A number of recent studies have explored how and when presidents use signing statements to shape the interpretation and implementation of statutes (e.g., Bradley and Posner 2006; Cooper 2005; Kelley 2007a; Kelley and Marshall 2008; Pfiffner 2007). Some emphasize the historical bases for the practice of attaching commentary to bills in the form of a signing statement (Cooper 2005; Kelley 2007a), while others work across administrations to describe presidential practices (Bradley and Posner 2006). Only recently have scholars sought to statistically estimate the impact of factors that drive those statements on the likelihood that a president signs a statement to accompany legislation that he signs. For instance, Kelley and Marshall (2008) show that presidents are more likely to sign statements when there is gridlock and also in the case of major legislation; Kelley and Marshall (n.d.) describe how presidents link veto threats and the use of the signing statement. Those papers, though, consider both types of signing statements: rhetorical (when the president lauds the Congress and others for their work on the legislation) and constitutional (when the president offers his view about how the legislation should be interpreted and implemented).
Political scientists often focus on the signing statement as part of the president's toolkit for shaping policy implementation. Likewise, the legal conversation about signing statements has come to emphasize their role in the president's arsenal of tools for bending policy closer to his own preferences. Walter Dellinger, then assistant attorney general, argued in a 1993 memorandum to Bernard N. Nussbaum (counsel to the president), that one "generally uncontroversial" purpose of signing statements was "to guide and direct Executive officials in interpreting or administering a statute." (1) For Magill, signing statements can be an effective "ex ante" method of presidential control since "The President could offer a wide variety of views in such statements, including his view of the substantive meaning of ambiguous statutory provisions, his view of how the statute should be implemented and enforced, or as Presidents in fact have frequently done, his view that certain provisions are or may be unconstitutional" (2007, 31). In her view, one of the most effective ways to use signing statements has been to "identify the priorities that should animate enforcement of the statute and those that should not" (2007, 50).
Yet, since the 1980s legal scholars have also debated the legitimacy of the president's use of statements to express his views about the meaning of statutes (e.g., Bradley and Posner 2006; Cross 1988; Garber and Wimmer 1987). For instance, Garber and Wimmer (1987) argue that courts should not refer to signing statements when deciding what Congress intended in a piece of legislation--that doing so would violate the constitutional separation of powers. In contrast, Cross (1988) argues that courts will find (at times) that signing statements are useful as significant sources of material for statutory interpretation, just as are the views of agencies. For political science studies like Kelley and Marshall (2008), the president's motives include influencing judges and bureaucrats, along with voicing their concerns about the constitutionality of statutes.
This article starts with the idea that presidential signing statements are both directives and defenses--useful now or in the future for presidents, their staff and appointees, bureaucrats, and judges--that help determine how a statute is implemented and interpreted. My focus is on the detail and complexity of the directive the president constructs rather than on whether the president signs a statement for a given piece of legislation. Presidents decide the shape and content of the directive, just like they use rhetoric for pushing their current or future policy goals (e.g., Campbell and Jamieson 1990; Tulis 1987). As Bradley and Posner (2006) note, the recent Bush administration's signing statements usually laid out multiple constitutional concerns; those collections of concerns are really packages of the president's position on the interpretation and implementation of individual statutes. As Cooper (2007) notes, presidents look to make "declaratory judgments," almost like a judge, in the form of a signing statement. Given that the president offers a signing statement, when is it more detailed and complex?
Yet, even if signing statements have interpretive consequences, they remain a political tool. I offer hypotheses that reflect the president's incentives to bend legislative outcomes closer to his preferences. Specifically, Cameron and McCarty (2004) describe how different arrangements of preferences in a separated powers system offer different opportunities for a president to pull outcomes closer to his own ideal point. The first hypothesis, which flows from the theory of legislative bargaining under incomplete information, is that the president will construct more detailed and complex statements when his ideal point is farther from that of Congress. The second is that a president constructs more detailed and complex statements over the course of his administration. Third, based on McCarty and Poole (1995), I hypothesize that the president constructs more detailed and complex statements when the president has lower public approval; I also test whether the president constructs more detailed and complex statements when public opinion is farther from his own ideal point. Finally, based on prior research, I account for factors such as whether the statute is important (an appropriations bill) (Cameron and McCarty 2004; Bradley and Posner 2006) or originated in the Senate (Kelley and Marshall 2008). One important attribute is that my approach is largely consistent with Huber and Shipan's landmark 2002 study of political constraints on bureaucratic discretion, in which discretion is measured as the detail of congressional directives.
I empirically test these hypotheses about signing statements, ideological proximity, and other attributes of the political bargaining environment using data from President George W. Bush's administration from 2001 to 2006. My data on signing statements and their complexity come from a database of the constitutional objections embedded in the statements and their application to individual sections of statutes, which were coded by experts in constitutional law (Kinkopf and Shane 2007). I estimate the impact of different factors using a negative binomial regression that accounts for the degree of variation in the detail and complexity of objections. My first measure is the number of objections attached to a given bill; my second measure is the number of different types of objections presented across the statements attached to a given bill.
I find that the president includes more objections for a given bill as the distance increases between the president's and Congress's ideal points and that the statement presents a greater number of types of objections. I also find evidence that Bush's strategy depended on the relative liberalism of the population and changed as time proceeded; I also discuss the relative impact of these factors on the detail and complexity of the objections embedded in the signing statement.
Of course, as often is the case in presidential studies, the use of evidence from a specific administration provides a specific type of information about the use of strategy for bargaining. The position in this article is that the Bush administration is a prominent example in the recent evolution of the unitary executive--and that examination of a large number of his signing statements and how they were constructed offers the best possible evidence about the evolution of that strategy. However, since the use of signing statements also evolved in the Reagan administration, I also discuss the Bush administration's use of signing statements in this way as part of the larger historical context. Just as presidential speechmaking reached new levels of use over time, so too has the use of constitutional signing statements changed in terms of levels. The Bush administration offers a useful lens to understand this evolution, but the broader context is important for understanding the importance of recent events.
The study proceeds as follows. I next offer a framework for understanding the president's use of signing statements. Second, I offer a specification of a model that accounts for previous research on signing statements. I then describe my estimation approach and findings. Finally, I discuss this theory, this test environment, and results.
Signing Statements and Bargaining
In 2006, the American Bar Association (ABA) put together a task force to examine the growing use of signing statements by presidents, following increasing awareness and concern about President Bush's use of signing statements in the case of several high-profile statutes. The centerpiece of the ABA report was that the president should not resort to the use of what they essentially saw as vetoes occurring after the president chose to sign bills. The report argued that the president had a "constitutional obligation to veto any bill that he believes violates the Constitution in whole or in part" (ABA 2006). The signing statement was becoming "a kind of back-door, override-proof, line-item veto power for Presidents, one that the Founders never intended Presidents to have" (Savage 2007, 5).
The fallout from the report was astounding. The ABA House of Delegates voted to voice the opinion that presidents should not employ signing statements since its use was "contrary to the rule of law and our constitutional separation of powers" (Savage 2007, 5). Constitutional law scholar Peter Shane framed the use of signing statements as the "flowering of faux law", an assault on the rule of law coming in the form of the Bush...