Presidential unilateral action: partisan influence and presidential power

Abstract

Focusing on executive orders (EOs) and extending previous models, we present a general theoretical framework of unilateral presidential action. This framework allows us to examine systematically how various possible party roles, such as agenda setting, result in legislative gridlock which, in turn, create or undercut presidential incentives, and how directional constraints on discretion undermine presidential leverage. In particular, negative agenda setting and party discipline intensify gridlock, enhancing presidential policy gains; positive agenda setting’s effect depends upon governmental regime. Empirically, majority parties consistently play some role, especially negative agenda setting, regardless of the threshold used to define EO significance, while party discipline is more pronounced with higher thresholds. Also, while a majority party median is crucial for constraining the direction of how presidents use discretion with lower thresholds, a chamber median (whose preference may be induced by party pressure) is key with higher thresholds.

Appendix: Formal presentation of model, proposition 1, and expected incentives of unilateral action

Assume that every player, except the court, has a symmetric single-peaked preference with an ideal point whose notation coincides with that of the player. So, in step (5) of the game sequence, each legislative pivot will not veto if and only if the new status quo established by the agenda setter in Congress makes him or her better off than the status quo generated from the unilateral presidential action (i.e., if the former is closer to his or her ideal point than the latter). Given these decisions made by legislative pivots, we can obtain the set of status quos unaltered in the legislative stage comprising the equilibrium gridlock interval EGI (Krehbiel 1998). By Chiou and Rothenberg (2009), G _L  = min{A, L} and G _R  = max{A, L}, where G _L and G _R are denoted as the EGI’s left and right bounds, respectively, and where A and L denote the agenda setter and the set of legislative pivots in a given game. Furthermore, denote y as the status quo in the beginning of the legislative stage (i.e., at the start of step (3)). The agenda setter A’s proposal z is a function of y, as given below³⁸:

$$z\left\{ {\begin{array}{ll} { = \hbox{max} \{ A,2G_{\it R} - y\} } & {{\text{if y > }}G_{\it{R}} } \\ { = y} & {{\text{if }}G_{\it{L}} < y \le G_{\it{R}} } \\ { = \hbox{min} \{ A,2G_{L} - y\} } & {{\text{if }}y \le G_{\it L} } \\ \end{array} } \right.$$

We formally express alternative assumptions about directional constraints on presidential discretion’s use as follows. Let M = min{M _H , M _S } be the chamber median further away from the president. Let MP = min{MP _H , MP _S } under divided government and MP = max{MP _H , MP _S } under unified government. For the president’s unilateral action x, we can state the following three rules, each of which corresponds to the mechanisms mentioned above sequentially:

C1:

unilateral presidential action will be sustained if and only if $x - q/ \le t$

C2:

unilateral presidential action will be sustained if and only if $\left| {x - q} \right| \le t$ and $(x - q)(M - q) \ge 0$ for $q \ne M$

C3:

unilateral presidential action will be sustained if and only if $\left| {x - q} \right| \le t$ and $(x - q)(MP - q) \ge 0$ for $q \ne MP.$

In words, rule C1 means that the president’s unilateral action cannot move the status quo more than t. Rule C2 (C3) implies that, in addition to the constraint in C1, the president’s action is permissible only for those moves in the direction of the chamber medians (majority party medians) in both legislative chambers.

Given the result above, we can now state the equilibrium unilateral action, denoted by $x*$, with unilateral action occurring when $x* \ne q$.³⁹

Proposition 1

Given that Nature draws t and q, ⁴⁰

• (1) Under rule C1, when $A < G_{R}$,

$$x^{*} \left\{ \begin{array}{ll} = q, \, &{\rm if} \, q \le \hbox{min} \{ G_{L} - t/2,A - t\} \, {\rm or} \, q > 2G_{R} - A + t \hfill \\ = \hbox{min} \{ q + t,G_{R} \} , \,& {\rm if} \, \hbox{min} \{ G_{L} - t/2,A - t\} < q \le G_{R} \hfill \\ = \hbox{max} \{ q - t,G_{R} \} , \,& {\rm if} \, G_{R} < q \le 2G_{R} - A + t. \hfill \\ \end{array} \right.$$

When $A = G_{R}$

$$x^{*} \left\{ \begin{array}{ll} = q, \, &{\rm if} \, q < \hbox{max} \{ G_{L} - t/2,2G_{L} - A\} \, or \, q > G_{R} \hfill \\ = \hbox{min} \{ q + t,G_{R} \} , \,& {\rm if} \, \hbox{max} \{ G_{L} - t/2,2G_{L} - A\} < q \le G_{R} . \hfill \\ \end{array} \right.$$

• (2) Under rule C2 (C3), $x^{*}$ is the same as under rule C1, except that $x^{*} = q$ for $q \in (M,G_{R} )$ (q locating between MP and G _R ).⁴¹

We calculate the expected incentives for unilateral action by integrating possible unilateral action over some distributions of discretion and status quos, i.e., $\Pr (x^{*} \ne q) = \iint {\Pr (x^{*} \ne q|t,q)}g(t)g(q)dtdq$, where $g(t)$ and $g(q)$ are the distributions of discretion and the status quos, respectively, and $\Pr (x^{*} \ne q|t,q)$ represents the probability of unilateral action given $t$ and $q$ being drawn. To reiterate, and as seen in Proposition 1, $\Pr (x^{*} \ne q|t,q) = 1$ for some $q$ regardless of the values of $t$, but $\Pr (x^{*} \ne q|t,q) = 0$ for some others. Therefore, $\Pr (act) = \Pr (x^{*} \ne q)$ formally defines the unconditional expected incentive for unilateral action. As mentioned, this implies that a higher $\Pr (act)$ is associated with more significant EOs being issued during a given Congress, a hypothesis which we can examine empirically for each of our 18 models.