Keyspan-Ravenswood, LLC v. Federal Energy Regulatory Commission

Notice: This opinion is subject to formal revision before publication in the
Federal Reporter or U.S.App.D.C. Reports. Users are requested to notify
the Clerk of any formal errors in order that corrections may be made
before the bound volumes go to press.

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued October 7, 2003 Decided November 18, 2003

No. 02-1115

KEYSPAN–RAVENSWOOD, LLC,
PETITIONER

v.

FEDERAL ENERGY REGULATORY COMMISSION,
RESPONDENT

CONSOLIDATED EDISON COMPANY OF NEW YORK, INC., ET AL.,
INTERVENORS

Consolidated with
Nos. 02-1125, 02-1150

On Petitions for Review of Orders of the
Federal Energy Regulatory Commission

Elaine M. Walsh argued the cause for petitioners. With
her on the briefs were Kenneth M. Simon, M. Eric Eversole,
Mitchell F. Hertz, and Ashley C. Parrish.
Bills of costs must be filed within 14 days after entry of judgment.
The court looks with disfavor upon motions to file bills of costs out
of time.
 2

Robert H. Solomon, Deputy Solicitor, Federal Energy Reg-
ulatory Commission, argued the cause for respondent. On
the brief were Cynthia A. Marlette, General Counsel, Dennis
Lane, Solicitor, and Timm L. Abendroth, Attorney.
Neil H. Butterklee argued the cause and filed the brief for
intervenor Consolidated Edison Company of New York, Inc.
Lawrence G. Malone and Michelle L. Phillips were on the
brief for intervenor Public Service Commission of the State of
New York. Jonathan D. Feinberg entered an appearance.

Before: SENTELLE, RANDOLPH and ROGERS, Circuit Judges.
Opinion for the Court filed by Circuit Judge ROGERS.
ROGERS, Circuit Judge: These consolidated petitions chal-
lenge the manner in which the Federal Energy Regulatory
Commission calculated a price cap for the New York City
electric capacity market when it authorized the New York
Independent System Operator (‘‘NYISO’’) to change its pric-
ing methodology. The NYISO was allowed to account for
forced outages by measuring the amount of electric generat-
ing capacity available for sale to the system (‘‘UCAP’’) rather
than installed generation capacity (‘‘ICAP’’), and the Commis-
sion adjusted the price cap to yield approximately the same
revenues from affected sales at the time of conversion to the
new methodology. At issue is the Commission’s determina-
tion that, in shifting to the new methodology, the most recent
twelve months constitute the appropriate period to estimate a
generating unit’s availability for the purpose of recalculating
the price cap. KeySpan–Ravenswood, LLC and Orion Power
New York GP, Inc., both electricity suppliers affected by the
price cap, petition for review of three orders in which the
Commission rejected their position that a longer period of
time was required. We find the Commission did not ade-
quately explain its decision. We therefore grant the petition.
 3

I.
The Commission has capped prices in the New York City
capacity market since 1998, when Consolidated Edison sold
generators serving the city to private energy suppliers, see
Consolidated Edison Company of New York, Inc., 84 FERC
¶ 61,287 (1998). In 2001, the NYISO filed a request to amend
its service tariff so that it could implement a market design
based on a shift in its methodology for measuring electric
generator capacity, see 16 USC § 824d (2003), 18 C.F.R.
§ 35.13 (2003), and requested that the Commission determine
the appropriate translation of the price cap in light of the new
methodology. The proposed translation of the price cap was
to reflect the shift from measuring capacity on the basis of
installed capacity (‘‘ICAP’’) to unforced capacity (‘‘UCAP’’).
The ICAP methodology calculates the amount of capacity a
supplier can sell based on the ideal performance of its genera-
tors, whereas the UCAP methodology accounts for the proba-
bility that a generating unit will be called upon to produce
energy but will be unable to do so because of ‘‘forced outag-
es,’’ i.e., unforeseen circumstances resulting in a generating
unit’s production of less than maximum net capacity. UCAP
thus requires predicting how often generators will be forced
out of service. The parties agree that, for the purpose of
determining UCAP available for sale, a 12-month rolling
average is appropriate. However, they disagree about what
period the Commission should use to determine UCAP for
the purpose of translating the price cap, which is fixed and
cannot be adjusted as outage rates fluctuate. The relevant
generators in New York City had performed far better in the
immediately preceding two years than in the years before
that, so a 1-year or 2-year history would predict an ‘‘equiva-
lent forced outage rate’’ of 6.92% or 6.59%, respectively,
whereas a 3-year history suggested a much higher rate of
12.58%.
The Commission published notice of the NYISO tariff
filing, see New York Independent System Operator, Inc.;
Notice of Filing, 66 Fed. Reg. 37,663 (July 19, 2001), and
received comment on the appropriate way to translate the
price cap. The higher the predicted forced outage rate, the
 4

higher the price cap necessary to offset the corresponding
drop in available capacity from ICAP to UCAP. Electricity
retailers (as well as the City) wanted the more current forced
outage data to be used, because that would result in a lower
price cap, and lower costs to consumers. Petitioners called
upon the Commission to look at a longer period of data
preceding their acquisition of the relevant generators, which
would have resulted in a higher cap, arguing that they should
be allowed to reap the benefits of reliability investments they
had made, and that the maintenance cycles of generators
necessitate use of a multi-year average to smooth out anoma-
lies. Each charged that translating the price cap using the
other’s methodology would result in a windfall to the other.
In the first order on review, the Commission authorized the
NYISO to change its methodology for measuring available
electric capacity to UCAP, and determined to use only the
past twelve months of data as a predictor of future outage
rates. Order Accepting Tariff Revisions and Directing
Translation of the In–City Price Cap (‘‘Order’’), 96 FERC
¶ 61,251 (2001). The Commission explained that the purpose
of the order was simply to translate the price cap, not to
change it, that the ‘‘translation TTT must be revenue neutral,’’
and that any arguments about changing the effective price
cap were ‘‘beyond the scope of this proceeding.’’ Id. at
61,994. The New York Commission had urged use of only
twelve months of forced outage data to ensure that ‘‘suppliers
do not derive financial benefits solely as a result of a change
of methodology.’’ Id. at 61,992. In its Order, the Commis-
sion rejected use of outage data from the period prior to the
divestiture, which would have resulted in a $126.14/kW-year
price cap, as compared to the pre-translation cap of $105/kW-
year, stating that the translation ‘‘must be based on operating
data from the most recent 12 months, as they reflect a more
current outage rate.’’ Id. at 61,994. The Commission reject-
ed petitioners’ argument that a price cap that incorporated
post-divestiture outage data would confiscate investments
they had made since acquiring the facilities in question,
observing that the price cap was set before the divestiture
and potential purchasers ‘‘were afforded an opportunity to
 5

adjust their bids for the generation being divested by the
amount necessary to compensate them for effects of mitiga-
tion measures.’’ Id.
Petitioners sought rehearing, renewing their confiscation
argument and arguing additionally that using only twelve
months of data is less accurate than a longer period of at
least five years because of year-to-year anomalies that
smooth out across several years. Prior to denying rehearing
the Commission requested that the NYISO supply data on
the outage rates using a 1, 2, or 3-year period. In the second
order on review, the Commission restated its reasons for
using only twelve months of data and made two additional
points: (1) the petitioners had misread its Order as changing
the in-city price cap, and (2) a 12-month period of forced
outage data is used throughout the state for calculating the
amount of UCAP available for sale. Order on Rehearing, 98
FERC ¶ 61,180, 61,665–66 (2002). Petitioner Orion sought
rehearing. In the third order on review, the Commission
denied rehearing as it had already dealt with Orion’s argu-
ments, and stated that NYISO data for 24 months indicated
no significant change from the twelve month average forced
outage rate and that the 36 month data were not ‘‘relevant to
the time period during which [petitioners] had operational
authority.’’ Order on Rehearing, 99 FERC ¶ 61,072, 61,335
(2002).

II.
On review, petitioners principally contend that the Commis-
sion never explained why their substantial objections to the
twelve month data limitation are wrong, and that the Com-
mission’s statement that twelve months of data are most
current begs the question by failing to justify why this limited
period is better than a longer period for reflecting the most
accurate prediction of usable capacity. They do not challenge
the inclusion of post-divestiture data in the calculation of the
forced outage rate, as they initially had before of the Commis-
sion, only the failure to include forced outage data from a
longer time period. The Commission responds that its judg-
 6

ment on this ‘‘rate design’’ is entitled to deference, because it
involves balancing interests at the core of its regulatory
function, that the Commission appropriately adhered to the
principle of revenue neutrality in the conversion from ICAP
to UCAP, and that the petitioners’ position is patently unrea-
sonable as capacity revenues (and rate payer costs) would
increase substantially at the time of the UCAP conversion
over what would have been expected had the ICAP methodol-
ogy been retained.
The court reviews whether the Commission engaged in
reasoned decisionmaking under the arbitrary and capricious
standard, which requires the Commission to ‘‘respond mean-
ingfully to the evidence,’’ for ‘‘[u]nless an agency answers
objections that on their face appear legitimate, its decision
can hardly be said to be reasoned,’’ Tesoro Ala. Petroleum
Co. v. FERC, 234 F.3d 1286, 1294 (D.C. Cir. 2000). The
underlying question, on which the reasonableness of the
Commission’s decision to use twelve months of data to trans-
late the price cap turns, is why the forced outage rates for the
1-year and 2-year periods are so much lower than those using
data averaged across 3 years or longer. The rate effectively
doubles from a 2-year to a 3-year average, and the parties
have suggested different explanations for why the rates vary
so much.
Petitioners contend that the variance is due in large part to
the maintenance cycles of generating units. Repair work
forces generators out of service, and major repairs that take a
generator offline for extended periods of time do not happen
every year. Therefore, a year in which major maintenance is
done will reflect an artificially high forced outage rate, and a
year in which no major maintenance is done will reflect an
artificially low forced outage rate. The consequence of this
theory is that, assuming the generators in the relevant mar-
ket are on somewhat similar maintenance cycles, the average
of several years (long enough for a generating unit’s full
maintenance cycle) is required to get an accurate prediction
of how often generators are forced offline, and using 12
months of data risks under- or over-estimating the forced
 7

outage rate, depending on when in the cycle the calculation is
made.
Evidence in the record also suggests that the variance
might be attributable to repairs and upgrades made by
petitioners since acquiring the relevant generating units from
Consolidated Edison in 1999. Petitioners initially argued
before the Commission that they had made significant im-
provements to the generators they had acquired. The conse-
quence of this theory is that post-1999 forced outage rates
may be better indicators of future outage rates because they
reflect the physical state of the electric plants as they are
today, rather than as they were before substantial mainte-
nance and repair work was done. However, the predictive
usefulness of such data might depend on the extent to which
any investments yielded permanent efficiency gains, as op-
posed to front-loading maintenance schedules bound to recur
in a few years.
Further, the record contains evidence suggesting that the
difference in the outage rates may be a product of changed
market circumstances. Before deregulation, there was argu-
ably less of a regulatory incentive for Consolidated Edison to
keep surplus capacity available when energy demands were
being met, whereas the new owners now have a market
incentive to keep capacity continuously available to maximize
the amount they can sell. The New York Public Service
Commission, for instance, argued to the Commission that the
regulatory structure before the divestiture caused data on
outages to be kept differently, so pre-1999 data are ‘‘stale.’’
The consequence of this theory is that data on forced outages
before 1999 are of little value in predicting future outage
rates.
These explanations are not logically inconsistent with each
other, and it may be that each is partly responsible for the
swing, or that one explains the variance in full and the others
are completely wrong. Or another explanation may account
for near-doubling of the forced outage rates between the 2-
year and 3-year averages. It matters which explanation is
adopted by the Commission. If the maintenance cycle causes
 8

the forced outage rates to vary as petitioners contend, the
Commission has translated the price cap in a way that
deprives them of anticipated revenue streams at the time of
purchase of the in-city generators. Because the amount of
capacity that the suppliers are actually allowed to sell is
recalculated continuously (on a 12-month rolling average), the
petitioners point out that if the forced outage rates swing up
again, they will not be able to sell enough capacity to bring in
the same revenues they had under the previous ICAP system
(i.e. the transition will not have been ‘‘revenue neutral’’). If
this explanation for the variations is plausible, then the
Commission’s decision to only use one year of data looks
arbitrary. The level of the price cap would, as petitioners
argue, depend entirely on whether it is set at a time when the
maintenance cycle is at a high point or a low point, and that
timing bears no relationship to the purpose of the cap itself.
The Commission cannot reasonably base its judgment on a
criterion if that criterion bears ‘‘no relationship to the under-
lying regulatory problem,’’ see ALLTEL Corp. v. FCC, 838
F.2d 551, 559 (D.C. Cir. 1988) (quoting Home Box Office, Inc.
v. FCC, 567 F.2d 9, 60 (D.C. Cir. 1977)).
On the other hand, if the variance is a product of the
changed market structure and recordkeeping, the forced out-
age rate over the next few years will likely be the same as it
has been in the past year, and pre-divestiture data are of
limited usefulness. Thus, a UCAP price cap based on a one-
year forced outage rate ought to keep revenue fairly identical
to what it was under the old ICAP system. If true, the
Commission accomplished its stated goal of ‘‘revenue neu-
tral[ity]’’ adequately. Similarly, if physical improvements to
the generators explain the lower forced outage rate over the
past two years, then post-improvement figures would seem to
be a more accurate predictor of future outage rates than a
multi-year average that factors in pre-divestiture data, de-
pending on the extent to which those physical improvements
represent permanent efficiency gains as opposed to simply
front-loading maintenance work bound to recur in a few
years. Although petitioners argued to the Commission that if
their investments are what caused the outage rates to fall,
 9

then using post-divestiture data has the effect of penalizing
them for improving the reliability of their generators, on
review they do not challenge the Commission’s decision to
include recent data in the calculation of the forced outage
rates, only the decision not to account for a longer time
frame.
It is obviously not the role of the court to decide why the
forced outage rates recently dropped. However, the petition-
ers contend the Commission did not adequately respond to
their argument that the maintenance cycle theory is correct.
The Commission’s treatment of petitioners’ objections was
quite curt. The Commission essentially relied on two rea-
sons: (1) The twelve-month data are ‘‘more current,’’ and
reflect the time over which the current owners ‘‘had opera-
tional authority;’’ and (2) The twelve-month data are also
what are used statewide to calculate how much capacity a
generator has available for sale. The second rationale is
somewhat of a nonsequitur: the sales allowance is continuous-
ly recalculated while the price cap is not — precisely petition-
ers’ point that as the maintenance cycle swings, future recal-
culations may cause the sales allowance to drop but the price
cap will be too low to compensate for the drop. As to the
first rationale, the Commission’s statement that the past 12
months was during the period in which the petitioners had
‘‘operational authority,’’ see Order on Rehearing, 99 FERC
¶ 61,072 at 61,335, appears to be a nod to either the ‘‘physical
improvements’’ explanation or the ‘‘market structure’’ expla-
nation for the drastic change in the equivalent forced outage
rates between 1998-1999 and 1999-2000. The Commission’s
implication seems to be that something about deregulation
and divestiture has caused a permanent improvement in
efficiency, and that the forced outage rates will not rise to
pre-1999 levels again. This may be the case, but the Com-
mission did not explain whether it had adopted this theory,
and if so, why.
The record evidence might support the notion that new
market conditions or the changed physical state of the gener-
ating units make pre-divestiture forced outage data unrelia-
ble. Consolidated Edison submitted the affidavit of Robert
 10

B. Stoddard claiming that when it managed the generators at
issue, it had an incentive to shut down units whose output was
not needed, and that those shutdowns (which private owners
do not do because they want to maximize the capacity avail-
able for sale) artificially raised the forced outage rate for
those years. Further, petitioners submitted to the Commis-
sion that they had invested in reliability improvements since
acquiring the facilities, lending plausibility to the theory that
the recent drop in the forced outage rates is due, at least in
part, to their improvements. Yet despite record evidence of
both theories, the Commission did not explain which, or what
other theory, it was adopting, thereby denying petitioners the
chance to respond to its reasoning.
On review, the Commission maintains that its orders make
clear that its choice among contending price cap translation
proposals was guided by the revenue neutrality principle, by
an effort to reflect current rather than stale outage condi-
tions, and by a desire to achieve consistency in UCAP treat-
ment throughout the state. However, in the orders on re-
view, the Commission did not respond to the petitioners’
argument and evidence that the maintenance cycle makes
twelve months too short for predicting future outage rates
accurately. Rather, the Commission stated only that twelve-
month data are more ‘‘current’’ and mirror the data used to
calculate the sales-allowance, see Order on Rehearing, 99
FERC at 61,335-36, even after receiving data from the NYI-
SO that revealed the large jump from 6.59% to 12.58%
between the two-year and three-year average. The Commis-
sion stated that it ‘‘considered [the petitioners’] viewpoint and
TTT disagreed.’’ Id. at 61,336. From all the court can tell
there may be good reasons to disagree with petitioners’
maintenance-cycle theory, and why more recent data are
better predictors of future outage rates, but the Commission
did not supply them. The NYISO asserted in its answer to
petitioners’ request for rehearing, and the Public Service
Commission of New York contends as intervenor before the
court, that even the one-year average is composed of enough
generators that variance due to maintenance cycles is aver-
aged out. Nothing in the Commission’s orders reflects adop-
 11

tion of this line of reasoning. There also is an assertion by
the New York Public Service Commission that Consolidated
Edison, which ran the facilities until 1999, recorded forced
outages differently than under current rules but this does not
appear as the Commission’s explanation in the orders on
review.
The Commission contends that the petitioners cannot claim
hardship from the new price cap because they knew about the
price cap when they purchased the relevant generating units
in 1999. This appears to be responsive to petitioners’ argu-
ment before the Commission that post-divestiture forced out-
age data should be disregarded so as not to penalize them for
investing in improved efficiency, but not to the question of
whether the Commission effectively lowered the price cap in a
proceeding the stated purpose of which was to translate it.
Intervenor Consolidated Edison’s contention that petitioners
lack standing because they did not challenge the ‘‘revenue
neutrality’’ principle before the Commission relies upon the
same misunderstanding; petitioners’ contention is that the
Commission’s methodology actually lowered the price cap,
and therefore did not comport with the Commission’s stated
goal of revenue neutrality.
In its brief on review the Commission also maintains that it
is sensible to use only twelve months data to predict the
future outage rate for price cap purposes because only twelve
months of data are used to calculate how much ‘‘UCAP’’ each
supplier has available for sale. Thus the amount suppliers
are allowed to sell will be reduced by the exact amount that
the price cap is increased. Other than the cryptic statement
that a 12-month period ‘‘ensures that the UCAP conversion
terms are consistent throughout the New York State,’’ Order
on Rehearing, 98 FERC ¶ 61,180 at 61,666, the Commission
did not adopt this rationale, however, and ‘‘post hoc salvage
operations of counsel’’ cannot overcome the inadequacy of the
Commission’s explanation. Florida Power & Light Co. v.
FERC, 85 F.3d 684, 689 (D.C. Cir. 1996), see generally SEC v.
Chenery Corp., 332 U.S. 194, 196-97 (1947). In any event, the
perfection of such a translation is hardly obvious: the price
cap remains at a fixed level until the Commission changes it,
 12

while the amount of UCAP that sellers are allowed to sell
keeps changing and can go up or down based on a 12-month
rolling average. If the petitioners are correct that the Com-
mission set the price cap when the forced outage rates were
below-average in their cycle, petitioners will not always have
as much UCAP available as they did the month of the
translation, and will lose revenue in the long run. The
question is whether the Commission used data that accurately
predict forced outage rates in the future, rather than simply
data that reflect those rates during the month of translation.
If not, the Commission has changed the price cap without the
requisite inquiry into whether the new rate is just and
reasonable, see 16 U.S.C. §§ 824d(a), (b); 18 C.F.R. § 35.13.
Consolidated Edison maintains that because the price cap’s
purpose is simply to prevent the improper exercise of market
power (rather than to guarantee any particular revenue
stream to petitioners), it would not matter even if the peti-
tioners are correct that the Commission chose a value that
does not fairly predict future outage rages: the price cap is
there to protect consumers, not guarantee a revenue stream.
This appears to be an argument that the validity of the price
cap translation does not depend on its effect on petitioners’
revenues. The scope of the Commission’s order was to
translate the price cap, not to change it, and it explained its
translation on that basis, referring to any change in the cap
as being ‘‘beyond the scope of this proceeding.’’ Order at
61,994. If the translated price cap was supposed to be
determined by consumer-protection rationales rather than on
the old price cap, petitioners presumably would have intro-
duced different evidence before the Commission, such as, for
example, evidence about the market structure and their own
price-setting market power. In any event, the Commission
relied nowhere in its orders on the rationale suggested by
Consolidated Edison.
For these reasons, we hold that the Commission did not
adequately explain why twelve months of historical data
would accurately reflect outage rates for use in translating
the price cap from ICAP to UCAP, nor why the two peri-
ods — the amount suppliers are allowed to sell and the
 13

amount the price cap is increased — must be based on the
same twelve months of data. It may be justifiable, for
purposes of establishing revenue neutrality, that the period
used to predict the availability of capacity should commence
with the petitioners’ acquisition of the equipment because
they are likely to be more efficient and therefore have
increased revenues, but that was not explained in the orders
on review. Although the petitioners did not specify in their
petitions that the five to seven year period they requested be
the most recent five to seven years (as opposed to the five to
seven years preceding divestiture), their rehearing request
can be construed in that fashion, and the Commission did not
explain why it was denying rehearing or state that the
petition did not adequately ask for relevant periods. Peti-
tioners presented a serious argument that a period long
enough to account for generators’ maintenance cycle should
have been employed to calculate the forced outage rate to be
factored into the translated price cap, and the Commission
neither responded nor based its decision on a procedural
default. On remand, petitioners can, if the Commission
deems it relevant, present evidence regarding the timing of
their generators’ maintenance cycles, to the extent it bears
upon whether a 12-month period sufficiently smooths out the
fleet-wide average of forced outages. Accordingly, we vacate
the orders and remand the cases to the Commission.