OPINION
Opinion by
Justice PEMBERTON.This case presents three sets of issues arising from Texas’s transition from a wholly regulated retail electricity market. First, we will consider the extent to which the Public Utility Commission had power to order electric utilities to refund alleged “over-mitigation” of their stranded costs, as determined from interim computer models, before the final 2004 true-up proceedings. Second, we will determine whether substantial evidence supports the Commission’s characterization of Nuclear Electric Insurance Limited (NEIL) account balances as generation-related rather than transmission-related. Third, we will address whether the Commission may set demand charges for large commercial customers greater than those it set before deregulation. Because we determine that the Commission exceeded its statutory authority in ordering refunds of “over-mitigated” stranded costs determined before the 2004 true-ups, we will reverse the portion of the district court’s judgment compelling such refunds and remand to the Commission for further proceedings. However, we will affirm the district court’s judgment affirming the Commission’s disposition of the issues concerning NEIL member accounts and demand charges.
GENERAL BACKGROUND
Finding that “the production and sale of electricity is not a monopoly warranting regulation of rates, operations, and services and that the public interest in competitive electric markets requires that, except for transmission and distribution services and for the recovery of stranded costs, electric services and their prices should be determined by customer choices and the normal forces of competition,” in 1999 the legislature enacted comprehensive legislation — commonly known by its bill number, S.B. 7 — providing for an ordered transition from Texas’s former wholly regulated electricity market to a more competitive retail electricity market. See Act of May 27, 1999, 76th Leg., R.S., ch. 405, 1999 Tex. Gen. Laws 2543, 2543-2625 (codified at Tex. Util.Code Ann. §§ 39.001-.910 (West Supp.2004-05)); Tex. UtiLCode Ann. § 39.001(a); In re TXU Elec. Co., 67 S.W.3d 130, 132 (Tex. 2001) (Phillips, C.J., concurring). In several of our prior opinions, we have described the basic steps in this transition. See, e.g., Reliant Energy, Inc. v. Public Util. Comm’n, 101 S.W.3d 129, 133-36 (Tex.App.-Austin 2003), rev’d in part sub nom CenterPoint Energy, Inc. v. Public Util. Comm’n, 143 S.W.3d 81 (Tex.2004); Reliant Energy, Inc. v. Public Util. Comm’n, 62 S.W.3d 833, 835-36 (Tex.App.-Austin 2001, no pet.). Under the former regulatory regime, each region of the state was served by a single vertically integrated utility that generated electricity, built and maintained the electricity distribution “wires” or grid, and sold the electricity to consumers at retail, all under the comprehensive regulation of the Public Utility Commission (Commission). Under S.B. 7, these utilities were required to “unbundle” themselves into three separate entities — a power generation company, a transmission and distribution utility, and a retail electric provider. Tex. Util.Code Ann. § 39.051(b). Power generation companies provide wholesale *685generation services in competition with other generators entering the market. In re TXU, 67 S.W.3d at 132 (Phillips, C.J., concurring). Retail electric providers (REPs) provide retail electric service to end-use customers in competition with other REPs. Id. Transmission and distribution utilities (TDUs) own and maintain the “wires” used to transport electricity from the power generation companies to all REPs and retail consumers in the utility’s geographic service area. Id. Because the legislature continued to regard TDU’s as monopolies within their respective service areas, their rates continued to be regulated by the Commission. See Tex. UtiLCode Ann. § 39.001(a), (b). A utility could “unbundle” through the creation either of separate unaffiliated companies or of separate affiliated companies owned by a common holding company (“affiliated companies” or “unbundled” companies), or through the sale of assets. Id. § 39.051(c).
Other aspects of the legislatively-mandated transition to a more competitive electricity market gave rise to the issues in this appeal. We explore each of these aspects below with its corresponding issues.
STRANDED COSTS
AEP Texas Central Company (AEP) brings three issues on appeal concerning the Commission’s order regarding stranded costs. We will first review the nature of stranded costs and the Commission’s decision to order credits to refund “over-mitigation” before the 2004 true-up. We will then turn to the specifics of AEP’s issues.
Nature of stranded costs
Although “stranded costs” have a precise, technical definition under chapter 39 of the utilities code, id. § 39.251(7), the supreme court has generally described them “as the portion of the book value of a utility’s generation assets that is projected to be unrecovered through rates that are based on market prices.” In re TXU, 67 S.W.3d at 132 (Phillips, C.J., concurring) (quoting City of Corpus Christi v. Public Util. Comm’n of Tex., 51 S.W.3d 231, 238-39 (Tex.2001)). The largest part of stranded costs are attributable to investments in nuclear power plants. See id.
Stranded costs are a potential byproduct of Texas’s transition from the former rate-regulated electricity system to competition. Under the former system, the Commission could set rates that would enable utilities to recover from consumers the costs of their generation-related assets. Utilities accordingly made considerable investments in generation-related assets with the expectation of being able to recover the costs of these investments and a reasonable return. See CenterPoint Energy, Inc. v. Public Util. Comm’n of Tex., 143 S.W.3d 81, 82 (Tex.2004). Theoretically, the existence of these costs would, upon the beginning of competition, create significant competitive disadvantages for incumbent utilities relative to new market entrants. Because the new market entrants would not have these embedded generation-related costs and opportunity cost reflected in the rate of return, their pricing structure would tend to be lower than those of incumbent utilities. This, in turn, would enable new market entrants to price electricity below a level at which incumbent utilities could recover their investments. See id. Hence, incumbent utilities would either have to charge uncompetitive higher rates or simply absorb these “stranded costs.” See id. at 82-83.2
*686The legislature thus gave careful attention to the issue of stranded costs when considering deregulation of the electricity market. The April 1998 Report to the Texas Senate Interim Committee on Electric Utility Restructuring contained an estimate of projected potential stranded costs, described as “excess cost over market,” or “ECOM,” for nine Texas incumbent utilities as of December 31, 2001, the last day before retail competition would begin. These “1998 ECOM Report” estimates were derived from computer models that took account of factors, such as the cost of fuel used to power generating plants, that would impact the market value of generating assets.
The legislature determined that, among its other foundational findings regarding electricity deregulation, it is in the public interest to “allow utilities with uneconomic generation-related assets and purchased power contracts to recover the reasonable excess costs over market of those assets and purchase power contracts.” Tex. Util. Code Ann. § 39.001(b)(2). It established a three-phase regulatory program intended to assist incumbent utilities in recovering or eliminating what otherwise would have been stranded costs in the competitive market. In re TXU, 67 S.W.3d at 132 (Phillips, C.J., concurring).
Under the first phase, which ended on December 31, 2001, the Commission froze retail electric rates (“freeze period”). Tex. Util.Code Ann. § 39.052; In re TXU, 67 S.W.3d at 133 (Phillips, C.J., concurring). Utilities that had been identified as having potential stranded costs in the 1998 ECOM Report were allowed to “mitigate” them by (1) shifting depreciation from the transmission and delivery assets to the generating assets, Tex. Util.Code Ann. § 39.256 (West Supp.2004-05), and (2) accelerating the cost recovery of stranded costs each year through the use of legislatively-approved “tools.” Id. § 39.254 (West Supp.2004-05); see also id. §§ 39.251-.265. Among the tools offered, the legislature set means for computing during the rate-freeze period positive annual revenues, annual costs, and invested capital. Id. §§ 39.257-259.
Under the second phase, from January 1, 2002, to December 31, 2003, the Commission was to determine whether any stranded costs remained to be recovered by entering updated data into the ECOM model. Id. § 39.201(a), (b)(3), (g), (h); In re TXU, 67 S.W.3d at 133 (Phillips, C.J., concurring). Based on these calculations, the Commission was authorized to consider any remaining stranded costs in setting the “competition transition charge” or “CTC.” Tex. Util.Code Ann. § 39.201(b)(3). The CTC is intended to cover a utility’s stranded costs through collection from ev*687ery customer taking power over the utility’s transmission and delivery system, thus making up the difference between a generating plant’s book value and its market value. In re TXU, 67 S.W.3d at 133 (Phillips, C.J., concurring). At the same time, the legislature set the rates each affiliated REP was allowed to charge residential and small commercial customers3 at six percent less than the rate charged on January 1, 1999. Tex. UtiLCode Ann. § 39.202(a) (West Supp.2004-05) (“price to beat”). Rates of competing unaffiliated REPs and of affiliated REPs for large commercial and industrial customers were not subject to the price to beat. See id. The legislature required the utilities to file their proposed tariffs with the Commission by April 1, 2000 (during the freeze period). Id. § 39.201(a). They were also to supply supporting cost data for determining “non-bypassable delivery charges,” including data establishing estimates of stranded costs “that are reasonably projected to exist on the last day of the freeze period.” Id. § 39.201(b), (g), (h).
Under the final phase, stranded costs are to be calculated in “true-up” proceedings beginning January 2004. Values generated at a “true-up” will emerge from market valuations of a utility’s generation assets, based on stock prices and anticipated income streams in a competitive market, as determined by updating the 1998 ECOM model. Id. §§ 39.201(i), 39.262(h), (i). If stranded costs remain, the Commission can extend the CTC collection period or increase the charge. Id. §§ 39.201(i), .262(c). At the utility’s option, it may sec-uritize4 any or all of the stranded costs. Id. § 39.262(c). Conversely, if the Commission finds in the true-up proceeding that the competition transition charge is larger than is needed to recover any remaining stranded costs, the commission may reduce the competition transition charge, reverse, in whole or in part, the depreciation expense, reduce the transmission and distribution utility’s rates; or implement a combination of these efforts. Id. § 39.201(¿).
Generic unbundled cost-of-service docket
In March 2000, the nine incumbent electric utilities in Texas, including Central Power and Light Company (the unbundled utility that owned the TDU that ultimately became AEP), filed applications with the Commission proposing rates based on a 2002 test year, and the Commission instituted separate contested case proceedings for each. See 16 Tex. Admin. Code § 25.344(d) (2005); In re TXU, 67 S.W.3d at 133 (Phillips, C.J., concurring). Because the nine dockets shared many of the same legal and policy issues concerning stranded costs, among other issues not now on appeal, the Commission concluded that a supplemental generic proceeding would be the most efficient method for resolving these common issues. Common issues resolved in the generic docket were then applied in each individual docket. The cases are generally referred to as the “unbundled cost of service” or UCOS cases.
The Commission segmented each individual docket into four phases. In Phase I, the Commission conducted a hearing on *688the business separation plan through which the utility proposed to divide itself into a power generation company, a transmission and delivery company, and an affiliated retail electric provider. In Phase II, the Commission conducted a hearing to project the amount of the utility’s stranded costs when retail competition began on January 1, 2002. See Tex. UtibCode Ann. § 39.201(g). In Phases III and IV, the Commission conducted hearings to determine the actual rates the transmission and delivery company could charge REPs. In re TXU, 67 S.W.3d at 133 (Phillips, C.J., concurring).
Identified in the 1998 ECOM Report as one of the utilities likely to have stranded costs, AEP had implemented procedures to mitigate its stranded costs by reducing the book value of its assets by the amount of its excess earnings. As a result of evidentiary hearings and the input of updated data into the ECOM model, in October 2001 the Commission revised its stranded-cost estimate for AEP to be negative $615,066 million.5 In other words, the Commission determined that the continuation of AEP’s mitigation efforts would result in an over-mitigation of $615,066 million by the time of the 2004 true-up. Because AEP had been utilizing mitigation tools from 1999 until 2001, the Commission applied its new data to AEP’s earnings report and determined that AEP had recovered actual excess earnings in the amount of $54,789 million by 2001.
Based on its determination from the 2001 interim ECOM calculations that several utilities had over-mitigated stranded costs, the Commission set a generic docket to decide the question of its authority to act with respect to the excess mitigation earnings. The Commission acknowledged that chapter 39 of the utilities code does not describe any method for addressing over-recovery of stranded costs before the 2004 true-ups. However, it determined that it could order utilities to refund the over-recovery of stranded costs it had determined through the interim ECOM estimates, relying on language in the first sentence of the section governing the 2004 true-up that “[a]n electric utility ... may not be permitted to overrecover stranded costs through the procedures established by this section or through the application of the measures provided by the other sections of this chapter.”6 See Tex. Util. *689Code Ann. § 39.262(a). In AEP’s individual case, the Commission then ordered AEP to refund those amounts through a credit (“excess mitigation credit” or “over-mitigation credit”) in transmission and distribution rates to the REPs, amortized over five years. It also decided that, if AEP is found at the 2004 true-up to have stranded costs after having refunded “over-mitigated” amounts, AEP could not recover interest on the over-refunded amounts.
AEP appealed the Commission’s orders to the district court, claiming that the Commission lacked statutory authority to halt mitigation or to order a refund of over-mitigation amounts and that it would be entitled to interest on any amount determined to be over-refunded at its 2004 true-up. The Cities and the Office of the Public Utility Council (OPC) also appealed, claiming that the refund of a TDU’s over-mitigation properly ought to be paid to the end-use residential and small commercial consumers, not REPs. The district court ruled that chapter 39 requires over-mitigation credits to be paid directly to the consumer rather than the REP and affirmed the Commission’s orders regarding all other issues. AEP now appeals the district court’s judgment resulting from the Commission’s stranded cost order.
Discussion
AEP brings three issues on appeal. It argues first that the Commission exceeded its statutory authority in requiring AEP to refund stranded cost amounts that the Commission had determined, based on the 2001 ECOM calculations, to have been over-recovered. AEP next argues that the district court erred in requiring over-mitigation credits to be paid to end-use consumers rather than to the REPs. Third, it asserts that the Commission violated chapter 39 in ordering that AEP would not be entitled to interest on any amount it had over-refunded. We agree with AEP that the Commission lacked authority to require a refund of amounts calculated in interim ECOM estimates to have been over-mitigated. As explained below, we need not reach AEP’s second or third issues in light of this disposition.7
Standard of review
The powers of the Commission include the powers delegated by the legislature in clear and express statutory language, together with any implied powers that may be necessary to perform a function or duty delegated by the legislature. GTE Southwest, Inc. v. Public Util. Comm’n, 10 S.W.3d 7, 12 (Tex.App.-Austin 1999, no pet.). We may imply that the legislature intended that an agency would have whatever power would be reasonably necessary to fulfill a function or perform a duty that the legislature has expressly placed in the agency. Id.; see also Kawasaki Motors Corp. U.S.A. v. Texas Motor Vehicle Comm’n, 855 S.W.2d 792, 797 (Tex.App.-Austin 1993, no writ); Texas Dep’t of Human Servs. v. Christian Care Ctrs., Inc., 826 S.W.2d 715, 719 (Tex.App.-Austin 1992, writ denied). However, even if the *690legislature intends that an agency created to centralize expertise in a certain regulatory area “be given a large degree of latitude in the methods it uses to accomplish its regulatory function,” Texas Mun. Power Agency v. Public Util. Comm’n, 150 S.W.3d 579, 586 (Tex.App.-Austin 2004, pet. granted), an agency may not, in the guise of implied powers, exercise what is effectively a new power, or a power contrary to a statute, on the theory that such exercise is expedient for the agency’s purpose, City of Austin v. Southwestern Bell Tel. Co., 92 S.W.3d 434, 441 (Tex.2002), nor may it contravene specific statutory language, run counter to the general objectives of the statute, or impose additional burdens, conditions, or restrictions in excess of or inconsistent with the relevant statutory provisions. State v. Public Util. Comm’n, 131 S.W.3d 314, 321 (Tex.App.Austin 2004, pet. denied).
To determine the scope of the Commission’s powers in this case, we must construe the relevant provisions of chapter 39 of the utilities code. Statutory construction is a question of law, which we review de novo. In re Forlenza, 140 S.W.3d 373, 376 (Tex.2004); McIntyre v. Ramirez, 109 S.W.3d 741, 745 (Tex.2003). "When interpreting a statutory provision, we must ascertain and effectuate legislative intent. Tex. Dep’t of Protective & Regulatory Servs. v. Mega Child Care, Inc., 145 S.W.3d 170, 176 (Tex.2004). In ascertaining legislative intent, we may consider the evil sought to be remedied, the legislative history, and the consequences of a particular construction. See Liberty Mut. Ins. Co. v. Garrison Contractors, Inc., 966 S.W.2d 482, 484 (Tex.1998). Further, we read every word, phrase, and expression in a statute as if it were deliberately chosen and presume the words excluded from the statute are done so purposefully. See Gables Realty Ltd. P’ship v. Travis Cent. Appraisal Dist., 81 S.W.3d 869, 873 (Tex.App.-Austin 2002, pet. denied); City of Austin v. Quick, 930 S.W.2d 678, 687 (Tex.App.-Austin 1996) (citing Cameron v. Terrell & Garrett, Inc., 618 S.W.2d 535, 540 (Tex.1981)), aff'd, Quick v. City of Austin, 7 S.W.3d 109 (Tex.1999); see also 2A Norman J. Singer, Sutherland Statutory Construction § 47.25 (6th ed.2000). In determining the scope of the Commission’s authority, we must read PURA as a whole to discover the underlying legislative intent. State v. Public Util. Comm’n, 883 S.W.2d 190, 196 (Tex.1994); Texas Building Owners & Managers Ass’n v. Public Util. Comm’n, 110 S.W.3d 524, 532-33 (Tex.App.-Austin 2003, pet. denied). We give weight to how the Commission interprets its own powers, but only if that interpretation is reasonable and not inconsistent with the statute. Southwestern Bell, 92 S.W.3d at 441-42; City of Austin v. Hyde Park Baptist Church, 152 S.W.3d 162, 166 (Tex.App.-Austin 2004, no pet.).
Commission power to order refunds of stranded cost over-recovery
In AEP’s first issue, as in In re TXU, the question is not whether stranded costs may be over-recovered. See 67 S.W.3d at 151 (Hecht, J., dissenting). It is clear that, at least at the 2004 true-up, “[utilities that are finally determined to have stranded costs will be entitled to recover only those costs and no more.” Id.; see also Tex. UtiLCode Ann. § 39.262(a). Rather, the question presented here is whether the Commission can intervene in a utility’s ongoing stranded cost mitigation before the 2004 true-up and compel refunds based on interim estimates of stranded costs. AEP argues that the Commission must wait until it makes its final determination of AEP’s stranded costs in the 2004 true-up proceedings before it can reconcile “actual values” of *691stranded costs with its mitigation efforts derived from the 1998 ECOM estimates. In response, the Commission argues that the legislature’s prohibition on over-recovery of stranded costs — which appears only in the section governing the 2004 true-up — confers implied authority for it to adjust, limit, or reverse utilities’ mitigation efforts before the 2004 true-up, at which point it would finally reconcile any over- or under-mitigation as determined in that proceeding. Our analysis of the text and structure of the relevant statutes compels us to agree with AEP.
To effectuate its policy to allow utilities to recover their stranded costs, the legislature established what the supreme court has described as a “comprehensive scheme” for stranded cost recovery. See Tex. UtiLCode Ann. § 39.201(b)(2); CenterPoint, 143 S.W.3d at 83. The legislature started with data presented in 1998 in the ECOM administrative model to project which utilities might have stranded costs on December 31, 2001. Tex. UtiLCode Ann. § 39.254. It then provided “a number of tools to an electric utility to mitigate stranded costs” between 1999 and the 2004 true-up. Id. It mandated that each identified utility use these tools “to reduce the net book value of ... its stranded costs each year.” Id. However, it provided no role for the Commission during this phase. In 2001, the Commission was to prepare revised stranded cost estimates by entering 2001 data into the ECOM model and, if necessary, allow the utility to recover stranded costs through the CTC. Id. § 39.201; In re TXU, 67 S.W.3d at 133 (Phillips, C.J., concurring). In the third and final phase, the 2004 true-up, the Commission is, essentially, to settle up based on a final calculation of each utility’s stranded costs and, as warranted, permit additional stranded cost recovery or, alternatively, require each utility to refund over-recovered stranded costs. Tex. Util. Code Ann. §§ 39.201(i), .262.
Only in the 2004 true-up phase, following the final calculation of each utility’s stranded costs, did the legislature explicitly contemplate over-recovery of stranded costs. See id. Likewise, the admonishment that “[a]n electric utility ... may not be permitted to overrecover stranded costs,” on which the Commission relies, appears solely in the statute governing the 2004 true-up. See id. §§ 39.201-.262. In contrast, the legislature did not mention any role for the Commission at all during the initial mitigation phase. Id. § 39.254. As for the second phase, the sole role the legislature provided for the Commission was to impose the CTC to permit additional stranded cost recovery as warranted by the 2001 ECOM estimates; the legislature said nothing about ordering refunds of any over-recovery ascertained through estimates at that juncture. Id. § 39.201. The literal text of these statutes, the comprehensiveness of this stranded cost recover scheme, see CenterPoint, 143 S.W.3d at 83, and the fact that the prohibition against over-recovery of stranded costs appears only in the provision governing the 2004 true-up convinces us that the legislature did not intend to confer power on the Commission to order refunds of stranded cost over-recoveries based on interim estimates before the 2004 true-up.
We find further support for our conclusion when we consider the unique nature of stranded costs and the difficulty of their measurement. See Tex. Gov’t Code Ann. § 311.023 (West 1998) (code construction act). Conceptually, stranded costs under chapter 39 of the utilities code exist as of the last day before the opening of retail competition, December 31, 2001. Tex. *692Util.Code Ann. § 39.251(7).8 However, accurate calculation of such costs could take years, as a utility may not know whether it has been able to recover the millions of dollars spent on a generation-related asset until it sells the last kilowatt generated by that asset. See In re TXU, 67 S.W.3d at 147 (Brister, J., concurring) (“it will be impossible to tell whether income stream estimates [on which true-up stranded cost estimates will be based] are accurate until decades from now when the last kilowatt is sold.”).9 Any estimates of stranded costs made before that time — whether in 1998, 2001, or even in the 2004 true-up — will thus be inherently inaccurate, especially because they depend on myriad, fluctuating economic variables. See CenterPoint, 143 S.W.3d at 101 (Brister, J., dissenting); In re TXU, 67 S.W.3d at 167 (Hecht, J., dissenting). The dramatic shift in ECOM estimates between 1998 and 2001, caused by unanticipated changes in natural gas prices, demonstrated the volatility of the estimates. Accordingly, periodic estimations of stranded costs have been aptly analogized to “a system in which a jury returns a different verdict every day for a period of years, each one very different from the verdict the day before, and each one correct.” CenterPoint, 143 S.W.3d at 101 (Brister, J., dissenting). For the same reasons, revisions “to the ECOM administrative model and variations in its data input necessarily produce stranded cost estimates that are kaleidoscopic.” In re TXU, 67 S.W.3d at 163 (Hecht, J., dissenting).
Against this backdrop, the legislature mandated that the 2004 true-up calculation would be the final, controlling calculation of each utility’s stranded costs. In exchange for sacrificing some accuracy in the calculation of stranded costs, the legislature provided finality regarding the issue to facilitate the transition to a competitive electricity market by 2008. See id.; see also Tex. UtihCode Ann. § 39.262(a); CenterPoint Energy, 143 S.W.3d at 101-02 (Brister, J., dissenting); In re TXU, 67 S.W.3d at 147 (Brister, J., concurring).10 The statute, then, reflects the intent of the legislature that only this final calculation, and not the “kaleidoscopic” interim computer estimates, could serve as the basis for Commission-ordered refunds of stranded cost over-mitigation. In re TXU, 67 S.W.3d at 163 (Hecht, J., dissenting).
The intended role of the interim estimates, in contrast, was solely to provide initial parameters for rapid stranded cost recovery in the period prior to the 2004 true-up. This reflects the legislature’s emphasis on such recovery as one of its principal policy objectives in S.B. 7, the fact that stranded costs were potentially very large, and the desire to finally resolve the issue to the extent possible by the 2008 advent of full competition. See Tex. Util. Code Ann. § 39.001(b)(2).
We thus reject the Commission’s position that the prohibition against over-re*693covery of stranded costs in section 39.262(a), the true-up statute, permits it to order refunds each year during the mitigation phase based on interim ECOM estimates. It is undisputed that the Commission has the power to set procedures governing the final stranded cost determinations in 2004 true-ups. See id. § 39.262(c). However, the express authority given in section 39.262, the absence of any language concerning the power of the Commission before 2004, the express burden placed on the utilities themselves to effectuate section 39.254, and the fluctuating nature of stranded cost valuation, together lead us to conclude that the Commission lacks the power to order a refund of any “over-mitigation” that interim computer models suggest has occurred prior to the 2004 true-up.
To suggest otherwise, the dissent divorces the stranded cost over-recovery prohibition from its context within the statutory framework and overlooks the role of the final 2004 true-up calculations in ensuring a clear and certain basis to guide any Commission-ordered refunds of overrecovered stranded costs. We agree with the dissent that chapter 39 does not permit utilities the “windfall” of overrecov-ered stranded costs, but whether or not such a windfall has actually occurred is to be determined in the 2004 true-up, not based upon continually shifting, “kaleidoscopic” interim estimates. The 2004 true-up calculations, in fact, may belie the earlier estimates of “windfalls” that the dissent decries. This is hardly an “ambiguous” statutory scheme, as the dissent urges, much less an “absurd” one. Moreover, we should be exceedingly hesitant to apply such labels to justify an expansion of agency power where, as here, the legislature has squarely rejected requests to explicitly confer such power on the agency. In 2001, the legislature was requested to amend chapter 39 to give the Commission power to reverse stranded cost mitigation efforts prior before the 2004 true-up. In the face of many of the same policy considerations that the dissent ably identifies here, the legislature declined. Tex. H.B. 2107, 77th Leg., R.S. (2001) (amending Tex. Util.Code § 39.201(d)); see also In re TXU, 67 S.W.3d 130, 165 (Tex.2001) (Hecht, J., dissenting).
We sustain AEP’s first issue. In light of this disposition, we do not reach AEP’s second issue concerning the district court’s ordering of excess mitigation refunds directly to consumers rather than to REPs. Nor do we reach AEP’s third issue concerning the award of interest on any overpayment AEP is ultimately found to have made in the 2004 true-up.11
NEIL MEMBER ACCOUNT BALANCES
We now turn to the issues presented by the Cities on appeal and begin with their first, in which they argue that the Commission erred in characterizing AEP’s NEIL member account balance as generation-related rather than as an asset of AEP’s transmission and distribution business.
Background
Nuclear Electric Insurance Limited is a mutual insurance company operated by utilities, including AEP, which own nuclear power plants. It issues policies covering property damage and losses caused by interruptions at nuclear power plants. NEIL’s 79 members own and control NEIL, have rights to its policyholder divi*694dends, and would share in its assets upon liquidation. AEP participates in NEIL directly, as a member in its own right based on its interest in the twin units of the South Texas Project Nuclear Station (STP), and indirectly, as a member of the South Texas Project Nuclear Operating Company (Operating Company). Each year since the STP entered commercial operation, AEP has paid ratepayer-funded premiums into the insurance fund. In other words, AEP has included the cost of these premiums in submitting rates under traditional ratemaking procedures. In a typical year, NEIL pays out a portion of its underwriting and investment income as distributions to member insureds. The distributions take the form of rebates of the prior year’s premiums. Distributions are credited to insurance expenses and thus decreased a utility’s reported operating expenses when proposing rates to the Commission under the traditional rate-making procedures.
NEIL also retains an amount of the premiums paid sufficient to cover losses in the event of two nuclear power accidents. Although NEIL retains this surplus, it tracks each member’s “share” of it for what NEIL terms “notational” purposes. An individual member’s share is known as the “Member Account Balance” (MAB). At the end of 1999, the NEIL surplus stood at $4.1 billion. AEP’s total MAB at the end of 1999 stood at $7.1 million, consisting of $3.1 million directly held for AEP for its direct NEIL coverage and $4.0 million, its 25.2% share of the Operating Company’s MAB. Were NEIL to have dissolved at the end of 1999, AEP would have been entitled to recover that $7.1 million of NEIL’s assets.
When AEP filed its application with the Commission proposing rates based on the 2002 test year, it assigned the generation portion of its NEIL premiums to its affiliated power generation company. The Cities contested this allocation, arguing that the MABs should be credited to transmission and distribution ratepayers (the REPs) rather than to the generating company. The Commission referred this question, among others, for a hearing at the State Office of Administrative Hearings (SOAH). After a hearing, the Administrative Law Judge (ALJ) concluded that AEP’s MAB is an “asset” towards which ratepayers had contributed in their rates. As a result, the ALJ required AEP to calculate its MAB at the end of 2001 (the beginning of deregulation), to establish that amount as a regulatory asset to remain with AEP, and to moderate rates in future TDU rate proceedings. Finally, the ALJ recommended that AEP’s MAB be credited to ratepayers as of the date of deregulation.
The Commission disagreed and found that NEIL assets are generation-related rather than transmission-related. On review, the district court affirmed the Commission’s conclusion.
Discussion
On appeal, the Cities argue that the Commission erred in determining that AEP’s MAB is generation-related and, thus, attributable to AEP’s affiliated generation company. We disagree.
Standard of review
Our review of this issue is under the substantial-evidence standard. See Tex. Util.Code Ann. § 15.001 (West 1998); Reliant Energy, Inc. v. Public Util. Comm’n, 153 S.W.3d 174, 184 (Tex.App.-Austin 2004, no pet.). We presume that the Commission’s findings are supported by substantial evidence, and the contestant bears the burden of proving otherwise. See Southwestern Pub. Serv. v. Public Util. Comm’n, 962 S.W.2d 207, 215 (Tex.App.Austin 1998, pet. denied). We will reverse and remand the cause to the agency when *695substantial rights of the appellant have been prejudiced by an agency’s findings that are not reasonably supported by substantial evidence considering the reliable evidence in the record as a whole. Tex. Gov’t Code Ann. § 2001.174(2)(E) (West 2000). However, we may not substitute our judgment for that of the agency on the weight of the evidence. Southwestern, 962 S.W.2d at 215. “Substantial evidence” does not mean a large or considerable amount of evidence but such relevant evidence as a reasonable mind might accept as adequate to support a conclusion of fact. Pierce v. Underwood, 487 U.S. 552, 564-65, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988); Lauderdale v. Department of Agric., 923 S.W.2d 834, 836 (Tex.App.-Austin 1996, no writ). We must first determine whether the evidence as a whole is such that reasonable minds could have reached the conclusion that the agency must have reached to take the disputed action. Texas State Bd. of Dental Exam’rs v. Sizemore, 759 S.W.2d 114, 116 (Tex.1988); Ramirez v. Texas State Bd. of Med. Exam’rs, 995 S.W.2d 915, 919 (Tex.App.-Austin 1999, pet. denied). The test is not whether the agency made the correct conclusion but whether some reasonable basis exists in the record for the agency’s action. Railroad Comm’n v. Pend Oreille Oil & Gas Co., 817 S.W.2d 36, 41 (Tex.1991); Texas Health Facilities Commission v. Charter Med.-Dallas, Inc., 665 S.W.2d 446, 452. The agency may accept or reject in whole or in part the testimony of the various witnesses who testify. Central Power & Light Co. v. Public Util. Comm’n of Tex., 36 S.W.3d 547, 557 (Tex.App.-Austin 2000, pet. denied). We must uphold an agency’s finding even if the evidence actually preponderates against it so long as enough evidence suggests the agency’s determination was within the bounds of reasonableness. Southwestern, 962 S.W.2d at 215. If the agency offers more than one ground as the basis for its decision, we will affirm if we find substantial evidence supporting one ground even if all bases given would be independently sufficient to support the decision. Texas State Bd. of Medical Exam’rs v. Scheffey, 949 S.W.2d 431, 436 (Tex.App.-Austin 1997, -writ denied).
Application
In this case, the record contains conflicting testimony concerning the proper characterization of the NEIL MABs. Nancy Bright, an accountant and a consultant, testified on behalf of the Cities that the MAB is an “asset” of AEP because it reflects the share of NEIL’s funds to which AEP has a right. According to her analysis, an REP’s rates included amounts to cover NEIL insurance premiums. AEP paid those premiums and, as a TDU, received insurance covering possible losses due to a disruption of service. NEIL makes distributions out of its surplus every year, and AEP, as a NEIL member, has a role in determining the amount of the distribution. Although NEIL does not classify the amounts in the MABs as “assets” for tax purposes, AEP will be able to recover the amount in its MAB upon liquidation of NEIL or upon a duly-approved distribution. Thus, she concluded AEP’s MAB acts as an asset for AEP’s transmission and distribution business.
On the other hand, David Carpenter, AEP’s director of Texas regulatory services, testified that the MABs are more correctly viewed as NEIL’s surplus, an equity. NEIL uses that surplus to purchase securities, which are NEIL’s assets, not the individual utilities. Historically, MAB accounting problems had arisen because some NEIL members had been in the process of selling their ownership rights in their nuclear power plants or were decommissioning their plants. Under NEIL bylaws, those utilities would have no longer been members of NEIL *696and would have lost any rights associated with their MABs, including the right to receive an MAB distribution upon NEIL’s theoretical dissolution. However, a utility’s MAB interest would not transfer to the new owner of that nuclear power plant. Instead, the value of the MAB would be distributed to the balances of the MABs of the remaining members. In part to resolve this problem, NEIL established nonnuclear insurance lines for utilities to purchase so that they might remain NEIL members and thus maintain their stakes in their MABs.
Carpenter further testified that MABs represent a surplus, not an asset, because they result from premium rates paid. In other words, NEIL charges insurance rates that have been determined to be “reasonable and prudent” for the purpose of providing insurance for nuclear accidents. REPs receive the benefit of the insurance coverage. In addition, distributions from the NEIL surplus are made in the form of reduction in rates of premiums, not in the form of money transfers.
The Commission made several findings in its order. First, it determined that NEIL assets are generation-related and thus remain with the unbundled generation company. Second, the Commission found that REP ratepayers have received benefits from the NEIL premiums through risk reduction and have received credits for rate expenses through NEIL distributions. Finally, it noted that “the value of the asset will be determined in the 2004 true-up proceeding at the generation plant valuation.” Therefore, the Commission concluded that the AEP’s NEIL member account balance be treated as a generation-related asset.
The Cities’ complaints on appeal center on a lack of evidence in the record to support the Commission’s statement about a 2004 true-up reconciliation. They do not assert a lack of evidence concerning the Commission’s other findings. The true-up reconciliation ground was only one of several on which the Commission based its conclusion. Reasonable minds could differ concerning the remaining grounds, but the Commission’s decision is reasoned. The assertions that the MABs are generation-related and that REP ratepayers have benefitted from NEIL insurance and distributions are supported by substantial evidence in the record. See Reliant Energy, 153 S.W.3d at 204. Substantial evidence exists, then, to support the grounds on which the Commission based its decision. As a result, we must affirm under the substantial-evidence rule. See Scheffey, 949 S.W.2d at 436. We overrule the Cities’ first issue.
DEMAND CHARGES
In their second issue, the Cities argue that the Commission erred in authorizing demand charges in excess of those charged under AEP’s bundled rate because the Commission allegedly shifted the burden of proof to the Cities when the burden should have remained on the utilities and because any demand charges greater than those approved before unbun-dling negatively impact competition in violation of section 39.001(d) of the utilities code.
Background
A demand-metered customer’s bill consists of a customer charge, a charge for delivered electricity, and a charge for “demand.” Demand is a measurement of a customer’s actual demand on the utility’s system at a given point in time. In other words, it is a measurement of the rate at which energy is consumed. Demand is physically measured by meters, and most small commercial and residential users have meters which measure maximum de*697mand over a period of a month in “per-kilowatt-hour” units. Large commercial customers often have more expensive meters that measure maximum demand over short intervals, such as fifteen or thirty minute periods, on a “per-kilowatt” basis. The function of the demand charge is to allow the utility to recover fixed costs arising from the demand placed on the system that are not reflected in the rate set for the electricity itself.
Transmission and distribution facilities are “fixed cost” facilities in that they are constructed to meet local or individual peak demands. A demand ratchet compensates a utility for initial cost and maintenance of those facilities over the course of the year, because those costs do not follow seasonal or other demand patterns. Use of a ratchet “flattens” these charges throughout the year. For example, with an 80% demand ratchet, as ultimately adopted by the Commission, a customer’s demand charge in a given month will be an amount based on the greater of the current month’s demand or 80% of the customer’s highest monthly demand in the preceding eleven months.
During the initial phase of unbundling, each utility in Texas except for El Paso Electricity Company had filed a rate case before the Commission. The Commission decided, because of the factors common in all the cases, to set the rate cases in the generic docket to be followed by company-specific hearings. In the generic docket, the Commission then adopted a rate design that included a demand charge rather than a rate design based on a “seasonal” differential system.12 It adopted an 80% demand ratchet for transmission and distribution rates because it decided that an 80% ratchet most appropriately recognized load diversity between different customers.13 The Commission further announced it could grant exceptions to the generic rate design,14 but only “if necessary to address extraordinary impacts on the ability of customers to obtain service from a competitive provider due to restrictions of the price to beat (i.e., ‘headroom concerns’[ 15 ]).” Headroom concerns, however, *698would not automatically mandate an exception to the generic rate design.
In AEP’s individual docket, AEP had originally proposed a demand charge for large commercial customers of $2.83/kW, based on their original argument in the generic case that the demand ratchet be set at 100% rather than at 80%, and supported its proposal with the testimony of Donald Moncrief, the manager of the regulated pricing and analysis section of one of AEP’s subsidiaries. The Cities argued instead that the demand charge should remain at the bundled rate level of $2.74/kW. They believed that customers “with demands that vary month to month” would be unlikely to have access to competitive services because the application of the demand ratchet to demand charges coupled with the proposed rate would reduce headroom to non-competitive levels. AEP responded that the Cities failed to justify that a headroom problem existed or the necessity of a shift. The ALJ found that AEP produced evidence that the proposed rate would not create a headroom problem. She also found that the Cities failed to produce any specific evidence of “an extraordinary headroom concern that warrants an exception to the generic rate design.” Thus, she concluded that reducing AEP’s demand charge to the bundled rate level would arbitrarily shift costs to “high-load-factor customers.” She made no conclusion about the proper demand charge rate. The Commission agreed with the ALJ’s analysis and ultimately set AEP’s demand charge at $3.27/kW.
Discussion
The Cities bring two challenges to the Commission’s order concerning AEP’s demand charge for large commercial customers.16 First, the Cities claim that the burden of proof for showing that a proposed rate is “just and reasonable” lies with the utility. See Tex. UtihCode Ann. § 36.006 (West 1998). As a result, they argue that the Commission erred in adopting the ALJ’s analysis. Next, the Cities argue that the Commission’s adoption of “demand charges exceeding those assessed for bundled service negatively impacts competition in violation of’ utilities code section 39.001(d).17 The Cities do not argue that the adopted rate negatively impacts competition. Rather, they argue that any rate greater than the bundled rate negatively impacts competition in violation of section 39.001(d). Nor do the Cities argue that the Commission acted outside its statutory authority in setting the demand charge. They argue only that the rate itself violates the statutory requirements. Assuming without deciding that the burden in this case properly lay with AEP,18 the essence of both challenges *699to the demand charge set by the Commission is that the evidence produced by AEP and relied upon by the Commission does not support the Commission’s decision to set a rate greater than the bundled rate. Thus, as when we considered the proper characterization of the NEIL MABs, above, in considering the Cities’ second issue we will apply the substantial-evidence test.
AEP initially proposed its demand charge for large commercial customers and offered Moncriefs testimony to support the proposition that its demand charge would leave sufficient headroom for competition and would provide an attractive rate for customers. Because AEP is entitled to recover its transmission and utility costs, any reduction in the demand charge for large commercial customers would result in an increase in rates, and a related decrease in headroom (and thus possible decrease in competition), for other customer classes. Large commercial customers, such as the Cities, place the largest share of demand on the system. As a result, Moncrief analyzed “typical customer bills” based on AEP’s set of proposed rates and concluded that the proposed demand charge for large commercial customers would result in bills that would adequately reflect their share of the demand placed on the system.
In response, the Cities offered the testimony of Steven Anderson, a consultant specializing in regulatory analysis and asset valuation, to argue that the Commission should set AEP’s demand charge at the level of its unbundled demand charge. He testified that a higher demand charge coupled with an 80% demand ratchet would create a financial hardship on REPs that elect “to serve customers with demands that vary significantly from month to month. As a result, it is unlikely such a customer will have access to competitive service.” He then suggested that the revenue shortfall that would result from lower demand charges be recovered by increasing energy charges. Moncrief responded to Anderson’s testimony by pointing out that Anderson’s recommendation would result in higher rates for residential and small commercial customers to support lower demand charges for large commercial customers. This cost-burden shifting, he argued, would violate “the rule that rates should be based on costs” and would reduce headroom for residential and small commercial customers.
Considering this testimony, the ALJ found that Moncrief produced analysis establishing that the proposed demand charge would produce no headroom problem for typical customers. She also accepted AEP’s argument that the Cities’ position would shift the burden created by the demand ratchet away from high demand, large commercial customers onto residential and small commercial customers. The Commission agreed with the ALJ’s conclusions.
We find that reasonable minds could have reached the conclusion that the Commission did. AEP produced evidence in its ease to support its proposed rate and rebutted the Cities’ proffered evidence. The Commission could have accepted in whole Moncriefs testimony, which supported setting a demand charge greater than the bundled rate, and rejected in whole Anderson’s testimony. Therefore, we find that the Commission based on substantial evidence its decision to set a demand charge greater than the demand charge approved for the bundled utility. We overrule the Cities’ second issue.
CONCLUSION
We have sustained AEP’s arguments that the Commission lacked authority to order refunds of allegedly “over-mitigated” *700stranded costs before a final determination is made at the 2004 true-up. As a result, we reverse the portion of the district court’s judgment concerning stranded cost over-mitigation and remand those issues for further proceedings consistent with this opinion. At the same time, we have overruled the Cities’ issues concerning the proper characterization of AEP’s NEIL member account balances and the adoption of demand charges greater than the demand charges approved for the bundled utility. Thus, we affirm the district court’s judgment in those respects.
Dissenting Opinion by Justice B.A. SMITH.
. As explained in CenterPoint,
The Legislature recognized that in fundamentally changing the industry, it was alter*686ing the assumptions that had led utilities to invest large sums in power generation assets. The Legislature understood that the cost of these assets likely would be recovered in a regulated environment, but might well become uneconomic and thus unrecoverable in a competitive, deregulated electric power market. The Legislature called such uneconomic assets stranded costs. The term "stranded costs” ... [means] the extent to which the book value of generation-related assets and purchased power contracts exceeds their market value.
The Legislature concluded that if generating plants became uneconomic as a result of legislatively mandated deregulation, it was in the public interest for utilities to be made whole by recovering their full investment in those generation plants, although the utilities would no longer receive a return on those investments. The Legislature determined that utilities should not be required to forfeit their investments in generating plants with the advent of deregulation.
CenterPoint Energy, Inc. v. Public Util. Comm’n of Tex., 143 S.W.3d 81, 82-83 (Tex.2004) (footnotes omitted).
. "Small commercial customers” are commercial customers having a peak demand of 1,000 kilowatts or less. Tex. UtiLCode Ann. § 39.202(o) (West Supp.2004-05).
. "Securitization” is a method to recover stranded costs by which a utility issues transition bonds that are secured by, or payable from, a nonbypassable transition charge, assessed for the use or availability of electric service, as approved in a Commission-issued financing order.
. In part, the difference between the 1998 and 2001 estimates could be related to an unprojected surge in natural gas prices between those years, which affected the market price of nuclear generating plants relative to those powered by natural gas. See In re TXU Elec. Co., 67 S.W.3d 130, 133 (Tex.2001) (Phillips, C.J., concurring).
. At that time, TXU Electric Co. filed a petition for writ of mandamus in the supreme court, arguing that the Commission lacked jurisdiction to order reverse mitigation credits based on the 2001 interim estimates. See In re TXU Elec. Co., 67 S.W.3d 130, 131 (Tex. 2001) (per curiam). Six members of the court voted to deny relief for different reasons. Id. Chief Justice Phillips, joined by two others, would not have exercised mandamus jurisdiction because he believed TXU had an adequate remedy at law. Id. at 132-36 (Phillips, C.J., concurring). Justice Baker and Justice Rodriguez felt that the court had no jurisdiction to mandamus a state board or commission. Id. at 136-45 (Baker, J., concurring). Justice Brister, at the time serving on the court of appeals' and sitting by assignment, believed that the court did have jurisdiction but that the Commission had power to order reverse mitigation efforts. Id. at 145-50 (Brister, J., concurring). Justice Hecht, joined by then-associate Justice Jefferson and Justice Owen, argued that the court had jurisdiction and that the Commission lacked statutory authority to order reverse mitigation efforts. Id. at 150-71.
In In re TXU, only four justices reached the issues we confront here regarding the Commission's power to order reverse mitigation based on interim stranded cost estimates. Although thus not binding authority, strictly *689speaking, these opinions provide especially helpful background regarding the applicable statutes, the nature of stranded costs, and the parameters of the debate regarding Commission power to order reverse mitigation efforts. For these purposes, we cite these opinions extensively in the foregoing discussion.
. For the same reasons, we will not separately address the three issues presented by cross-appellant Constellation New Energy, an unaffiliated REP. Constellation joins AEP in arguing that the district court erred in requiring over-mitigation credits to be paid to end-use consumers rather than to the REPs. It also argued that applying over-mitigation credits to REPs does not discriminate against residential and small commercial customers and does not permit XDUs to over-recover stranded costs, thus joining AEP’s second issue.
. Thus, the differing stranded cost estimates in 1998 and 2001 (and possibly in the 2004 true-up) are not snapshot views of continually accruing costs at different points in time, but different estimates of the same figure.
. We note that, were the issue of over-recovery of stranded costs directly addressed in In re TXU, Justice Brister would have held that the Commission has authority to address "over-recovery” of stranded costs before the 2004 true-ups. See 67 S.W.3d at 145-50. We cite his opinion here for its helpful discussion of the nature of stranded costs and the problems associated with their recovery.
.As Justice Brister noted in In re TXU, that the 2004 valuations will be final does not mean that they will be accurate. The legislature guarantees only finality in this phase of the transition to competition, not the ultimate accuracy of the stranded cost valuations.
. We assume that the Commission will consider on remand of this case the implications of CenterPoint Energy, Inc. v. Public Util. Comm’n of Tex., 143 S.W.3d 81 (Tex.2004).
. Under a "seasonal” differential charge system, a utility would be permitted to charge a higher rate during summer months (typically from June through September) and a lower rate during the rest of the year to reflect the demand put on the system during the peak demand summer months. A seasonal differential system is also termed a "flat kWh” charge. To further confuse matters, a demand charge system may also be termed a "seasonal kWh charge.” Typically, a demand charge system results in charges that remain relatively "flat” over the course of a year. A seasonal differential charge system, on the other hand, yields higher charges during the summer and lower charges the rest of the year. Generated revenue for TDUs is generally the same from either system. The differences lie mostly in the flow of the revenue stream.
. In doing so, it rejected AEP’s proposed 100% demand ratchet.
. The Commission exempted seasonal agricultural customers from the demand ratchet on finding that those customers only use electricity in significant amounts one or two months a year. Thus, seasonal agricultural customers are only billed demand charges during months of significant demand.
. "Headroom” refers to the margin between the "price to beat” and the new REPs’ costs of providing electricity. From January 1, 2002 until January 1, 2007, electric providers formerly affiliated with regulated utilities must provide electricity at rates that are six percent lower than their rates before deregulation. This rate is known as the "price to beat.” See Tex. Util.Code Ann. § 39.202 (West Supp.2004-05). In enacting the price-to-beat statute, the legislature intended to create incentives for new REPs not affiliated with the regulated utility industry to enter the market and compete for customers with affiliated REPs, those that were formerly part of the bundled utility companies. Thus, the *698greater the headroom, the more room for new market entrants to engage in pure competition with affiliated REPs.
. Neither party argues about the evidence supporting the rate ultimately adopted by the Commission, $3.27/kW. They argue only about the Commission’s adoption of any rate greater than the bundled rate.
. The Commission "shall authorize or order competitive rather than regulatory methods to achieve the goals of [chapter 39 of the utilities code] to the greatest extent feasible and shall adopt rules and issue orders that are both practical and limited so as to impose the least impact on competition.” Tex. Util.Code Ann. § 39.001(d) (West Supp.2004-05). This statutory mandate does not forbid the Commission from setting rates greater than the unbundled rate. It only requires that orders “impose the least impact on competition.” Id.
. See Office of Pub. Util. Counsel v. Public Util. Comm'n of Tex., No. 03-03-00462-CV, slip op., at *17-18, 2005 WL 1787434 (Tex. App.-Austin July 28, 2005, no pet. h.) (claim that Commission improperly shifted burden of proof in fuel-factor case, when utility offered evidence in support of its position, decided under substantial-evidence review).