In 2018, Attorney General Jeff Sessions reached into a single immigration case, certified it to himself, and overruled the precedent that had let domestic violence survivors qualify for asylum. In 2021, Garland certified it back. In 2025, Bondy certified it away again. Three reversals on the same doctrine in seven years — no statute changed, no rulemaking happened. One signature each time.
In this episode — the second in a three-part series on how agencies actually produce outcomes — Gwen and Marc move up a level from the hearing room and ask whether internal appellate review fixes the variation E13 documented. The textbook answer is yes: that's what review is for. The real answer is that "internal review" describes three very different machines, and none is primarily in the business of producing consistency.
- The APA baseline is permissive. 5 USC 557 gives the reviewing body "all the powers which it would have in making the initial decision" — including the power to redo credibility findings from a cold transcript. The ALJ who sat three feet from the witness gets reversed; the federal court then defers to the agency under substantial evidence review. The one person who actually saw the testimony is the one whose findings get displaced.
- Three agencies, three pullbacks. Social Security walks the default back via 20 CFR 404.970 — the Appeals Council acts only on legal error or want of substantial evidence. Immigration walks it back furthest via 8 CFR 1003.1 — IJ fact findings get clear-error review. The SEC operates closest to the default, where "all the powers" actually bites.
- Model 1 — Social Security: review mostly doesn't happen. The Appeals Council grants review in roughly 15-20% of cases. The rest get a one-paragraph denial. That denial opens the 60-day window to federal court, but reads to a lay claimant like a final rejection — because in ordinary English "denied" means you lost, while in admin-law English it means you've exhausted and the clock is running. Represented claimants know this. Unrepresented claimants miss the window and lose by default.
- Model 2 — Immigration: review is a policy instrument. The BIA designates decisions as precedential, binding every IJ in the country. Under 8 CFR 1003.1(h), the Attorney General can pull any case — pending or decided — and write a precedential opinion on it. Matter of A-R-C-G (2014), Matter of A-B- (2018), Garland's vacatur (2021), Matter of S-F-M (2025): the same doctrine swung four times across four administrations through a single referral mechanism. That isn't error correction. It's substantive policymaking through the review channel.
- Model 3 — Enforcement (SEC, FTC, NLRB): review is automatic and institutionally tilted. Every ALJ decision goes up. About three-quarters of SEC reversals favor the enforcement division — finding violations the ALJ didn't or raising penalties the ALJ imposed. The Commission supervises the enforcement staff, sets priorities, allocates their budget, and then sits as the appellate tribunal reviewing whether that same staff proved its case. No bad faith required; the incentives only run one way.
Three failure modes, one conclusion: SSA fails by not reviewing; immigration fails by reviewing through a politically directed mechanism whose priorities shift with administrations; enforcement fails by reviewing through a body with an institutional stake in the result. Review corrects some errors and generates precedent where precedent gets designated. What it does not do — anywhere — is pull similar cases toward similar outcomes.
Next episode: agencies don't shape outcomes by correcting decisions. They shape the conditions under which decisions get made.
00:00:14 --> 00:00:17 Hello, and welcome to Administrative Remedies,
00:00:17 --> 00:00:19 because you can't fix what you don't understand.
00:00:19 --> 00:00:21 Brought to you in part by the University of Tulsa
00:00:21 --> 00:00:24 College of Law. I'm Gwendolyn Savitz, an associate
00:00:24 --> 00:00:26 professor here at TU and the associate dean of
00:00:26 --> 00:00:28 research and intellectual life. And I'm Mark
00:00:28 --> 00:00:31 Rourke. I'm the dean of the College of Law. We'll
00:00:31 --> 00:00:33 be breaking down complex doctrines with real
00:00:33 --> 00:00:35 -life analogies and examples to demystify the
00:00:35 --> 00:00:37 world of administrative law for everyone trying
00:00:37 --> 00:00:39 to understand how government actually works.
00:00:40 --> 00:00:41 Agencies are the main way the federal government
00:00:41 --> 00:00:44 gets things done. It's not through Congress for
00:00:44 --> 00:00:46 reasons we'll be addressing over the course of
00:00:46 --> 00:00:50 this series. Mark, last episode, we spent 30
00:00:50 --> 00:00:52 minutes on the hearing level picture. We talked
00:00:52 --> 00:00:54 about judge level variation of up to 46 points
00:00:54 --> 00:00:57 on the same claim and another 10 or 15 points
00:00:57 --> 00:01:00 of variation between hearing offices. And we
00:01:00 --> 00:01:02 also talked about the representation effect that
00:01:02 --> 00:01:04 your approval odds are roughly doubled if you
00:01:04 --> 00:01:07 have counsel representing you. And finally, we
00:01:07 --> 00:01:09 talked about the fact that that is likely because
00:01:09 --> 00:01:11 sophisticated lawyers can navigate this better
00:01:11 --> 00:01:14 because they built up private knowledge about
00:01:14 --> 00:01:17 specific judges and specific offices. Yeah. And
00:01:17 --> 00:01:19 the point was that none of that is controlled
00:01:19 --> 00:01:22 by any top -down mechanism at the ALJ level.
00:01:22 --> 00:01:26 That the agency has some binding interpretive
00:01:26 --> 00:01:28 rulings, but they cover a narrow band of questions.
00:01:29 --> 00:01:32 Everything else is private. It literally is like
00:01:32 --> 00:01:35 a free market of dispute resolution and that
00:01:35 --> 00:01:38 ALJ judgment on one side with the experienced
00:01:38 --> 00:01:41 counsel's ambient knowledge on the other. So
00:01:41 --> 00:01:44 the natural move, once you've seen that, is to
00:01:44 --> 00:01:46 say, OK, fine, maybe the hearing level doesn't
00:01:46 --> 00:01:49 produce consistency on its own, but there's another
00:01:49 --> 00:01:51 layer above that. Every agency has some sort
00:01:51 --> 00:01:53 of review. We've got the Appeals Council at Social
00:01:53 --> 00:01:56 Security. We have the Board of Immigration Appeals.
00:01:56 --> 00:01:58 We have the Commission at the SEC, the Board
00:01:58 --> 00:02:02 of the NLRB. So even if these are producing inconsistent
00:02:02 --> 00:02:05 decisions themselves, there should be these bodies
00:02:05 --> 00:02:07 above them that can correct errors, smooth out
00:02:07 --> 00:02:10 conflicts and pull the outcomes towards consistency.
00:02:10 --> 00:02:12 That should fix what the hearing level doesn't.
00:02:13 --> 00:02:17 Yeah. And today's question is really, does it?
00:02:18 --> 00:02:21 Right. So today's question is what it does, because
00:02:21 --> 00:02:23 the answer is it's not really just one thing.
00:02:23 --> 00:02:25 It's three things. It's three different kinds
00:02:25 --> 00:02:29 of agencies. And none of the three is what the
00:02:29 --> 00:02:32 textbook might suggest. So we're not going to
00:02:32 --> 00:02:36 land on internal review doesn't work. Right.
00:02:36 --> 00:02:38 We're going to land on something more interesting.
00:02:38 --> 00:02:40 Internal review does different work at different
00:02:40 --> 00:02:43 agencies. And the kind of work it does at each
00:02:43 --> 00:02:45 one tells us something about what it actually
00:02:45 --> 00:02:48 means for an agency to review itself. OK, so
00:02:48 --> 00:02:51 let's start with what they're looking at. When
00:02:51 --> 00:02:54 the Appeals Council or the BIA or the SEC Commission
00:02:54 --> 00:02:58 opens a file, what are they actually looking
00:02:58 --> 00:03:02 at? They're looking at a transcript and the written
00:03:02 --> 00:03:04 exhibits and whatever briefs got filed on appeal.
00:03:04 --> 00:03:06 And that's essentially it. They're not really
00:03:06 --> 00:03:09 looking at the hearing itself. Right. So the
00:03:09 --> 00:03:12 APA says that the hearing record is the exclusive
00:03:12 --> 00:03:16 basis for decision. Here's 5 U .S .C. 556. Okay.
00:03:16 --> 00:03:19 It says the transcript of testimony and exhibits
00:03:19 --> 00:03:22 together with all papers and requests filed in
00:03:22 --> 00:03:25 the proceeding constitutes the exclusive record
00:03:25 --> 00:03:28 for decision. The exclusive record. So this means
00:03:28 --> 00:03:30 the reviewing body is working from paper. They
00:03:30 --> 00:03:32 don't see the witnesses testify. They don't watch
00:03:32 --> 00:03:34 the claimant shift in their chair. They don't
00:03:34 --> 00:03:36 see them wince when they stand up. Everything
00:03:36 --> 00:03:38 that the ALJ observed in the room that informed
00:03:38 --> 00:03:41 that credibility finding. None of it is in the
00:03:41 --> 00:03:44 transcript. And a couple episodes ago, we talked
00:03:44 --> 00:03:46 about how credibility findings are essentially
00:03:46 --> 00:03:50 unreviewable on appeal because you just weren't
00:03:50 --> 00:03:53 in the room. Right. So that's how federal courts
00:03:53 --> 00:03:55 are viewing it. But the agency reviewing bodies
00:03:55 --> 00:03:57 can operate under a slightly different rule.
00:03:57 --> 00:04:00 So here's 5 URC 557. Okay. It says on appeal
00:04:00 --> 00:04:04 from or review from the initial decision, the
00:04:04 --> 00:04:06 agency has all the powers which it would have
00:04:06 --> 00:04:10 in making the initial decision. All the powers,
00:04:10 --> 00:04:13 like meaning they can redo the case from scratch,
00:04:13 --> 00:04:16 even the credibility findings. Yes, even the
00:04:16 --> 00:04:19 credibility findings on paper from a cold record.
00:04:19 --> 00:04:22 So the ALJ who sat three feet away from the witness
00:04:22 --> 00:04:25 and watched her testify for 45 minutes makes
00:04:25 --> 00:04:27 a credibility finding and a commissioner board
00:04:27 --> 00:04:28 can reverse it based on reading a transcript.
00:04:29 --> 00:04:31 That's not them abusing their power. It's what
00:04:31 --> 00:04:34 the APA says they can do. Okay, that seems kind
00:04:34 --> 00:04:36 of messed up, actually. I mean, when we think
00:04:36 --> 00:04:39 about what the powers of discretionary review
00:04:39 --> 00:04:43 are, that seems wildly out of the park. Federal
00:04:43 --> 00:04:45 courts reviewing a trial court's credibility
00:04:45 --> 00:04:50 finding vastly differ on what they find. They
00:04:50 --> 00:04:53 just can't redo it from a cold record. The trial
00:04:53 --> 00:04:57 judge saw the witness. Yes. But the agency review
00:04:57 --> 00:05:01 body is able to redo the ALJ's credibility finding?
00:05:02 --> 00:05:04 And then the federal court reviewing the agency
00:05:04 --> 00:05:07 defers again because the agency is the finder
00:05:07 --> 00:05:10 of fact. So the one party who actually saw the
00:05:10 --> 00:05:12 witness, the ALJ, is the one whose credibility
00:05:12 --> 00:05:15 finding gets replaced. You can't just swap in
00:05:15 --> 00:05:17 a finding that she did make eye contact when
00:05:17 --> 00:05:19 the ALJ who was there said she didn't. But you
00:05:19 --> 00:05:21 can change the outcome that flows from that finding.
00:05:22 --> 00:05:24 The asymmetry runs in the wrong direction. OK,
00:05:25 --> 00:05:28 give me a concrete example. All right. Let's
00:05:28 --> 00:05:31 take an SEC insider trading case. So the respondent
00:05:31 --> 00:05:32 goes through an evidentiary hearing in front
00:05:32 --> 00:05:36 of an SEC ALJ. The ALJ watches the respondent
00:05:36 --> 00:05:39 testify across multiple days, watches the government's
00:05:39 --> 00:05:41 cooperating witness testify, makes a series of
00:05:41 --> 00:05:44 credibility findings that the respondent's explanation
00:05:44 --> 00:05:47 of the trades are plausible, that the cooperator
00:05:47 --> 00:05:49 is shading testimony to fit their cooperation
00:05:49 --> 00:05:52 agreement, and the ALJ finds... insufficient
00:05:52 --> 00:05:54 evidence of center and dismisses the proceedings.
00:05:55 --> 00:05:57 So under the SEC rules, that initial decision
00:05:57 --> 00:06:00 doesn't become final on its own. The enforcement
00:06:00 --> 00:06:02 division petitions the commissioner for review,
00:06:02 --> 00:06:04 and the commission, or five politically appointed
00:06:04 --> 00:06:07 commissioners who oversee the enforcement division,
00:06:07 --> 00:06:10 take the case. They read the transcript. They
00:06:10 --> 00:06:12 didn't see the respondent testify. They don't
00:06:12 --> 00:06:14 see the cooperator. What they see is a paper
00:06:14 --> 00:06:16 record where the trades look suspicious and the
00:06:16 --> 00:06:18 respondent's explanation looks constructed after
00:06:18 --> 00:06:21 the fact. So they re -waive credibility on a
00:06:21 --> 00:06:23 cold record and reverse. Violations found, sanctions
00:06:23 --> 00:06:26 imposed. Okay, and the federal court reviewing
00:06:26 --> 00:06:30 the commission? So then that court applies substantial
00:06:30 --> 00:06:32 evidence review to the commission's determination,
00:06:32 --> 00:06:36 not to the ALJ's. The original credibility finding
00:06:36 --> 00:06:38 is gone and substantial evidence is a deferential
00:06:38 --> 00:06:40 standard. The court's asking whether a reasonable
00:06:40 --> 00:06:42 person looking at the record could reach the
00:06:42 --> 00:06:45 conclusion the agency reached, not whether the
00:06:45 --> 00:06:47 agency reached the right conclusion. Which means
00:06:47 --> 00:06:49 if the record contains evidence pointing in either
00:06:49 --> 00:06:53 direction, the agency wins. I mean, that's usually
00:06:53 --> 00:06:56 possible to reach either conclusion from a record
00:06:56 --> 00:06:59 that has evidence that runs on both sides. Usually
00:06:59 --> 00:07:02 possible. So the practical effect is that the
00:07:02 --> 00:07:04 ALJ who saw the witness gets overturned, the
00:07:04 --> 00:07:06 commissioner board that didn't see the witness
00:07:06 --> 00:07:09 gets deferred to, and the person who lost in
00:07:09 --> 00:07:12 that room can't recover. The asymmetry survives
00:07:12 --> 00:07:15 all the way through federal court. Okay. I want
00:07:15 --> 00:07:17 to flag something before we move on. You picked
00:07:17 --> 00:07:21 the SEC for that example. Is that because every
00:07:21 --> 00:07:23 agency works the same way or because the SEC
00:07:23 --> 00:07:27 is the place where all the powers language really
00:07:27 --> 00:07:30 bites? So that actually does matter for this
00:07:30 --> 00:07:33 episode. The APA default, the all the powers
00:07:33 --> 00:07:36 language, that's the statutory baseline. But
00:07:36 --> 00:07:38 individual agencies have some regulations that
00:07:38 --> 00:07:41 can pull back from that baseline. Social Security
00:07:41 --> 00:07:43 has one. Immigration has one. And they pull back
00:07:43 --> 00:07:47 in different ways and to different degrees. Let's
00:07:47 --> 00:07:51 look at the SSA regulation then. First, since
00:07:51 --> 00:07:53 that's the one that affects the most people.
00:07:53 --> 00:07:59 All right. So here's 20 CFR 404970. Okay. The
00:07:59 --> 00:08:01 appeals counsel will review a case at a party's
00:08:01 --> 00:08:04 request or on its own motion if there appears
00:08:04 --> 00:08:06 to be an abuse of discretion by the administrative
00:08:06 --> 00:08:10 law judge, there is an error of law, the action
00:08:10 --> 00:08:12 findings or conclusions in the hearing decision
00:08:12 --> 00:08:14 are not supported by substantial evidence, or
00:08:14 --> 00:08:16 there is a broad policy or procedural issue.
00:08:16 --> 00:08:19 Okay. That language, not supported by substantial
00:08:19 --> 00:08:22 evidence, that's the same deferential standard
00:08:22 --> 00:08:24 the federal court then will apply later. Right.
00:08:25 --> 00:08:27 Same standard, and they've moved it back to the
00:08:27 --> 00:08:30 review stage. So here, the Social Security Administration
00:08:30 --> 00:08:32 wrote its own regulation that constrains the
00:08:32 --> 00:08:35 appeals counsel below the APA default. The counsel
00:08:35 --> 00:08:38 can't just decide it would have weighed the credibility
00:08:38 --> 00:08:40 differently. It has to find that the ALJ's conclusion
00:08:40 --> 00:08:42 isn't supported by substantial evidence in the
00:08:42 --> 00:08:45 record. Meaning a reasonable person couldn't
00:08:45 --> 00:08:47 have reached that conclusion. That's a much higher
00:08:47 --> 00:08:48 bar than we read the transcript differently.
00:08:49 --> 00:08:51 So the scenario where the appeals counsel just
00:08:51 --> 00:08:54 substitutes its own credibility call from a cold
00:08:54 --> 00:08:57 transcript, that's not really what's happening
00:08:57 --> 00:09:00 at SSA. Right. Not on the front end. The constraint's
00:09:00 --> 00:09:03 real. The SSA problem is different, and we'll
00:09:03 --> 00:09:05 get to it. It's mostly about review not happening
00:09:05 --> 00:09:07 at all and about what the agency does with new
00:09:07 --> 00:09:10 evidence. Immigration walks back the APA default
00:09:10 --> 00:09:12 even further, though with a different regulation
00:09:12 --> 00:09:15 we'll be... when we get to model two. So the
00:09:15 --> 00:09:17 picture is a spectrum. We have the APA default
00:09:17 --> 00:09:20 at one end, agency heads can redo everything.
00:09:20 --> 00:09:23 We've got social security in the middle, there's
00:09:23 --> 00:09:25 substantial evidence trigger for review, and
00:09:25 --> 00:09:27 immigration at the far end, a clear error review
00:09:27 --> 00:09:30 of facts, the strictest constraint. The enforcement
00:09:30 --> 00:09:33 agencies operate closest to the default. They're
00:09:33 --> 00:09:35 the place where all the powers isn't a theoretical
00:09:35 --> 00:09:38 baseline, it's the operating standard. So each
00:09:38 --> 00:09:41 agency has effectively decided how much of the
00:09:41 --> 00:09:44 APA default it actually wants to use. Yes. And
00:09:44 --> 00:09:46 those choices weren't necessarily made by Congress.
00:09:46 --> 00:09:48 They are made in many cases by the agency themselves
00:09:48 --> 00:09:51 for their own rulemaking. That's worth remembering
00:09:51 --> 00:09:55 for the rest of the episode. Okay. Let's talk
00:09:55 --> 00:09:58 about the record. The record the reviewing body
00:09:58 --> 00:10:02 sees, can parties add to it? If it's new evidence
00:10:02 --> 00:10:04 that comes in after the hearing, can it come
00:10:04 --> 00:10:07 in at review? It can sometimes, but the window
00:10:07 --> 00:10:10 is narrower and it narrows further at every stage.
00:10:10 --> 00:10:13 So let's look at Social Security, for instance.
00:10:13 --> 00:10:17 Okay. That was the rule that required the evidence
00:10:17 --> 00:10:20 to be new, material, tied to the period before
00:10:20 --> 00:10:23 the hearing, and likely to actually change the
00:10:23 --> 00:10:25 outcome. Right. It was four requirements stacked
00:10:25 --> 00:10:28 on top of each other. And on top of that, you
00:10:28 --> 00:10:30 also needed good cause for why the evidence wasn't
00:10:30 --> 00:10:33 submitted earlier. So what qualifies as good
00:10:33 --> 00:10:36 cause? The regulation lists a few things, like
00:10:36 --> 00:10:38 the agency might have misled you, or you had
00:10:38 --> 00:10:39 a limitation that prevented you from submitting
00:10:39 --> 00:10:42 it, some unusual circumstance like a serious
00:10:42 --> 00:10:45 illness or destroyed records, evidence that you
00:10:45 --> 00:10:48 were actively seeking it but didn't receive it
00:10:48 --> 00:10:51 in time. Okay, walk me through what this looks
00:10:51 --> 00:10:53 like when it goes wrong for someone, though.
00:10:54 --> 00:10:56 All right, so here's the common version. A claimant's
00:10:56 --> 00:10:58 treating physician wrote an opinion a year before
00:10:58 --> 00:11:00 the hearing that would have... directly addressed
00:11:00 --> 00:11:03 with the ALJ ends up denying the person based
00:11:03 --> 00:11:06 on. And nobody submitted it, maybe because the
00:11:06 --> 00:11:08 representative didn't know it existed, or maybe
00:11:08 --> 00:11:09 the claimant didn't mention it. Maybe it got
00:11:09 --> 00:11:12 misfiled. The hearing happens without it and
00:11:12 --> 00:11:15 the ALJ denies it. And the representative finds
00:11:15 --> 00:11:17 it after the hearing. Right. And so they try
00:11:17 --> 00:11:19 and submit it to the appeals counsel, but it's
00:11:19 --> 00:11:22 too late. There's no good cause for why it wasn't
00:11:22 --> 00:11:23 in the hearing record. So the appeals counsel
00:11:23 --> 00:11:26 declines to consider it. Then the case goes to
00:11:26 --> 00:11:28 federal court. And the federal court reviews
00:11:28 --> 00:11:30 the record the agency had. Right, which didn't
00:11:30 --> 00:11:33 include this opinion. So the court doesn't ask
00:11:33 --> 00:11:35 whether the opinion was persuasive. It asks whether
00:11:35 --> 00:11:38 it's in the record. It isn't, so the person loses.
00:11:38 --> 00:11:42 Not on the merits of this opinion by the doctor
00:11:42 --> 00:11:44 that didn't get in, but on the procedural failure
00:11:44 --> 00:11:47 to get it there. Okay, but could she refile?
00:11:48 --> 00:11:50 So she absolutely can. You can always refile
00:11:50 --> 00:11:52 in Social Security. But the problem is that when
00:11:52 --> 00:11:56 you refile, you have a new onset date. If her
00:11:56 --> 00:11:59 original application was filed in 2020 and she
00:11:59 --> 00:12:03 refiles in 2026, she's losing years of back benefits.
00:12:03 --> 00:12:06 This can be tens of thousands of dollars. So
00:12:06 --> 00:12:08 every stage after the hearing is damage limitation,
00:12:09 --> 00:12:12 not really a fresh opportunity. Yes, that's a
00:12:12 --> 00:12:14 good way to put it. And the implication for review
00:12:14 --> 00:12:16 is important. If the record at the hearing is
00:12:16 --> 00:12:18 what it is, if the evidence that would have changed
00:12:18 --> 00:12:21 the outcome wasn't in the file, then even a well
00:12:21 --> 00:12:23 -intentioned review body can't fix it. They're
00:12:23 --> 00:12:25 looking at the wrong record. The hearing is the
00:12:25 --> 00:12:27 only real opportunity to shape what every subsequent
00:12:27 --> 00:12:31 reviewer will see. All right. Let's get to the
00:12:31 --> 00:12:33 question we opened with. What does the review
00:12:33 --> 00:12:37 layer actually do? Because I guess just from
00:12:37 --> 00:12:40 the volume issues alone, the different agencies
00:12:40 --> 00:12:43 handled this very differently. Yes. dramatically
00:12:43 --> 00:12:44 differently. So we're going to look at three
00:12:44 --> 00:12:46 different models. Each one represents a genuinely
00:12:46 --> 00:12:49 different answer to what review means at an agency.
00:12:49 --> 00:12:52 So the first model is Social Security. The Appeals
00:12:52 --> 00:12:54 Council receives hundreds of thousands of requests
00:12:54 --> 00:12:57 for review every year, and it grants review in
00:12:57 --> 00:13:00 maybe 15 to 20 percent of cases, which is still
00:13:00 --> 00:13:02 quite a lot if we're talking hundreds of thousands
00:13:02 --> 00:13:04 of requests for review. But the rest get a one
00:13:04 --> 00:13:06 page denial that says the appeals counsel has
00:13:06 --> 00:13:09 denied your request for review. There's no analysis.
00:13:09 --> 00:13:11 There's no engagement with the arguments. There's
00:13:11 --> 00:13:13 no explanation of what they would have looked
00:13:13 --> 00:13:16 at. OK, so most Social Security claimants who
00:13:16 --> 00:13:19 appeal to the appeals counsel do not actually
00:13:19 --> 00:13:21 get reviewed. Right. They don't. And it's not
00:13:21 --> 00:13:23 just that. It's that the design of the denial
00:13:23 --> 00:13:26 letter itself is one of the cleanest examples
00:13:26 --> 00:13:28 we can see. of the consequences of procedural
00:13:28 --> 00:13:31 design, even if it's not intended to be that
00:13:31 --> 00:13:34 way. So unrepresented claimants can misread it,
00:13:34 --> 00:13:35 and when they misread it, they lose their case.
00:13:36 --> 00:13:38 We talked about this a couple of episodes ago,
00:13:38 --> 00:13:40 but there's more to say. This isn't something
00:13:40 --> 00:13:42 anyone designed to produce this effect. It's
00:13:42 --> 00:13:43 a procedural artifact, but the effect is real.
00:13:44 --> 00:13:48 Okay, take a second. Walk me through the intricacies
00:13:48 --> 00:13:52 of this. All right. So the Appeals Council denial
00:13:52 --> 00:13:55 of review is the procedural event that triggers
00:13:55 --> 00:13:56 the claimant's right to file in federal court.
00:13:56 --> 00:13:59 The denial letter is the gate being opened. The
00:13:59 --> 00:14:01 claimant has 60 days from the date of the letter
00:14:01 --> 00:14:03 to file a civil action in federal court. Once
00:14:03 --> 00:14:06 they miss this window, the right is gone. And
00:14:06 --> 00:14:09 the letter itself? So the letter itself is a
00:14:09 --> 00:14:11 paragraph. It tells the claimant that the Appeals
00:14:11 --> 00:14:13 Council denied the request for review. It doesn't
00:14:13 --> 00:14:15 analyze their case. It doesn't engage with their
00:14:15 --> 00:14:17 arguments. It doesn't explain in plain terms
00:14:17 --> 00:14:18 that they have the right to sue the commissioner
00:14:18 --> 00:14:21 of Social Security in federal court within 60
00:14:21 --> 00:14:23 days. That information is somewhere on the boilerplate.
00:14:23 --> 00:14:26 But the letter reads to a late reader like a
00:14:26 --> 00:14:29 final rejection. Because it's a final rejection
00:14:29 --> 00:14:32 at the agency level. Yes. This is why we have
00:14:32 --> 00:14:35 this consequential mismatch. So denied an order
00:14:35 --> 00:14:38 of usage means you lost. Denied in this administrative
00:14:38 --> 00:14:41 law usage means you've now exhausted your administrative
00:14:41 --> 00:14:43 remedies. The 60 -day clock to federal court
00:14:43 --> 00:14:45 is running. Those are practically opposite, but
00:14:45 --> 00:14:47 the letter doesn't make clear that it's the second
00:14:47 --> 00:14:51 one. OK, so unrepresented claimants see the letter
00:14:51 --> 00:14:54 and apply the ordinary language meaning and think,
00:14:54 --> 00:14:57 I'm done. Right. So they think it's over. They
00:14:57 --> 00:15:00 don't file. The clock runs out. And by the time
00:15:00 --> 00:15:02 they figure out that they did have a right, they
00:15:02 --> 00:15:05 didn't exercise it. Their time is gone. And so
00:15:05 --> 00:15:07 is the case. Represented claimants, meanwhile,
00:15:07 --> 00:15:09 know that this letter is just a checkpoint before
00:15:09 --> 00:15:11 they can finally file in federal court. They've
00:15:11 --> 00:15:13 been expecting it. They filed the appeal in federal
00:15:13 --> 00:15:16 court and the case proceeds. Which connects back
00:15:16 --> 00:15:18 to everything we said in the last episode about
00:15:18 --> 00:15:21 privatized knowledge. Right. So the meaning of
00:15:21 --> 00:15:23 this denial letter, which is the most consequential
00:15:23 --> 00:15:25 piece of paper the claimant potentially receives
00:15:25 --> 00:15:28 in this entire process, is information that's
00:15:28 --> 00:15:30 easily available to experienced counsel and not
00:15:30 --> 00:15:32 meaningfully available to somebody who's unrepresented.
00:15:33 --> 00:15:36 It doesn't really explain itself. at least not
00:15:36 --> 00:15:38 in a way that the people it's impacting are likely
00:15:38 --> 00:15:41 to understand. And I'm not saying that the people
00:15:41 --> 00:15:43 designing this were trying to mislead. This is
00:15:43 --> 00:15:45 just an artifact of administrative law vocabulary
00:15:45 --> 00:15:48 meeting ordinary language expectations. But the
00:15:48 --> 00:15:50 effect on these underrepresented claimants who
00:15:50 --> 00:15:51 take the letter at face value is still that they
00:15:51 --> 00:15:53 lose a case they wouldn't have necessarily lost.
00:15:54 --> 00:15:56 Yeah. So in this model, review mostly doesn't
00:15:56 --> 00:15:59 happen. And the review denial itself produces
00:15:59 --> 00:16:02 outcomes through a mechanism unrelated to the
00:16:02 --> 00:16:04 merits. Right. So the summary denial functions
00:16:04 --> 00:16:08 as a case -ending event for a huge percentage
00:16:08 --> 00:16:10 of unrepresented claimants. The ALJ's decision
00:16:10 --> 00:16:13 stands. Whatever inconsistency existed across
00:16:13 --> 00:16:16 the thousands of ALJs is preserved, not corrected.
00:16:16 --> 00:16:19 So Model 1 is this exhaustion step, not necessarily
00:16:19 --> 00:16:22 a correction mechanism because they're not correcting
00:16:22 --> 00:16:25 so many of these. And the exhaustion step itself
00:16:25 --> 00:16:27 is where a meaningful percentage of unrepresented
00:16:27 --> 00:16:30 claimants lose by default. Let's talk about Model
00:16:30 --> 00:16:33 2. All right. So this one's immigration. Immigration
00:16:33 --> 00:16:36 judge decisions get forwarded to the BIA if either
00:16:36 --> 00:16:38 party appeals. So the BIA then either conducts
00:16:38 --> 00:16:41 full review or issues a summary affirmance, a
00:16:41 --> 00:16:43 brief order that sometimes might just be a sentence
00:16:43 --> 00:16:45 saying affirmed for the reasons stated by the
00:16:45 --> 00:16:48 immigration judge. OK, but what happens when
00:16:48 --> 00:16:51 the BIA does do full review? So this is where
00:16:51 --> 00:16:53 things can get different from Social Security.
00:16:53 --> 00:16:56 In Model 1, the case is the review counsel. doesn't
00:16:56 --> 00:16:59 review just disappear. The ALJ's decision stands.
00:16:59 --> 00:17:02 Nothing about it travels anywhere. The BIA can
00:17:02 --> 00:17:04 designate decisions as presidential. And when
00:17:04 --> 00:17:06 it does, those decisions bind every immigration
00:17:06 --> 00:17:08 judge in the country. So the same review body
00:17:08 --> 00:17:11 that issues one sentence affirmances in most
00:17:11 --> 00:17:14 cases is also selectively making law that governs
00:17:14 --> 00:17:17 every case that comes after it. The stakes of
00:17:17 --> 00:17:18 which cases get full review are a lot higher
00:17:18 --> 00:17:21 than they look. Before we get to which cases
00:17:21 --> 00:17:24 get reviewed and what becomes precedent. So.
00:17:24 --> 00:17:26 Before we really look back to the spectrum we
00:17:26 --> 00:17:29 set up earlier, you said that immigration walks
00:17:29 --> 00:17:33 back the APA default at its furthest. So here.
00:17:34 --> 00:17:38 Let's look at 8 CFR 1003 .1. The board will not
00:17:38 --> 00:17:40 engage in a de novo review of findings of fact
00:17:40 --> 00:17:43 determined by an immigration judge. Facts determined
00:17:43 --> 00:17:46 by the immigration judge, including findings
00:17:46 --> 00:17:49 as to the credibility of testimony, shall be
00:17:49 --> 00:17:51 reviewed only to determine whether the findings
00:17:51 --> 00:17:54 of the immigration judge are clearly erroneous.
00:17:54 --> 00:17:57 Clear error. That's a standard the federal appeals
00:17:57 --> 00:18:00 court uses in district judges' findings. Yeah.
00:18:00 --> 00:18:03 So the BIA can't do what the SEC commissioner
00:18:03 --> 00:18:06 can do. It can't sit down with a transcript and
00:18:06 --> 00:18:09 decide that the asylum applicant's flat affect
00:18:09 --> 00:18:11 reads as evasive rather than traumatized and
00:18:11 --> 00:18:14 substitute its own credibility call. If the IJ
00:18:14 --> 00:18:16 found the testimony credible, the board is stuck
00:18:16 --> 00:18:18 with that unless it's left with the definite
00:18:18 --> 00:18:22 and firm conviction the IJ got it wrong. Immigration's
00:18:22 --> 00:18:23 constraints on fact -finding is the tightest
00:18:23 --> 00:18:26 of the three. Which makes immigration look like
00:18:26 --> 00:18:29 the agency that has its house in order. The fact
00:18:29 --> 00:18:33 finder is protected. So, yes and no. The review
00:18:33 --> 00:18:36 problem with immigration doesn't live at that
00:18:36 --> 00:18:39 factual level. The BIA reviews applications of
00:18:39 --> 00:18:42 facts to law de novo, whether a credible applicant's
00:18:42 --> 00:18:44 experience adds up to persecution, or whether
00:18:44 --> 00:18:46 a group counts as a particular social group,
00:18:46 --> 00:18:48 or whether there's a nexus to a protected ground.
00:18:49 --> 00:18:51 Those are where outcomes live, and the bigger
00:18:51 --> 00:18:53 lever sits above the board entirely. Which sounds
00:18:53 --> 00:18:56 more like what a listener might hope internal
00:18:56 --> 00:18:59 review looks like. It's closer, but the piece
00:18:59 --> 00:19:01 that makes immigration different from the model
00:19:01 --> 00:19:03 you might be imagining is that the attorney general
00:19:03 --> 00:19:05 can reach into any case at any stage and decide
00:19:05 --> 00:19:08 it for himself. Okay, just to be clear, though,
00:19:08 --> 00:19:12 the attorney general personally? Yeah, so let's
00:19:12 --> 00:19:30 look back at 8 CFR 1003 .1. So when the BIA does
00:19:30 --> 00:19:32 decide to write, it can bind every immigration
00:19:32 --> 00:19:35 judge in the country, except the board itself,
00:19:35 --> 00:19:37 which is still free to change their minds here.
00:19:38 --> 00:19:41 Okay, so it says, except as board decisions may
00:19:41 --> 00:19:43 be modified or overruled by the board or the
00:19:43 --> 00:19:46 attorney general, decisions of the board and
00:19:46 --> 00:19:48 decisions of the attorney general are binding
00:19:48 --> 00:19:51 on all officers and employees of DHS or immigration
00:19:51 --> 00:19:54 judges in the administration of the immigration
00:19:54 --> 00:19:56 laws of the United States. So let's look at the
00:19:56 --> 00:19:58 regulation that lets the attorney general pull
00:19:58 --> 00:20:01 in the case in the first place. This is 8 CFR
00:20:01 --> 00:20:05 1003. We looked at D, we looked at G, now we're
00:20:05 --> 00:20:08 looking at H. OK, it says the board shall refer
00:20:08 --> 00:20:10 to the attorney general for the review of its
00:20:10 --> 00:20:13 decision on all cases that the attorney general
00:20:13 --> 00:20:17 directs the board to refer to them. So based
00:20:17 --> 00:20:19 on that, the attorney general can pick a case,
00:20:20 --> 00:20:23 any case pending or decided, and issue a decision
00:20:23 --> 00:20:25 that binds every immigration judge in the country.
00:20:25 --> 00:20:28 Any case. And this has been used in ways that
00:20:28 --> 00:20:30 dramatically shift doctrine. Let's talk about
00:20:30 --> 00:20:33 asylum for survivors of domestic violence. In
00:20:33 --> 00:20:38 2014, the BIA published Matter of ARCG, where
00:20:38 --> 00:20:40 it held that a Guatemalan woman in a relationship
00:20:40 --> 00:20:42 she couldn't escape could be a member of a particular
00:20:42 --> 00:20:45 social group for asylum purposes. Survivors of
00:20:45 --> 00:20:48 domestic violence had a presidential anchor.
00:20:49 --> 00:20:52 Then in 2018, Attorney General Sessions certified
00:20:52 --> 00:20:56 the case to himself and overruled ARCG. and wrote
00:20:56 --> 00:20:59 that claims involving domestic violence by private
00:20:59 --> 00:21:02 actors generally won't qualify for asylum. So
00:21:02 --> 00:21:04 one certification, and now every immigration
00:21:04 --> 00:21:07 judge is bound the other way. No statutes have
00:21:07 --> 00:21:09 changed. There's no notice of comment rulemaking.
00:21:09 --> 00:21:13 And then in 2021, Attorney General Garland certified
00:21:13 --> 00:21:15 A .B. to himself and vacated Sessions' decision,
00:21:16 --> 00:21:19 reinstating ARCG. Domestic violence is back as
00:21:19 --> 00:21:24 a viable basis. Then in 2025... A .G. Bondi certified
00:21:24 --> 00:21:27 a different case, the Matters S .F .F .M., and
00:21:27 --> 00:21:30 used it to overrule Garland and reinstate A .B.
00:21:30 --> 00:21:33 Three swings on the same doctrine in seven years,
00:21:33 --> 00:21:36 all through the same mechanism. Same story for
00:21:36 --> 00:21:39 family -based asylum. The Matter of L .A. has
00:21:39 --> 00:21:41 been certified and recertified across administrations.
00:21:41 --> 00:21:44 These aren't individual case decisions. They're
00:21:44 --> 00:21:46 substantive reinterpretations of who qualifies
00:21:46 --> 00:21:48 for protection imposed through a single referral.
00:21:49 --> 00:21:52 And because it's binding precedent, it flows
00:21:52 --> 00:21:54 down to every immigration judge going forward.
00:21:54 --> 00:21:57 Exactly. This is why review and immigration is
00:21:57 --> 00:22:00 so different. Review isn't just error correction,
00:22:00 --> 00:22:01 and it isn't just case -by -case consistency
00:22:01 --> 00:22:04 building. At immigration, it's also an active
00:22:04 --> 00:22:07 policy instrument wielded by a politically appointed
00:22:07 --> 00:22:09 head of a department who can reach into pending
00:22:09 --> 00:22:12 adjudications or decided adjudications and rewrite
00:22:12 --> 00:22:16 the rules. The BIA exists formally as the appellate
00:22:16 --> 00:22:18 body. The attorney general sits above it with
00:22:18 --> 00:22:20 the power to pull in any case and change the
00:22:20 --> 00:22:23 doctrine. That's not error correction. That is
00:22:23 --> 00:22:25 policy control executed through the review mechanism.
00:22:25 --> 00:22:28 And the reason to emphasize that structurally
00:22:28 --> 00:22:30 is that other agencies have versions of the same
00:22:30 --> 00:22:33 agreement. Right. So the commissions and enforcement
00:22:33 --> 00:22:36 agencies have less sweeping versions. Their review
00:22:36 --> 00:22:38 authority is real, but it doesn't carry this
00:22:38 --> 00:22:41 attorney general certification's sheer doctrinal
00:22:41 --> 00:22:43 reach. The appeals council at Social Security
00:22:43 --> 00:22:46 has a weaker version. It reviews little. It doesn't
00:22:46 --> 00:22:48 designate precedent. It doesn't really make doctrine
00:22:48 --> 00:22:51 like that. What the immigration example shows
00:22:51 --> 00:22:53 is what the review mechanism can look like when
00:22:53 --> 00:22:55 the agency head has both the legal authority
00:22:55 --> 00:22:58 and the political will to use it. And that's
00:22:58 --> 00:23:01 a completely different kind of thing from what
00:23:01 --> 00:23:03 the appeals council does. The appeals council
00:23:03 --> 00:23:05 at Social Security mostly doesn't review. The
00:23:05 --> 00:23:07 BIA plus the attorney general certification reviews
00:23:07 --> 00:23:10 a lot. And the review includes reshaping doctrine
00:23:10 --> 00:23:12 in real time through the cases that the attorney
00:23:12 --> 00:23:15 general selects. The variable across agencies
00:23:15 --> 00:23:18 isn't whether the review mechanism exists. It's
00:23:18 --> 00:23:20 how aggressively the mechanism gets used. OK,
00:23:20 --> 00:23:24 so we've talked about two models. How about the
00:23:24 --> 00:23:26 third? All right. So now we're talking about
00:23:26 --> 00:23:29 enforcement agencies. This is like the SEC, the
00:23:29 --> 00:23:32 FTC, the NLRB. At these agencies, review isn't
00:23:32 --> 00:23:35 discretionary. It is automatic. When an ALJ issues
00:23:35 --> 00:23:37 an initial decision, it goes to the commissioner
00:23:37 --> 00:23:40 board by default. The ALJ's decision is essentially
00:23:40 --> 00:23:42 just a recommendation. It doesn't become final
00:23:42 --> 00:23:44 until the agency heads either affirm it, modify
00:23:44 --> 00:23:47 it, or reverse it, or let the review window expire.
00:23:48 --> 00:23:52 So at the SEC, every single case gets a full
00:23:52 --> 00:23:54 review by the commissioners? Functionally, yes.
00:23:55 --> 00:23:57 The commission is expected to issue its own final
00:23:57 --> 00:23:59 decision. The ALJ hearing is always preliminary.
00:23:59 --> 00:24:03 The commission has the last word. Okay, why does
00:24:03 --> 00:24:06 this exist? difference exist? I mean, why does
00:24:06 --> 00:24:09 Social Security, which deals with real people
00:24:09 --> 00:24:12 and real problems, barely review while the SEC
00:24:12 --> 00:24:17 reviews everything? This is partly based on volume
00:24:17 --> 00:24:19 and partly by design. Social Security is processing
00:24:19 --> 00:24:22 millions of disability claims, and there are
00:24:22 --> 00:24:25 hundreds, thousands of appeals filed every year.
00:24:25 --> 00:24:28 It's mechanistic. Mechanical. It's really impossible
00:24:28 --> 00:24:30 for the appeals counsel to meaningfully review
00:24:30 --> 00:24:33 every case. The SEC brings maybe a couple hundred
00:24:33 --> 00:24:36 enforcement actions a year, and five commissioners
00:24:36 --> 00:24:38 can genuinely look at a couple hundred cases.
00:24:39 --> 00:24:41 Okay, so at one end you have benefits claimants
00:24:41 --> 00:24:45 who mostly get no meaningful review, but at the
00:24:45 --> 00:24:47 other end you have enforcement respondents who
00:24:47 --> 00:24:50 get guaranteed review by the same institution
00:24:50 --> 00:24:52 that brought the case against them. Both of those
00:24:52 --> 00:24:54 sentences are doing work in that tradeoff. We'll
00:24:54 --> 00:24:55 come back to the second one in a minute. But
00:24:55 --> 00:24:58 first, I want to stay with this. So in each of
00:24:58 --> 00:25:01 the three models, the review body exercises selection
00:25:01 --> 00:25:04 power over its docket. This is a big deal since
00:25:04 --> 00:25:06 it's a structural feature of agency review that
00:25:06 --> 00:25:09 courts don't have. OK, because courts don't get
00:25:09 --> 00:25:12 to do that, then they take what comes to them.
00:25:12 --> 00:25:14 Yeah. So think about how a circuit court gets
00:25:14 --> 00:25:17 cases. The parties appeal. The court gets whatever
00:25:17 --> 00:25:20 the parties bring. The Supreme Court can deny
00:25:20 --> 00:25:22 cert if it wants. Eric can issue summary affirmances.
00:25:22 --> 00:25:25 But historically, it can't reach down into a
00:25:25 --> 00:25:27 trial court's docket and pull up a case it wants
00:25:27 --> 00:25:29 to decide. The court is working from the pool
00:25:29 --> 00:25:31 of cases that other people brought. What about
00:25:31 --> 00:25:35 the agencies? So agencies can reach. The Appeals
00:25:35 --> 00:25:37 Council can review a Social Security case on
00:25:37 --> 00:25:39 its own motion, even if nobody appealed. The
00:25:39 --> 00:25:42 BIA can do the same. The SEC Commission can initiate
00:25:42 --> 00:25:45 review. And the Attorney General, as we saw,
00:25:45 --> 00:25:48 can certify any case and decide it himself. So
00:25:48 --> 00:25:50 this is really about what the reviewing body
00:25:50 --> 00:25:53 picks. Right. And that isn't a small thing. In
00:25:53 --> 00:25:55 a court system, we have law that develops because
00:25:55 --> 00:25:57 the litigants themselves decide to push the question
00:25:57 --> 00:26:00 up. The Fifth Circuit rules on the cases that
00:26:00 --> 00:26:03 come to it. Parties can force the court to confront
00:26:03 --> 00:26:05 a question by appealing. That gives litigants
00:26:05 --> 00:26:07 a meaningful role in how the law develops. They
00:26:07 --> 00:26:09 pick which issues get resolved when. And in the
00:26:09 --> 00:26:13 agency system? The agency leadership picks which
00:26:13 --> 00:26:14 questions reach the precedent -making level,
00:26:15 --> 00:26:17 which facts get adjudicated at that level, which
00:26:17 --> 00:26:19 doctrines get announced. The president is curated
00:26:19 --> 00:26:21 from the top down, not pushed from the bottom
00:26:21 --> 00:26:24 up. So the selection is its own form of control.
00:26:24 --> 00:26:26 Yes, and it is a significant form of control.
00:26:26 --> 00:26:29 So an appeals court can decide which decisions
00:26:29 --> 00:26:32 are going to be precedential, but it still is
00:26:32 --> 00:26:34 going to have to address every issue. But if
00:26:34 --> 00:26:36 the agency wants the law to develop in a particular
00:26:36 --> 00:26:38 direction, it can just pick the cases that go
00:26:38 --> 00:26:40 in that direction. The attorney general's certification
00:26:40 --> 00:26:42 power is the most visible version of this, but
00:26:42 --> 00:26:46 every review's body exercises some version of
00:26:46 --> 00:26:48 it. Deciding what to review is deciding what
00:26:48 --> 00:26:50 becomes law. Which means the review layer isn't
00:26:50 --> 00:26:54 neutral error correction. It's a mechanism that
00:26:54 --> 00:26:56 combines error correction with agenda setting.
00:26:56 --> 00:26:59 Yes. So the textbook version of internal review
00:26:59 --> 00:27:01 would treat it like quality control. The agency
00:27:01 --> 00:27:04 had sample incorrect. The real version is agenda
00:27:04 --> 00:27:07 setting plus quality control, plus at least the
00:27:07 --> 00:27:09 enforcement agencies, a structural conflict we
00:27:09 --> 00:27:12 haven't talked about yet. OK, you flagged this
00:27:12 --> 00:27:15 a minute ago, but let's get back to the enforcement
00:27:15 --> 00:27:18 side. Yeah. So at the SEC, the commissioners
00:27:18 --> 00:27:20 oversee the enforcement division. They approve
00:27:20 --> 00:27:23 investigation targets. They set enforcement priorities.
00:27:23 --> 00:27:25 They allocate resources to the enforcement staff.
00:27:25 --> 00:27:27 And then they serve as the appellate tribunal
00:27:27 --> 00:27:29 reviewing whether that same staff proved its
00:27:29 --> 00:27:32 case. Okay. So the investigator -adjudicator
00:27:32 --> 00:27:35 overlap we've been talking about at the institutional
00:27:35 --> 00:27:39 level. Right. So in an ordinary appellate system,
00:27:40 --> 00:27:42 the reviewing body has no institutional stake
00:27:42 --> 00:27:44 in the dispute. The Ninth Circuit isn't supervising
00:27:44 --> 00:27:47 the plaintiff's lawyers. The D .C. Circuit doesn't
00:27:47 --> 00:27:49 set policy for the agencies whose rules it's
00:27:49 --> 00:27:52 reviewing. Enforcement agency review looks completely
00:27:52 --> 00:27:54 different. OK, how does that pattern show up
00:27:54 --> 00:27:57 in the data? So when the SEC reverses the ALJ's
00:27:57 --> 00:27:59 decision, the large majority of those reversals,
00:27:59 --> 00:28:02 about three quarters, favor the enforcement division.
00:28:03 --> 00:28:05 They're either finding violations that the ALJ
00:28:05 --> 00:28:07 didn't find or increasing penalties that the
00:28:07 --> 00:28:10 ALJ imposed. So when the commission steps in,
00:28:10 --> 00:28:12 though, it usually goes against the respondent.
00:28:12 --> 00:28:14 Yes, and you don't necessarily need bad faith
00:28:14 --> 00:28:16 to see why. The enforcement division works for
00:28:16 --> 00:28:19 the commission. If the commission regularly rejects
00:28:19 --> 00:28:20 enforcement cases, they're effectively telling
00:28:20 --> 00:28:23 their own staff they're getting it wrong. The
00:28:23 --> 00:28:25 incentives line up in one direction, and they
00:28:25 --> 00:28:27 do that without anyone necessarily acting improperly.
00:28:27 --> 00:28:30 And they're exercising that while applying the
00:28:30 --> 00:28:33 all the powers authority to redo credibility
00:28:33 --> 00:28:35 from a transcript. Yeah. So we have this combination
00:28:35 --> 00:28:37 where the reviewing body has an institutional
00:28:37 --> 00:28:40 preference and the review includes the authority
00:28:40 --> 00:28:42 to overturn credibility findings made by someone
00:28:42 --> 00:28:45 who was actually in the room. And that's not
00:28:45 --> 00:28:46 really a neutral error correction mechanism.
00:28:46 --> 00:28:49 That's something else. OK, what about the benefits
00:28:49 --> 00:28:52 agencies? The Appeals Council doesn't have the
00:28:52 --> 00:28:55 same prosecutorial stake in those cases. Right.
00:28:55 --> 00:28:57 So nobody wins if a claimant loses at the Social
00:28:57 --> 00:29:00 Security Administration. But the Appeals Council
00:29:00 --> 00:29:02 operates within an agency that has budget constraints,
00:29:02 --> 00:29:05 that has processing targets and institutional
00:29:05 --> 00:29:07 views about how disability standards should be
00:29:07 --> 00:29:10 applied. If agency leadership believe the ALJs
00:29:10 --> 00:29:11 are being too generous, the review function can
00:29:11 --> 00:29:13 become a tool to sort of tighten the outcomes.
00:29:14 --> 00:29:16 The structural pressure is different from enforcement,
00:29:16 --> 00:29:19 but it's not absent. OK, is there a counter argument
00:29:19 --> 00:29:22 to all of this? I mean, because I can hear someone
00:29:22 --> 00:29:25 saying, of course, agency heads review their
00:29:25 --> 00:29:29 own agencies, ALJs. They're the experts. They're
00:29:29 --> 00:29:32 accountable. Somebody has to be the final word.
00:29:33 --> 00:29:36 So. Yeah, that's really three arguments. We have
00:29:36 --> 00:29:39 expertise that the agency heads know the statutes
00:29:39 --> 00:29:40 and regulatory landscape better than anyone,
00:29:41 --> 00:29:43 and that without review, these ALJ interpretations
00:29:43 --> 00:29:45 would drift in different directions, as we saw,
00:29:46 --> 00:29:48 which they do in the last episode. And we've
00:29:48 --> 00:29:50 got accountability that agency heads are politically
00:29:50 --> 00:29:52 appointed by the president, so they're more democratically
00:29:52 --> 00:29:56 accountable and career ALJs. And we've also got
00:29:56 --> 00:29:58 uniformity that without this, different ALJs
00:29:58 --> 00:29:59 interpret the same statute in different ways,
00:30:00 --> 00:30:03 with no real mechanism to reconcile them. Yeah,
00:30:03 --> 00:30:06 and those sound like practical arguments, though.
00:30:06 --> 00:30:09 Do they really answer the due process concerns?
00:30:09 --> 00:30:11 They're arguments for why we want some form of
00:30:11 --> 00:30:13 central review. They don't answer the question
00:30:13 --> 00:30:15 of what to do when that review is institutionally
00:30:15 --> 00:30:17 tilted. The system that the APA built leaves
00:30:17 --> 00:30:21 that question open. Okay, let's try to pull these
00:30:21 --> 00:30:23 three models together because I want to make
00:30:23 --> 00:30:26 sure I'm seeing what unifies them. Fabulous.
00:30:26 --> 00:30:29 All right, so at Social Security, review mostly
00:30:29 --> 00:30:33 doesn't happen. The ALJ's decision stands. Hearing
00:30:33 --> 00:30:36 level variation is preserved, not corrected.
00:30:36 --> 00:30:40 At immigration, review does happen, sometimes
00:30:40 --> 00:30:42 in depth, sometimes it produces binding precedent
00:30:42 --> 00:30:47 through designation or AG certification. But
00:30:47 --> 00:30:49 the selection of what to designate or certify
00:30:49 --> 00:30:52 is itself politically directed, and doctrine
00:30:52 --> 00:30:56 can shift with administrations. And at the SEC,
00:30:56 --> 00:31:00 review happens in every single case, but with
00:31:00 --> 00:31:02 a structural institutional lean towards enforcement
00:31:02 --> 00:31:05 division. Yes. And none of those three models
00:31:05 --> 00:31:07 primarily produce consistency. Not by design,
00:31:08 --> 00:31:09 not by accident, not as a side effect of doing
00:31:09 --> 00:31:12 something else. Each one fails to produce consistency
00:31:12 --> 00:31:14 in a different way. The Social Security Administration
00:31:14 --> 00:31:17 fails by not reviewing most cases. Immigration
00:31:17 --> 00:31:19 fails by reviewing through a mechanism directly
00:31:19 --> 00:31:22 controlled by political leadership whose priorities
00:31:22 --> 00:31:25 shift with administrations. And enforcement fails
00:31:25 --> 00:31:27 by reviewing through a body that has an institutional
00:31:27 --> 00:31:29 stake in the outcome. We have these different
00:31:29 --> 00:31:32 failure modes, but the same bottom line. Internal
00:31:32 --> 00:31:34 review is not the mechanism that pulls similar
00:31:34 --> 00:31:37 cases towards similar outcomes. So review can
00:31:37 --> 00:31:40 correct egregious errors. It produces some precedent
00:31:40 --> 00:31:44 where precedent gets designated. It's not zero,
00:31:44 --> 00:31:48 but as the mechanism that's supposed to fix what
00:31:48 --> 00:31:50 the hearing level couldn't, which is what the
00:31:50 --> 00:31:52 textbook description of internal review suggests,
00:31:53 --> 00:31:56 none of these three models is doing that work.
00:31:56 --> 00:31:58 Right. So they're all going to correct some errors
00:31:58 --> 00:32:02 and resolve some conflicts and potentially produce
00:32:02 --> 00:32:04 some precedent. It's not a system with zero value,
00:32:04 --> 00:32:06 but as a mechanism for ensuring that similar
00:32:06 --> 00:32:09 cases come out the same way, which is the thing
00:32:09 --> 00:32:11 a listener might hope for after last week's episode,
00:32:11 --> 00:32:15 it is not that. So it's not the fix for the hearing
00:32:15 --> 00:32:18 level picture. Right. So now we've covered two
00:32:18 --> 00:32:20 layers. We've talked about how ALJs don't produce
00:32:20 --> 00:32:22 consistency on their own. That was last week.
00:32:23 --> 00:32:25 And that the review layer doesn't really produce
00:32:25 --> 00:32:27 it either. That's this week. And neither of these
00:32:27 --> 00:32:30 mechanisms, the formal adjudicated system was
00:32:30 --> 00:32:32 built around, produce the thing we... potentially
00:32:32 --> 00:32:34 want from it most, that consistency. And yet
00:32:34 --> 00:32:37 agency preferences do translate into outcomes.
00:32:37 --> 00:32:40 Aggregate approval rates shift across administrations.
00:32:40 --> 00:32:42 Enforcement priorities become enforcement outcomes.
00:32:43 --> 00:32:45 Doctrinal positions announced by agency leadership
00:32:45 --> 00:32:48 become practice. Which means that whatever is
00:32:48 --> 00:32:51 actually producing those patterns has to be operating
00:32:51 --> 00:32:53 somewhere other than where we've been looking.
00:32:53 --> 00:32:55 Right. So agencies don't have to wait for appeals
00:32:55 --> 00:32:57 to shape outcomes. They shape behavior before
00:32:57 --> 00:33:00 the ALJ even decides the case. Through metrics,
00:33:00 --> 00:33:02 through performance reviews, through guidance
00:33:02 --> 00:33:04 that can function like law even if it's technically
00:33:04 --> 00:33:06 not, and through the quiet institutional pressure
00:33:06 --> 00:33:08 that comes with working inside a hierarchical
00:33:08 --> 00:33:11 organization where your bosses have views. That's
00:33:11 --> 00:33:13 what we haven't really looked at yet, and that
00:33:13 --> 00:33:15 is the next episode. How agencies actually control
00:33:15 --> 00:33:17 outcomes, not by correcting decisions, but by
00:33:17 --> 00:33:19 shaping the conditions under which the decisions
00:33:19 --> 00:33:22 are made. So that does it for today's episode
00:33:22 --> 00:33:24 on administrative remedies. Thank you for joining
00:33:24 --> 00:33:27 us today. Please, if you enjoy this podcast,
00:33:27 --> 00:33:30 enjoy this episode, give us a like on Spotify,
00:33:30 --> 00:33:33 iTunes, or whatever platform you're listening
00:33:33 --> 00:33:35 on. And be sure to tune in next time where we'll
00:33:35 --> 00:33:37 continue to dive into the contours of administrative
00:33:37 --> 00:33:40 law, because remember, you can't fix what you
00:33:40 --> 00:33:41 don't understand.