abstract. In November 2014, President Obama announced his intention to dramatically reshape immigration law through administrative channels. Together with relief policies announced in 2012, his initiatives would shield nearly half the population of unauthorized immigrants from removal and enable them to work in the United States. These events have drawn renewed attention to the President’s power to shape immigration law. They also have reignited a longstanding controversy about whether constitutional limits exist on a central source of executive authority: the power to enforce the law.

In using the Obama relief policies to explore these dynamics, we make two central claims. First, it is futile to try to constrain the enforcement power by tying it to a search for congressional enforcement priorities. Congress has no discernible priorities when it comes to a very wide swath of enforcement activity—a reality especially true for immigration law today. The immigration code has evolved over time into a highly reticulated statute through the work of numerous Congresses and political coalitions. The modern structure of immigration law also effectively delegates vast screening authority to the President. Interlocking historical, political, and legislative developments have opened a tremendous gap between the law on the books and the law on the ground. Under these conditions, there can be no meaningful search for congressionally preferred screening criteria. Far from reflecting a faithful-agent framework, then, immigration enforcement more closely resembles a two-principals model of policymaking—one in which the Executive can and should help construct the domain of regulation through its independent judgments about how and when to enforce the law.

Second, when exploring limits on the enforcement power, we should focus not on who benefits from enforcement discretion but on how the Executive institutionalizes its discretion. The Obama relief initiatives are innovative: they bind the exercise of prosecutorial discretion to a more rule-like decision-making process, constrain the judgments of line-level officials by subjecting them to centralized supervision, and render the exercise of enforcement discretion far more transparent to the public than is customary. These efforts to better organize the enforcement bureaucracy ultimately advance core rule-of-law values without undermining deterrence or legal compliance, as some critics have worried. Moreover, while our focus on discretion’s institutionalization requires contextualized judgments that may rarely translate into clear doctrinal rules to govern the enforcement power, we believe it is generally unnecessary and unwise to use constitutional law to limit the President’s authority over how to organize the enforcement bureaucracy.

authors. Adam B. Cox is the Robert A. Kindler Professor of Law, NYU School of Law. Cristina M. Rodríguez is the Leighton Homer Surbeck Professor of Law, Yale Law School. For their sharp insights and generous comments, many thanks to Muneer Ahmad, Rachel Barkow, Ming Hsu Chen, John Ferejohn, Jamal Greene, Rick Hills, Dan Ho, Lewis Kornhauser, Marty Lederman, Steve Legomsky, Ethan Leib, Daryl Levinson, Dave Martin, Nick Parrillo, Dave Pozen, Daphna Renan, Judith Resnik, Ricky Revesz, Shalev Roisman, Adam Samaha, Peter Schuck, David Sklansky, David Strauss, Peter Strauss, and participants in the faculty workshops at NYU, Berkeley, and Stanford Law Schools, the legal theory workshop at Columbia Law School, and the Law, Economics, and Politics Workshop at NYU. We are immensely grateful for the outstanding research support of Meg Braun, Olivia Gibbons, Kate Huddleston, Katie Kavanagh, Beezly Kiernan, Gabe Panek, and Susan Smelcer. Adam Cox thanks the Filomen D’Agostino and Max E. Greenberg Research Fund, and Cristina Rodríguez thanks the Oscar M. Ruebhausen Fund at Yale Law School, for generous support of their research.

Introduction

On November 20, 2014, President Obama announced sweeping executive reforms of immigration law.1 The centerpiece of his announcement was an initiative designed to provide a measure of security to millions of unauthorized immigrants. Under it, executive branch officials would exercise discretion to defer the deportation of unauthorized immigrants who have lived for years in the United States and have U.S. citizen (or green-card holding) children. Parents who received this “deferred action” also would be eligible to receive work permits. As many as 3.6 million noncitizens may be eligible for relief under the program—a number that jumps to more than five million when the program for parents is combined with an earlier-announced initiative for unauthorized immigrants who arrived in the United States as children.2 Together, President Obama’s efforts could protect nearly fifty percent of today’s unauthorized immigrant population.3

The President’s decision to defer the deportation of millions of immigrants sparked sharp debate among scholars and political figures about his authority to create such a large-scale relief program. The Administration provided an unusually meaty framework for the debate by releasing an opinion, prepared by the Office of Legal Counsel (OLC) in the Department of Justice, concluding that the initiative was well within the Administration’s statutory and constitutional authorities.4 Critics disagreed with OLC’s conclusion, decrying President Obama’s actions as not just unwise but unconstitutional—the latest installment in the rise of an imperial presidency.5 The debate quickly made its way to the federal courts, as nearly two dozen states challenged the relief programs in a lawsuit that, as of this writing, remains pending and has resulted in the temporary injunction of the President’s initiatives.6

These events have drawn renewed attention to the President’s power to shape the substance of immigration law through the exercise of his enforcement power. They have also reignited the longstanding controversy over whether any limits exist on this central source of executive authority. Both of these issues were at the heart of our previous work, The President and Immigration Law.7 Published in these pages six years ago, that article provided a historical account of the distribution of immigration lawmaking authority between the President and Congress. Our core claim in that piece was that a series of twentieth-century developments—constitutional, historical, and institutional—had, as a functional matter, given the President tremendous power over the immigrant-screening system: power to determine which immigrants would be permitted to remain in the United States, and which would be forced to leave.8 We labeled this constellation of developments “de facto delegation” and argued that it constituted one of the most important features of modern American immigration law.

Developments since we last wrote, culminating in President Obama’s recent announcement, have both confirmed our earlier account and raised important new questions. While our previous work was mostly descriptive and historical, intervening developments have sharpened the legal and theoretical separation of powers questions raised by our argument. Moreover, whereas in 2009 we chiefly addressed the allocation of power between the branches in immigration law, the passage of time has highlighted the importance of power allocations within the Executive Branch for understanding the on-the-ground practice of presidential immigration law. Thus, this Article seeks to move beyond our earlier arguments in two ways—by squarely confronting the legal and normative questions about the President’s power over immigration policy, and by carefully unpacking the “unitary” Executive to develop better purchase on these questions and on our earlier descriptive account of the President and immigration law.

This Article makes two central claims about the relationship between enforcement discretion and the separation of powers, both in immigration law and more generally. The first concerns the substantive limits on enforcement discretion: what (if anything) constrains executive branch choices about which immigrants will be protected through the exercise of enforcement discretion? The second concerns the institutionalization of that discretion: what (if anything) constrains executive branch choices about how to institutionalize the exercise of enforcement discretion within the bureaucracy? While we address these questions by focusing on the Obama relief initiatives, the questions themselves implicate broader separation of powers debates and will remain pressing even if opponents of the President’s relief initiatives emerge victorious in the pending federal litigation.9

With respect to our first argument, we show that efforts to constrain the President’s enforcement authority with reference to “congressional enforcement priorities”—an approach taken by both defenders and critics of the President—are doomed to fail.10 We recognize the appeal of this approach. By tying the exercise of enforcement discretion to inferences about congressional intent drawn directly from immigration statutes, the Administration can claim to be acting as Congress’s faithful agent, following the principal’s wishes rather than making policy unmoored from the dictates of immigration law’s elaborate statutory scheme. On this account, Congress makes the tough value judgments, not the President. He or she simply extracts those underlying value judgments from the statute through sophisticated legal analysis. The approach also provides a seemingly clear limiting principle to prevent the enforcement power from devolving into dispensation of the law—something that supporters of large-scale administrative relief had failed to provide until OLC shifted the tenor of the debate.

The trouble is that this faithful-agent model obscures the role that enforcement discretion plays in our modern system of separated powers. Even outside the immigration context, it would be passing strange to argue that the myriad discretionary decisions made by law enforcement officials should always be motivated and constrained solely or even primarily by the value judgments those officials can trace to a code enacted by Congress. Moreover, this model is especially limited as an account of immigration law. Our historical account of separation of powers in this domain highlights the ubiquity of presidents exercising discretionary immigration authority in ways that cannot be characterized as consistent with clearly identifiable congressional priorities.11 That history has combined with a series of other developments—most notably the growth of the deportation regime and the size of the unauthorized population—to create the de facto delegation model of immigration policymaking. The tremendous authority wielded by the President under that model to shape our immigrant screening policies renders talk of “congressional priorities” for enforcement inapposite. We do not think it possible to coherently identify a set of congressional priorities for immigration enforcement through a careful, lawyerly exercise of intertextual fidelity to the 300-page immigration code.12

Far from fitting into a faithful-agent framework, therefore, our modern system of presidentially driven, ex post immigration screening is better understood as embodying a “two-principals” model of immigration policymaking. One possible response to the emergence of this model would be to decry it as lawless. But that would be a mistake. We see significant value in a model of the enforcement power according to which executive priorities stand alongside congressional ones. As the history of immigration law has demonstrated, this model empowers the Executive to address the unanticipated costs and epistemic limits of ex ante congressional lawmaking, calibrate the policies enacted by Congress to changed circumstances, provoke constructive and innovative policy reforms in both branches, and guard against the perils of legislative stasis. Policymaking through enforcement may not advance these objectives all the time, and it could certainly be abused. But given the reality of de facto delegation and the benefits that flow from the President’s current role, it would be a mistake to dismiss policymaking through enforcement as lawless.

While we reject substantive limits derived from congressional priorities, our second claim is that we can still meaningfully address the desirability or legality of particular regimes of enforcement discretion. As we explore in Part III, the better inquiry into the legality of President Obama’s relief programs, and the use of the enforcement power more generally, asks whether the Executive should be constitutionally prohibited from institutionalizing enforcement discretion in particular ways. The most important aspects of the President’s immigration initiatives have nothing to do with the substantive criteria for relief; the program’s focus on children, families, long-term residence, and clean criminal records strongly resembles the approach contained in many earlier, much less controversial guidance documents intended to channel prosecutorial discretion.13 Instead, the more important innovation was to make the exercise of discretion more rule-like, centralized, and transparent. These features have been the focus of prominent critics, who have argued that the President has wielded prosecutorial discretion in an impermissibly “categorical” way, rather than in a valid “individualized” fashion, or that he has extended substantive “legal benefits” to unauthorized immigrants, rather than mere forbearance.14

The institutional choices embodied in the President’s initiatives thus raise issues far beyond immigration law: they concern broader debates about centralization, transparency, and bureaucratic justice. How one evaluates the choices embodied in the President’s plans, therefore, cannot be divorced entirely from one’s views on some classic debates about the theoretical and legal underpinnings of the American administrative state. In that sense, the President’s critics are correct that much more is at stake than the justice of deferring the removal of long-term residents of the United States.

At the same time, critics err in thinking that those debates can be resolved in this instance without a historically grounded understanding of the immigration separation of powers. The institutional account of immigration law that we have jointly developed over the course of the last several years ultimately helps explain exactly why the President’s immigration initiatives are both lawful and desirable. They promote important rule-of-law values, such as transparency and accountability, as well as the age-old aim of treating like cases alike. And they do so without threatening to undermine another rule-of-law value—legal compliance—that some have claimed will be compromised by the President’s initiatives.15 Conjuring out of Article II ether a constitutional prohibition on the way the President has institutionalized discretion in his recent immigration initiatives would significantly undermine these values, and for essentially no benefit. Moreover, it would entrench the authority of low-level bureaucrats against alternative judgments about how best to arrange power within the bureaucracy—even judgments by the very Congress that created the bureaucracy.

Our complementary arguments—against the congressional priorities approach and in favor of a focus on discretion’s institutionalization—ultimately show how the leading critiques of the President’s relief initiatives go wrong.16 Yet our two central claims are important not only (or even primarily) because they help us properly evaluate the legality of the most important presidential immigration initiative in several decades. They also address a set of shortcomings in modern separation of powers and administrative law theory. Principal-agent models borrowed from contract theory and positive political theory have been invaluable tools for analyzing the administrative state. But those models also have serious limitations. In this Article, we illuminate one crucial area of executive power where standard principal-agent models obscure much more than they illuminate. We also show that the project of fleshing out separation of powers theory, descriptively and normatively, must occur with much more institutional and domain-specific context than is typical in contemporary constitutional scholarship. Far from an argument for immigration exceptionalism, our analysis highlights how immigration is just like multiple other domains of regulation, in that each evolves according to particular legal, practical, and political dynamics. Though we may be able to identify abstract goals that a system of separated powers should serve, how power has been and ought to be allocated among the branches to serve those goals will differ across time and setting.

This emphasis on context does not mean that the search for generalizable limiting principles or theories in separation of powers contexts is doomed. In fact, the arguments we make in Parts II and III together provide a framework, which we develop in Part IV, for thinking about limiting principles that can serve separation of powers values while accounting for institutional and historical context. Moreover, our defenses of presidential immigration law in general, and President Obama’s immigration initiatives in particular, do not amount to a conclusion that current congressional-executive dynamics are optimal. We conclude in Part IV, therefore, by taking seriously the second-best nature of immigration law’s current structure. We consider reforms—both modest and radical—that would promote and discipline the role that the President currently plays in American immigration law.

I. a brief history of presidential immigration law

Before we can evaluate the immigration enforcement initiatives announced by President Obama and understand the scope of the contemporary enforcement power, some history is in order. This Part situates the initiatives within a century-long story of administrative innovation that produced modern American immigration law. Only with this context can we make sense of the motivations for, and the legality of, the President’s deportation relief programs.

We show that the Obama relief initiatives represent only the most recent examples of the executive policymaking that has been part and parcel of immigration history. The President has always been an immigration policymaker alongside and sometimes in competition with Congress. President Obama’s recent actions simply reinforce the ways in which the content and scope of the President’s regulatory authority have evolved in response to the actions of Congress, as well as underlying historical and social factors. That evolution has been complex, involving a combination of partisan politics, economic and demographic forces, social movement pressures, and institutional demands. This specificity of context, however, does not turn our account into a tyranny of particularism. The trajectory we trace provides important, generalizable lessons that, as we will show in Parts II and III, have direct implications for how we judge the legality and desirability of the President’s relief initiatives and the use of the enforcement power more generally.17

These lessons ultimately differ considerably from the ones that some supporters of the President’s initiatives have drawn from pieces of the history we recount below. Some commentators have argued that the initiatives are lawful because they sufficiently resemble actions by previous administrations—in particular, the use of administrative relief by Presidents Reagan and Bush during the implementation of a legalization program enacted by Congress in 1986. We neither treat this history as quasi-legal precedent, nor rely on debatable notions of congressional acquiescence to executive branch practice to make claims about constitutional settlements between the branches. Instead, we use this history to provide a thorough account of the structure of modern immigration law, identify the imperatives and temptations that attend the use of the enforcement power in light of that structure, and explain the motivations for present-day uses of that power.18 Our history underscores what critics fail to understand about the nature of enforcement today, and in that sense it provides the context for a reality-based articulation of the scope of the enforcement power.

In this Part, we begin by summarizing our 2009 account of how the President historically has used the powers expressly delegated to him to advance his own policy agenda, resulting in what we term executive unilateralism. We then turn to the central source of power at issue in this Article—enforcement discretion. We demonstrate how the underenforcement of certain parts of the immigration code, as in many domains, has transformed the law enacted by Congress into regulation that reflects executive branch priorities. We then elaborate on the concept of de facto delegation introduced in our earlier work and explain its relevance to current controversies. In keeping with our focus on the internal organization of the Executive Branch, we close by documenting the trend in recent decades toward the Executive’s centralization of its enforcement discretion. Taken together, these perspectives on executive power help make the descriptive case for the two-principals model defended in Part II and provide the institutional detail required to understand what precisely is at stake with the Obama relief initiatives.

A. From Delegation to Unilateralism

In our 2009 work, we identified three models of executive authority that emerged over the course of the twentieth century: inherent presidential authority,19 express delegation,20 and de facto delegation.21 Each model arose through institutional practice and amidst confusion in the courts about the constitutional role each branch was supposed to play in the exercise of the federal government’s immigration power. By the late twentieth century, consonant with the dramatic expansion of the delegated administrative state, the first tradition of inherent authority had receded.22 But presidents looking to mold immigration law to advance their own objectives have rarely needed to resort to claims of inherent constitutional authority. Instead, they have used authorities expressly delegated to them by Congress, or taken advantage of their role in enforcing congressional schemes (the source of de facto delegation) to advance their own agendas.

Throughout the twentieth century, and up to the present, the President has used powers expressly delegated to him by Congress to advance his own immigration agenda. Importantly, these uses have often been innovative, accomplishing objectives Congress almost certainly did not intend and expanding or repurposing Congress’s original design. Congress has at different moments resisted and accommodated these efforts, in some moments moving to limit the originally delegated power in an effort to rein in executive branch efforts, while at others creating new statutory frameworks to accomplish some of the Executive’s objectives.

Perhaps the best twentieth-century example of this phenomenon is the President’s use of the parole power. Contained within the original Immigration and Nationality Act (INA) of 1952, the parole power permits the President to exercise discretion and allow otherwise inadmissible noncitizens into the United States.23 As we explained in 2009, beginning with President Eisenhower’s admission of 15,000 Hungarians fleeing the communist crackdown in their country, the power served as “the central tool of American refugee policy,” enabling the President to control refugee admissions for over twenty years.24 Though Congress attempted to curtail the President’s use of the power by enacting a refugee preference regime in 1965, presidents continued to wield the discretionary power that Congress intended only for “emergent, individual, and isolated situations” in order to admit large groups of noncitizens, including during refugee crises from Cuba, Haiti, and Vietnam.25 A combination of settled expectations and political pressures eventually led Congress to make those temporary admissions permanent, underlining the President’s agenda-setting power.26

With the Refugee Act of 1980, Congress directly responded to the executive-driven agenda in two ways. First, it added language to the parole provision requiring that the discretionary act serve compelling reasons in the public interest—an addition many in Congress (perhaps mistakenly) regarded as a means of “bring[ing] the admission of refugees under greater Congressional and statutory control.”27 Second, and more importantly, it created a scheme for overseas refugee selection that expressly delegated power to the President to set the number of annual refugee admissions and to select the countries from which they would be accepted.28 In 1990, Congress further systematized the process of admitting noncitizens fleeing disaster by creating the Temporary Protected Status (TPS) designation, which authorizes the President to permit categories of noncitizens to remain in the United States on a temporary basis, provided they meet statutory criteria defining the types of calamities Congress deemed worthy of response through protection.29 The combination of these new provisions suggests that Congress sought to replace the nontransparent use of parole and other discretionary mechanisms with semi-supervised and controlled schemes of delegation that required the President to submit his recommendations to congressional committees and to consult with various agency heads in the process.30

As we will explain later, the substitution of delegated and visible authority for discretionary and opaque authority generally should be welcomed. But here, it is important to see how the President made these supposed new constraints on his authority his own. A common critique of the President’s implementation of the refugee selection system in the 1980s and 1990s, for example, was that admissions during that period skewed toward nationals of then-Communist regimes, suggesting that the President used the system in order to advance his particularistic foreign policy goals rather than the more universal humanitarian objectives of the 1980 Act.31 This critique simultaneously assumes that the two goals are mutually exclusive and that Congress had a clear purpose it thought should drive refugee selection.

Whether either of these claims has merit is beside the point for our purposes. Instead, what matters is that the President utilized his delegated authority to serve a decidedly executive agenda. The creation of the refugee selection process in 1980 and TPS authority in 1990 may have diminished the need for sweeping and categorical use of the parole power, as well as the political and legal flexibility of the President to rely on parole as he had in the past. But these effects have been more modest than one might suppose, and parole remains an important alternative route of admission for those who may not qualify for refugee status.32 The authority also continues to serve as a basis for innovation. Most recently, the Obama Administration has invoked parole in place—itself an innovation on the parole power33—to provide relief for a large group of unauthorized immigrants already in the United States—relatives of members of the military. Though the application for and granting of parole continues to be framed as case-by-case, the memorandum announcing parole in place for military families clearly reflects an intent to provide relief to a favored category of unauthorized immigrants.34

This sort of creative unilateralism, which we identified in our 2009 article, has arisen in many other instances. A vivid example is the once obscure but now frequently invoked “family fairness” policies adopted by Presidents Reagan and George H.W. Bush after Congress enacted a large-scale legalization program in 1986. The legalization program, part of the Immigration Reform and Control Act (IRCA), provided a path to legal status for millions of unauthorized migrants, but it did not extend to many of the spouses and children of those immigrants. Despite this, President Reagan’s Immigration and Naturalization Service (INS) in 1987 elected to defer the removal of many of these family members35—a deferral President Bush continued, and then expanded in 1990 when legislation to legalize their status stalled in Congress.36 Later that year, Congress enacted a statutory legalization for the group.37

These deferrals of removal can be cast in two very different lights. First, we might see them as nothing more than a form of transitional relief. On this account, Presidents Reagan and George H.W. Bush operated within a statutorily created legalization framework, but in the course of implementation identified inequities (and perhaps oversights) in the design of IRCA’s original program. They used their discretion to ameliorate those inequities—to prevent the removal of family members who eventually would be eligible for immigration status through their newly legalized spouses or parents. Once debate began in Congress over new legalization legislation that would reach family members left out of the initial legislation, thus obviating the need for those family members to petition through the ordinary immigration process, the actions of the Presidents truly became transitional amelioration pending congressional action.38 If the statutory legalization scheme would soon encompass those family members, it would make little sense—as a matter of resource allocation or justice—to deport large numbers of them during the period of legal transition.39 Far from being oppositional, the President’s actions could be seen to exemplify cooperation between the Executive and Congress in the implementation of a large new initiative.

Of course, the family fairness regulations could also be seen as an act of executive defiance. On this account, Congress’s intent as reflected in IRCA was to provide legal status to a precisely defined group of unauthorized immigrants. And President Reagan’s actions, in particular, amounted to a kind of executive rejection of the parameters of IRCA’s legalization program and a unilateral decision to protect a group that the President, but not Congress, regarded as deserving. Perhaps these very actions forced the issue onto Congress’s agenda and helped secure the statutory change adopted in 1990. Such unilateralism might have made Presidents Reagan and George H.W. Bush’s judgments at the time more subject to question, but this characterization would also make so-called family fairness more of an on-point precedent for the Obama relief initiatives,40 which emerged through the President’s use of quintessentially executive authority, rather than in the implementation of a congressional legalization scheme.41

Whatever the appropriate characterization of family fairness, the episode embodies two of the characteristics of the separation of powers in immigration law that we have emphasized here and in other work. First, the particular tool Presidents Reagan and George H.W. Bush used to extend relief to the “ineligible spouses and children of legalized aliens”—extended voluntary departure (EVD)—was an innovation on enforcement discretion that emerged to address particular contingencies and grew in scope over time. The origins of, justifications for, and evolution of EVD are somewhat obscure and poorly understood. But it appears to have developed in an ad hoc fashion in the 1960s and 1970s, as a class-based form of relief from deportation. The Executive typically, though not exclusively, directed it at nationals of particular countries, often for humanitarian reasons or because conditions in the noncitizens’ home countries were dangerous or chaotic.42 Certain Cuban nationals permitted by President Eisenhower to remain in the United States in 1960, for example, benefitted from EVD.43 And though it was most often used to address foreign policy-related exigencies, presidents came to use EVD to exert considerable authority over who could remain in the United States even when foreign policy was not at issue.44

The innovative nature of EVD extended to the legal justifications for the power: executive branch officials appear to have toggled between at least two different sources of legal authority to support its use. In 1985, officials in the Reagan Administration testified that EVD stemmed from the “Executive’s constitutional authority in the areas of foreign and prosecutorial policy (supplemented by the general delegation of power over immigration in 8 U.S.C. § 1103(a)).”45 In 1987, however, officials claimed a more specific statutory source for the authority, contending that the power expressly delegated in the INA to grant voluntary departure (an alternative to formal removal whereby a noncitizen departs of his own volition) implied the power to grant EVD, or a temporary reprieve from removal.46 Though the latter justification appears to have prevailed, probably because it points to a firmer statutory foundation than the former, it is clear that the legal authority for the practice emerged and evolved alongside (and not in advance of) the practice itself.

Second, the family fairness episode highlights the dynamic nature of the congressional-executive relationship. Executive actions like those taken by Presidents Reagan and George H.W. Bush can powerfully shape the congressional agenda and the future path reform takes in the legislature. They can also, as the broader history of EVD highlights, prompt Congress to attempt to control executive discretion in order to advance Congress’s own policy goals. In the 1980s, for example, House and Senate subcommittees called hearings to insist that the President exercise his EVD power to defer the removal of noncitizens from El Salvador.47 The appropriations process was used for a similar purpose: appropriations bills for fiscal years 1982-1983 and 1984-1985 contained statements, admittedly nonbinding, that it was “the sense of the Congress” that Salvadorans should be granted EVD.48 In other words, Congress sought to use and constrain novel forms of executive decision making to advance the congressional agenda, ultimately replacing the Executive’s ad hoc discretionary tool with clear statutory authority to extend relief under circumstances specified by Congress.49 These dynamics are by no means unique to immigration law, but they have been notable throughout its history.

B. Policymaking Through (Under)Enforcement

A central feature of the examples of innovation discussed above is that they all emanated in some way from either express congressional delegation or the process of implementing a discrete congressional program over which Congress had given the Executive expansive implementation authority. These historical instances of executive policymaking therefore differ in significant respects from the Obama relief initiatives.50 The latter were formulated pursuant to the President’s determination as to how to go about enforcing the INA as a whole, not as the result of an express statutory delegation to defer the removal of certain categories of noncitizens or as part of the implementation of a larger program. To be sure, the INA expressly grants the Secretary of the Department of Homeland Security (DHS) broad authority to enforce the Code—a provision numerous defenders of the Administration have cited to support the Obama relief initiatives.51 But this general authority to enforce the Code cannot reasonably be characterized as an express delegation of any particular form of authority; it is instead a recognition that the Executive will need to develop policies and protocols to accomplish all that the INA does expressly delegate.

The scope for executive policymaking in law enforcement contexts is vast, as commentators have emphasized with respect to numerous domains.52 Immigration law is no exception to this basic fact of the American system of separated powers. As is true in other arenas, the way the Executive exercises its enforcement discretion over time powerfully shapes the meaning and significance of the law. In enforcing the INA—a multi-faceted and complex code—the Executive must make numerous decisions, large and small, about how and when to wield its power.53 In so doing, it must navigate the vagaries of ideologically diverse public and congressional opinion; observers will often criticize the same enforcement strategy as both feckless and draconian. In addition to addressing the basic question of how to allocate enforcement resources between the border and the interior,54 the President and officials within DHS must determine the specific means for each sort of enforcement. At the border, should it rely on fencing and technology as deterrents,55 or apprehensions and quick returns? In the interior, should its focus be on employers who hire unauthorized workers, on identifying and removing noncitizens who have committed crimes,56 or on removing unauthorized noncitizens generally? Congress sometimes sets the stage for or constrains these choices through authorization and appropriations laws,57 but in the main, the complexity and breadth of the tradeoffs required of the Executive transform him into a policymaker.

Historically, the President has exercised this power using a variety of legal tools. Prosecutorial discretion in the form of “deferred action”—the mechanism for the Obama relief initiatives—represents just one. We reserve analysis of that law enforcement tool until Part III and focus here instead on another crucial but oft-overlooked example of the President’s use of the enforcement power to advance his agenda (arguably at the expense of Congress): underenforcement of the employer sanctions regime. The history of the Executive’s weak (some might say irresponsible) implementation of this major congressional initiative illuminates how the President’s constitutionally assigned role to execute the laws gives him power over the contours and significance of a statutory scheme.

Created in 1986 by Congress in tandem with IRCA’s large-scale legalization program in 1986, the employer sanctions regime imposes both civil and criminal sanctions on employers who hire immigrants not authorized to work in the United States.58 The theory behind employer sanctions was that penalizing employers who hired unauthorized workers would reduce the labor market incentive for illegal immigration. Congress also included employer sanctions to help justify legalization—as a promise that future legalizations would be unnecessary because IRCA would eliminate one of the primary reasons for illegal immigration.59

From its inception, however, the employer sanctions regime has been largely ineffectual.60 Its weaknesses stem in part from the statute itself and the tradeoffs built into it. As the Supreme Court recognized in Arizona v. United States, IRCA reflects Congress’s efforts to balance the desire to prevent the hiring of unauthorized immigrants with the concern that overzealous prosecution could give employers incentives to discriminate against potential workers on the basis of race or national origin.61 But these legislative tradeoffs form only part of the story. The Executive itself has taken much of the bite out of this signature congressional enforcement initiative. Across administrations, different combinations of political desires and institutional concerns have led to varying degrees of underenforcement of the statute, leading lawmakers and scholarly commentators to doubt that IRCA has played a significant role in curbing unauthorized immigration to the United States.62 This underenforcement may reflect the Executive’s desire to satisfy business or labor constituencies (to varying degrees depending on the party in office). It could also reflect the government’s desire to target enforcement resources in the direction most saleable to the general public—toward safety risks and border enforcement—goals also reflected in Congress’s expansion of the criminal law grounds for removal and appropriation of funds for border enforcement.63

Considered at a more granular level, it also becomes clear how each administration has calibrated its enforcement judgments under IRCA to address the particular mix of political and institutional pressures it has faced, managing the domain of enforcement according to its own policy preferences. Studies of the early years of implementation point to low levels of enforcement that declined over time, accompanied by the failure to develop strong incentives for compliance in immigrant-heavy industries.64 One leading history of IRCA argues that the Reagan and George H.W. Bush Administrations’ commitments to deregulation led to INS policy focused on educating businesses, rather than imposing penalties on them.65

Not much changed in later years. With one limited exception, the number of investigations, warnings, and fines directed at employers all declined steadily and dramatically from around 1990 into the 2000s.66 A brief uptick in enforcement occurred during the mid-1990s,67 following a Clinton Administration directive that called for “strengthening worksite enforcement and work authorization verification . . . to better protect American workers and businesses that do not hire illegal immigrants.”68 But this strengthening of enforcement dissipated within two years as the Clinton Administration shifted its efforts away from both sanctions and worksite raids and toward targeting the removal of noncitizens with criminal convictions.69

This trend toward targeting noncitizens who had committed crimes continued under the George W. Bush Administration, which seemed uninterested in employer sanctions and was focused on national security targets in the wake of the attacks of September 11, 2001. While we know that millions of unauthorized immigrants have long been employed by hundreds of thousands of employers,70 for years during the Bush Administration, DHS fined fewer than one hundred employers for violating IRCA.71 For several years, the number of both final orders issued and fines levied hovered close to zero,72 suggesting the Administration was doing next to nothing to enforce the statute.73 Even when the Administration increased scrutiny of workplaces near the end of President Bush’s second term, the enforcement that resulted took the form of a series of high-profile worksite raids that led to the arrests of hundreds of workers. DHS issued only thirty administrative fines against employers.74

The Obama Administration disavowed high-profile raids in favor of employer audits, while continuing the focus on noncitizens who had committed criminal offenses.75 The emphasis on audits led to increasing numbers of employer sanctions during President Obama’s first term.76 Yet even with the Obama Administration’s increased attention to employer compliance, weak enforcement of IRCA has been a perennial feature of the immigration system. This underenforcement likely stems in part from Congress’s ever-growing focus on border enforcement in the appropriations process77—an indirect means of de-prioritizing employer sanctions. But the Executive has been more directly responsible for deflating the 1986 statute. It has consistently chosen to focus its enforcement strategy elsewhere, rendering a signature congressional enforcement initiative largely irrelevant to immigration policy.

The complex enforcement history of IRCA reflects a crucial feature of the enforcement power we take up in more detail in Part II—that the President (through enforcement) and Congress (through appropriations and oversight) together continue to make policy and redefine the meaning of a statutory regime long after its enactment, as the regime unfolds in practice. This history also underscores how the President’s enforcement judgments drive much of that development, constructing over time the domain of regulation. These enforcement judgments may take the scheme in practice far from the intentions of the enacting Congress, but such is the consequence of enforcement.78

C. The Rise of De Facto Delegation

In our 2009 article, we juxtaposed how presidents have used authorities expressly delegated to them, such as the parole power, with a phenomenon we termed de facto delegation. The concept relates to the sort of ordinary enforcement discretion that requires priority setting and judgment and can lead to the underenforcement we describe above. But the phenomenon is also more radical, amounting to a system of executive decision making about who may remain in the United States—a system that effectively substitutes for congressional judgment. Importantly, this delegation of de facto screening authority comes not from specific statutory enactments, but emerges instead from the modern structure of immigration law as a whole.

At bottom, de facto delegation is the result of a profound mismatch between the law on the books and reality on the ground, which has resulted from a series of legal, political, and demographic developments that have accelerated over the last four decades.79 Three features of the immigration code produced by Congress helped establish the conditions for de facto delegation. First, the Code renders removable any noncitizen who enters the United States without authorization. This seemingly simple legal command has intersected with complex demographic and social trends—in particular, record levels of migration, both legal and illegal, over the last thirty years80—to produce an unauthorized population that reached over twelve million at its peak in 2007 and has remained above eleven million in recent years.81 This is an arrestingly large number. It means that nearly half of all noncitizens currently living in the United States are formally deportable under the immigration code.82

Second, since the late 1980s, Congress has made increasing numbers of criminal offenses predicates for removal, sweeping even minor drug crimes into the Code and expanding the definition of so-called “aggravated felonies.” That term of art initially only encompassed very serious crimes, such as murder and rape, but has come to encompass numerous minor offenses, including many misdemeanors.83 This makes the pool of deportable noncitizens significantly larger, adding to those who are unauthorized many legal immigrants, including large numbers of lawful permanent residents. The size and complexity of the population thus eligible for removal, coupled with the fact that removal requires investigations, arrests, and charging decisions by immigration police and prosecutors, means that the Executive wields tremendous screening power—functional authority to make judgments about the types of noncitizens who should be permitted to remain in the United States.

Finally, the scope of the Executive’s discretion at the enforcement stage has only been augmented by recent congressional decisions to constrain the authority of immigration judges to grant relief from removal at the end of deportation proceedings.84 These restrictions on relief have often been conceptualized as limiting the role discretion plays in immigration enforcement. But far from eliminating executive discretion, these provisions have simply moved the power to provide relief to the arrest and charging phase, shifting the exercise of discretion from immigration judges to prosecutors and immigration police.85

Given the central role Congress has played in the rise of de facto delegation, we chose in 2009 to describe it as a cousin to ordinary delegation. But in using the term “delegation,” we do not mean to suggest that Congress clearly intended at any moment in time to create a system of vast ex post executive screening. Instead, the concept describes a structural reality inherited from a series of choices overtime—choices that have created a parallel executive screening regime through which the Executive exercises its own value judgments about the scope of our immigration policy.

Moreover, it would be a mistake to describe de facto delegation as the product solely of congressional choices. Enforcement judgments also have contributed to its rise during the era of mass migration, through the sorts of enforcement tradeoffs described in Part B. For instance, the Executive arguably has contributed to the scope of illegal immigration by declining to either prevent the entry of or remove in large numbers unauthorized immigrants who do not pose public safety or national security risks.86 Even in a world of ever-increasing resources for immigration enforcement,87 the gap between law on the books and on the ground has failed to close in any meaningful way, in part because of executive policy judgments. Seen in this light, Congress’s decision to shower the enforcement bureaucracy with resources has served only to further increase the Executive’s capacity to shape the pool of immigrants living in the United States.

Even the Supreme Court has embraced the central role the President plays in structuring the modern immigration screening system. In Arizona v. United States, the Court struck down most provisions of an Arizona law designed to augment federal immigration enforcement. While many of the provisions of the Arizona statute precisely tracked the INA—statutory text one might think embodied Congress’s enforcement priorities—the Court rejected Arizona’s attempt at redundant enforcement. Under the Court’s theory of preemption, federal immigration law consists not only of the legislature’s work, or the terms of the Code, but also of the enforcement choices the Executive makes. These choices elevate certain elements of the Code over others or reflect the Executive’s desire to emphasize “human concerns” that the Executive has come to appreciate in the course of its enforcement but that might not be embedded in the Code.88 By conceptualizing immigration law in this way, the Court converted the ordinary exercise of prosecutorial discretion into binding federal law that preempted Arizona’s immigration initiatives.89 Thus, even though the state sanctions mirrored the federal statute, they were preempted because they conflicted with the way in which the Executive Branch had wielded its enforcement discretion. Whether we think federal enforcement priorities ought to have preemptive effect—a move that could be quite disruptive to federalism in the administrative state—the Court’s move speaks powerfully to the independent role the Executive plays in the development of the very meaning of a statutory scheme.90

For our purposes here—dissecting and understanding the enforcement power—the signal feature of de factodelegation has been the priority setting it entails, which can result in profound and widespread policy effects. As we argued in 2009, “[T]he President’s inability to set formal admissions and removal criteria has not precluded him from playing a major role in shaping screening policy.”91 This ex post form of screening authority has amplified the President’s control over our immigration policy, despite the fact that (and paradoxically because) Congress has maintained a virtual monopoly over ex ante screening. Congress, perhaps unwittingly, has borne considerable responsibility for expanding the domain of enforcement in a way that has magnified executive policymaking power, by making the INA more and more complicated and rule-bound since its adoption in 1952. As we will argue in Part II, the fact that Congress may not have contemplated or intended these effects does not render the presidential actions producing them unlawful. Instead, the rise of de facto delegation underscores how regulatory domains evolve over time through the interplay of legislative acts and discretionary enforcement choices.

D. Centralizing Enforcement Within the Executive

As we explained in 2009, there are reasons to be concerned about the increasingly outsized role enforcement policy has come to play in the formulation of immigration policy. The scale of de facto delegation, in particular, has given rise to a variety of good governance and rule-of-law concerns. As has been emphasized recently in debates about policing and criminal justice, enforcement judgments are often opaque and, for that reason, frequently resist accountability.92 In addition, enforcement imperatives often empower low- and mid-level officials, especially as the size of the enforcement pool expands.93 The diffusion of responsibility that results may make it more difficult to structure and control enforcement policy according to priorities established by executive branch leadership, let alone by Congress.

Presidential administrations are, of course, attentive to these concerns, if only because they implicate the territorial tussle between political leadership in Washington and agents in the field. Without control over the bureaucracy, it can be difficult, if not impossible, for a modern administration to implement its agenda. Accordingly, the modern history of presidential immigration law is as much a story about the organization of the Executive Branch, and dynamics among actors within it, as it is an account of the relationship between the Executive and Congress. Whereas in our 2009 article we focused exclusively on the latter, here we also seek to highlight how the former should factor into our account of the enforcement power and the separation of powers.

The Obama Administration, in particular, has responded to the demands engendered by de facto delegation with systemic and organizational changes, of which the 2014 relief policies represent only one example. The Administration’s enforcement policy as a whole has become increasingly directed at regularizing and making more consistent the operation of the de facto, ex post screening system—a system executive leadership came to see as too random and overly subject to the views of low-level bureaucrats and state and local officials.94 The motivations for its various regularization efforts have been simultaneously institutional and political. As a Democratic administration, the politics of immigration required that it commit to enforcing the law, but also that it respond to enforcement’s perceived excesses. And from DHS’s institutional point of view, agents in the field (both federal and local) had become too powerful in dictating the direction of administration policy.

The Administration’s centralizing and regularizing moves have been myriad, but three prior to the announcement of the first relief initiative in 2012—Deferred Action for Childhood Arrivals (DACA)—stand out. First, the Administration’s virtually unprecedented decision to file preemption lawsuits against Arizona and several other states, challenging state laws designed to buttress the federal enforcement regime, reflected a desire to retake federal control over the immigration debate and suppress state efforts to shape both immigration policy and politics.95 Second, in 2011, DHS released the so-called Morton Memos, a pair of agency memoranda designed to regulate the use of prosecutorial discretion by line-level enforcement officials. These memos grew out of a long tradition of similar efforts in previous administrations to provide guidance to immigration enforcement officials.96 But the Morton Memos reached significantly beyond their predecessors by creating a tiered enforcement scheme. Accordingly, the immigration law world regarded them as more serious efforts to regularize discretion than past guidance documents. Finally, the Obama Administration’s decision to make Secure Communities (launched during the waning days of the Bush Administration) the centerpiece of its enforcement strategy reflected a turn to technology to systematize enforcement against criminal offenders. Secure Communities promised to displace the unpredictable human element of formal and informal cooperation with local police.97 Even though the President, as part of his November 2014 announcement, declared an end to the program and DHS replaced it with the Priority Enforcement Program, the core centralizing, data-sharing feature remains in place and likely reflects a permanent shift in the way DHS collects the information essential to its enforcement activities.98

With these moves as prelude, the motivations for the Obama relief initiatives come into sharper focus. Two of the centralizing moves succeeded, but one resulted in limited, if any, success and required recalibration. Federal lawsuits in Arizona and elsewhere successfully muted the state policymaking initiatives that had been accelerating prior to the litigation.99 Secure Communities took center stage and has come to account for the vast majority of removals from the interior of the country, superseding the more limited 287(g) Program, and systematizing long-standing informal cooperation between local and federal officials.100 But for reasons we discuss in detail in Part III, the Morton Memos did not achieve their objectives, at least to the extent they were motivated by a genuine desire to significantly curb line-officer discretion to initiate the removal of unauthorized immigrants without criminal records. These institutional developments, in turn, coincided with a powerful social movement of unauthorized youth demanding recognition of their rightful place in the United States. The movement eventually made its way into the White House, while the larger campaign against deportations made a strong impression on powerful local officials in places such as California, Chicago, and New York City, who began to resist participation in federal enforcement.101

The combination of these institutional and social movement pressures, along with the imperatives of the 2012 election, created the context in which the White House announced DACA and then, a little more than a year later, Deferred Action for Parents of Americans and Lawful Permanent Residents (DAPA). The “Obama relief initiatives,” as we call them, emerged in two phases. In June 2012, then-Secretary of Homeland Security, Janet Napolitano, announced what became known as DACA.102 According to the DHS memorandum accompanying the announcement, noncitizens without legal status who met certain criteria were eligible to apply for a renewable two-year period of relief from removal, as well as for the authorization to work in the United States.103 The central feature of DACA was that it covered blameless youth with longstanding presence in the United States, namely, unauthorized immigrants who had come to the United States before the age of sixteen and had resided continuously in the United States for at least five years.104

The Administration styled relief for that group as an exercise of prosecutorial discretion—a large-scale extension of the “deferred action” immigration authorities had utilized for decades as a case management and humanitarian relief tool.105 To underscore that the initiative fell within the President’s enforcement powers, the Administration emphasized that DACA would not confer a lawful status on its recipients, that the adjudicators of DACA petitions in United States Citizenship and Immigration Services (USCIS) retained discretion to deny applications of even those who satisfied the eligibility criteria, and that DHS retained the discretion to terminate the status at any time. By the end of 2014, approximately 638,897 noncitizens had been granted relief under DACA.106

In an address to the nation in November 2014, the President himself announced a second round of administrative actions designed to advance a variety of long-sought policy objectives.107 The centerpiece again consisted of a large-scale deferred action initiative, this time for the unauthorized parents of U.S. citizens and lawful permanent residents. Pursuant to this program, known as DAPA,108 eligible noncitizens who are not otherwise enforcement priorities for the government would be permitted to apply for the deferral of their removal, as well as work authorization, for three years.109 Alongside this new deferred action initiative, the Administration proposed to tweak the existing DACA program, expanding eligibility and extending the relief period to three years.110 Together with the announcement of DAPA, DHS Secretary Johnson also issued a memorandum identifying department-wide guidelines intended to govern removal and detention policies and budget requests more generally. The “Johnson Memo” reinforced the Department’s longstanding emphasis on public safety, national security risks, and border enforcement. To implement these priorities, however, the memo superseded all previous enforcement guidance with a new three-tiered scheme for prioritizing enforcement efforts.111

These Obama relief policies are thus best understood as the most dramatic and politically salient examples of a larger effort to centralize the vast enforcement authority that modern de facto delegation has given to the Executive. This centralization has entailed experimenting with different means of ensuring that political leadership within the agencies, as well as the White House, exert greater control over the structure of the immigrant screening system in order to advance the policy objectives of leadership, as well as to promote consistency and predictability in enforcement. Some aspects of this centralization have elevated decision-making authority within the bureaucracy—moving it from lower-level to higher-level decision makers. Other aspects have drawn power into the bureaucracy that otherwise might lie outside it—as is true in efforts to reduce the role of state and local actors in shaping the enforcement system.

All recent administrations have reflected some centralizing tendencies, but the combination of politics and current events has made centralization efforts particularly pronounced in the Obama years. Of course, the fact that the Obama relief initiatives arose in response to these intertwined institutional and political forces does not tell us that those initiatives are lawful. Nor are they lawful simply because they fit comfortably within the tradition of executive branch policymaking that we have brought to the fore in this Part and in our 2009 work. Indeed, the propriety of any one of the forms of executive action highlighted in this Part could be debated, and the mere historical rootedness of the particular exercise of a power is not sufficient to endow it with constitutional status.112

The emergence of the relief initiatives as administration policy does, however, highlight the dynamic evolution of the content and reach of executive power. In particular, the initiatives embody recent efforts by the President and political leadership to reorganize this power. Their importance stems from what they reveal to us about the Executive Branch’s internal operations and those operations’ relationship to core constitutional and legal values, and not just their substantive outcomes. In Parts II and III, we turn from this historical account to critique and justification, to explain how such internal reorganization can promote transparency and accountability and thus serve the objectives of the separation of powers, even as it might deviate from congressional design and advance the Executive’s own policy agenda.

II. the substantive grounds of enforcement discretion

Everyone debating the Obama relief initiatives agrees on two basic points. First, all acknowledge that executive branch officials have some discretion to decide whether and when to initiate a prosecution in an individual case. This understanding represents the paradigm case of the Anglo-American concept of “prosecutorial discretion.” Even those who insist most strongly on a constrained Executive accept this discretionary authority over charging decisions in both criminal and civil contexts.113 Second, all participants agree that the President cannot decline to enforce altogether a law that is constitutional. Such an effort to “suspend” the law would amount to an abdication of his Article II obligation to “take Care that the Laws be faithfully executed.”114

But how do we distinguish the constitutional exercise of prosecutorial discretion from an impermissible abdication of the President’s duty to enforce the law? Putting aside purely formal arguments about the distinction between permissible “underenforcement”115 and impermissible “suspension,” which suffer from serious conceptual problems,116 claims about how to draw this distinction typically take one of two forms. The first searches for principles that limit the substantive criteria that can serve as a basis for prosecutorial discretion. The second focuses on the way the Executive institutionalizes the criteria—that is, on how the Executive structures its decision making to take account of substantive criteria it has defined as relevant.

Prior to President Obama’s November 2014 announcement, few commentators had taken the first tack of focusing on whether the substantive grounds of relief in the President’s potential programs were themselves unlawful.117 But that changed when the Office of Legal Counsel (OLC) released a legal opinion to accompany the President’s unveiling of DAPA. Before the President announced his new relief policies, the Secretary of DHS and the White House Counsel turned to OLC, proposing two deferred action programs and seeking advice as to whether they were lawful. OLC found one within the Executive’s authority and the other not.118 OLC’s opinion honed in on the President’s substantive priorities, asking whether the central criteria for relief—being a parent of a U.S. citizen, for example—were lawful. The answer, according to OLC, could be found by asking whether providing relief to those singled out advanced “congressional priorities” embedded in the INA.

Though OLC developed its congressional priorities approach in response to a direct question about the lawfulness of DAPA, the opinion’s analytic framework transcends the details of any one scheme of enforcement discretion. In our assessment of it, then, we aim simultaneously to address the particularities of DAPA (as well as DACA) in order to help resolve the debate currently raging about these specific programs, as well as to consider the viability of a congressional priorities framework for understanding any general limits on enforcement discretion, which can take numerous forms. In other words, even if DAPA were never implemented119 and DACA were invalidated as the result of final federal court judgments120—outcomes we are skeptical will come to pass—evaluating the congressional priorities approach would remain an important task.

Putting aside one puzzling aspect of OLC’s congressional priorities approach—that it elevates an ordinary argument about agency compliance with statutory obligations into a constitutional argument about the President’s Article II obligations—the basic analytic framework of the “congressional priorities” approach seems straightforward.121 But its seemingly straightforward quality turns out to be an illusion. As we explain in this Part, tying executive discretion to congressional priorities cannot provide a satisfying limiting principle within immigration law because, for the vast majority of enforcement choices that must be made, there are no coherent congressional priorities to be extracted from the Code. Any inquiry into congressional priorities is thus likely to be futile, which is why the dueling accounts of those priorities supplied by OLC and its critics are both unpersuasive. Moreover, in addition to providing little interpretive guidance, the congressional priorities approach perpetuates a “faithful-agent” model of law enforcement that is neither descriptively accurate nor normatively attractive. Executive branch policymaking through enforcement actually advances certain goals of our scheme of separated powers. When it comes to the exercise of the enforcement power, therefore, we should embrace what we refer to as the two-principals model of decision making that has emerged in practice.

A. Congressional Priorities and Faithful Agents

Though we ultimately disagree with the OLC opinion’s approach, the opinion reflects the best instincts of OLC: that significant and novel executive branch policies ought to be scrutinized and that such scrutiny is doubly important when the exercise of power is unlikely to be reviewed by courts and raises potential separation-of-powers concerns.122 The independence of the Office’s judgment is also reflected in an aspect of the opinion Administration detractors seem to overlook: its conclusion that one of the President’s proposed initiatives was beyond his authority. Though OLC frequently advises the President that a proposed course of action would not be lawful,123 such advice is rarely made public, making the release of the opinion itself a remarkable event. In taking on the task of crafting a principle to limit a highly malleable form of executive authority, OLC’s actions highlight that law constrains the President’s actions.

Two crucial legal conclusions structure the analysis in the OLC opinion. First, the opinion rejects the idea that “resource constraints” provide a meaningful principle for limiting enforcement discretion.124 Many defenders of broad deportation relief had pressed that as a limiting principle. But OLC was right to reject it; as a limiting principle, it is virtually meaningless.125 The existence of resource constraints obviously provides a sufficient condition for the exercise of prosecutorial discretion. If the Executive lacks the resources to pursue every violator of the law, she must make choices about which ones not to pursue—that much is a truism. But resource constraints are not a necessary condition for the exercise of discretion: the paradigmatic historical justifications for prosecutorial discretion have little or nothing to do with resource constraints. And even were one to reject this history and conclude that resource limits should be considered necessary, the ubiquity of resource constraints would prevent this principle from providing any meaningful constraint on the exercise of executive authority. DHS has been showered with resources and operates with a budget larger than all other federal law enforcement agencies combined. Yet DHS could ignore broad swaths of the immigration code and still spend its appropriated dollars. After all, DHS currently spends its full appropriation every year and still manages to deport only a tiny fraction of the potentially removable noncitizens living in the United States.126

Instead of looking to financial constraints, OLC concluded that a limiting principle could be supplied by “congressional priorities” embedded in the Code: these priorities, it concluded, constrain the substantive criteria that can lawfully serve as the basis for deportation relief.127In its opinion, OLC determined that, where the decision to grant relief tracked priorities the Office unearthed from the statute, such as keeping intact the families of citizens and lawful permanent residents, relief fell within the permissible zone of discretion.128 But where OLC believed that the relief could not be tightly linked to priorities embodied in existing statutory provisions, it concluded that the Executive was without legal authority to act.129 The opinion surveys numerous executive branch uses of deferred action and emphasizes that Congress was aware of them, seeming to use past practice as a form of precedent. But the opinion then turns to determine whether the President’s new proposals building on that history are, in fact, “consonant with, rather than contrary to,”130 priorities derived from the statute itself.131 OLC ultimately determined that the decision in DAPA to provide relief to the parents of U.S. citizens and green card holders would promote congressionally articulated priorities, but that a proposed program to provide relief for the parents of DACA recipients would not.

To our knowledge, the notion that the exercise of enforcement discretion is lawful only if consistent with congressional priorities had not yet emerged as a claim in the debate at the time OLC issued its opinion. At that moment, we had not yet seen defended elsewhere the idea that executive priority setting ought to be informed by the Executive’s own analysis of the enforcement obligations (and forms of relief) Congress thought most important. At the same time, the approach feels familiar. It aligns analysis of presidential enforcement authority with the way courts (and offices such as OLC) decide whether administrative agencies have lawfully exercised their delegated authority. This focus on consistency with congressional priorities in the context of administrative rulemaking reflects the dominant approach to administrative law, in which principal-agent models—both informal and formal—are used to conceptualize and evaluate the administrative state.132 When we characterize Congress as the principal and the Executive as its agent, the obvious question becomes whether the agent is promoting his principal’s goals or, instead, advancing his own. The turn to congressional priorities in the OLC Memorandum thus reflects a larger commitment to a delegation-centric model of congressional-executive relations—call it the faithful-agent model of prosecutorial discretion.

On the surface, the faithful-agent model might seem to have even stronger purchase in the enforcement context than in other administrative settings. In rulemaking, Congress has expressly delegated policymaking and thus interpretive authority to the Executive, but the duty to enforce is more akin to a straightforward obligation to follow the law on the books.133 Congress passes laws, the Executive enforces them—or so the argument goes. Under this reasoning, the constitutional allocation of enforcement power to the Executive assumes that the President and the bureaucracy will enforce Congress’s policies and priorities.

Of course, elucidating those priorities will likely involve a more freewheeling, inference-based inquiry than entailed by ordinary statutory interpretation, because Congress does not typically draft statutory enforcement priorities to accompany its substantive rules.134 Priorities can be gleaned from any of a statute’s provisions and not just the provisions being enforced or interpreted.135 Any executive branch effort to limit its enforcement judgments based on its own understanding of the goals Congress sought to achieve with the statutory framework in question is thus likely to give the Executive Branch considerable interpretive authority.

But even with these caveats, it might remain appealing to ground enforcement judgments in an argument that they advance goals set by Congress. Under this view, enforcement judgments emanate from tough value choices made by Congress, not the President. The President simply extracts those judgments from the statute, using sophisticated legal analysis. Analytically, this approach preserves congressional supremacy in the lawmaking process. The strongest version of this model would treat the Executive as a functionary, though both OLC136 and commentators137 wedded to the principal-agent model recognize the reality that the Executive must exercise judgment when determining how to enforce the law. They simply seek to discipline that judgment in a way that ensures Congress, not the President, remains responsible for substantive policy.

B. The Limits of Congressional Intent

The appeal of the congressional priorities approach is understandable. But we do not believe it provides an effective principle for limiting executive branch enforcement judgments in immigration law and many other domains. The congressional priorities approach fails because those priorities are a mirage.

Little meaningful congressional guidance exists about how to appropriately structure the ex post screening rules for immigration law. As we explained in Part I, the modern structure of immigration law effectively delegates vast screening authority to the President. The interlocking statutory and political developments we describe have opened up a tremendous gap between law on the books and on the ground. In a world where nearly half of all noncitizens living in the United States are formally deportable, there can be no meaningful search for the congressionally preferred screening criteria.138 The keys to the immigrant screening system effectively belong to the Executive, which has the authority (and some might even say obligation) to define screening criteria.139 As we described earlier, administrations have wielded this authority to reshape the screening system over time, a practice the Obama Administration has continued.140

In theory, of course, Congress could constrain de facto delegation by complementing its substantive statutory enactments with detailed enforcement instructions or prohibitions. In practice, Congress has rarely done this—in immigration law or any other regulatory arena. Occasionally Congress blandly obligates DHS to do something like “prioritize the identification and removal of aliens convicted of a crime by the severity of that crime,”141 or to fund a particular number of beds for immigrant detention (34,000, to be exact).142 But loose language of prioritization does little to constrain the Executive’s authority,143 and even numerical prescriptions like the bed-space mandate only scratch the surface of the decisions the Executive must make when enforcing immigration law. Negative injunctions issued by Congress have the potential to be more powerful; prohibiting DHS from granting any immigrant deferred action, for example, would more seriously constrain the President’s power to structure the immigrant screening system. But these sorts of prohibitions are also rare.

In this world, it will generally be futile to search for “congressional priorities” that legally constrain executive branch decisions about which immigrants, from within the vast pool of eleven million unlawfully here, may be deprioritized for deportation (not to mention congressional views as to how such deprioritization ought to be structured). And given the absence of such priorities, efforts to invoke them ultimately only obscure the reality that executive branch officials are making important value judgments about our immigrant-screening system.

Our argument should not be confused with the claim that presidential immigration law grows out of inherent Article II authority and exists independently from Congress. To the contrary: the argument is perfectly consistent with the claim that the President has no inherent constitutional authority over immigration policy.144 In such a world, the Executive makes enforcement judgments within the domain Congress has created. Congress’s statutory grounds of removal, for example, specify necessary conditions for the exercise of the enforcement power against a noncitizen, and the President cannot act outside the domain defined by those conditions. Thus, if DHS decided to start deporting immigrants who had failed to pay child support—not a ground of deportability under the INA—that decision would be unlawful.

Nor is our argument that the very idea of “congressional priorities” is incoherent in principle, or that such priorities can never be identified in practice. Ordinary interpretation often entails the search for Congress’s specific intent or overarching legislative “plan.”145 The idea of legislative priorities (or purposes, or intent) is, in our view, crucial to the construction of any persuasive interpretive theory (though the fact that it has been embraced by so many conservative legal scholars arguing against DAPA’s lawfulness is perhaps ironic).146 When a court confronts the question of whether an immigrant’s criminal conviction amounts to a ground of deportability under the Code, statutory interpretation arguments grounded in legislative intent will be perfectly plausible.147

But statutory interpretation questions of this sort typically have as their focus a discrete piece of statutory text. While interpreting that text might require placing it in the context of a larger code or in relation to other statutory provisions, the inquiry will typically be much more grounded in a discrete set of legislative materials than in inquiry into enforcement priorities.148 Because, as we have noted, legislatures are not in the habit of writing enforcement instructions to accompany the substantive rules of a code, the congressional priorities approach will almost always be unmoored from any particular text and will require drawing inferences from a wide, amorphous range of statutory provisions and legislative materials. These materials are unlikely to contain much guidance. And the lack of guidance should come as no surprise, once we recognize that the pervasive failure of legislatures to write down enforcement instructions reflects the implicit delegation of those choices to the Executive.

That general challenge is only magnified in the specific context of modern American immigration law, where de facto delegation has given the Executive tremendous authority to manage the ex post screening rules by picking deportees from among a population of immigrants who are all obviously, and incontrovertibly, deportable. That is not to say, we reiterate, that the notion of congressional intent is conceptually incoherent. It is always possible to construct fanciful examples in which enforcement judgments would clearly contradict congressional purposes. Immigration law is no different in this respect. If the President announced that no enforcement resources would be directed toward immigrants with criminal convictions, and instead all resources would go toward deporting only long-term residents who were married to Americans, we would not hesitate to conclude that such an enforcement decision is prohibited by the congressional priorities embedded in the Code, as well as appropriations law. But no President is likely to adopt such a policy. Thus, within extremely broad limits—limits that, we show below, easily sweep up programs like DACA and DAPA—the structure of modern immigration law simply leaves us with no discernable congressional enforcement priorities.

To see the failure of the congressional priorities approach in practice, we need look no further than OLC’s efforts to extract such priorities from the INA in order to evaluate the two relief initiatives proposed by the Administration.149 OLC ultimately determined that the INA’s goal of promoting family unity justified DAPA, which provides relief to the parents of U.S. citizens and green card holders. It observed that the statute creates a path to lawful immigration status for immediate relatives of U.S. citizens without numerical limitation, and that “numerous provisions of the [INA] reflect a particular concern with uniting aliens with close relatives who have attained lawful immigration status in the United States.”150 But it rejected an initiative that would have provided relief from removal and work authorization for the unauthorized parents of the beneficiaries of the DACA program of 2012. The Office determined that such relief was beyond the President’s authority because the INA did not reflect “comparable concern for uniting persons who lack lawful status (or prospective lawful status) in the United States with their families. Extending deferred action to the parents of DACA recipients would therefore expand family-based immigration relief in a manner that deviates in important respects from the immigration system Congress has enacted and the policies that system embodies.”151

In the wake of the opinion’s release, critics of DAPA disagreed strongly with OLC’s view about how to cash out the congressional priorities embedded in the INA.152 The Code does not promote family unity in some abstract and general way, they argued. Instead, the Code sometimes makes immigration benefits available for family members and at other times conspicuously declines to do so.153 In other words, Congress has clearly and specifically defined the limited circumstances in which it values family unity, and the circumstances of DAPA recipients are not among them. For decades the INA has prohibited children born in the United States from immediately sponsoring their parents’ entry into the United States. A U.S.-born child must turn twenty-one before she can do so—a restriction that prevents the Fourteenth Amendment’s birthright citizenship rule from enabling unauthorized immigrants to acquire status quickly by having children in the United States.154 But U.S.-born children are precisely the group who, under DAPA, serve as the basis of relief for their unauthorized parents. This shows, say Administration critics, that OLC got things exactly backwards.155 To the extent the INA expresses priorities about when family unity should be the basis of immigration benefits, it has expressly rejected the priorities reflected in DAPA.

Similar arguments have also been made that DACA is inconsistent with the INA’s priorities. DACA treats early childhood arrival in the United States as the touchstone criterion for relief from deportation. But the INA nowhere privileges young arrivals in its immigrant screening rules. Moreover, Congress has repeatedly rejected the so-called DREAM Act,156 which would provide a path to legalization for many of the young migrants covered by DACA—further evidence, critics argue, that the Code cannot be read to reflect a congressional priority to provide protection to these young migrants.157

Who has the better argument? In our view, neither side persuades. OLC’s critics are correct that there is no general policy in favor of family reunification that applies consistently throughout the Code. But critics are wrong too: the mere fact that U.S. citizen children cannot file green card petitions for their parents until age twenty-one does not tell us that the Code prohibits their parents from being provided with some lesser form of relief from deportation. DAPA simply defers a parent’s deportation; it does not provide any lawful immigration status, let alone the right of permanent residency that comes with a green card. For the same reason, critics are mistaken in thinking that the Code’s inclusion of specific, limited grounds for “relief” from removal—like those contained in the Code’s “Cancellation of Removal” provision—undercuts DAPA’s legality. The relief provided under the cancellation provision is, again, green card status, not deferred action. If all forms of relief from removal, including deferred action, really had to be limited to the enumerated grounds of “relief” in the Code, then nearly every grant of deferred action would be unlawful—not just the President’s current policies—because DHS generally extends deferred action to noncitizens who are not eligible for more robust forms of relief like cancellation.158

If we attempt to abstract from any particular statutory provision to the claim that a web of provisions—really the whole Immigration Code read intratextually—dictates the result that either OLC or its critics are correct, we are left with an all-too-familiar level-of-generality game. At some high level of generality (i.e., does the INA prioritize families?) OLC’s view looks more persuasive. At some lower level of generality (i.e., does the INA endorse deferred action for the out-of-status parents of U.S. citizens and lawful permanent residents?) it looks less persuasive. But we have no way to determine which level of generality to choose, given the way the INA evolved over time. The INA, initially adopted in 1952 and amended in significant fashion many times in the decades since, consists of a long series of legislative accretions. Each addition to the Code reflects a complicated mix of conflicting priorities either balanced against one another by a single Congress or across Congresses. The provisions for family-based immigration benefits have, for example, evolved in complex ways over more than a century.159 There is little doubt that American immigration law makes family ties more important than do the immigration systems of many other nations.160 But the devil is in the details: the general principle of family unity has been defined, qualified, and cabined in numerous ways, as have the general policy goals of augmenting the U.S. labor supply and providing protection for noncitizens fleeing disasters of various sorts, for that matter. A statute like the INA—one constructing a comprehensive regulatory scheme that has evolved in dynamic fashion over time and that embodies such a high level of complexity—will often not be amenable to many common intratextualist interpretive moves. The legislative “plan” of the INA is so full of internal contradictions and complexities as to be nearly impossible to characterize as pursuing concrete “priorities” at anything other than the highest level of generality.

This problem is not unique to immigration law. Today, it is common to many regulatory arenas, and looking for congressional priorities to constrain enforcement discretion will therefore pose a more difficult problem than those typically posed by statutory interpretation. When it comes to the INA, no individual relief provision points to Congress’s intent to prohibit the adoption of a particular prioritization scheme for dealing with the eleven million removable noncitizens in the United States. Nor does the Code as a whole, read intratextually, do so. And at bottom the reason goes back to the general theory we laid out at the top of this Part: the rise of de facto delegation consolidated in the Executive the authority to make these sorts of judgments.

C. The Two-Principals Model of Immigration Policymaking

At a general level, debates over the scope of executive power traffic in two competing frames of reference. The congressional priorities approach embodies a faithful-agent model according to which the Executive, when fulfilling its responsibilities through rulemaking, administration, or enforcement, should always ask itself: “What would Congress do?” The President’s obligation is to reflect as nearly as possible the policy Congress would adopt, were Congress itself making the regulatory or enforcement decision. Under this framework, the Executive exercises no policymaking autonomy and refrains from making contested value judgments, even as it exercises judgment and sets priorities.161

When it comes to understanding the enforcement power, we believe this framework is mistaken as a descriptive matter and unappealing as a normative matter. Instead, we offer a contrasting account—a two-principals model162—according to which the President possesses his own policymaking power. This model appears most clearly in the foreign affairs context and in debates over the extent of inherent authority the President possesses as Commander-in-Chief or as a function of Article II. It also characterizes theories of administration and statutory interpretation that capture the power of the modern Executive to displace Congress as a policymaker.163 One of our core contributions in this Article is to elucidate how a version of the two-principals model also characterizes the enforcement domain, not as a matter of inherent presidential authority, but as a function of the imperatives of the President’s obligations under the Take Care Clause, which emanate from but are not wholly controlled by Congress.

In this Part, we begin by reinforcing this two-principals claim descriptively. We then move to establish why the Executive serves rather than undermines certain core separation-of-powers values when acting as a kind of second principal in the exercise of the enforcement power. But first, we should say a few words about what we mean by two principals. We do not mean to suggest that the President and Congress are substitutes. Instead, we envision the Executive as a principal because, using the tools conferred by both the Constitution and the historical development of a particular regulatory arena, the President acts as a policymaking counterpart to Congress. He does so by playing a major and independent role in constructing the domain of enforcement over time, defining whom and under what circumstances the law will regulate. Moreover, to say that the Executive is a co-principal does not mean that the President himself is authorized or obligated to act as a pure unitary principal of the sort sometimes imagined in the separation-of-powers scholarship. Enforcement power can be lodged in a variety of institutional locations within the Executive Branch. Part III explores this explicitly, disaggregating the Executive and considering the possibility that the policymaking potential of the enforcement power may necessitate, or at least justify, some degree of high-level political control of or supervision over priority setting.164 In the remainder of this Part, however, we bracket this institutional complexity and think about executive power in general terms.

1. Executive Construction of Enforcement Domains

The faithful-agent model cannot be squared with the reality of prosecutorial discretion in immigration and many other regulatory arenas. It is a descriptive impossibility. Outside the immigration context, for example, it would be strange to argue that the myriad discretionary decisions made by federal prosecutors and other law enforcement officials are (or should be) motivated only by a sense of the value judgments Congress made when enacting the criminal law. To the contrary, when a prosecutor makes a plea deal, she is much more likely to describe the choices embodied in the plea in terms of an all-things-considered pragmatic calculation that is guided by oversight within her office and, ultimately, by what justice requires. Her time would not be spent scouring the criminal code to unearth some latent congressional priorities that somehow compelled the particular plea deal.165 Prosecutorial discretion has long entailed executive branch officials’ legal authority (and responsibility) to make difficult value judgments about the exercise of the state’s coercive authority.166

This dynamic of prosecutorial policymaking is perhaps even more vivid when we move from the retail level of the line prosecutor to the level of the agency head or the President himself. As we documented in detail in Part I, a simple principal-agent model does not accurately capture the history of immigration law and enforcement. While the nature and scope of executive power in immigration law has evolved over time in response to events and structural phenomena, as it has in other regulatory domains,an especially notable fact of immigration history is that the President has regularly acted as an independent policymaker, pursuing agendas that his corresponding Congresses may or may not have shared. In the first century of immigration law, Presidents used quintessentially executive powers—namely the negotiation of treaties—to advance their agendas, and they were able to do so because Congress had yet to occupy the field of immigration regulation with an elaborate code.167 But even in the twentieth-century context of domesticated executive power, amidst the rise of immigration delegation, the President has played the independent policymaking role to at least as robust an effect, both in exercising delegated authorities and in large part through the exercise of the enforcement power in the context of de facto delegation.

Our account of IRCA, in Part I, presents a good example: the substantive legal regime that determines whether and how to regulate employers and their unauthorized workers has evolved over the last three decades through the application of the Executive’s enforcement judgments. IRCA as a regulatory system in 2015 looks quite distinct from IRCA as a statute enacted in 1986. The combination of partisan politics and the institutional dynamics of enforcement itself (the assessment of its costs and the efficacy of different methods of enforcement, for example) have reconstructed the regulatory domain Congress created with its initial statutory enactment.

But as the evolution of IRCA highlights, the modern immigration system should not be understood to embody a simple static and uncontested shift of authority to the Executive. Nor is this joint federal lawmaking necessarily collaborative or harmonious. Rather, it consists of the branches responding to one another’s regulatory choices. Presidential immigration law has precipitated a variety of responses from Congress, which has both ratified and resisted the Executive’s use of his authority.168 On occasion Congress has responded with actual lawmaking, as in the case of refugee policy.169 In other moments, Congress has wielded the power of the purse, using appropriations measures to shape executive branch conduct.170

In response to DAPA, in particular, House Republicans have sought to tie funding for the Department of Homeland Security to riders that would block implementation of the Obama relief initiatives; this is but the latest example of this phenomenon.171 That they have not succeeded could either demonstrate the limited utility of appropriations threats, or that these Republicans were simply grandstanding, exerting a less formal form of control through politics. The congressional-executive dynamic often does not rise to the level of lawmaking. Congress’s response to the President’s use of his de facto delegated power frequently takes the form of political posturing, whether during election campaigns or in hearings called to bring attention to opposition among members of Congress.172 But rather than think of these responses as reflecting the petulance or dissatisfaction of a principal whose agent has gone astray, we should understand them as embodying the rivalry of two principals and the friction that can result when the center of policymaking gravity moves from one to the other.

This two principals understanding of executive power differs in important respects from other influential accounts of interbranch relations. As noted in Part I, we disclaim the position that the legality of the President’s actions turns on whether precise historical precedents or analogs exist. In that sense, our argument diverges from and is more radical than the view, present in some scholarship as well as executive branch practice, that congressional acquiescence over time to a particular executive branch practice is what makes it lawful.173 Instead, on our account, the President effectively acts as a principal within a regulatory space that has been constructed over time, even if Congress has not acquiesced. In immigration law, that space is breathtakingly broad in part because of the rise of de facto delegation. And within that space, the President shapes immigration law by continually revising and restructuring enforcement authority.174At the same time, our account is more restrained than the one contained in the historical gloss literature. That literature concludes that practices to which Congress has acquiesced at Time One become immune from congressional override—that is, constitutionally entrenched—at Time Two. In contrast, in our account, Congress can defeat presidential power at Time Two, producing a more fluid politics of congressional-executive relations over time.

This defeasibility does not mean Congress is, ultimately, the only “true” principal, or that our account can be reduced to the claim that presidential immigration law is nothing more than the product of agency “slack.”175 The idea that mere agency slack is all that is at stake is misleading for the same reason that the faithful-agent framework (from which the idea of slack is drawn) leads us astray. Conceptualizing the tremendous divergence between congressional statutes and executive branch outcomes that we document in Part I as the product of slack suggests that we should find ways to control that divergence. Slack is undesirable—something one always wishes to squeeze out of the principal-agent relationship, something that we are saddled with only because principals are incapable of perfectly monitoring their agents. The history of immigration law we tell, however, suggests that presidential policymaking is too pervasive and autonomous to fit this model. And as we explain in Part II.C.2 below, there is value in that relationship that would be quashed by an insistence that the goal of administrative law and design should be to tighten up slack. It is true that, as a matter of formal game theory, the President can be labeled principal only if his authority is indefeasibly by Congress. But we think that model obscures the interbranch dynamics that have existed in practice, and we believe our conceptions of those dynamics must take account of that practice.

2. Against Faithful Agents

If we have succeeded in our descriptive account of two principals, at least within the domain of immigration enforcement, the question then becomes what to think as a normative matter about the system we now have.176 Our goal is not to erect a theory of separated powers from the ground up, or to identify an “optimal” separation of powers. It can be hard to avoid abstract generalities when attempting to articulate the reasons for horizontal divisions of power. The Supreme Court’s regular references to the prevention of tyranny or the protection of individual rights as the purposes of the separation of powers may ring true as far as they go,177 but they are little more than platitudes when a genuine competition for power is at stake, in part because those power struggles do not themselves involve a clear battle between tyranny and freedom. We are deeply skeptical that a true first-principles inquiry can succeed, and a central conceit of our work is that any theory of power allocation must emerge from institutional and historical context.178 That is not to say that we don’t think current arrangements can be improved—a task we take on in Part IV. But if we reason from abstractions rather than from practice, we will lose sight of a number of benefits that flow from the two-principals model of immigration lawmaking that has emerged over time. Given the system we’ve inherited, it will be useful to understand its upsides.

In so doing, we focus on a second-order set of tradeoffs that the separation of powers entails: the classic struggle between the exercise of power to accomplish the ends of government and the overarching need to ensure that power is constrained, or exercised in a nonarbitrary fashion.179 A central feature of this inquiry, for our purposes, is a debate about how best to ensure that the Executive acts in an accountable fashion, and then whether that accountability should be to apolitical norms of reasoned decision making or to popular and political conceptions of accountability.180 In this sense, the inquiry sounds as much in administrative law as it does in constitutional law. In the balance of this Part, we explain why the two-principals model as applied to the enforcement power resonates with aspirations for constrained and accountable governmental power across the branches. We reserve for Part III a full discussion of the accountability tradeoffs embodied in the Obama relief initiatives. A key insight of our account of immigration enforcement is that independent priority setting by the Executive can, within the scheme of separated powers, actually facilitate the constrained use of power. Much separation-of-powers scholarship concerns itself with the rise of the imperial presidency, which prompts views that range from cheerful acceptance to alarmism,181 though whether to cheer or warn often depends on the sources of power the Executive purports to be using.182 A crucial dynamic missing from this account, which revolves around the assumption that the Executive is the branch in need of constraint, emerges from our observation of how de facto delegation has operated in immigration law. Executive action and priority setting in the exercise of the enforcement power can serve to constrain power in a world of overbroad legislation. For example, the enforcement priorities articulated across administrations to emphasize the removal of security and safety risks constitute executive efforts to construct a more rational screening system within the overinclusive sweep of today’s immigration code.183

Moreover, the act of actually enforcing the law—of confronting its real-world effects—can help point to limits or unintended consequences of the law as drafted.184 Enforcement brings to life the consequences of legislation—one concrete manifestation of the informational advantages of the presidency. We should want the Executive Branch to have the power to grapple with those consequences based on judgments forged through its own experience. Indeed, these informational benefits can often only be acquired in a dynamic context, in which executive branch officials have authority to make decisions subsequent to congressional policymaking. For the Executive to respond to the lived experience of the law by shifting priorities can help hold the legislature accountable, but also advance a policy debate by pointing a regulatory regime in better directions.

These epistemic benefits of executive action also bring with them increased policy responsiveness. While responsiveness is no unalloyed virtue, executive-branch initiative taking can perform a valuable constitutional function in immigration policy, particularly within a system that governs a polity marked by deep ideological differences and in a domain where a significant legislative reform occurs, at best, once a generation.185 One way executive-driven priority setting has fostered responsiveness is by offering a counterpoint to the interest groups that dominated Congress, with the White House serving as an alternative site for organizing and advocacy for the immigrants’ rights social movement. Whatever we think about the merits of their various positions, multiplying outlets for interest-group competition, and expression of popular preferences through policy, can promote the responsiveness of government.

To the extent that the concern about executive policymaking through enforcement stems from the desire for accountability, we do not think the Executive suffers from a democracy deficit as compared to Congress in any meaningful sense.186 The mechanisms of democratic influence and accountability may differ from those that operate on the legislature, but they exist not only in the President’s election mandate, but also in the corresponding organization of interest groups that pressure the White House and the agencies, and in the modern media’s insistence that the President explain and justify his actions and take his policy initiatives “on the road.”187 In fact, the Obama relief initiatives and the reformulation of Secure Communities offer prime examples of the Executive’s responsiveness to popular dissent from administration enforcement policies and the underlying statutory framework that sets the stage for them. Relatedly, formal and informal interactions with Congress will themselves constrain the Executive. In the war-powers context, scholars have drawn attention to such potential. Stephen Griffin, for example, writes of a “cycle of accountability” that consists of interbranch interaction over time through which “mutual testing and deliberation results,” such that the branches learn from their mistakes.188

It may be that robust policymaking through enforcement creates disincentives for Congress to act. But while we could certainly characterize recent immigration history as embodying a failure of congressional will, we doubt that the causes of legislative stasis include executive initiative taking. Whatever the reasons for congressional inaction—whether they be dysfunction, paralysis, or simply the choice to not act—the President’s decision to offer an affirmative, substantive vision can expand the policymaking domain in constructive and idea-generating ways. In this sense, the Executive acts as an engine in the policymaking process. This agenda-setting function may be particularly vital during times of polarization that produce legislative stasis,189 such as the one we seem to be living through, but this function of executive policymaking is by no means limited to moments like ours.190 Presidential immigration law can help mitigate one of the ordinary costs of our separation-of-powers regime—the reluctance or inability of government as a whole to act—thus furthering the interest of a healthy tradeoff between power and constraint.

Again, action in and of itself is not necessarily good. And whether executive action of the sort initiated by President Obama in his relief policies will actually prompt further policy deliberation is an empirical question. Executive action often amounts to a second-best alternative to legislation. Nowhere is that more true than in the immigration enforcement context, where prioritization cannot provide legal immigration status to those who receive deferred action under Obama’s relief initiatives.191 Thus, we should remain concerned with the possibility that executive policymaking will disable or displace Congress in some way—a point we take up in more detail in Part IV. But having a rival or complementary policymaker in the Executive can be good for the democratic and problem-solving features of government.

***

Our dynamic understanding of the relationship between Congress and the Executive, and our conception of the domain of regulation as one that evolves over time through the exercise of the enforcement power, reflect what we believe to be an important moment in the intellectual history of separation-of-powers scholarship—both at the descriptive and the normative levels. The constitutional framework sets up institutional rivalries as forms of constraint on government power. But these constraints do not come exclusively from the formal powers the Constitution assigns each branch and the “checks” each one possesses over the other.192 Our immigration history shows how constraints can also arise from the push and pull of politics and institutional design.193 Unique to our account is the way in which we illuminate, through history, how constraints on government can arise when each branch acts as an institutional source of policymaking in the same domain.

This idea that the roles and powers of the political branches are defined through a complex historical process that cannot be easily captured through either formal models or deductive, judicial-style reasoning is far from limited to the immigration arena.194 In fact, it appears even in settings where the Constitution clearly allocates overlapping power to Congress and the President, as in the war-powers context.195 As we have emphasized, whether this sort of relationship between the branches proves productive really depends on context. Moreover, policymaking through enforcement judgments presents concerns that policymaking through rulemaking does not: the source of delegated authority is less clear, and it can be difficult to externally police the executive decision-making process—concerns we flagged back in 2009. In Part III, we take up this dilemma and explore how the Obama relief policies simultaneously harness the benefits of policymaking through enforcement, enhance accountability, and promote constrained government by making enforcement judgments more transparent. Even within our vision of executive policymaking through enforcement, we do believe there should be a limiting principle on executive action, in service of the basic value of constraining government power. We turn now to a more fruitful source of such a principle.

III. the institutionalization of enforcement discretion

The faithful-agent model of congressional-executive relations we reject in Part II focuses on the substantive relationship between the choices or values reflected in the INA and those reflected in the President’s enforcement of the statute. This approach, we have argued, leads us astray. But that does not mean we think all bets are off. In this Part, we turn from substance to process and argue that the inquiry into the legality of President Obama’s relief programs and other similar exercises of the enforcement power should revolve around whether the Executive should be constitutionally prohibited from institutionalizing prosecutorial discretion in certain ways.196 On this account, the concern is not who gets protected from deportation, but how they come to be protected.

Numerous critics of the Obama relief initiatives have focused on these process concerns and declared his actions unconstitutional because of the way they institutionalize the Executive’s discretion. One prominent critique holds that the initiatives are unlawful because they provide “categorical” forms of relief, rather than resting on the exercise of “individualized” discretion.197 A second claim emphasizes that the way the relief initiatives institutionalize discretion threatens to undermine the “rule of law,” by which critics mean legal compliance.

This Part explains why these claims are misguided as a matter of both law and theory. As a theoretical matter, these claims about President Obama’s relief initiatives are actually just retail-level examples of more general debates that have long raged in legal theory and administrative-law scholarship. We will show that these critiques all boil down to claims about one or more of three choices: the choices between (1) rules versus standards, (2) centralized versus decentralized control over prosecutorial discretion, and (3) secret versus public norms regarding the exercise of that discretion. The Obama relief initiatives’ central “innovation” is to bind the exercise of prosecutorial discretion to more rule-like criteria, to centralize the supervision of discretion to a greater extent than is typical in enforcement contexts, and to make the exercise of discretion predictable and transparent. In other words, DACA and DAPA choose rules over standards, centralization over decentralization, and transparency over secrecy.

These choices ultimately advance the core rule-of-law values of consistency, transparency, and accountability: they ensure that similar cases are more likely to be treated alike, that the exercise of discretion is more predictable, and that enforcement outcomes align more closely with the policy preferences of the agency’s political leadership and the President himself.198 This does not mean, of course, that the choices embodied in DACA and DAPA are legally required, or even that they would in all contexts be legally permissible. At a high level of generality and abstracted from the details of the Obama relief initiatives, these choices are contestable. In many situations, good reasons exist to prefer standards to rules, decentralization to centralization, and secrecy to transparency; these choices involve tradeoffs between values at the very core of the American legal tradition. But for this very reason, it is impossible to make much progress in evaluating how those tradeoffs cash out without careful attention to institutional context. And once we understand the institutional realities against which the Obama Administration developed its relief initiatives, the choices embodied in DACA and DAPA become easy to defend.

Importantly, we ground our defense of the institutional choices reflected in the Obama relief initiatives by accepting the importance of identifying limiting principles on the enforcement power—principles that arise from constitutional structure, not just extralegal sources.199 But we acknowledge that these limits must, given the structure of the modern administrative state, inevitably derive from the contextual application of the broader objectives we identify in this Part and Part II. Critics have been blind to both that institutional context and to the realities of administrative governance. Their view that the institutional choices embodied in DACA and DAPA are unconstitutional (rather than just undesirable) embodies a radical theory that the Constitution sharply restricts the ways in which the Executive may organize itself.

A. Rules and Standards

The Obama Administration’s relief initiatives institutionalize discretion in an innovative way—just not in the way critics charge.200 According to critics, DACA and DAPA are unconstitutional because they exercise discretion on a “categorical” rather than an “individualized” basis. The individualized exercise of discretion comports with the canonical form of prosecutorial authority, the argument goes. But the categorical exercise of discretion amounts to an unconstitutional act of executive “lawmaking.”201

It might be tempting to dismiss this claim by pointing to the conceptual flaw in this dichotomy. Every exercise of prosecutorial discretion, including those authorized by the Obama relief initiatives, is “individualized” in the sense that individual persons seek or will be granted relief as individuals.202 Given this, it is unclear what it even means to say that DACA and DAPA do not involve “individualized” determinations regarding the exercise of prosecutorial discretion. The fact that large numbers of noncitizens who meet the announced eligibility criteria will come forward to apply does not mean that each application will not be adjudicated on an individual basis. This sort of confused talk about individualized decision making is part of what plagued debates about profiling for years, and the history of that debate shows that there is little profit in trying to make this analytic distinction do much work as a matter of law or theory.203

But while the focus on “categorical” decision making is confused, we can recharacterize this argument to capture what seems to be at the heart of critics’ concern. The core of their objection appears to be that the Obama relief initiatives have substituted rules for standards in the exercise of prosecutorial discretion. As a descriptive matter, this claim is essentially accurate. The initiatives innovate because they move from a system of suggestive enforcement guidelines to a much more rule-bound enforcement system. Previously, grants of “deferred action” were made on an ad hoc basis, guided by loose priorities laid out in a series of agency memos on the exercise of prosecutorial discretion. These memos specified dozens of factors relevant to relief determinations (many of which embodied vague standards), and said little about the appropriate relationship between the factors.

DACA and DAPA reshaped this decision-making process to make it much more rule-bound. (They also formalized the application process, a point we will take up in a moment.) First, DACA and DAPA reduced considerably the number of criteria relevant to eligibility for relief; rather than dozens of factors being relevant, the initiatives selected just a handful. Second, the initiatives replaced loose criteria with more objective ones. Under DAPA, for example, the two most important criteria the applicant must establish are that she has resided in the United States for at least five years, and that she has a child who is a U.S. citizen or green card holder.204 Third, DACA and DAPA clearly specify the logical relationship among the listed criteria. Each criterion is a necessary condition, meaning that an immigrant must show that she satisfies all of the enumerated criteria in order to be eligible for the exercise of discretionary relief.

Defenders of the relief initiatives have tried to resist the claim that DACA and DAPA have a rule-like structure. The Department of Justice has taken great pains to emphasize—both in the OLC opinion and in the Texas litigation over DAPA—that the relief programs authorize agency personnel to exercise discretion to deny relief to otherwise-eligible noncitizens.205 But the preservation of formal discretion does not mean that DACA and DAPA are not more rule-like than the regime they would replace. The relevant question is whether, as a causal matter rather than as a formal one, discretion plays as large a role in deferred-action determinations under the new initiatives as under the old regime. The answer to that question is unequivocal: while discretion previously pervaded every aspect of each decision, it plays a very limited role under the new initiatives. Government-provided data for DACA show that almost no eligible applicants have been denied relief as a matter of discretion. From DACA’s inception until the end of 2014, USCIS approved 638,897 applications and denied 38,597.206 Most denied applicationswere rejected “based on a determination that the requestor failed to meet certain threshold criteria.”207 In other words, a full ninety-four percent of adjudications have resulted in grants.208 And of the six percent that were rejections, most were based on the failure to satisfy DACA’s eligibility criteria, not on the exercise of discretion to deny relief to an otherwise-eligible applicant.209 Thus, the program appears to have operated, as intended, to ensure that certain types of immigration-law violators—those who satisfied the rule-like eligibility criteria—succeeded in their applications and therefore became protected from deportation.

These outcomes do not mean that the discretion left to USCIS officials is not in some sense “real.” Out of the hundreds of thousands of DACA applications, there do appear to have been a small number in which adjudicators have denied relief to applicants who otherwise satisfied the eligibility criteria.210 Thus, our claim is not that the formal preservation of discretion is somehow a sham. The formal discretion left to adjudicators may have been intended to preserve some of the case-by-case flexibility for truly exceptional cases; or perhaps it was mainly intended to insulate the new policies from certain kinds of (misguided) legal challenges.211 But even if discretion remains relevant in rare cases, it is hard to resist the conclusion that the more rule-like components of the decision will be dispositive of ultimate relief decisions in the vast majority of cases.212 This reality is, in fact, precisely why the Administration replaced a wholly discretionary regime subject to suggestive guidance and ad hoc supervision with a rule-like application process in which discretion is limited to a backstopping role. As a practical matter, the structure of DACA and DAPA significantly constrains, if not functionally eliminates, the discretion of those adjudicating relief applications.

Critics are thus correct that the Obama initiatives replaced the old regime of discretion with a new regime bound by rules. Contrary to the critics, however, we do not believe this renders the programs unconstitutional. For these critics, the articulation of objective criteria and the overwhelming grant rates for DACA applicants who met the criteria render the program an unconstitutional act of executive “lawmaking.”213 Under this view, the distinction between rules and standards maps onto the constitutional division between “legislative” authority under Article I and “executive” authority under Article II; vague standards are less “legislative” than clear rules, and vice versa. But even for those who subscribe to formalistic accounts of the constitutional separation of powers, under which each branch exercises power of a particular type (and we are not among them), this argument makes little sense. Taken seriously, it would lead to the conclusion that Congress improperly exercises Article II “executive” authority when it enacts vague standards into law. The Sherman Antitrust Act would be unconstitutional, along with myriad other laws.214 Such an argument would require a robust reinvigoration of the nondelegation doctrine. It also would require courts to conclude that administrative agencies improperly usurp Article I “legislative” authority whenever they issue regulations that embody bright-line rules. So much for the modern administrative state.

More generally, American law rarely constitutionalizes the choice between rules and standards in public administration. Courts have in some cases constitutionally required reliance on rules—generally in cases where courts perceive a significant risk that discretionary decision making will serve as cover for discriminatory decision making. This idea runs through a number of First Amendment doctrines and helps explain the Warren Court’s criminal procedure revolution.215 In other circumstances, courts have constitutionally prohibited reliance on rules, generally where courts have concluded that adjudicatory discretion (and often a particular adjudicatory forum) must be preserved in order to secure the liberty or property interests of individuals protected by the Due Process Clause. The prohibition on categorical rules for pretrial detention in the criminal context offers a prominent example.216 But these instances are clear exceptions to the general agnosticism of constitutional law on this question.

Even if the choice whether to structure decision making rigidly or flexibly rarely raises constitutional questions, it does implicate a question at the heart of twentieth-century administrative law and bureaucratic design: what is the best way to mete out mass justice, including in contexts like the one at issue here, in which millions of cases demand the attention of the Executive?217 Attempts to answer that question generally must grapple with the foundational tradeoff between rules and standards: rules promote equal treatment across cases, but they necessarily define “like cases” in a more reductionist way than standards. Broad standards and “individualized discretion” can foster more fine-grained judgments about when justice or other equitable factors support relief.218 But they necessarily achieve nuance at the expense of equal treatment across cases—especially in a world where the same decision maker cannot decide all cases and power is diffused across a bureaucracy.219

There is no single answer to how best to strike these tradeoffs. In the context of mass administrative justice, however, the choice of whether to adopt rules or open-ended discretion is closely linked to the question of whether to centralize authority within the bureaucracy and exert significant supervisory authority over line-level executive officials. Within the Executive Branch, rules facilitate oversight and make it easier for high-level executive branch officials, many of whom are politically accountable, to prevent low-level agents from imposing their own views about when and how the law should be enforced.220 But rather than eliminating discretion from the system, as critics charge, constraining low-level decision makers with rules simply relocates discretion to a point higher up in the bureaucracy. It is in this sense that discretion most truly remains within the system under the President’s relief initiatives. That discretion simply belongs primarily to the Secretary of Homeland Security, Jeh Johnson, the officer to whom the INA formally delegates discretionary enforcement authority.221 He retains the power to alter the criteria for relief in any way he sees fit, or even to cancel the relief programs.222 And once we acknowledge that the formal discretion left to line-level adjudicators to deny relief is of little practical importance, this transfer of discretion becomes all the more pronounced. We doubt Secretary Johnson (of his own volition or at the President’s direction) will exercise his discretion to alter or terminate the program. But this doubt does not mean that Secretary Johnson actually lacks discretion; it simply means that we are confident about how he will exercise it.223

B. Supervision (Not Separation) of Powers

The argument that “categorical rules” violate some requirement of “individualized discretion” really amounts to an argument that the supervision and centralization of discretion in immigration law are prohibited. The prohibition on centralization could be cast as a statutory directive, as a constitutional requirement, or as an imperative of good institutional design. So far as we are aware, no one has advanced the statutory argument—that Congress embedded in the immigration code a requirement that enforcement discretion be located exclusively in the hands of line-level enforcement personnel. This argument’s absence is unsurprising, given that nearly all the statutory developments of the last several decades point in the other direction, promoting consolidation rather than diffusion within the Executive Branch of authority to make discretionary decisions about who should be deported.224

Accordingly, we focus in this Part on the institutional and constitutional versions of the anticentralization claim. As a matter of institutional design, we show that the efforts in DACA and DAPA to centralize decision making have been significant administrative improvements on the practices of diffused prosecutorial discretion that preceded them. Second, we show that the notion that either the separation of powers generally, or the Take Care Clause in particular, constitutionalizes decentralization within the Executive Branch in the way imagined by critics of the relief programs is wildly implausible.225

1. Institutional Design

The long-term trend in American bureaucracy has been toward centralization—elevating decisions within agencies themselves, as well as above agencies into the Executive Office of the President. Political scientists and legal scholars from Terry Moe to Justice Elena Kagan have documented this trend,226 and both unitary theorists on the right and advocates of presidential administration on the left have defended it.227 Even recent developments in the Supreme Court’s administrative law canon have promoted and (implicitly) defended administrative centralization.228

Considered abstractly, of course, it would be difficult to identify a single, optimal level of centralization that applies across the bureaucracy to all (or even many) agencies and regulatory contexts. It is unsurprising, therefore, that a number of administrative law scholars have resisted Kagan’s normative gloss on the centralization Moe describes.229 In the critics’ telling, centralization can diminish transparency, obscure lines of accountability, undermine expert decision making, and politicize agency action.

But our argument that the Obama relief initiatives promote the more disciplined and accountable use of executive power does not depend on taking a side in this general debate. If the last two decades of scholarship prove anything, it is that the appropriate level of centralization cannot be determined in the abstract; whether and how to centralize depend on how the relevant institutions operate in practice. Those who have argued that Obama’s relief programs are unconstitutional have mostly elided this institutional detail. But it is precisely this detail in the immigration setting that offers us a unique policy experiment with which we can actually assess the centralizing tradeoffs made by Obama’s relief initiatives. If we evaluate the initiatives in terms of what they replaced, we see how they promise to significantly improve the ex post screening system by regularizing it through the supervision of line officials.

Prior to DACA and DAPA, the Administration launched a prosecutorial discretion initiative that sought to preserve and guide line-level enforcement authority. The Obama Administration announced this initiative in June 2011 with the release of the so-called Morton Memos—directives that laid out the criteria the political leadership of Immigration and Customs Enforcement (ICE) and DHS wanted to govern the exercise of prosecutorial discretion by ICE employees.230 The memos defined “prosecutorial discretion” broadly to encompass nearly every sort of enforcement decision made by ICE agents, including, crucially, the decision to initiate removal proceedings and the decision to grant deferred action.231 As we noted in Part I, the memos were far from the first such documents; officials in both the Bush and Clinton Administrations issued guidance documents listing criteria intended to inform myriad discretionary enforcement judgments.232 Yet both the content of the Morton Memos and the timing of their release, coinciding as they did with broader agency efforts designed to bring consistency to the system of screening noncitizens for deportability, led many advocates to see the memos as heralding a new era in which immigration discretion would be wielded on a more widespread and consistent basis.233 This assumption may have been overly optimistic. The memos only articulated priorities; they did not indicate an intention not to remove low-priority targets, nor did they identify the means by which the priorities would inform the actual judgments of the line agents scattered across the country. By touting the memos, however, the Administration made a kind of political promise to shift the brunt of the enforcement system away from status violators and toward more serious offenders.

Many of the Morton Memos’ factors for exercising prosecutorial discretion (and granting deferred action) bear a marked similarity to the deferred action criteria eventually embodied in the Obama relief initiatives. The factors included an immigrant’s length of residence in the United States, as well as educational history, family ties, and criminal record (or lack thereof)—factors closely related to the eligibility criteria for both DACA and DAPA.234 Having been a child when one migrated to the United States—the keystone criterion under DACA—was also deemed important under the Morton Memos.235 Although the factors in the Morton Memos were less precise and more numerous, and although the logical relationship among them was not well defined, their resemblance to the DACA and DAPA priorities is unmistakable.236

Despite this resemblance, however, the Morton Memos did not have the immediate and obvious effects of DACA (and presumably of DAPA, once implemented), and perhaps for precisely that reason, they provoked much less public controversy than either of the Obama relief initiatives.237 In the months following the memos’ June 2011 release, there were few observable changes in the exercise of immigration prosecutorial discretion. According to widespread accounts, ICE continued to place immigrants who should have been among the lowest enforcement priorities in removal proceedings, routinely ignoring individual requests for deferred action.238 Moreover, a large-scale review of over 300,000 ongoing removal cases, implemented in conjunction with the memos’ release in order to identify those cases in which prosecutorial discretion was warranted, resulted in a very small number of case closures.239 And while the fraction of criminal deportees did go up somewhat through this period, that trend appears to be largely the product of changes to other enforcement initiatives, not the Morton Memos themselves.

Additional evidence of the Morton Memos’ ineffectiveness can be seen in the operation of Secure Communities, another signature Obama enforcement initiative. That program, launched in the fall of 2008 by the Bush DHS, turned every local criminal arrest in the country into a point of immigration screening.240 Universal screening at the point of arrest provides a tremendous amount of information to the federal government—information that can be used (and that the government has argued was designed to be used) to make decisions about removal both more consistent and more responsive to federal priorities. Under the program, federal officials, not local police, decided whether to place an immigrant identified through arrest data in removal proceedings.241

If the Morton Memos had actually significantly impacted the decisions made by agency personnel about whether to place a particular immigrant in removal proceedings, one would have expected to see that impact reflected in the pool of immigrants arrested by ICE under Secure Communities; those memos applied directly to arrest decisions made by ICE agents under Secure Communities. Yet no effect was apparent in the wake of the Morton Memos’ release. Figure 1 shows the composition of that pool over time, broken down by the criminal history of those apprehended and placed in deportation proceedings following notification to DHS as part of Secure Communities.242 Two aspects of the Figure stand out. First, a large percentage of people placed in proceedings under the program had no criminal history at all: nearly a third had no criminal conviction, despite the fact that the program was publicly touted as a means of targeting “criminal aliens.”243 Second, the composition of the arrestee pool did not change at all after the Morton Memos were released. The arrest decisions of line-level ICE agents under Secure Communities looks much the same before and after June 2011.244 While the memos formally singled out noncitizens without prior convictions as lower priorities for removal, the reality on the ground was that they were just as likely to be arrested by ICE after the memos’ announcement.245

Figure 1.

criminal history of noncitizens arrested by ice under secure communities246

These data highlight the limitations of the Morton Memos’ approach to producing meaningful changes in the exercise of immigration discretion.247 In retrospect, this limitation of a guidance-document-oriented approach should not be surprising. The memos embody an effort to shift the culture of enforcement at the agency through an articulation by leadership of best practices. But even if agency leadership sought to monitor compliance with the memos’ priorities through vigilant supervision buttressed by the disciplining of officials who consistently failed to respect the memos’ priorities, observable changes in enforcement practices would have taken considerable time to emerge. And, producing dramatic results of the sort achieved by DACA would have been elusive. The Morton Memos would have had to change the behavior of large numbers of ICE agents—the line-level enforcement personnel principally responsible for making decisions about whether to place an immigrant in removal proceedings.248

Many agents were extremely resistant to the memos’ central goal, some quite vocally.249 They work within a law enforcement agency that has an enforcement-oriented and results-driven institutional culture, not unlike the culture of the FBI and the DEA. It should not be surprising to find resistance within the ranks to the premise of the Morton Memos. Men and women who see their jobs as punishing lawbreakers could have felt as if they were being directed to ignore the transgressions of immigration violators.250 This discontent ultimately bubbled up through the employees’ union,251 and once DACA was announced, some members of the union filed suit against the Secretary of DHS, arguing that the policies required them to violate their legal duties to enforce immigration law.252

But even if the inability of the Morton Memos to significantly (and quickly) reshape the exercise of prosecutorial discretion comes as no surprise, it was quite consequential. The limitations of the approach bolstered forces—both inside and outside the agency—that sought to draw critical attention to the Administration’s deportation policies generally. Many advocates initially regarded the memos as a strong promise of protection, and their disappointment with enforcement practices in the wake of the memos helped create the political conditions that ultimately persuaded the Administration of the need to centralize enforcement judgments in order to better insure protection of status violators.253 One way of understanding the Obama relief initiatives, then, is as a determination that the Administration could no longer wait for the indirect guidance of the memos to take root and shift the culture of the agency. Political leadership thus turned to a more decisive and reliable approach to insulate mere status offenders from law enforcement.

For critics of the Obama relief initiative, this influence of politics on the formulation and timing of DACA underscores the argument that the President engaged in impermissible policymaking. We contend, however, that the inter-weaving of political and institutional incentives for administrative reorganization is to be expected generally and can often be constructive. The move to a more rule-bound and centralized regime provided the rule-of-law benefits associated with promoting consistency in official decision making, amplifying political control and, most importantly, instituting accountability over the enforcement power.

On top of that, we are aware of no evidence that the pre-DACA regime of prosecutorial discretion guided by informal memoranda was serving the salutary function typically associated with discretion—promoting fine-grained judgments involving individual equities about when the initiation of removal proceedings was warranted. In the case of the Obama initiatives, then, the benefits of centralization and the rule-like inquiry it entailed came without the costs typically associated with a move away from discretion. The fact that the same initiatives advanced the political goals of the Administration, or may have been timed in response to political pressures, or the fact that the Administration could have been more patient with the Morton Memos’ approach, are not reasons to declare them suspicious, much less unconstitutional. Instead, the evolution of the Morton Memos into the Obama relief initiatives underscores our claims in Part II about the potential value of a two-principals model of enforcement.

The trajectory of the Morton Memos also explains what we believe to be the most salient features of Obama’s relief initiatives for the purposes of evaluating their legality. DACA made two interrelated institutional changes to the prior regime of discretion. In addition to the turn to rules that we discussed in the previous Part—a move that facilitated oversight of the bureaucracy and constrained lower-level decision makers—DACA changed the decision makers themselves. Not only did the Obama initiative locate the bulk of discretion in the hands of DHS leadership, it took the process of individual adjudication out of the hands of line-level ICE agents and the enforcement arm of the immigration bureaucracy and handed it over to personnel in USCIS, the arm of DHS responsible for conferring immigration benefits.254

This shift to USCIS might seem to have further fractured enforcement responsibilities across the bureaucracy, since ICE remains responsible for enforcement generally. In fact, however, it had the effect of further centralizing discretion. USCIS decided that all DACA applications would be processed in one of four major service centers, rather than dispersing them to eighty-seven field offices around the nation.255 Moreover, the agency required that nearly every discretionary denial recommended by a service center be reviewed by decision makers at USCIS headquarters in northern Virginia.256

Even more important, the shift from ICE to USCIS amounted to DHS leadership, presumably in consultation with the White House, selecting the agency component more likely to share the views of the President and agency leadership and therefore more amenable to oversight in its administration of the program.257 As the benefits-granting component of DHS, USCIS is more likely than ICE to be institutionally predisposed to viewing immigrants as claimants with potential entitlements; ICE is more likely to see them as lawbreakers. This change in decision makers, combined with the adoption of rules, thus facilitated the centralization of prosecutorial discretion decisions within DHS, making it possible to overcome the Department’s seeming inability to supervise the exercise of discretion under the Morton Memos regime.

Seen through this lens, the President’s relief initiatives form part of a broader trend in recent years toward the centralization and reorganization of immigration enforcement authority. We documented some of this movement in The President and Immigration Law, which showed how a series of statutory changes shifted discretionary authority that had previously been held by individual immigration judges within the Department of Justice to DHS.258 Other data points in this trend include the rollout of Secure Communities, along with the scaling back of section 287(g) agreements that sometimes gave considerable discretionary immigration authority to local officials.259 To be sure, the pattern of centralization has been complicated. It would be a mistake to suggest that significant authority over the shape of immigration law no longer exists outside executive leadership, or that such diffusion of authority could ever be extinguished, since immigration enforcement depends on public and private institutions beyond the federal bureaucracy.260 Nonetheless, these recent developments show that DACA and DAPA, far from being anomalous, reflect the latest significant moves in the ongoing reorganization and centralization of the immigration bureaucracy.

This series of recent institutional changes has helped constrain and control the use of the enforcement power in an immigration regime that today gives the Executive capacious authority. The Obama relief initiatives promise to do the same, if and when the Administration can fully implement them. The tradeoffs between rules and standards and centralization and diffusion may be intractable in the abstract, but in this context they point clearly in one direction.

2. Constitutionalized Decentralization

Our account of the institutional dynamics leading up to the Obama relief initiatives also has the benefit of highlighting a strange idea implicit in the conception of the separation of powers advanced by the President’s critics. On the one hand, these critics complain about the failure of the Executive Branch to serve as Congress’s faithful agent. They worry about the lack of sufficiently strong checks on principal-agent problems that arise across the branches. But they would address this problem by constitutionally prohibiting the President from attempting to ameliorate principal-agent problems within the Executive Branch, arguing that the President cannot take a centralizing step to ensure that the priorities reflected in immigration enforcement match his agenda, instead of being the product of tens of thousands of line-level agents within the immigration bureaucracy.261

Perhaps critics who make this claim believe that bureaucratic insulation from either politically appointed agency heads, or from the Executive Office of the President itself, actually furthers congressional control. The perennial argument in administrative law in favor of empowering the civil service sometimes takes this form, emphasizing that these employees are less likely than a political appointee or the President himself to ignore the wishes of Congress, or to be motivated by aggrandizement of the President and his political party.262 This idea also sometimes animates arguments for independent agencies.263

To the extent the arguments in favor of insulating low-level bureaucrats intend to promote Congress’s substantive enforcement priorities, we are deeply skeptical of their purchase in this context. As we explained in Part II, the work of the Congresses that enacted the various provisions that have constructed the Executive’s domain over immigration enforcement do not embody any coherent enforcement priorities. Insulating low-level bureaucrats from the President in this setting, therefore, will not facilitate their compliance with congressional priorities. It will simply enable them to freely pursue their own agendas.

More importantly, when theorists argue that insulating bureaucrats from the President may empower Congress, they do so primarily to defend the claim that the Constitution permits Congress to legislate such insulation by adopting new organizational structures for administrative entities, such as staggered-term commissions, or restrictions on the President’s removal power.264 Here, in sharp contrast, the constitutional critique of the President’s relief initiatives amounts to a claim that the Constitution requires this sort of insulation—even if neither Congress nor the President prefers it. This claim amounts to radical constitutional theory. On this account, the Constitution imposes stringent dictates on the internal organization of the Executive Branch, precluding even the modest effort to discipline the exercise of prosecutorial discretion within an agency by subjecting its exercise to somewhat more rule-like criteria. Even if one thought that the idea of empowering low-level bureaucrats to resist supervision might be attractive in certain situations, constitutionalizing those views in a way that prohibits other organizational judgments by either Congress or the President would be a mistake.

In light of the foregoing analysis, the scope of our two-principals account in Part II comes into further relief. Our concept of the Executive as a second principal necessarily entails permitting politically accountable leadership to exert supervisory authority over line-level agents. But our account certainly does not require disabling line-level officials. After all, the informational advantages the Executive possesses—the learning that comes through enforcement and that should inform ongoing policymaking—enter the system first and foremost through the work of line-level agents and bureaucrats and their operations in the regulatory field. The Executive’s capacity to act as a second principal depends on diffusion in this sense—on the actual officials who produce regulatory reality and see the operation of the law in practice. But what the Obama relief initiatives seek to do is to channel and control the information flow from the field to the center, and our claim in this Part has been that such control should not be constitutionally prohibited (indeed, it might sometimes be constitutionally required). As in the criminal justice system, each individual immigration prosecutor or law enforcement official possesses small-scale policymaking power. When structured with the sorts of rule-of-law values we explore in this Part, the power that this gives to the Executive as a whole should be understood as both legitimate and productive.

C. Transparency and the Rule of Law

As we noted at the outset of this Part, the Obama relief policies implicate a third tradeoff—between transparency and secrecy. One might worry that increased transparency pits two laudable rule-of-law values against each other. On the one hand, through their transparency about how enforcement judgments will be exercised, the initiatives secure greater consistency and predictability in the exercise of discretion, reducing the extent to which decisions about who will be deported appear arbitrary or random.265 Curbing inconsistency and arbitrariness in the exercise of government power is commonly defended as a boon for the rule of law.266 But clarity can come at a cost. Critics of the Obama initiatives worry that the very predictability of enforcement—or, more precisely, the predictability of who will be protected from enforcement—will undercut compliance with the INA and reduce the deterrent effect of the law, thereby threatening the rule of law.267

On this account, any effort to rationalize the exercise of prosecutorial discretion will necessitate a tragic choice between rule-of-law values: the value of making enforcement predictable on the one hand, and the value of legal compliance on the other.268 This claim applies far more broadly than to the President’s immigration actions: it amounts to an argument that no enforcement regime can simultaneously maximize both fairness and legal compliance. For the President’s relief initiatives in particular, the argument suggests that critics would favor more emphasis on legal compliance, while supporters would be happy to trade reduced compliance for increased fairness.

This account poses an interesting dilemma, but we believe the choice to be illusory, not tragic, in our particular immigration setting. To make this case, we first show that this “tragic choice” logic rests on fuzzy thinking about deterrence. Second, and more important and controversial, we explain why concern about legal compliance rests on an incomplete understanding of modern American immigration law.

1. The Logic of Deterrence

There is an entirely mundane reason to doubt that DACA or DAPA will undermine the deterrent value of the law: both of the President’s initiatives apply retrospectively. They grant relief only to past immigration violators, not to future ones. Because eligibility for DACA or DAPA requires at least five years’ continuous residence in the country, the immigration violations of most eligible immigrants will have occurred more than five years ago, years before the President announced the policies. In fact, given estimates that the majority of the unauthorized population has been living in the United States for at least ten years,269 DAPA relief may well end up helping mostly those whose immigration violations are more than a decade old.

The relief initiatives’ retrospectivity makes the programs very different from conventional policies of prosecutorial discretion, which should reduce concern about their impact on law’s deterrent value. Most prosecutorial discretion policies set out criteria that guide decisions about which future offenders will be prosecuted for certain legal violations. The Department of Justice’s recent guidance about low-level marijuana offenses provides a run-of-the-mill example. The policy tells legal subjects—you and me—that a U.S. Attorney’s office is extremely unlikely to prosecute us for possessing a few marijuana cigarettes. This announcement might well affect our decisions about whether to abide by the federal prohibition on marijuana possession laid out in 21 U.S.C. § 841. Indeed, the Department of Justice almost certainly wanted to shape people’s decision making in this way, freeing states like Colorado and Washington to experiment with regimes that opted to decriminalize marijuana, regulating its distribution and use through noncriminal means.270

DAPA and DACA, by contrast, should not produce such a behavioral response. They are best conceptualized as exogenous legal shocks that will affect the size of the existing pool of unauthorized migrants without altering the legal regime that applies to future immigration violators (or to those who do not receive relief under the program). This feature enables the policies to drive a wedge between the values of deterrence and predictability; DHS can make enforcement more predictable without undermining future legal compliance. In fact, these programs might even improve future compliance with immigration law. With fewer resources devoted to identifying and deporting those who receive relief, DHS can redirect enforcement resources to increase the likelihood that future immigration law violators will be caught (thus increasing deterrence). DHS directives released alongside the relief policies in November 2014 reveal this strategy: DHS will devote more resources to the border, as well as to identifying those who become deportable by virtue of convictions for certain crimes.271

Critics who claim that DACA and DAPA will undermine the rule of law might resist this account of legal compliance in two ways. First, they might reject the notion that the programs grant relief for past immigration violations, arguing instead that immigration violations, by their very nature, constitute ongoing offenses. Every day an unauthorized migrant spends in the country amounts to a legal violation, and only by leaving the country can she put an end to her lawbreaking. Thus, the argument goes, even if DACA and DAPA do not affect the behavior of other migrants, they undermine legal compliance by the very migrants granted relief under the programs. Absent the programs, some number of these migrants might self-deport to bring themselves into compliance with the law, but in the presence of the program no one granted relief will do so.

For a number of reasons, we reject this conceptualization of immigration violations. We need not delve into the complexity of what it means for conduct to constitute an ongoing legal violation, because the argument contains a more basic flaw: by focusing on the individuals eligible for relief, it ignores the effect the relief programs are likely to have on legal compliance more generally. DAPA might make compliance via self-deportation less likely for those eligible for relief, but it enhances the likelihood of compliance by those not eligible for relief by raising the risk that the latter will be deported. In other words, even on an account that treats immigration status violations as ongoing violations, neither DACA nor DAPA undermines the overall level of deterrence;272 they simply shift the brunt of deterrence from one population to another.273

Second, critics might challenge the sharp analytic distinction we have made between retrospective and prospective relief. The distinction depends on the credibility of the government’s commitment not to extend relief in the future to those who violate immigration law after DACA or DAPA’s announcement. Commitment presents a perennial problem for amnesties and other forms of relief for past legal violations. If people begin to believe that similar relief will be granted again in the future, they might break the law in anticipation of a future grant of amnesty. If DACA grants relief to some immigrants who arrived as children, perhaps more children will attempt to enter the country in hopes of some future DACA-like program. Some commentators made precisely this claim last summer, when apprehensions of unaccompanied minors at the Texas border suddenly skyrocketed. And similar claims have been made about past immigration relief programs, most notably the legalization program that was a part of the 1986 Immigration Reform and Control Act (IRCA), which granted green cards to nearly 2.7 million unauthorized immigrants, more than half of the then-existing unauthorized population.274

Evidence from IRCA’s legalization gives us reason to doubt strong claims that immigrants will make different migration decisions simply because of a slight increase in the highly uncertain prospect of some unspecified relief years down the road. Economists studying IRCA’s effect on unauthorized migration found no evidence of an increase in the flow of migrants hoping for a future program.275 Moreover, in the present political environment, it seems equally possible that DACA and DAPA will serve as anchors that lower, rather than raise, the probability of more expansive relief down the road. Prior to these programs, a legislative legalization program formed a central component of the immigration reform bills passed by the Senate in 2006 and 2013.276 Most commentators think a legislative amnesty will eventually come to pass, in part because of the sheer magnitude of the unauthorized population. The open questions include when that legislative relief will ultimately come, and how far it will extend. It may be that the Obama relief initiatives will shape future legalization by limiting any program to those who have benefited from DACA and DAPA. But we think any such predictions would be foolish. What effect, if any, DACA and DAPA will have on future reform turns on political dynamics so complex that even those who are central players in the drama, right down to the congressional leadership and President Obama himself, seem to have little idea how it will play out. Given this extreme uncertainty, we think it implausible that DACA or DAPA’s influence on the prospect of relief for future immigration violators will meaningfully affect the migration decisions of prospective migrants contemplating coming to America, either by encouraging or discouraging them.

2. Underenforcement and Ex Post Screening

The retrospective nature of DACA and DAPA make them special from the perspective of deterrence. But even if the compliance critique fails at the retail level, the turn to transparent, predictable, and rule-bound enforcement strategies could still raise compliance concerns, thus implicating the tragic choice between rule-of-law values. For this reason, it is important to explain why the choice between transparency and compliance is illusory in light of the structure of modern American immigration law—a structure other regulatory contexts may share and that in the immigration setting transcends the Obama relief initiatives.

The argument against transparency in the exercise of prosecutorial discretion conflates two very different types of enforcement settings. In one, the legal system makes clear that the desired level of some conduct is zero. In such a setting, there ideally would be perfect compliance with the legal prohibition. But compliance might fall short of that, and the government may lack the resources to punish every single violator. In that context, secrecy about enforcement strategy can be valuable, and transparency can threaten legal compliance. The threat to compliance will be most palpable when an offense can be committed in multiple ways, or in multiple places. In such situations, publicizing what law enforcement officials will be looking for, or where they will be looking, can make it easier for would-be violators to avoid having their legal violations detected. Secrecy about law enforcement tactics, even to the extent of randomizing those tactics, can often increase compliance, both by raising the risk of detection and by creating more uncertainty about the level of risk.277 For these reasons, the IRS works hard to keep its audit algorithms secret, state highway patrols do not disclose the locations of speed traps, Customs and Border Patrol frequently moves the roving checkpoints it uses along the southern border, and the CIA and other intelligence agencies resist disclosure of their surveillance tactics.278

But for reasons we already have explained in our development of the concept of de facto delegation, this account does not describe the immigration enforcement setting. In the world of de facto delegation, where exceptionally broad grounds of deportability and long-standing acceptance of high levels of unauthorized immigration have made half of all noncitizens living in the United States formally deportable,279 a perfect world is not a world of perfect enforcement.280 This is so not simply because the government has limited resources—the claim around which most defenses of the President’s actions have turned.281 Nor is it true only because the social costs of deporting all unauthorized migrants would be enormous (though they would be). Rather, it is because immigration law’s formal prohibitions do not accurately reflect the structure of the immigrant screening system. The INA does not establish fully the contours of culpability under immigration law. Instead, our system of de facto delegation requires the Executive to assume responsibility for sorting “deserving” violators out from non-deserving ones—for making the sorts of policy judgments we defend in Part II.282 The screening decisions of executive branch officials who decide which formally deportable noncitizens deserve deportation thus shape the true limits of the law.

In our view, the screening the law requires will be most fair if the Executive conducts screening universally and makes its rules transparent. Universality makes it possible to treat like cases alike, and transparency provides notice to immigrants about the actual structure of the screening system (as opposed to the formal structure of that system). We need not worry that transparency will undermine compliance with the formal screening rules of the INA, because the rules are not meant to be followed to the letter.283 Instead, public behavior would track the screening system established by executive discretion, instead of the (overbroad) criteria articulated in the statute itself. The Obama relief initiatives aid this objective through transparency.

Other areas of the law share this structure whereby the legally desirable level of an action is not zero, and according to which overbroad laws have the effect of empowering public officials to screen for culpable violators from among a larger pool of formal violators.284 Consider criminal law. Bill Stuntz has argued that it has just such a structure: legislators draw substantive rules of criminal liability to sweep in far more persons than are deserving of punishment, effectively delegating power to prosecutors and police to sift through the universe of violators, screening for those who are worthy of punishment.285 In such a system, a rationalized, predictable, and transparent system of prosecutorial discretion promotes the equal treatment of similarly situated defendants without undermining compliance with the “true” criminal law norms that the system seeks to enforce. If we think that low-level marijuana possession by a person with no other criminal record should not be punished, then we should hope that judgment applies to all cases, not to a random subset of violators. And we would not worry that publicizing the policy would undermine legal compliance, because the very point of the exercise of discretion would be to communicate that possession under those circumstances does not deserve punishment.

We recognize, of course, that critics of the Obama relief initiatives, in Congress and in the commentariat, might dispute our account of the modern structure of immigration law—we view this claim as our most controversial and contestable one. The self-deportation strategy advanced by opponents of illegal immigration reflects a total compliance worldview leavened by the reality of limited enforcement resources. Measures that induce self-deportation become necessary in a world of limited resources precisely in order to achieve the total compliance goals of the system. But the core of our argument here is that these critics operate with the wrong understanding of the nature of underenforcement in immigration law—one that ignores the actual history and practice of immigration law and puts them on the wrong side of the Supreme Court in Arizona. Supporters of DACA and DAPA have reinforced this confusion about underenforcement’s significance; their persistent focus on resource constraints as both a necessary (and sufficient?) condition for the exercise of broad deportation discretion promotes the misconception that, in the absence of resource constraints, the Executive’s duty would be to enforce against all violators (save perhaps for the occasional exercise of humanitarian discretion). Budget constraints are ubiquitous. But the structural delegation of screening authority present in immigration law—and some other areas like criminal law—is not.

D. Benefits Versus Penalties

In choosing rules over standards, centralization over diffusion, and transparency over secrecy, the Obama relief initiatives highlight some key debates over how best to organize the modern bureaucracy. In their particular context, these choices not only make acceptable structural tradeoffs, they actually promote rule-of -law values and accountable and constrained executive power. Far from striking a blow on behalf of the imperial presidency, the relief initiatives represent responsible uses of the enforcement power, even as they advance the President’s political agenda. Any unseemliness of the latter does not detract from the value of the former.

Of course, in offering this conceptual alternative to the faithful-agent framework we reject in Part II, the question becomes what sorts of efforts to structure the enforcement power would give rise to constitutional concerns. Can we imagine any limits on the way in which the Executive institutionalizes its discretion or makes the inevitable tradeoffs among rule-of-law values? We turn to these most difficult questions in Part IV below.

But before developing our account of limiting principles, we pause to address one last, frequently heard claim about the way in which DACA and DAPA institutionalize relief loosely related to our arguments in this Part. The claim is as follows: by choosing rule-like criteria for relief, by centralizing control over the application of those criteria, and by establishing a transparent application process, the Obama Administration has conferred on DACA and DAPA recipients a promise of nonenforcement that differs in kind from a mere guideline that de-prioritizes removal on the basis of certain characteristics. This claim—that there is an important substantive difference between the form of relief provided by these programs and ordinary prosecutorial discretion—rests at the heart of the Texas district court’s decision to enjoin DAPA and the expansion of DACA. In the words of the district court, the programs provide the benefit of “three years of immunity from [the] law.”286 At another point the decision describes the benefit as the conferral of “legal presence status.”287 In other words, the relief initiatives confer new legal benefits on recipients rather than simply declining to prosecute them.288 Providing these legal benefits, in the critics’ view, departs from any coherent understanding of prosecutorial discretion and exceeds the authority of executive branch officials.289

This argument appears to turn on the legal consequences for immigrants (or other regulatory subjects) of institutionalizing enforcement discretion in particular ways. Its subtext might also be that the relief policies’ beneficiaries do not deserve an open and notorious relief from prosecution, even if they might escape prosecution as the result of case-by-case choices. To the extent this argument depends on the claim that DAPA and DACA recipients receive a legally binding promise that they will not be deported for three years,290 the argument is mistaken. The Administration has made no such promise.291 As a formal matter, the promise entailed by the relief initiatives is no more a legal entitlement than the promise that would have accompanied an “individualized” grant of deferred action in the years prior to DACA. While grants of deferred action have in the past often been made for multi-year terms, and sometimes for indefinite periods, these grants have never been understood as legal entitlements. As with any other instance of prosecutorial discretion, executive branch officials remain free to reverse course and charge a person previously granted deferred action. Of course, in practice, such reversals were rare historically, and we agree that Jeh Johnson almost certainly will not reverse course next month and order ICE agents to initiate removal proceedings against those who have been granted relief under DACA. As a matter of law, however, this reality does not convert a permissible nonenforcement decision into an impermissible grant of a legal benefit. If it did, grants of prosecutorial discretion by criminal prosecutors would be widely unlawful, as it is often clear in the case of such grants that criminal justice officials have no intention of pursuing changes in the future for the conduct at issue.

We could try to rescue the “benefits” critique by recasting it in a functional rather than formal light (though critics themselves have cast it in a formalist way). We could say that the core of the benefits claim is not that the deferred action granted amounts to a legal entitlement, but instead that, as a practical matter, DACA and DAPA provide relief that will be more durable than ordinary decisions to defer or forgo enforcement. This durability, critics might argue, applies not only within the current Administration, but also beyond it, as it would be very costly politically for a new President, regardless of party, to begin removing in significant numbers the beneficiaries of DACA and DAPA who will have developed strong reliance interests. On this account, the practical entrenchment likely to arise from the President’s actions separates constitutional exercises of prosecutorial discretion from unconstitutional ones.

These programs may in fact be durable, but that would not distinguish DACA and DAPA from earlier exercises of immigration enforcement discretion. Whereas President Obama’s initiatives promise only three years of relief, noncitizens granted deferred action in the past have in many cases been granted indefinite relief—relief that, in practice, often lasted for periods of more than three years and sometimes spanned more than one administration.292 And, as a theoretical matter, it seems equally plausible that the institutionalization of relief in high-level agency decisions will ultimately undermine the durability of relief over time. A single decision of a future administration could reverse the nonenforcement decisions with respect to millions of noncitizens.293 Nothing of this sort would have been possible under the regime that preceded DACA and DAPA. The literature analyzing centralization appreciates this possibility, emphasizing the fact that agencies become more responsive and policies less entrenched—not the other way around—as decision making becomes centralized in high-level officials who are less subject to the slow-to-change culture of an institution.294

As a descriptive matter, therefore, we are skeptical of the claim that DACA and DAPA entrench nonenforcement promises to a greater degree than other forms of enforcement discretion. More importantly, we see no reason why the practical durability of the policy should be constitutionally relevant: there is no plausible constitutional theory of which we are aware under which a promise not to prosecute becomes unconstitutional whenever that promise might be politically durable.

IV. whither limiting principles?

As we showed in Part II, principles to limit the exercise of enforcement discretion based on substantive factors grounded in congressional priorities will be elusive across many statutory schemes, especially as those schemes become more reticulated over time. This reality underscores that the “parade of horribles” invoked by many critics of the Obama relief initiatives is not so much a series of hypotheticals about a dystopian post-DACA future as a simple description of the actual history of many regulatory arenas. Ronald Reagan dramatically scaled back environmental and antitrust enforcement. George W. Bush transformed the enforcement culture of the Environmental Protection Agency, altering threshold regulatory requirements for new source review under the Clean Air Act and abandoning investigations that had been commenced under the pre-existing legal regime.295 His Administration also nearly shuttered the Department of Justice office that pursues structural reform of police misconduct,296 and the Voting Section of his Civil Rights Division did not file a single lawsuit alleging discrimination against minority voters for several consecutive years.297 Indeed, Eric Holder came into office as Attorney General promising to restore the stature and power of the Division. The first chair of the Securities and Exchange Commission appointed by President Obama immediately ended a Bush Administration pilot program that required enforcement staff to seek permission from the Commission before negotiating a civil monetary penalty against a public company—a policy that had delayed the enforcement process.298

These shifts in enforcement policy, while perhaps dramatic, simply reflect the consequences of politics and presidential elections. Though it might be magnanimous and perhaps even judicious for the President to tread lightly when making enforcement judgments, in order to set good precedents for his or her successors who will have distinct ideological preferences, it may also be foolishly high-minded. Exercising the enforcement power necessitates making value judgments, particularly in circumstances where the laws being enforced either reflect a variety of compromises made by the enacting Congress or leave the Executive Branch with wide-ranging discretion in implementation. And as we explained in Part II, there are reasons to value rather than lament the scope of judgment the enforcement power entails—reasons that parallel common defenses of Congress’s delegation of vast rulemaking and, therefore, policymaking power.

Yet the inevitability and desirability of presidential priority setting does not mean that the President can exercise the enforcement power without constraint. In the absence of express and specific statutory direction as to how to prioritize enforcement, we would still want the Executive to abide by constitutional norms that expect the exercise of coercive power to be well supervised and accountable. But these norms will be difficult to translate into limiting principles based on substantive priorities. Instead, we should devise limiting principles that operate as forms of constraint on the way the Executive institutionalizes enforcement priorities. Understanding the choices at stake in the institutionalization of prosecutorial discretion, as we have presented them, can help us begin to identify when the form discretion takes might raise red flags as a matter of constitutional law or culture.

To say that such limiting principles can be identified is not to say that they will always be judicially manageable. Some might be easily embodied in doctrine, but others will be difficult to formulate into clear legal rules. For constitutional scholars who believe the very definition of a “limiting principle” is that it must be amenable to enforcement by an Article III judge, our view will be unsatisfying. But we believe that tangling up debates about the existence of limiting principles with longstanding disagreements about the extent to which constitutional norms must be judicially enforceable stymies genuine inquiry into how best to conform the enforcement power to conceptions of constrained and accountable government.

In what follows, we begin by identifying the sorts of limiting principles that might apply to the exercise of the enforcement power. But the inquiry into constraint need not end there. Even if broad use of the enforcement power according to executive judgment will almost always be constitutional, and even if it will be difficult to conclude that any given institutionalization of discretion crosses a constitutional line, we can still evaluate on the merits the Executive’s decisions regarding how to structure its power, as well as any arrangement the political branches might have reached through the political process. In other words, it would be a mistake to limit the analysis of the enforcement power to the constitutional register. We therefore close by considering what might be deficient with the Obama relief initiatives and the state of affairs that produced them, as a matter of more general legal and political theory.

A. Current Constraining Principles

Constraints on the President’s authority to determine how to institutionalize discretion within the Executive Branch could arise from a number of sources, many of which have nothing to do with constitutional limiting principles. First, as we discussed in Parts II and III, Congress possesses a variety of tools to constrain discretion, though some will be easier to employ than others. Congress certainly could place constraints on the substance of discretionary choices. As we noted in Part II when describing how enforcement judgments help construct the regulatory domain, statutes themselves place limits on the bases of enforcement. The President could not, for example, declare as a ground of removal an offense Congress has not listed in the Code. Congress also could draft statutory prohibitions against certain exercises of discretion,299 or use its appropriations power to shape enforcement choices, though we have discussed the limits of the latter300 and are skeptical that the former approach would be a good one to adopt with frequency, given the affirmative value of executive policymaking through enforcement articulated in Part II.

Congress could also use institutional design and oversight in tandem to constrain discretion. It could limit the President’s capacity to supervise officials’ discretionary judgments; for example, Congress could require that particular low-level adjudicators, such as administrative law judges, make certain decisions without interference from agency leadership. The development of statutory protections for civil service employees reflects this sort of judgment and promotes the professionalization of government employment by insulating employees from the pressure of patronage politics.301 Congress might also seek to diffuse power to make agencies more responsive to individual members of Congress and oversight committees through day-to-day, informal interactions. Strategies of this sort might be difficult to launch, given the fraught legislative process. But Congress could, at the very least, use hearings and appeals to the press to advance its point of view in a way that might create political pressure on the Executive to change its behavior.302

But our primary interest here is not in the myriad political and institutional forces that constrain the President as a matter of fact. We focus instead on the ways in which the Constitution itself will sometimes directly constrain how executive officials institutionalize enforcement discretion.303 Constraints can crop up along any of the three dimensions we identified in Part III. Take the choice between rules, standards, and unfettered discretion. Sometimes the Constitution requires government-by-rules; other times it prohibits their use. First Amendment doctrine, for example, sometimes requires that executive branch officials make decisions pursuant to rules clearly specified ex ante, in order to constrain discretionary judgments about decisions to issue permits and the like—judgments that present a high risk of impermissible discrimination.304 In contrast, the Due Process Clause sometimes prohibits the use of highly structured decision-making rules, requiring that an adjudicator retain discretion to take account of any evidence or arguments offered by an individual claimant.305

It could also be the case that particular choices between rules and standards result in impermissible tradeoffs among rule-of-law values. These limits will be hard to characterize as hard-and-fast constitutional requirements and may be more appropriately characterized as features of a theory of constitutionalism. But some criticisms aimed at particular institutional design choices might sound in constitutional concerns, even though a court would be unlikely to strike them down. If, for example, the President’s initiative permitted any non-citizen to apply for relief but then left the judgment entirely to the whims of adjudicators, the loss of supervision resulting from this standards-based approach might, depending on the practice of the adjudicators, amount to a loss that cannot be offset by the benefits of individualized, fine-grained decision making. We could describe this tradeoff as a constitutionally irresponsible choice, even if it would be difficult to describe it as unconstitutional such that a court could strike it down.

We could similarly analyze the choice to centralize or diffuse power. Centralization designed to facilitate preferential treatment of the President’s cronies could present a constitutionally problematic form of institutionalization. A centralization initiative designed to serve partisan goals might ordinarily be unexceptional, but it could present a source of constitutional concern in a context in which custom demands independence. The centralized and politicized hiring of immigration judges in the Ashcroft Department of Justice, for instance, contravened civil service regulations and customs surrounding the hiring of officials otherwise removable by the Attorney General.306

The tradeoff between transparency and secrecy could also be governed by principles or presumptions that would constrain presidential choices about how to institutionalize discretion. If the structure and integrity of a given enforcement domain depend on self-compliance by regulated parties, then the costs of transparency regarding enforcement priorities might be so high as to be deemed impermissible. In the tax arena, for example, the enforcement machinery depends heavily on the in terrorem effect of legal regulation; the system’s goal is maximal compliance with the law, and so it is crucial for the IRS to keep its enforcement priorities hidden from view, in order to maintain public incentives for widespread compliance. If the IRS were to announce that a portion of the Code simply would not be enforced because of the Service’s scarce resources, legal compliance would be undermined for little benefit and for arguably questionable reasons (perhaps to curry favor with a tax-skeptical public).307 In such a circumstance, the charge of “abdication” would have bite, because enforcement policy would reflect less an attempt to better structure the location of discretion than an effort to undermine the law itself.

Again, it bears emphasizing that these limits we imagine may not be judicially enforceable, and we have not conceptualized them as doctrinal principles. Defining and then mobilizing limits as doctrinal rules would require that we formulate a comprehensive theory of executive power that would inevitably be unmoored from constitutional text and have an ambiguous relationship to constitutional practice. As we emphasized throughout Part III, we are reluctant to constitutionalize the internal structures of the Executive Branch, given the complex tradeoffs among rule-of-law values that must be made when an administration seeks to organize and wield its power.

The difficulty of devising rules of constraint from a set of general principles does help explain the appeal of OLC’s substantive approach, or of the prophylactic, bright-line framings of critics who have tried to draw the conceptual distinctions we reject between individual and categorical judgments, or between non-prosecution and the granting of benefits. But as we hope we have shown in Parts II and III, the lawyerly appeal of these frameworks cannot save them as descriptive or normative accounts of the scope of executive authority. The theoretical framework we have offered as an alternative may not result in clear lines around the enforcement power, but it does provide a vocabulary well suited to legal conversations about constitutional norms with which we (scholars, lawmakers, internal executive watchdogs) can assess the merits of executive branch practice.

B. Future Discipline

The fact that the President acted lawfully and reasonably when announcing DACA and DAPA does not mean that our current institutional arrangements are ideal. The two-principals model of immigration policymaking is the one we have inherited, and we showed in Part II that important benefits flow from this model. But we do not maintain that the current regulatory structure is optimal, and current interbranch dynamics do present downsides. We therefore close by considering various ways of improving the constitutional and theoretical grounding of the President’s enforcement power over immigration law.308

1. Meta: The Process of Institutionalization

Let’s start with the development of the Obama relief initiatives themselves. As we have argued, they represent transparency-enhancing and regularizing improvements on the status quo that preceded them—a world in which the Morton Memos and other guidance documents provided far too little information about how the Executive actually exercised its significant screening authority, and far too little supervision of line-level officials. Yet while the Obama initiatives themselves are transparent, the process that produced them was opaque. Mobilized interest groups may well have informed the ultimate shape of the initiatives, but there were no formal avenues for public input into the policymaking process. The policies were drafted and vetted only within the Executive Branch and its self-defined spheres of influence.

One means of addressing this flaw might be through the Administrative Procedure Act (APA). The notion that DACA and DAPA count as legislative rules subject to the APA’s notice-and-comment requirement, rather than as general statements of policy, led the district court in Texas to declare that the Administration had violated the APA.309 That decision remains pending before the Fifth Circuit as we write, though the court of appeals appears poised to affirm the district court, teeing up the APA issue, if not the underlying constitutional question, for the Supreme Court.310 In denying the government’s request for a stay of the district court’s injunction,311 and during oral arguments on the merits,312 two different Fifth Circuit panels (with overlapping membership) appeared skeptical of the federal government’s case (though one judge subjected Texas’s claims to withering criticism, as well).

As a matter of existing administrative law doctrine, we are skeptical of this outcome (though, to be frank, the case law attempting to sort legislative from nonlegislative rules is a mess).313But as a matter of principle, the claim that the Obama relief policies would have benefited from more procedural formality, or transparent public input, should be taken seriously. The Administration’s defenders have tried valiantly to frame the relief initiatives as entirely ordinary. But this framing obscures the innovative nature of DAPA and DACA that we described in Part III—a characteristic that, when combined with the scale of the programs, marks them as significant acts of policymaking by the Executive, much as the historical precedents the Administration cites were. DACA and DAPA may not confer formal legal status, but they enable millions of unauthorized immigrants to live and work free of the fear of removal, further entrenching their interests in remaining in the United States. The Obama relief initiatives thus significantly increase the political and humanitarian costs of removing this population at some future point.

Significant policymaking of this sort would have benefitted from public scrutiny and involvement.314 Open debate could have informed the Administration’s judgments on questions such as the relevant criteria, the scale of the program, and the range of “benefits” that should flow from the granting of relief, which may have improved the design of the program and certainly would have enhanced the legitimacy of the President’s initiatives.315 Such public deliberation also would have facilitated a central goal underlying the APA of increasing the accountability of the policymaking process while also bolstering public confidence in the measures ultimately adopted.

These potential benefits do not mean, however, that courts should overhaul existing legislative rules jurisprudence in order to force programs like DACA and DAPA to go through notice-and-comment rulemaking, as the litigants in the Fifth Circuit aim to do. The protracted multi-year nature of modern rulemaking would likely have made such a process unworkable from the Administration’s point of view,316 especially to the extent both DACA and DAPA were timed to maximize the political payoff of the announcements.317 Perhaps interest group meetings in the White House were all that could reasonably have been expected by way of public input into an initiative of this sort. But that points to a more general dilemma posed by modern administrative law, where the choice too often is between a cumbersome notice-and-comment regime and minimal procedural formality.318

Faced with this choice, the incentives of executive branch officials have predictably produced more and more informality in the sphere of administrative action. In a way, it would have been shocking had the development of DACA and DAPA unfolded any differently. Nonetheless, the absence of manageable channels for public input highlights the basic failure of administrative law to address the central role that enforcement discretion plays in important regulatory arenas. Some form of public input into the development of enforcement priorities with more formality than private meetings convened by the Executive and less than notice-and-comment rulemaking would be a valuable contribution to regulatory spheres in which the enforcement power drives application of the law, as well as the politics and substantive policy of the area.

2. Prosecutorial Discretion in a Second-Best Regulatory Environment

Perhaps the most fundamental problem still in need of a solution stems from the facts on the ground that gave rise to DACA and DAPA in the first place. As we have chronicled here and in previous work, de facto delegation endows the President with asymmetrical screening power, giving him much more power at the back end of the system than the front. While exclusion and deportation are clearly substitute mechanisms for screening migrants, restricting most regulatory innovation to the ex post screening environment leaves us stuck in a second-best regulatory environment.

If the President is to have primary responsibility for the structure of the immigrant screening system, he should be able to determine the optimal mix of ex ante and ex post screening mechanisms. A better-designed system would prevent such a large pool of potentially removable noncitizens from arising in the first place, reducing the need for the coercive power of the state and therefore the sort of policymaking through enforcement that can tend toward the opaque and create the impression, if not the reality, of arbitrary decision making. In other words, the fact that policymaking through enforcement can play a desirable function within a scheme of separated powers does not mean that alternative forms of more transparent policymaking are not preferable.319

We can imagine numerous ways to address this problem of an overly large enforcement realm. Eliminating or narrowing some of the grounds of removability and scaling up border enforcement represent two opposite ends of the political spectrum (and each addresses a different source of de facto delegation). We doubt that either will be especially effective, however, as a means of curbing the enforcement power. Whereas the former will have an effect primarily on the margins (at least as long as unauthorized presence remains a ground of removal), the latter offers a blunt instrument for reducing the deportable pool and only magnifies the unreviewed power of the Executive by focusing enforcement where law enforcement power is at its most robust and judicial review and due process norms are at their weakest. What is more, the pathologies of de facto delegation have not arisen solely from the legal structure of immigration law. As we emphasized earlier, the intersection of this legal structure with powerful social and demographic forces has produced the current state of affairs. For those whose answer to our dilemma of de facto delegation would be to use law to prevent the unauthorized pool from arising in the first place, we suggest that such thinking is likely wishful.

For all of these reasons, we have advocated in the past delegating greater ex ante screening authority to the Executive to enable the government to respond to demographic and labor market factors with sensitivity to their fluctuations, trading explicit delegation for de facto delegation.320 Seen in light of our analysis in this Article of the role of the President as independent policymaker, this option should seem normatively attractive.321 An ex ante process would make room for far more significant public input than an ex post enforcement regime—even one as transparent as DACA and DAPA. It also would channel executive power into less coercive forms than the operation of a law enforcement bureaucracy. We happen to be at a moment in time when net illegal migration appears to have approached zero,322 and so the need for such ex ante authority may be less pressing than it would have been a decade or two ago. But our very point is that the Executive should have substitute tools at its disposal to adapt to circumstances.

This call for delegation ultimately feeds into one final point. The President’s powers over immigration policy remain limited as compared with those of Congress. The President has significant control over our shadow immigration system, but he cannot confer legal status directly on unauthorized immigrants; only Congress can do that through a legalization program. It has become commonplace for defenders of the President’s actions to emphasize this point as a way of underscoring that the President’s actions have remained within his domain.323 Within our framework, however, this point highlights the centrality of Congress to addressing the policy issues raised by the persistence of a large, unauthorized population. In our view, however, Congress has been a poor participant in the debate. In 2013, the Senate passed an astoundingly comprehensive bill that would have launched a legalization program, but that bill has languished. Congress has contributed to the debate precipitated by the Obama relief initiatives largely through symbolic appropriations riders forbidding the President from implementing the initiatives, as well as threats to defund the Department of Homeland Security. These forms of debate serve primarily to escalate political conflict while offering no real hope for policy reform. Perhaps enough members of Congress prefer the pre-DACA state of affairs, such that we can read these symbolic gestures as reflective of a considered policy position. But even if one does not support a legalization program as a matter of policy, the rule-of-law concerns we have identified as emblematic of the pre-DACA world should be cause for legislative debate and action.

Some commentators seem to fear that the President’s actions have compounded congressional policy passivity and partisan grandstanding—that DACA and DAPA have had the effect of disabling Congress, or at least pushing it into an oppositional posture rather than a lawmaking one.324 We are skeptical of the claim that they have prevented Congress from acting, because it seems equally plausible that such actions could spur congressional action. Historically, politicians, advocates, and strategists have offered two different strategies for prompting Congress to fix the immigration laws. Some advocates argue that the President should publicly grant relief to millions (or even declare a moratorium on deportation) in order to highlight the broken nature of the system and prompt Congress to act. Others argue, instead, that the President should do his best to enforce the law to the hilt in order to expose the harshness and futility of the formal rules and thereby create political pressure for legislative change.325 In other words, not even those who are enmeshed in the congressional-executive dynamic seem to agree on how presidential action will affect Congress’s ability and willingness to legislate.

Moreover, there is an additional reason to be skeptical of the claim that executive action reduces the likelihood of congressional action. This worry typically arises in contexts where executive action can serve as a substitute for congressional action. If the Executive takes action on its own to regulate greenhouse gases, for example, some worry that its measures will reduce the pressure for Congress to take steps that will lead to the same outcome—the regulation of greenhouse gases. Or if courts engage in judicial review to evaluate the constitutionality of legislation, then Congress will stop worrying about constitutional questions when it drafts legislation. The crucial difference between these contexts and the present immigration context is that the actions of the Executive and Congress are emphatically not substitutes. President Obama did not legalize five million unauthorized migrants with his relief initiatives, and he lacks the authority to do so. Action by Congress will be required to provide a lawful immigration status to most unauthorized migrants living in the United States. And given that only Congress can confer such status, it is hard to understand why the President’s limited relief programs would somehow eliminate the pressure for future congressional action on immigration reform.

Recent history ultimately suggests that Congress would not have acted on immigration, even if President Obama had not pursued his relief initiatives. Generally speaking, the partisanship reflected in the debate over deferred action dominates the relationship between the branches, making it unlikely that continuation of the pre-DAPA and DACA status quo would have resulted in more meaningful cross-branch debate. But even so, when turning to executive branch policymaking as a viable alternative to congressional stasis, we should not lose sight of what can be lost when the Executive becomes the primary engine of policy—not just the open and transparent decision making more likely to come from a less disciplined but more multi-faceted congressional debate, but also the collaboration between Congress and the Executive that defines any legislative process.

To be clear, we do not mean to suggest that executive policymaking as a general matter is always a second best option to congressional action. We hope that our arguments in Part II, highlighting the dynamic, iterative conception of the separation of powers and the role the enforcement power plays in shaping a regulatory domain over time, shine through as reasons to appreciate executive policymaking within a proper and healthy distribution of powers. But we also believe that, when the President must take bold action to address threats to the rule of law within the domain of enforcement, a more fundamental recalibration of political branch responsibilities may be necessary.

Conclusion

Presidential immigration law is ascendant. The dominant policymaking role long played by the President, combined with the twentieth-century rise of de facto delegation, destabilizes a simple principal-agent model as a way of understanding the separation of powers in immigration regulation. These developments have produced an immigration regime in which the President has significant responsibility for—not just power over—the rules for screening immigrants. The Executive Branch has actually helped construct the screening system over time as it has wielded its enforcement power and decided how to put the INA’s statutory framework into effect—a process that has entailed considerable executive policymaking dynamically related to, but still separate from, congressional policy. The separation-of-powers framework that emerges from this history is thus far from static. De facto delegation has not entailed a simple transfer of power from one branch to another; as we have documented here and in our 2009 work, the structure of modern immigration law has produced an iterative relationship between the branches in which Congress has played an important, if not dominant, role. Without both an awareness of and appreciation for these dynamics, the President’s immigration enforcement power cannot be properly understood—let alone cabined.

The separation-of-powers framework we have uncovered is not just an institutional reality to be bemoaned. We believe there are reasons to accept, and even to endorse, presidential policymaking through enforcement, perhaps especially in the immigration context. Such acceptance does not mean that the President’s (and the agencies’) exercise of discretion should go entirely unchecked. But it does suggest that the constraint in most cases will not come from an inquiry into whether the substantive policy choices embodied in enforcement initiatives such as DACA and DAPA promote congressional priorities. In place of that sort of doomed Take Care Clause inquiry, enforcement policies should be evaluated for whether they make reasonable rule-of-law tradeoffs and thereby advance the general purposes of the constitutional separation of powers—constraining and rendering accountable government power.

Crucially, this rule-of-law inquiry requires that one attend as much to relationships of power within the branches as across them, as a growing body of scholarly work has come to appreciate with respect to the administrative state as a whole. Within this framework, we think it clear that the Obama relief initiatives are lawful. By using rules to centralize discretionary decision making, DACA and DAPA make visible the political and policy choices the Executive Branch has made while enhancing the consistency of the government’s use of its coercive powers. Further, the initiatives tame the faceless prosecutor by imposing politically accountable constraints on the decisions of low-level officials. Critics have argued that the programs are dangerous because they permit the President to replace Congress’s judgments with his own. In reality, however, the relief initiatives have enabled him to discipline the judgments of low-level enforcement officials, bringing order and discipline, along with his own substantive preferences, to an immigration enforcement regime in which the Executive has significant responsibility for the structure of screening. Advocates for the President’s actions who tout the virtues of “individualized” prosecutorial discretion in this setting obscure the institutional reality of how that discretion has operated in immigration law, and they distract attention from the crucial benefits that flow from the way that DACA and DAPA actually centralize and limit discretionary judgments.

Presidential policymaking has always provoked political controversy, at least as much because of deep ideological disagreements over immigration policy as because of the perception it creates of an aggressive or boundless Executive. The fact that today’s particular controversy over the Obama relief initiatives has vivid partisan overtones is not to say that limits on policymaking through enforcement do not exist, even when they do not appear visibly or clearly in the statutes that set the parameters for executive action. But those limits will be excruciatingly difficult to define without disabling legitimate and desirable executive action, given the inevitability of enforcement discretion and the values-based judgments that attend it. The bright-line rules critics have offered may have the appeal of ease of application, but they constrain executive power in ways that are neither constitutionally required nor necessarily consistent with the goal of keeping executive power in check. The separation of powers ultimately amounts to a messy political contest, and the search for clear, lawyerly lines to draw around the powers of the branches, we have come to believe, is misguided. Even if the current lawsuits succeed in scuttling the Obama initiatives, the imperatives of enforcement will not disappear, and any President will respond to those imperatives while pursuing his own objectives. We can only hope that he has the necessary freedom to structure the enforcement power in ways that serve the goals of accountability and constraint, and that he faces public and congressional pressure to do precisely that.