Share this:

Joshua Kennedy 80x108This week has brought the news that House Speaker John A. Boehner is planning a lawsuit against President Obama over his use of executive orders after months of concerns from Republicans over what they term Obama’s ‘imperial presidency’. But are the executive orders issued by the president actually acted upon by the agencies meant to carry them out? Using twenty years of data on executive orders, Joshua Kennedy finds that, on average, any given executive order has only a 2.5 percent probability of being implemented. He finds that the president can increase this probability by naming the agency involved and if the agency is headed by an ideological friend. He also warns that if the president appoints agency staff on the basis of their politics, rather than competence, then they may risk ending up with less competent management, which will drag down their implementation of executive orders.

Concerns about presidential power are as old as the presidency itself, irrespective of which party may be in control of the White House at a given time. These concerns are understandable given our broad expectations as citizens for a government that is both representative and that does not concentrate too much power in any sector.

One often-discussed facet of presidential power is the executive order, an administrative tool that presidents can use to direct the behavior of federal agencies. Though these directives carry the force of law, they cannot contradict congressional statute and can easily be undone at the president’s direction. They are, therefore, more ephemeral methods of establishing policy than standard laws, but nevertheless their impact can be particularly significant: as one familiar example, EO 9981 issued by Harry Truman led to the desegregation of the military, a goal the president was able to accomplish by using his authority as commander-in-chief to direct the military departments. A noble order, to be sure, yet many worry about the ability of the president to use these directives for purposes that may not be so noble.

The question remains, however: just how impactful are executive orders (EOs) in affecting real policy change within the federal bureaucracy? Though the presidency may seem more oriented for action than the Congress, given the fact that there is a single actor at the head of the institution, the truth of the matter is that the executive branch is comprised of many individuals with their own preferences and goals. The president may nominally head the bureaucracy, but with so many administrative layers in play it is worth asking when, how, and if these orders are faithfully implemented. In new research I set out to demonstrate the conditions under which agencies respond to executive orders from the president and offer some important conclusions about the true limitations of these supposedly “unilateral” directives. I find that if the president names the agency to carry out the EO, requests that the rule be formally promulgated, and if the agency head is ideologically aligned with the president, then the agency will be much more likely to implement the order.

Put simply, executive orders are not self-executing by their very nature; their implementation depends on the willingness of agency employees to put them into place. It would be unreasonable (and indeed impossible) for a single individual, the president, to make certain that his or her directives are being effectively enacted at the departmental level. What agency-specific factors, then, increase or decrease the likelihood of an order being carried out?

I compiled unique data to answer this question, counting as a response the publication of a final rule or regulation by an agency in the Federal Register or Code of Federal Regulations that directly acknowledges the executive order as part of the regulation’s authority. The data covers 1989-2011, and the baseline probability that any single order will be implemented is slightly under two and a half percent. At first blush this seems like a remarkable rate of non-responsiveness, but it is critical to note that orders can be wildly disparate in their content, and not every directive requires a response. For example, many EOs deal with the creation of temporary advisory committees, and there is no need for a regulation to accomplish this.

Not surprisingly, the two best predictors of agency response are whether the president names affected agencies directly and whether formal rule promulgation is requested within the order. Naming the agency increases the likelihood of a response to approximately eight and a half percentage points. Requesting formal promulgation raises the likelihood of a response to around thirty-five percentage points, and doing both simultaneously predicts a response rate of over sixty-seven percent.

Of note here is that sixty-seven percent is still short of what presidents would want, so it is clear that even when agencies receive seemingly forceful directives there is no guarantee of a response. What other factors affect the chance that an agency will implement an executive order? I considered three explicitly. The first is the agency’s level of decentralization, by which I mean how focused are the agency’s responsibilities and mission? An organization like the Department of Defense is comprised of numerous sub-agencies with employees that have varying degrees of specialization, but not every agency is as complex.

The second factor is the agency’s general ideological orientation. We could imagine that an organization with serious substantive policy oppositions to the president might resist the heavy hand of an executive order. The third factor is the agency’s level of politicization, which I define as the percentage of employees who are political appointees rather than career officials.

As it happens, neither the agency’s decentralization nor ideological opposition to (or symmetry with) the president has a significant effect on the likelihood of a response to an executive order. We might expect that decentralized agencies, because of their complexity, are less likely to implement directives as they may get “lost in the shuffle.” This, however, is not the case. It is perhaps more surprising that ideological congruence between the president and the agency has no noticeable effect on the likelihood that a directive will be implemented. In other words, for example, conservative agencies seem just as responsive to liberal presidents as they are to conservative presidents.

A separate model I tested, however, also incorporated the ideological orientation of the agency head, rather than just the agency more generally speaking. In this case, an agency head more opposed to the president’s ideological position predicts lower rates of response to an executive order. This suggests that it is the organization’s upper management, and perhaps not its more career-oriented bureaucrats, that can drive the effect of ideology on implementation.

Figure 1 – Effect of Ideological Distance Between Agency Head and President on Responsiveness

Kennedy Fig 1

Politicization also has a noticeable effect. Though presidents may seek greater responsiveness by appointing more loyal bureaucrats to positions of influence in the bureaucracy, politicization actually leads to lower levels of responsiveness. What explains this finding? Research by David Lewis provides a possible answer, as he argues that there can be an important trade-off between responsiveness and competence when it comes to matters of politicization. This is to say the presidents seeking greater influence in agencies via the appointments process may subsequently wind up with a less efficient organization.

Figure 2 – Effect of Agency Politicization on Responsiveness

Kennedy Fig 2

It is possible that this loss of efficiency is to blame for lower rates of responsiveness to executive orders in more politicized agencies. Presidents may seek control through their appointees, but perhaps instead they are ending up with less competent management. It is critical to note, however, that this explanation is purely speculative: I do not explicitly test for an agency’s competence, and it is certainly not the case that all political appointees are necessarily less competent than their career-oriented counterparts.

While much more work remains to be done to understand how executive orders work in practice, what is clear is that these directives are multifaceted and the act of changing policy hardly stops once the president issues an order. It must still be effectively carried out by bureaucrats, and two lessons are clear: first, management matters, as an agency head who does not share the president’s ideology may lead to lower rates of responsiveness. Second, presidents who attempt to gain greater control through the appointments process may paradoxically hinder their chances at affecting real policy change because of a possible loss in organizational efficiency. 

This article is based on the paper ‘“‘Do This! Do That!’ and Nothing Will Happen”: Executive Orders and Bureaucratic Responsiveness’, in American Politics Research.

Please read our comments policy before commenting.

Note:  This article gives the views of the author, and not the position of USApp– American Politics and Policy, nor of the London School of Economics. 

Shortened URL for this post:


About the author  

Joshua Kennedy 80x108Joshua KennedyGeorgia Southern University
Joshua Kennedy completed his Ph.D. at the University of Colorado, Boulder in 2014. Beginning Fall 2014 he will be an Assistant Professor of Political Science at Georgia Southern University. His research interests principally focus on the American presidency, unilateral presidential power, and the federal bureaucracy.