the value of the property which it employs for the convenience of the public equal to that generally being made at the same time . . . in other business undertakings which are attended by corresponding risks, and uncertainties;. . . . The return should be reasonably sufficient to assure confidence in the financial soundness of the utility and should be adequate ... to maintain and support its credit and enable it to raise money necessary for the proper discharge of its public duties.

Two decades later, the principle of basing a regulated utility's return on the financial risks of other comparable firms was firmly established in the Court's 1944 Hope decision, in which die Court stated as follows:

"[T]he return to the equity owner should be commensurate with returns on investments in other enterprises having corresponding risks. That return, moreover, should be sufficient to assure confidence in the financial integrity of the enterprise, so as to maintain its credit and to attract capital."10

For the last seven decades, the quoted language from Hope, along with the comparable risk standard (once called "comparative" risk) has been the sine qua non for determining whether regulated rates of return set by federal regulators, such as the FERC, and state utility regulators are just and reasonable. Yet, the ways that regulators determine which firms have comparable risk remain quite arbitrary. Perhaps this is inevitable. Every firm is unique, and, therefore, it is impossible to say that any two firms face identical financial and business risks.

Since 2000, financial risk in regulated electric utilities, transmission owners, and generators has been affected by volatile fuel markets and regulatory uncertainty that has affected the industry as it has evolved. For example, the restructuring efforts of the 1990s continue in some segments of the industry (such as transmission markets), while a move towards re-regulation appears to be underway in the generation market.

The task of establishing separate rates of return for transmission, generation, and distribution functions has also been complicated by restructuring, generation mandates for renewable resources, new FERC incentives for transmission investment,11 and the risks associated wim looming greenhouse gas regulations. These changes, along with die upheaval in global financial markets, have made it more difficult to identify risk-comparable firms that can be used to determine "zones of reasonableness" within which allowed returns can be set. As a result, the criteria traditionally used by the FERC and state utility regulators to establish risk comparability may no longer be relevant or adequate. Consequentially, allowed rates of return may be straying ever further away from satisfying the Court's mandate that returns must be sufficient to maintain financial integrity and attract capital.12

This article first examines how the FERCs policy toward defining comparable risk for the firms it regulates has changed over time. As we discuss, die general approach taken by the FERC in defining comparability has not been consistent. It is reasonable to expect mat the major factors influencing me financial risks faced by regulated firms might change over time, but, even given this inevitability, the FERCs approach to defining comparability has proven to be unnecessarily arbitrary. In a 1987 Notice of Proposed Rulemaking (NOPR), the Federal Communications Commission (FCC) recognized that the common "screening" approach used to determine groups of risk-comparable firms (the approach most used today by both the FERC and state utility regulators) fails to allow for any substitution across various risk measures.13 To avoid the weaknesses of the direct screening approach, the FCC NOPR recommended using a statistical technique called cluster analysis to identify firms of comparable risk. Oddly, the FCC ultimately did not adopt the approach, although it has been used since in several instances in both telecommunication and electric rate cases. In this article, we review this technique and demonstrate several examples of its application. Although cluster analysis is not a panacea for selecting risk-comparable firm - that process will always include some degree of subjectivity - we believe cluster analysis aids in the selection of the most comparable firms and, thus, better serves the Court's requirements set out in Bluefleld and Hope.

II. The FERC's Evolving Approach to Comparability

The Bluefleld and Hope decisions require all state and federal regulators to ensure that the companies they regulate earn rates of return sufficient to continue their operations and attract capital, while at the same time guaranteeing that the rate is just and reasonable to ratepayers.14 Typically, rate of return is determined on a case-by-case basis through administrative procedures.15 In these "rate cases," regulators rely on evidence presented by various parties to determine the allowed return on equity (ROE) as well as an appropriate capital structure.16

Economists define the required rate of return on a particular investment as the return that investors forego by making that investment instead of an alternative investment of equal risk. This is known as the opportunity cost of capital. Although the cost of debt is easily observable, the required return on equity is not. Moreover, in many cases, regulated firms are not publicly traded, either because they are privately held or are subsidiaries of parent companies which may or may not be publicly traded. Thus, in setting an allowed rate of return on invested capital that meets the requirements established by the Bluefleld and Hope decisions, regulators like the FERC must: (1) identify riskcomparable firms, (2) determine an appropriate capital structure for the regulated firm, and (3) apply one or more analytical methodologies to estimate an appropriate allowed ROE.17 This article focuses on step (l)- identifying riskcomparable firms. The group of risk-comparable firms (the "proxy group") forms the basis from which the FERC determines an allowed return for the firm under investigation. This section describes how the FERC has historically defined comparable risk and, when used, how the FERCs various definitions of comparable risk were translated into screening parameters for establishing proxy groups.

A. Early Regulation and the Absence of Proxy Groups

The FERCs approach for determining allowed rates of return has changed repeatedly over time. Prior to passage of the Energy Policy Act of 1992,18 the primary regulatory focus of the FERC was setting allowed rates of return for interstate natural gas and oil pipelines. The Energy Policy Act of 1992 created a broad class of exempt wholesale generators, established open access to highvoltage transmission systems, and set in motion the restructuring of the electric industry.19 After 1992, the FERCs approach to rate regulation necessarily began to change.

At first, the FERC did not even require the use of proxy groups to determine comparable risk. It instead focused on a single company or, where die regulated entity was a subsidiary, the return of the parent company. Until the early 1990s, many of the FERCs determinations of allowed returns focused on me relative risk of a wholly owned subsidiary relative to its parent.20 For example, in Williston Basin the FERC relied on a stand-alone DCF analysis of the company's parent, MDU Resources, Inc., and then adjusted the resulting value downwards based on Staffs analysis using the Capital Asset Pricing Model.21 When proxy groups were used, they were broadly defined and were employed more as guidelines, rather than as a means to determine zones of reasonableness for a given regulated entity's allowed return. For example, in Tennessee Pipeline the FERC relied on a hodgepodge of recommendations, none of which was based on a well-defined proxy group.22 In that case, the company's witnesses presented a risk-premium analysis tied to (1) the pipeline's parent company, Tenneco; (2) "the earned returns of the top 25% of unregulated industrial companies"; and (3) a stand-alone discounted cash flow (DCF) study applied to Tenneco itself.23 Oddly enough, in its decision the FERC stressed the importance of determining the company's risk relative to its parent Tenneco, stating as follows:

Having concluded that Tennessee is lower risk than Tenneco and so has a lower cost of equity, we must determine how much lower Tennessee's cost of equity is than Tenneco's. Unfortunately the record evidence concerning the degree (as distinguished from direction) of the difference in their risks and costs of equity is somewhat limited. The [FERC] has consistently required that rate of return be set based on the risks of the regulated entity [rather than that of the parent.]

Thus, in establishing an allowed return for the pipeline, the FERC did not rely on any specific risk criteria or develop a zone of reasonableness by contemporaneously estimating returns for other pipeline companies. Instead, based on the return derived from an analysis of Tenneco,25 the FERC next evaluated Tennessee Pipeline's risk relative to other pipelines and the past allowed returns the FERC had granted to those pipelines. The FERC also took into account, although in no discernable analytical way, changes in interest rates since those prior decisions.

The FERC ultimately set a zone of reasonableness between 15.0% and 16.9%.26 The low value was Staffs recommendation. The high value equaled the discounted-cash flow analysis performed for Tennessee's parent. The FERC then determined that the pipeline "is of about average risk" and set an allowed return at 15.95%, the midpoint of the zone.27

The FERC used this same approach in other proceedings, either to determine the subject company's return directly or to determine upper and lower bounds of the zone of reasonableness. In the 1980 case, Arkansas Louisiana Gas Company, the FERC relied on the highest return on equity that it had previously granted and then adjusted that value downwards by fifty basis points to determine the upper end of the zone of reasonableness.28 Arkansas Louisiana Gas Company appealed the FERCs Final Order, asserting that the FERC ignored issues of comparable risk when it used a historical benchmark to establish the upper limit of the zone of reasonableness.29 The Court of Appeals affirmed the FERCs right to rely on historical returns, provided they were "recent," and stated as follows:

While we are concerned that continuation by the [FERC] for too prolonged a time in a practice of relying primarily on allowed returns might deprive a rate of return decision of a rational relationship with market realities, we do not believe that problem exists here where the cases relied upon are recent and where those cases themselves established their allowed [return] directly on the basis of evidence of market realities.30

Throughout the 1980s, the FERC continued to rely on benchmarks as its method for setting allowed returns in both natural gas and electric cases. In 1984, the FERC issued regulations that established generic benchmarks for electric utility ROEs that would to be amended every year.31 The FERC established a generic rate of return benchmark to (1) introduce consistency, (2) increase FERC input into the return-setting process, and (3) reduce uncertainty and costly litigation for electric utilities.32 The FERC stressed its belief that the benchmarking would reduce uncertainty: "[t]he generic procedure adopted here should reduce uncertainty, since the advisory and later the presumptive rate of return will be known to all parties by the time each [rate case] proceeding begins."33

In an abrupt reversal, just seven years after it formally established the procedure, the FERC abolished the generic benchmark approach.34 The benchmarks had been adopted to reduce regulatory uncertainty, but over a period of a few years, it became clear that the approach was ineffectual.35 The FERC reversed its position on generic benchmarks for electric utilities and publicly stated its reasons for doing so (i.e., that since returns were still being determined on a case-by-case basis, the generic benchmarks had not achieved the FERCs goals) as follows: "[t]he benchmark has not reduced parties' uncertainty in rate cases as to what will be the [FERC]' s ultimate determination. Thus, hopes of conserving resources and of enhanced certainty have not been fulfilled."36

B. The 1990s through Today: Changing Requirements for Proxy Groups

Benchmarking, while convenient, departed from the degree of comparability required under Hope and Bluefleld. Thus, by the mid- 1990s, the FERC preferred rate of return recommendations that were supported by the use of well-defined proxy groups of comparable firms. As discussed below, however, the requirements underlying establishment of those proxy groups changed significantly over time. Moreover, the FERC imposed far different requirements for establishing proxy groups of interstate natural gas (and oil) pipelines than it did for establishing proxy groups of regulated electric companies and interstate transmission companies.37

1. Natural Gas and Oil Pipelines.38

The FERCs treatment of proxy groups in natural gas pipeline rate cases has changed over time because of the consolidation and restructuring of the pipeline industry that has occurred over the years. The most recent change occurred in April 2008, when the FERC issued its Policy Statement, "Composition of Proxy Groups for Determining Gas and Oil Pipeline Return on Equity."39 This Policy Statement signified a methodological break from well-established precedent that had been used in connection with setting natural gas and oil pipeline proxy groups: (1) a requirement that natural gas or oil transportation constitute a significant portion of the proxy group's business (usually fifty percent) and (2) the exclusion of Master Limited Partnerships (MLPs) from proxy groups.40 The FERC explained that:

Historically, in determining the proxy group, the [FERC] required that pipeline operations constitute a high proportion of the business of any firm included in the proxy group. However, in recent years, there have been fewer gas pipeline corporations that meet that standard, in part because of the greater trend toward Master Limited Partnerships (MLPs) in the gas pipeline industry. Additionally, there are no oil corporations available for use in the oil pipeline proxy group. These trends have made the MLP issue one of particular concern to the [FERC] and are the reason that the [FERC] issued the Proposed Policy.

2. Pipeline Operations Requirement

Historically, the FERC required that all gas pipeline proxy group members: (1) have publicly traded stock, (2) be recognized as a natural gas or oil pipeline company and be tracked by one or more investment information services, and (3) have pipeline operations that constitute at least fifty percent of its business (Pipeline Operations Requirement).42 The latter would be satisfied if pipeline operations accounted for at least fifty percent of a company's assets or operating income.43 Both the natural gas and oil pipeline industries consolidated over time through mergers and the establishment of MLPs and, as a result, fewer and fewer firms satisfied the FERCs three-prong proxy group parameters. The FERCs 2008 Policy Statement eliminated the restrictive Pipeline Operations Requirement.44

The Pipeline Operations Requirement historically excluded natural gas distribution companies (e.g., LDCs) from pipeline proxy groups because LDCs were considered to have lower risk than gas pipeline companies.45 Given multiple mergers, very few publicly traded corporations, four in 2008 to be exact, were primarily engaged in the transportation of natural gas, as many firms had significant natural gas distribution operations. The number of gas pipeline firms satisfying the three-prong proxy group parameters shrunk further as many became MLPs and, merefore, were ineligible to become proxy group members for determining a given pipeline's allowed rate of return.46 As the number of eligible companies mat satisfied the three-prong proxy group parameters shrank, the FERC was forced to revise the parameters, first in a rate proceeding and later with the 2008 Policy Statement.

The FERC relaxed me Pipeline Operations Requirement for the first time in a rate case involving Williston Basin Interstate Pipeline Company when it expanded the proxy group to include companies with gas pipeline segments that constituted less than fifty percent of their total operations. 7 The FERC deemed die change necessary because only three companies at me time satisfied the historical three-prong proxy group parameters, and the FERC determined that a tiiree-member proxy group was not sufficient to determine a just and reasonable ROE.48 In pipeline rate cases immediately prior to Williston Basin II, the FERC had used a four-company proxy group comprised of: Coastal Corporation, El Paso Energy Corporation, Enron Corporation, and Williams Companies, Inc.