CRISIS BUREAUCRACY: HOMELAND SECURITY AND THE POLITICAL DESIGN OF LEGAL MANDATES

INTRODUCTION .................... 674

I. THE EVOLUTION OF HOMELAND secURITY AFTER SEPTEMBER 11 .................... 680

A. The Status Quo Before September 11 .................... 680

B. Shocks and Responses: The Immediate Aftermath .................... 684

C. Initial White

House Resistance to Reorganization .................... 687

D. Shaping a Reorganization and Striking Legislative Bargains .................... 689

E. The Final Bill .................... 696

II. UPDATING POLITICAL THEORIES OF BUREAUCRATIC ORGANIZATION AND PERFORMANCE: THE POLITICAL-BUREAUCRATIC SYSTEM .................... 700

A. A Theory of Legislation: Inconsistent Objectives .................... 702

B. The Political-Bureaucratic System: Institutional Solutions to Delegation Problems .................... 704

1. The imprint of legislative and executive politics on the delegation of statutory authority to bureaucracies .................... 704

2. Two theoretical refinements .................... 707

C. Theoretical Conclusion: Policies May Not Be Designed to Succeed .... 712

III. APPLYING THE THEORY: THE ARCHITECTURE OF THE DEPARTMENT OF HOMELAND SECURITY .................... 714

A. The Political Influence of Congress .................... 714

B. The Political Influence of the White House .................... 722

C. The Role of the Post-September 11 Crisis .................... 730

D. Budget Politics and Legacy Mandates .................... 732

E. The Consequences of Reorganization .................... 738

IV. IMPLICATIONS .................... 743

A. Structural Politics and Agency Legal Interpretations .................... 743

B. Parallels with Previous Reorganizations .................... 746

CONCLUSION .................... 749

APPENDIX .................... 756

INTRODUCTION

Modern governments implement most legal mandates through bureaucracies. Politicians delegate authority by Grafting legislative compromises, which lawyers and judges then seek to interpret. But bureaucratic agencies are often the entities that most directly wield the power to spend money, impose penalties, provide public services, and regulate individuals and organizations. Consequently, a central question in public law concerns who exactly controls the bureaucracy's power to interpret and execute law. Although legal scholars are consumed by normative debates concerning who should exercise such control, those debates are difficult to resolve or even follow in the abstract without some knowledge of the techniques used in the political process to control bureaucratic power over legal interpretations and over the execution of regulatory mandates.1

Surprisingly, the creation or reorganization of bureaucratic units-such as the new Department of Homeland Security (DHS)-remains among the least-understood techniques for controlling bureaucracies.2 We know politicians may create or reorganize agencies for multiple reasons: to appear as if they are addressing a salient policy,3 to please organized interests most likely to be directly impacted by the agencies,4 to create procedures that bias agency policy in particular directions,5 and (perhaps more occasionally) genuinely to address a major problem of public concern in a prescriptively defensible manner.6 We know far less, however, about how these different potential motivations interact, how agency structure is affected by major crises such as the September 11 terrorist attacks, or why politicians allocate different chunks of legal responsibility to distinct bureaucratic units.7

These gaps are evident in the persistence of many unsolved puzzles about the largest government reorganization in a half-century-the creation of the DHS.8 For instance, why did the President support the creation of DHS after initially opposing it? Why did the Department become so vast, including in the reorganization a wide range of components with little or no responsibility for homeland security? We also understand little about whether the crisis enabled or forced politicians to forge a bureaucracy that actually enhanced the government's capacity to undertake security-related functions. Even as the creation and operation of DHS continues to inspire controversy, policymakers and scholars have yet to address these questions.9 Nor have they been resolved in the wide-ranging criticisms leveled at DHS following the Katrina disaster, or in light of the national security threats the Department was nominally designed to address.10

The colossal new DHS melded the functions of twenty-two previously existing agencies, from Treasury's Customs Service, to Agriculture's Plum Island Animal Disease Center, to the previously independent Federal Emergency Management Agency (FEMA). Upon its creation, the Department gained regulatory authority over transportation security and matters as disparate as marine ecosystems and refugee admissions. Its ranks swelled with nearly a quarter of a million federal employees ranging from border inspectors to environmental compliance officers. Nothing of this scope had happened in the United States since the creation of the Department of Defense a half-century earlier.11

Even for reorganizations of smaller scope than that of the DHS or the Defense Department, the structural changes are unlikely to be solely symbolic, devoid of legal and policy consequences. Such an assumption ignores the aggressive infighting over structure among legislators, the executive branch, and organized interests. Ignoring the significance of changes in bureaucratic structure also neglects the findings of work in political science and sociology,13 and the legal doctrines vesting valuable discretion to interpret statutes in specific administrative agencies.14 Yet we are only beginning to understand precisely how changes in structure shape the implementation of legal mandates, and how those changes would affect legislative bargaining over the contours of agencies such as DHS.

We propose to address these questions by combining a detailed analysis of the legislative process creating DHS with a new theory of the impact of bureaucratic structure on the execution of legal mandates. Our theoretical approach extends existing accounts of bureaucratic structure to address key features of the DHS case that also arise in other cases of bureaucratic change, especially the role of crises in loosening the constraints of organizational interests and the impact of senior legislators guarding their committee jurisdiction.15 In the process, our analysis fills several gaps in the legal and political science literature concerning some matters, such as how reorganizations differ from familiar procedural techniques for controlling the bureaucracy, including environmental impact requirements or cost benefit analyses; how reorganizations may be enacted despite their adverse impact on the performance of widely held goals; and how Presidents, legislators, and organized interests sometimes bargain about bureaucratic structure in the shadow of an engaged, rather than disconnected, mass public.

As crises enlarge windows of opportunity for legislative action, policy changes in the area of concern-in our case, homeland security-can be driven by the efforts of politicians trying to affect regulatory and administrative activities in a different domain. Changes in the nature and scope of security policy may powerfully affect other legal and policy domains, such as the Coast Guard's environmental regulatory functions or the application of immigration laws. Moreover, politicians use the occasion of legislation to force changes in other areas having little to do with the principal issue being addressed.16 While these themes are particularly relevant in the context of national and homeland security, they also hold important implications for the more-often-studied aspects of bureaucratic politics, affecting domains such as pharmaceutical and environmental regulation. In fact, politicians may endeavor to achieve policyor control-related goals by strategically mixing security and nonsecurity functions within the same bureaucracy.

Against this theoretical backdrop, our account also yields answers to the DHS-specific questions. We argue that the President changed his mind about the reorganization in part because he did not want to be on the losing side of a major issue. But, more importantly, the Administration appears to have supported reorganization on such a massive scale to further domestic policy priorities independent of homeland security. By moving a large set of agencies to the new Department and giving them new homeland security responsibilities without the promise of additional budgets, the President all but forced these agencies to draw resources away from their legacy mandates.

Though such changes have unquestionably become part of the President's own legacy, fixing the precise extent to which he and his top advisers consciously schemed to weaken domestic legacy mandates without regard for a corresponding homeland security benefit must await the judgment of history. But our analysis does establish three crucial realities. First, the Administration eventually pressed for the largest possible Department despite the security-related risks of the merger identified by some of the Administration's own aides. second, many of the key players participating in or affected by the Department's creation-including legislators and bureau employees-explicitly grasped how the merger threatened legacy mandates. Third, key features of the legislative progression culminating in the creation of DHS-in particular, the President's pledge of revenue neutrality and the White House's willingness to consider including agencies such as the Nuclear Regulatory Commission and the Federal Aviation Administration-make little sense without assuming that the White House harbored the goal of affecting the performance of legacy mandates, even if doing so failed to yield a corresponding security benefit. At a minimum, these realities suggest that the Administration became the fertile soil in which arguments supporting reorganization became deeply rooted-arguments that had glaring prescriptive problems, yet happened to serve many of the White House's political objectives.

From a prescriptive point of view, our conclusions are sobering. Our analysis shows how the merger adversely affected even those legal mandates plainly relevant to homeland security.17 More generally, we explain how decisions about whether to create a new security agency, what scope and size to give it, and how to organize congressional jurisdiction over it are unlikely to have been driven primarily by meaningful prescriptive concerns. Yet such decisions are also unlikely to be merely symbolic. They can powerfully (and covertly) reshape how laws are implemented while making it more difficult for government to achieve broadly shared prescriptive goals. Marginal improvements depend on solving problems of legislative oversight, and on whether competent bureaucrats will succeed in forging autonomy and capacity in a world unlikely to support it. While these scenarios remain elusive, our analysis does not yield a blanket condemnation of bureaucracies created through high-profile reorganizations. Bureaucracies forged in crisis may not be inexorably doomed to fail in carrying out their legal responsibilities, and there may yet be reasons to defer to their legal interpretations. Instead, we highlight the difficulties in averting such failure.

Our argument proceeds as follows. Part I presents the homeland security story since just before September 11. Part II develops our theory of bureaucratic organization and performance, with an emphasis on policy change in response to crises. Part III applies the theory to the creation of DHS and related legislative enactments, providing empirical support for the theoretical conclusions. Our purpose there is not merely to demonstrate how the architecture of DHS was politicized, but to elucidate more specifically in what manner political considerations shaped the agency's structure, and to what effect. Part IV discusses extensions and implications. We conclude by discussing promising avenues for further research and noting that the prospects for improving homeland security depend crucially on understanding the political forces that constantly pervade, and often warp, the work of organizations entrusted with this crucial mandate.

I. THE EVOLUTION OF HOMELAND SECURITY AFTER SEPTEMBER 11

Public bureaucracies decide where dams are built, whether nuclear power plants will add to energy production, how intelligence operations are conducted, who gets turned away at the border, and what environmental standards must be met. As with the imaginary lines that subdivide metropolitan areas into distinct jurisdictions, enormous practical significance flows from the legal rules allocating power among bureaucracies. Lurking behind the design of those rules may be a complex political story.

In this Part, we begin tracing such a story. We describe the interwoven origins of DHS and the sprawling statute (the Homeland security Act, or HSA) from which the new Department was forged. We discuss the political context and substance of the Act at length for two reasons. First, our story contrasts with certain canonical descriptions of legislative developments following the September 11 terrorist attacks, which tend to emphasize resolute presidential leadership and relative legislative passivity.18 In contrast, our account reveals presidential policy reversals, the Generality of congressional bargaining even in the midst of a crisis, and the resulting statutory intricacies governing the new cabinet agency. Second, certain puzzles emerge from the story of the Act and the Department, setting the stage for our analysis in the sections that follow.

A. The Status Quo Before September 11

The end of the Cold War dramatically affected debates about American security. By the middle of 2001, American policymakers had largely altered a national security argot once replete with references to a balance of power, containment, and mutually assured destruction. Instead the rhetoric of national security policymakers and analysts increasingly focused on terrorism, asymmetric warfare, and above all "homeland security."19 Terrorist attackers had struck several times during the previous eight years, most notably at the World Trade Center in 1993 and in Oklahoma City in 1995.20 Numerous blue-ribbon commissions had called for heightened attention to the threat of terrorism.21 In response, the new President-elect created a structure within the White House National security Council (NSC) to coordinate matters involving terrorism, its prevention, and the nation's ability to prepare and respond to such attacks.22 Unlike previous directives, the focus was primarily on attacks targeting the United States itself.23

The perceived need for coordination arose in part from the sprawling nature of modern government. Numerous bureaus were responsible for preventing, preparing for, and responding to man-made threats against the United States. Homeland security encompassed aspects of the work of the Departments of State and Defense, as well as the NSC.25 Rounding out the coterie of national security bureaucracies were agencies devoted to intelligence, criminal investigation, and prosecution. The Central Intelligence Agency (CIA) combined an explicit core function of intelligence-gathering with its covert operations.26 By 2001, it had multiple task forces working on terrorism-related issues. An elaborate group focused almost entirely on Osama bin Laden and al Qaeda. Specialized intelligence entities, such as the National security Agency (NSA), further complemented these activities by engaging in electronic eavesdropping outside the United States and gathering considerable signals intelligence. Homeland security and terrorism prevention were also considered the province of federal special agents and the law enforcement agencies for which they worked. As hearings in three Senate committees during the week of May 7, 2001, demonstrated, law enforcement agencies were routinely considered to be responsible for protecting the American public. The Federal Bureau of Investigation (FBI) commanded vast budgets and statutory responsibility, serving as the lead counterterrorism law enforcement agency.29 Foreign attacks on American interests, such as the bombing of the U.S.S. Cole in Yemen, invariably led to the deployment of an FBI team.

Several other bureaus also performed missions relevant to homeland security. A host of specialized law enforcement agencies existed, such as the Immigration and Naturalization Service (INS) and the Drug Enforcement Administration (DEA). The INS served multiple inspection, detention, investigation, quasi-adjudication, and policy functions related to controlling the flow of people into the country. Customs had the similarly daunting task of preventing prohibited items, including drugs and explosives, from entering the nation.30 Within the U.S. Treasury Department, Customs was one of the largest bureaus in terms of budget, staff, and enforcement responsibilities. Like INS, it performed more than just investigative functions (e.g., tracking down money launderers, drug traffickers, and illicit brokers of technology subject to export controls). It also played a regulatory function. While INS regulated the entry of people, Customs controlled the vast flow of goods into (and, in theory, out of) the United States. The secret Service investigated counterfeiting and fraudrelated financial crimes in addition to serving its most visible role of protecting the President. In addition to collecting excise taxes, the Bureau of Alcohol, Tobacco, and Firearms had become a law enforcement agency focused on firearms and explosives, with a wealth of technical expertise on these subjects unrivalled elsewhere in the federal government.31

Transportation and coastal security were handled largely through a tangle of overlapping functions nominally overseen from within the Department of Transportation. The Federal Aviation Administration (FAA) looked after the security of the aviation infrastructure, imposing (among other things) mandates on airlines and airports requiring them to pay for employees to screen passengers and their luggage. The Coast Guard similarly shared with Customs responsibility for key aspects of port security. It also performed coastal search and rescue operations along with a multitude of safety, rate-setting, and environmental regulatory functions.32

Presumably, the work of these agencies could forestall a disaster that would have had to be handled by emergency response bureaucracies, who together formed the final pillar of homeland security-emergency response.33 Of these, the Federal Emergency Management Agency (FEMA) was the most important. In addition to fielding emergency response teams and serving as a conduit for disaster relief money, FEMA also encompassed insurance programs to help mitigate the longer-term impact of various natural disasters.34

Three features characterized the homeland security status quo before the September 11 attacks. First, policymakers assumed homeland security bureaucracies to be capable of operating reasonably effectively even though they had largely separate reporting structures and bureaucratic identities. Though some legislators and independent commissions complained about the fragmentation of responsibility for security-related problems, legislators tolerated the decentralization of bureaucratic power over national and homeland security.35 second, the description of the agencies above demonstrates that virtually every bureaucratic unit that had a role to play in homeland security also had separate functions-such as INS's role in providing immigration services-that were different in scope and, therefore, potentially in conflict with security. Finally, enormous variation existed in the degree of coordination across relevant units. Some problems were undeniable, such as the relationship between the FBI and the CIA (and, for that matter, between the FBI and just about everyone else). But there were also apparent successes, as when federal officials foiled a plot to bomb traffic tunnels leading into New York City and some of its major landmarks.

B. Shocks and Responses: The Immediate Aftermath

The Administration's initial response to the September 11 attack focused on proposing substantive legal changes. Working groups at the Justice Department soon pulled together legislative proposals from preceding years to fashion an outline of what would become the USA PATRIOT Act (Patriot Act).36 The White House supported federal agents' aggressive use of immigration and material witness authority to detain scores of people almost immediately following the attacks, and the President used his authority under the International Emergency Economic Powers Act (IEEPA) to block the assets of various individuals and organizations suspected of being tied to terrorists.37

The White House staff also oversaw the implementation of two noteworthy changes in organizational structure, though its approach to each demonstrated a great deal of caution about major changes in the allocation of bureaucratic jurisdiction. On October 8, using existing statutory authority, the President created the position of Homeland security Advisor within the Executive Office of the President and appointed Pennsylvania Governor Tom Ridge to fill it. Ridge sought to build a structure around his position to match the President's ambitious rhetoric that the new Office of Homeland security would "coordinate" policy by creating a Homeland security Council, paralleling the structure of the National security Council.38

An implicit presumption that underlay the creation of Ridge's office concerned the value of coordinating separate agencies mixing homeland security missions with other functions. White House officials presumed a gap to exist not only in the provision of advice, but in the extent of coordination among a great many agencies and bureaus.39 These officials believed that success in the arena of homeland security depended on enhancing such coordination.40 Ridge sought to provide that coordination, or at least the trappings of it. But even delivering the image of greater coordination to the public proved daunting. The National Counterterrorism Coordinator structure set up at the NSC now had a mandate overlapping with that of the new Homeland security Office (HSO). There were no precedents for how to resolve the potential jurisdictional conflicts, nor was it obvious precisely what it meant for Ridge to coordinate, what his role would be in a crisis, or whether the conflict between the NSC and the HSO would prove a major impediment to the goal of coordination.41

Sensing disarray, some legislators proposed alternative structures. They insisted that the new Homeland security Advisor should be subject to Senate confirmation and have statutory powers over budgets. Senator Bob Graham, a Democrat from Florida, introduced a bill to transform Ridge's entity into a new National Office for Combating Terrorism to achieve the aforementioned purposes.42 Other legislators went even further, reiterating occasional calls made earlier by selected legislators and blue-ribbon commissions for the creation of a new cabinet department focused on domestic security.43

The President strongly opposed these efforts. Instead, White House aides emphasized the advantages of the status quo. From October 2001 until at least March of 2002, the White House Press secretary insisted that creating a cabinet department was unnecessary and possibly counterproductive.44 Unfortunately for the White House, the performance of the new Office during the anthrax attacks belied the President's argument that coordination had been sufficiently bolstered by the creation of Ridge's Office. During the anthrax episode, some observers described Ridge's response as tentative and uncertain.4 Despite the new Homeland security Advisor's declaration that he was in charge of the response, Health and Human Services secretary Tommy Thompson appeared to contradict Ridge. During this difficult period, however, despite both the criticisms and the absence of formal budget authority, Ridge pressed for, and helped the White House achieve, a $1.2 billion increase in the immigration enforcement budget.46

A second structural change took shape in response to discussions between Congress and a reluctant White House. These negotiations culminated in the creation of a new federal bureaucracy to consolidate responsibility for transportation security. Upon its creation in 2002, the Transportation security Administration (TSA) assumed complicated responsibilities over the security of the nation's transportation infrastructure.48 The White House initially opposed the idea. The President preferred to forego creating a new bureaucracy and to keep the screeners private. Whether that opposition was rooted in ideology or in concern for the organized interests likely to be affected, the Administration later abandoned its reluctance and endorsed the idea. Some observers with access to the deliberation now report that the Administration's acquiescence reflected not only mounting pressure from congressional Democrats but also the recognition that the mass public was unlikely to trust private screeners given their inability to prevent the hijackings.49 The new law placed TSA within the Transportation Department. The new agency's creation was also accompanied by an initial dismemberment of the FAA's security capacity (provoking bitter opposition by some employees at the FAA),50 lodging it elsewhere in the Transoortation Deoartment.

C. Initial White House Resistance to Reorganization

Whether TSA would even remain at the Transportation Department would soon become a matter of intense political debate. Well before September 11, a number of legislators and blue-ribbon commissions had called for consolidating some bureaus with a homeland security mandate in a cabinet-level agency.5 Various plans on Capitol Hill focused on three functions: border security and enforcement, disaster response functions relevant to terrorist attacks, and policymaking activities to facilitate the prevention of attacks.

Following the attacks, Senator Joseph lieberman, then serving as Senate Governmental Affairs Committee Chair, re-introduced legislation to centralize certain government functions into a single homeland security department. While some Republican legislators, such as Arlen Specter, expressed some interest, the President did not. He believed that such consolidation would constitute a waste of time at best. On March 19, 2002, for instance, in response to a questioner who asked "[w]hy ... the White House continue[s] to resist the idea of making the Office of Homeland security a Cabinet-level department," Press secretary Ari Fleischer insisted:

I'm not aware of a single proposal on Capitol Hill that would take every single one of those agencies [dealing with terrorism] out from their current missions and put them under Homeland security. So even if you took half of them out and put them under a Cabinet level Office of Homeland security, the White House would still need, in the President's estimation, an advisor on how to coordinate all that myriad of activities the federal government is involved in. So creating a Cabinet office doesn't solve the problem. You still will have agencies within the federal government that have to be coordinated.53

Several factors might have made the creation of a new department seem problematic from the President's perspective. The substantive benefits of a consolidation were unobvious, indeed, highly uncertain-a point to which we return below. Major changes were likely to provoke opposition from powerful legislators whose committees stood to lose some jurisdiction and from the interest groups they served.54 Moreover, career officials and political appointees within the Administration were likely to resist the transfer. Opposition among the bureaucracies could have proven politically costly to the President,55 increasing the risk that reorganization would backfire and potentially exposing the Administration to criticism in the press or on Capitol Hill.56 Critics of previous reorganizations had, after all, pointed out that they had created such problems in the past.57 Finally, to the extent that prescriptive concerns mattered at all (something we explore and question below), they might cut sharply against the sort of reorganization that might seem superficially appealing to the public. Reorganizations almost inevitably cost money and create friction among people and organizations scrambling to understand the consequences of the new hierarchy under which they must work. Moreover, reorganizations create new authority structures that typically engender friction, which hinders one of the main reasons for reorganizationnamely, coordination.

Even in the midst of its crisis mode, White House aides may have appreciated certain risks inherent in taking responsibility for a massive reorganization. In the short run, there was a substantial chance that reorganization would actually decrease agencies' effectiveness in responding to security threats, at a time when the Administration would have thought these would almost certainly persist or grow. Though little is known about the impact of reorganizations on bureaucratic performance, it is widely acknowledged that performance suffers at the outset.5 One account of the frantic days following September 11 underscores the extent to which these prescriptive concerns, intermingled with an appreciation of the political costs, were on the White House radar screen even two days after the terrorist attack:

By Thursday, Abbot, Kuntz, and Libby [aides to Vice President Cheney] had concluded that the first thing the Bush administration should do would be not to reorganize all those agencies, but to hire a heavyweight to come work in the White House and coordinate them, much the way Condoleezza Rice, the National security Advisor, coordinated the various agencies involved in foreign and defense policy. They could never get all the agencies with some role in domestic security into one department, they reasoned, because so many also did so many other, unrelated jobs. (FEMA, for example, administrates flood insurance in addition to coordinating the federal response to disasters.) The goal should be to coordinate whatever they did related to homeland security, rather than spend a lot of time and money dislodging them from their current departments.5

The White House emphatically followed that path. In the weeks following the attacks, aides insisted that Ridge's office fit the bill, coordinating both the sprawling federal security apparatus and the thousands of local police and fire departments, from Manhattan to Minnesota to Manhattan Beach, still scrambling to enhance security in their local jurisdictions.60

Despite the Democrats' control of the Senate, Congress broadly supported the President in the two months following the attack. The Administration achieved rapid passage of the Patriot Act and a resolution authorizing the use of force abroad in response to the attacks. Even individual Democratic legislators seemed initially inclined to cooperate. Senator Graham, for example, agreed to the President's proposed legislation. The extent of congressional support contributed to an impression of considerable (if not frantic) policy change and implementation. The Administration's burst of activity since September 11-including the Patriot Act, the creation of a large new transportation security bureaucracy, the private sector's thrust to crack a new homeland security market, state and local officials' regional exercises, the invasion of a Central Asian nation, and forging a new White House staff office-seemed to push the limits of what the nation's political machinery could digest in such a short time.

D. Shaping a Reorganization and Striking Legislative Bargains

Legislators were not entirely passive participants in the policymaking process, however. Emboldened by White House reluctance and public opinion surveys, a score of legislators called for a new cabinet department focused on homeland security.61 White House aides thus encountered a more complex political terrain. By late October 2001, Democrat Joseph Lieberman in the Senate and Republican Mac Thornberry in the House led what had begun as an unlikely (if not downright outlandish) crusade to forge agencies into a new super-bureaucracy that began to pick up support among both Republican and Democratic legislators.62 The response from the White House through the rest of 2001 and early 2002 remained an emphatic "no."

But this negative response was not the last word from the White House. On June 6, President Bush unveiled his own proposal for the new Department of Homeland security.63 Hints that something was in the works had appeared beginning in April, when Budget Director Mitch Daniels publicly stated that the President could propose reforms at a later date.64 What Daniels did not say was that the President had already set the process in motion. In late 2001 and early 2002, the President had several conversations with Ridge and Chief of Staff Andrew Card about the merits of creating a new department to administer homeland security.65 Responding to congressional resistance to an earlier border consolidation plan, the President apparently noted the plan "seems kind of small to me," and then added:

You know . . . maybe we should stop getting pecked to death like this. Maybe it 's time to think big. When you do something piecemeal, all the interests here come at you one by one and kill you. Let's just make believe we are recreating the government from scratch and map out what we'd put in a new homeland department and then maybe we'll go for it.

By March, aides to Ridge, Card, and Daniels were holding secret meetings in an underground White House bunker.67 Participants in the meetings now suggest that their deliberations were driven largely by prescriptive concerns about the organizational merits of consolidating various units. The group was also concerned about what could be sold on Capitol Hill, as underscored by the fact that the initial small group soon expanded to include staff from the White House legislative affairs operation.

Although the small amount of information available regarding these early meetings makes it impossible to determine the participants' precise mix of concerns, the discussions soon yielded a rough picture of a department with two significant features. First, it would be significantly larger in scope and size than anything that had been proposed by the Democrats or previous independent commissions. "The PEOC [Presidential Emergency Operations Center] group," noted one commentator (referring to the underground bunker where White House aides were meeting to plan the new agency), "had now created a mega-agency that far exceeded Senator lieberman's relatively modest proposal for a Department of Homeland security, and they weren't finished."68 The working group demonstrated a willingness to contemplate an even larger department by its inclination to consider moving the FBI, the FAA, and ATF into the Department (moves that were ultimately rejected). Precisely why the White House process contemplated and produced such a sprawling department is not immediately clear, a matter to which we return in Parts II and III. Second, the PEOC group intended the new department to serve as a showcase for the value of flexibility in presidential control of personnel. The goal of watering down civil service protections appealed to the President's aides, particularly Daniels.69

The thirty-five page legislative proposal that emerged from the meetings of the "PEOC group" sought to establish four primary "directorates" at the core of the new Department: border and transportation security, information analysis and critical infrastructure protection, science and technology, and preparedness and emergency response. It included provisions allowing the President to appoint over a half-dozen assistant secretaries without Senate confirmation, and sought to imbue the President with power to redistribute appropriations among several different agencies. It called on political appointees to rewrite civil service protections governing many of the agency's new employees and to replace them with a "flexible" system, presumably vesting greater power over career officials in the hands of political officials.70

The starkest feature of the plan was its scope. It sought to move some twenty-two agencies into DHS, despite the fact that not all of their functions conformed to even the most expansive definition of homeland security. The marine environmental portions of the Coast Guard, for instance, were to be entirely absorbed by the Department, as were the revenue collection and trade enforcement functions of the Customs Service, and the agricultural regulatory functions of the Animal and Plant Health Inspection Service (APHIS).71

In contrast, many previous proposals for the creation of a homeland security agency had considered more modest changes. For example, Republican Representative Mac Thomberry's pre-September 11 bill, introduced in March of that year, essentially advocated moving FEMA, Customs, the Coast Guard, and border patrol to the new Department.72 Unlike the President's plan, agencies such as the secret Service, APHIS, the investigative and regulatory functions of immigration authorities, health-related functions such as the national vaccine stockpile, and the Treasury's Federal Law Enforcement Training Center were left untouched.73 The Administration-supported reorganization's mixing of a wide range of legacy missions with new homeland security responsibilities raised the question of how the tradeoffs were to be made across these missions.74

A flurry of activity followed the White House's June 7 announcement. The White House briefed cabinet members (many of whom were just learning about the plan at that point) and legislative leaders. The President's aides spoke to the media, and at 8 p.m. Eastern Time, the President spoke to the nation about the plan.75 The elaborate roll-out confirmed that the President and his staff were now not only joining the chorus of support for the reorganization but sought to lead the reorganization drive. The building blocks of the new proposal broke from past plans in the larger scope of agencies to be included and in the provisions weakening civil service rules. Despite these differences, publicly the plan was premised on the same logic that the alternatives were: the value of centralization.

The Administration's decision to develop that plan did not unfold in a political vacuum. Several factors may have underscored to the White House that it would face rising costs by continuing its opposition to the creation of a new cabinet department. Its legislative affairs staff documented rising support for consolidation among legislators. Security issues continued holding much of the public's attention, particularly given congressional hearings about the September 11 attacks, and public debate about whether an independent commission would ultimately be created to investigate the attacks, the creation of which the White House opposed.77 White House officials may have anticipated risks from opposing a new department along with an independent commission heading into the midterm congressional elections. In addition, creating a new department may have had particularly strong political salience because of its appeal to latent, if potentially superficial, notions of effective governance.78 But these developments fail to account for the choices the White House made regarding the size, scope, and prescriptive merits of the new Department.

The President's June announcement found Congress still mired in divisions about the merits of creating a new department. Support remained vigorous among members of the Senate Government Affairs Committee, who had proposed renaming the committee "Homeland Security and Governmental Affairs" and who almost certainly stood to gain prestige, power, and influence if their expectations were fulfilled and the new department fell under their jurisdiction. At the same time, a stubborn core of opposition persisted among lawmakers who had committee jurisdiction to lose, or who saw position-taking opportunities in opposing substantive civil service changes, earmarks, and liability protections in the President's proposal.79

With a growing number of legislators joining the President in supporting the creation of a new department, there followed a period of intense bargaining. In the House, Speaker Dennis Hastert and the Republican leadership created a two-track process to evaluate the bill, christened the "Homeland Security Act" (HSA). Over a dozen committees with existing jurisdiction over various aspects of homeland security marked up the bill, but their votes were considered advisory in nature. Meanwhile, Hastert empaneled a Select Committee on Homeland security, which included most of the chairs of existing committees with jurisdiction over homeland security, to make final decisions on the House version.80

Hastert's move was understandable. If the leadership had left the decisionmaking solely to the existing standing committees with existing authority, they would likely oppose the major reorganization that the President was now publicly committed to support and that a growing segment of the public appeared to support. Alternatively, if Hasten created his own handpicked committee, existing members and committees would likely be opposed to the result.

The markups revealed widespread concern among the committees regarding potential changes in their jurisdiction. For example, the House Judiciary Committee voted to transfer the Secret Service to the Justice Department, over which it had jurisdiction, instead of letting it go to the new Department. The House committee with jurisdiction over transportation issues sought (like the one in the Senate) to prevent or delay moving the new Transportation Security Agency to the new Department. And many legislators sought to limit the presidential powers in the new bill, such as those allowing the White House to appoint assistant secretaries without Senate confirmation.81

Although the first stage of advisory markups appears symbolic since the Select Committee would have the final say, the procedure resulting in these votes could also be understood as serving as a critical information collecting device for party leaders who favored the reorganization. The House leadership appeared inclined to support the President's push for reorganization. Nonetheless, the membership was likely to be quite wary of a wholesale redistribution of power within the legislature, which was an almost inevitable consequence of the reorganization legislation. The markups thus allowed the committees to reveal what portions of the proposed changes were politically most costly to them and which were less so. The Select Committee could then take these committee actions into account in its decisions, either by incorporating the committees' changes or by searching for other means to assuage the committees' ostensible concerns. The Select Committee reported its version of the HSA on July 19, 2002, on a straight 5-4 party-line vote. This legislation became the basis for the final bill, described below, and passed by the full house on July 26, 2002.82

Partisan divisions on the House Select Committee foreshadowed greater conflict in the Senate, where Democrats controlled the chamber by a tiny margin. Already, the President's June announcement had probably begun to blunt the perception, which Senator Lieberman had intensely sought to foster, that creating the new super-agency was a Democratic initiative. Lieberman now sought to recapture the initiative. In late July, the Senate Governmental Affairs Committee approved a Lieberman-sponsored version of the homeland security bill with civil service provisions more acceptable to the Democrats and provisions transforming Ridge's existing office at the White House into an Office of Counterterrorism with a director subject to Senate Confirmation.83 The Senate then received the House version of the HSA, which gave the President, among other things, the power to exempt parts of government from federal labor management relations statutes.84 Lieberman and his allies sought to substitute his new bill for the House version. But Senator Phil Gramm filibustered cloture motions to limit debate. In the end, Senate Democrats were unable to pass a cloture motion to force a vote on their preferred version of the bill, which would have triggered a House-Senate conference on the creation of the new Department. And they were unwilling to compromise on the civil service provisions. Thus, when the midterm elections arrived, the Senate had not agreed to support the President's and the House Republican leadership's version of the HSA. Ironically, the Democrats were exposed to the charge that they opposed the creation of a department that they had played such a key role in forcing the President to accept.86

The elections brought further unwelcome news for the Democrats, who lost the Senate and were dealt an even more lopsided minority in the House. After a final attempt to strip provisions allowing the President to suspend collective bargaining protections, the Democrats compromised and allowed cloture to be invoked in the Senate by a vote of 83 to 16 on November 19, 2002. The Senate then passed the House bill with minor amendments that were approved in the House by voice vote, and the bill was sent to the President on November 22, 2002.87

E. The Final Bill

When the President declared victory three days later, he signed a bill that was far more detailed than what the White House had initially proposed. The details reflected protracted presidential bargaining with Congress. On the surface, the final bill established a department that was quite similar to what President Bush had proposed. Consistent with the President's proposal, the core functions of DHS were grouped into four directorates: Border and Transportation Security (including the bulk of the Department's employees and resources), Intelligence and Infrastructure Protection (incorporating some of the smaller infrastructure protection offices absorbed from the Commerce Department's Critical Infrastructure Assurance Office and the FBI), Science and Technology (including the Homeland Security Advanced Research Projects Agency, or HSARPA, initially projected to administer a $500 million fund supporting innovative research and development projects), and Preparedness (primarily FEMA). As Appendix Table 1 indicates, not every agency that the White House working group considered placing within DHS ended up in the Department. The sprawling agency had nonetheless come to encompass functions ranging from international child labor investigations to marine fuel leaks, and included nearly every entity that the President ultimately proposed to move into the new agency.

Despite the White House's relative success, the legislative bargaining process also introduced some important changes. The final Homeland Security Act contained nearly two hundred separate legislative provisions (with some individual provisions stretching over a half-dozen pages). In contrast, the President's original proposal contained fewer than fifty sparsely-written provisions focusing primarily on the structure of the four aforementioned directorates. This disparity reflects complexities lurking beneath the surface of the HSA.

Unlike the original White House bill, for instance, the resulting HSA simultaneously included language explicitly emphasizing the importance of non-homeland security missions along with the terrorism-focused language90 and provisions establishing the secretary's power over the bureaus.91 It could not have been lost on legislators that the first three of the Department's six functions concerned terrorism. At the same time, lawmakers supplemented the blanket entreaty for the new Department to "[e]nsure that the functions of the agencies and subdivisions within the Department that are not related directly to securing the homeland are not diminished or neglected except by an explicit Act of Congress" with additional agency-specific language.92 In the case of the Coast Guard, legislators actually allowed some (ostensibly limited) diminution of non-homeland security functions, but sought to monitor changes in its nonsecurity regulatory and safety missions by requiring regular reports from the Inspector General and the Secretary.93 The HSA also contained similarly detailed provisions governing a plethora of other agencies transferred to the new Department, specifying (for example) that some revenue-collection regulatory functions of Customs would remain at Treasury94 while the Secretary of Homeland Security could administer others, and providing that FEMA should carry out an "all hazards" mission while simultaneously allowing the Secretary the flexibility to refocus FEMA's actual operations.96

The resulting bill also denied to the White House many of the sweeping presidential powers contained in the original proposal. Despite the united Republican government, the bill did not allow the White House directly to control the timing of agency transfers, to redistribute appropriations among different agencies, or to appoint assistant secretaries without Senate confirmation. The HSA also created a host of research institutes and centers of excellence with mandates to focus on exceedingly broad conceptions of homeland security (including, for example, one center focused on developing new prison-related technologies).97 Over time these institutions would almost certainly serve as conduits for federal spending benefiting particular regions or industries.98

Finally, the legislation accomplished a proliferation of other goals, many of which were initially addressed in separate legislative proposals. For instance, although the Justice Department lost virtually all its immigration enforcement power when INS was transferred into the new Department, it gained most of Treasury's Bureau of Alcohol, Tobacco, and Firearms." Pilots gained a right to be armed.100 Airlines obtained new insurance and financial protections,101 and the Department gained new regulatory powers to protect manufacturers of "anti-terrorism" technologies from liability.102

Together these features evince the importance of four recurring themes associated with the legislative bargaining process. First, although the new bill granted the secretary of Homeland security sweeping powers of "direction, authority, and control" over the new Department,1 legislators recoiled from granting the President the sweeping powers he had requested to reallocate appropriations, appoint assistant secretaries without confirmation, and control the timing of agency transfers. second, legislators showed predictable interest in creating conduits for the transfer of federal money to particular regions or industries-in short, pork. Third, lawmakers used the fast-moving HSA to advance discrete legislative projects that allowed them to signal desirable positions to the public (as with the provision arming pilots) or achieve major substantive policy goals sought by organized interests (such as the expansion in airline liability protections). Fourth, even as they ultimately voted for legislation that transferred major agencies into a new bureaucracy, legislators insisted on asserting control over those agencies by including provisions governing how those agencies were supposed to discharge their missions. In particular, legislators showed some awareness that the Department would-true to its name-emphasize homeland security over a plethora of legacy missions. In response, lawmakers made modest efforts to stress the continued importance of the agencies' myriad non-homeland security responsibilities.104

Equally noteworthy is what the bill omitted-congressional organization. The HSA describes the "sense of Congress that each House ... should review its committee structure in light of the reorganization of responsibilities within the executive branch by the establishment of the Department,"105 but the bill requires no changes in structure. Thus, as tens of thousands of inspectors, agents, and government employees began a long journey towards their positions in DHS in late November 2002, the congressional oversight structure over the Department's components remained largely unchanged.106

Earlier the White House had sought to bolster its reorganization plan by arguing that too many congressional committees were involved in overseeing homeland security. It ultimately acquiesced to a status-quo-driven congressional oversight structure. Although the Senate's Homeland security and Governmental Affairs Committee appears to have gained some degree of jurisdiction at the expense of other committees, the Senate's Appropriations, Judiciary, Armed Services, and Finance Committees (among others) all retain substantial homeland security oversight responsibilities. In the House, even less centralization occurred in the legislative oversight structure. The relative preservation of the status quo in the House probably indicated the leadership's reaction to the repeated standing committee "advisory" markups seeking to limit the size and scope of the Department. Although such votes had not succeeded in limiting the scope of the sprawling new Department, little had changed with respect to congressional oversight as late as mid-2004:

In reality, jurisdiction [over DHS] in both chambers remains allocated to dozens of committees and subcommittees. From January to June 2004, DHS officials testified before 126 hearings, or about 1? per day of legislative session, not including briefings or other meetings. secretary Ridge estimated that he has been called to appear before 80 different committees and subcommittees on the Hill....108

Potential problems with congressional oversight did little to dampen the political enthusiasm for the new Department when it finally opened its doors in mid-2003. The President had switched from opposing the merger to fashioning-with legislative allies-a new homeland security agency larger than anything previously proposed. Neither the Administration nor members of Congress involved in forging the Department expressed much uncertainty about how well this sprawling arrangement would function. But history would soon extinguish any certainty about the legislation's merits.

II. UPDATING POLITICAL THEORIES OF BUREAUCRATIC ORGANIZATION AND PERFORMANCE: THE POLITICAL-BUREAUCRATIC SYSTEM

To understand the history of DHS, we must begin by considering what politicians want to accomplish when they bargain over the creation of a new bureaucracy. Over the past three decades, political scientists have developed a new approach to studying questions of bureaucratic performance and organization.109 This perspective emphasizes a series of non-obvious implications about bureaucratic structure, incentives, and performance. In contrast to the traditional literature on bureaucracy,110 which primarily seeks to explain bureaucratic performance and inefficiency from an "internalist" perspective focusing largely on the goals and routines of the bureaucracy itself, the new approach discussed here places greater emphasis on an "externalist" perspective-one based on factors in a bureaucracy's political environment.111

This Part uses insights about the importance of external political dynamics to develop a theory explaining the creation and performance of bureaucracies such as DHS. Our approach refines existing theories by accounting for the impact of crises, and of uncertainty about the prescriptive effects of reorganization. These refinements allow us to better understand the evolution of DHS. They also shed light on the political design of legal mandates at other points in history.

The basic outline of our theory can be summarized briefly. First, interest groups have influence on the policy process because they affect the reelection prospects of political officials.112 Interest group activity implies a bias toward those groups that are active. But this influence is not simply general influence. It works through the political system. Interest groups within the President's support constituency have more influence on the White House than those who are not among that constituency. Similarly, those interests closely connected with important members of the relevant congressional subcommittees that oversee various agencies have more influence on the oversight process. second, a range of interbranch dynamics shape the legislative process. As lawmakers bargain, key legislators negotiate with organized interest groups and with the President. The resulting congressional process-combining internal negotiations among legislators and pressure from outside interest groups, and bargaining with the President-puts its distinctive stamp on both bureaucratic structure and bureaucratic incentives. Third, a range of intrabranch politics affects design, involving both bureaucratic politics within the executive branch as well as congressional bargaining among committees and among legislators with different vested interests in the existing structure. Fourth, mass politics affects design. The public's level of attentiveness affects the induced preferences of representatives over various institutional design questions. Fifth, a crisis dramatically transforms public attentiveness and can therefore dramatically change the political pressures on a given issue. Together these factors interact in potentially complex ways.

A. A Theory of Legislation: Inconsistent Objectives

The civics textbook view of national legislation is that it is designed to solve various social, economic, and security problems.113 In practice, legislation rarely addresses these problems particularly well when measured against the standard of what politicians publicly claim their goals to be. The reason is politics.114 First, as designed by the Founders, the separation of powers system assures that the two houses of Congress and the President have different electoral constituencies and therefore respond to different interests. The different constituencies lead officials in the different branches to favor different ways to address each policy issue. second, legislation requires majority support (and sometimes supermajority, as in a filibuster or veto override), endowing majority pivots with bargaining power. Third, each house of Congress has a range of institutions that grant further bargaining leverage and control of veto gates, notably committees and the majority party leadership.

All these institutional features of national politics combine to grant bargaining power to a range of legislators with divergent goals. Routinely, many such lawmakers use their leverage to alter the legislation in ways that benefit their constituents who may have very narrow interests or interests that conflict with the overall goals of the legislation. Below we survey a series of implications that congressional institutions have for the shape of legislation. For now, we mention one implication that we believe is the most important, namely, the effect of political compromise on the shape of legislation.

Typically, legislation begins with proponents who favor legislation that addresses a particular problem."5 Often, the initial legislation is designed efficiently in the sense that, given the sponsors' definition of the problem, the legislative proposal addresses the problem directly and effectively. This proposal is often relatively short. More importantly, the proposal typically has little chance of passing in this form. Opponents seek to preserve the status quo by defeating the legislation in any form.

Although most popular accounts depict legislative struggles between the legislation's proponents and opponents, a critical-indeed, pivotal-third group exists: the moderates. As the moderates go-for or against-so goes the legislation. To succeed, a bill's sponsors must bargain with the moderates for their support. The inherent need to negotiate to pass legislation implies that nearly all successful congressional legislation is the product of legislative compromise between the bill's sponsors and the moderates. Moderates generally seek to pass a weaker form of the legislation, a bill that simply does less of what the sponsors originally proposed. Typically, compromise that brings the moderates on board requires limits on the legislation, including restrictions, exceptions, and cumbersome procedures that afford interested parties the ability to contest or delay the implementation of the legislation's effects. These provisions often compromise the legislation's purpose in the sense that they make it harder to achieve the law's purported goals.116

An example: in the mid-1970s, environmentalists sought to control sulfurdioxide emissions, a major source of which was sulfur in coal burned to generate electricity. To reduce these emissions, utilities could either switch to low-sulfur coal or add "scrubbers" to their waste stacks that would eliminate the sulfur from the exhaust emissions. Although the prescriptively optimal way to address the problem would have been to allow some utilities to burn lowsulphur coal instead of installing scrubbers, this approach engendered political opposition because it endangered union jobs in the high sulfur coal regions. In the end, supporters of the legislation agreed to compromise with the representatives of the unionized coal regions to require that every utility in the country-regardless of whether it burned high or low sulfur coal-add scrubbers. This requirement at once raised the cost of removing sulfur from emissions, lowered the overall environmental cleanup per dollar spent, but saved union jobs.117

In short, major legislation often approaches incoherence, in the sense that it contains provisions designed to address a particular problem along with provisions that limit the efficacy of those solutions in the form of exceptions, exemptions, limitations, and cumbersome procedures.118 The latter provisions meant to qualify the act are a necessary feature of the legislation: without them, the legislation would not pass. Put differently, Members of Congress knowingly pass an incoherent measure, in that its provisions are internally inconsistent to the point of profound ambiguity, when they would not be able to pass the more coherent version.

B. The Political-Bureaucratic System: Institutional Solutions to Delegation Problems

1. The imprint of legislative and executive politics on the delegation of statutory authority to bureaucracies

The divisions among legislators-and between specific legislators and executive branch officials-discussed above have predictable implications for the production of statutes and the structure of the bureaucracies that implement statutes. In particular, previous research identifies the following factors as especially likely to shape legislative bargaining when lawmakers are divided about their goals. The general consequence of this system is goal distortion, the process by which politics inevitably distorts the goals of the legislation as legislators transform a proposal into a vehicle that will pass Congress.

The distributive tendency and goal distortion. An inevitable effect of the legislative process is that benefits from a program must be widespread or the program will not gain sufficient support to pass. Consider legislation designed to alleviate poverty. If the proposal does this efficiently, it will concentrate resources in states and congressional districts with high concentrations of poor people. Because poor people are a relatively small minority of the countryperhaps fifteen percent-the legislation concentrates resources in a small number of congressional districts. This concentration also means that the legislation will have difficulty passing. A small minority of districts benefit from the legislation, while most of the polity gain little while bearing the tax costs of the program.

In reaction to the problem of passage, bill sponsors have a tendency to distribute the funds more widely than efficiency dictates. This distributive tendency implies a high likelihood of goal distortion.119 By distributing the funds from a program more widely, this tendency breaks the link between the legislative solution and the problem the program is designed to address. Programs regularly distribute funds widely in a way that distorts their purpose. Some examples include the space program, various urban and housing programs, the Department of Energy's National Laboratory system, and (as discussed below) DHS's programs to fund emergency preparedness grants as well as research and development.120

Multiple veto points. We have already seen that the need to command a majority typically implies legislative compromise that affects both the legislation's goals and the means by which it addresses those goals. Other institutional aspects of Congress work in the same way.

First, at least two committees (one in the House, and one in the Senate) share jurisdiction on any given issue, and often many more committees have somewhat overlapping jurisdiction. Because of these multiple veto points, members can sometimes hold up legislation desired by others as bargaining leverage over legislative issues wholly independent of the legislation. During the consideration of the early legislation addressing the savings and loan crisis of the 198Os, both House and Senate passed versions of legislation aimed at mitigating the growing savings-and-loan problem that would ultimately cost taxpayers several hundred billion dollars. Both versions of the bill contained similar provisions addressing the crisis. But a compromise failed because House bill contained different add-ons to the legislation than the Senate bill. Because members of the two houses could not agree on how to compromise these additional parts, the legislation died in 1986, allowing the problem to mushroom.121

second, another type of intrachamber conflict concerns the distribution of power and authority within the chamber. Committee members typically seek to enhance the discretion, scope, and authority of their committees, even at the expense of others.122 For example, when a new issue arises that does not readily fit with the existing pattern of authority, members on different committees jockey for control of this issue. Energy provides the canonical example. Prior to the energy crisis of 1973, energy was a relatively minor issue. With the energy crisis, it suddenly became a significant national issue. Within each chamber, a wide range of committees sought to control a piece of this issue. Indeed, negotiations over how to divide this authority within each chamber delayed a national response to the crisis for several years, from 1973 until 1977.

The upshot is simple. As new policy issues emerge and as political support for existing issues evolves, committees negotiate over who has authority on a given issue or a given aspect of an issue. Because the policy preferences on the different committees jockeying for power can often be quite different, different allocations of authority within the chamber have significant policy implications.

Congressional jurisdictions. Congressional committees and subcommittees are intimately involved in bureaucratic oversight.123 These bodies are not only each chamber's agents charged with overseeing a bureaucracy's implementation of policy, but they put their own stamp of interest on the direction ofthat policy.

To an important degree, the structure of Congress and the bureaucracy parallel one another. These institutions work together. Moreover, complex policies, such as the environment and energy, are often divided into a great many pieces, with different subcommittees overseeing different portions of a bureaucracy's activities.124

Because members on the different subcommittees have different interests, they pull policymaking within their domain in different directions. For policies that are completely independent, this is fine, but when the policies interact-as they must, for example, because of budgetary trade-offs-the inconsistent views on different subcommittees can create potential problems.

These interactions plainly have implications for bureaucratic reorganization. Consider a set of bureaus that work on related policies but were created by different legislation and are overseen by different subcommittees or committees. They are likely to pursue different goals, in part because the legislation creating them differs and in part because the interests of the members of the relevant subcommittees differ.

Suppose that the President seeks to achieve efficiency gains by coordinating the bureaus' activities though a bureaucratic reorganization that combines the two. The degree to which these efficiency gains are realized in practice depends in part on whether there is a parallel congressional reorganization. Allowing the two separate subcommittees with different goals to retain jurisdiction over the different pieces of the now reorganized bureau impedes coordination: the different interests on the two subcommittees lead them to continue to pull the two portions of the reorganized bureau in different directions. Members of each subcommittee face a common pool problem: both prefer the efficiency gains, yet both prefer more benefits from their own portion of the whole. Because they control only their portion, each has a tendency to take more for itself. To the extent that coordination achieves an increase in benefits that comes at the expense of one of these pieces, members of the relevant subcommittee will use their oversight powers to work against coordination. In contrast, if the congressional jurisdictions are also reorganized so that one subcommittee now gains sole jurisdiction over the bureau, it will be better able to make coherent trade-offs between the two activities and hence realize the efficiency gains.

Partisan electoral goals. A fourth principle relevant for bureaucratic structure is that each of the two national parties has incentives to use legislation to enhance their members' electoral goals at the expense of the other. The majority party in Congress typically has the advantage.

A particular instance of partisan warfare is "baiting" the opposition on popular legislation. Suppose the public is strongly supportive of some legislation. Because voters rarely follow or understand legislative details, the majority party has an incentive to add extreme components to popular legislation in an effort to bait the minority. They do so in hope that members of the opposition will object to or obstruct the legislation because of these features. Often, the public fails to understand the nuances and instead sees the opposition as simply objecting to the legislation. This gives an electoral issue to majority party candidates who challenge opposition incumbents. Sure, the incumbent will try and explain-"I would have supported the legislation, but this one feature made it objectionable." Such a strategy may succeed in some cases. In others it may have the hollow ring of an excuse.

2. Two theoretical refinements

Despite its usefulness in understanding the broad outlines of how statutes and bureaucracies are designed by the legislature, existing work largely leaves out some crucial elements, notably the role of crises and uncertainty. Consider each in turn.

The role of crises. For our purposes a crisis is (a) a circumstance perceived by the public as an exogenous shock sharply raising demand for policy changes in a particular domain, and (b) costly for politicians to ignore.

A political crisis has four interrelated effects on the forces underlying bureaucratic creation and structure. First, it implies a far more attentive public. Although the public cannot attend to the policy details, an attentive public pressures political officials to address the problem underlying the crisis.125 second, as a consequence of public attentiveness, interest groups often have less power to protect their interests. Whereas interest groups may predominate in the relevant policy areas prior to a crisis-in part because of relatively inattentive public or because the portion of the public that is attentive is a small subset of the larger population-a crisis that brings public attention provides a new set of rewards for public officials to counterbalance the rewards generated by interest groups. In some cases this change allows new interest groups to become relevant, as in contractors following a widely publicized natural disaster.126 Third, a crisis typically means a far greater urgency than in most policy areas so that political officials must act fast; failing to do so will leave political officials electorally vulnerable for having failed to address a critical issue in a timely fashion. Finally, these effects are sometimes most pronounced for the President in the sense that he is seen as the national leader.127

Taken together, these four effects have several implications for a crisis response. Crisis responses tend to lack deliberation. Because an attentive public demands timely action but cannot understand details or the implications of all bureaucratic-institutional choices, elected officials are tempted to act too quickly so that they can demonstrate their responsiveness, even if their legislation is ill-considered.128

As an example, consider the response to the thalidomide episode. Thalidomide was a drug given to pregnant women in the 1950s and early 1960s to reduce the effects of morning sickness, but had disastrous side effects impairing fetal development. Many so-called thalidomide babies were bom without arms or legs.129

Prior to the thalidomide episode, Senator Estes Kefauver (D-Tenn.) had criticized the drug regulatory process by the Food and Drug Administration (FDA), which had regulatory control over the introduction and oversight of drugs. For several years, he had held hearings and pushed legislation that would force drug manufacturers to prove their drugs were efficacious in that they actually produced the effects that the manufacturers claimed. Of course, the problem of efficacy is different from the problem of drug safety, which involves side effects.130 In the wake of the thalidomide episode, Congress quickly passed Kefauver's proposals even though the episode involved drug safety while the legislation largely addressed the problem of efficacy.

This case exhibits two separate ironies. First, although the 1962 Drug Amendments promoted a laudable policy goal, they may have had some negative effects on the introduction of new drugs in the United States relative to Europe, Canada, and Japan. second, thalidomide had never been introduced in the United States even under the existing regulatory scheme (the side effects occurred from its distribution in Europe and Canada).131

Thus, the policy imperative created by a crisis combines with public inattentiveness to policy details to push political officials to act quickly, sometimes more quickly than is advisable. For many officials, it is better to have some policy-any policy-than to accept a delay that eventually yields a policy better aimed at the problem.132

Another effect of crises concerns interest groups. In contrast to the pattern governing legislation and policymaking in ordinary times, crises can make it attractive for politicians to act despite the opposition of interest groups. The policy equilibrium in ordinary times generally reflects the influence of (sometimes competing) interest groups. Crises that draw public attention provide a new source of political rewards to political officials. Public attention thus allows legislators and the President more room to bargain over policy changes given the extent of public demand.

Nonetheless, while this window for policy innovation may occasionally result in prescriptively attractive policies, it is wrong to assume such an outcome is rendered more likely during a crisis. With interest groups less able to stop or water down a legislative change, the President and the legislature take center stage. Presidents may seek new legislative authority to layer political appointees over existing independent agencies, while enterprising legislators lacking the most desirable committee assignments may scheme to enhance their own committees' jurisdiction. But, as will emerge below, their competing objectives are likely to lead them to create bureaucracies ill equipped to achieve their stated purposes, except under special conditions. In contrast, politicians might endeavor to build an effective bureaucracy after a crisis if the relevant goals are widely supported, if accurate information is widely available about the relationship between structural or legal changes and the advancement of those goals, and if bureaucratic performance to achieve those goals is easily observable over time. If these conditions exist, a crisis can free politicians from some of the interest group pressures that often contribute to bureaucratic failure. In contrast, in the absence of the aforementioned conditions, crises have the potential to introduce distinctive pathologies into the legislative process as legislators and the President scramble to produce changes simultaneously advancing their political agendas and pleasing a more attentive public.

The role of uncertainty. In their continuing efforts to advance a specific political agenda, Presidents and their staff members may be able to predict certain implications of reorganization, such as those impacting the performance of domestic regulatory mandates.133 By pressing agencies to do more with similar or more constrained resource endowments, for example, reorganizations can subtly force agencies to reshape their activities. From a prescriptive point of view, however, the relationship between bureaucratic institutions and performance is not an exact science. This implies a significant degree of uncertainty about the prescriptive benefits associated with any reorganization. Moreover, when we parse the different costs and benefits, we find that some are far more uncertain than others.

Consider a reorganization ostensibly designed to create greater coordination among agencies with related functions. Putting the agencies together under a single umbrella with greater centralization holds the promise for greater coordination. Reducing redundancy, sharing of information, and coordination of a wide range of many separate activities potentially increases efficiency and total output.

Yet major reorganizations also have a range of potentially negative effects. First, by creating a far larger and more complex organization, massive centralization makes it harder for organizational leaders to master their organization, to understand its separate parts, and to understand the complex ways in which better coordination can be achieved. Even if a new agency were sharply focused on a narrow definition of homeland security, the full scope of activities involving matters such as immigration enforcement would require officials to understand sprawling intricacies.134 Second, centralization can diminish the competition among agencies and risks creating a bureaucracy with a monopoly of control over a massive portion of the government's operation.135 Many areas of government feature competition because of the presence of multiple, semi-independent bureaucratic units, including the three branches of the military, the various agencies dealing with agriculture, and the many agencies focused on urban problems. The absence of competition tends to imply less efficient performance. Third, reorganization creates considerable uncertainty for individual bureaucrats by changing career patterns and promotion possibilities. Bureaucrats who lose power, authority, and promotion possibilities are far less likely to work toward the new goals sought by the reorganization. Bureaucratic rivalries within a single organization can often be counterproductive when one group seeks to promote itself over another. Fourth, some scholars suggest that increasing the presence of levels of hierarchy slows down bureaucratic responses to legislative signals designed to control an agency's work through the budget process. Politicians trying to insulate policy from legislative control can therefore use layers of hierarchy to frustrate lawmakers' control of agency actions.136 Fifth, on a pragmatic level, reorganization and centralization are likely to decrease efficiency in the short run, as addressing the transaction costs of combining computer systems and designing compatible operating procedures and routines across formerly separate agencies takes considerable time and effort.

While these subtleties may not be apparent to the mass public, particularly not in the midst of a crisis, they suggest that centralization is likely to have two separate effects heightening the difficulty of achieving prescriptive benefits from reorganization. On the negative side, centralization creates two different categories of problems. First, centralization involves large transaction costs as separate agencies merge their activities. second, greater centralization creates a series of other costs: the agency becomes harder to manage, and sheer size works against the ability of leaders to coordinate so many separate activities; centralization also creates monopolization that reduces incentives for bureaucracies to efficiently carry out their supposed purposes; and new hierarchies diminish incentives for those bureaus that lose stature. Moreover, both sets of costs rise with the scope of centralization and reorganization.

On the positive side, reorganization holds promise for greater coordination of effort, potentially allowing the efforts of many previously separate agencies to add up to a whole that is greater than the sum of its parts. Because of the nature of the costs involved in large organizations, the uncertainty surrounding the potential benefits of reorganization rise with the scope of the reorganization.

These two opposing effects imply that the prescriptive question of how to structure bureaucracies to accomplish particular goals is a complex one, made all the more so because the full constellation of relevant interests rarely agree completely on what those goals should be. Reorganization could be complicated from a policy perspective and could have counterintuitive effects.137

Although key actors trying to assess the long-term security payoffs of creating the Department would have faced considerable uncertainty, we do not believe that the probabilities of gains and losses are symmetrical. The evidence suggests that reorganizations as complex and massive as that creating the DHS would be beset by large costs with relative certainty, while the benefits are both more uncertain and less likely to outweigh the costs.138

C. Theoretical Conclusion: Policies May Not Be Designed to Succeed

In combination, the preceding principles imply that legislative allocations of bureaucratic authority over legal mandates are often, perhaps even typically, not designed to succeed at achieving their stated goals. Barring unusual circumstances, the need to pass legislation through a complex legislative process with many potential veto gates implies that a wide range of interests can hold up the legislation. Their price for allowing the legislation to move forward tends to be that the legislation's proponents alter the proposal in a way that advantages those holding veto power. Unless they choose unusual and difficult-to-sustain strategies, Presidents and legislative leaders are unlikely to jeopardize political gains. In particular, they are unlikely to insist on broadly unpopular but prescriptively attractive statutory details, or to oppose popular statutory changes made possible during a crisis that are prescriptively questionable but deliver partisan or long-sought policy benefits.139 Political competition thus routinely distorts the goals of the legislation. Legislation rarely addresses policy problems directly. Indeed, sometimes it is designed to fail. Moe's conclusions about the design of the Occupational Safety and Health Administration (OSHA) during the early 1970s are instructive, and likely generalize to a host of agencies and legal mandates:

While [the creation of OSHA] had the appearance of a systematic attack on the problem, in fact it was an administrative nightmare that did a thorough job of protecting business's interests. Authority was divided among an independent board, the secretary of labor, the states, HEW [the Department of Health, Education, and Welfare], and the courts. This would create confusion, lack of coordination, and multiple veto points. No one was in charge, and the secretary of labor, in particular, was kept weak.

The problems associated with OSHA also arise frequently elsewhere in government. Writing and passing of legislation creates pressures that often weaken the link between the goals of legislation and what the legislation actually does. The distributive tendency reflects legislation's tendency to benefit a wide variety of districts rather than concentrate resources where the problem lies. Multiple veto points allow a wide variety of legislators to hold legislation hostage in an effort to gain favorable adjustments. Indeed, the general need for legislative compromise means that almost all legislation is incoherent: the different legislative components often work at cross purposes whereby one section promotes a particular goal and another qualifies and limits the ability of an agency to attain that goal. These tendencies are aggravated by a mismatch between congressional committee jurisdiction and the organization of a bureaucracy. Put simply, the greater the dispersal of jurisdiction, the less coherent bureaucratic policymaking will be. Finally, partisan electoral goals often exacerbate difficulties associated with Grafting legislative solutions. Successful efforts to harness statutory substance, budgets, bureaucratic structure, and congressional oversight in genuinely addressing a difficult policy problem must therefore overcome substantial (though perhaps not insurmountable) challenges.

By placing the focus beyond position-taking and emphasizing the role of crises and the intricacies that impact the real-world effects of reorganization, our theoretical approach builds on existing theories of how legislators allocate bureaucratic jurisdiction over legal mandates. We posit not only that reorganizations are driven by politics, but that they can be used to further presidential control and policy objectives in specific ways-such as by reshaping the regulatory missions of bureaus, or redistributing authority from career officials and bureau heads to political appointees. Our challenge is not just to the questionable assumption that reorganizations would naturally tend to achieve prescriptively desirable objectives, but also to the theories that primarily emphasize position-taking or the role of politics generally without considering what policy or distributive goals are actually being achieved and through what structural techniques they are being achieved. As the next Part chronicles, each of these tendencies shaped the political bargains that resulted in the creation of DHS.

III. APPLYING THE THEORY: THE ARCHITECTURE OF THE DEPARTMENT OF HOMELAND SECURITY

Our theory of the political-bureaucratic system provides a lens through which to interpret the legislative developments that forged the architecture of DHS. In this Part, we use the theory to better understand the statutory and institutional features that have become characteristic of the Department. In the process, we revisit our two principal questions concerning the Department's formation: Why did the President change his mind to propose reorganization as a means of mega-centralization? And, why did the President create a DHS that contains so many bureaucratic units, many of which are only tangentially related to homeland security? The answers to these questions turn out to be intimately related. They can be understood by reviewing the intersecting effects of legislative influence, presidential choices, political responses to crises, budgets, and bureaucratic structure in this particular context.

A. The Political Influence of Congress

Congressional influence permeated nearly every aspect of the HSA legislation and the bureaucratic machinery it spawned. The Administration's bill was short-fifty brief provisions in comparison with the congressional legislation that included two hundred provisions-many of them in the convoluted legislative lexicon characteristically associated with lawmakers' desire to control the bureaucracy.141 The difference between the President's proposal and the final congressional legislation reflects more than just filling in details and gaps; it reflects the effects of the congressional politics engineering the new bureaucracy to serve the interests of its members; that is, to conform with the political-bureaucratic system.

Each of the principles discussed in the theoretical Part applies to this case: goal distortion and the distributive tendency; multiple veto points forcing alterations in the legislation; intracongressional committee jurisdictional issues; electoral goals of the majority party against the minority party; uncertainty about the reorganization; and the role of the crisis. As the theory predicts, these principles add up to a set of policies that are not obviously designed to obtain the stated objective of homeland security.

Consider, for example, the impact of goal distortion as it plays out through the distributive tendency. Calculating the optimal allocation of funds is a complex task. As Robert Powell suggests, this calculation must take into account a wide range of characteristics, including: (i) estimations of risk, themselves subject to uncertainty, such as the differential risks associated with targets in high profile cities, in places such as New York and Washington; (ii) factors that reduce risk everywhere, such as increased border security; and (iii) the notion that making one target far more secure makes the next most vulnerable target more attractive to strategic terrorists.142

Despite the difficulty with creating an optimal spending plan, nearly everyone agrees that the allocation of funds should be based on the factors noted above, especially assessments of differential risk. Yet, as noted in the theory, spending money according to the optimal factors often implies a high concentration of funds in particular districts, making these programs less popular in Congress. The congressional tendency is therefore to alter the criteria for spending in a way that spreads the money around, even at the expense of efficient pursuit of the legislation's goals. And this seems exactly what has happened.

DHS Secretary Michael Chertoffs statements on January 4, 2006 provide evidence for this story. In response to sharp criticism, Chertoff made the astounding public announcement that thenceforth the Department would base part of its homeland security grant allocation on risk factors (DHS could not allocate all the funds based on objective factors because the legislation required minimum-percentage spending in every state). Chertoff announced new rules about the distribution of such funds based on the risk of terrorist attack to thirty-five urban areas deemed especially vulnerable. Chertoff also stated that homeland security grants are "not party favors to be distributed as widely as possible," thereby suggesting that the previous approach to distributing grants amounted to such "party favors."143 This admission acknowledged that the Department had not based its assessments on risk factors prior to this time.144

Indeed, a wide range of commentators imply that some DHS spending may have degenerated into another source of congressional pork, especially through spending money in rural states with relatively low risks of terrorist attack.145 While the port of New York and New Jersey is widely regarded as at the highest risk, it received only $6.6 million in FY 2005, about equal to Memphis and far behind Houston's $35.3 million.146 The attempts to renew the Patriot Act in late 2005 witnessed attempts in the House to place greater emphasis on risk factors and to lower the guaranteed minimum percentage going to each state from the prevailing 0.75 to 0.25 percent. The Senate, with its greater rural bias, beat back this plan so that the original law would prevail.1 This topic may well be revisited as Congress reconsiders the Act's renewal. Similarly, an analysis of per capita homeland security grant spending for FY 2003 and FY 2004 indicates that in both years Wyoming-the best-funded state-received $35.30 and $37.74 per capita, respectively. New York State, on the other hand, received only $5.10 and $5.41 per capita in each respective fiscal year.148 The small-state bias seems rooted in legislators' distributive interests, filtered through institutions enhancing the political power of small states, rather than in meticulous analyses of why such funding should be allocated to Wyoming or similar states.

As the theory suggests, legislators' distribution of federal funds reflects a common pool problem: while all are better off from a homeland security program that fulfills its objectives, each is better off if his or her district gains a bigger share of the total. When all members seek greater funds for their districts, however, the consequences can be enormous. Members of Congress have greatly hindered DHS's ability to address the pressing problems of terrorism in America by prescribing constraints on spending that have little or nothing to do with homeland security and everything to do with their reelection prospects.

Another policy realm where legislators' parochial concerns seem paramount to security concerns is the creation of the so-called "Homeland security Centers of Excellence." Each such "HS-Center" has received a grant of between $10 million and $18 million over a three- or five-year period to study topics ranging from network analysis to the economic consequences of terrorist attacks.149 Instead of creating a mechanism for choosing center locations on the basis of defensible analytic criteria, the structure set up by the statute essentially dictated the location of the new centers. The Department created six HS-Centers located at the following universities: Johns Hopkins, the University of Southern California, Texas A&M, the University of Minnesota, the University of Maryland, and Michigan State. One would expect that these centers would be created in the districts of legislators facing a risk of losing committee jurisdiction as part of the transfer. As Appendix Table 2 indicates, it is telling that the statute happened to place the centers in jurisdictions where at least one lawmaicer was in danger of losing committee jurisdiction, a pattern that suggests the outlines of a potential logroll.15 Although such a loss of jurisdiction is almost never welcome among legislators, funding for the new homeland security centers may have served as part of the political exchange to increase support for the new legislation among members facing the prospect of diminished jurisdiction.

A third example of political bias in the distribution of funds concerns the new structures for the transfers of funds. After 2001, Congress slightly reduced funding for natural disaster grants and dramatically increased funding for counterterrorism grants.151 One example of an explosion in grant funding can be seen with grant opportunities provided through what was once the Department of Justice's tiny Office of Domestic Preparedness. The office was transferred to DHS, and thereafter its grant-making abilities have grown exponentially. In FY 1998, the Office of Domestic Preparedness awarded $12 million through a single grant program. As Appendix Table 3 indicates,152 by FY 2003, the office was in charge of meting out funds in seven separate programs, each ranging in total funding from $65 million to $1.5 billion.153 Far from reluctant participants in this growth, legislative majorities voted to fund the grants well beyond what the President requested-adding over $800 million to the President's request in this category for the FY 2004 budget.154

While the funding of preparedness and research has proven to be an important aspect of the new Department's activity, the legislative process did more than simply inject distributive concerns into the architecture of DHS. It also diluted the extent to which lawmakers considered the prescriptive merits of reorganization, or reorganized the internal distribution of legislative committee jurisdiction to realize those benefits.

The frequently stated rationale for creating the massive Department was coordination. Yet the prescriptive benefits of reorganization are highly uncertain. Centralization creates a far more massive organization, implying that organizational leaders have much greater difficulty mastering the various pieces. Department leaders' difficulties in managing FEMA and its natural disaster mission have made this plain, particularly in the aftermath of the Katrina disaster.155 Centralization may also contribute to greater monopolization of functions within the government, yielding less competition among bureaus and dissipating the potential benefits of competition. Reorganization creates considerable uncertainty about future career paths for bureaucrats. Those whose futures have been downgraded or who face the most uncertainty are most likely to work less hard or leave the agency. This too has become evident in FEMA, as many of its former employees simply left the Agency.156 Finally, substantial short-run costs arise from centralization as agencies undergo the transaction costs of integrating personal, information, financial, management, and field systems. These problems raise serious questions-almost entirely neglected in the legislative process-about whether the massive centralization of DHS has had a net increase in the effective provision of homeland security.157

During most of the Department's history, legislators also neglected the mismatch of congressional jurisdictions and bureaucratic centralization. Although the reorganization made massive changes in bureaucratic organization, Congress declined to engineer parallel changes in congressional oversight. The House Judiciary Committee is particularly illustrative. Among the Judiciary Committee's many amendments, the committee voted to transfer only the law enforcement functions of INS to DHS, keeping the citizenship functions at the Department of Justice (DOJ) and, obviously, under the purview of the committee. Besides retaining their oversight functions, the Judiciary Committee also voted to increase their responsibility by approving an amendment to transfer the secret Service and the Federal Law Enforcement Training Center from Treasury to the DOJ.158 The House leadership did create a special nine-member Select Committee on Homeland security in July 2002 to screen the changes made by individual committees,159 but House Speaker Dennis Hasten largely populated this panel with committee chairs who could protect their jurisdiction.160 Legislators also showed significant resistance at the time the HSA was passed. President Bush reportedly made some early attempts to encourage Congress to solve the jurisdictional issue,161 but legislators essentially ignored him.

The massive bureaucratic reorganization unaccompanied by any congressional reorganization implies that the structure of bureaucratic incentives induced by congressional oversight works against the effects of centralization. This problem raises another variant of the congressional common pool problem: though all members may have wanted to achieve improved homeland security coordination, they also sought to control a piece of the bureaucracy. Doing so, legislators could claim credit for steering policy in a highly salient domain, and for steering funds in directions benefiting constituents. Many members of the relevant subcommittees have specialized in helping existing constituents of the agencies being moved to DHS. To the extent that coordination lowers the level of service legislators could provide to their constituents, these members are likely to use their oversight jurisdiction to impede coordination.16

Because the creation of DHS made massive changes to the bureaucracy while leaving the existing structure of congressional jurisdictions in place, congressional incentives cut against the goals of centralization and coordination.164 The piecemeal set of congressional jurisdictions reflected the old set of priorities; in particular, a set of agencies that did not coordinate. Much of the lack of coordination under the old system represents a set of diverse agencies serving diverse constituencies overseen by a diverse set of subcommittees. Leaving the old congressional jurisdictions intact allows representatives of the old, uncoordinated system to pull their agencies away from the coordination-related goals of the new system and to continue to serve their old constituents' interests. In addition, leaving the existing distribution of committee jurisdiction yields another problem interfering with the potential benefits of centralization: DHS's leaders must now report to all of the separate congressional committees, depleting the time they can devote to coordinating. In a less sprawling reorganization, such reporting demands might prove less problematic. Not so with DHS; during the first half of 2004, Department officials were called to testify-on average-almost twenty times a month.165

Yet most of these prescriptive concerns-whether about the merits of centralization, or about the need to overhaul congressional committee jurisdiction in order to capture the benefits of centralization-were cast aside. Instead, senior legislators from both parties maneuvered to preserve their committees' power. And the parties themselves competed in trying to take credit for an impending reorganization with growing public salience. In the process, Republicans appear to have baited the Democrats on several issues, notably the drug liability and civil service exe