The University of Chicago Law School Faculty Blog

Notice and Comment, Behavioral Economics, and United States v. Texas

Assistant Professor Daniel Hemel on the Supreme Court’s certiorari grant in United States v. Texas:

In many respects, the Supreme Court’s cert grant earlier this week in United States v. Texas was utterly unsurprising. The Fifth Circuit Court of Appeals affirmed a nationwide injunction blocking a Department of Homeland Security policy that would have allowed approximately 4 million parents of U.S. citizens and lawful permanent residents to seek “deferred action” and temporary authorization to work in the United States. The Supreme Court’s own Rule 10 says that in deciding whether to grant cert, the Court considers whether a lower court has ruled on “an important question of federal law that has not been, but should be, settled by this Court.” United States v. Texas meets that standard by any measure.

In one respect, though, the cert-stage proceedings in United States v. Texas pose a puzzle. (And no, I’m not referring the Court’s decision to add a fourth question regarding the Take Care Clause of Article II; as Marty Lederman and others have explained, the Court had reason to add that question even if it has no intention of going Texas’s way on the constitutional issue.) I’m referring instead to the third question: “Whether the Guidance was subject to the APA’s notice-and-comment procedures.” Of course, there is no mystery as to why the Supreme Court granted cert on that question: the notice-and-comment issue is one of two independent bases for the Fifth Circuit’s ruling, so the Supreme Court would need to reach that issue if it wants to overturn the Fifth Circuit’s decision on the merits. The puzzle is why the Obama administration is still litigating this issue despite the fact that DHS could moot the notice-and-comment question if it wanted to.

A bit of background: In November 2014, Homeland Security Secretary Jeh Johnson sent a memo to top DHS officials outlining new policies for exercising prosecutorial discretion. Most significantly, the Johnson memo directs DHS officials to establish a process that will allow parents of U.S. citizens and permanent residents to apply for “deferred action.” The memo states that applications will be evaluated on a “case-by-case basis,” and that deferred action will be limited to parents who—among other criteria—have not been convicted of a felony or significant misdemeanor. DHS’s approval of an application for deferred action means that the Department has decided to forbear from removing the applicant for three years. Successful applicants also may seek work authorization for a three-year period. The program is known as “Deferred Action for Parents of Americans and Lawful Permanent Residents” (DAPA), and DHS planned to begin accepting applications in May 2015.

In December 2014, Texas and 25 other states sued in federal district court seeking to enjoin implementation of DAPA. Texas’s substantive arguments are (1) that DHS lacks statutory authorization for DAPA, and (2) that DAPA violates the Take Care Clause of Article II. Texas’s procedural argument is that DAPA is subject to the Administrative Procedure Act’s notice-and-comment requirements, and that DHS failed to comply with those requirements entirely. In February 2015, a federal district court held that Texas was likely to prevail on its notice-and-comment argument, and the court entered a nationwide preliminary injunction against implementing DAPA. In November, a Fifth Circuit panel voted to 2-1 to affirm the injunction—on the procedural ground that DHS failed to go through notice and comment and on the substantive ground that DHS lacks statutory authority for the DAPA program.

I am not the first to observe that DHS could moot the notice-and-comment question by publishing the Johnson memo in the Federal Register and then finalizing it 30 or more days later. That’s all that the APA’s notice-and-comment provision requires. To be sure, the notice-and-comment process is not costless, but it is also not prohibitively expensive. The White House’s regulations.gov website shows 221 proposed rules published in the last month—ranging from a proposal to end the endangered-species status of the scarlet-chested parakeet to a proposal to clarify the number of charity stamps that the Postal Service can sell at any one time. The costs of notice-and-comment rulemaking are likely higher for higher-profile rules (for one thing, DHS is likely to receive more comments on its deferred action policy than on its ammonium nitrate security regulations). On the other hand, the deferred action policy is a top administration priority, and it seems like it would be well worth DHS’s time to wade through a pile of public comments (most of which are electronic anyway) if going through the notice-and-comment process increases the probability that the deferred action policy will pass judicial muster.

So why doesn’t DHS just go ahead with the notice-and-comment process? (And why didn’t DHS do this last year, when the district court first ruled on the issue?) Here are a handful of hypotheses:

-- (1) The administration may believe that taking the notice-and-comment issue off the table won’t measurably increase its chances of victory. The issues in United States v. Texas fall into three categories: (i) threshold questions of standing and reviewability; (ii) substantive questions as to whether the administration has the statutory and constitutional authority to implement DAPA; and (iii) the procedural question (notice and comment). If the administration wins on a threshold question, then the whole case goes away and DHS can implement DAPA. If the administration loses on the threshold questions, then it must win on both the substantive and procedural questions in order to have the Fifth Circuit’s decision reversed.

Let p_t be the administration’s probability of success on the threshold issue; let p_s be the administration’s probability of success on the substantive questions; and let p_p be the administration’s probability of success on the procedural point. The administration’s overall probability of success (P) is equal to:

p_t + (1 - p_t)(p_s x p_p)

The administration may think that p_t is so close to 1 that the rest of the terms in the equation don’t matter. That is, the administration might think its arguments on standing and reviewability are airtight, and that the Court won’t even reach the substantive and procedural questions. Or, the administration may think that p_p is close to 1: its arguments on notice-and-comment are so strong that if the Court rules for Texas on the threshold issues, all that really matters are the substantive questions. Less likely, the administration might think that p_s is close to zero—if it can’t win on standing, it won’t win on the substance—in which case the procedural question also won’t be outcome-determinative. A fourth possibility is that the administration thinks that a five-Justice majority will only rule for Texas on standing if it’s dead-set on striking down DAPA—in which case an ideologically motivated majority will side with Texas on the substance as well. Put differently, the administration may think that p_t and p_s are closely correlated—in which case the notice-and-comment issue also wouldn’t matter.

-- (2) The administration may think that proceeding with notice-and-comment will reduce its chances of securing Supreme Court review. The administration may want to resolve the DAPA issue ASAP so that it can begin to implement the policy before President Obama’s second term is out. Maybe the worry is that if DHS proceeds with notice-and-comment, the Supreme Court will remand to the Fifth Circuit to determine whether DHS had rendered the procedural issue moot. To be sure, there would still be a live case and controversy on the substantive questions (insofar as there is a “case or controversy” to begin with). But a remand to the Fifth Circuit would delay matters for months, pushing oral argument over to the Supreme Court’s October 2016 Term.

It seems pretty clear, though, that the procedural question would be moot once DHS completed the 30-day notice-and-comment process. The Court resolved an analogous question in the tax context four decades ago, and there is no reason why the Court would need to remand to the Fifth Circuit to figure it out. If there were any lingering doubt on the matter, the Supreme Court could hear arguments on the threshold and substantive questions and then remand to the Fifth Circuit if necessary. (There would be no remand if the Court rules against Texas on standing/reviewability or if the Court rules for Texas on the substance.) Which is to say: If the Supreme Court wants to hear United States v. Texas this spring, DHS’s initiation of notice-and-comment proceedings won’t be an obstacle. And if the Supreme Court wanted to delay consideration until the fall, it could have relisted the case for a few more weeks before granting cert.

-- (3) The administration may think that going through with notice and comment on DAPA would establish a bad precedent. Notice and comment is not required under § 553 for “general statements of policy.” The administration maintains that the Johnson memo qualifies as such a statement. Note, though, that executive agencies routinely publish planned actions in the Federal Register and solicit comments from the public even while maintaining that the APA’s notice-and-comment requirement does not apply. (For a few recent examples, see here, here, and here.) DHS could easily take that route in this case, going through with notice-and-comment while sticking to its view that notice-and-comment is unnecessary.

-- (4) The administration may think that Supreme Court review on the notice-and-comment question will establish a favorable precedent. That is, maybe the administration actually wants the Supreme Court to weigh in on the notice-and-comment question. In recent years, the Fifth and D.C. Circuits have whittled down the notice-and-comment exemption for “general statements of policy.” Those courts say that a guidance document only falls within the exemption if it “genuinely leaves the agency and its decisionmakers free to exercise discretion.” Even when the language of the guidance appears to allow discretion, the Fifth and D.C. Circuits will look to the agency’s post-guidance conduct to decide whether the guidance is binding “as a practical matter.” Put differently, the Fifth and D.C. Circuits decide whether notice and comment was required at time t by looking at the way the agency acted at time t + 1. The Fifth Circuit decision in this case extends that approach even further. The Fifth Circuit held that the Johnson memo was a rule subject to the notice-and-comment requirement rather than a “general statement of policy” because there is “a substantial likelihood that DAPA would not genuinely leave the agency and its employees free to exercise discretion.” In other words, the Fifth Circuit held that notice and comment was required at time t based on the court’s prediction as to how DHS would act at time t + 1.

The Fifth Circuit’s approach has no basis in statutory text or in Supreme Court case law. Just last Term, in Perez v. Mortgage Bankers Association, the Supreme Court reversed the D.C. Circuit for conjuring up a “judge-made procedural right” to notice and comment that had no root in the APA. The administration may be looking for a similar ruling here. But if so, United States v. Texas would seem like a strange vehicle. One would think that the administration would choose to challenge the Fifth and D.C. Circuit’s approach in a case with low ideological valence—like Mortgage Bankers, in which the underlying dispute was about overtime pay for loan officers. And if one were trying to choose a case with low ideological valence, United States v. Texas would be just about the last pick.

-- (5) A final possibility—perhaps not the most likely, but probably the most interesting—is that the administration’s litigation strategy is informed by insights from behavioral economics. (President Obama issued an executive order in September 2015 calling on agencies to incorporate behavioral science insights into their programs and policies; maybe DHS and the Solicitor General are taking a cue from the White House.) One of the more robust results in the behavioral economics literature is the “compromise effect.” As stated by Itamar Simonson, the “compromise effect” is the finding that “an alternative’s choice probability tends to increase when it becomes a compromise choice in the set.” In one well-known study, Simonson and co-author Amos Tversky offered consumers a $170 camera and a $240 camera; 50% of subjects chose the $170 camera and 50% chose the $240 option. But when Simonson and Tversky added a third, more expensive alternative (a $470 camera), the share that chose the $240 camera jumped from 50% to 57%, and the share that chose the cheaper camera fell from 50% to 22%. (The remaining 21% opted for the high-end option.) That is, the $240 camera became more attractive when it was presented as the middle of three options, and the $170 camera became less popular when it was the cheapest of the three.

Retail stores make use of the compromise effect all the time: Starbucks knows that offering a “venti” size in addition to a “tall” and a “grande” makes the “grande” look more attractive because it appears to be a compromise. Mark Kelman and coauthors have also documented a “compromise effect” in legal decisionmaking. Experimental subjects role-playing as jurors gravitate toward whichever punishment is framed as the middle option rather than one of two extremes.

What does any of this have to do with United States v. Texas? If the Supreme Court reaches the merits questions in the case, then it (1) could reverse the Fifth Circuit on the substantive and procedural questions; (2) could reverse the Fifth Circuit on the substance but affirm on the procedural question; or (3) could affirm the Fifth Circuit on both the substantive and procedural questions. (There technically is a fourth option—affirm on substance and reverse on procedure—though if DAPA fails on substantive grounds, it is difficult to see why the Court would bother reaching the notice-and-comment issue.) So framed, Option 2 is the moderate option bookended by two extremes. The behavioral economics literature on the compromise effect suggests that this framing makes it more likely that the Court will choose Option 2. By mooting the notice-and-comment issue, however, DHS would leave only Option 1 and Option 3 on the table if the Court reaches the merits.

Of course, the administration would prefer that the Court choose Option 1. (Actually, it would prefer that the Court reverse on standing/reviewability and thus never reach the merits.) But most of all, the administration wants the Court not to choose Option 3. Option 3 means the death of DAPA. Option 2 means that the administration can implement DAPA after it goes through a 30-day notice-and-comment process. At least in relative terms, Option 2 would be a resounding victory for the administration. And if the compromise effect holds here, keeping Option 2 in the mix makes Option 3 less probable.

Take this last hypothesis with several grains of salt. It’s worse than predicting what’s going on in the mind of Anthony Kennedy; it’s predicting what’s going on in the minds of Donald Verrilli and Jeh Johnson as they predict what’s going on in the mind of Anthony Kennedy. The bottom line, though, is that while the administration’s strategy on the notice-and-comment issue initially seemed befuddling, it might turn out in the end to be brilliant.

Friedrichs, Free-Riding, and Life After the Agency Shop

Assistant Professor Daniel Hemel and UC Berkeley Jurisprudence & Social Policy Ph.D. candidate David Louk on next Monday’s Supreme Court oral argument in Friedrichs v. California Teachers Association:

The Supreme Court will soon decide whether public-sector employers can maintain “agency shops” in the roughly two dozen states that now allow these arrangements. In an “agency shop,” an employee who chooses not to join her local union must nevertheless pay the union an “agency fee” to cover her pro rata share of its collective bargaining expenses. Agency shop supporters justify these arrangements as a means of preventing non-union members from free-riding off the union’s bargaining efforts. Opponents—including the plaintiffs in Friedrichs—say that agency shop arrangements violate the First Amendment rights of non-union members who are forced to pay a fee to an organization whose views they reject.

In the 1977 case Abood v. Detroit Board of Education, the Supreme Court held that agency shop arrangements in public-sector workplaces do not violate the First Amendment as long as nonmembers’ fees are applied toward collective bargaining costs rather than unrelated “ideological activities.” More recent decisions have suggested that Abood may be ripe for reconsideration, and the question presented in Friedrichs explicitly asks whether Abood should be overruled. We take no position on that constitutional question. We do think, however, that the practical implications of overruling Abood could be more muted than either side recognizes. Even if Abood is overruled, public-sector employers in sympathetic states still will be able to ensure that unions are reimbursed for their collective bargaining costs (including the cost of representing nonmembers). They just might have to take a different (and more straightforward) approach than the agency shop.

In a new  essay in the University of Chicago Law Review Dialogue, we explain in detail how this alternative arrangement might work, but here’s a quick synopsis: Let’s say a union in California represents a school district’s 100 teachers and has collective bargaining costs of $100,000 per year. To make things simple, let’s also assume that all teachers in the district earn $50,000 annually. Under the existing agency shop arrangement, every teacher must pay $1,000 to the union to cover her share of the union’s collective bargaining costs. (Union teachers also pay additional dues to cover expenses unrelated to collective bargaining.) So in our example, although teachers nominally earn $50,000, their effective pay is really $49,000, because each pays a minimum—and mandatory—$1,000 agency fee. A more straightforward way to accomplish the same result would be to adopt what we call the “direct payment alternative,” whereby the district would reimburse the union directly for its bargaining costs. To offset these costs, the district might then want to reduce each teacher’s salary from $50,000 to $49,000. The teachers, however, would be no worse off: They would no longer have to pay $1,000 in agency fees or union dues, so their net wage (before taxes) would be the same as in the agency shop arrangement.

Even if five justices vote to overrule Abood, it seems unlikely that the Court’s decision would prevent public-sector employers from adopting the direct payment alternative. Money may be considered speech under current First Amendment doctrine, but expenditures of government money count as “government speech.” And the Supreme Court’s “government speech” precedents, such as Rust v. Sullivan, establish that the government has wide latitude to make policy judgments when it chooses what speech to fund. Indeed, the direct payment alternative may actually be preferable for unions, even if the court declines to discard Abood. Under the court’s post-Abood decision in Lehnert v. Ferris Faculty Association, unions are subject to a number of constitutional limitations on how they may collect and spend agency fees. Those limitations would be unlikely to apply under a direct payment method, given the breadth of permissible government speech.

An added benefit of the direct payment alternative—at least from the perspective of public-sector employees—is that it would also result in most of them paying less in federal taxes. In the agency shop example above, each teacher’s adjusted gross income for federal tax purposes would be $50,000, and her wages for Social Security and Medicare tax purposes would be $50,000 as well. While union dues and agency fees are deductible for federal income tax purposes, many employees cannot take advantage of this benefit—either because they don’t itemize deductions on their returns or because they fall below the 2% floor on miscellaneous itemized deductions. Under the direct payment alternative, the school district’s reimbursements to the union would be excluded from employees’ income, and so teachers in our hypothetical school district would be taxed on $49,000 instead of $50,000. Even employees who itemize on their returns and whose miscellaneous deductions exceed the 2% floor would be better off under the direct payment alternative—both because the reduction in gross income would lower the 2% floor and because of the payroll tax savings.

Why, then, haven’t public-sector unions prodded employers to switch to the direct payment alternative already? The straightforward answer is that in roughly a dozen states with agency shops—including the Friedrichs plaintiffs’ state of California—direct payments from public-sector employers to labor unions are prohibited by state law. But states can change their laws, so this doesn’t explain why states maintain agency shop arrangements notwithstanding the availability of the direct payment alternative. Nor does it explain why states chose agency shop arrangements instead of the direct payment alternative in the first place.

One possible reason is that even though public-sector employees would be better off post-tax under the direct payment alternative, employees don’t realize that fact. The required payment to the union may be an example of what behavioral economists call a “low-salience” fee. In other words, an annual salary of $50,000 with a required agency fee of $1,000 might appear more lucrative than a $49,000 salary and no fee. This may be so even though under the direct payment alternative, a public-sector employee actually pays less in federal taxes and takes home more of her salary as a result. So public-sector employers may favor agency shop arrangements in order to inflate the sticker salary that they can offer potential employees—even though after-tax pay is lower under the agency shop than the direct payment setup.

A second—and perhaps more likely—reason is political: A direct payment alternative might initially be met with hostility from taxpayers and lawmakers alike. Voters and legislators might object to the appearance of states and municipalities subsidizing public-sector unions. But appearances can be deceiving—and revealing: state and local governments in agency shop states already subsidize public-sector unions when they require employees to pay agency fees. The direct payment alternative simply makes this subsidy more transparent.

Does the possibility of the direct payment alternative mean states could neutralize the effects of a decision to overrule Abood? In at least eight states with agency shops, we believe direct payments from public-sector employers to unions are arguably already legal. In those states, public-sector employers could minimize any potential fallout of Friedrichs without any change to state law—and they might want to do so regardless of the outcome for the reasons noted above.

In about a dozen more states with agency shop arrangements, direct payments from public-sector employers to unions are prohibited by law, and those laws would need to change. This might not be so difficult in some instances, since agency shops arrangement exist only in states that already look favorably on public-sector unions. As a result, political support for public-sector unions in those states may be so strong that a decision to overrule Abood would have limited practical effect: those same states that presently enable agency shop arrangements may just change their laws to allow direct payments instead.

To be sure, support for public-sector unions is not equally strong in all agency shop states. In California, for example, the state Senate, Assembly, and Governor’s mansion are all controlled by Democrats, and public-sector unions presumably would find powerful allies willing to support a direct payment alternative in the event Abood is overruled. Thus even if the Friedrichs plaintiffs win before the Supreme Court, California could very well implement the direct payment alternative instead, continuing to prevent the free-rider problem among non-union public employees. In other agency shop states, however, divided government might make Friedrichs’ holding more consequential. In Illinois, for example, it appears unlikely that Governor Bruce Rauner—who has clashed with public-sector unions throughout his tenure—would sign off on direct payments.

Does the availability of the direct payment alternative mean that the fight over agency shops is “much Abood about nothing”? We don’t think so: the outcome in Friedrichs really will determine whether public-sector unions in some states (e.g., Illinois) are reimbursed for the cost of representing nonmembers in collective bargaining. Moreover, the direct payment alternative and the agency shop differ meaningfully—in the way they are taxed and in the way they are perceived. We do think, though, that whichever way Friedrichs goes, states will still have the option of adopting arrangements that address the free-rider concern that motivated the agency shop to begin with. Not every state will adopt our direct payment alternative, but the chief obstacle will be political—not constitutional.

The Tax Returns of the Top 400: A Deeper Dive

Assistant Professor Daniel Hemel on taxes paid by the highest earners in the United States:

The IRS released data this week on the 400 individual income tax returns with the highest adjusted gross incomes (AGIs). According to the IRS data, the average federal income tax rate for the top 400 was 22.9% in 2013, down from a peak of 29.9% in 1995 (though up from a low of 16.6% in 2007). Much has been written about the IRS data already (and no doubt more will be written in the coming days and weeks), but three trends deserve more attention than they have drawn thus far.

First, and most remarkably, charitable contributions by taxpayers in the top 400 have increased dramatically since the mid-1990s—both in absolute dollar terms and as a percentage of income. In 1995, the average taxpayer in the top 400 reported charitable contribution deductions of $2.9 million, slightly less than 5.6% of AGI. In 2013, that number was $32.8 million—more than 12.1% of AGI. That fact alone explains a sizeable share of the decline in the average tax rate of the top 400. Assuming that charitable contribution deductions offset income that otherwise would have been taxed at the top marginal rate, then charitable deductions reduced the average tax rate of those in the top 400 by 2.2 percentage points in 1995 and by 4.8 percentage points in 2013. If so, then 37% of the decline in the average tax rate of the top 400 (i.e., 2.6 percentage points) would be attributable to the increase in charitable contribution deductions.

Second, deductions for taxes paid to state and local governments (as well as taxes paid to foreign governments for which no foreign tax credit was claimed) rose from 5.2% of AGI for the top 400 in 1995 to 7.8% of AGI in 2013. Some of this increase might be explained by the fact that taxpayers have had the option of deducting state and local sales taxes in lieu of income taxes since 2004; however, the available data suggests that the 2004 change made at most a minor difference for the top 400. (The taxes-paid deduction as a percentage of AGI for the top 400 barely budged when the change in law went into effect—from slightly more than 4.0% in 2003 to slightly less than 4.1% in 2004.) More likely, the rise in taxes paid reflects the rise in state income tax rates: California’s top rate rose from 11% in 1995 to 13.3% in 2013, and New York’s top rate rose from 7.5% in 1995 to 8.82% in 2013. And assuming that taxes-paid deductions offset income that otherwise would have been taxed at the top marginal rate, then 14% of the decline in the average tax rate of the top 400 (i.e., 1 percentage point) would be attributable to the increase in taxes-paid deductions. (Those in the top 400 who are subject to the alternative minimum tax wouldn't be able to utilize the taxes-paid deduction, but the AMT affects very few among the very rich: only 2.4% of income taxes paid by the top 400 comes through the AMT.)

Third, foreign tax credits claimed by the top 400 increased from less than 0.7% of AGI in 1995 to nearly 1.2% of AGI in 2013. The increase in foreign tax credits claimed explains roughly 8% of the decline in the average tax rate of the top 400 between 1995 and 2013 (more than 0.5 percentage points). If foreign tax credits claimed reflect foreign taxes actually paid, and if the taxes-paid deduction also reflects state and local taxes actually paid, then taxes paid by the top 400 to all governments (federal, state, local, and foreign) amounted to 35.8% of AGI in 1995 and 31.9% of AGI in 2013. On those assumptions, total taxes paid by the top 400 fell by 3.9 percentage points from 1995 to 2013—still a substantial decline, but much less than 7 percentage point drop that one sees when one focuses exclusively on federal income taxes.

These calculations are, concededly, crude. The IRS data shows deductions and credits claimed by the top 400 for taxes paid to state, local, and foreign governments; the data does not show actual taxes paid to those governments. (It omits, for instance, state and local sales taxes paid by individuals who elected to deduct state and local income taxes.) Moreover, a number of the above calculations assume that the marginal tax rate for all those in the top 400 is 39.6%; however, the IRS data indicates that some in the top 400 actually face a lower marginal rate (likely because their income primarily consists of dividends and capital gains taxed preferentially). Finally, none of this is to deny that the very rich paid less in taxes as a percentage of income in 2013 than they paid in 1995. What it does suggest is that a singular focus on federal income taxes as a percentage of AGI may miss part of the picture: the very rich appear to be giving much more to charity—and paying more in taxes to state, local, and foreign governments—now than they did two decades ago.