Bureaucratic Autonomy

“Just because you do not take an interest in politics does not mean politics will not take an interest in you,” – this quotation is often attributed to Pericles, the prominent Athenian statesman of the 5th century BC. The point Pericles is making here is that citizens ignore politics at their own peril, as the decisions of politicians will bear on people in very real ways whether they know it or not. Social scientists and other academics have long been interested in the politics of administrative agencies for a similar reason: bureaucracy is not typically interesting to learn about (that’s not exactly true, some people surely find bureaucracy interesting), but the activities of bureaucrats touch virtually every aspect of modern life, and thus are certainly deserving of serious attention. Here at DPP, agencies like FEMA, the Army Corps of Engineers, and the EPA (among many others) are of particular interest because of the work they do in disaster policy and management.

One topic of much debate in scholarly circles concerns the tension between bureaucratic efficiency and democratic responsiveness. That is, one of the virtues of administrative governance is (in theory) political independence – making decisions and implementing policies on the bases of sound economic and scientific determinations made by experts in their respective fields instead of being decided on the basis of congressional log-rolling or parochialism. The presumptive downside is the same as the virtue – independence from politicians means that bureaucratic practices may run far from the mainstream of political will (which might be good or bad), and policies may be unresponsive to the needs of affected groups [see my previous posts on the different goals of politicians and bureaucrats for more on these issues].

For these reasons I think it is worthwhile to give some thought to the relative independence of those bureaucratic agencies that are tied up in the politics of property and disasters – and note that “independence” is not a fixed quantity, rather it varies over time. A solid understanding of the relative autonomy of an agency is necessary to understanding bureaucratic activity and policy – that is, a fairly autonomous agency will be motivated to act by very different considerations than an agency that is tightly dependent on Congress or some Congressional committee.

In a short history of the Army Corps of Engineers, Todd Shallat argues that the Corps has actively sought large projects to extend federal responsibility for natural resources, and that in doing so, it has “blurred the line between policymaking and program implementation” (1989, 7). Shallat argues that the Corps has generally built its authority in three main ways: through it’s “military approach” to action and problems, using “discretionary powers,” and by taking advantage of “the flexibility of its original mission” (1989, 9). Over time, the Corps has frequently sought high cost, highly complex, and innovative (often experimental) projects to bring “scientific glory” to nation (and to the Corps itself, no doubt). That is to say, there has always been a political side to the Corps’ considerations. Shallat also notes that the Corps has been known to influence policy by “broad”, perhaps even dubious, interpretations of the law (1989, 17).

A more recent study of the Army Corps of Topographical Engineers has yielded some similar – and some new – insights on the topic. William Adler argues that “the Topogs” enjoyed “conditional bureaucratic autonomy” because of the long tenure of agency officials, their expertise on their subject, and lax oversight due to relative ignorance and disinterest by their principals (i.e. Congress) (2012, 111). Further, Adler argues that semi-autonomous, uniformed bureau chiefs at the War Department were crucial to many of these socioeconomic projects. As such, bureaus “such as the Corps of Topographical Engineers developed the ability to help shepherd their preferred policies into law or to create new policies on their own initiative when Congress declined to accede to their wishes” (2012, 110).

In an excellent case-study of FEMA, Patrick S. Roberts argues that a reputation for competence and efficiency can be a source of bureaucratic autonomy and independence (it is an unstable source, however). More specifically, he argues that under James Witt, the director of FEMA during the Clinton presidency, FEMA clarified its mission (by adopting the “all hazards” approach), made a concerted effort to improve customer service, and very importantly, consciously aligned its goals with reelection goals of Congressmen and the president. In doing so, FEMA went from laughing stock to a genuine model bureaucracy. FEMA’s reputation as the effective “all hazards” disaster agency allowed it to act independently – but also left the agency exposed to destructive forms of politicization when it failed to live up to that reputation (e.g. after Katrina).

These scholars have pointed to several general conditions that foster bureaucratic independence: 1) discretionary powers, and relatedly, 2) the ability to broadly interpret Congressional grants of authority, 3) reputation for expertise, 4) reputation for efficiency, 5) lax oversight, and 6) aligning agency goals with Congressional and presidential reelection goals. [Additionally, Shallat argues that the Corps’ militaristic approach is itself an independent source of autonomy, though he does not really tell us how that militarism imbues autonomy – I think this in and of itself deserves some serious study.] It is not necessary that an agency meet all of these conditions to be autonomous – autonomy is clearly a matter of degree, not an all-or-nothing proposition.

Further, all of these factors are conditional – it must be remembered that Congress can exercise significant influence over agencies, and no amount of reputation or discretionary interpretation can save an agency from concerted Congressional discipline. Still, Congress rarely goes after agencies with much force (typically Congress is satisfied to scapegoat agencies for benefit of media soundbites). Reputation, as Professor Roberts points out, is a very unstable source of independence. Thus these factors vary over time, as the exigencies of the moment changes, as political and social interests move from topic to topic, and as personnel and leadership in agencies turn over. Still, these factors do functionally increase the ability of an agency to act as an autonomous governor.

These ideas are of particular interest to us here at DPP because they suggests that the Corps of Engineers and FEMA –federal bureaucracies responsible planning and implementing many disaster and emergency management programs – as well as other agencies may be more independent than most bureaucracies are generally thought to be, both today and in the past. Moreover, they give us a good way to think about how any agency might be able to act independent of – or perhaps even contra to – the will of Congress and/or the president.

By the same token, these features suggest ways in which agencies are able circumvent or ignore public pressure. For example, many of the factors necessary for conditional autonomy have (arguably) been met within the Corps of Engineers at numerous points in its history. Further study is undoubtedly necessary, but by reminding us that the Corps has some unique characteristics and, perhaps, that it should not be analyzed in the same manner as other agencies, These scholars’ arguments could pave the way for a significantly better understanding of how flood mitigation policy in the Mississippi River Valley has developed over the last two centuries. It might tell us, for example, that “normal” Congressional politics cannot explain much of the development of flood mitigation policy – or it might tell us that policy developed as it did because the Corps was held close under the Congressional lash.

Again, bureaucratic autonomy is neither an unmitigated evil nor an unalloyed good. Bureaucratic autonomy was considered a great virtue, and thus was a primary goal of Progressive reformers reacting to the vestiges of the spoils system which frequently saw policy created and implemented in ways detrimental to the public interest (see for example, Carpenter, The Forging of Bureaucratic Autonomy). Autonomy is independence, however – including independence from “We, the People”, the ostensible American sovereigns (and our agents in Congress). This is to say, we should be cognizant of the give-and-take between independence and responsiveness. A wholly responsive agency is one that is prone to capture by special interests, and, history suggests, quite unlikely to produce any good policy. A wholly independent agency, on the other hand, is prone abuses and fundamentally antidemocratic. Striking a balance is never easy – and the “right” balance in any particular case is likely to depend more on one’s policy preferences than on some principled theory of bureaucratic politics. Still, effort to strike some principled, empirically grounded, generally applicable balance between autonomy and independence would surely not be effort wasted.


Adler, William D. 2012. “State Capacity and Bureaucratic Autonomy in the Early United States: The Case of the Army Corps of Topographical Engineers.” Studies in American Political Development 26 (October), 107-124.

Roberts, Patrick S. 2006. “FEMA and the Prospects for Reputation-Based Autonomy.” Studies in American Political Development 20(Spring): 57-87.

Roberts, Patrick S. 2006. “FEMA After Katrina.” Policy Review 137(June/July): 15-33.

Shallat, Todd. 1989. “Engineering Policy: The U.S. Army Corps of Engineers and the Historical Foundations of Power.” The Public Historian 11(3): 6-27.

(Editor’s Note: While some of the articles above may be gated — meaning you’ll need institutional access or have to pay for the article — we’ve provided links to them for those who do have access or those who are willing to pay. We would love for all articles to be available to everyone, but publishers just aren’t willing to do that at this time.)

Resilience and the “Engineering Mindset”

In a recent article in Risk Analysis, Park and several colleagues argue that recent disasters such as Hurricane Katrina, the Deepwater Horizon oil spill, and the crisis at the Fukushima nuclear power plant have renewed interest in “resilience” in system design. I strongly recommend this article to readers, as it offers an interesting take on catastrophe management and mitigation from the perspective of systems engineering (systems engineering is concerned with managing complex systems throughout their life cycles – including during crises; when the concept is applied to communities [i.e. understanding a community as a system], a system engineer seeks to plan for the community in such a way that it fails minimally and recovers quickly). One thing that is of particular interest to our readers is their argument regarding the engineering mindset, and its relevance to “resilience” versus traditional hazard mitigation approaches. In the words of Park et al., engineers tend to look for “problems to be solved rather than conditions to be managed” (2013, 360). While the engineering mindset is clearly crucial to disaster preparation and response, it is not necessarily (or always) the best, and certainly not the only mode of response. That is, the engineering mindset is useful and important, but it should not animate the sole voice in any design or response situation, including disaster situations.

Resilience is typically defined as “the capacity to adapt to changing conditions without catastrophic loss of form or function” (pp. 356). Park and colleagues argue, however, that resilience is not a property that system has or lacks (as the preceding definition suggests), rather, they suggest that resilience is an emergent property – an “outcome of a recursive process that includes: sensing, anticipation, learning, and adaptation” (pp. 356). That is, resilience is a dynamic property like risk itself, a function of the system. Resilience derives from the basic elements of a system, and the interplay between those elements. Thus argue Park et al., systems (whether a household, a bridge, a city, etc.) should be engineered not to be disaster-proof, but rather to survive and effectively weather a disaster (think fail-safe versus safe-fail; a “fail-safe system is one that can never fail, a safe-fail system is one designed to fail safely (as opposed to catastrophically) in the event of a failure). A resiliency-oriented system, then, is one whose designers recognize that it is genuinely impossible to build something that is truly fail-safe, or “disaster-proof”, and to instead emphasize the importance of designing and building systems that can fail without imposing massive harm to people or infrastructure.

I’ll try to clarify with an example of the “engineering mindset” from research conducted by the editors of this blog (and published here, if you’re interested in reading further) [note that this is a military-engineering mindset, not a perfect representation of the presumptive “typical” engineering mindset]. In post-Katrina New Orleans, the Army Corps of Engineers was tasked with cleaning up debris in wrecked portions of the city (the Corps had many tasks, this was only one of them). The Corps pursued this objective with vigor that one would expect from a militarized bureaucracy. Unfortunately, in their haste to fulfill their demolition mission, the Corps ran roughshod over citizens’ due process and property rights, routinely violated its own administrative procedures, and consequently demolished a large number of homes before home owners were able to make any claims as to whether they were actually so badly damaged that they actually need to be demolished. Many property owners claimed that their homes or businesses should never have been destroyed – and some were destroyed before the owners were allowed to return to the city to view the damage. The Corps claimed that the work was necessary for the health and safety of people of New Orleans. Unfortunately, the work was done so quickly that we’ll never know which of these claims is true (or how true each is). For these and similar reasons, the Corps was subject to numerous lawsuits from citizens and activists seeking relief from the Corps’ ardor for its task.



Figure 1 shows the number of demolitions conducted in the Lower 9th Ward (black bar) as compared to the total number of demolitions conducted in all other neighborhoods combined (gray bar). The data presented cover all months from February of 2006 through September of 2007.

This is not intended to disparage the Corps, per se, but rather to point out that engineers, including the Corps, also create unintended consequences with their laser-like foci on “solving problems.” In post-New Orleans the Corps could have avoided some major headaches by slowing down, thinking about due process and property rights issues, the importance of community in places like the Lower 9th Ward, and so (not only would they have avoided violating peoples’ rights, they also would have obviated more than a few lawsuits).

This example is a modest departure from the core concepts I’m discussing here. In physical terms, the New Orleans flood-protection system was theoretically engineered to protect the city from a direct hit from a hurricane; for many well-documented reasons, that system failed. In reality, it is almost certainly impossible (and is certainly cost-prohibitive) to build a flood-protection system around New Orleans that is truly “hurricane proof.” If the Corps would heed the advice of Park et al., they would of course take pains to protect the city from flooding, but would also invest in resilient infrastructure designed to weather inevitable flooding disaster. The example I offer though concerns social resilience. Specifically, the Corps’ collective “engineering mindset” motivated it to solve a physical problem without thought of the non-physical costs of its activities.

In short, I am inclined to think that the rising tide of resilience-oriented systems (both social and physical) reflects a positive step in the development of disaster management and mitigation techniques and theories. The work of Park et al. (2013) offers a cogent and accessible case for resilience, and is well worth reading for individuals interested in disaster and emergency management – perhaps especially from a political perspective.



Park, J. et al. 2013. “Integrating Risk and Resilience Approaches to Catastrophe Management in Engineering Systems.” Risk Analysis 33(3): 356-367.

Hatcher, Laura J., Logan Strother, Randolph Burnside and Donald Hughes. 2012. “The USACE and Post-Katrina New Orleans: Demolitions and Disaster Clean-Up,” Journal of Applied Social Science 6(2): 176-190.

See also:

Aguirre, B.E. 2007. “Dialectics of Vulnerability and Resilience.” Georgetown Journal on Poverty Law & Policy 14(1): 39-59.