Finality Explained: When Has the Agency Actually Done Something You Can Sue About?

In November 2013, a man who had been turned down for a job with the Texas Department of Public Safety filed a charge with the Equal Employment Opportunity Commission. The Department had a total ban on hiring convicted felons — not just as troopers, for any job. His theory was disparate impact under Title VII. A year and a half earlier, the EEOC had issued enforcement guidance saying that automatic across-the-board exclusion of people with records disproportionately screens out applicants by race and national origin, and employers probably couldn't justify it. The guidance wasn't a regulation. It had never gone through notice and comment. The EEOC's chair when the agency first took this position in 1987 was Clarence Thomas, so the theory wasn't new. What was new in 2012 was the specificity — a playbook telling employers what the EEOC would accept. Parts of Texas law forbade exactly the individualized assessment the guidance recommended. Texas sued the EEOC and the Attorney General before anyone could sue it. The government's first answer wasn't that the guidance was right. It was that no one could challenge it, because guidance isn't final agency action. That argument bought the agency almost six years and two trips to the Fifth Circuit.

This is the finality episode, and the fourth in the season's judicial review arc. The last two episodes were about standing and the ballroom case — can you get into the courthouse at all? This episode assumes you've cleared that gate and asks the next one: has the agency actually done enough for you to challenge it? Finality decides whether what the agency has done to you counts as "done." And "done" turns out to be much harder than it sounds.

The core rule is one sentence in APA § 704: "Agency action made reviewable by statute, and final agency action for which there is no other adequate remedy in a court, are subject to judicial review." Unless some other statute makes action reviewable, that's two requirements — final, and no other adequate remedy. For fifty years after the APA passed in 1946, "final" was decided case by case. The modern test comes from Bennett v. Spear (1997), a unanimous Scalia opinion in a water rights fight on the Oregon-California line. The Fish and Wildlife Service had issued a biological opinion telling the Bureau of Reclamation to keep lake levels up to protect two endangered sucker fish species. Ranchers who were about to lose irrigation water sued. The government argued the biological opinion was just advice — the Bureau made the actual decision. Scalia's answer was that agencies very rarely ignore a biological opinion, because departing from one creates serious legal exposure. The government's own brief admitted it. The label said advice. The effect was a command.

Bennett set up two requirements for finality. First, the action must mark the consummation of the agency's decision-making process — not tentative or interlocutory. Second, it must be an action by which rights or obligations have been determined, or from which legal consequences will flow. Both prongs have to be met. The consummation question is often the easy half — a proposed rule isn't consummated, a final rule is — but it gets harder in the middle. If the FDA sends a warning letter saying your product is misbranded, posts the letter on its public website, and distributors start pulling the product off shelves, that pressure is real, but the D.C. Circuit held in Holistic Candlers (2012) that pressure isn't the same thing as legal consequences. Warning letters, the court said, aren't the agency's last word, and the agency isn't bound by them. For informal documents that do represent the agency's settled position, the fight usually turns on the second prong — whether legal consequences flow.

Two Supreme Court cases out of the Clean Water Act define what counts as a legal consequence. Sackett v. EPA (2012) involved a couple who bought a lot near Priest Lake, Idaho, and started putting gravel and dirt down to build a house. The EPA issued a compliance order calling it a wetland, demanding restoration, and barring further work. Penalties ran up to $37,500 a day, and the government's position was that ignoring the compliance order was itself a second violation — doubling the exposure to $75,000 a day. The EPA argued the compliance order wasn't final and the Sacketts could challenge the wetlands determination when the EPA sued them. The Court rejected it unanimously. The compliance order changed the Sacketts' legal position by doubling their potential exposure and making an after-the-fact permit harder to get. Scalia also answered the agency's efficiency argument — that making compliance orders reviewable would mean the EPA uses them less — with language that applies to every finality dispute: "The APA's presumption of judicial review is a repudiation of the principle that efficiency of regulation conquers all." Every agency makes the efficiency argument. If it were a reason to deny review on its own, nothing would be reviewable.

Army Corps of Engineers v. Hawkes (2016) pushed the same analysis into a case with no order at all. Hawkes mines peat in Minnesota and wanted to expand onto land it owned. The Corps issued an approved jurisdictional determination — a formal written document saying the property contained waters of the United States. The determination doesn't order anyone to do anything. It just tells you whether you need a permit. The Corps said that if Hawkes disagreed, it could apply for the permit and challenge a denial, or build without one and defend the enforcement action. Chief Justice Roberts took the first option apart with numbers. One study found the average applicant for the specialized permit at issue spent 788 days and $271,596 completing the process — more than two years and a quarter of a million dollars just to put the question in front of a judge. That isn't an adequate alternative. The second option — building without a permit and risking serious criminal and civil penalties — isn't either. And the jurisdictional determination met both Bennett prongs: it was the consummation of the Corps' process (a formal document, good for five years, binding on EPA too under an interagency agreement), and legal consequences flowed because a negative determination worked as a five-year safe harbor from enforcement, while a positive one took the safe harbor away. The Corps had built a definitive, reliable document because definitive and reliable was useful — to landowners and to the Corps. That definiteness was what made it reviewable.

The Texas case is harder than either of those because the EEOC hadn't ordered Texas to do anything or made a determination about Texas. It had just published a guidance document for employers generally. Guidance exists precisely because it lets agencies tell the public how they're reading the law without going through notice and comment. There's no proposed rule, no obligation to answer significant comments, no final rule. The 2019 executive order on guidance (issued by the first Trump administration) acknowledged the regulated party's complaint directly: "Even when accompanied by a disclaimer that it is non-binding, a guidance document issued by an agency may carry the implicit threat of enforcement action if the regulated public does not comply." But the order didn't make coercive guidance judicially reviewable. It told agencies to make guidance public, label it non-binding, and generally run significant guidance through comment. Finality still gets decided one lawsuit at a time.

When the Texas case reached the Fifth Circuit the second time, the court didn't focus on how aggressively the EEOC had used the guidance. It looked at what the guidance told the EEOC's own staff to do. The guidance bound staff to an analytical method in Title VII investigations and directed decisions about which employers to refer for enforcement. It also described two ways employers would consistently meet the business necessity defense — effectively safe harbors, the same feature that made the Hawkes jurisdictional determination final. The court's conclusion followed the logic to its endpoint: a document that binds agency staff and creates safe harbors isn't guidance. It's a substantive rule. And Title VII gives the EEOC power to write only procedural rules — meaning the EEOC couldn't have issued the guidance through notice and comment either. The district court had barred enforcement until the EEOC completed notice and comment. The Fifth Circuit struck that condition, because no process could fix the underlying problem, and ordered that the EEOC and the Attorney General couldn't treat the guidance as binding in any respect. Texas won outright. The guidance itself is still posted on the EEOC's website, marked as remaining in effect until rescinded or superseded, for every other employer in the country covered by Title VII.

The pattern across all three cases is the same: the question moved from what the agency called the thing to what the thing did to the person holding it. Compliance orders, jurisdictional determinations, and guidance documents all look different, and the agencies issuing them described each as non-binding or preliminary. In each case, the court looked at the effect. If the document is the agency's settled position and legal consequences flow from it — mandatory language that binds staff, safe harbors for compliant conduct, formal determinations that other agencies honor — the label the agency chose doesn't control. The agency controls the drafting. It controls whether the document is a rule, a letter, a Dear Colleague memo, or an FAQ page. It controls whether to treat the document as binding in practice. The 2011 Education Department Dear Colleague letter on campus sexual assault — telling schools to use preponderance of the evidence for Title IX discipline — reshaped procedures at colleges and universities nationwide for six years before the Department withdrew it in September 2017. No court ever reached the question of whether it was a substantive rule. The agency decided what the document looked like before anyone ever applied Bennett to it.

Listen Now

Apple Podcasts | Listen on our site

What We Cover

Full Rough Transcript
Opening

Gwen: Welcome to Administrative Remedies, because you can't fix what you don't understand. Brought to you in part by the University of Tulsa College of Law's Mabee Legal Information Center. I'm Gwendolyn Savitz, an associate professor here at TU.

Marc: And I'm Marc Roark, a professor at the College of Law. We'll be breaking down complex doctrines with real-life analogies and examples, to demystify the world of administrative law for everyone trying to understand how government actually works.

Gwen: Agencies are the main way the federal government gets things done. It's not through Congress, for reasons we'll be addressing over the course of the series.

Marc, the last couple of weeks we've been talking about the threshold question of standing. Are you hurt enough to be in court at all? This week, we assume that you are, but there's still another gate. Has the agency actually done enough for you to challenge it?

Marc: Meaning the agency can take a position that affects you, and still say you sued too early.

Gwen: Yes, exactly. Finality asks whether the agency is done, and what "done" means turns out to be much harder than it sounds.

As an example: thirteen years ago, the Texas Department of Public Safety didn't hire convicted felons. It didn't hire them as troopers. It didn't hire them for any job in the department. This was a total ban. In November 2013, a man who'd been turned down for a Department of Public Safety job filed a charge with the Equal Employment Opportunity Commission, saying that the bar violated Title VII.

Marc: On what theory was he bringing this claim? A felony bar isn't a racial classification.

Gwen: He was claiming disparate impact. A year and a half earlier, the EEOC had issued enforcement guidance on arrest and conviction records. This wasn't a regulation. They hadn't gone through notice and comment to do it. This was just guidance. And it said that an automatic, across-the-board exclusion of everyone with a record, from any sort of job, disproportionately screens people out by race and national origin, and the employer probably can't justify it.

Marc: Like in Griggs v. Duke Power. A power company had a diploma requirement and adopted an aptitude test the day Title VII took effect. The Court struck them both down, even if there was no bad intent. A neutral practice that screens out Black applicants violates Title VII unless the employer shows that it's job-related. Seems to fit the disparate impact test, but applied instead to background checks?

Gwen: Yeah. This idea isn't a new one at the agency. The EEOC's first policy statement on conviction records came out in 1987. And at that point, the chair was Clarence Thomas. So the 2012 document was just consolidating and updating those.

Marc: Which makes it hard to call the theory a partisan invention. What changed in 2012 was the document, not the idea.

Gwen: Right. What changed was how specific it got. The 2012 version was a playbook. It told employers what the EEOC would accept. It was a targeted screen that looks at the offense, how long ago it happened, and the job, and then an individualized look at anybody the screen excluded. The guidance said it wasn't required in every case, but it was plainly the safest route. And parts of Texas law forbid exactly that kind of individualized assessment. Texas's own complaint put the choice this way: start considering felons for trooper, jailer, and teacher jobs, or ignore the EEOC and risk an enforcement action.

Marc: And as they say, sometimes the best defense is a good offense. It chose to sue before anyone could sue it.

Gwen: Yes. So it sued the EEOC and the Attorney General. And the government's first answer wasn't that the guidance was right. It was that nobody could challenge it. Guidance isn't final agency action. The EEOC went even further. It argued that the document couldn't carry legal consequences — that's part of the analysis about whether something is final — because the EEOC has no authority to issue binding interpretations of Title VII in the first place.

Marc: So basically the agency was saying, look, we don't even have the power to do the thing that you said that we did. Which is kind of an unusual thing to lead with.

Gwen: It definitely isn't how they generally start. But this did buy the government almost six years. This was two trips to the Fifth Circuit.

Segment 1 — The Bennett test

Gwen: The core of it is one sentence in the APA.

Marc: Agency action made reviewable by statute, and final agency action for which there is no other adequate remedy in a court, are subject to judicial review.

Gwen: Right. Unless some other statute makes action reviewable, that's two requirements. It needs to be final, and there needs to be no other adequate remedy. The next sentence of § 704 matters too. Preliminary or intermediate steps aren't reviewable on their own, but they can be reviewed along with the final action.

The APA was passed in 1946. For half a century, "final" got decided case by case. Abbott Labs in 1967 set the mood. Review is presumed. Finality is read pragmatically. But it wasn't a test. The test came in 1997, over water rights for the Klamath Irrigation Project. This is on the Oregon-California line. The Fish and Wildlife Service had issued a biological opinion telling the Bureau of Reclamation that they needed to keep lake levels up to protect two endangered species of sucker fish. Ranchers who were about to lose irrigation water sued.

Marc: And the government said the biological opinion was only advice. The Bureau of Reclamation made the actual decision.

Gwen: Yes, that was the argument — that this was just an advisory opinion by the agency, and that the Bureau of Reclamation was free to ignore it. Scalia's answer was that agencies very rarely ignore one. The government's own brief admitted this, because departing from a biological opinion creates serious legal exposure. So the label said advice, but the effect was actually a command.

Marc: So, Scalia opening the courthouse to ranchers suing under the Endangered Species Act. Not the lineup anyone would have predicted.

Gwen: And it was unanimous, too. This is Bennett v. Spear. Most people meet Bennett v. Spear in a standing course. This is the zone of interests holding that we talked about before, but it matters for finality too. Here's the important part.

Marc: As a general matter, two conditions must be satisfied for agency action to be final. First, the action must mark the consummation of the agency's decision-making process. It must not be of a merely tentative or interlocutory nature. And second, the action must be one by which rights or obligations have been determined, or from which legal consequences will flow.

So, "as a general matter." Scalia here is hedging a little bit.

Gwen: The hedge leaves room in both directions, and courts have used it both ways. But in practice, these two prongs are the test — whether it's the consummation of the agency's decision-making process, and whether legal consequences flow. For something to be final, you need both of them.

Marc: And every litigator already lives with some version of this story, right? The final judgment rule. You can't appeal until the trial court is done, unless you fit within a short list of exceptions for interlocutory orders. And this seems like it's following the same instinct.

Gwen: It is the same instinct, but they play out differently. The final judgment rule protects a trial judge's process. Finality protects an agency, and there are reasons for it. An agency in the middle of a decision could change its mind. The record isn't built yet. But the difference here is that a trial court can't decide to just never enter a judgment. And agencies can potentially decide to never make something final.

Marc: And so consummation here sounds like the easy half. A proposed rule isn't consummated. A final rule is. Or an appeal taken as far as the agency lets you take it is.

Gwen: At the end, yes. But this gets harder in the middle. If FDA inspects a manufacturer and sends a warning letter saying your product is misbranded, fix it within fifteen days or face enforcement action — this letter goes up on the FDA's public website. Distributors read it. Retailers pull the product.

Marc: And that company isn't going to litigate whether it had to comply. It's just going to comply.

Gwen: Yes, almost always. And when companies have litigated, they mostly lost. One of the leading cases here is Holistic Candlers, from the D.C. Circuit in 2012. This was saying that warning letters are informal, FDA isn't bound by them, FDA can still decide not to bring enforcement action, so they aren't final. But the pressure is real. However, the court said pressure isn't the same thing as legal consequences.

Marc: OK. A warning letter looks more like a prosecutor's letter than a judgment does. Nobody gets to sue a prosecutor for sending a warning letter.

Gwen: Yes. That's why warning letters are where this label-versus-effect argument usually loses. The D.C. Circuit there held they failed both prongs. They weren't the agency's last word, and there were no legal consequences.

Marc: So then, legal consequences will flow. So having legal consequences flow is absolutely critical to finality.

Gwen: Yes. For informal documents that do represent the agency's settled position, the fight usually turns on the second prong. An interpretive letter, an FAQ page, a guidance document that the agency calls non-binding but then uses in every investigation.

Segment 2 — The compliance order cases

Gwen: The Supreme Court cases define the second prong, and both come out of the Clean Water Act. So we have Chantell and Michael Sackett, who bought a lot near Priest Lake, Idaho, and they wanted to build a house on it. In 2007, they started putting gravel and dirt in it. The EPA issued a compliance order saying, that's a wetland. You've discharged fill into the wetland without a permit. You need to restore the property. Give us access to your land and your records. And they asked for a hearing, and EPA said no.

Marc: And with penalties that were attached to it.

Gwen: Right. This Act allowed up to $37,500 a day per violation by the time the case had reached the Supreme Court. That was because the amounts kept ratcheting up with time. The government's position was that ignoring the compliance order was a second violation on top of the first. So you could double that. They were facing $75,000 a day.

Marc: $75,000 a day usually ends most fights before they start.

Gwen: EPA's position here is the one you'd predict. The compliance order isn't final agency action. If you don't think it's a wetland, wait for us to sue you — with the meter running the whole time.

Marc: And nobody's going to make that choice.

Gwen: Right. Nobody would call it a choice, and the Court didn't. This wasn't a partisan position. The order went out in 2007 under the Bush EPA, and the Obama Justice Department was defending it at the Supreme Court. The Court rejected it unanimously in 2012, but the reasoning was narrower than the headline. The penalties existed before the order. But what the order did was change the Sacketts' legal position. Under the government's own reading, it doubled their potential exposure going forward, and it made an after-the-fact permit from the Army Corps of Engineers harder to get.

But the EPA also had a policy argument the agencies still make — that compliance orders get problems fixed fast, through voluntary compliance. Make them reviewable, and the EPA will use them less.

Marc: Which is probably true.

Gwen: And Scalia conceded it might be.

Marc: That may be true, but it will be true for all agency actions subjected to judicial review. The APA's presumption of judicial review is a repudiation of the principle that efficiency of regulation conquers all. And there is no reason to think that the Clean Water Act was uniquely designed to enable the strong-arming of regulated parties into voluntary compliance without the opportunity for judicial review — even judicial review of the question whether the regulated party is within the EPA's jurisdiction.

So this idea that efficiency of regulation conquers all — every agency makes some version of that argument, about every kind of review, it seems.

Gwen: Well, it's pretty much true every time. Review will always slow an agency down. If that were a reason to deny review on its own, nothing would be reviewable.

Army Corps of Engineers v. Hawkes pushed the same idea into a case with no order at all. Hawkes mines peat in Minnesota, and they wanted to expand onto land that they owned. The Corps issued an approved jurisdictional determination saying that the property contained waters of the United States. That jurisdictional determination doesn't order anyone to do anything. It tells you whether you need a permit. And the Corps said, if you disagree, apply for the permit and challenge a denial, or go ahead without one and then defend the enforcement action. The Chief Justice took this apart with numbers. He said:

Marc: It is often difficult to determine whether a particular piece of property contains waters of the United States, but there are important consequences if it does. The costs of obtaining such a permit are significant. For a specialized individual permit of the sort at issue in this case, for example, one study found the average applicant spends 788 days and $271,596 in completing the process, without counting costs of mitigation or design changes.

So more than two years and a quarter of a million dollars just to put the question in front of a judge.

Gwen: That led the Court to say that that route really isn't an adequate alternative. And the enforcement route fared no better. A landowner doesn't have to expose themselves to serious criminal and civil penalties just to get a judge to look at the question.

So on the first prong, there was no fight at all. The Corps conceded that. This jurisdictional determination was the consummation of its process. It's a formal written product. It's good for five years. And under an agreement between the agencies, EPA also treats it as binding. The only Bennett prong in dispute was the second — whether legal consequences flow. And the Court found that there were legal consequences, because a negative jurisdictional determination works as a five-year safe harbor from enforcement, and a positive jurisdictional determination takes that safe harbor away.

Marc: The Corps built a definitive, reliable document, because definitive and reliable is useful. Useful to landowners and useful to the Corps. But that definiteness is what made it reviewable. It couldn't have it both ways.

Gwen: Right. This was eight to nothing. So across both cases, the question moved from what the agency called the thing to what the thing did to the person holding it.

Segment 3 — Guidance documents

Gwen: Sackett had an order. Hawkes had this formal determination. But let's finally circle back to Texas. That Texas case at the beginning is harder. The EEOC hadn't ordered Texas to do anything. It hadn't made a determination about Texas. It had just published a guidance document for employers generally.

Marc: Which isn't supposed to be binding. So why would guidance ever be a final agency action?

Gwen: And that's the problem. Guidance lets the agency tell the public how it's reading the law and how it intends to enforce it, without going through the notice and comment process. There's no proposed rule. There's no obligation to take comments and answer the significant ones. There's no final rule at the end. That's part of what makes guidance useful. It's faster and more flexible.

Marc: But if I'm the regulated party, "non-binding" may not mean I can safely ignore it. The people who wrote the guidance are also the people who decide whether to investigate me.

Gwen: Yes. And in 2019, the first Trump administration said that explicitly in an executive order about guidance. So they said:

Marc: Even when accompanied by a disclaimer that it is non-binding, a guidance document issued by an agency may carry the implicit threat of enforcement action if the regulated public does not comply.

That's basically the regulated party's argument. You call it guidance. I experience it as a command.

Gwen: Yeah, but look at what the order did with that problem. It didn't say that coercive guidance is now judicially reviewable. It told agencies to make guidance public, to identify it as non-binding, and generally put significant guidance through a public comment process. And the order expressly said it created no rights that anyone could enforce in court.

Marc: So the executive branch was willing to say, yeah, supposedly non-binding guidance can pressure people into compliance — without giving those people a new way to challenge that guidance.

Gwen: Yes. That puts us back at Bennett, which is why finality still gets decided one lawsuit at a time.

OK, back to Texas. So the district court dismissed. In 2016, a divided Fifth Circuit panel reversed. It said the guidance was final and Texas had standing. And then the panel withdrew its own opinion.

Marc: OK, I have to ask why.

Gwen: That Hawkes case that we were talking about. The panel sent the case back down so that the district court could apply Hawkes first. When the case came back in 2019, the Fifth Circuit didn't look mainly at how aggressively the EEOC had used the guidance. It looked at what the guidance told the EEOC's own people to do.

Marc: So why was it that Texas brought this in the first place, and not some trucking company with a no-felons policy?

Gwen: The Fifth Circuit said that states have a sovereign interest in making and enforcing their own laws, and that this guidance put direct pressure on Texas to change its hiring laws and policies. One Texas agency had already had to respond to an EEOC charge as well.

Marc: So actually being a state here matters. Kind of like in Massachusetts v. EPA's special solicitude for states.

Gwen: Yes. A sovereign's interest in its own legal code counts, although the regulatory burden alone could have been enough. Once Texas was in the door, the finality question's waiting. Texas wasn't just predicting that somebody might enforce it someday. The guidance applied to state employers, and it condemned the kind of policy Texas was actually using. And it put Texas's own law on one side of the conflict. The first time up, the court took standing first. But on that second round, it started with whether it was actually final agency action. In that case, the Fifth Circuit said:

Marc: The guidance indicates that it binds EEOC staff to an analytical method in conducting Title VII investigations, and directs their decisions about which employers to refer for enforcement actions. Finality, however, cannot vary depending on who sued the agency. It depends on the rule itself.

And that second sentence of the reading is aimed at the argument that the EEOC can't sue a state.

Gwen: Yes. EEOC can investigate a state employer, but it can't bring the federal government's enforcement suit. That has to come from the Attorney General. So the EEOC's position was, whatever this guidance does to private employers, it has no legal consequences for Texas, because we can't enforce anything against Texas. The court's answer was that the same document can't be final for a private company and non-final for a state.

Marc: And the safe harbors?

Gwen: Now we're talking about Hawkes again. Yeah, the guidance described two ways that employers would consistently meet the business necessity defense. The court treated these like safe harbors. That was the same feature that made that jurisdictional determination final. And then it took the next step. A document that binds agency staff and that creates safe harbors isn't guidance. It is a substantive rule. And Title VII gives the EEOC power to only write procedural rules.

Marc: So then the EEOC couldn't have issued it through notice and comment either.

Gwen: The court rewrote the injunction. The district court had barred enforcement against Texas until the EEOC went through notice and comment. The Fifth Circuit struck that, because there was no process that could fix it, and ordered that the EEOC and the Attorney General could not treat the guidance as binding in any respect.

Marc: And then Texas wins outright.

Gwen: Yeah, Texas wins. So for Texas, this injunction is protecting it. The guidance itself, though, is still posted on the EEOC's website. It's still there right now. And it's marked as remaining in effect until rescinded or superseded — so for every other employer in the country covered by Title VII. And the man whose charge started all of this isn't even named in the opinion.

Marc: So by 2019, the case had become almost entirely about Texas and the guidance.

Gwen: Partly. In January 2018, the Associate Attorney General issued a memo that was supposed to tell Justice Department lawyers they couldn't use enforcement to turn agency guidance into binding rules. And the government also told the court it no longer shared the EEOC's approach in several respects. Its argument was that Texas had nothing left to fear from the Attorney General. The court then raised mootness on its own, because nobody else had. That's the idea of voluntary cessation. But here, the defendant is free to return to his old ways. So the burden was on the government to make it absolutely clear it couldn't recur. And a general memo didn't do that.

Marc: Okay. And I think there's something that's really telling here, right? That the court reached for it even though the government had not. And that a policy change isn't a promise — especially one the next administration can just undo.

Segment 4 — Who controls the clock

Gwen: In April 2011, the Education Department's Office for Civil Rights sent a Dear Colleague letter about campus sexual assault. And among other things, it told schools to use the preponderance of the evidence standard in Title IX discipline. It never went out for comment, and schools across the country rewrote their procedures to match it, because federal funding sits behind Title IX compliance. And then a student and a university sued in 2016, arguing it was a substantive rule issued without notice and comment. The Department withdrew the letter in September 2017, and the parties stipulated to dismissal five months later. No court ever reached the merits.

Marc: So you had six years of nationwide effect, and no court ever deciding whether it was a rule.

Gwen: Yes, that's the structure here. The agency controls the drafting. It controls whether this document is issued as a rule, or a letter, or frequently asked questions, or a framework — and whether to treat it as binding in practice. And each one of those choices feeds the finality analysis. The court decides what final means, but the agency decides what the document looks like before anyone's applying the test.

Marc: I think you're overstating it, though. Guidance that isn't binding can't be enforced as if it were the law. If the EEOC wants to win against an employer, it has to prove a Title VII violation to a judge, and the judge decides what the statute means. The guidance gets only whatever persuasive weight it earns. And an employer who ignores the guidance gives up nothing it had a legal right to do. A lot of guidance exists because regulated companies ask for it.

Gwen: Yes, industry really likes guidance. Most companies, if they're told what to do that is a safe harbor, they will happily do that. But the fact that it isn't binding on the employer doesn't answer whether it's final agency action. Those are different questions. A guidance document can be the agency's finished position, even though the agency won't point to the document itself as law. And if it tells the agency's own people what position to take, or what counts as compliance, or when to refer a case for enforcement, that can produce the legal consequences that Bennett is looking for.

Marc: So calling it guidance doesn't resolve either prong.

Gwen: Right. You can still ask whether the agency's done, and whether legal consequences flow from what it did. In Texas, the Fifth Circuit said the answer is yes, even though the document said it wasn't binding.

Marc: Though the EEOC is a special case, because the court held it has no power to issue substantive rules under Title VII at all.

Gwen: Yeah, it's true. An agency with substantive rulemaking authority could potentially take the same document and issue it after notice and comment and have it stand.

Marc: But the more you make these reviewable, the less likely the agency is to issue them to begin with.

Gwen: Yes. And that's a tradeoff. It's one reason not to make all guidance reviewable. One strong indicator in the Fifth Circuit is whether the document binds the agency's own staff. If there's mandatory language, or withdrawal of staff discretion, or it creates safe harbors, those all point toward legal consequences. A document that genuinely leaves the agency free to decide cases differently on a case-by-case basis looks much more like advice. Agencies can operate on either side of the line. What they can't do is write the first kind and label it the second kind.

Marc: Okay. It seems to cut in both political directions, which is what we've seen this past year.

Gwen: Yeah. The same informality that lets the EEOC issue that 2012 guidance without comment lets an agency take guidance down without comment. So on June 29th, the Commission voted two to one to rescind its 1979 affirmative action guidelines and the compliance manual sections that went with them. Its regulatory agenda, which it published a week later, put more rescissions at the final rule stage, with no proposed rules ahead of them. The regulatory agenda is like, this is what we're planning to do over the next year. So they're saying, and we're doing this to more. This included the interpretive portions of the Uniform Guidelines on Employee Selection Procedures, which are the validation framework — the part that tells an employer how to show a screen is job-related. Final action on this is projected for November. It'll be effective in January. And the recordkeeping requirements, by contrast, are still at the proposed-rule stage. But none of this is issued yet. This agenda is just a statement of their intent.

Marc: And the premise of the 2012 guidance — disparate impact under Title VII?

Gwen: So on June 9th, the Justice Department's Office of Legal Counsel concluded that these disparate impact guidelines are unconstitutional, and it supplied a much narrower reading of the statute in their place. The EEOC's own chair had asked for an opinion. This binds no court, it binds no employer, but it does change how the executive branch reads Title VII.

Marc: So suppose the EEOC rescinds the 2012 criminal record guidance outright. Could anyone challenge that?

Gwen: Potentially, but you still have to identify the legal consequence of the rescission, and somebody injured by it. Remember, the Fifth Circuit didn't treat the guidance as final because it was codified. It wasn't. It found legal consequences because the guidance bound EEOC staff and created safe harbors for employers.

Marc: So taking the guidance away could matter if taking it away changes one of those consequences.

Gwen: Yes. And if you compare that with the affirmative action guidelines that the EEOC just rescinded, those have even more concrete consequences. Section 713 of Title VII can protect an employer from liability or punishment when it acts in good-faith reliance on a written EEOC interpretation. And the 1979 guidelines have been held to expressly qualify for that kind of protection.

Marc: So rescinding those takes away something employers could actually invoke as a statutory defense.

Gwen: Yes. That's a pretty easy legal-consequences argument. A lot easier than simply saying the agency took a guidance document off its website.

Closing

Marc: I have to say, finality doesn't feel very final.

Gwen: Nothing is final. This is administrative law. Next week, we have three more timing doctrines. We're going to do ripeness, exhaustion, and mootness all in one day. This is three ways of asking when. And we will, again, be talking about the Sacketts.

So that does it for today's episode of Administrative Remedies. Thank you for joining us today. Please, if you enjoy this podcast, enjoy this episode, give us a like on Spotify, iTunes, or whatever platform you're listening on. And be sure to tune in next time, where we'll continue to dive into the contours of administrative law. Because remember, you can't fix what you don't understand.

Related Guides

Administrative Remedies is produced by Professor Gwendolyn Savitz and Professor Marc Roark at the University of Tulsa College of Law.