The Federal Energy Regulatory Commission is responsible for ensuring that interstate electricity rates are “just and reasonable.” 16 U.S.C. §§ 824d(a), 824e(a). To do so, it approves electricity providers’ proposed rate changes, and it can require them to change their rates if the rates become unreasonable. This case is about one of FERCs rate determinations.
Midcontinent Independent System Operator, Inc. administers the electric grid on behalf of the companies that own transmission lines. Those transmission owners invested money to build their transmission lines, and MISO must charge customers electricity-transmission rates that provide those companies an appropriate return on their investment. That return-on-equity component of the transmission rates, which well just call the Return, is at issue in this case.
In this case, a group of customers thought MISO provided transmission owners a too-generous Return. They asked FERC to reduce that aspect of MISOs rates. FERC did. In the process, it completely overhauled its approach to setting an appropriate Return.
Both the customers and transmission owners now challenge several aspects of the FERC proceedings as unlawful or arbitrary and capricious.
We agree with the customers that FERCs development of the new Return methodology was arbitrary and capricious, so we vacate its rate-determination orders and remand for further proceedings. Because the other challenged aspects of FERCs orders flow from FERCs rate determination, we do not reach them.
I
We start this section with some background on the general regulatory framework for electricity-transmission rates. Then we describe the history of FERCs approach to Return determinations. Finally, we explain what happened in these proceedings.
A
For most of the twentieth century, vertically integrated state and local utilities monopolized electricity markets. See Atlantic City Electric Co. v. FERC, 295 F.3d 1, 4 (D.C. Cir. 2002). When technological progress enabled competitors to offer lower prices for electricity, the incumbent utilities used their control of transmission lines to keep competitors out of the market. Id. That exclusion caused higher prices. So in 1996, FERC required utilities to provide open access to transmission lines. Id. To help achieve its open-access goals, FERC created a framework for independent companies, called independent system operators, that would impartially operate transmission lines. Id. at 5.
MISO performs that service for fifteen states in the middle of the country from Louisiana up to Minnesota (and beyond to Manitoba). In exchange for its services, it charges transmission rates that approximate the costs it incurs plus an appropriate return on equity for the transmission owners’ original investment in building the lines. See FERC, Energy Primer: A Handbook of Energy Market Basics 59-60 (2020).
Like all public utilities, MISO must file its proposed rates with FERC for approval. As part of its review, FERC ensures that the Return portion of the rates is appropriate to compensate transmission owners for the risks they took and to attract future investment in transmission lines. Emera Maine v. FERC, 854 F.3d 9, 20 (D.C. Cir. 2017).
There are two ways that MISOs rates can change.
One, called a Section 205 proceeding, is utility-initiated. If MISO wishes to change its rates, it can file a new set of proposed rates with FERC. 16 U.S.C. § 824d(d). FERC then reviews the proposed rates to determine whether they are just and reasonable. Id. § 824d(e). If they are, MISO can charge them. NRG Power Marketing, LLC v. FERC, 862 F.3d 108, 114 (D.C. Cir. 2017). If not, FERC rejects them. Id.
The other, called a Section 206 proceeding, is customer-or FERC-initiated. A customer can file a complaint alleging that a current rate is unjust and unreasonable, or FERC can set a hearing on its own motion. 16 U.S.C. § 824e(a). At step one, FERC decides if the old rate is unjust and unreasonable. Id. If so, then FERC proceeds to step two and sets a new rate. Id.
Until FERC sets a new rate in a Section 206 proceeding, customers continue to pay the challenged rates. See City of Anaheim v. FERC, 558 F.3d 521, 525 (D.C. Cir. 2009). So Congress gave FERC limited refund authority. At the beginning of the proceeding, FERC sets “a refund effective date.” 16 U.S.C. § 824e(b). It can then give refunds of any excess payments for fifteen months after that refund effective date. Id. Those excess payments are calculated as the difference between the old, challenged rate and the new rate ordered by FERC. Id.
This case is about a Section 206 proceeding.
B
To understand what FERC did in this proceeding, it helps to have some historical background on FERCs methodology for assessing the reasonableness of the existing Return and, if necessary, setting a new one.
Since the 1980s, FERC calculated the Return with the aid of a financial tool called the discounted-cash-flow model. That model uses a companys stock price to represent the companys value to investors. Canadian Association of Petroleum Producers v. FERC, 254 F.3d 289, 293 (D.C. Cir. 2001). It assumes that the stock price is equal to all the dividends the company will pay out in the future “discounted at a market rate commensurate with the stocks risk.” Id. A simplified version of that baseline formula is P = D/(r-g), “where P is the price of the stock at the relevant time, D is the dividend to be paid at the end of the first year, r is the rate of return and g is the expected growth rate of the firm.” Id. That is the version investors use to try to calculate a companys stock price. But to calculate an appropriate Return for transmission owners, FERC rearranges the equation to be:
r = D/P + g.1
For publicly traded companies, calculating an appropriate Return with the discounted-cash-flow model is relatively easy because of its publicly traded stock price. But for privately held companies like the transmission owners, which have no public stock price, FERC uses a proxy group of comparable, publicly traded companies. Id. at 293-94. With that proxy group of public companies, FERC can approximate what a discounted-cash-flow analysis should look like for the privately held companies at issue. Id.
When FERC chooses a proxy group and conducts a discounted-cash-flow analysis for each company in the group, it gets a range of possible Returns that FERC calls the “zone of reasonableness.” Emera Maine v. FERC, 854 F.3d 9, 15 (D.C. Cir. 2017). A Return must be a single value, so FERC then needs to choose a point within the zone. It typically uses the midpoint, at least for independent system operators like MISO. See Southern California Edison Co. v. FERC, 717 F.3d 177, 186 (D.C. Cir. 2013).
That was the state of play until 2014: FERC would produce a zone of reasonableness using a discounted-cash-flow analysis of proxy group companies, then set the Return at the midpoint.
Then FERC changed things up. In a rate-review proceeding for New Englands independent system operator, FERC found that anomalous market conditions required a higher Return than the one provided by the midpoint of the discounted-cash-flow models zone of reasonableness. Emera Maine, 854 F.3d at 18. It looked at several other models to determine how much higher the Return should go and ultimately set the Return at the midpoint of the upper half of the zone of reasonableness. Id.
That brings us to this case.
C
This case started with two separate Section 206 complaints against MISOs rates.
In 2013, a group of customers believed the Return component of MISOs existing rate was too high. They filed a Section 206 complaint asking FERC to lower it. That was this cases first complaint.
FERC set a refund effective date of November 12, 2013, which meant that customers could only get refunds for overpayments through February 11, 2015. But FERC did not resolve the first complaint by February 11, 2015. The following day, a different group of customers filed a complaint challenging the same MISO rate. That was this cases second complaint.
Finally, on September 28, 2016, FERC resolved the first complaint in Opinion No. 551. It agreed with the customers and reduced the Return from 12.38% to 10.32%. In doing so, it used the same Return-setting methodology that it had developed in the New England proceeding.
The next year, in Emera Maine v. FERC, we vacated FERCs orders from the New England proceeding. 854 F.3d at 30. We identified two infirmities in FERCs analysis. First, as the transmission owners had argued, FERC “never actually explained how” the New England transmission owners’ existing Return “was unjust and unreasonable.” Id. at 26. And second, as the customers had argued, FERC failed to justify its decision to set the Return at the three-quarters point of the zone of reasonableness. Id. at 28-29.
Because FERC had relied so heavily in this proceeding on the orders that we vacated in Emera Maine, FERC chose to revisit Opinion 551. It set the first complaint for rehearing and informed the parties that it planned to resolve the second complaint in the same rehearing proceeding.
In its rehearing order, FERC proposed an entirely new methodology for calculating a just and reasonable Return. The proposal used four different financial models, giving each equal weight:
• Model 1, discounted cash flow (as described three pages ago);
• Model 2, capital-asset pricing;2
• Model 3, expected earnings;3 and
• Model 4, risk premium.4
FERC planned to use the first three models, each of which produce a zone of reasonableness, to answer the threshold question whether an existing rate is unjust and unreasonable. Because risk premium (Model 4) produces only a single point, FERC intended to leave it out of that first step. It planned to create a composite zone produced by the average of the first three models’ zones of reasonableness, then divide the composite zone to create presumptively just and reasonable ranges for utilities based on their risk profiles, as this image shows:
Then, if FERC found an existing Return unjust and unreasonable, it would set a new Return by averaging the midpoint (or the one-quarter or three-quarters point for utilities of below-average and above-average risk respectively) of the first three models with the single point that the risk-premium model (Model 4) produces.
A year later, when FERC issued its second order in this proceeding — Opinion No. 569 — it abandoned expected earnings (Model 3) and risk premium (Model 4), and made other, more minor tweaks to its proposed Return methodology. It then applied the new methodology, again found the pre-complaint 12.38% Return unjust and unreasonable, and set a new Return of 9.88%. FERC backdated that new Return to make it effective as of September 28, 2016, requiring the transmission owners to refund — for the period between the first and second orders — the difference between the 10.32% FERC had set in its first order and the 9.88% it had set in its second order.
5
As it had promised, FERC also resolved the second complaint in Opinion 569. It determined that the currently effective Return was the new 9.88% Return that it had just imposed. Then it found that 9.88% was not unjust and unreasonable. It therefore did not order a new rate in response to the second complaint. And because it had not ordered a new rate, FERC concluded that it could not order a refund for the second complaints fifteen-month refund period.
The customers and transmission owners alike found fault with Opinion 569, so both petitioned for rehearing on several grounds. FERC granted rehearing and, in its third order — Opinion 569-A — FERC again changed its Return methodology. It added risk-premium (Model 4) back into the mix and shifted the presumptively just and reasonable zones, among other things.
After explaining its changes, FERC applied the new Return methodology to, yet again, find the pre-complaint 12.38% Return unjust and unreasonable. FERC then set a new Return of 10.02%, which it again backdated to September 28, 2016. Finally, it used that 10.02% Return to again reject the second complaint.
The parties again sought rehearing before FERC. In response, FERC issued Opinion No. 569-B, which tweaked the Return methodology a bit without making any further major changes.
This chart summarizes the relevant FERC proceedings.
II
Under the Administrative Procedure Acts arbitrary-and-capricious standard, our review of FERCs ratemaking choices is limited. 5 U.S.C. § 706(2)(A); see also Emera Maine v. FERC, 854 F.3d 9, 21-22 (D.C. Cir. 2017). We must deny the petitions for review as long as FERC “has made a principled and reasoned decision supported by the evidentiary record.” Id. at 22 (quoting Southern California Edison Co. v. FERC, 717 F.3d 177, 181 (D.C. Cir. 2013)). That inquiry includes verifying that FERC had a reasoned basis for any changes of heart. Verso Corp. v. FERC, 898 F.3d 1, 7 (D.C. Cir. 2018).
III
The customers challenge FERCs new Return methodology on five grounds. First, they argue that FERC should not have altered its previous approach to balancing long-term and short-term growth rates in the discounted-cash-flow model (Model 1). Second, they challenge three aspects of FERCs approach to the capital-asset model (Model 2). Third, they argue that FERCs creation of presumptively just and reasonable ranges at step one of the Section 206 analysis was arbitrary and capricious. Fourth, they argue that FERC should have set the new Return based on the median of the zone of reasonableness rather than the midpoint. And fifth, they challenge FERCs decision to resuscitate the risk-premium model (Model 4) in its second rehearing order shortly after interring the model in its first rehearing order.
We find the first four of those arguments unpersuasive. But we agree with the customers’ final argument. And that conclusion is alone enough to make FERCs rate orders arbitrary and capricious.
A
The customers take aim at a change that FERC made to its discounted-cash-flow analysis (Model 1). Remember, the simplified version of that is r = D/P + g, with the letters representing the Return, dividend, stock price, and expected growth rate.
In conducting a discounted-cash-flow analysis for a company, FERC balances short-term and long-term expected growth to pick an expected growth rate. Before 1999, FERC used a fifty-fifty split. Canadian Association of Petroleum Producers v. FERC, 254 F.3d 289, 292, 297 (D.C. Cir. 2001). After 1999, FERC used a two-thirds-short-term versus one-third-long-term split. Id. at 297.
In this proceeding, FERC changed to a four-fifths-short-term versus one-fifth-long-term split. When we approved the 1999 change (from the pre-1999 fifty-fifty split), we noted that because this kind of weighting doesnt lend itself to “strict rules, it would likely be difficult to show that [FERC] abused its discretion in the weighting choice.” Id.
That remains true. Short-term rates are more reliable projections; long-term rates just “normalize any distortions” in the short-term rates. Id. (cleaned up). Recently, the normalizing value of long-term rates has declined as the short-term and long-term projections have converged. So as the importance of long-term rates has declined, FERC decided that their role in the discounted-cash-flow analysis should too. That was not arbitrary and capricious.
B
The customers next challenge three aspects of FERCs application of the capital-asset model (Model 2). We reject each challenge.
That model begins with the following formula:
Return = risk-free rate + beta(expected return – risk-free rate).
Lets break down each term in that formula, as FERC applied it in this proceeding:
• The risk-free rate is the Treasury-bond rate.
• The “beta” is a company-specific value that industry experts assign to measure a companys riskiness as an investment. A beta value of one represents average risk, such that a beta below one represents a lower-risk company and a beta above one represents a higher-risk company.6
• The expected return is the result of a discounted-cash-flow analysis of all dividend-paying companies in the S&P 500.7
Although FERCs application of the model begins with that formula, it doesnt end with it. Before running the formula, FERC adjusts the beta towards 1.0 because some finance scholars believe that betas “converge to 1.0” in the long run. JA 611 (quotation marks omitted). Then, after running the formula, FERC takes the formulas Return result and applies a “size-premium adjustment” to that result. The adjustment is a value meant to ensure that the model adequately accounts for companies’ sizes.
For this model, the customers challenge: (1) FERCs decision not to include long-term growth rates in its analysis of the S&P 500; (2) its use of adjusted betas (as part of the formula) with a size-premium adjustment derived from unadjusted betas (applied after running the formula); and (3) its use of betas based on the market risk of the New York Stock Exchange with an expected market return based on the S&P 500. We will address each individually.
1
For the dividend-paying S&P 500 companies that FERC used to determine the “expected return,” no one knows for sure how much they will grow. But those companies’ growth rates are necessary to calculate the expected return. So FERC filled in that blank with five-year growth projections from the Institutional Brokers’ Estimate System. It rejected the customers’ request that it average those five-year projections with longer-term growth projections.
FERC adequately explained that decision. It cited financial research that supported the use of only short-term growth rates. And it explained that the short-term rates better reflect an investors expected return on an investment in the S&P 500 as an index. Thats because the S&P 500 is regularly updated to include only companies with high market capitalization. Further, FERC explained that the S&P 500 includes companies at all stages of growth, so older companies with lower growth potential will balance out younger companies with higher growth potential. In light of the “great deference” that we afford FERCs ratemaking analysis, that explanation is sufficient. See FERC v. Electric Power Supply Association, 577 U.S. 260, 292, 136 S.Ct. 760, 193 L.Ed.2d 661 (2016) (cleaned up).
2
The second issue concerns the size-premium adjustment that FERC applied to the result of its formula. Ibbotson, the company that calculated the size premium, analyzed a large group of companies in the New York Stock Exchange. To grossly simplify, Ibbotson applied a capital-asset formula to those companies and then saw if there were any differences in the results that were best explained by size. See Frank Torchio & Sunita Surana, Effect of Liquidity on Size Premium and Its Implications for Financial Valuations, 9 J. Bus. Valuation & Econ. Loss Analysis 55, 56-57 (2014).
Ibbotson used unadjusted betas in its capital-asset formula. But recall that FERC used adjusted betas for its capital-asset formula. The customers argue that FERCs decision to use both despite that mismatch was irrational.
FERC acknowledged the “imperfect correspondence” between the two. JA 611. But it decided that the size-premium adjustment sufficiently improved the capital-asset models accuracy to justify the mismatch.
We can only judge FERCs logic based on the evidence it had before it. See FCC v. Prometheus Radio Project, ––– U.S. ––––, 141 S. Ct. 1150, 1160, 209 L.Ed.2d 287 (2021). Here, because FERC had a size-premium adjustment based on unadjusted betas and believed that adjusted betas were the most appropriate input to use in the capital-asset model, it had to choose between “imperfect correspondence” and no size adjustment at all. That is the kind of technical choice to which we are “particularly deferential.” Public Service Commission of Kentucky v. FERC, 397 F.3d 1004, 1006 (D.C. Cir. 2005) (quoting Time Warner Entertainment Co. v. FCC, 56 F.3d 151, 163 (D.C. Cir. 1995)). We do not find it arbitrary and capricious.
3
That same logic persuades us to reject the challenge to FERCs decision to combine adjusted betas based on the New York Stock Exchange with an expected return based on the S&P 500. Here, too, FERC acknowledged the “imperfect correspondence” between the New York Stock Exchange and the S&P 500. JA 873. But FERC concluded that it would not be reasonable to calculate an expected return using all 2,800 companies in the New York Stock Exchange. And no party provided adjusted betas from the appropriate time frame based on the S&P 500. It was not arbitrary and capricious for FERC to do the best it could with the data it had. See Prometheus, 141 S. Ct. at 1160.
8
C
From there, the customers level an array of challenges to FERCs creation of presumptively just and reasonable ranges at step one of the Section 206 analysis. Recall, if youll suffer another reminder, that FERC created ranges within the zone of reasonableness based on the companys risk profile to analyze the step-one question of whether an existing rate is unjust and unreasonable. Rates within the appropriate range are presumed to be just and reasonable.
1
First, the customers argue that we did not require FERC to adopt its presumption scheme when we vacated FERCs New England opinion in Emera Maine. See 854 F.3d 9 (D.C. Cir. 2017). That is true, but it misses the point. FERC is entitled to adopt any methodology it believes will help it ensure that rates are just and reasonable, so long as it doesnt adopt that methodology in an arbitrary and capricious manner. See Southern California Edison Co. v. FERC, 717 F.3d 177, 182 (D.C. Cir. 2013).
As FERC recognized, our opinion in Emera Maine held that FERC had failed to sufficiently explain why the existing rate was unjust and unreasonable at step one of the Section 206 inquiry. 854 F.3d at 26-27. We had explained that “the zone of reasonableness creates a broad range of potentially lawful” Returns, such that FERC needed to do more than identify a single new Return that it preferred. Id. at 26. So in response, FERC developed this new framework to more effectively verify that an existing rate is in fact unjust and unreasonable. The customers have provided no persuasive reason to think that doing so was arbitrary and capricious.
2
Second, the customers contend that the presumption scheme unlawfully heightens the burden of proof that they must carry. It doesnt. The presumption is just that: a presumption. FERC provided several types of evidence that could rebut it, from non-utility stock prices to expert testimony.
3
Next, the customers claim that FERC created an irrebuttable presumption in this particular case by using the Return it had set in the first-complaint proceeding to adjudicate the second complaint. Their argument has two layers. First, they argue that it was unlawful for FERC to use the new Return (the 10.02% it had just set earlier in Opinion 569-A) instead of the pre-complaint 12.38% Return they had originally challenged. Second, they say that even if that was lawful, FERCs adjudication of both proceedings in one order denied them any meaningful opportunity to rebut the presumption because they didnt know what presumptively just and reasonable number they had to rebut. They are wrong on both fronts.
To the first point, Section 206 says:
Whenever the Commission, after a hearing held upon its own motion or upon complaint, shall find that any rate, charge, or classification, demanded, observed, charged, or collected by any public utility for any transmission or sale subject to the jurisdiction of the Commission, or that any rule, regulation, practice, or contract affecting such rate, charge, or classification is unjust, unreasonable, unduly discriminatory or preferential, the Commission shall determine the just and reasonable rate, charge, classification, rule, regulation, practice, or contract to be thereafter observed and in force, and shall fix the same by order.
16 U.S.C. § 824e(a) (emphases added).
Two aspects of the statute show that FERC was correct to use the 10.02% Return it had set earlier in Order 569-A when it resolved the first complaint.
9
First, the statute uses the present-tense verb “is,” which means that FERC must look to the current Return at the time of decision. See Carr v. United States, 560 U.S. 438, 447-48, 130 S.Ct. 2229, 176 L.Ed.2d 1152 (2010) (explaining the importance of verb tense). Second, the statute commands that FERC set a Return “to be thereafter observed and in force.” Once FERC sets a new Return in the first proceeding, it must observe and enforce that Return until it lawfully changes, including in ongoing proceedings.
On top of those points, the customers’ theory would upend the strict fifteen-month refund limit that Congress placed on Section 206 proceedings. 16 U.S.C. § 824e(b). If customers can just file new complaints challenging the same Return every fifteen months, the limit accomplishes nothing. Absent some clearer indication of congressional intent, we will “not assume that Congress left such a gap in its scheme.” Jackson v. Birmingham Board of Education, 544 U.S. 167, 181, 125 S.Ct. 1497, 161 L.Ed.2d 361 (2005).
10
To the customers’ second argument, there is some awkwardness in the fact that FERC chose to act on the first and second complaints in one order. But FERC has “broad discretion to manage” its docket. Florida Municipal Power Agency v. FERC, 315 F.3d 362, 366 (D.C. Cir. 2003) (cleaned up). And the customers do not point to any evidence that they would have marshaled to challenge the new 10.02% Return that they did not offer to challenge the old 12.38% one. So on these particular facts, we cannot conclude that FERC abused its broad discretion.
4
The customers’ final step-one challenge says that the presumption unlawfully creates a difference between Section 205 proceedings and Section 206 proceedings. But Congress required that difference. “Section 206s procedures are entirely different and stricter than those of section 205.” See Emera Maine, 854 F.3d at 24 (cleaned up).
D
Next, remember that for step two of the Section 206 analysis — setting the new just and reasonable rate — FERC returned to its customary practice of using the midpoint of the zone of reasonableness. The customers argue that it should have set aside the midpoint in favor of the median.
11
But we have already held that FERC can reasonably use the midpoint of the zone of reasonableness when setting a Return for “a diverse group of companies.” Public Service Commission, 397 F.3d at 1011. That decision, Public Service Commission of Kentucky v. FERC, even involved MISO. Id. at 1006.
The customers try to cabin Public Service Commission to Return analyses where FERC uses a proxy group made up of companies from within the same region as the transmission owners. But that was not the reason FERC chose the midpoint in Public Service Commission, so it is not the reason we deemed FERCs choice reasonable. Id. There, FERC focused on “the rates across-the-board applicability to MISO” transmission owners. Id. at 1011. FERC did the same here, so precedent requires that we reach the same result.
E
Finally, the customers challenge FERCs about-face on the risk-premium model (Model 4). As FERC applied it in this proceeding, the model compares past Returns that FERC itself set or approved to contemporaneous corporate-utility-bond rates. FERC took the difference between those rates and added it to the current corporate-utility-bond rate. So for example, if the past corporate-utility-bond rates were always 6% and Return rates were always 10%, FERC would take that difference (4%) and add it to the current corporate-utility-bond rate. If the current corporate-utility-bond rate is 5%, the new Return would be 9%.
12
See James Bonbright, Albert Danielsen & David Kamerschen, Principles of Public Utility Rates 323 (2d ed. 1988) (offering a similar example).
In FERCs first rehearing order, Opinion 569, it concluded that any “additional robustness” the risk-premium model added to its methodology was “outweighed by the disadvantages of its deficiencies.” JA 628. It then spent several pages demonstrating the impressive extent of those deficiencies. For example:
• The model, at least as applied in this case, “defies general financial logic” by keeping the Return stable regardless of capital-market conditions. JA 629.
• There was insufficient evidence in the record to conclude that investors rely on this kind of risk-premium model.
• The model is less accurate than the discounted-cash-flow model (Model 1) or capital-asset model (Model 2) because it relies on previous Return determinations that may not have been market-based.
• It “is largely redundant with” the capital-asset model (Model 2), so adding it would overweight risk-premium methodologies against the long-used discounted-cash-flow model (Model 1). JA 628.
• It presents “particularly direct and acute” circularity problems because it uses past FERC-allowed Returns to set the new ones. JA 628.
And those are just from the first two pages of criticisms. Suffice it to say that in Opinion 569, FERC found the risk-premium model quite defective.
Then, in Opinion 569-A — on rehearing of Opinion 569 — FERC changed its tune. It decided “that the defects of the Risk Premium model do not outweigh the benefits of model diversity” after all. JA 882.
FERC is, of course, entitled to change its mind. FCC v. Fox Television Stations, Inc., 556 U.S. 502, 515, 129 S.Ct. 1800, 173 L.Ed.2d 738 (2009). But to do so, it must provide a “reasoned explanation” for its decision to disregard “facts and circumstances that” justified its prior choice. Id. at 515-16, 129 S.Ct. 1800. Here, FERC failed to do that.
First and worst, FERC did not explain how its changes brought the analysis into line with “general financial logic.” JA 629. FERC cant ignore the basic financial principles that otherwise undergird its analysis — at least not without a compelling explanation. See Tennessee Gas Pipeline Co. v. FERC, 926 F.2d 1206, 1210-11 (D.C. Cir. 1991); id. at 1213 (Thomas, J., concurring) (“At the very least, FERC was obliged to offer some convincing evidence in support of its facially implausible economic assumption”).
Second, FERC failed to adequately explain why it no longer mattered that investors dont use this model. Instead, it simply noted that investors expect a premium on a stock investment over a bond investment, and that investors track the Returns FERC allows. Both statements are true, but neither offers a persuasive reason to think that the risk-premium model as FERC applied it here offers meaningful insight into investor behavior.
Third, FERC failed to meaningfully address its own concerns about the risk-premium models circularity. Instead, it just said that “all of the models contain some circularity” and decided that averaging the risk-premium models results with the other models’ results helps mitigate the circularity. JA 882. That explanation doesnt meaningfully engage with the “particularly direct and acute” circularity problems presented by using old rates to set new ones. JA 628.
Finally, FERC never engaged with its earlier concerns about the overweighting of risk-premium theory. It briefly discussed the redundancy of the capital-asset and risk-premium models (Models 2 and 4), saying that because they used different inputs to calculate the risk premium they were not too redundant to use. But it failed to reckon with its own serious concerns about “variations of the risk premium model” receiving twice the weight of the discounted-cash-flow model (Model 1) that FERC “has long used and, over time, refined.” JA 628. An agency ignoring its own qualms is not reasoned decisionmaking.
* * *
FERC failed to offer a reasoned explanation for its decision to reintroduce the risk-premium model (Model 4) after initially, and forcefully, rejecting it. Because FERC adopted that significant portion of its model in an arbitrary and capricious fashion, the new Return produced by that model cannot stand. We therefore vacate FERCs orders.
IV
In addition to the customers’ challenge to FERCs new Return methodology, the customers challenged FERCs determination that it could not order a refund for the second complaints refund period. But to the extent that any of that argument survives our earlier rejection of the customers’ statutory basis for their “irrebuttable presumption” argument, see Part III.C.3, we decline to opine on the customers’ argument because we have already granted their petition to vacate FERCs rate orders. See Southwest Airlines Co. v. FERC, 926 F.3d 851, 859 (D.C. Cir. 2019).
For the same reason, we dismiss the transmission owners’ petitions challenging those now-vacated orders. They had challenged FERCs right to require transmission owners to pay the difference between the amount FERC ordered in its first decision and the rate it ordered on rehearing. But because we vacate FERCs rehearing order, there is no longer a new rate to base a refund on.
Until FERC sets a new Return, a decision on the refund issue will not alter the parties’ rights and obligations. Nor will a decision on the transmission owners’ argument that FERC lacked the authority to adjudicate the second complaint. When “it is not necessary to decide more, it is necessary not to decide more.” PDK Laboratories, Inc. v. DEA, 362 F.3d 786, 799 (D.C. Cir. 2004) (Roberts, J., concurring in part and in the judgment).
V
We grant the customers’ petitions for review, dismiss the transmission owners’, vacate the underlying orders, and remand for FERC to reopen proceedings.
FOOTNOTES
1
. As we said, r = D/P + g is a simplified version of FERCs formula. The actual, more complicated formula includes a dividend multiplier, which accounts “for the fact that dividends are paid on a quarterly basis.” JA 514. It is r = D/P(1 + .5g) + g. But because the dividend multiplier affects none of the analysis in this case, well use the simplified formula when discussing the discounted-cash-flow model.
2
. The Return for this model depends on, among other things, a risk-free rate like the Treasury-bond rate, an analysis of the returns in the market, and an estimate of the companys riskiness. Part III.B of this opinion explains it in more detail.
3
. This model produces a Return based on the earnings investors in comparable stocks expect to receive based on those stocks’ “book value,” which measures the difference between a companys assets and liabilities. Spoiler alert: FERC will later drop this model from its methodology.
4
. This model subtracts past corporate-utility-bond rates from past Returns to calculate an average risk premium that FERC has given in the past. The new Return is that number added to the current Treasury-bond rate. We will explain more about this model in Part III.E.
5
. The MISO transmission owners’ primary challenge focuses on the lawfulness of this backdating decision. Because we do not reach that question, we wont delve into the sides’ conflicting positions.
6
. Specifically, the beta looks at risk as compared to the full market. So an investment that “fluctuates exactly in step with the market,” which means that the investments “rate of return increases on average by 1 percent when the markets return increases 1 percent,” will have a beta of one. A. Lawrence Kolbe, James Read, Jr. & George Hall, The Cost of Capital: Estimating the Rate of Return for Public Utilities 70 (1984).
7
. For some concrete examples of that formula in action, imagine three companies with slightly different risk profiles at a time when (1) the risk-free rate is 3% and (2) the discounted-cash-flow analysis of dividend-paying companies in the S&P 500 produces an expected return of 10%. Lets calculate the three companies’ Returns using the formula above:• If Company A has a completely average risk profile, its Beta is 1. So Company A has a Return of 10%. Thats because 10 = 3 + 1(10 - 3).• Say Company B is slightly riskier than Company A. If its Beta is 1.05, then its Return is 10.35%. Thats because 10.35 = 3 + 1.05(10 - 3).• Finally, say Company C is slightly safer than Company A. If its Beta is 0.95, then its Return is 9.65%. Thats because 9.65 = 3 + 0.95(10 - 3).
8
. In a more recent proceeding FERC did have access to adjusted betas based on the S&P 500, so it used them. Constellation Mystic Power, LLC, 176 FERC ¶ 61,019, 61,102 (2021).
9
. Although the statute uses “rate,” in this case the only component of the rate that was at issue was the Return, so that is what FERC focused on.
10
. This should not be read to endorse the transmission owners’ argument that customers cannot file successive complaints. FERC has an explanation for allowing successive complaints that it says reconciles the practice with this provision. Because we decide that FERC was correct to use the Return from the first complaint to adjudicate the second, and therefore that FERC was right to dismiss the second complaint, we need not decide this issue.
11
. For a series of numbers, the midpoint is the halfway point between the biggest number and the smallest number (calculated by adding the two together and dividing by two). The median is the middle number in the series. So, for example, the midpoint of 1, 3, 5, 9, and 11 is 6. The median is 5.
12
. This explanation omits one step that no one questions, which is therefore not relevant to our analysis. Before FERC adds the risk premium it calculated from the past corporate-utility-bond rates and Returns to the current bond rate, it adjusts that number to “reflect the tendency of risk premiums to rise as interest rates fall.” JA 249. Basically, it calculates the inverse relationship between bond yields and risk premiums to determine how much higher the risk premium needs to be to incentivize investment when the bond rate is lower. Here, for example, the calculation determined that for every 1% bond rates dropped, investors required an extra .77% Return. So when the bond rate had dropped by 1.35%, FERC multiplied 1.35 and .77 to get an adjustment of 1.04%, which it added to the average difference between the past bond rates and past FERC-allowed Returns. It then added the sum of those numbers to the current corporate-utility-bond rate to get the value for what the new Return should be.
Walker, Circuit Judge: