George BRAULT, Plaintiff–Appellant, v. SOCIAL SECURITY ADMINISTRATION, COMMISSIONER, Defendant–Appellee.
Plaintiff–Appellant George Brault appeals from the judgment of the United States District Court for the District of Vermont (Murtha, J.) affirming the decision of the Commissioner of Social Security (“the Commissioner”) denying Brault's application for disability benefits. We conclude the decision of the administrative law judge (“ALJ”) was supported by substantial evidence and was not the product of legal error, and affirm the district court's judgment.
Brault filed an application for a period of disability and Disability Insurance Benefits in September 2007. He claimed that he became disabled in September 2006 because of nerve damage in his left arm and a cervical spine injury he sustained in a motor-vehicle accident. After his application was denied initially and on reconsideration, he requested an administrative hearing before an ALJ.
The ALJ found that Brault had carried his burden of proof on steps one through four of the five-step process the Commissioner employs to determine disability. See 20 C.F.R. § 404.1520; Butts v. Barnhart, 388 F.3d 377, 383 (2d Cir.2004), as amended on reh'g in part, 416 F.3d 101 (2d Cir.2005). At that point, the burden shifted to the Commissioner to show there is other work that Brault can perform. See DeChirico v. Callahan, 134 F.3d 1177, 1180 (2d Cir.1998). The ALJ concluded, based on the testimony of the vocational expert (“VE”) the government had retained, that Brault was “capable of making a successful adjustment to other work that exists in significant numbers in the national economy.” Accordingly, Brault's application was denied.
Although Brault's counsel stipulated to the VE's expertise, he asserted a Daubert-like objection to the VE's actual testimony, contending it was unreliable.2 In response to hypotheticals from the ALJ roughly describing Brault's specific limitations, the VE had identified eight occupations an individual with such limitations could perform.3 He based that determination on his own expertise, as well as on the position descriptions in the Dictionary of Occupational Titles (the “DOT”), a United States Department of Labor publication. The DOT gives a job type a specific code—for example, “295.467–026 Automobile Rental Clerk”—and establishes, among other things, the minimum skill level and physical exertion capacity required to perform that job. Because of the detailed information appended to each DOT code, the codes are useful for determining the type of work a disability applicant can perform. In fact, the DOT is so valued that a VE whose evidence conflicts with the DOT must provide a “reasonable explanation” to the ALJ for the conflict. See Social Security Ruling (SSR) 00–4p, Policy Interpretation Ruling: Titles II and XVI: Use of Vocational Expert and Vocational Specialist Evidence, and Other Reliable Occupational Information in Disability Decisions, 2000 WL 1898704 (Dec. 4, 2000).
The DOT, however, just defines jobs. It does not report how many such jobs are available in the economy. Consequently, the VE turned to the Occupational Employment Quarterly II (the “OEQ”), prepared by a private organization called U.S. Publishing, to assess whether positions exist for each of the eight DOT codes, both in the national economy and in Vermont, Brault's home state. The VE testified that according to the OEQ more than 1,750,000 light-exertion, unskilled, positions exist in the national economy, and more than 3,600 exist in Vermont.
Brault argued, somewhat in passing, that the VE's report improperly counted part-time positions. But his main objection to the VE's testimony was that it did not reliably match the DOT codes to the OEQ data. According to Brault's submissions to the ALJ, the OEQ does not compile data by DOT code, but rather by Standard Occupational Classification System (“SOC”) code, a new system the Bureau of Labor Statistics has embraced to replace the DOT code regime. SOC codes, however, are not useful for disability proceedings because they do not contain the same detailed occupational information as DOT codes. Thus a VE must use some method for associating SOC-based employment numbers to DOT-based job types. The problem, however, is that DOT codes are much more granular than SOC codes—according to Brault, there were nearly 13,000 job titles in the 1991 edition of the DOT, but only about 1,000 SOC titles.
Citing this inexact matching, Brault submitted a memorandum arguing that “the underlying numbers [are] unscientific and fail to meet the Daubert standard for reliability.” According to him, “the numerical data provided by the SOC code do[ ] not enable a vocational expert to accurately determine the number of jobs within that SOC code for a particular DOT title.” As such, he maintained that the VE “has no scientific basis to break down between the various DOT titles” and to match them to SOC codes. He then explained—without any citation—that an expert “must use a ‘crosswalk,’ ” in other words, a data-matching algorithm, “to cross-reference the occupational detail for a particular DOT code to a SOC code [and then must] use the statistical data to define the number of jobs related to that DOT code.” 4
Brault's counsel addressed most of these points while cross-examining the VE. While acknowledging Brault's objections, however, the VE denied having reported the numbers for the entire SOC. Instead, he claimed to have “reduced” the numbers from “the entire [SOC] code” to only count “jobs ․ that I know exist.” With the ALJ's permission, Brault's counsel submitted additional briefing fully setting forth his objections to the VE's SOC–to–DOT mapping methodology.
The ALJ never directly responded to those objections. Instead, about a month after the hearing, the ALJ issued a ruling which relied on the VE's testimony, agreed that positions existed in the eight DOT positions the VE had identified at the numbers the VE had given, and denied Brault's application for benefits.
Brault appealed to the district court, which rejected Brault's challenge to the reliability of the VE's testimony, noting that it was appropriate for the VE to consult the OEQ in rendering his testimony. See Brault v. Soc. Sec. Admin., Comm'r, No. 1:10–CV–112(JGM), 2011 WL 1135014, at *3–4 (D.Vt. March 24, 2011). It affirmed the Commissioner's decision as supported by substantial evidence. Brault timely appeals.
When we review the Commissioner's denial of Social Security benefits, “our focus is not so much on the district court's ruling as it is on the administrative ruling.” Schaal v. Apfel, 134 F.3d 496, 500–01 (2d Cir.1998). Indeed, “[i]t is not our function to determine de novo whether [a plaintiff] is disabled.” Pratts v. Chater, 94 F.3d 34, 37 (2d Cir.1996). Instead, “we conduct a plenary review of the administrative record to determine if there is substantial evidence, considering the record as a whole, to support the Commissioner's decision and if the correct legal standards have been applied.” Moran v. Astrue, 569 F.3d 108, 112 (2d Cir.2009); see also 42 U.S.C. § 405(a) (on judicial review, “[t]he findings of the Commissioner of Social Security as to any fact, if supported by substantial evidence, shall be conclusive.”).
Substantial evidence is “more than a mere scintilla.” Moran, 569 F.3d at 112 (quotation marks omitted). “It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Id. (quotation marks omitted and emphasis added). But it is still a very deferential standard of review—even more so than the “clearly erroneous” standard. See Dickinson v. Zurko, 527 U.S. 150, 153, 119 S.Ct. 1816, 144 L.Ed.2d 143 (1999). The substantial evidence standard means once an ALJ finds facts, we can reject those facts “only if a reasonable factfinder would have to conclude otherwise.” Warren v. Shalala, 29 F.3d 1287, 1290 (8th Cir.1994) (emphasis added and quotation marks omitted); see also Osorio v. INS, 18 F.3d 1017, 1022 (2d Cir.1994) (using the same standard in the analogous immigration context).
Brault argues the ALJ erred by relying on VE testimony which Brault considers of dubious reliability. According to Brault, once that testimony had been challenged, the ALJ was required: (1) to grant an opportunity to inspect and challenge the proffered evidence and (2) if the ALJ relied on the challenged evidence, to explain why the challenge was rejected. Brault claims to find support in Seventh Circuit case law, but he candidly acknowledges a split among our sister circuits on the matter—one we have yet to address. Compare Bayliss v. Barnhart, 427 F.3d 1211 (9th Cir.2005) with McKinnie v. Barnhart, 368 F.3d 907 (7th Cir.2004). We address his arguments in reverse order.
A. Statement of Reasons
There is no question that the ALJ, in his written ruling, did not mention Brault's objection to the VE's testimony. In accepting that testimony, the ALJ necessarily rejected Brault's grievances, but Brault argues this implied rejection was insufficient—the ALJ needed to do more. In his view, he was owed an explanation.
He claims support from Donahue v. Barnhart, 279 F.3d 441, 446 (7th Cir.2002), purporting to quote from what he characterizes as the Seventh Circuit's holding that, when an expert's conclusions have been challenged, ALJs must “make an inquiry” and “explain how any conflict that has been indentified [sic] was resolved.” This is a misquotation and mischaracterization of the Seventh Circuit's language. The actual unaltered quotation is from a Social Security ruling that the court is discussing in its opinion—a ruling it cited only by way of analogy. See Donahue, 279 F.3d at 446. Nor do McKinnie, 368 F.3d at 911, and Lawrence v. Astrue, 337 Fed.Appx. 579, 585 (7th Cir.2009) (unpublished), which Brault also cites, support his argument.
This outcome, of course, is no surprise. An ALJ does not have to state on the record every reason justifying a decision. “Although required to develop the record fully and fairly, an ALJ is not required to discuss every piece of evidence submitted.” Black v. Apfel, 143 F.3d 383, 386 (8th Cir.1998). “An ALJ's failure to cite specific evidence does not indicate that such evidence was not considered.” Id. (citation omitted). That principle holds true not just for Social Security determinations, but also the determinations of other agencies, see, e.g., Chen v. United States Department of Justice, 471 F.3d 315, 341 (2d Cir.2006) (immigration), and even some decisions of the federal district courts, see, e.g., United States v. Fernandez, 443 F.3d 19, 31 (2d Cir.2006) (sentencing).
Assuming the ALJ had to consider Brault's objection to the VE's testimony, we are satisfied that he did so. There is no requirement that the ALJ discuss his specific analysis of it.
B. Opportunity to Challenge
1. Nature of the Right
Furthermore, the proposition that the ALJ was required to inquire into Brault's objection, whether or not the ALJ discussed it on the record, is dubious.
Currently, such a “duty to inquire” exists only in the Seventh Circuit. In Donahue, the Seventh Circuit drew inspiration from Fed.R.Evid. 702 and Daubert. See 279 F.3d at 446. While recognizing that Rule 702 does not apply in disability proceedings, the court held that the spirit of Rule 702 did. In that court's view, “the idea that experts should use reliable methods ․ plays a role in the administrative process because every decision must be supported by substantial evidence,” and thus evidence is per se not substantial if “vital testimony has been conjured out of whole cloth.” Donahue, 279 F.3d at 446. Therefore in the Seventh Circuit, if the basis of a vocational expert's conclusions is questioned, “the ALJ should make an inquiry (similar though not necessarily identical to that of Rule 702) to find out whether the purported expert's conclusions are reliable.” Id. McKinnie built upon this precedent by requiring that a VE make his or her data “available on demand” if it is challenged. 368 F.3d at 911. Lawrence reemphasized the rule of McKinnie but explained that the failure to provide data is subject to harmless error review. See 337 Fed.Appx. at 586.
The Donahue rule, however, has not been a popular export. See Babb v. Astrue, 2:10–CV–49–DBH, 2010 WL 5465839, at *3 (D.Me. Dec. 29, 2010), report and recommendation adopted, CIV. 10–49–P–H, 2011 WL 672438 (D.Me. Feb. 16, 2011) (disagreeing with Donahue ); Pritchett v. Astrue, No. 5:09–CV–144 (CAR), 2009 WL 4730326, at *3–*4 (M.D.Ga. Dec. 4, 2009) (same); Masters v. Astrue, No. 07–123–JBC, 2008 WL 4082965, at *4 n. 8 (E.D.Ky. Aug. 29, 2008) (same). And the Ninth Circuit's Bayliss decision, while not explicitly rejecting Donahue, reached a conclusion necessarily in conflict with it. See 427 F.3d at 1218 (“A VE's recognized expertise provides the necessary foundation for his or her testimony. Thus, no additional foundation is required.”). In fact, no court outside the Seventh Circuit has agreed with Donahue and its offspring.
And there appear to be good reasons to question Donahue's approach. Donahue relied heavily on the principles, if not the actual authority, of Fed.R.Evid. 702 and Daubert, 509 U.S. 579, 113 S.Ct. 2786. See Donahue, 279 F.3d at 446. But Congress has provided, quite clearly, that the Federal Rules of Evidence do not apply in Social Security proceedings. 42 U.S.C. § 405 (“Evidence may be received at any hearing before the Commissioner of Social Security even though inadmissible under rules of evidence applicable to court procedure.”); see also Richardson v. Perales, 402 U.S. 389, 400–01, 91 S.Ct. 1420, 28 L.Ed.2d 842 (1971) (“[S]trict rules of evidence, applicable in the courtroom, are not to operate at social security hearings․ [A]dministrative procedure, and these hearings, should be understandable to the layman claimant, should not necessarily be stiff and comfortable only for the trained attorney, and should be liberal and not strict in tone and operation. This is the obvious intent of Congress so long as the procedures are fundamentally fair.”). It is unclear, therefore, why the Seventh Circuit would acknowledge in Donahue that ALJs are not bound by the Rules of Evidence, but then turn around and require ALJs to hew so closely to Daubert's principles.
It is also unclear, given our standard of review, why a Daubert-like hearing is even useful. As deferential as the “substantial evidence” standard is, it is also extremely flexible. It gives federal courts the freedom to take a case-specific, comprehensive view of the administrative proceedings, weighing all the evidence to determine whether it was “substantial.” Other federal courts have proven the validity of this approach—while declining to impose a strict Daubert rule, reviewing the entirety of a VE's testimony, including the expert's methods, to make sure it rose to the level of “substantial” evidence. See, e.g., Palmer v. Astrue, No. 1:10–CV151–JGM, 2011 WL 3881024, at *6 (D.Vt. Sept. 2, 2011).
Galiotti v. Astrue, 266 Fed.Appx. 66 (2d Cir.2008) (summary order), while nonprecedential, makes the same point.5 There, the claimant challenged the credibility of a VE's testimony because the expert was unable to fully explain how he arrived at the number of certain jobs in the economy. Id. We explained that the VE, while not providing the specific information the claimant wanted, “identified the sources he generally consulted to determine such figures.” Id. We also noted the marked absence of any “applicable regulation or decision of this Court requiring a vocational expert to identify with greater specificity the source of his figures or to provide supporting documentation.” Id. We thus affirmed, not on any Daubert basis, but instead on typical “substantial evidence” grounds. We do the same here.6
Yet, although we harbor doubts about the Seventh Circuit's approach, the extent to which an ALJ must test a VE's testimony is best left for another day and a closer case. We do not hold that an ALJ never need question reliability, and we agree with the Seventh Circuit that evidence cannot be substantial if it is “conjured out of whole cloth.” See Donahue, 279 F.3d at 446. There is no need to fully resolve the matter now, however, because, assuming arguendo Brault had a right to have the ALJ consider his challenge to the VE, that is exactly what the ALJ did.
At the beginning of the hearing, the ALJ asked the VE to affirm that he would impartially evaluate the vocational evidence, and that, in the event of conflict between his testimony and the DOT, he would advise the ALJ “of the differences and the basis for [his] opinion.” The ALJ also identified a specific issue in Brault's case where such a conflict might arise. Further, the ALJ sought and received a stipulation from Brault's counsel regarding the VE's expertise and qualifications. When Brault's counsel cross-examined the VE, he was given a full opportunity to explore the limitations of the SOC–to–DOT mapping methodology, including challenging some of the expert's specific numbers. After cross-examination, the ALJ permitted counsel to submit briefing questioning the VE's “flawed” process. In sum, Brault's attorney had a full opportunity to explain his objections in significant detail. Nothing more was required.
For the foregoing reasons, the judgment of the district court, which affirmed the Commissioner of Social Security's denial of Brault's disability benefits, is affirmed.
2. Before the ALJ, his counsel in fact called it a Daubert objection, even though the rule of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), does not apply in Social Security administrative proceedings. See infra Section II.B.1.
3. In response to a second hypothetical with additional physical limitations, the VE pared the number of occupations down to two, but the ALJ did not rely on the second hypothetical, and no one argues he should have.
4. Brault's argument to the ALJ was somewhat inartful. But he stumbled onto a classic academic problem with data aggregation—a many-to-one mapping, such as the one the VE used to associate DOT titles to SOC codes, necessarily creates information loss. See generally Guy H. Orcutt, et al., Data Aggregation and Information Loss, 58 Am. Econ. Rev. 773 (1968); Margaret St. Pierre & William P. LaPlant, Jr., Issues in Crosswalking Content Metadata Standards (National Information Standards Organization White Paper, 1998). If, for example, ten DOT codes map to a single SOC code, saying there are 100,000 total positions available in that SOC code gives no information at all about how many positions each of the ten DOT codes contributed to that total. This becomes a problem if DOT titles with different exertion or skill levels map to the same SOC code. In such a situation, the OEQ apparently uses a rough weighted average algorithm—if ten DOT codes correspond to one SOC code, and four of those codes are light-duty, unskilled positions, then the OEQ will list 40% of the positions available in that SOC as light-duty, unskilled positions. That estimate may deviate significantly from the actual number of existing positions.
5. We are, of course, permitted to consider summary orders for their persuasive value, and often draw guidance from them in later cases. See United States v. Payne, 591 F.3d 46, 58 (2d Cir.2010).
6. For the same reason, we reject Brault's argument that the ALJ erred by allowing the VE to provide employment numbers that might have included part-time positions (he offers no proof at all that the VE's numbers actually included such positions). We decline to create a per se rule prohibiting an ALJ from considering part-time positions. See Liskowitz v. Astrue, 559 F.3d 736, 745 (7th Cir.2009) (“[A] VE may ․ testify as to the numbers of jobs that a claimant can perform without specifically identifying the percentage of those jobs that are part-time.”). The ALJ did not need to find specific numbers of jobs—all he was required to do was find that “substantial” positions exist. There was substantial evidence supporting that finding.Brault argues Liskowitz is in conflict with the Eleventh Circuit's decision in Kelley v. Apfel, 185 F.3d 1211 (11th Cir.1999). We disagree. See Liskowitz, 559 F.3d at 745 (“Our conclusion is not at odds with [Kelley ].”). Kelley specifically declined to “confront the issue of whether part-time work, as opposed to full-time work, will prevent a claimant from being found disabled at Step Five of the sequential analysis.” 185 F.3d at 1215. And although Kelley stated in dicta that “a claimant could pass Step Five and be entitled to benefits even though capable of working on a part-time basis,” id., it never stated part-time work could never count at step five, and certainly “did not say that a VE may testify only as to the existence of full-time jobs,” Liskowitz, 559 F.3d at 745.