dissenting.
The majority does something truly remarkable: It holds that a regulation which was valid when adopted became invalid because it wasn’t adjusted for inflation. There are many problems with this ruling, the most fundamental of which is that we have no jurisdiction to consider the issue. No action of the agency caused this regulation to lose its validity, but judicial review under the Administrative Procedure Act extends only to agency “action.” See 5 U.S.C. § 702. The applicable substantive statute, moreover, gives us no standard against which to judge the legality of the agency’s non-action, but the APA forbids judicial review unless there is a meaningful legal standard. See 5 U.S.C. § 701(a)(2); Heckler v. Chaney, 470 U.S. 821, 832, 105 S.Ct. 1649, 1656, 84 L.Ed.2d 714 (1985). Were the merits properly before us, we would have no choice but to uphold the regulation as a sensible application of what statutory guidance there is. Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Finally, the majority not only raises a direct conflict with every other circuit that has considered the issue, see Brown v. Secretary of Health and Human Serv., 46 F.3d 102 (1st Cir.1995); Hazard v. Shalala, 44 F.3d 399 (6th Cir.1995); Champion v. Shalala, 33 F.3d 963 (8th Cir.1994); Falin v. *1352Shalala, 6 F.3d 207 (4th Cir.1993) (per curiam), cert. denied, — U.S. -, 114 S.Ct. 1551, 128 L.Ed.2d 200 (1994), it actually overrules them by precluding the agency from applying the regulation even where the regional circuit has upheld it. Maj. op. at 1351 (vacating the regulation). Few opinions have taken us so far afield.
Jurisdiction
The APA does not give us a roving commission to second-guess agency judgments; it is a limited grant of authority to review agency “action.” In the run-of-the-mill APA case, the reviewable action isn’t hard to find because the agency has formally adopted a regulation. See, e.g., Chevron, 467 U.S. at 840, 104 S.Ct. at 2780. Less commonly, an agency’s refusal to adopt a regulation may amount to a reviewable action. For example, the APA allows interested parties to petition for rulemaking, see 5 U.S.C. § 553(e), and requires the agency to give an explanation if it denies the petition, see 5 U.S.C. § 555(e). Such a denial can be an action subject to judicial review. See American Horse Protection Ass’n v. Lyng, 812 F.2d 1, 4 (D.C.Cir.1987).
. The rule of American Horse Protection Ass’n doesn’t apply here, however, because plaintiffs never petitioned the agency for a change in the regulation.1 Thus, the only agency action pertaining to the regulation was its initial adoption in 1982, which the majority holds was valid. Maj. op. at 1344-1346. It was only later — at a time the majority does not identify — that the regulation supposedly lost its validity. But there was nothing the agency did to cause this; rather, it was caused by macroeconomic forces which reduced the purchasing power of the dollar. But standing still in a changing world surely isn’t the type of “action” that triggers judicial review under the APA.2
Plaintiffs’ failure to petition the agency for a modification of the regulation is thus far more than a failure to exhaust; it is a jurisdictional defect because it denies us the irreducible minimum for APA review of agency action, namely, agency action.3
But there is a second jurisdictional problem: Were there agency action to review, we would have no legal yardstick by which to review it. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 410, 91 S.Ct. 814, 820-21, 28 L.Ed.2d 136 (1971).4 According to the applicable statute, a family’s eligibility for AFDC benefits must be determined by excluding from the calculation of resources “so much of [its] ownership interest in one automobile as does not exceed such amount as the Secretary may prescribe.” 42 U.S.C. § 602(a)(7)(B) (emphasis added). As the First Circuit observed in rejecting an identical challenge to this regulation, there is a “total absence of any standards within the statute.” Brown, 46 F.3d at 111.
The statutory language here is as vague as that in Chaney, where the Supreme Court held unreviewable the FDA’s decision not to *1353challenge states that used prescription drugs in lethal injections. See 21 U.S.C. § 372(a) (“The Secretary is authorized to conduct examinations and investigations”); 21 U.S.C. § 334 (offending food, drugs or cosmetics “shall be liable to be proceeded against” in seizure actions). The language is vaguer than that in Webster v. Doe, 486 U.S. 592, 600, 108 S.Ct. 2047, 2052, 100 L.Ed.2d 632 (1988), which held that the federal courts have no authority to review the Director of Central Intelligence’s decision to fire an employee under 50 U.S.C. § 403(c) (termination authorized whenever the Director “shall deem such termination necessary or advisable in the interests of the United States”). The language here is as vague or vaguer than that in other cases holding that review was barred by section 701(a)(2). See, e.g., Flint v. United States, 906 F.2d 471, 474-76 (9th Cir.1990) (Secretary of Interior may enter into water-sale contracts “at such rates as in the Secretary’s judgment will produce revenues at least sufficient to cover an appropriate share of the annual operation and maintenance cost and an appropriate share of such fixed charges as the Secretary deems proper”); State of Nevada v. Watkins, 914 F.2d 1545, 1563 (9th Cir.1990) (Secretary of Energy shall act “[i]f [he] at any time determines” that a location is unsuitable for use as a waste depository), cert. denied, 499 U.S. 906, 111 S.Ct. 1105, 113 L.Ed.2d 215 (1991); Slyper v. Attorney General, 827 F.2d 821, 823 (D.C.Cir.1987) (two-year waiting period shall be waived “upon the favorable recommendation of the Director of the United States Information Agency”), cert. denied sub nom., Slyper v. Meese, 485 U.S. 941, 108 S.Ct. 1121, 99 L.Ed.2d 281 (1988). The majority’s attempt to review the regulation in the absence of a meaningful statutory standard thus conflicts with our own caselaw and that of other courts, including the Supreme Court.
The jurisdictional rules the majority discards are not hollow formalities; they are integral parts of the mechanism giving us authority to review agency action. Where these rules are followed, they set the stage for meaningful judicial review. Thus, when there is agency action, it is normally accompanied by a statement of reasons and an evidentiary record — two things sorely lacking here. See pp. 1356-1358 infra.5 Moreover, a statutory standard enables the court to determine the rationality of the Secretary’s action, as any action may be rational or not, depending on the standard by which it is measured. Here, there’s no standard. As the First Circuit recognized in Brown, Congress gave “to the Secretary, and to no one else, the unqualified authority to prescribe the amount of the automobile resource exemption.” 46 F.3d at 107.
Because there is no action to review, nor any standard by which to review it, that’s the end of the case for me. We needn’t speculate as to what might have animated the Secretary had she been asked to make a decision. Nevertheless, because the majority’s rationale for striking down the regulation is so plainly wrong and so dangerous, I proceed to address the most obvious flaws in its analysis.6
*1354 The Merits
The majority’s analysis is a textbook illustration of why courts are ill-equipped to conduct judicial review in the absence of a meaningful statutory standard. Finding no law to apply, my colleagues plunge deep into questions of policy. Thus, they recognize that Congress intended to cut spending in 1981, but then take the curious view that the cost-cutting was not meant to continue past that year: “While it is true that the purpose of OBRA was to reduce spending, there is no evidence to suggest that Congress intended to cut spending progressively past 1981.” Maj. op. at 1347. Nor, according to the majority, is there any “indication in OBRA or its legislative history that Congress intended to use the automobile equity limit itself as a cost-cutting measure.” Id. Hunting for a limitation on the Secretary’s authority that simply does not exist, the majority finds it necessary to balance OBRA’s cost-cutting purpose with AFDC’s purpose of “foster[ing] ‘self-support and personal independence’ for needy families.” Id. Based on its own balancing of these two policy objectives, the majority concludes that “periodically adjusting the automobile equity limit for inflation is the only reasonable way to achieve Congress’s competing goals.” Id. (emphasis added).
Finding the bounds of rationality so narrowly circumscribed is certainly unusual. The statute itself contains no such requirement, even though Congress in 1981 was well aware of the effects of inflation. In 1981, Members of Congress — like everyone else— had lived through the runaway inflation of the Vietnam era, President Nixon’s wage and price controls, President Ford’s WIN (Whip Inflation Now) program, President Carter’s “misery index” and interest rates in the 20% range. The erosion of purchasing power caused by years of double-digit inflation, and its effect on everyone, especially the poor, was headline news for some time prior to OBRA’s enactment. See, e.g., Harry Anderson et al., The Gold Rush of’79, Newsweek, Oct. 1, 1979, at 48 (“According to most analysts, the U.S. economy is in — or at least on the brink of — a recession created as much as anything by a huge inflationary drain on consumer purchasing power.”); Nancy L. Ross, Small Business Capital Investment Aid Proposed, Wash. Post, Feb. 11, 1978, at C12 (“Alan Greenspan, former chairman of the Council of Economic Advisors [warned that] [ujnless the purchasing power of the dollar can be stabilized [there could be] a massive run on the dollar by foreign holders, resulting in a rapid increase of interest rates around the world, and especially in this country.”); A.F. Ehrbar, The Trouble With Stocks, Fortune, Aug. 1977, at 89 (“Inflation was a ... serious matter, as ... the reduction in the purchasing power of the dollar cut the real rate of return [on stocks from 7 percent] down to a meager 2.2 percent.”); The Squeeze on the Middle Class, Business Week, Mar. 10, 1975, at 55 (“Over-all, the inflationary squeeze brought a 3.4% drop in last year’s real per-capita disposable personal income — the first such drop since 1958. Obviously, the poor feel the tightest pinch.”).
Members of the 97th Congress not only knew about inflation, they grappled with it. “Bracket-creep” may have little meaning to anyone who came of age in the 1980s and thereafter, but to those among us who paid income taxes in the ’60s and ’70s, the term has real bite. The reason for this generation gap is that the very same Congress that passed OBRA — -within the space of a week— passed the Economic Recovery Tax Act of 1981. Compare 1981 U.S.C.C.A.N. 396 (OBRA passed both houses of Congress in final form by July 31, 1981) with 1981 U.S.C.C.A.N. 105 (ERTA passed both houses of Congress in final form by August 4, 1981). *1355ERTA did away with bracket-creep by weaving inflation-indexing into the fabric of the Internal Revenue Code. As the House Conference Report accompanying ERTA explained:
Under present law, the individual income tax is based on various fixed amounts including the amounts that define the tax brackets, the zero bracket amount, and the personal exemption. These amounts are set by statute and are not adjusted for inflation. [The conference agreement is thatl the income tax brackets, zero bracket amount, and personal exemption are adjusted for inflation (as measured by the Consumer Price Index), starting in 1985.
H.R. Conf. Rep. No. 215, 97th Cong., 1st Sess. (1981), reprinted in 1981 U.S.C.C.A.N. 285, 290.
Even as it passed OBRA, then, Congress was working on major legislation that contained an inflation adjustment mechanism. Over the years, Congress has passed several other such statutes, usually including indexing in the legislation, see, e.g., 7 U.S.C. § 2014(g)(2) (food stamp program vehicle exemption); 42 U.S.C. § 415(i) (Social Security benefits); 29 U.S.C. § 720(c) (vocational rehabilitation grants); 20 U.S.C. § 1087rr(b) (student aid); 20 U.S.C. § 2007 (stipends for federal scholarship recipients); 5 U.S.C. § 8340 (retired federal employees’ annuities); 10 U.S.C. § 1401a (retired military personnel pay), but at least once directing an administrative agency to adjust benefits for inflation, 10 U.S C. § 945(e) (retired military judges’ annuities).
When it passed OBRA, therefore, Congress was well aware that federal programs based on fixed-dollar amounts would be affected by inflation, and it knew how to correct the problem. It could, for example, have directed the Secretary to set the automobile exemption at “such amount (periodically adjusted for inflation) as the Secretary may prescribe.” Congress did no such thing; instead, it left the Secretary wide-open dis-eretion to set and maintain the value of the exemption.
Congress again considered the automobile exemption in 1988. By then the exemption had been in effect six years, and its 1982 value had been reduced from $1,500 to approximately $ 1,270.7 Still, Congress didn’t adjust it for inflation or instruct the Secretary to do so; it expressed no unhappiness with the phenomenon that captures the majority’s attention — erosion of the automobile exemption and consequential tightening of AFDC eligibility standards. In fact, Congress rejected a law that would have allowed five states to raise the automobile exemption to $4,500 and gave the Secretary a vote of confidence by instructing him to adjust the exemption amount only “if he determines revision would be appropriate.” H.R. Conf. Rep. No. 998, 100th Cong., 2d Sess. (1988), reprinted in 1988 U.S.C.C.A.N. 2879, 2976-77.
Finally, less than three years ago, Congress decided to index the automobile exemption for Food Stamp benefits, see Omnibus Budget Reconciliation Act of 1993, Pub.L. No. 103-66, § 13923, 107 Stat. 312, 675 (codified as amended at 7 U.S.C. § 2014(g)(2)), but took no similar action as to the closely analogous exemption for AFDC benefits.
Erosion of the car exemption is thus not a mysterious phenomenon Congress was unaware of, could not anticipate and didn’t know how to deal with. Indexing is a well-known and powerful tool for adjusting the burdens of inflation; whether to invoke it in a particular context is a hot-button policy issue with vast fiscal implications. It is precisely the kind of issue Congress should— and does — address directly. And here it has. The clearest proof that Congress meant for AFDC eligibility standards to gradually be raised by inflation is on the face of the statute. As our Sixth Circuit colleagues have observed, “OBRA itself cut in half the general limit on allowable resources under AFDC, and Congress has not adjusted that amount *1356for inflation.” Hazard, 44 F.3d at 404. Erosion of the basic asset limit ratchets up AFDC eligibility standards just like erosion of the automobile exemption: It makes it harder and harder for applicants who own property to qualify. Thus, the majority is plainly wrong when it finds “no evidence ... that Congress intended to cut [AFDC] spending progressively past 1981.” Maj. op. at 1347.
The majority also seems to be saying that Congress’s cost-cutting zeal was spent once it reduced the basic resource limit from $2,000 to $1,000, leaving no room for the Secretary to conduct further cost-cutting by reducing the automobile equity limit. Id. The statute, however, contains no such constraint on the Secretary’s authority; the majority comes up with its conclusion only by weighing the law’s competing policies. Id. But reconciling a statute’s diverging policy objectives is the province of the agency charged with its administration. Chevron, 467 U.S. at 844-45, 104 S.Ct. at 2782-83. This is particularly so where hard choices must be made as to the allocation of scarce resources: “The distribution of public funds among competing social programs is an archetypically political task, involving the application of value judgments and predictions to innumerable alternatives.” Int’l Union, United Automobile, Aerospace & Agricultural Implement Workers of America v. Donovan, 746 F.2d 855, 862 (D.C.Cir.1984) (Scalia, J.). By holding that the only rational judgment the Secretary could make under these circumstances is to inflation-proof the automobile exemption, the majority has assumed the prerogatives of the political branches.
Nor is the majority’s ruling limited to this particular regulation. My colleagues go much farther by holding that periodic review and adjustment for inflation is required whenever “a regulation’s rationality depends on economic conditions.” Maj. op. at 1348. This reasoning affects all manner of regulations that set fixed-dollar amounts and are therefore subject to erosion by inflation. See, e.g., 20 C.F.R. § 416.1218(b)(2) (excluding up to $4,500 of market value of one automobile from resources of person applying for Supplemental Social Security Income); 20 C.F.R. § 416.1222(a) (excluding up to $6,000 of income-producing property from resources of person applying for Supplemental SSI); 20 C.F.R. § 404.1574(b) (establishing one fixed monthly earning limit for years 1979 to 1990 and another for years 1990 to present, for purposes of determining eligibility for SSI disability benefits); cf. Garnett v. Sullivan, 905 F.2d 778, 782-83 (4th Cir.1990) (rejecting argument that Secretary was required to adjust 20 C.F.R. § 404.1574’s monthly earning limits for inflation).
Another in the smorgasbord of reasons the majority serves up in support of its conclusion is that “Congress did not necessarily want to exclude from the AFDC program needy families that owned safe, functioning cars,” maj. op. at 1348, and “functioning cars have increasingly cost more than $1500,” id. But Congress said nothing at all about what types of cars, if any, AFDC recipients should be allowed to own. The majority relies instead on vague assertions about the policy of encouraging “self-sufficiency and independence among AFDC recipients.” Id. at 1347. But the weight to be accorded this policy is entrusted to the Secretary, who is given authority to set the automobile exemption at “such amount as [she] may prescribe.” 42 U.S.C. § 602(a)(7)(B). If we are going to reject the Secretary’s policy judgment, we must come up with some very persuasive reasons. Have we done so? Not even close. The majority notes that
$1500 in 1992 represented just 9% of the average selling price of a new domestic car. Cars worth only $1500 are now between ten and fifteen years old and have more than 85,000 miles on them. Compared to newer cars, these cars are unreliable as well as more costly in terms of gas and repairs.
Maj. op. at 1346. This is very interesting but largely beside the point. Of course older cars are more trouble than newer ones, which is usually why people who can, do buy new cars. But what does all this tell us about the self-sufficiency of AFDC recipients?
To make that determination, we must have reliable data about the economics of car ownership — just the type of data the Secretary *1357develops during the course of administrative rulemaking. As the Secretary had no opportunity to develop such data, the majority relies entirely on an advocacy piece, suggestively titled “An Economic Investigation into the Basis for and Consequences of the H.H.S. Rule Eliminating from Eligibility for AFDC Benefits Families Who Have Over $1500 Equity in a Motor Vehicle.” See Maj. op. at 1346-1347 & nn. 6-7. What assurance do we have that this “study” reflects reality? The author hasn’t been qualified as an expert; his data haven’t been subjected to peer review; he wasn’t tested on cross-examination; there was no opportunity to consider competing views.8 Yet, this is the only evidence the majority considers in vacating the regulation.9
In fact, the report contains patent flaws. To begin with, AFDC excludes $1,000 in assets, in addition to an automobile, a home and, at each state’s option, items such as “clothes, furniture and other similarly essential items of limited value.” 45 C.F.R. § 233.20(a)(3)(i)(B); see 42 U.S.C. § 602(a)(7)(B).10 If a family has no unex-cluded assets, it can assign this added value to its automobile, which means it can qualify for AFDC while owning outright a ear worth up to $2,500. We have no idea whether some, many or most AFDC recipients are in this position since plaintiffs’ report, and the majority, ignore this point altogether.11
The report and the majority also err in treating the $1,500 figure as if it were a limit on the total value of the car. See maj. op. at 1346-1347 & nn. 6-7(citing report). In fact, the regulation limits equity. 45 C.F.R. § 233.20(a)(3)(i)(B)(2). AFDC recipients can thus own cars worth anywhere from $3,000 (twice $1,500) to as much as $10,000 (four times $2,500) if they buy on credit. This opens up the question whether, and to what extent, AFDC recipients can obtain and service loans secured by their cars. The issue is not addressed by plaintiffs or considered by the majority.12 In fact, it isn’t something we *1358could possibly judge for ourselves because it is the type of decision that calls for the Secretary’s administrative resources and her exercise of policy judgment. What we know for sure, however, is that an analysis which treats equity value as if it were total value omits a key feature of the statutory scheme.13
Nor does the majority give the least thought to what other means of locomotion might be available to AFDC recipients.14 Isn’t it effete to assume that self-sufficiency and independence can only be found behind the wheel of a motor car? In fact, the federal government has poured tens of billions of dollars into public transportation, no doubt hoping people will abandon polluting, congestion-prone automobiles. See, e.g., Thomas C. Hayes, Dallas Negotiates Its Way Out of the Commuter Business, N.Y. Times, Feb. 2, 1986, at E5 (federal government has spent $43 billion on mass transit since 1964). The Secretary — if given a chance — might determine that AFDC recipients can get around well enough by means of heavily subsidized public transport, and should be discouraged from using AFDC dollars to buy and service cars.
The degree to which a car is a necessity, moreover, varies with local conditions. Residents of New York City and San Francisco are in a much different position than, say, those who live in Texas and Wyoming. Hawaii is unique: Not only is traffic a serious and growing problem there, see Susan Hooper, Steering Toward Solutions, Hawaii Business, Sept. 1989, § 1, at 16, but distances are short and automobiles are of limited use, as they can’t be driven between many of the state’s populated areas, or from Hawaii to anywhere else.15
Which brings me to the majority’s last, and perhaps most troubling, rationale:
[T]he Secretary has allowed various states to waive the $1500 limit and set a higher automobile equity limit.... Such waivers are not only a recognition that in many states the $1500 limit is unreasonable but, more importantly, are consistent with this court’s view that the purpose of the automobile equity limit is not to reduce federal spe: ding but to ensure that eligible families are not excluded from the AFDC program simply because they own a reliable car.
Maj. op. at 1349. The logic of this argument escapes me. In 1962, Congress authorized the Secretary to grant waivers to a wide variety of AFDC program requirements on an experimental basis and for a limited time. See Public Welfare Amendments of 1962, Pub.L. No. 543, § 122, 76 Stat. 172 (codified at 42 U.S.C. § 1315). This enables the Secretary to “test out new ideas and ways of dealing with the problems of public welfare recipients.” S.Rep. No. 1589, 87th Cong., 2d Sess. (1962), reprinted in 1962 U.S.C.C.A.N. 1943,1961. By drawing an adverse inference from the grant of such waivers, the majority impermissibly burdens the Secretary’s discretion and undermines the policy of 42 U.S.C. § 1315. If the Secretary must worry that granting a temporary waiver will be *1359construed by the federal courts as an admission that the requirement being waived is irrational, such waivers will be granted less freely. This will not only make it more difficult for states to tailor welfare programs to local conditions, it will undermine the congressional goal of “testing] new ideas and ways of dealing with the problem of public welfare recipients.” S.Rep. No. 1589, 87th Cong., 2d Sess. (1962), reprinted in 1962 U.S.C.C.A.N.1948,1961.
The Remedy
The majority’s ruling is extraordinary in one further respect. It doesn’t merely prevent the Secretary from applying the $1,500 automobile resource limit in Hawaii or even the entire Ninth Circuit. No, it “vacate[s] the $1500 automobile equity limit regulation” altogether, maj. op. at 1351, with the stated consequence that all AFDC applicants will immediately qualify for benefits no matter how fancy a car they may own, at least until the Secretary can put a revised regulation into place, id. at 1343. The fiscal consequences of the majority’s ruling would no doubt be substantial were its sweep confined to our own territorial jurisdiction, but the majority raises the stakes by giving its ruling nationwide scope. The majority thus not only creates a conflict with the four other circuits which have upheld the regulation, see p. 1351-1352 supra, it overrules their decisions by precluding the Secretary from enforcing the regulation even where the law of the local circuit allows it. Worse still, the majority holds that the Secretary may not set the automobile exemption at $1,500 no matter what justifications she comes up with, because any such amount would be “struck down ... as arbitrary and capricious.” Maj. op. at 1343. We are thus in the unusual position of holding that an amount that four other circuits have upheld as rational is so irrational that it could not be upheld no matter what the justification.16
The majority could have ruled much more narrowly, of course, by merely adjudicating the denial of Gamboa’s application for benefits. Such a ruling would have precluded the Secretary from applying the regulation within the Ninth Circuit, as a matter of stare decisis, but would have left her free to apply the regulation elsewhere. This course would have given proper deference to those other circuits that have reached a contrary conclusion. The majority, instead, goes out of its way to nullify the considered judgments of our colleagues of co-ordinate authority.
Were the majority nevertheless bent on affording nationwide relief, it could still adopt a less drastic remedy. The majority holds that the Secretary’s rationale for the regulation does not support retaining it without periodic adjustments for inflation. See maj. op. at 1347-1348. The more prudent remedy under such circumstances isn’t to vacate the regulation forthwith; it is to remand so the agency can come up with adequate reasons for retaining the regulation, or commence rulemaking. See, e.g., American Horse Protection Ass’n, 812 F.2d at 7-8 (where “the Secretary has not presented a reasonable explanation of his failure to grant the rule-making petition ... [he] ... must be given a reasonable opportunity to explain his decision or to institute a new rulemaking proceeding.”); see also 3 Kenneth C. Davis & Richard J. Pierce, .Jr., Administrative Law Treatise § 18.1, at 165 (3d ed.1994). Such a ruling would have the added virtue of avoiding an immediate collision with the decisions of our sister circuits, as the Secretary would remain free to apply the current regulation while she pondered her options on remand. Of course, since the majority has prejudged any effort by the Secretary to re-adopt the $1,500 automobile exemption under any rationale, this would only be a temporary reprieve. But it would at least allow other circuits to reconsider their decisions in light *1360of our ruling, rather than having it foisted on them as if from above.
* # *
My colleagues review an action that is not an action by applying a standard that is not a standard. They rush in where other circuits fear to tread, putting us first among equals. They adopt a rationale with explosive potential, one that applies equally to every regulation setting fixed-dollar amounts subject to erosion by inflation. They implement a policy device — indexing—that Congress has guarded jealously and used only sparingly. And, at a time when the federal government regularly grinds to a standstill because of widespread concerns about excessive spending, my colleagues set in motion a legal flywheel that will substantially ratchet up expenditures nationwide for what is already a very expensive federal program. This is as egregious a case of judicial overreaching as I have seen in my years on the bench. I emphatically dissent.
. Indeed, plaintiffs have expressly disclaimed any construction of the record that would find an implied petition for rulemaking. Plaintiffs-Appellants/Cross-Appellees' Reply/Answering Brief at 19.
. That the majority is reviewing a non-action is perhaps best illustrated by its anachronistic conclusion: "Based on these changed economic circumstances, we find that in 1996, there is no rational connection between the facts found and the choice made by the Secretary in 1982.” Maj. op. at 1346-1347 (internal quotation marks omitted).
. I also disagree with the majority that exhaustion here would have been futile. See nn. 5-6 infra.
. While there is a presumption favoring judicial review of agency action, there’s a presumption against judicial review of an agency’s failure to act. See Heckler v. Chaney, 470 U.S. 821, 832, 105 S.Ct. 1649, 1656, 84 L.Ed.2d 714 (1985). Accordingly, an agency’s "refusal to institute rulemaking 'is to be overturned only in the rarest and most compelling of circumstances.' ’’ Brown v. Secretary of Health and Human Serv., 46 F.3d 102, 110 (1st Cir.1995) (quoting American Horse Protection Ass'n, 812 F.2d at 5). Because the majority holds that the Secretary abused her discretion by failing to revise a regulation that was validly adopted, plaintiffs had to overcome the inertia of this presumption before the district court could review this aspect of the case. Yet, as I explain below, the statutory language contains no standard against which we can measure the Secretary’s decision to set the automobile exclusion at any particular amount. There is thus nothing with which to overcome the presumption against judicial review.
. I thus disagree that exhaustion was unnecessary because “[a]t stake in this case are primarily issues of law, which courts are equipped to handle.” Maj. op. at 1351. The Supreme Court has admonished that "courts ordinarily should not interfere with an agency until it has completed its action ... particularly ... where the function of the agency and the particular decision sought to be reviewed involve exercise of discretionary powers granted the agency by Congress, or require application of special expertise.” McKart v. United States, 395 U.S. 185, 194, 89 S.Ct. 1657, 1663, 23 L.Ed.2d 194 (1969). A petition for rulemaking would have focused the Secretary's attention on plaintiffs’ claim at the administrative level. Had she chosen not to grant the petition, the denial would have contained reasons and, quite possibly, supporting data we could now consider in conducting our review. See 2 Kenneth C. Davis & Richard J. Pierce, Jr., Administrative Law Treatise § 15.2, at 309 (3d ed.1994) ("judicial review of agency action can be hindered by failure to exhaust administrative remedies because the agency may not have an adequate opportunity to assemble and analyze relevant facts and to explain the basis for its action.").
. Normally, of course, we would be reviewing the Secretary's actual reasons for her action. Because the Secretary here was given no opportunity to consider the matter administratively, we are cut adrift, having to invent reasons why the Secretary might or might not have acted. For example, the majority observes that "at oral argument, the Secretary conceded that it has never been the Secretary's position that the $1500 limit was set and not adjusted for inflation to make it *1354more difficult for needy families to qualify for AFDC benefits.” Maj. op. at 1348. This offhand comment by counsel is at odds with the position taken by the Secretary in litigation. See Hazard, 44 F.3d at 404 ("The Secretary responds that inflation’s effect of gradually tightening eligibility standards is consistent with congressional intent in enacting OBRA.”). In any event, the Secretary may not have adopted this position because there has never been a rulemaking, during the course of which she could have developed and articulated reasons for changing the exemption amount or for refusing to do so. I therefore don’t believe counsel’s concession can bind the Secretary or foreclose our consideration of this or any other plausible reason the Secretary might have developed for her action.
. This figure is based on the increase in the CPI-U — the Consumer Price Index for All Urban Consumers — for used cars, which increased from a base of 100 in 1982 to a level of 118 in 1988. See Bureau of the Census, U.S. Dep't of Commerce, Statistical Abstract of the United States 1994, at 490 (114th ed.).
. The report was authored by one Dr. Peter Fisher. All we know about him is that he received his doctorate in economics from the University of Wisconsin in 1978 and that he has published ‘‘a substantial number of articles in academic journals.” ER 102. Dr. Fisher was retained to prepare the study by Pine Tree Legal Assistance, Inc., of Augusta, Maine, for the purpose of challenging the regulation now before us, in the case of We Who Care, Inc. v. Sullivan, 756 F.Supp. 42 (D.Me.1991) (striking down regulation), overruled by Brown v. Secretary of Health and Human Serv., 46 F.3d 102 (1st Cir.1995). We have held that an expert report prepared for litigation must be considered with skepticism. See Daubert v. Merrell Dow Pharmaceuticals, Inc., 43 F.3d 1311, 1317 (9th Cir.1995).
. The Secretary’s failure to submit contrary data would ordinarily justify reliance on plaintiffs’ data at summary judgment. This isn’t the case here, however, as plaintiffs have short-circuited the process whereby the Secretary could have been expected to develop contrary data. See nn. 5-6 supra.
. AFDC also excludes one burial plot and bona fide funeral agreements for each family member. 45 C.F.R. § 233.20(a)(3)(i)(B)(3) & (4). In addition, AFDC excludes for a period of six consecutive months real property the family is making a good faith effort to sell, provided that the family signs an agreement to dispose of the property and to repay aid to which the family would not have been entitled had the property been sold at the beginning of the period (with repayment not to exceed the net proceeds of the sale). 45 C.F.R. § 233.20(a)(3)(i)(B)(5).
. Dr. Fisher’s report contains tables demonstrating that you can get an automobile up to three years newer by adding $1,000 to the purchase price. See ER 99, 112. His discussion, like the majority’s, ignores this inconvenient fact. Compare maj. op. at 1346-1347 with ER 99, 112.
. My colleagues are mistaken if they assume that anyone who qualifies for AFDC cannot possibly be driving a car bought on credit. To begin with, there are those who qualify for AFDC for short periods, but are not permanent residents on the welfare rolls. Two examples are mentioned in Dr. Fisher’s report: a family where one spouse is recently divorced or widowed, or where both adults are suddenly unemployed, perhaps because of layoffs. ER 97. Such families may have qualified for a car loan before they got into financial straits and would only need to keep up payments after they became eligible for AFDC. Also, families may qualify for a variety of social welfare benefits — federal, state and local— over and above AFDC. AFDC recipients thus may have a far more positive financial picture (making them more credit-worthy) than would appear if one were to consider only what they receive under AFDC. See Michael Tanner et al., The Work vs. Welfare Trade-Off 5 (Cato Institute Policy Analysis Series No. 240 1995). The availability of such other assistance varies from state to state but, ironically, Hawaii is at the very top of the list in terms of available assistance: A typical *1358AFDC family in Hawaii can get yearly assistance equivalent to a salary of $36,400, in addition to assistance available at the local level. Id. at 25. Owning a car bought on credit may be common among AFDC families. We don’t know and have no way to find out.
.Dr. Fisher sidesteps the issue by assuming AFDC recipients own their cars outright. ER 97-98, 113. He offers no support for this assumption, except a brief reference to the fact that, when "considering the situation of a woman recently divorced or widowed, or a couple with both adults unemployed ... [one can assume] it is probably common to own a decent car outright." ER 97. Such guesswork strikes me as an extraordinarily thin basis on which to ignore a key term of the statute, and poor science to boot.
. Dr. Fisher’s report implicitly recognizes that public transportation may afford many AFDC recipients a perfectly reasonable alternative to owning a car. ER 94, 106. But he chooses to focus instead on those recipients for whom the lack of a car may prove to be most problematic. Id. Thus, the report assumes, without argument or supporting authority, that a reasonable AFDC automobile exclusion "would permit ownership of an adequate car under the worst conditions prevailing in any state." ER 85 (emphasis added). In administering a nationwide program, the Secretar)' need not, of course, rely on the worst-case scenario in making her policy judgments.
. Once again raising the question of why Hawaii has Interstate Highways.
. There could be an interesting jurisprudential standoff if the Secretary were to deny benefits to an applicant in, say, Massachusetts based on the fact that the applicant owns a car with more than $1,500 equity value. Under one view of the matter, the applicant would win as a matter of res judicata, as he would be a member of the class in our lawsuit, which consists of all persons who have been or will be denied benefits because of their excess auto equity, CR 28, at 9. Under another view, the courts of the First Circuit might refuse to give effect to our ruling because it conflicts with their own circuit precedent, see Brown, 46 F.3d 102.