with whom WIGGINS, O’SCANNLAIN and TROTT, Circuit Judges, join, concurring only in the judgment.
While concessions made by government counsel at oral argument require that I join in remanding the two cases before us, I can agree with almost nothing else the majority says in its opinion. As I explain below, the opinion is supported by neither the applicable statutes nor the regulations; it is an exercise of common-law decisionmaking spuriously imposed on a complex regulatory scheme. My distinguished colleagues may be convinced they are doing the right thing and, in some cosmic sense, perhaps they are. But in our tripartite system of government, the judiciary does the right thing by applying statutes and regulations as written1 and letting the political branches resolve the intractable policy conflicts that inevitably arise in the implementation of social welfare legislation.
I
A. While the majority reaches out— wholly gratuitously — to overrule Bates v. Sullivan, 894 F.2d 1059, 1064 (9th Cir. 1990) (Wallace, C.J., .and Wright, J., concurring),2 it never comes to grips with that opinion’s analysis of the statute, the applicable regulations and the legislative record pertaining to the issue of pain. I can do no better than to quote that analysis at length:
Prior to its enactment of the Social Security Disability Benefits Reform Act of 1984, Pub.L. No. 98-460, 98 Stat. 1794 (the Reform Act), Congress had not specifically addressed whether an individual’s subjective pain could be considered disabling. Congress, however, had clearly stated:
The term “disability” means—
*349(A) inability to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment which can be expected to result in death or which has lasted or can be expected to last for a continuous period of not less than 12 months.
42 U.S.C. § 423(d)(1) (1983) (emphasis added). In section 423(d)(3), the statute further stated:
For purposes of this subsection, a “physical or mental impairment” is an impairment that results from anatomical, physiological, or psychological abnormalities which are demonstrable by medically acceptable clinical and laboratory diagnostic techniques.
(Emphasis added.) While this statutory language does not mention subjective pain testimony, it surely indicates that Congress intended that an alleged impairment had to be “medically determinable” before it could be considered disabling.
The Secretary had also promulgated a regulation concerning pain which provided:
If you have a physical or mental impairment, you may have symptoms (like pain, shortness of breath, weakness or nervousness). We consider all your symptoms, including pain, and the extent to which signs and laboratory findings confirm these symptoms. The effects of all symptoms, including severe and prolonged pain, must be evaluated on the basis of a medically determinable impairment which can be shown to be the cause for the symptom. We will never find that you are disabled based on your symptoms, including pain, unless medical signs or findings show that there is a medical condition that could be reasonably expected to produce those symptoms.
20 C.F.R. §§ 404.1529 and 416.929 (1983) (emphasis added). Although similar to the statutory language, the Secretary’s regulation further clarified the importance of medical evidence of pain by requiring that the medical condition alleged to cause the complained of pain be of the type “reasonably expected to produce” that pain. The wording of the regulation was clear: a claimant would never be found disabled unless he could point to a medical condition which caused him to suffer from the degree of pain which he alleged.
Given the plain language of the statute and the Secretary’s regulation, our task should have been: where a claimant’s allegation of pain were not corroborated by medical evidence, we would conclude from the record as a whole that there was substantial evidence justifying the Secretary’s finding that the claimant was not disabled.
Id. at 1064-65.
As Bates points out, however, the “plain language of the statute and the Secretary’s regulations” was not enough; courts still chose to ignore the direction of the political branches and accorded great weight to a claimant’s subjective allegations of pain. See 130 Cong.Rec. S6211 (daily ed. May 22, 1984) (criticizing Polaski v. Heckler, 585 F.Supp. 1004, 1008-09 (D.Minn.), aff'd, 739 F.2d 1320 (8th Cir.), ordered remanded on other grounds, 751 F.2d 943 (8th Cir.1984), vacated, 476 U.S. 1167, 106 S.Ct. 2885, 90 L.Ed.2d 974 (1986)). This judicial recalcitrance led Congress to enact 42 U.S.C. § 423(d)(5)(A) in 1984:
An individual shall not be considered to be under a disability unless he furnishes such medical and other evidence of the existence thereof as the Secretary may require. An individual’s statement as to pain or other symptoms shall not alone be conclusive evidence of disability as defined in this section; there must be medical signs and findings, established by medically acceptable clinical or laboratory diagnostic techniques, which show the existence of a medical impairment that results from anatomical, physiological, or psychological abnormalities which could reasonably be expected to produce the pain or other symptoms alleged and which, when considered with all evidence required to be furnished under this paragraph (including statements of the individual or his *350physician as to the intensity and persistence of such pain or other symptoms which may reasonably be accepted as consistent with the medical signs ■ and findings), would lead to a conclusion that the individual is under a disability. Objective medical evidence of pain or other symptoms established by medically acceptable clinical or laboratory techniques ... must be considered in reaching a conclusion as to whether the individual is under a disability.
(Emphasis added). The focus of that amendment, as Bates emphasized, was clear:
Thus, according to the amended statute, in order to prove that pain is disabling, “there must be medical signs and findings ... which show the existence of a medical impairment ... which could reasonably be expected to produce the pain ... alleged.” Significantly, the language of the statute uses a definite article, referring to “the pain” alleged. It does not refer to “some of the pain” alleged by the claimant, nor does it refer to medical impairments expected to produce “a certain amount of pain,” nor to “excess pain.” See Cotton v. Bowen, 799 F.2d 1403, 1407 (9th Cir.1986) (Cotton ). Thus, read in its most direct and plain sense, section 423(d)(5)(A) requires evidence of a medical condition which “could reasonably be expected to produce” the actual pain, in amount and degree, alleged by the claimant. Unless a claimant can produce “objective medical evidence of” such a medical impairment, he or she is not disabled under the statute.
Bates, 894 F.2d at 1066. Furthermore, the Bates opinion exhaustively reviewed the legislative record, demonstrating beyond dispute that Congress approved of the regulations and that it meant what it said in the statute. Supported by numerous quotations, Bates concluded:
Plainly, Congress concurred in the Secretary’s regulations and intended a claimant to produce medical evidence corroborating the “severity” of the pain that he alleged.
Indeed, the comments in the legislative record are unanimous in concluding that section 423(d)(5)(A) requires that a claimant’s allegation of subjective pain, including the severity of the alleged pain, be confirmed by objective medical evi-dence_ [T]he legislative history confirms that the statutory language means just what it says: the Secretary may reject allegations of pain unless the claimant is able to point to a medical condition which could reasonably be expected to produce that pain.
Id. at 1066-68.
This careful analysis by Chief Judge Wallace and Judge Wright commands a response, but the majority does not join issue. While the majority dutifully quotes the language of the statute and regulations, Majority at 344 n. 1, 344-45, it does not analyze the language it quotes, and it certainly does not cite or address the legislative record Bates so persuasively marshals. The majority overrules Bates and approves Cotton, but does not explain why Bates was wrong in concluding that “Congress appears to have specifically rejected the position taken by Cotton. ” Bates, 894 F.2d at 1068.
B. Bates aside, it is clear that the majority reaches its conclusion by misreading the applicable regulations. As Bates noted and the majority acknowledges, the regulations in question are deserving of special deference because they have explicit congressional approval. The majority quotes these regulations, but it does not make even a colorable attempt to analyze them. Were it to do so, it would learn that the regulations are consistent with Bates and wholly at odds with our decision today.
The relevant language of 20 CFR §§ 404.1529 and 416.929 states that the Social Security Administration “will never find that you are disabled based on your symptoms, including pain, unless medical signs or findings show that there is a medical condition that could be reasonably expected to produce these symptoms” (Emphasis added). This language articulates a two-pronged objective standard: To be found medically disabled, the claimant *351must first present proof of a “medical condition” and must next present proof that his medical condition is such that it “could be reasonably expected” to produce disabling pain.
Purporting to apply sections 404.1529 and 416.929, the majority simply drops the second of these requirements, stating that the claimant need only prove that his alleged pain is “associated with a clinically demonstrated impairment.” Majority at 345 (quoting Cotton, 799 F.2d at 1407). Thus, in the eyes of the majority, the regulations may require a claimant to produce medical evidence of his underlying impairment, but they free him from having to produce medical evidence of the alleged severity of his pain. But if that is all the regulations require, why would the Secretary use the phrase “a medical condition that could reasonably be expected to produce” disabling pain? The majority sayeth not. As the majority reads the regulations, this language is rendered a nullity.
Equally unaccounted for in the majority opinion is language in another portion of the regulations which states that “[t]he effects of all symptoms, including severe and prolonged pain, must be evaluated on the basis of a medically determinable impairment which can be shown to be the cause of the symptom.” 20 CFR §§ 404.-1529 & 416.929 (emphasis added). Once again, the regulations place the burden on the claimant to show a causal link between his medical condition and the disabling pain he alleges. Merely showing that the pain is “associated” with a medical condition, as the majority holds, ignores this requirement. I respectfully suggest that any interpretation of the regulations that overlooks such key phrases cannot be correct.
II
Having cut itself free of the applicable statute and regulations, the majority relies instead on conjecture about what Congress could and could not have intended. Thus, the majority discards the standard clearly adopted by the statute and regulations because “[i]f an adjudicator could reject a claim for disability simply because a claimant fails to produce medical evidence supporting the severity of the pain, there would be no reason for an adjudicator to consider anything other than medical findings.” Majority at 347. The majority also announces that “ ‘Congress did not intend to render the claimant’s pain testimony irrelevant/ ” id. (quoting Cotton, 799 F.2d at 1407), and that Congress could not have “intended to require objective medical evidence to fully corroborate the severity of pain while aware of the inability of medical science to provide such evidence.” Id.
The fact is, Congress and the SSA did precisely what the majority says they could not have meant to do. The statutory language is clear; the applicable regulations are clear; the congressional record, as detailed in Bates, is clear. Hard as it may be for my colleagues to believe, Congress actually made a hard policy judgment and directed the SSA to deny disability benefits in dubious cases. It may not be the policy judgment that makes sense to us, but we are not empowered to second-guess this judgment simply because we are incredulous that Congress chose the path it did.
An important clue that we have strayed from the path of statutory interpretation and wandered deep into the briar of policy comes from the fact that “excess pain”— the concept which lies at the heart of the majority’s approach — is nowhere to be found in the statutes, the regulations or the extensive congressional debate on this issue. In fact, the phrase was invented in Cotton, where it is defined as “pain that is unsupported by objective medical findings.” Cotton, 799 F.2d at 1407. Rather than following the clear mandate of Congress, the SSA and our prior caselaw — all of which are loyal to the view that such pain cannot support a disability determination — we held in Cotton that it was legal error for an ALJ to reject such “excess pain” testimony. We did not explain in Cotton where we found the concept of “excess pain” or why it had never surfaced before.
“Excess pain” is a concept only a lawyer could love: vague, statutorily unsupported, *352metaphysically incongruous. As the phrase implies, “excess pain” presupposes that there is some normal level of pain associated with a particular impairment and that some individuals, by virtue of personal idiosyncrasies, suffer more than their fair share. Pain, however, like beauty, is entirely subjective; it is impossible to compare one person’s suffering with that of another, much less determine the “correct” amount of pain someone should feel because of a particular impairment. See generally L. Wittgenstein, Philosophical Investigations (1953). To one who suffers pain, any degree of it is too much; everyone justly feels that his pain exceeds an acceptable level. Yet, if Congress had wanted to make subjective testimony of pain conclusive of disability, it would surely have said so. In fact, it said exactly the opposite.
Our focus on some undefined (and undefinable) level of pain in excess of the norm does little more than give claimants a chance to circumvent the disability rules set up by Congress, as the grant or denial of benefits now turns on a wholly subjective and totally unverifiable circumstance, one to which no one but the claimant can speak. See Stewart v. Sullivan, 881 F.2d 740, 746 (9th Cir.1989) (Sneed, J., concurring) (“I am hard-pressed to describe what showing the Secretary must make to refute an excess pain claim. It is possible, of course, for the Secretary to enlist the investigatory resources of the FBI to ascertain the genuineness of these claims. This does not strike me as a reasonable course of action.”). In fact, as each claimant experiences only his own pain, having no clue how it compares to the pain experienced by anyone else, the entire inquiry is absurd. Its only effect is to rob the Social Security disability system of the objectivity Congress has clearly declared it should have.
The cases that have endorsed this freewheeling concept of “excess pain” — starting with Cotton and continuing through our decision today — demonstrate a disturbing mistrust of an administrative agency that tries to follow its congressional mandate of holding disability claims up to scrutiny. They reflect a philosophy that anyone who claims entitlement should be awarded benefits unless the Secretary can prove otherwise. While this approach may have much to commend it, it is not that adopted by Congress, which put the burden of proving entitlement squarely on the claimant.
By endorsing this approach, the majority steps on one side of a difficult and sensitive policy issue, namely how to deal with the question of pain in disability cases. As we have recognized, because pain is subjective, it is easy to falsify or exaggerate claims thereof. See, e.g., Fair v. Bowen, 885 F.2d 597, 602 (9th Cir.1989). Decisions to grant or deny benefits thus “often hinge entirely on whether or not the claimant’s description of what he is feeling is believed.” Id. How to resolve this is a hard policy question, one that Congress has been struggling with for years. See 130 Cong.Rec. S6211-36 (Senate debate on 1984 amendment).
The difficulty in reaching a satisfactory solution lies in the intractable nature of the problem. Any rule of decision must strike a balance between two types of potential error — alpha and beta. See generally G. Keppell, Design and Analysis 65-66 (1973). For example, in the criminal context, the two types of error are unjust convictions of innocent persons (alpha errors) and wrongful acquittals of guilty ones (beta errors). In the Social Security disability context, the two types of error are wrongful denials of benefits to deserving claimants (alpha), and wrongful awards of benefits to undeserving claimants (beta).
Any rule of decision reflects a policy judgment as to the proper trade-off between alpha and beta errors. The rule of decision we’ve adopted in criminal cases tilts heavily away from alpha errors (erroneous guilty verdicts), embodying the popular notion that it is better for ten guilty persons to go free than to convict an innocent one. Even there, however, we do not adopt a rule that calls for absolute certainty. Rather, we accept as inevitable that some, although hopefully few, alpha errors will slip through; the occasional conviction of an innocent is the bitter price we pay for *353avoiding the wholesale acquittal of the guilty.
For disability determinations, Congress has adopted a rule of decision that reflects a particular concern about beta errors (wrongful awards of benefits). As always, there is a trade-off. Trying to hold down beta errors inevitably results in some alpha errors (wrongful denials of benefits). This is unfortunate indeed, but it reflects a policy judgment Congress made about the relative harm to the integrity of the social security disability insurance system from errors of a particular type.
Because wrongful grants of disability benefits are not appealable, beta errors never come to our attention; we see only the alpha errors. Our natural human tendency is to shift the rule of decision so as to minimize, or eliminate altogether, the type of error we do see. Because of the inevitable trade-off between the two types of error, however, a consequence of our tinkering is that we increase the number of beta errors, the type Congress was most concerned with avoiding, undermining the actuarial assumptions upon which the system is built.
Ill
The final irony in the majority’s opinion is that it is totally uncalled for in light of the posture of this case. At oral argument counsel for the government took the position that Social Security Ruling 88-13, which applies to AU evaluations of pain, adopts “a standard that is very much like this court’s concerns in the Cotton-Vamey line of cases” and sets forth a specific methodology by which AUs must consider claims of pain that are not fully corroborated by objective medical evidence. Because this SSR was not available when the ALJs reached their decisions in these cases, the government’s counsel conceded that remand was appropriate here.3 In light of this concession, we have no occasion to consider whether the statute and the regulations require the Secretary to do that which he has conceded he’s willing to do anyway.
Nonetheless, the majority does reconsider, and overrule, Bates. It eschews reliance on SSR 88-13 presumably because it does not consider the SSR binding. Majority at 346 n. 3. The majority is mistaken. While SSR 88-13 may not be binding on the courts, it is binding on the Secretary. See 20 CFR § 422.406(b)(1). If the Secretary chooses to apply a standard that is less rigorous than that envisioned by the statute and the regulations, he will be bound thereby during the course of the litigation.
By reaching beyond SSR 88-13, the majority denies the Secretary the flexibility to issue a superseding SSR and require adherence to the principle that pain testimony should be rejected to the extent it is unsupported by objective medical evidence. To this I cannot agree. Thus, although I concur in the majority’s decision to remand, I cannot accept, let alone join, its reasoning.
. Unless, of course, they are unconstitutional, which no one claims is the case here.
. That we have no occasion to overrule Bates, or even address the complex statutory question presented there, is clear beyond reasonable dispute. See Section III infra.
. When asked if the Secretary would "have any trouble with sending the whole shootin’ match back," counsel responded “[i]n these two cases, no."