Within the last two decades a number of researchers (Chiou and Rothenberg 2013; Cooper 2002; Howell 2003, 2005; Krause and Cohen 1996; Mayer 2001; Rottinghaus and Maier 2007; Rudalevige 2010; Savage 2007; Warber 2006) have sought to document presidents' capacity to make policy administratively, via executive orders (EOs), national security directives, proclamations, signing statements, and other nonlegislative means. Much of this research suggests that presidents, by pursuing policy administratively, can avoid the need to bargain with Congress via the legislative process. That is, they can produce policy "unilaterally," with a "stroke of the pen" (Mayer 2001). (1)
The implicit (and often explicit) assumption underlying many of these studies is that presidents face a choice: they can either pursue a policy objective by submitting a legislative proposal to Congress or they can "proceed on their own" through administrative means (Howell 2003, 96; Mayer 2009; Moe and Howell 1999, 132). But how do presidents choose? What factors determine whether they proceed legislatively or administratively? Previous research (Deering and Maltzman 1999; Howell 2003, 2005) suggests that the decision turns in large part on the degree to which Congress is able to legislate, either on the president's behalf, or against his policy objectives. Thus, Will Howell shows that when Congress has a strong majority party, as measured by the number of seats it possesses and its ideological cohesion relative to the minority party, presidents will be less likely to pursue policy objectives through administrative means such as executive orders. This is because, in the case of a sympathetic Congress, presidents can achieve their objectives via legislation. And if a strong majority party does not share the president's policy preferences, it can overturn his administrative directives. However, when Congress is ideologically fragmented or lacks a strong majority party, it has less capacity to legislate on his behalf, or to block him.
Significantly, most of this prior research primarily addresses one side of the either/ or decision; it documents the logic for why presidents issue administrative directives but does not demonstrate whether these same reasons also explain when presidents submit legislative requests to Congress. Howell does show that a Congress dominated by a large and ideologically cohesive parties is more likely to produce nontrivial legislation (Howell 2003, 96-99). However, looking at the nontrivial legislation Congress produces does not necessarily tell us whether that legislative output fulfills the president's policy goals. To more accurately assess the either legislation or administrative action hypothesis, then, scholars need to compare presidents' use of administrative directives, such as executive orders, with their legislative requests to Congress. When do presidents request legislative action from Congress and when do they opt to pursue policy objectives via administrative action?
This article presents an initial effort to answer that question. We show that administrative directives and legislation are not, as some previous research suggests, equivalent methods by which presidents can make policy, and presidents do not view them as such. Instead, presidents choose which policy instrument to use based in part on its particular structural attributes.
As a policy-making mechanism, administrative directives, such as executive orders, provide important advantages to presidents. Most notably, they allow presidents to capitalize on their first-mover status to make policy quickly and efficiently. But EOs cannot appropriate money or remake existing law, and they are susceptible to tampering by future presidents, which makes them less durable. Moreover, because these directives are based on existing statute or other formal grants of power, they tend to change policy in a more limited fashion than does legislation. To enact broad and lasting policy change on important issues, then, presidents prefer to work with Congress via the legislative process.
In some instances, particularly during periods of crisis, presidents may initially pursue significant policy change via direct administrative action. But rather than a means to bypass Congress, as previous research suggests, these substantively important unilateral acts are often meant to put down an initial policy marker pending eventual congressional action; presidents act administratively in the expectation or even hope of further legislation action, or to prod Congress to act. In short, administrative directives can, under some circumstances, serve as a first step in a multistep policy process that culminates in legislative action by Congress.
Presidents, we suggest, understand the distinct advantages and disadvantages offered by each policy instrument and act accordingly. Most notably, except when speed is paramount, presidents prefer to pursue their most significant policy goals through legislation rather than administrative means.
Our argument unfolds as follows. Building on previous research, in the next section, we examine the circumstances under which presidents during the post--World War II period are likely to issue executive orders and when they will propose legislation. Consistent with prior research, we find that when congressional majorities are small and ideological cohesion in Congress is weak, presidents are increasingly likely to issue executive orders. However, there is no statistically significant relationship between the strength of congressional parties and the number of legislative proposals the president submits to the body. Moreover, the number of proposals sent to Congress in each period dwarfs the number of policy-making executive orders signed, indicating that presidents seeking significant policy change prefer to work through the legislative process.
To explain these somewhat unexpected findings, we examine the establishment of the Department of Homeland Security (DHS) during George W. Bush's presidency. In the crisis conditions existing in the immediate aftermath of the 9/11 terrorist attacks, Bush moved quickly via executive order to establish a Homeland Security Council (HSC) administered by a White House--based assistant for homeland security. While his EO was being implemented, however, several bills were introduced in Congress designed to put homeland security organization on a more durable statutory footing. Ultimately Bush was able to bargain successfully with Congress to pass legislation establishing a cabinet-level DHS on terms largely agreeable to him.
Although a single case study, Bush's experience creating the DHS suggests to us that although EOs and legislation serve somewhat different roles in the policy process because of their different structural attributes, under some circumstances they can be used in complementary fashion to achieve the president's important policy objectives. Executive orders, we hypothesize, offer advantages in speed of action, symbolic impact, and flexibility of terms, while legislation--although slower to reach fruition--tends to be more durable because passage typically requires political coalition building. As a consequence, particularly in response to a crisis, and for symbolic reasons, presidents sometime proceed first through administrative means, but then seek to secure these initial policy gains through legislation.
How often does this occur? In the penultimate section, we present a new data set linking executive orders to subsequent presidential legislative proposals during the period 1947-2003, from the start of Harry Truman's administration through George W. Bush's first term. This data set differs from previous efforts that link executive orders to subsequent actions by Congress (e.g., Howell 2003, who traces congressional attempts to revoke or codify executive orders) or that identify executive orders issued in response to pending legislation, either to support or preempt its enactment (Belco and Rottinghaus 2014). We identify twenty orders for which presidents then submitted legislative proposals, either to codify the orders' authority or to expand it. For an important subset of policies, then, EOs are best understood as the president's first step in a bargaining process with Congress.
Our findings indicate the need for scholars to dig beneath the aggregate numbers and look more closely at the substance of legislative and executive orders and the context in which they are issued. Although in many instances administrative procedures are unilateral, it is also true that a significant portion are subsequently amended or even revoked by presidential action. Moreover, a few of the most important EOs are best viewed as part of the congressional bargaining process. This interpretation suggests that, rather than offering a "direct challenge to the strategic presidency" (Mayer 2009, 429), presidents' use of administrative directives, such as executive orders, is consistent with prior research (Jones 2005; Neustadt 1990) regarding how presidents must bargain to achieve policy goals. Even in the realm of unilateral administrative action, it appears that presidents cannot always escape the constraints imposed on them by operating in a system of separated institutions sharing powers.
The Policy Process: EOs or Legislative Requests?
When do presidents pursue policy change through administrative means, and when do they rely on the legislative process? And are these approaches equivalent means of achieving policy goals, as some scholars suggest? Will Howell (2003, 2005) provides one of the more thoroughly developed explanations for why presidents issue executive orders in the post--World War II era. His 2003 book, Power Without Persuasion and a subsequent 2005 journal article present a model that posits an inverse relationship between a president's likelihood to issue significant executive orders and...