The Contemporary Presidency: Executive Orders and Presidential Unilateralism
Rudalevige, Andrew, Presidential Studies Quarterly
In the last decade or so, students of the American presidency have renewed their interest in the formal authorities and unilateral possibilities of presidential power, driven both by methodological logic and by events. On the theoretic side, scholars working within the broad framework of the "new institutionalism," especially its rational choice variant, have made a case that the formal, legal, and organizational aspects of the presidency--and the incentives and constraints for presidential behavior these implied--had been too long neglected in favor of impressionistic accounts of the "personal presidency." A focus on the formal powers that underlay the presidential office, and the way these could be used to enhance an incumbent's influence, was needed to fill that gap (e.g., Howell 2003; Kelley 2007; Moe 1985, 1993; Moe and Howell 1999). After all, as Kenneth Mayer argued (2001, 11), "in most cases, presidents retain a broad capacity to take significant action on their own, action that is meaningful both in substantive policy terms and in the sense of protecting and furthering the president's political and strategic interests."
The assertive--even "imperial"--stance taken by recent presidents provided empirical grist for this mill. President George W. Bush was particularly notable in acting aggressively to expand his office's powers vis-a-vis other political actors (Cooper 2002; Goldsmith 2007; Rudalevige 2005, 2010; Savage 2007). Redressing the perceived constriction of the presidential office after the Watergate/Vietnam years provided a new rationale for unilateral command--even before the terrorist attacks of September 11, 2001. Barack Obama, while disavowing some of his predecessor's rationales, has acted in a similar manner in a number of areas. The assassination of American citizens acting with al-Qaeda in Yemen; the evasion of the War Powers Resolution in Libya; the use of the state secrets act in fending off judicial inquiry--all these suggest a continuing approach to presidential authority that overrides shifts in the incumbent's personality.
From either direction, the upshot has been important recent work on a presidential administrative toolkit that includes appointments (Lewis 2008), signing statements (Evans 2011; Kelley and Marshall 2010; Korzi 2011), executive agreements (Krutz and Peake 2009), proclamations (Rottinghaus and Bailey 2010; Rottinghaus and Maier 2007), rulemaking and guidance (Graham 2010; Kerwin and Furlong 2010), and especially executive orders (Gibson 2009; Howell 2003; Mayer 1999, 2001; Rodrigues 2007; Warber 2006; Wigton 1996). Indeed, at this point it is safe to say that a standard textbook in the field could not--as it did even after Watergate--exclude "executive orders" and "signing statements" from the index (Koenig 1975). The study of the contemporary presidency thus requires serious attention to that office's executive authority.
The aim of this article is to do just that while extending and complicating the empirics and implications associated with unilateralism. It takes seriously the conclusion drawn by Professor James Hart some 85 years ago. "In theory, but only in theory, the President controls all executive formulation of policy, all ordinance making," Hart wrote in The Ordinance-Making Powers of the President (1925, 297). "In practice the heads of departments may themselves merely accept ordinances drawn up by bureau chiefs, while the President may have only passing knowledge of the whole affair." The data presented here are preliminary--but serve as cautionary correlatives of Hart's point and as a reminder that the "old" institutionalism still has lessons for its heirs.
Unilateralism, in Theory
We know a fair bit of the "what" regarding executive directives, especially executive orders. That is, we have a literature that explores in some depth the timing of order issuance and some idea of the systemic factors, many of them legislative in nature, that explain when it is issued--in the maximum likelihood sense of calculating how those external factors increase or decrease the likelihood of that issuance actually occurring. A variety of hypotheses have been tested quantitatively in order to show the impact of such contextual variables as unified vs. divided government, the size of a partisan majority, shifts between presidents of different parties, ongoing campaigns, and presidential popularity (see, e.g., Deering and Maltzman 1999; Howell 2003; Krause and Cohen 2000; Mayer 2001; Warber 2006).
But the essence of the "why," as contained in the decision-making process itself surrounding the order, is a black box. Since the main unit of analysis is the fact of an order's issuance, orders tend to spring from a sort of immaculate conception.
Again, this is more than fair enough, given the aims of the literature to date. Further, it matters little so long as we can safely assume that presidential issuance of an order reflects presidential preferences, more-or-less purely enacted--an assumption commonly extended here. For instance, that the presidency and the executive branch can be treated as a single, unitary actor is what, Howell argues (2003), distinguishes unilateral power from all other sources of presidential influence. (1) Terry Moe puts it most starkly: unlike Congress, he says, "presidents are not hobbled by these collective action problems and, reigning supreme within their institution, can simply make authoritative decisions about what is best" (1995, 417). Such a claim, imbued with moral authority by political actors like Justice Antonin Scalia (in his dissent to Morrison v. Olson ) and scholars like Steven Calabresi and Christopher Yoo (2008), underlies "unitary executive" theory as it has advanced over time. This notion of collective action, or rather of its absence, goes back to Alexander Hamilton's claim of "unity" as a key component of executive energy.
We tend to take it as given that an executive order, which, after all, is a command to some part of the executive branch, shapes the actions of that branch to conform to the president's will. A recalcitrant department is told to shape up and fly right, told to admit gays to the military, to enforce integrated housing policies, to submit their proposed regulations for White House scrutiny. And surely this sort of order is issued. Is it the norm? Adam Warber rightly observes (2006, 8) that "the presidency literature contains no studies addressing the importance of persuasion in relation to the president's executive order authority because scholars consider executive orders as a tool of command." The notion of unitary action has great utility as a working premise. But we need to be careful about how far it is extended.
After all, is it really safe to assume that the executive branch is a single actor? Most of the history of public administration is predicated on the notion it is not (see, e.g., Seidman and Gilmour 1986). Can we safely assume that the president incurs no transaction costs of collective action when dealing with the bureaucracy, even when dealing with the presidential branch's own "counter-bureaucracy" (Nathan 1983) within the Executive Office of the President (EOP)? Even Mayer (2001), in his magisterial book on the importance of executive orders, provides important cautions against that claim. The empirical literature suggests that some orders are not issued at all, in the face of interior objections (see Gibson 2009); that they often move policy closer to congressional preferences (Law 2011); and that many rely for their content on internal and external coalition-building (Morgan 1970; Rodrigues 2007; Wigton 1996). Dickinson (2009, 757) argues that "the difference in administrative and legislative policymaking is not that it signifies a shift between a 'unilateral' and a 'multilateral' process. Instead, it is a change in where, and with whom, bargaining takes place."
Indeed, an established body of literature suggests that even within the EOP, coordination has clear managerial limits. This is true even when everyone concerned is well meaning. And the relationship grows more fraught as we move outwards from the White House complex (e.g., Dickinson 1997; Heclo 1977). In a branch that collects nearly three million persons within its purview, we can expect collective action problems. Richard Nixon's aide William Satire (1975, 246-47) called the executive branch a "lethargic behemoth" that presidents struggle to control: "the so-called Chief Executive," he added, "can tug and haul all day and never rip up the bureaucracy." More analytically, Krause (2009, 75) notes that "analyses grounded exclusively in formal executive powers understate organizational complexity, and thus overstate presidential capacity for controlling the bureaucracy." Recent cases in this vein might include Bill Clinton's 1993 effort to end the ban on gays' open service in the military, George W. Bush's program of domestic surveillance (including a dramatic showdown between Department of Justice and White House staff in Attorney General John Ashcroft's hospital room), or Barack Obama's so-far unheeded order to close the detention facility at Guantanamo Bay. To take a different sort of example, consider one of Gerald Ford's first executive orders. It dealt with the early promotion of outstanding army officers--but Ford (or even his staff) had nearly nothing to do with it. It had been proposed in July 1974, and sent to the White House later that month, well before President Nixon's resignation. (2)
These examples hardly prove that presidents have no power of command (even Neustadt  did not argue that). They do, however, speak "to the illusion that administrative agencies comprise a single structure, 'the' executive branch, where presidential word is law, or ought to be" (Neustadt 1990, 33-34). They suggest the value in breaking up the "unitary executive" into the fragments and routines with which presidents must daily contend. Those lie across the executive branch, and even within the Executive Office. As such, the case moves in rather parallel fashion to the shift traced by Graham Allison when he wrote about the 1962 Cuban Missile Crisis. Beginning with a model that assumed a unified rational actor, Allison discovered this left odd but important details unexplained. To understand the nuance of what happened over those "thirteen days" it was necessary to bring in two other models more closely associated with bureaucratic policy making, dealing with the standard operating procedures relied upon by agencies and the intragovernmental politicking they engaged in (Allison and Zelikow 1999).
I hasten to add that as with Allison's Model I, so with executive orders: simplifying assumptions provide parsimony that oftentimes provides valuable clarity. The complications I introduce below are supplements, rather than substitutes, for such work. Further, such a discussion hardly rules out other routes to presidential influence, even in the Neustadtian sense. As Mayer and others make clear, executive orders are often about establishing process rather than effecting immediate policy change. (Think, for instance, of processes governing regulatory review or setting up advising systems.) This fits Howell's (and Alexander Hamilton's) emphasis on the "first mover" strategy.
Thus, the permutations of power play out along crooked paths. Nonetheless, there seems wide scope for extending the current literature. If the logic above is correct, we should expect that the provenance of orders contains more consultation and less fiat than currently assumed.
A logical place to test this is the universe of executive orders. As Mayer argues, "executive orders have played a critical role in the development and exercise of presidential power"; and of the various tools, "executive orders combine the highest level of substance, discretion, and direct presidential involvement" (2001, 31, 35). Further, they are (mostly) matters of public record. (3) Finally and usefully, executive orders are aimed within the administration, thus relating directly to presidential control of the bureaucracy and questions of a unitary executive--both across the executive branch and within the presidential branch.
Unilateralism, in Practice?
Along these lines, consider Bill Clinton's issuance of E.O. 13045 on April 21, 1997. (4) E.O. 13045 was entitled "Protection of Children from Environmental Health Risks and Safety Risks." Children, President Clinton declared in the order, "may suffer disproportionately" from such environmental risks; thus, each federal agency was to "make it a high priority to identify and assess" them and to "ensure" they were addressed by "its policies, programs, activities, and standards." The executive order also created a task force cochaired by the secretary of Health and Human Services (HHS) and by the administrator of the Environmental Protection Agency (EPA) to monitor, gather, and publish research on environmental risks to children as well as an "interagency forum" convened by the Office of Management and Budget (OMB) charged with producing an "annual compendium of the most important indicators of the well-being of the Nation's children." Along the way it revoked a Reagan-era order requiring government-wide rulemaking attention to "family policymaking criteria," notably "the marital commitment." (5) Most critically, the order added a new hook to the regulatory review function conducted by OMB's Office of Information and Regulatory Affairs (OIRA). Agencies now had to justify to OIRA their "evaluation of the environmental health or safety effects of the planned regulation on children" and provide "an explanation of why the planned regulation is preferable to other potentially effective and reasonably feasible alternatives considered by the agency."
If viewed solely from the point of issuance--as an output--E.O. 13045 has many attributes of unilateralism, suggesting a president imposing his will on the executive branch. Centralized processes were created to reflect presidential preferences and priorities. A predecessor's preferences were set aside. (6) Bureaucrats resisted its imposition. Outside critics grumbled--a front-page Washington Times story (Wetzstein 1997) quoted riled representatives of the Heritage Foundation and the Family Research Council, while Senator Jeff Sessions (R-AL) introduced a bill to overturn Clinton's action. Neither gained much traction. (7) President one, agencies (and enemies) nil.
Or so it seems, until a closer look at the formulation of the order complicates that conclusion. In an August 12, 1996, memo to more than a dozen cabinet and White House policy staff,8 EPA administrator Carol Browner noted that agency would soon release a new National Agenda to Protect Children's Health from Environmental Health Risks.9 Various initiatives tied to that agenda were touted, since "protecting children from environmental health threats has been one of EPA's highest priorities during the Clinton Administration": one "opportunit[y] for additional action" by the administration, EPA suggested, was a "potential executive order" that would tighten the standards set by government regulation to be "protective enough of the potentially heightened risks faced by children." By January 1997, EPA--which is of course a cabinet-level executive bureaucracy, not a White House unit--had drafted such an order and sent it to OMB, which sent it out for wider comment and review. Some seventeen executive agencies and EOP staff offices became involved, as EPA meantime sought to "generat[e] support within the White House" for its text.
While agencies uniformly claimed to support the idea of protecting children from environmental hazards, objections to the mechanisms anticipated by the draft order arose immediately across the wider bureaucracy. Various staffs raised a number of objections, most colorfully perhaps the idea that the order put a "kick-me sign" on their own backs. That is, since it would have required them to evaluate not only the effects of the planned regulation on children, but also to explain if their proposal failed to protect children fully and to justify why it was issued anyway, it effectively directed opponents to the best grounds for a lawsuit. How, for instance, HHS asked, were they supposed to say why tobacco remained a legal product? Banning it would clearly be better for children's health.
A series of negotiations ensued over four months of meetings; "we have made significant drafting changes to accommodate concerns," Domestic Policy Council (DPC) staffer Diane Regas told her boss Elena Kagan in March. Those concerns came from within the EOP as well as the wider bureaucracy; indeed, a Kagan memo to White House chief of staff Erskine Bowles that same month noted that DPC, National Economic Council (NEC), Council on Environmental Quality (CEQ), and (and Technology Policy (OSTP) were all involved. It was not until late March that Regas could note that "I think we have resolved all the kids e.o. issues among WH offices." (10) And then, "serious last-minute objections" from the agencies remained (as Kagan told Bowles), even as EPA continued to press for the order in a meeting with key domestic staffer Bruce Reed. Notes by Kagan from an April 1 meeting note EPA's objections to Treasury "nervousness" and to others' continued queries: "we've redone [the order] to address concerns. Weakened already."
As it moved on, the draft order continued to be more and more tightly targeted. Agencies were to comply "to the extent permitted by law and appropriate," and independent regulatory agencies were merely "encouraged" to participate in the order's implementation. The contentious section requiring agencies to identify better alternatives was tweaked to make the justification more positive and less defensive (and therefore, not trivially, less work-intensive).
In the end, then, it was something of an understatement to note (as the decision memo sent to the president did, on April 11, 1997) that "the proposed Executive Order ... has been the subject of extensive discussion with affected agencies." Even as the president was urged to issue the order, several departments continued to press their reservations; and in his note approving the order, President Clinton requested still more changes aimed at addressing those objections--"might want to ease burden a bit," he scrawled, with regard to the scope of the analysis of the alternative routes not taken. One more revision thus ensued before the order was issued. (11)
From Anecdote to Data
If this narrative is typical, the issuance of executive orders follows extensive consultation across the executive branch and even the EOP--and suggests that presidents must cajole, rather than directly command, departments and agencies in pursuit of an order in a given form.
The remainder of this article seeks to probe that suggestion more systematically. Uniquely in the literature to date, it makes an attempt to discover whether the E.O. 13045 saga is, in fact, typical: we simply do not know. To do so, it utilizes data from the OMB and presidential archives that shed light on the provenance of executive orders in administrations ranging from Harry Truman to Ronald Reagan. Opening the black box shows us that the Clinton example above is broadly representative: the issuance of executive orders seems to involve as much about consultation as command. There are exceptions, of course; but this general finding has important implications for our conception of contemporary presidential power and how we conceive of "unilateralism" itself, in the context of a not-very-unitary executive branch.
Executive Orders: Definitions and Data
Executive orders are not, as is now well known, defined or even described in the Constitution; they "do not dwell amid the comfortable certainties of administrative law" (Bruff 2006, 154). But they have been utilized as an implication of the vesting of the "executive power" in the president from the Washington administration forward (see Mayer 2001, chap. 2, for background and context), and justified by an ad hoc variety of constitutional and statutory citation. The Congressional Research Service cites a widely accepted definition from a 1957 congressional study:
Executive orders and proclamations are directives or actions by the President. When they are founded on the authority of the President derived from the Constitution or statute, they may have the force and effect of law.... In the narrower sense Executive orders and proclamations are written documents denominated as such.... Executive orders are generally directed to, and govern actions by, Government officials and agencies. They usually affect private individuals only indirectly. (Burrows 2010, 1)
However, even an indirect impact can be important; by changing how stringently cost-benefit analysis is applied to regulatory review, or how strictly anti-discrimination laws are enforced by government contractors, for instance, presidents can readily influence private sector behavior.
The scale of that influence varies with the scope of the federal government. It is perhaps not surprising, then, that recent studies have found an upswing in the number of "significant" executive orders issued by presidents over time. Kenneth Mayer (2001), sampling just over one thousand orders from 1936 to 1999, and using a multistage selection process parallel to David Mayhew's treatment of significant legislation, found that about one in seven orders were significant. As a proportion of the total issued, significant orders jumped to 22% in the 1970s and to 28% by the 1990s, up from between 9 and 16% in the prior decades. (12) William Howell (2003), using a somewhat different procedure over the 1900-98 period, also found the number of significant orders issued each year had increased as the century wore on. Certainly by the late 1990s there was much gnashing of teeth by pundits and scholars alike over the scope and sweep of executive orders (e.g., Olson and Woll 1999); it is worth noting, though, that such complaints persist throughout the whole era of the modern presidency. The Kennedy White House, for instance, received so many letters accusing JFK of dictatorship in connection with a civil defense order that it was forced to develop a form letter defending the president. (13)
In examining executive orders a number of substantive categories recur, sometimes quite frequently. Mayer (2001) divided his sample into eight main subjects: executive branch administration, civil service, public lands, defense and military policy, foreign affairs, war and emergency powers, labor policy, and domestic policy. The relative proportion of orders falling within each category varies over the period of his study, with public lands orders seeing a sharp drop-off over time and foreign affairs, including trade questions, rising significantly. The EOP's own sense of how orders should be classified can be inferred from the indices OMB creates for its internal filing system. These change over time but, for 1983, for instance, it divided orders into categories comprised of the following: (14)
* [Advisory] Committees (23 orders issued)
* Delegations of Presidential Authority (10)
* Federal personnel, to include:
** Pay and Allowances (2)
** Military Compensation (1)
** Civil Service Retirement and Disability System (2)
** The Executive Schedule (2)
** Office of Personnel Management (1)
** Labor-Management Relations (1)
** Federal [Office] Space Management (1)
* Trade issues, to include:
** Export Administration Act (2)
** Beneficiary Developing Countries [a trade designation] (2)
** Arms Export Controls (1)
** Nuclear (1)
* Other economic issues:
** National Mediation Board15 (2)
** Minority Enterprise (1)
** Powerplant and Industrial (1)
** Violations of Anti-Trust Law (1)
* Public International Organizations (2)
* Diseases (1)
Generally symbolic matters (proclaiming a given week, or month, or day of remembrance) are reserved for proclamations, rather than executive orders (Rottinghaus and Maier 2007). But from the 1983 Reagan list, it seems clear that executive orders vary in their substance and significance. Discerning the latter, as noted above, is tricky. Should we only examine "significant" orders? Simply counting orders in the aggregate seems clearly insufficient--such an approach equates (just to take examples from the current administration) such issues as a ban on text messaging while driving federal vehicles with a reshaping of the terms under which terrorism suspects held at Guantanamo Bay might be held without trial. (16) On the other hand it is artificial to constrict the range of what executive orders, and thus presidents, actually do. If one of seven orders is significant, should our analysis bypass the remaining 85% of the order universe? As importantly, perhaps--given that the question of interest is one of intra-executive relations--it is tricky to define significance in terms external to the executive branch (i.e., by reference to an order receiving attention from another branch or from the media; cf. Howell 2003). For instance, E.O. 11602, issued by President Nixon in June 1971, appears on no such list--it simply provides for "Administration of the Clean Air Act with Respect to Federal Contracts, Grants, or Loans," in apparently innocuous bureaucrat-ese. But that order, when proposed, prompted a five-inch thick file full of denunciation and proposed revision--for the intended audience of government procurement officers (and for a wide range of potential contractors) this was a hugely significant order. Similarly, another Nixon-era order dealt with whether coyotes and other predators could be poisoned on federal land--sparking a massive outcry, a prolonged battle between ranchers and what one staffer imprudently called "eco-freaks," and at least two subsequent orders reaching into the Ford administration. (17) Likewise, exempting employees from the statutory retirement age is hardly something that will show up as important--Mayer omits these sorts of orders entirely from his sampling. But what if that employee is J. Edgar Hoover? (18)
At least at this stage, it seems better to cast the net widely to get a broad sense of the substantive universe of executive orders: any order that bears the president's signature, if only by virtue of its claim on his scarce attention, must be worth at least cursory examination. But it is certainly true that further research will need to control for differences between significant and clerical executive orders, both in formulation and implementation.
Exhuming Executive Orders
The data in this article come from archival sources that, to my knowledge, have not been utilized in previous treatments of the subject. As discussed below, the OMB (until 1970 known as the Bureau of the Budget, or BoB) has long been the key presidential player in supervising the formulation of executive orders. OMB keeps a file on every executive order and proclamation, documenting the formulation and clearance of each proposal and ending with the order's publication in the Federal Register. (19) Past files are available in the National Archives and Records Administration (NARA) facility in College Park, Maryland. From 1939 (when BoB moved to the EOP) until 1983, the executive order files seem to be more or less complete, if not always well indexed. At least some executive order files are available up to late 1987. These can be supplemented with material documenting internal discussions about executive orders held in various presidential libraries. (20)
The present stage of research draws on four decades of orders from the Truman, Eisenhower, Kennedy, Johnson, Nixon/Ford, Carter, and Reagan administrations, covering 1947 through 1987. This sweep provides a wide range of potential variation, in terms of partisan affiliation and scope of presidential agenda. Institutionally, it covers the pre-"Imperial" presidency as well as the "imperiled" version of the late 1970s, while the post-Watergate 1980s mark the resurgence of presidential efforts to assert their authority. In practical terms, as noted above, the Reagan years are the last for which archival records are currently available from NARA. (21)
In any case, while these data will surely need to be expanded backward and forward in time, they provide sufficient evidence to work with for the present. For each administration a group of executive orders was sampled at random. The target sampling frame was 30 orders per four-year presidential term (thus, 30 for Carter, say, and 60 for Reagan). To date, sufficient information has not been found to support the confident coding of every item in the overall sample; thus, n will vary by president. Still, at present, total n = 293, well large enough to give a sense of whether the intuitions above are worth pursuing more systematically (see Appendix A for the distribution of the sample by administration).
The Formulation Process
The files reveal a template for the consideration of executive orders. Attention to the stylistic formatting of executive orders had begun in the 1920s, most notably with Herbert Hoover's 1929 effort (via E.O. 5220) to standardize punctuation and place names and to require a certain number of copies on a specific size of paper (8" by 12 1/2" for the record.) But it was Franklin Roosevelt's order in August 1933 (E.O. 6247) that created a standard process for their formulation. FDR decreed that "the draft of an Executive order or proclamation shall first be submitted to the Director of the Bureau of the Budget," and if approved at BoB, to the attorney general. The latter was given the routine--though hardly unimportan--job of analyzing the order "for form and legality," which the attorney generally normally delegated to Justice's Office of Legal Counsel.
A follow-up order three years later strengthened the veto points at BoB and Justice: "If [the proposed order] is disapproved by the Director of the Bureau of the Budget or the Attorney General," read E.O. 7298, "it shall not thereafter be presented to the President unless it is accompanied by the statement of the reasons for such disapproval." The standard process for receiving BoB approval mirrored the system already in place for the president's legislative program, and for many of the same reasons (Wann 1968). A letter went to any executive agencies BoB thought might be affected by the issuance of a given order, asking for their comments. BoB staff assessed the feedback returned--ranging from strong opposition or support to baffled apathy--and, as appropriate, edited the order or asked the originating agency to defend the extant draft or produce a revised one.
In 1962, several additional changes were made to reflect administrative experience with this process. John F. Kennedy's E.O. 11030, entitled "Preparation, Presentation, Filing and Publication of Executive Orders and Proclamations," required that departments and agencies seeking issuance of an order had to include documentation "explaining the nature, purpose, background, and effect of the proposed Executive order or proclamation and its relationship, if any, to pertinent laws and other Executive orders or proclamations." This should be done, BoB general counsel Arthur Focke argued, because the BoB staff (and through them, the president) frequently did not have "from the agencies adequate information in support of the proposed order.... Such information has frequently been meager and has necessitated requests for additional material." (22)
E.O. 11030 still governs how executive orders are drafted. (23) And it should be stressed that a basic assumption of the process is that such orders are normally proposed by actors outside the EOP. Indeed, the process was put in place to prevent the president from being forced into what Dick Cheney once termed "oh, by the way ... " decisions, made on the fly in informal bilateral encounters with administration officials (Sullivan 2004, 104). A 1962 order to modify eligibility for the armed services' Ready Reserve is a good example. The BoB file consists of a single index card, which notes that the order was not cleared by the BoB--or anyone else. The explanatory text placed in the file reads, rather plaintively:
The Secretary of Defense brought the order to the White House ... and took it back to the Pentagon with him after obtaining the President's signature. Someone in his office then transmitted it to the Federal Register ... the next morning, when we became aware of its existence for the first time. A request has been made of the Defense Department for any file or papers that may have accompanied it but as yet none have been received. This [card] constitutes the file so far. (24)
In this instance, McNamara and the Defense Department evaded the long months of clearance that had been part of an earlier effort by Treasury to amend the Ready Reserve rules as they pertained to the Coast Guard, then a part of that department. Responding to that sort of outcome, BoB stressed to Kennedy aide Ted Sorensen the need for an orderly process that would allow all aspects of an issue to be considered before an order was "presented to the President. In most cases," Focke went on, "a proposed order ... handled under this procedure can be issued more expeditiously, and with greater protection to the President, than one which is presented and processed outside the normal channels." (25)
In early 1969, likewise, Focke was at pains to explain to the incoming Nixon administration the value of clearing orders through BoB, rather than issuing them direct from the White House. (26) He reminded them that this would help the president learn of budgetary, management, and organization implications raised by a draft order, and provide "the best judgment of the Administration as a whole." It would also avoid "the confusion and embarrassment," which could result from endorsing a request without wider coordination and consultation. Chief of Staff H. R. Haldeman, always nervous that the bureaucracy would try to get one over on the new Nixon insiders, tersely noted in reply that "the procedure outlined by Mr Focke ... should become standard procedure immediately."
In a unitary executive it would surely be notable that the president needed "protection" from the agencies, or that he needed procedural insulation against being manipulated by his own executive branch.
To be sure, the files emphasize a good deal of variance in the formulation of executive orders. Some orders are drafted directly in the BoB/OMB, and others in the White House proper: examples of various originations are laid out in the next section. And OMB did, to be sure, have clear opinions as to the utility and quality of any given order. But even White House-driven orders were subject (usually) to clearance and consultation--agencies were given notice of the proposed order, invited to comment, and their comments (sometimes) heeded. OMB's goal seems to have been centered on gaining consensus--on ensuring that the wider executive branch agreed, to the extent possible, on the text of an order moving forward. This could spur a long process of sequential appeasement. In one 1962 case, the Department of the Interior produced a draft order (complete with a draft letter from the president to the secretary of the interior, directing him to submit such an order), in mid-February. The order gave the Interior Department the authority to coordinate administration policy with regards to outdoor recreation. The BoB had some qualms, made some changes, and wound up sending out for clearance both its own draft and Interior's original. Other agencies, most vehemently the Department of Agriculture, objected to what they saw as Interior's power play: "this Department has consistently opposed" such proposals, shot back Agriculture Secretary Orville Freeman, attaching a redrafted order that downgraded Interior and elevated Agriculture instead. "We believe our position to be fundamental."
A long back-and-forth ensued; by the end of March, BoB staff reported that all objections had been met except for Agriculture's--and that Agriculture was, not to put too fine a point on it, wrong. An assertive EOP would presumably have decided the matter there. Instead the leadership of BoB did not cease striving for consensus and asked for another draft. The next offer was of a BoB-brokered summit meeting between Freeman and Interior Secretary Stewart Udall (to include Sorensen's deputy Lee White at the White House)--a placating letter from BoB deputy director Elmer Staats to Udall "hope[d] that something along the lines of the attached might meet as far as possible the different views of the two agencies in this order." And by mid-April the logjam was finally broken, with Agriculture and Interior effectively rotating the chair of the new coordinating committee the order established. As well he might have, Freeman sent Staats an appreciative note: "Thank you for your patience and thoughtful consideration of our points of view." (27)
In a similar instance during the Ford administration, this time dealing with the delegation of functions under the Foreign Assistance Act, an OMB staffer noted to another that "this matter is at such a relative impasse between State and Treasury that we recommend the following course: Return the file to Treasury, with agency letters, indicating State's counter proposal and requesting Treasury's reaction thereto. Then, let's see what happens." Two months later, the impasse not resolved, the recipient noted, rather wistfully, "we could, of course, kill the whole matter by requesting DOD [the Defense Department] to come up with a complete rewrite." Another three months later, and the order was finally issued, more or less in Treasury's favor. (28)
As an addendum it is worth noting that as the White House and EOP staff grew, EOP staff units--the CEA, the National Security Council (NSC), the CEQ, the Domestic Policy Staff (by whatever name), and the like--became regular recipients of clearance letters as well, feeding into the order drafting process. Far from showing a unified presidential front, the EOP units often proffered divergent opinions. Nor were their "internal" comments always privileged over departmental views. In one tortured iteration of this with particularly high intra-EOP transaction costs, the consideration of an order (ultimately issued as E.O. 11742) blew up into what one Nixon OMB staffer called a "tempest in a teapot" when the CEQ and NSC loudly misunderstood (OMB thought) the import of an order delegating authority to the State Department. (29)
A Systematic View of Executive Orders, 1947-87
This template seems compelling but again does not show that these cases are typical. For data we need, at the least, plural anecdotes. Thus, we now turn to what can we say more systematically, albeit in relatively preliminary fashion.
The basic analysis here is built on the sample of 293 executive orders described above. It adjudges the provenance of each order: where was it predominantly drafted and at whose request? The basic question of interest is whether an order is "presidential" or whether it came from the outside (i.e., the wider bureaucracy, though sometimes, intriguingly, from Congress.)
The data suggest three major categories. A brief example of each may help clarify the coding scheme.
First are orders that are prepared by OMB/BoB itself, sometimes prompted by the White House. Often these deal with advisory arrangements. For instance, E.O. 10938, which established the President's Foreign Intelligence Advisory Board in 1961, was prepared in BoB from a draft that originated in McGeorge Bundy's office at the NSC; E.O. 12412 in 1983, which reconstituted a Peace Corps Advisory Council was, likewise, Director David Stockman noted, drafted "in this office, at the request of the White House Office." E.O. 12350, which terminated the Urban Impact Statements imposed on agencies by Jimmy Carter, was prepared rather straightforwardly in OMB, since it revoked an OMB circular the agency thought had "not made a useful contribution to the decision process and is redundant of other agency or OMB analysis." (30) Sometimes multiple orders are issued in quick succession by OMB, usually to extend deadlines that have not been met--in 1948-49 five different EOs were issued to amend another one requested by the Defense Department, each time giving the department more time to develop regulations required by a recent statute. (31)
Other orders are prepared elsewhere in the EOP, such as the White House Counsel's Office, the NSC, or even the OSTP. In this category, I include orders that were drafted by a department or agency but on the clear directions of the White House rather than on their own initiative. In 1982, for example, the Department of Transportation drafted an executive order (later E.O. 12358) creating a Commission on Drunk Driving. While enthused about the work of the commission, Transportation Secretary Drew Lewis noted it was submitted "in response to a request from the President"; Stockman reminded the president after the order was finalized, with OMB and WHO input, that it "was proposed by [DOT] to implement your decision to establish such a Commission." (32)
These first two categories can be seen as "unilateral" in the way we normally think of it. But there is another set of orders as well: those drafted in the departments and executive agencies, either at their own behest or--quite commonly--in response to congressional action, either because legislators requested an action or because they demanded it in statute. Agency-initiated actions include such things as changes to internal operating rules (as when Harry Truman authorized the armed forces to suspend the eight-hour workday for its employees or when Ronald Reagan redefined hazardous pay guidelines). (34)
Sometimes Congress--or a key member of Congress--makes a political request that winds up as an executive order. Demands that the President order the IRS to allow an investigating congressional committee to see tax returns are routine. But other more idiosyncratic requests creep in too. For instance, in the spring of 1972, Sen. John Stennis (D-MS) asked the Nixon Justice Department to allow the US Marshal for the Southern District of Mississippi to stay in his post past his 70th birthday, when he would normally have to retire. The Marshal, Jack Stuart by name, did not even bother to ask for this exemption through formal channels; as Stennis noted, "Mr. Stuart and I have been close personal friends since high school days"--and Stennis was chair of the Armed Services Committee, in a presidential election year. This was something the chairman wanted-"again, let me express my deep interest in favorable action"--and he got it. (35)
But orders prompted by statute are more common, and indeed make up the bulk of the sample. Often they involve delegating authority granted in the new law from the president to a given agency (as in the 1974 State-Treasury imbroglio noted above.) More contentiously, agencies may also seek to gain that sort of authority through their own initiative (as in the Agriculture-Interior spat noted above). Or they may be more substantive in seeking to implement a law. For example, E.O. 12340, entitled "Court Martials," amended court martial procedures and regulations, prompted directly by the Military Justice Amendments of 1981 (Public Law 97-81), which had a rapidly-approaching effective date.
The results of this first cut at the sampled orders are presented in Table 1, above. It makes clear that departments serve as the major source of executive orders: more than six in 10 come from departments rather than from the president proper or via his staff at the OMB.
This 60-40 (indeed, nearly 2-1) split provides important support for the idea that executive orders are not, necessarily, purely presidential. But it is an aggregate view. It might be thought that, with the growth of the institutional clutter around the presidency, it has shifted over time. Figure 1 therefore presents the data above disaggregated by administration.
[FIGURE 1 OMITTED]
The figures show a drop in the number of externally generated executive orders occurring around the time when the literature might predict it--with the Nixon administration. Yet the drop is far smaller than one might expect given the shift in the political environment over the 40 years the data span. It is true that the least centralized period is at the start of the sequence, in the latter part of the Truman administration, and that the percentage of departmental orders decreases through the 1970s, and (intriguingly) especially with the Carter administration, only to move up slightly again in the Reagan years. But at no point does it fall below 50% of the total. (36)
The next iteration of analysis will involve multivariate analysis, to try to tease out this variation. A number of contexts need exploration. Based on the extant literature, they should include at a minimum:
1. The legislative backdrop: Does the president face divided government? What is the size of his partisan delegation in Congress? How distant is he from them, ideologically?
2. The statutory backdrop: does the order follow recently passed law?
a. Does it, by contrast, seek to preempt pending legislation?
3. The bureaucratic backdrop: how many departments had an interest in the order?
4. The policy backdrop:
a. Is the order "significant"? (however measured)
b. Is the order foreign, domestic, "intermestic"? Does it involve distributive or redistributive policy?
5. The political backdrop:
a. How close is a presidential election?
b. Has there just been a change of party in the presidency?
c. How popular is the president at the time of issuance?
This work is in progress. But even without dramatically expanding the coding, can we can at least ask whether significant orders are, well, significantly different?
The answer is, "sort of." The sample size to date is small, even in the aggregate, and the methodology rather crude. As noted above, current definitions of "significance" are derived from outside mentions of a given order, rather than from internal importance. From the OMB/White House files, however, it is possible to directly estimate significance, and I have done so here. (37) The evidence, however limited, does seem to hint at more centralized attention to important orders. The totals may be seen in Table 2.
While the total still shows that more than half of significant orders were proposed by the departments, the individual administration totals suggest this is an artifact of the Truman administration's figures, of Eisenhower's, and, intriguingly, of Reagan's. Of the 11 significant Truman orders in the sample so far, only two were formulated in the EOP, a ratio below the sample as a whole. Only eight significant Eisenhower orders have been coded so far, but 62.5% were formulated in the departments. And of the 15 significant Reagan orders, six, or 40%, were formulated in the EOP, very close to the ratio in the overall sample.
If Truman is removed from the sample, the EOP-centered orders become a (bare) majority. This is because the Kennedy through Carter administrations show more reliance on the EOP. Each had more significant orders drafted in the EOP than in the departments--in aggregate, a split of about 56-44%.
Three quick impressions, not yet strong enough to be conclusions, follow. One is that these findings suggest some enhanced tendency toward EOP influence over important unilateral actions under some presidents but not under others. If so, this is similar to other findings of "contingent centralization" (Rudalevige 2002). Second is simply that, even if so, 56% is still not especially high--that figure suggests a substantial proportion were at the departments' request and in their purview, especially since in most cases these orders received clearance across the affected executive branch agencies even if they came from the White House originally. Third, the distinction over time--if it is not simply an artifact of the sample or the definition of significance--is intriguing. It might suggest differences in the institutional development of the office or executive branch generally, perhaps displaying a lower level of effective politicization in the 1960s and 1970s as compared to the 1940s and 1980s, and thus more use of centralization as a substitute strategy. But this possibility is raised only to spur discussion at present.
Implications for Unilateral Power
It is worth recalling the metaphor of a "policy hill" that has been used to characterize the Eisenhower national security decision-making process (Bose 1998, 12). Detailed policy research was formulated on the upward slope (via the NSC's Policy Planning Board.) Then, at the NSC meeting, decisions were taken about which options to select. Finally, the chosen policy rolled down the hill towards its destination target. To make sure it got there, without winding up stuck against a rock or drowned in a bureaucratic brook, the Operations Coordinating Board kept an eye on its progress.
If we transfer that imagery to the world of presidential unilateralism, we find we know a fair bit about the very top of the hill and what the landscape looks like from there. But we know far less about either slope---either in terms of formulation or implementation.
On the formulation side, even the snapshot of preliminary research presented here makes clear that we need to be cautious in simply equating the power associated with executive orders with unilateral power. It suggests that the caution is indeed warranted in thinking about the source of executive orders and whether they are a function purely of presidential preferences. The bargaining arena may shift, as