Unilateral Presidential Powers: Significant Executive Orders,

Transcription

1 University of Wisconsin-Madison From the SelectedWorks of Kenneth R Mayer June, 2002 Unilateral Presidential Powers: Significant Executive Orders, Kenneth R Mayer Kevin Price Available at:

2 PRESIDENTIAL Mayer, Price / UNILATERAL STUDIES QUARTERLY PRESIDENTIAL / June POWERS 2002 Unilateral Presidential Powers: Significant Executive Orders, KENNETH R. MAYER KEVIN PRICE University of Wisconsin Madison The conventional sense of presidential power remains anchored in Neustadt s notion of persuasion in a fragmented constitutional system. Here, the authors add to an emerging literature that redirects attention to formal sources of presidential authority. They examine the frequency of executive orders from 1949 to 1999 and offer new evidence that presidents rely on executive orders to effect significant policy change and send strategic signals to other actors in the political system. They contend that executive orders enable presidents to recast the organization and activities of the federal government and, at times, the larger contours of American politics. After assessing the political and temporal logic behind this manifestation of institutional power, the authors conclude with several observations about the implications of the findings for the study of the American presidency. In the typical rendering of the American presidency, chief executives encounter formidable barriers to decisive leadership: a far-flung, entrenched bureaucracy (Wilson 1989); a largely independent and often recalcitrant Congress (Mayhew 1974; Peterson 1990; Jones 1994); an array of news media operating under distinctive incentive structures (Bennett 1988; Entman 1989); and a sporadically attentive, increasingly cynical public (Zaller 1992; Black and Black 1994; Cooper 1999). Richard Neustadt s (1990) description of presidential power as the power to persuade captures the conventional scholarly wisdom of a constrained executive office. Indeed, many of us deploy the premise of the impossible presidency as we teach undergraduates that presidents struggle to achieve and sustain success. Presidents clearly labor under the dual burdens of high expectations and contested power, a challenge apparent to many occupants of the White House. In a typically forthright moment, President Truman suggested a man would be crazy to want the office if he knew what it required (McCullough 1992). Decades earlier, President Grover Cleveland offered a young Franklin Roosevelt one wish as they shook hands: that the boy not grow up to be president of the United States. Only a handful of postwar presidents Eisenhower, Reagan, and Kenneth R. Mayer is professor of political science at the University of Wisconsin Madison and author of With the Stroke of a Pen: Executive Orders and Presidential Power (Princeton University Press, 2001). Kevin Price is a graduate student at the University of Wisconsin Madison. Presidential Studies Quarterly 32, no. 2 ( June) 2002 Center for the Study of the Presidency 367

3 368 PRESIDENTIAL STUDIES QUARTERLY / June 2002 Clinton have left office with public approval ratings substantially above the low point of their respective administrations. Notwithstanding the conventional view of the impossible presidency, political scientists have rediscovered arguments that chief executives enjoy meaningful unilateral authority even in a system of separated institutions sharing power. Amassing a rich body of descriptive data about unilateral presidential action, this literature has elaborated a theoretical framework to fit the pieces together and advance our understanding of presidential power (Moe and Howell 1999; Howell 2000). Of particular interest is the executive order, a tool presidents use to bring their constitutional or congressionally delegated powers into play unilaterally. Formally, an executive order is a presidential directive that draws on the president s unique legal authority to require or authorize some action within the executive branch (Mayer 1999, 445). Political scientists now recognize executive orders as important unilateral policy tools, however constrained by legal and political considerations they may be (Krause and Cohen 1997; Moe and Howell 1999; Mayer 1999, 2001; Deering and Maltzman 1999). In this article, we examine the frequency of executive orders from 1949 to 1999, offering new evidence that presidents rely on executive orders to implement significant domestic and foreign policies. We contend that executive orders enable presidents to recast the organization and activities of the federal government and, at times, the larger contours of American politics. We assess the political and temporal logic behind this manifestation of institutional power and conclude with several observations about the implications of our findings for the study and practice of the American presidency. The Unilateral Presidency in Action Careful observers of American politics cannot miss the persistent importance of unilateral presidential authority. To take a recent example, Bill Clinton s 1996 establishment of the 1.7 million acre Grand Staircase-Escalante National Monument reflected not only the president s programmatic ambitions but the imperatives of election-year politics. Clinton recognized that the Republican Congress would certainly refuse to join him in a legislative effort to protect large parts of the West from industrial development, but he recognized the salience of western lands policy to his environmentalist supporters. In choosing to protect a vast tract of land in Utah, Clinton sought political benefits (the bolstered support of environment-minded voters) without incurring political costs (because he knew he would not win Utah s electoral votes in any case). A University of Colorado professor called Clinton s decision a brave and true act, but a representative of the Utah Association of Local Governments called it the most arrogant gesture I have seen in my lifetime and suggested that the only comparable act I can think of is when a country is ruled by a king and he sweeps his hand across a map and says, It will be thus! (Maraniss 1996). By deploying the authority vested in his office by the Antiquities Act, a 1906 law enacted to allow presidents to protect historically significant sites, Clinton announced his commitment to an important element of his constituency, but he sought these broad electoral advantages by concentrating costs in

4 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 369 the mountain states of the West where he fully expected to lose in the 1996 election. 1 The political logic was clear, and his institutional authority allowed President Clinton to pursue political and programmatic gains without incurring significant political costs (at least in the short run; one might argue that mounting western opposition to the administration s lands policy hindered Al Gore s effort to make inroads in the region in the 2000 campaign). As his presidency reached its final stages, Bill Clinton used executive orders to bypass a largely recalcitrant Republican Congress (Shogren 1998). The political incentives driving the Clinton strategy were not lost on contemporary observers. Editorial writers at the Washington Times, for instance, noted that Clinton s unilateral actions had two central objectives: To appear relevant and to juice core constituents (Government by executive order 1998). In a scathing indictment of Clinton s use of executive orders, a contributor to the Union Leader of Manchester, New Hampshire, opined that Chairman Clinton rules by decree in People s Republic of America (Lessner 1998). One can make a principled case against presidents exercise of unilateral authority to achieve wide-ranging policy objectives, but much of the conservative opposition to the use of such authority when wielded by President Clinton receded when George W. Bush considered using his authority to overturn several key Clinton decisions. Like so many other features of the American political system, one s opinion of ostensibly institutional questions turns in a fundamental sense on whose interests are better served by structural arrangements at a given moment (see Binder and Smith [1997] for a similar argument in the specific context of the Senate filibuster). During the second half of his second term, Bill Clinton resorted to executive orders and other forms of unilateral authority to enact important policy changes, to draw what he saw as politically helpful contrasts between his party and the GOP, and to pursue an appealing programmatic legacy in the face of an opposition Congress. For example, he signed an order requiring firms that provide health care to federal employees to abide by the requirements of the 1996 Kennedy-Kassebaum legislation, invoked the Antiquities Act to protect land in the West, banned road building in large sections of the national forests, secured Medicare coverage for patients involved in clinical trials, and adopted water quality standards that set pollution limits based on local conditions. Amid this flurry of executive activity, critics railed against Clinton for his liberal, often creative, use of executive orders to steer the ship of state in ways and directions that might not pass muster in the normal system of checks and balances (Paige 2000). But this is precisely the point: like other presidents before and after, President Clinton used this repository of power to achieve political and policy objectives that otherwise faced grim prospects. When George W. Bush came to office in 2001, he immediately set to work overturning key Clinton orders the Bush ban on aid to international family planning organizations drew widespread attention and testing the political waters for other opportunities to reverse Clinton administration enactments such as the ban on road building in national forests, limits on arsenic in drinking water, and new meat safety rules. The point is not that Bush can overturn every Clinton action he finds objectionable; like Clinton himself, Bush will use 1. Largely obscure before Bill Clinton s invocation of it in the Grand Staircase decision, the Antiquities Act achieved a degree of cultural salience when it played a key role in an episode of the NBC television series about the White House, The West Wing.

5 370 PRESIDENTIAL STUDIES QUARTERLY / June 2002 political and institutional calculations to probe for soft spots in his predecessor s program. Indeed, in July 2001, Bush and the House Republican leadership backed away from threats to systematically reverse Clinton s last-minute executive orders and regulatory policies (Eilperin 2001). September 11 obliterated the previous calculus of presidential power. Bush, as wartime presidents have done without exception, moved quickly to assert control over what was by any definition a national crisis. Within weeks, he froze the assets of terrorist groups (Executive Order 13224, September 25, 2001), redefined the scope of airline security, established the Office of Homeland Security within the Executive Office of the President (Executive Order 13228, October 10, 2001), authorized military tribunals for trials of suspected terrorists (Military Order of November 13, 2001), revised Justice Department regulations concerning the treatment of attorney-client conversations by certain federal detainees (Department of Justice/Bureau of Prisons regulation of October 31, 2001), and began prosecuting the war against the Taliban in Afghanistan. While some of these actions were clearly authorized by Congress, some of the more significant steps were taken unilaterally, either in the face of congressional acquiescence or even opposition. 2 In its essentials, this sequence illustrates two processes. The first is one of dueling presidencies, each one exploiting the prerogatives of executive authority when political incentives recommend it and other conventional means of achieving policy change appear foreclosed. The second is one of executive initiative in response to a direct threat to U.S. security. Both reflect a consistent pattern. As these familiar events suggest, presidents can deploy their constitutional and statutory authority to reshape national governance and the contours of the political system itself. In short, presidential power is more than the power to persuade; it is also the power to command, though such authority is always contingent on the political and institutional constraints of a given moment. Executive Orders and Formal Presidential Power Neustadt s conception of presidential power remains dominant four decades after he first developed the argument. In Presidential Power, Neustadt (1990) contends that presidents cannot normally secure their objectives by executive fiat or command. Presidential commands do not signify presidential power, Neustadt argues, but a painful last resort, a forced response to the exhaustion of other remedies, suggestive less of mastery than of failure (p. 24). Instead, presidents must rely on their strategic bargaining skills to induce compliance from other key actors, many of whom have independent bases of political support and institutional authority. In our view, this dichotomy between formal power and persuasion is artificial; the two categories are neither mutually exclusive nor exhaustive of all possible avenues of presidential influence. Here, we merge conceptions of presidential power in particular with more theoretical work about the nature of political power in general. 2. One order had little to do with national security or the war against terrorism. In October, George W. Bush issued an executive order (Executive Order 13233, November 5, 2001) pertaining to the Presidential Records Act. In it he, in effect, limited access to the papers of ex-presidents beyond the twelve-year period specified in that act and gave both ex-presidents and current presidents a concurrent veto over release of disputed papers.

6 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 371 Most accounts of the presidency and its occupants stress the decidedly limited extent of direct presidential authority, emphasizing instead the requisite exercise of informal skills and persuasive political leadership (see Fisher [1995]; Edwards, Barrett, and Peake [1997]; and Cooper [1986] for noteworthy exceptions. An emerging line of research following Krehbiel [1998] also attends systematically to the programmatic implications of the presidential veto; see Cameron [2000] for example). In Presidential Power, Richard Neustadt (1990) observed that even presidents most explicit formal prerogatives the veto, the military order, the federal directive require compliance from others: Literally, Neustadt writes, no orders carry themselves out; self-executed actually means executed by others (p. 17). As a consequence, virtually every case of presidential action involves persuading other political actors. In practice, then, even the exercise of unilateral authority requires a president to shape the perceptions and political calculations of other actors to secure their cooperation. Our point here is not to question the enduring relevance of Neustadt s construct but to advance an argument connecting the study of the presidency to the tradition of public law in which institutional prerogatives played a central analytical role. Before Neustadt introduced this notion of persuasive power, Edward Corwin (1948) noted the ambiguity of the constitutional provision that the president take care that the laws be faithfully executed. Executed, Corwin wondered, by others? (p. 4). As in most forms of institutional authority, the exercise of presidential power is indirect in Neustadt s literal sense that it requires compliance from any number of distinctly situated actors. Nevertheless, such compliance is highly likely when presidents marshal the formal powers of their office. Within constraints established by political, organizational, and legal circumstances, presidents enjoy substantial latitude in exercising official privileges such as the executive order (Mayer 1999; Moe 1993; Moe and Howell 1999). How might presidents take advantage of this deference, and to what ends? A president can commit strategic resources to induce compliance in a specific interaction with another political actor (a transaction, to use the new institutional economics, or NIE, parlance). Such a principal-agent relationship raises many of the same problems that attend to contractual relationships among economic actors: specifying the terms of agreement, monitoring behavior, enforcement, moral hazard, shirking. An alternative to this sort of case-by-case persuasion is to commit resources to create new institutional patterns or procedural mechanisms and to create new and possibly enduring arrangements that shape the incentives of other actors. In the NIE framework, firms are a way to institutionalize contractual interactions between a principal and an agent. In political settings, argues Moe (1993), institutions can serve much the same function. By using their formal powers, presidents structure the institutions that surround them to standardize their interactions with other actors. To convert the bargains that would otherwise require skill and scarce political capital into manageable leadership opportunities, presidents seek routines that encourage compliance from other actors. By creating institutions and processes that make these once-expensive bargains part of the political landscape, presidents alter default outcomes, leaving it to other actors to expend resources to undo what the president has done. If presidents may use the rhetorical, personal, and persuasive resources of the office, they may also take advantage of the formal prerogatives vested in the chief executive by the

7 372 PRESIDENTIAL STUDIES QUARTERLY / June 2002 Constitution, statutory grants of authority, and the evolution of relevant legal doctrine. (Cash 1963; Cross 1988; Fleishman and Aufses 1976; Jack 1986; Neighbors 1964; Rosenberg 1981; Sunstein 1981; Tauber 1982). To the extent that executive orders alter institutional structures or processes, they provide potent resources for presidential leadership based on formal authority. In short, presidents enjoy significant but not unfettered latitude in governing the government through the issuance of executive orders. The administrative authority vested in the modern presidency is significant in its own right, but we claim that the exercise of intragovernment powers can yield consequences for the larger political system as well. In a highly permeable system in which executive policy decisions can spill over into other institutions and the broad political culture, executive orders can effect changes that presidents may or may not have intended or even considered in the first place (Weir 1989). In other words, politically significant executive orders are not merely executive phenomena. As illustrated by the long history of executive orders concerning integration and civil rights Truman ordering the integration of the armed forces, Kennedy and Johnson requiring affirmative action in federal contracting, Reagan attempting to limit the role of ethnic preferences in federal affirmative action programs these instruments of presidential authority can animate contending forces, facilitate innovations in the legislative process, codify ideological commitments, and drive social change. Obviously, formal and informal presidential powers are not mutually exclusive. The exercise of formal powers at one point can affect subsequent opportunities to exploit informal resources. As Neustadt (1990, 4) writes, the crucial question for any president is whether he will have influence when he wants it. Recognizing the dynamic quality of national leadership, we consider executive orders to be substantively and theoretically important manifestations of presidential discretion. Shaped by political calculations, advisors offerings, and the possibility of countervailing actions by other institutional actors, executive orders reflect a uniquely presidential mode of action. Each order bears the chief executive s signature in a symbolically telling and potentially clear statement of presidential priorities. Executive orders provide an important link to a crucial result of presidential leadership: public policy. According to Paul Light (1993, 161), lasting programmatic innovations represent the most important legacy of any administration. Other types of presidential legacies symbolic (Tulis 1987; Kernell 1986), partisan (Skowronek 1993; Milkis 1993), and organizational (Moe 1985; Weko 1995) also provide telling clues to the effects of presidential leadership on the larger political system. Executive orders, of course, can impinge on each of these dimensions. It is important to note that the next occupant of the White House may summarily retract a president s executive orders. Any decision to overturn an existing order will be shaped by the usual suspects political, organizational, and legal concerns but presidents possess the authority to do so if they wish. Statutes may be overturned, but such an effort requires much more of its proponents as they negotiate the complications of the congressional lawmaking process (Rauch 1994). In the simplest sense, executive orders are presidential directives that require or authorize some action within the executive branch of the federal government. Presidents have used them to reorganize executive branch agencies, alter administrative and regulatory processes, shape legislative interpretation and implementation, and make policy within the bounds of their constitutional or statutory authority. Those bounds are somewhat fluid, as

8 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 373 the Supreme Court has generally but not exclusively taken an expansive view of presidential discretion. Recent research into these unilateral executive powers has found consistent evidence that presidents utilize them for strategic purposes. The overall frequency of executive order issuance varies in predictable ways, as presidents tend to issue more orders and thus rely more heavily on their own powers when unpopular, when faced with crises that demand swift and decisive action, when running for reelection, and immediately after the White House switches party control (Mayer 1999). But it is possible to extend this research by sifting through the population of executive orders to isolate especially significant directives. From a theoretical standpoint, one should consider not simply the number of executive orders a president issues even though such a measure is a useful indicator of overall energy but how many of those orders are recognized as genuinely important. Over the past four decades, presidents have issued roughly forty to sixty executive orders per year, some important, others quite mundane. Orders clearly recognized as significant such as Johnson s Executive Order 11246, which laid the framework for federal affirmative action programs stand in the record alongside undeniably trivial orders (such as Ford s Executive Order 11943, which merely corrected a typographical error in an earlier order setting salary levels for federal employees). The noise level of these unimportant orders might obscure substantive patterns among the important ones. In what follows, we describe our data and methods, provide the results of our statistical estimations, and conclude with several theoretical points concerning the American presidency. Isolating Significant Executive Orders Our first cut at the data analysis begins in March 1936, the first month in which executive orders were systematically recorded in the Federal Register. Determining which of the roughly seventy-five hundred executive orders issued since then qualify as significant is an inexact and subjective process. 3 Nevertheless, it is possible to outline several criteria that permit distinctions between relatively significant and relatively trivial orders. In Divided We Govern, David Mayhew (1991) constructed an inventory of major legislative enactments from 1946 to Using a two-stage process, Mayhew first drew on journalists year-end accounts of legislative action to note whether contemporary observers considered a law significant. In a second pass, he looked at retrospective evaluations by policy experts in various fields: did the legislation attract scholarly attention as something that changed a policy domain in important ways? Based on these criteria, Mayhew identified 267 acts as particularly important. 4 In his study of executive-legislative relations, Mark Peterson (1990) coded presidential policy initiatives as innovative or as part of a broadly important program. Here, Peterson s 3. The first executive order to appear in the Federal Register was number 5,674. The final order in the series, issued December 21, 1999, was number 13,144. The total population of orders is thus 7,471 orders. 4. Mayhew (1991, 49). During this span, Congress enacted a total of 24,671 bills, of which 15,875 were public and the remainder private (data from Ornstein, Mann, and Malbin 1996, 165). By Mayhew s criteria, just over 1 percent of these bills constituted major legislation (267 out of 24,671).

9 374 PRESIDENTIAL STUDIES QUARTERLY / June 2002 declaration of importance depended on whether specialized sources of political coverage (Congressional Quarterly, CQ Almanac, National Journal) mentioned the policy. A measure s level of innovation turned on whether the proposal departed significantly from existing policy. In his sample of presidential proposals, Peterson found that more than one in five constituted issues of the greatest consequence. Working with a random sample of 1,028 orders issued between March 1936 and December 1999, we characterized orders within that sample as significant if they met any one of six criteria. 5 First, we considered whether the order received attention at the time from the press or from other political actors. At times, journalistic coverage has focused more on the subject of an order than on the order itself. When Eisenhower placed the Arkansas National Guard under federal control to force the integration of Little Rock schools, he took action within a much broader legal and political process. Press accounts at the time tended to focus not on the fact that Eisenhower had issued an executive order but rather on the substance of what he did. 6 In looking at press attention, we considered both the national press (the New York Times, Washington Post, or Wall Street Journal) and more specialized publications covering politics or specific professional fields. Second, we ascertained whether Congress considered an order important enough to justify committee hearings and whether specific legislation to override the order made it to the floor of either chamber. Congressional hearings, in particular, have long been recognized as an important measure of the issues legislators deem significant (Edwards and Wood 1999). Again, some orders receive more congressional attention than others, and many pass unnoticed, but it is not unusual for legislators to pay attention to those orders that might threaten congressional prerogatives. Third, we asked if students of law or the presidency refer to the order (or a class of orders) in their scholarly work. Here, we effectively replicate Mayhew s (1991) second sweep of expert opinion. In this regard, the legal literature is extensive, though the social science and history literatures are less developed. Nevertheless, the extent to which an order receives mention in one or more of these forums is an indicator of importance. For example, we counted all of the emergency seizures of private property during wartime or not as significant. While some of these seizures made the news at the time and others did not, Rossiter (1956, 59) depicted them as a spectacular display [of] the President s power of martial rule. 7 Similarly, Corwin (1948, 301) dealt at length with the wartime property seizures as an example of presidents intruding into the clearly legislative territory of industrial policy. In 5. The original sample was one thousand orders, drawn from the March 1936 to December 1995 period. This yielded a sample of 17.6 percent of the population. We later updated the sample, drawing the same fraction of all orders issued between January 1996 and December The resulting pooled sample can be considered a single random sample for the purposes of statistical inference. 6. Conversely, the press often refers to unilateral presidential actions in general as executive orders, even when the actions do not take that form. When Clinton reversed the existing ban on fetal tissue research and overturned a ban on abortions at overseas military hospitals, he did so through presidential directives to agency heads, not through formal executive orders. Nevertheless, these actions were widely reported to be executive orders. 7. Rossiter (1956, 60) discounts the fact that in 1943, Congress granted the president formal legislative authority to make such seizures, arguing that since Roosevelt had relied on the commander-in-chief power to take over plants well before this act, the delegation was at best declaratory, even supererogatory.

10 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 375 addition, we examined the law review literature to determine whether an order spurred any attention by legal scholars. Fourth, we identified cases in which presidents themselves demonstrated in public statements or emphasis in the public record that they considered an order important. Presidents have placed varying emphasis on executive orders. Carter and Reagan tended to publicize most of their orders, placing them in their collections of public papers. Others did not routinely publicize their orders, instead releasing order texts and public statements only for those they considered especially noteworthy. We thus include orders about which a president made a nonroutine statement (that did not include typical press releases). Fifth, we determine if the order triggered federal litigation. A few orders have generated legal challenges of major importance Youngstown v. Sawyer, Korematsu v. U.S., and E.P.A. v. Mink, for example. A lawsuit is an indication that a plaintiff considered the order a sufficiently important infringement on some right or interest to seek judicial remedy. Similarly, a Supreme Court or appeals court decision is clear evidence that the judiciary considers the case important and controversial as a matter of law. Executive Order 11246, for example, has been cited in more than three hundred federal district court cases since In deploying this criterion, we identify executive-order-related litigation as catalogued in the Lexis and Westlaw databases. Finally, we ask whether an order created a new institution with substantive policy responsibility, expanded or contracted significant private rights, or constituted a significant departure from existing government policy. Again, we distinguish between orders creating ceremonial or symbolic organizations (such as the American Battle Monuments Commission) from those establishing or reorganizing agencies with substantive duties (such as the Office of War Information or the Office of Strategic Services). We counted an order as significant if it met at least one of the conditions outlined here: press attention, congressional notice, scholarly treatment, presidential emphasis, litigation, or creation of institutions with substantive policy responsibility. Most significant orders met more than one of these criteria. To be sure, this process involved subjective judgments, but we attempted throughout to adopt a conservative view of what constituted a significant order. In cases of ambiguity about an order s significance, we considered it insignificant. We deemed insignificant those orders that exempted individuals from retirement or other civil service requirements and all but four public land orders. In making these designations, we erred on the side of underestimating the number of significant orders. According to these criteria, 149 of the 1,028 executive orders in the original sample about one in seven qualify as significant (the appendix contains a list of these orders, giving the number, date, and title of each). From this sample statistic, we can estimate within a 3 percent margin of error that 1,083 significant executive orders were issued during the sixty-five-year period from 1936 to 2000, an average of about 17 per year. We can use this database of significant executive orders to understand the political and structural forces that prod presidents to issue them. Extending Mayer s (1999, 2001) analysis of the full sample, our working hypothesis is that the frequency of important orders will vary with the strategic context in which presidents operate. In the ensuing section, we describe our models and the results of our statistical analysis.

11 376 PRESIDENTIAL STUDIES QUARTERLY / June 2002 Models and Results Because the number of executive orders issued in any year can only assume nonnegative integer values, an event-count model provides an appropriate estimation procedure for our data. Specifying the appropriate count model requires care because of the reasonably strict assumptions required to invoke the Poisson model (Greene 1997; King 1989). In the Poisson, one assumes no contagion among observations. In other words, the Poisson requires that each significant order be statistically independent of all others. In the more general negative binomial model, a dispersion parameter α is estimated to account for any contagion in the data. The Poisson is thus a special case of the negative binomial appropriate only under certain substantive and technical circumstances. As Tables 1 and 2 suggest, the Poisson model is entirely adequate for our data set a likelihood ratio test reveals that the negative binomial provides no improvement over the Poisson. Drawing on previous research concerning the exercise of unilateral presidential authority, our primary theoretical expectations concern the political and institutional environment in which presidents operate. We begin with the observation that the ambiguities of the president s constitutional powers, combined with the assumption that presidents are motivated to assert control over governing structures, provide opportunities for presidents to probe the boundaries of their power and extend those boundaries when they can. A president s proclivity to do so depends on the specifics of his political context and his relationship with other participants in the governing process. 8 First, we expect newly elected presidents to issue relatively more significant executive orders. This pattern should be more pronounced in cases where party control of the presidency has just switched. With their partisan objectives frustrated by the previous regime, presidents reclaiming the White House for their party should issue important orders in an effort to demonstrate their authenticity to campaign supporters and important segments of their electoral coalitions. They should also try to reverse the policies of their predecessors. For example, President Clinton promised to issue an order permitting homosexuals to serve in the military, as part of a larger effort to consolidate and reward the political support of the gay and lesbian community. Owing to congressional and military opposition, of course, he stopped well short of the policy he first proposed, opting instead for the much weaker don t ask, don t tell policy that seemed to satisfy no one. But the landscape changed with even this step, as during the 2000 campaign Vice President Al Gore vowed to extend the policy and George W. Bush promised to reverse it. Notwithstanding this case, however, new presidents should exercise their unilateral powers with unusual vigor soon after assuming office. Executive orders are notable because of their potential transience: successors in the White House may simply overturn previously issued orders with new orders of their own. In light of this core institutional characteristic, and in light of the political logic underlying early issuance, new presidents should be more likely to sign important executive orders in the first months of their terms. 8. For an explication of these two threads, see Mayer (2001) and Moe and Howell (1999).

12 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 377 Additional theoretical expectations include the following. First, new presidents who have just wrested control of the White House from the other party should issue significant orders with greater frequency than new presidents who inherit the White House from their own party (or from themselves in the case of second-term presidents). Second, presidents running for reelection should seek to avoid unnecessary controversy by issuing fewer significant executive orders. Conversely, presidents not running for reelection should issue more orders in an effort to create some programmatic or institutional legacy in the waning months of their administrations. Third, if Neustadt (1990) was right that presidents need robust public prestige to govern effectively in the separated system, approval ratings should be inversely related to the number of significant executive orders issued in a given year. Fourth, if presidents encounter recalcitrant opposition parties in Congress, divided government should encourage the issuance of significant executive orders. Finally, we expect Democrats to pursue their typically more assertive programmatic ambitions by issuing more orders than their Republican counterparts. For the event-count model, the coefficient vector is µ i = exp(x i, β), where x i β = b 0 + b 1 First Change + b 2 First No Change + b 3 Reelection + b 4 No Reelection + b 5 Public Approval + b 6 Divided Government + b 7 Party + ε i. Variable Descriptions and Values Orders t : number of significant executive orders issued in year t from 1949 to First Change: 1 during the first year of a new administration when party control has changed from the previous administration, 0 otherwise. First No Change: 1 during the first year of a new administration when party control has not changed from the previous administration, 0 otherwise. Reelection: 1 in presidential election year, when incumbent president is running for reelection; 0 otherwise. No Reelection: 1 in presidential election year, when incumbent president is running for reelection; 0 otherwise. Public Approval: Average presidential approval rating by year, as measured by the Gallup poll. Divided Government: 1 if the president s party does not control both houses of Congress, 0 otherwise. Party: 1 if the president is a Democrat, 0 otherwise. Although the full series of executive order begins in 1936, we were forced to begin our event count analysis in 1949, the year public approval ratings for presidents became generally available. While this reduces the number of significant executive orders in the sample, we believe the results remain valid. The truncation eliminates two clearly extraordinary periods the Roosevelt presidency and World War II in which an unusually large number of orders were issued; including them in the model would overestimate the impact of our inde-

13 378 PRESIDENTIAL STUDIES QUARTERLY / June 2002 pendent variables. Given that the frequency of order issuance has been relatively stable since 1949, our model is more likely to produce coefficient estimates that are comparable across time. 9 We do not include indicator variables for each presidency for two reasons. Theoretically, we are much more interested in the behavioral regularities than in the idiosyncratic exercise of presidential discretion. Statistically, we found no telling results from presidency dummies (with the exception of the Carter years, which we believe stand out because of the number of orders relating to the Iranian Hostage crisis). We want to look beyond specific presidencies to the larger contours of the institution and the prerogative of the executive order. As the results from the first model presented in Table 1 indicate, significant executive orders respond in predictable ways to the political and institutional circumstances in which presidents operate. First, a new president who has regained control of the White House for his party issues more orders, but presidents in the first year of a term not following a party switch clearly issue fewer important orders. Though the coefficients fall short of statistical significance (note, however, that each is larger than its standard error), fourth-year presidents tend to issue fewer major executive orders. Though divided government and partisan affiliation do not seem to relate systematically to order issuance, public approval appears very important. As Neustadt (1990) suggested, presidents who can marshal public prestige need not resort to this version of command to achieve desired outcomes. To ascertain the statistical leverage generated by our distinction between first-year presidents with and without changes in party control, we conducted a likelihood ratio test of the two models presented in Table 1. By calculating the marginal gain in the likelihood function in the more fully specified model 1, one can determine whether that full specification yields meaningful information. In this case, it clearly does; the LR statistic is 6.96 (distributed as χ 2 with one degree of freedom, p <.01). We can thus conclude that the specification in model 1 provides both statistically and substantively meaningful information. To interpret the substantive effects of the coefficients presented in model 1, we calculated predicted probabilities for a group of plausible outcomes (from zero to four significant orders in a year) given certain covariate profiles. We present these results in Table 2. Conclusion A growing literature contends that presidents use executive orders to achieve programmatic and institutional objectives, and that these instruments of unilateral authority empower chief executives in a separated system not typically conducive to decisive presiden- 9. Because of the small absolute number of significant executive orders over the period, we replicated the event-count analysis using King and Zeng s (2001) rare events data method. Although the rare-events model uses a logistic regression rather than an Poisson event-count model, the overall logic is comparable: under the rare-events method, we are predicting the probability that a significant executive order occurs in a given month, rather than an estimate of the number of orders issued, given a particular configuration of independent variables. There were no substantive differences in the overall predictive power of the model or in the coefficient estimates between the two models, and we concluded that the event count model is the more appropriate method given the nature of our dependent variable.

14 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 379 TABLE 1 Poisson Regression Results Model 1 Model 2 Constant 2.00*** (.566) 2.02*** (.567) First Change.532** (.309) First No Change.639* (.407) First Year.001 (.259) Reelection.338 (.319).323 (.319) No Reelection.430 (.407).435 (.407) Public Approval.026*** (.009).023*** (.009) Divided Government.085 (.319).126 (.310) Party.205 (.308).085 (.308) n Log likelihood Likelihood ratio χ ** 10.92* Note: The dependent variable is the number of significant executive orders issued in a calendar year, with appropriate adjustments for the month of January in the last year of a presidency. Cell entries are maximum likelihood coefficients, with standard errors in parentheses. *p <.10. **p <.05. ***p <.01 (one-tailed tests for coefficients, two-tailed tests for likelihood ratio). TABLE 2 Predicted Probabilities Number of Significant Orders per Year Covariate Profile Zero One Two Three Four Public approval at 30 percent Public approval at 50 percent Public approval at 70 percent First year with party change First year without party change Note: Cell entries are predicted probabilities for each outcome under specific covariate profiles using the results from model 1 in Table 1. For the public approval calculations, we hold all other variables at zero. For the first-year calculations, we hold the other indicator variables at zero and public approval at 50 percent. tial leadership. Our results show that the use of significant executive orders relates in theoretically tractable ways to a president s political environment. As our statistical findings suggest, presidents who can leverage robust public approval resort to the executive order less frequently than do those who lack strong popular support. In addition, partisanship and the ideological differences reflected in partisan affiliations affect order issuance in sensible ways. New presidents recapturing the White House for their party tend to issue more orders relative to those operating amid partisan continuity. In asserting such a conclusion, we are hardly suggesting that scholars should cast aside Neustadt s (1990) construct of presidential persuasion; the enduring notion that presidents

15 380 PRESIDENTIAL STUDIES QUARTERLY / June 2002 operate in a contested, fragmented system remains valid and important. Nevertheless, presidential power turns not just on the executive s persuasive capacities but on the constitutional and statutory sources of unilateral authority on which any president might draw. Today, students of the presidency might profitably explore both facets of presidential power unilateral authority, here manifested in unusually important executive orders, and persuasive authority and agenda control, as articulated most clearly in Neustadt s work. Whatever the subtleties of significant executive orders, the emerging literature on formal executive authority suggests that students of the American presidency should revise the prevailing view of presidential power to include the brute facts of presidents important unilateral capacities. Appendix List of Significant Executive Orders Order Date Title/Description 7856 Mar. 31, 1938 Rules Governing the Granting and Issuing of Passports in the United States 8233 Sept. 5, 1939 Prescribing Regulations Governing the Enforcement of the Neutrality of the United States 8344 Feb. 11, 1940 Withdrawal of Public Land for Classification Alaska 8629 Jan. 7, 1941 Establishing the Office of Production Management 8683 Feb. 14, 1941 Establishing Naval Defensive Sea Areas around Naval Airspace Reservations 8693 Feb. 25, 1941 Prescribing Regulations Concerning the Exportation of Articles and Materials 8751 May 2, 1941 Establishing the Division of Defense Aid Reports in the Office for Emergency Management of the E.O.P June 9, 1941 Government Possession and Control North American Aviation, Inc June 14, 1941 Regulating Transactions in Foreign Exchange and Foreign-Owned Property 8802 June 25, 1941 Equal Employment in Government Contracting, Establishing the Fair Employment Practices Committee 8890 Sept. 3, 1941 Establishing the Office of Defense Health and Welfare 8944 Nov. 19, 1941 Government possession and control, Grand River Dam Authority 9017 Jan. 12, 1942 Establishing the National War Labor Board 9040 Jan. 24, 1942 Adding duties to the War Production Board 9083 Feb. 28, 1942 Transfer of authority for maritime functions 9102 Mar. 18, 1942 Establishing the War Relocation Authority 9253 Oct. 9, 1942 Administration of Contracts of the Immigration and Naturalization Service 9279 Dec. 5, 1942 Mobilization of National Manpower, Administration of Selective Service System 9337 Apr. 4, 1943 Transfer of authority to withdraw public lands to the Secretary of the Interior 9341 May 13, 1943 Government possession and control, American Railroad Co. Of Puerto Rico

16 Mayer, Price / UNILATERAL PRESIDENTIAL POWERS 381 Order Date Title/Description 9351 June 14, 1943 Government possession and control, Howarth Pivoted Bearings Co Oct. 15, 1943 Establishing Advisory Board on Just Compensation 9425 Feb. 19, 1944 Establishing Surplus War Property Administration 9427 Feb. 24, 1944 Establishing Retraining and Reemployment Administration 9437 Apr. 18, 1944 Revocation of Executive Order 9165 (Protection of Essential Facilities from Sabotage and Other Destructive Acts) 9459 Aug. 3, 1944 Government possession and control, Philadelphia Transportation Co Aug. 31, 1944 Government possession and control, Ford Colleries Co. and Rochester and Pittsburgh Coal Co Sept. 3, 1944 Government possession and control, certain coal companies 9484 Sept. 23, 1944 Government possession and control, Farrell Cheek Steel Co Dec. 6, 1944 Government possession and control, Cudahy Brothers Co Dec. 27, 1944 Government possession and control, Montgomery Ward & Co Jan. 12, 1945 Government possession and control, Cleveland Electric Illuminating Co Feb. 18, 1945 Government possession and control, American Enka Corp June 14, 1945 Government possession and control, Scranton Transit Co July 25, 1945 Government possession and control, Springfield Plywood Corp July 30, 1945 Government possession and control, United States Rubber Co Aug. 30, 1945 Revoking forty-eight-hour minimum work week 9608 Aug. 31, 1945 Termination of Office of War Information 9621 Sept. 21, 1945 Termination of Office of Strategic Services and transfer of its functions 9622 Sept. 21, 1945 Revocation of Executive Order 9103 (Publication and Use of Federal Statistical Information) 9639 Sept. 29, 1945 Government possession and control, certain petroleum companies 9661 Nov. 29, 1945 Government possession and control, Great Lakes Towing Co Dec. 21, 1945 Establishment of National Wage Stabilization Board, Termination of the National War Labor Board 9674 Jan. 4, 1946 Liquidation of War Agencies 9691 Feb. 4, 1946 Resumption of peacetime civil service regulations 9801 Nov. 9, 1946 Removal of wage and price controls 9808 Dec. 5, 1946 Establishment of the President s Committee on Civil Rights 9863 May 31, 1946 Designation of certain international organizations as having diplomatic privileges 9981 July 26, 1946 Establishing the President s Committee on Equality of Treatment and Opportunity in the Armed Services 9989 Aug. 20, 1948 Transferring authority over blocked assets to the Attorney General Dec. 12, 1949 Defining Noncombatant Service and Noncombatant Training Aug. 17, 1950 Incentive Pay for Hazardous Duty Sept. 9, 1950 Delegating functions under the Defense Production Act Oct. 18, 1950 Protection of vessels, harbors, ports, and waterfront facilities of the United States Dec. 16, 1950 Providing for the conduct of the mobilization effort Feb. 2, 1951 Authorizing exercise of powers in Title II of First War Powers Act, June 7, 1951 Suspension of eight-hour law for certain Department of Defense personnel (continued)